article_title
stringlengths 1
1.44k
| abstract
stringlengths 10
22.8k
| contributor
stringlengths 29
38.3k
| journal
stringlengths 3
246
| full_text
stringlengths 22
2.53M
|
|---|---|---|---|---|
Arsenic Cancer Risk Confounder in Southwest Taiwan Data Set
|
<p>Quantitative analysis for the risk of human cancer from the ingestion of
inorganic arsenic has been based on the reported cancer mortality experience
in the blackfoot disease (BFD)–endemic area of southwest
Taiwan. Linear regression analysis shows that arsenic as the sole
etiologic factor accounts for only 21% of the variance in the
village standardized mortality ratios for bladder and lung cancer. A previous
study had reported the influence of confounders (township, BFD
prevalence, and artesian well dependency) qualitatively, but they have
not been introduced into a quantitative assessment. In this six-township
study, only three townships (2, 4, and 6) showed a significant positive
dose–response relationship with arsenic exposure. The other
three townships (0, 3, and 5) demonstrated significant bladder and
lung cancer risks that were independent of arsenic exposure. The data
for bladder and lung cancer mortality for townships 2, 4, and 6 fit
an inverse linear regression model (<italic>p</italic> < 0.001) with an estimated threshold at 151 μg/L (95% confidence
interval, 42 to 229 μg/L). Such a model is consistent
with epidemiologic and toxicologic literature for bladder cancer. Exploration
of the southwest Taiwan cancer mortality data set has clarified
the dose–response relationship with arsenic exposure by
separating out township as a confounding factor.</p>
|
<contrib contrib-type="author"><name><surname>Lamm</surname><given-names>Steven H.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001077">1</xref><xref ref-type="aff" rid="af2-ehp0114-001077">2</xref><xref ref-type="aff" rid="af3-ehp0114-001077">3</xref></contrib><contrib contrib-type="author"><name><surname>Engel</surname><given-names>Arnold</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001077">2</xref></contrib><contrib contrib-type="author"><name><surname>Penn</surname><given-names>Cecilia A.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001077">1</xref><xref ref-type="aff" rid="af2-ehp0114-001077">2</xref></contrib><contrib contrib-type="author"><name><surname>Chen</surname><given-names>Rusan</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001077">4</xref></contrib><contrib contrib-type="author"><name><surname>Feinleib</surname><given-names>Manning</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001077">1</xref></contrib>
|
Environmental Health Perspectives
|
<p>Southwest (SW) Taiwan has been the site for health studies for more than 45 years, initially
because of the discovery of a unique peripheral
vascular disease, blackfoot disease (BFD), that led to gangrenous amputation
of the extremities and later because of the associations between
high arsenic levels in local artesian well water and a variety of diseases, including
cancers. A review of the history of these studies is
useful as they have been used to estimate the carcinogenic risk from
the ingestion of inorganic arsenic.</p><p>Epidemiologic studies on BFD were conducted by the Institute of Public
Health of the National Taiwan University. All cases of BFD were found
to have used the local artesian wells (<xref rid="b7-ehp0114-001077" ref-type="bibr">Chen and Wu 1962</xref>), which were found to have high levels of arsenic (0.35–1.10 ppm) and
algae (<xref rid="b8-ehp0114-001077" ref-type="bibr">Chen et al. 1962</xref>). A high level of arsenic was suspected to be the most likely causal factor, although
organic toxins from the algae were also considered (<xref rid="b8-ehp0114-001077" ref-type="bibr">Chen et al. 1962</xref>), as later were fluorescent or humic substances (<xref rid="b15-ehp0114-001077" ref-type="bibr">Lu 1990</xref>). BFD was particularly prevalent in the townships of Pei-men, Hsieh-chia, and
Pu-tai (<xref rid="b7-ehp0114-001077" ref-type="bibr">Chen and Wu 1962</xref>).</p><p>In 1965 a dermatologic survey was conducted in 37 villages in the BFD-endemic
area, which found close associations between BFD, signs of chronic
arsenicism (hyper-pigmentation and keratosis), and skin cancer (<xref rid="b24-ehp0114-001077" ref-type="bibr">Tseng et al. 1968</xref>). This study showed a dose–response relationship for both BFD
and skin cancer with respect to arsenic level, when well water arsenic
level was stratified as < 0.30 ppm, 0.30–0.59 ppm, and ≥ 0.60 ppm (mg/L).</p><p>A death certificate study (1968–1982) was conducted on 84 villages
in the BFD-endemic area that found a dose–response relationship
between standardized mortality ratios (SMRs) of certain cancers
and BFD prevalence rates of the villages and townships (<xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. 1985</xref>). The highest cancer rates were for the townships of Pei-men, Hsieh-chia, and
Pu-tai, and the lowest for the township of I-chu. Cancer SMRs
were greatest in the villages where only artesian wells were used as the
drinking water.</p><p>This study was revamped in order to study arsenic levels as a risk factor. Because
well water arsenic levels from the 1964–1966 survey (<xref rid="b12-ehp0114-001077" ref-type="bibr">Kuo 1968</xref>) were known only for 27 of the 84 villages, the study was extended to
include an additional 15 villages from the neighboring townships of Yen-shui
and Hsia-in. The expanded study (<xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. 1989</xref>) also stratified villages by their median well water arsenic level as < 0.30, 0.30–0.59, and ≥ 0.60 ppm. The data of <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> with the addition of an exposure stratum of median well water arsenic
level < 0.1 ppm were used to calculate cancer potency indices (<xref rid="b5-ehp0114-001077" ref-type="bibr">Chen et al. 1992</xref>). The potency indices for excess lifetime risk due to an intake of 10 μg/kg/day
of arsenic were about 1.5 × 10<sup>−2</sup> for male and female bladder cancers and lung cancers.</p><p><xref rid="b17-ehp0114-001077" ref-type="bibr">Morales et al. (2000)</xref> used the <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> raw study data to model arsenic-attributed cancer risk, using village-specific
census and mortality data and village-specific median well water
arsenic levels rather than using the three-level arsenic strata methodology
employed by <xref rid="b24-ehp0114-001077" ref-type="bibr">Tseng et al. (1968)</xref> and <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref>. Risks were calculated using a variety of models, either with SW Taiwan
or all of Taiwan as the comparison population, and with no comparison
population. The exposure–response curves showed a wide range
of slopes at low-dose levels, depending upon choice of comparison population, model, and
model parameters. <xref rid="b17-ehp0114-001077" ref-type="bibr">Morales et al. (2000)</xref> performed a stratified SMR analysis using narrower strata than those of
the <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> study.</p><p>The U.S. Environmental Protection Agency (<xref rid="b25-ehp0114-001077" ref-type="bibr">U.S. EPA 2001</xref>, <xref rid="b26-ehp0114-001077" ref-type="bibr">2005</xref>) and the National Research Council (<xref rid="b18-ehp0114-001077" ref-type="bibr">NRC 1999a</xref>, <xref rid="b20-ehp0114-001077" ref-type="bibr">2001</xref>) have each used particular fits from the <xref rid="b17-ehp0114-001077" ref-type="bibr">Morales et al. (2000)</xref> data set analysis as their basis for estimating the carcinogenic risk
from the ingestion of inorganic arsenic. These estimates, like those of <xref rid="b17-ehp0114-001077" ref-type="bibr">Morales et al. (2000)</xref>, used the median village arsenic level as the sole explanatory variable
for estimating the risk. The estimates also assumed a linear no-threshold
model for extrapolation to low doses.</p><p>Previous analysis of the cancer mortality in the BFD-endemic area had shown
township to be a discriminating factor for carcinogenic risk. <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> stated that “the higher the BFD prevalence rate of a township, the
greater the SMRs for cancers of bladder, kidney, skin, and lung of
the township.” This article adds township as an explanatory
variable in the quantitative analysis of the study by <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref>.</p><p>We have expanded a data set of <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> received from A. Schulman (U.S. EPA) and L. Ryan (Harvard University) by
adding township and well water arsenic level information published
by the <xref rid="b19-ehp0114-001077" ref-type="bibr">NRC (1999b)</xref> and have examined the dose–response relationship for both the
low-dose villages and high-dose villages with respect to possible explanatory
variables, including township, apart from arsenic level (<xref ref-type="table" rid="t1-ehp0114-001077">Table 1</xref>). Data can be obtained at StatLib website hosted by Carnegie Mellon University (<xref rid="b4-ehp0114-001077" ref-type="bibr">Carnegie Mellon University 2006</xref>); click on “Get Data,” then search the term “arsenic.”</p><sec sec-type="materials|methods"><title>Materials and Methods</title><p>Data underlying the study by <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> are analyzed in this report. That study examined the 1973–1986 cancer
mortality for the 42 study villages, based on death certificates
that were coded to the <italic>International Classification of Diseases: Manual of the International Statistical
Classification of Diseases, Injuries, and Causes of Death, 8th
Revision</italic> (<xref rid="b27-ehp0114-001077" ref-type="bibr">World Health Organization 1965</xref>), person-year distributions based on Taiwan household registration office
data and in the data set of <xref rid="b17-ehp0114-001077" ref-type="bibr">Morales et al. (2000)</xref> for study population and bladder and lung cancer deaths, and median village
well water arsenic level as reported by the <xref rid="b19-ehp0114-001077" ref-type="bibr">NRC (1999b)</xref>. The outcome variables in the Morales et al. data set were limited to
bladder and lung cancer mortality.</p><p>We performed linear regression analyses of the village data using Excel (Microsoft
Corp., Redmond, WA) least-squares analysis with the village
SMRs as dependent observations and with median village well water arsenic
level as a continuous predictor. Approximate 95% confidence
intervals (CIs) on <italic>x</italic>-intercepts from inverse linear regression were calculated. SMR = 100 was
used as the reference value because it represents the SMR value
at which the risk is not increased above that of the reference population. Stratified
analyses have been based on township (individual
or grouped), number of reported wells per village (one vs. multiple), artesian
well dependency, and exposure strata (low vs. higher: < 0.13 ppm
vs. > 0.25 ppm). The townships were identified by number by
the <xref rid="b19-ehp0114-001077" ref-type="bibr">NRC (1999b)</xref>. The number of wells in each village was also identified by the <xref rid="b19-ehp0114-001077" ref-type="bibr">NRC (1999b)</xref>. Twenty villages had only one reported well, 10 had two wells, 7 had three
to five wells, and 5 had six or more wells. Questions of measurement
error or exposure misassignment have been raised, either because of
the wide range of measurements that might be behind a median (<xref rid="b16-ehp0114-001077" ref-type="bibr">Morales et al. 1999</xref>) or because the measured well may not be the actual well of use (<xref rid="b3-ehp0114-001077" ref-type="bibr">Brown and Chen 1995</xref>). Well multiplicity was examined as a surrogate for distinguishing between
those villages that entered the analysis on the basis of a “median” village
well water arsenic level (i.e., those villages
with multiple wells) and those villages whose exposure assignment
was not based on a median (i.e., single-well villages). A village was
operationally defined as being “artesian well dependent” if
all its wells had arsenic levels > 0.325 ppm, based on the observation
of <xref rid="b8-ehp0114-001077" ref-type="bibr">Chen et al. (1962)</xref> of arsenic levels in artesian wells in the endemic area, which was cited
by both <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> and <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref>. The definition of “low-dose village exposure” was based
on the gap in the median village well water arsenic levels between 0.13 and 0.25 ppm
and the observation by <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> that arsenic content of well water samples had two clusters, with the
low-dose cluster < 0.25 ppm.</p><p>We calculated SMRs for individuals 20 or more years of age for individual
villages or groups of villages, with SW Taiwan data used for the reference
population. We also calculated confidence intervals on SMRs from
a Poisson distribution on the basis of the observed number of cancer
deaths. Analysis of variance (ANOVA), regression analyses, and hierarchical
regression analyses have been performed using SAS software (SAS
Institute, Cary, NC).</p></sec><sec sec-type="results"><title>Results</title><p>The present analysis is based on the 1973–1986 bladder and lung
cancer mortality experience of the 42 villages studied by <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref>—27 from the study by <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> of the four BFD-endemic townships and 15 from two neighboring townships. In
total, this comprises 478,776 person-years of observation (≥ 20 years
of age) and 181 bladder and 268 lung cancer deaths (449 total
deaths). The primary exposure variable is the median village well
water arsenic level that represents one well for 20 villages and multiple
wells (range, 2–47) for the other 22 villages. Overall, linear
regression analysis of the 42-village bladder and lung cancer SMR
data on the median village well artesian level (parts per million) showed
an explanatory model that accounted for only 21% of the
variability (<italic>p</italic> = 0.03) (<xref ref-type="fig" rid="f1-ehp0114-001077">Figure 1</xref>). The <italic>y</italic>-intercept is at SMR = 189, above the no-increased risk line at
SMR = 100.</p><p>A number of investigators have considered various alternative sources of
variability in order to better understand the data. For example, the
median village well water level may to some degree misrepresent the village-specific
exposure because it ignores the marked variability in
well water arsenic measures for a number of villages. The extreme example
is that of village G in township 0, which has a range of well water
arsenic measurements from 0.010 to 0.770 ppm but enters the analysis
with a median of 0.030 ppm. <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref>, the study that provided data for two-thirds of these villages (27/42 = 64%), pointed out that township, BFD prevalence, and
artesian well dependency were each significant determinants of the cancer
risk. <xref rid="b13-ehp0114-001077" ref-type="bibr">Lamm et al. (2003)</xref> demonstrated that artesian well dependency was at least as strong a determinant
of bladder cancer risk in this data set as was median village
well water arsenic level.</p><p>We have used the village-specific data that are available for the villages
studied by <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> to examine singly and collectively the various hypothesized alternative
sources of variability in the regression of cancer mortality risk (bladder
and lung) on arsenic exposure level (median village well water
arsenic level). Unfortunately, village-specific BFD prevalence ratios
are not in the available data.</p><p>The NRC data (<xref rid="b19-ehp0114-001077" ref-type="bibr">NRC 1999b</xref>) allowed us to examine three dichotomous potential sources of variability—well
multiplicity, artesian well dependency, and exposure
strata (low dose vs. higher dose)—as well as township. Each of
the dichotomous variables was found to be a predictive factor for bladder
and lung cancers in the one-way ANOVA, with <italic>p</italic>-values of 0.01–0.02.</p><p>We have looked at township as a stratifying variable and examined the dose
response between bladder and lung cancer and median village well water
arsenic level for each of the six showed a statistically significant
association with <italic>p</italic>-values of 0.019, 0.011, and 0.019, respectively. Townships 0 and 3 did
not show significant association, with <italic>p</italic>-values of 0.24 and 0.91, respectively. As township 5 had only two villages
in the study, the <italic>p</italic>-value is undefined. The townships were grouped into those that showed
a significant dose–response relationship with the arsenic exposure (townships 2, 4, and 6) and those that did not (townships 0, 3, and 5). Township
group (as a dichotomous variable) was also a significant
determinant of risk (<italic>p</italic> = 0.006) in a one-way ANOVA analysis.</p><p>Stepwise regression analysis demonstrated that median village well water
arsenic level (parts per million) and township group were the only significant
factors. The inclusion of either well multiplicity or artesian
well dependency did not significantly increase the explanatory power
of the model. The same result was found for bladder and lung cancer
individually and combined and by sex individually and combined. The single
exception was that the mortality risk for female lung cancer was
best explained by township group and well multiplicity but not by median
arsenic level.</p><p>The combined bladder and lung cancer mortality was significantly higher
in the higher-dose villages (SMR = 402.5; 95% CI, 358 to 447) than
in the low-dose villages (SMR = 178.5; 95% CI, 148 to 209). The
dose–response relationship was examined
within each subgroup (i.e., the low-dose and higher-dose villages), with
the expectation that the slope of the dose–response relationship
for the higher-exposure villages and the slope of the dose–response
for the low-dose exposure villages would be similar (<xref ref-type="fig" rid="f2-ehp0114-001077">Figure 2</xref>). However, the regression lines for the linear regression analysis of
the cancer risk against median village well water arsenic level (parts
per million) for the low-dose (<italic>n</italic> = 18) and higher-dose villages (<italic>n</italic> = 24) were dissimilar. The higher-dose villages showed a significant
positive dose–response relationship for bladder and lung
cancer SMR (<italic>n</italic> = 24; <italic>R</italic><sup>2</sup> = 0.24; <italic>p</italic> = 0.02), whereas the low-dose villages showed a nonsignificant
negative dose response for bladder and lung cancer SMR (<italic>n</italic> = 18; <italic>R</italic><sup>2</sup> = 0.04; <italic>p</italic> = 0.42).</p><p>The data for the low-dose villages were then examined to seek a basis for
the negative slope. Township was examined as a potential source of
cancer risk variability for the low-dose villages. For instance, the five
low-dose range villages in township 3 had median well water arsenic
levels of 10, 32, 32, 56, and 65 ppb (μg/L), with bladder and
lung cancer SMRs of 649, 348, 568, 587, and 154, respectively, whereas
the five low-dose range villages in township 4 had median village well
water arsenic levels of 42, 60, 80, 123, and 126 ppb (μg/L) and
bladder and lung cancer SMRs of 0, 81, 75, 96, and 114, respectively. The 18 villages
that make up the low-dose group were from five townships. There
were no villages from township 5 among the low-dose villages. Stratification
of the bladder and lung cancer SMR analysis by
township showed that the SMRs for townships 2, 4, and 6 were similar (and < 100) and
that the SMRs for townships 0 and 3 were each significantly
elevated (<xref ref-type="fig" rid="f3-ehp0114-001077">Figure 3</xref>).</p><p>Township-stratified linear regression analysis for bladder and lung cancer
against median village well water arsenic level (micrograms per liter; ppb) for
the low-dose villages yielded positive slopes (dose–response
relationships) for townships 4 and 6 and negative slopes (dose–response
relationships) for townships 0 and 3. Township 2 had
only one village in the low-dose village group, and township 5 had
none.</p><p>Thus, among the low-dose villages, townships 0 and 3 each showed a bladder
and lung cancer mortality risk that was significantly increased over
the background, greater than the risks in townships 2, 4, and 6 and
independent of the median village well water arsenic exposure levels. The
identity of this independent township factor is not known, although
it may relate to the BFD prevalence township factor that <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> had previously reported. Nonetheless, these analyses indicate that some
geographically related risk factor exists in townships 0 and 3 that
may be independent of arsenic exposure levels and appears to confound
the dose–response analysis in the data set. As township 5 had
no village among the low-dose villages and only two among the higher-dose
villages, there were no data to examine risk at low dose in township 5.</p><p>The data for the 42 villages were separated into two groups—the
data for townships 2, 4, and 6 that each showed a significant relationship
to the median village well water arsenic level (<italic>p</italic> = 0.01–0.02) and the data for townships 0, 3, and 5 that
showed some other possible nonarsenic carcinogenic risk factor and
no apparent relationship to arsenic level [<italic>p</italic> = 0.24, 0.91, and undefined (based on only two data points), respectively].</p><p><xref ref-type="fig" rid="f4-ehp0114-001077">Figure 4</xref> shows the dose–response relationship for bladder and lung cancer
mortality risk and median village well water arsenic level by township
group (townships 2, 4, and 6 vs. townships 0, 3, and 5) for the full
set of 42 villages (range, 0.01–0.934 ppm). Median village
well water arsenic level explains 75% of the variability in the
bladder and lung cancer SMRs for the 20 villages in township group 2, 4, 6 (<italic>p</italic> < 0.001) and only 5% of that for the 22 villages in township
group 0, 3, 5 (<italic>p</italic> = 0.30). The regression line for bladder and lung cancer SMR for
township group 2, 4, 6 meets the no-increased-risk line (SMR = 100) at 151 μg/L with a 95% CI of 42 to 229 μg/L. The
regression line for bladder and lung cancer SMR for township
group 0, 3, 5 is above an SMR of 350 at 0 μg/L.</p><p>The regression line from bladder and lung cancer SMR for township group 2, 4, 6, meets
the no-increased-risk line (SMR = 100) for males
at 119 μg/L with a 95% CI of −70 to 229 μg/L
and for females at 191 with a 95% CI of 66 to 280 μg/L.</p><p>Separate examination of the bladder cancer SMRs for the villages in township
group 2, 4, 6 showed a significant dose–response relationship
with median village well water arsenic level (<italic>p</italic> < 0.001) that meets the no-increased-risk line (SMR = 100) at 139 μg/L
with a 95% CI of −7.5 to 233 μg/L. For
the villages in township group 0, 3, 5, the dose–response
relationship is not significant (<italic>p</italic> = 0.61), and the regression line is above SMR = 750 at 0 μg/L (data
not shown). Likewise, examination of the lung cancer
SMRs for the villages in township group 2, 4, 6 showed a significant
dose–response relationship with median village well water
arsenic level (<italic>p</italic> < 0.001) that meets the no-increased-risk line (SMR = 100) at 164 μg/L
with a 95% CI of 26 to 257 μg/L. For
the villages in township group 0, 3, 5, dose–response relationship
is not significant (<italic>p</italic> = 0.16) and the regression line is above SMR = 250 at 0 μg/L (data
not shown).</p><p>Further analysis was conducted among those townships (townships 2, 4, and 6) that
appeared to show some relationship between cancer risk and
arsenic exposure, eliminating those townships that appeared to have some
other strong risk factor for bladder and lung cancer that was not related
to the arsenic exposure (townships 0, 3, and 5).</p><p><xref ref-type="fig" rid="f5-ehp0114-001077">Figure 5</xref> demonstrates, for male bladder cancer mortality in township group 2, 4, 6, a
regression line (<italic>p</italic> < 0.001) that meets the no-increased-risk line (SMR = 100) at 125 μg/L
with a 95% CI of −19 to 218 μg/L, and
for female bladder cancer mortality, a regression line (<italic>p</italic> = 0.001) that meets the no-increased-risk line (SMR = 100) at 163 μg/L
with a 95% CI of −80 to 294 μg/L.</p><p><xref ref-type="fig" rid="f6-ehp0114-001077">Figure 6</xref> demonstrates for male lung cancer mortality in township group 2, 4, 6, a
regression line (<italic>p</italic> < 0.001) that meets the no-increased-risk line (SMR = 100) at 117 μg/L
with a 95% CI of −274 to 273 μg/L, and
for female lung cancer mortality, a regression line (<italic>p</italic> = 0.001) that meets the no-increased-risk line (SMR = 100) at 217 μg/L
with 95% CI of 31 to 273 μg/L.</p><p><xref ref-type="fig" rid="f7-ehp0114-001077">Figure 7</xref> summarizes for township group 2, 4, 6 the exposure levels at which the
regression line meets the no-increased-risk line (SMR = 100) with
the 95% CI of that intercept in micrograms per liter arsenic. Most
of these levels are in the range of 100–200 μg/L
and statistically significantly different from zero overall. The
fact that the intercept levels for females are greater than those for
males and tend to be statistically significantly different from zero while
those for the males do not, may reflect the effects of cigarette
smoking, which is far less prevalent among the females than among the
males.</p></sec><sec sec-type="discussion"><title>Discussion</title><p>Analysis of the bladder and lung cancer mortality data from the 42 study
villages of the BFD-endemic area of SW Taiwan from the <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> study showed only a 21% explanatory power when median village
well water arsenic level is used as the sole explanatory variable in the
linear regression model (<italic>p</italic> = 0.03). Separation of low-dose villages from higher-dose villages
showed that the explanatory power resided among the higher-dose villages
and did not appear among the low-dose villages. Previous analysis
of the cancer mortality in the BFD-endemic had shown that township
was a discriminating variable for the cancer mortality risk in this area (<xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. 1985</xref>). The mortality experience for the low-dose villages was examined to explore
the contribution potentially made by township.</p><p>Within that analysis, certain townships (townships 0 and 3) were found
each to have an increased cancer risk that was not related to arsenic
exposure level, and other townships (townships 2, 4, and 6) were found
to have cancer risks similar to each other with SMRs less than 100. The
full 42-village data set was stratified by township group in order
to examine the dose–response relationship in the absence of the
interfering township factor. A significant positive dose response was
seen with the data from townships 2, 4, and 6, with the explanatory
power raised to 75% (<italic>p</italic> < 0.001). No significant positive dose response was seen with the data
from townships 0, 3, and 5 (<italic>p</italic> = 0.30). Removal of the data from the townships influenced by
the township factor markedly improved the fit of the data and the explanatory
power of the model. If the SW Taiwan data are going to be used
in formal risk analysis for ingested arsenic, the analysis should be
restricted to the data from townships 2, 4, and 6.</p><p>The identity of this independent township factor is not known and may relate
to the BFD prevalence township factor that <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> had previously demonstrated. The data from <xref rid="b7-ehp0114-001077" ref-type="bibr">Chen and Wu (1962)</xref> showed among the townships that the average number of wells per village, the
proportion of villages with artesian well dependency, and the BFD
prevalence were significantly correlated. Without knowing the specific
identity and location of each study village, these various factors
cannot be disentangled. <xref ref-type="fig" rid="f8-ehp0114-001077">Figure 8</xref> shows that the BFD case distribution (<xref rid="b9-ehp0114-001077" ref-type="bibr">Ch’i and Blackwell 1968</xref>) was not uniform over the study area. The number of BFD cases seems to
be heavily concentrated in the three townships to the left of the figure.</p><p>The township factor may be a reflection of a selection bias occurring because
the well water sampling was focused on the villages with high prevalence
of BFD. Because the villages in the <xref rid="b28-ehp0114-001077" ref-type="bibr">Wu et al. (1989)</xref> study were specifically selected because well water arsenic data from
the 1960s existed for them, such a selection bias may have entered that
could skew the interpretation of the results. High BFD prevalence was
found in most but not all of the townships of Pei-men, Hsieh-chia, and
Pu-tai; moderate prevalence, in the southern part of I-chu; and near
absence in Hsia-in and Yen-shui. Again, it may not be possible to disentangle
the relationship between BFD and any other possible etiologic
factor for any outcome of interest, at least until each village can
be specifically located on this map.</p><p>The “true” underlying structure of the arsenic dose–response
relationship is more likely to be seen in the analysis of
the cancer risks in townships 2, 4, and 6 than in that of the entire
data set that contains the influence of the township factor. Removal
of this extraneous source of variability has allowed for a clearer examination
of the dose–response relationship. Interestingly, analysis
of the bladder cancer mortality data for townships 2, 4, and 6 appears
to display a fit to a threshold model. This finding for bladder
cancer mortality is independently seen in males and females, suggesting
that it may reflect a real phenomenon.</p><p>A thresholdlike model indicating that the bladder cancer mortality risk
does not increase with exposure levels < 150 μg/L is consistent
with other epidemiologic data. An ecologic analysis of the white
male bladder cancer risk in the United States found no increase over
an arsenic exposure range of 3–59 μg/L (<xref rid="b14-ehp0114-001077" ref-type="bibr">Lamm et al. 2004</xref>). Case–control bladder cancer studies found no increased risk
in the United States for exposures < 80 μg/day (<xref rid="b23-ehp0114-001077" ref-type="bibr">Steinmaus et al. 2003</xref>) or < 160 μg/L (<xref rid="b2-ehp0114-001077" ref-type="bibr">Bates et al. 1995</xref>) or in Argentina with exposures > 200 μg/L (<xref rid="b1-ehp0114-001077" ref-type="bibr">Bates et al. 2004</xref>). The prospective cohort study in northeastern Taiwan reported that the
multivariate-adjusted relative risk of urinary tract cancer was statistically
significant for residents who drank well water containing arsenic
at levels > 100 μg/L (<xref rid="b10-ehp0114-001077" ref-type="bibr">Chiou et al. 2001</xref>). Cigarette smoking still remains a risk factor for bladder cancer.</p><p>A threshold, sublinear, or hormetic model for bladder cancer and inorganic
arsenic exposure has been proposed from toxicologic studies, based
on a number of modes of action (<xref rid="b22-ehp0114-001077" ref-type="bibr">Snow et al. 2005</xref>). All three model forms have in common an inflection point wherein the
risks at exposure levels greater than the inflection point do not predict
the risks for exposures less than the inflection point exposure level. Suggested
modes of action include generation of oxidative stress, perturbation
of DNA methylation patterns, inhibition of DNA repair, and
modulation of signal transduction pathways (<xref rid="b21-ehp0114-001077" ref-type="bibr">Schoen et al. 2004</xref>). Cellular proliferation has been recognized as an early step in the development
of bladder cancer by substances that are not mutagenic, such
as inorganic arsenic (<xref rid="b11-ehp0114-001077" ref-type="bibr">Cohen 2002</xref>).</p><p>A similar inflection point is seen in the analysis of the lung cancer mortality
data and is statistically significant in females but not in males, although
that question may not be answerable in the absence of smoking
history. Perhaps analysis of lung cancer mortality in the northeastern
Taiwan prospective study where smoking histories have been obtained
would allow for a better assessment.</p></sec><sec sec-type="conclusion"><title>Conclusion</title><p>Exploration of the SW Taiwan cancer mortality data set from the BFD-endemic
area has identified significant confounding variables. Certain townships
in the study demonstrate a significantly increased cancer risk
at low-dose exposures that is independent of the village arsenic exposure
levels. Removal of the data confounded by the township factor reveals
an underlying dose–response curve for bladder and lung cancer
mortality and arsenic level (median village well water arsenic level) that
displays as a thresholdlike model. Such a model would not be
contradicted by either the epidemiologic or the toxicologic literature, at
least for bladder cancer.</p><p>The variability in the village SMRs, particularly the low-dose high-rate
villages, is not explained by any biological model that uses arsenic
as the sole explanatory variable. It is reasonable to assume that they
are high for some nonarsenic reason. As <xref rid="b6-ehp0114-001077" ref-type="bibr">Chen et al. (1985)</xref> demonstrated, the cancer mortality in the study area showed a dose–response
relationship with BFD prevalence at both a village and township
level of analysis. We have followed through from these findings
and have analyzed separately the data from the townships that do not
appear to have been strongly influenced by this second factor that may
be related to BFD prevalence. It is likely that the cancer dose–response
relationship is more clearly seen in those areas that have
little or no confounding by BFD prevalence.</p></sec><sec><title>C<sc>orrection</sc></title><p>The following figures have been modified from the originally published
article online:</p><p><xref ref-type="fig" rid="f1-ehp0114-001077">Figure 1</xref>: Now includes the horizontal regression line at SMR = 100 to indicate
the level at which no increased risk is observed.</p><p><xref ref-type="fig" rid="f3-ehp0114-001077">Figure 3</xref>: Now includes township group 0, 3, 5 (Twn grp 0, 3, 5). A vertical line
has been added to separate the township-specific results within each
township, and the 95% CI values have been added to the Twn 2 data
point.</p><p><xref ref-type="fig" rid="f7-ehp0114-001077">Figure 7</xref>: 95% CI values have been corrected for female bladder and female
lung cancers.</p></sec>
|
Chronic Obstructive Pulmonary Disease Mortality in Diesel-Exposed Railroad
Workers
|
<p>Diesel exhaust is a mixture of combustion gases and ultrafine particles
coated with organic compounds. There is concern whether exposure can
result in or worsen obstructive airway diseases, but there is only limited
information to assess this risk. U.S. railroad workers have been
exposed to diesel exhaust since diesel locomotives were introduced after
World War II, and by 1959, 95% of the locomotives were diesel. We
conducted a case–control study of railroad worker deaths
between 1981 and 1982 using U.S. Railroad Retirement Board job records
and next-of-kin smoking, residential, and vitamin use histories. There
were 536 cases with chronic obstructive pulmonary disease (COPD) and 1,525 controls
with causes of death not related to diesel exhaust or
fine particle exposure. After adjustment for age, race, smoking, U.S. Census
region of death, vitamin use, and total years off work, engineers
and conductors with diesel-exhaust exposure from operating trains
had an increased risk of COPD mortality. The odds of COPD mortality increased
with years of work in these jobs, and those who had worked ≥ 16 years
as an engineer or conductor after 1959 had an odds ratio
of 1.61 (95% confidence interval, 1.12–2.30). These
results suggest that diesel-exhaust exposure contributed to COPD mortality
in these workers. Further study is needed to assess whether this
risk is observed after exposure to exhaust from later-generation diesel
engines with modern emission controls.</p>
|
<contrib contrib-type="author"><name><surname>Hart</surname><given-names>Jaime E.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001013">1</xref><xref ref-type="aff" rid="af2-ehp0114-001013">2</xref></contrib><contrib contrib-type="author"><name><surname>Laden</surname><given-names>Francine</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001013">1</xref><xref ref-type="aff" rid="af2-ehp0114-001013">2</xref><xref ref-type="aff" rid="af3-ehp0114-001013">3</xref></contrib><contrib contrib-type="author"><name><surname>Schenker</surname><given-names>Marc B.</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001013">4</xref></contrib><contrib contrib-type="author"><name><surname>Garshick</surname><given-names>Eric</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001013">1</xref><xref ref-type="aff" rid="af5-ehp0114-001013">5</xref></contrib>
|
Environmental Health Perspectives
|
<p>Diesel exhaust is a complex mixture of particles (< 1.0 μm in
diameter) and combustion gases. These particles have organic compounds
adsorbed on an elemental carbon core that may be inhaled deep into
the lung. Regulation of diesel-exhaust exposure in the United States has
been largely based on its potential to be a human lung carcinogen [<xref rid="b51-ehp0114-001013" ref-type="bibr">U.S. Environmental Protection Agency (EPA) 2002</xref>]). Information regarding the occurrence of nonmalignant respiratory
effects in humans was determined to be inadequate (<xref rid="b51-ehp0114-001013" ref-type="bibr">U.S. EPA 2002</xref>). Diesel exhaust is a common exposure because many occupational groups (underground
miners; bridge and tunnel workers; dockworkers; truck drivers; farm-workers; auto, truck, and bus maintenance garage workers; operators
of heavy construction equipment; and railroad workers) are regularly
exposed to diesel exhaust at work [<xref rid="b35-ehp0114-001013" ref-type="bibr">National Institute for Occupational Safety and Health (NIOSH) 1988</xref>]. Although occupational exposures to dusts and fumes have been
shown to contribute greatly to the burden of chronic obstructive pulmonary
disease (COPD) (<xref rid="b5-ehp0114-001013" ref-type="bibr">Balmes et al. 2003</xref>; <xref rid="b14-ehp0114-001013" ref-type="bibr">Christiani 2005</xref>; <xref rid="b25-ehp0114-001013" ref-type="bibr">Hnizdo et al. 2002</xref>; <xref rid="b34-ehp0114-001013" ref-type="bibr">Meldrum et al. 2005</xref>; <xref rid="b48-ehp0114-001013" ref-type="bibr">Trupin et al. 2003</xref>), previous studies have had limited ability to examine the health effects
of a specific occupational group or exposure.</p><p>Workers within the U.S. railroad industry have been exposed to diesel exhaust
since the industry converted from steam to diesel locomotives after
World War II (<xref rid="b50-ehp0114-001013" ref-type="bibr">U.S. Department of Labor Bureau of Labor Statistics 1972</xref>). There have been few epidemiologic studies assessing whether exposure
to diesel exhaust is associated with increased mortality due to non-malignant
respiratory diseases (<xref rid="b7-ehp0114-001013" ref-type="bibr">Bergdahl et al. 2004</xref>; <xref rid="b49-ehp0114-001013" ref-type="bibr">Ulvestad et al. 2000</xref>). In this case–control study, we investigated a possible association
between exposure to diesel exhaust from operating locomotives and
COPD mortality.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Study population</title><p>The data set of railroad workers obtained with the assistance of the U.S. Railroad
Retirement Board (RRB) has been described elsewhere (<xref rid="b22-ehp0114-001013" ref-type="bibr">Garshick et al. 1987a</xref>; <xref rid="b29-ehp0114-001013" ref-type="bibr">Larkin et al. 2000</xref>). Briefly, the RRB manages the retirement benefits for all U.S. railroad
workers with ≥ 10 years of railroad employment. Next of kin
can draw benefits only by notifying the RRB of the worker’s death; therefore, it
was possible to conduct a case–control study
of deaths in the railroad industry that included workers with long-term
railroad employment.</p></sec><sec><title>COPD case series</title><p>Between 1 March 1981 and 28 February 1982, with the cooperation of the
RRB, we collected incident deaths of U.S. railroad workers. Workers were
eligible for inclusion in the study if they were born in 1900 or later
and had no mention of suicide, accidental causes, or unknown causes
on the death certificate. The original case–control study was
designed primarily to study the association of exposure to diesel exhaust
from locomotives with lung cancer mortality. Results from the lung
cancer mortality case–control study have been published previously (<xref rid="b22-ehp0114-001013" ref-type="bibr">Garshick et al. 1987a</xref>). During death certificate coding, because a specific code for COPD was
not available in the <italic>International Classification of Diseases</italic>, Eighth Revision [ICD-8; <xref rid="b55-ehp0114-001013" ref-type="bibr">World Health Organization (WHO) 1967</xref>], we noted deaths where COPD was specifically listed as the underlying
cause of death. These deaths, plus ICD-8 codes 490–493 (bronchitis, including
chronic bronchitis, emphysema, and asthma), were
considered to be cases for this analysis. There were 536 cases with
COPD or these related conditions listed as the underlying cause.</p></sec><sec><title>Control series</title><p>Controls were selected from the pool of remaining deaths. This pool included
cancer deaths other than lung cancer and deaths originally selected
to be controls in the lung cancer study. Persons were excluded from
the control group if they had lung cancer (ICD-8 code 162) listed anywhere
on the death certificate, or other causes potentially related to
exposure to diesel exhaust or fine particulate matter (PM). Because
there is a potential association between diesel exhaust or fine particle
exposure with selected cardiovascular diseases (deaths with ICD-8 codes 410–414, 420–425, or 427–429) or with bladder
cancer (ICD-8 code 188) (<xref rid="b8-ehp0114-001013" ref-type="bibr">Boffetta and Silverman 2001</xref>), individuals with these causes listed anywhere on the death certificate
were not included in the control group. The selection of these cardiovascular
ICD-8 codes was based on associations between cardiovascular
mortality and fine PM reported by <xref rid="b38-ehp0114-001013" ref-type="bibr">Pope et al. (2004)</xref>. To assess the sensitivity of the results to the selection of the control
group, we also conducted analyses excluding all cardiovascular deaths (ICD-8 codes 390–458) as controls. Deaths with COPD or related
conditions listed as a secondary cause of death or other chronic
respiratory diseases (ICD-8 codes 517–518) as primary or secondary
cause were also not included as controls. After these exclusions, 1,525 individuals
remained to serve as controls. The protocol was approved
by the Brigham and Women’s Hospital and VA Boston Healthcare
System institutional review boards.</p></sec><sec><title>Exposure to diesel exhaust</title><p>Diesel exposure was determined based on yearly job code provided by the
RRB. Since 1959, the RRB has maintained a computerized work history for
all workers including a yearly Interstate Commerce Commission job code
and number of months worked during the year. Transition from steam
to diesel locomotives began largely after World War II. In 1946, 10% of
the locomotives in service were diesel powered, and by 1959, 95% of
locomotives in use were diesel powered (<xref rid="b50-ehp0114-001013" ref-type="bibr">U.S. Department of Labor Bureau of Labor Statistics 1972</xref>). Therefore, we chose 1959 as the effective start of diesel exposure for
our primary analyses. Alternatively, we conducted analyses using 1946 as
the start of exposure to account for work during the transition
period. Because it is unusual for railroad workers to change jobs and
because yearly job codes were unavailable before 1959, job in 1959 was
used to indicate pre-1959 exposure category. Using information from an
industrial hygiene survey that we conducted (<xref rid="b56-ehp0114-001013" ref-type="bibr">Woskie et al. 1988a</xref>, <xref rid="b57-ehp0114-001013" ref-type="bibr">1988b</xref>) and review of industry practices, subjects were characterized as exposed
or unexposed to diesel exhaust based on yearly job code. The engineers (engineers
and firemen) and conductors (conductors, brakemen, and
hostlers) who worked on operating trains were determined to be “diesel
exposed.” Other jobs, including ticket agents, station
agents, signal maintainers and other maintenance of way workers, car
repair workers, and clerks, were considered “unexposed.” We
determined that the shop job codes were not specific for locomotive
shops and that the shop workers should be considered a separate
group containing a mixture of diesel-exposed and unexposed workers. From
the industrial hygiene survey, information was available on mean level
of cigarette-smoke–adjusted respirable PM for each of the
major job groups. Mean levels for workers on operating trains, that is, engineers
and conductors, were 71 μg/m<sup>3</sup> and 89 μg/m<sup>3</sup>, respectively. Mean levels were lower for workers with clerical jobs (33 μg/m<sup>3</sup>) and signal maintainers (58 μg/m<sup>3</sup>). Because of uncertainties in assigning individual-level historical exposures, analyses
were conducted assessing COPD risk between exposed and
unexposed workers rather than specifically incorporating the PM exposure
estimates. Diesel-exhaust exposure was defined by cumulative years
of work in the engineer or conductor job group starting in 1959. Shop
exposure was defined by cumulative years of work in the shop job group
starting in 1959.</p></sec><sec><title>Possible confounders</title><p>Information was available on several potential confounders. Given the small
number of minorities in the database, individuals were coded as Caucasian
or other race. Using the state of death from the death certificate, U.S. Census
region of death was assigned as Northeast, South, Midwest, or
West (<xref rid="b10-ehp0114-001013" ref-type="bibr">Bureau of the Census 1993</xref>), to control for geographic variation. Questionnaires completed by next
of kin provided information on smoking history; ever use of vitamins
C, A, and E and multivitamins (which may protect against the development
of COPD) (<xref rid="b2-ehp0114-001013" ref-type="bibr">Anto et al. 2001</xref>; <xref rid="b32-ehp0114-001013" ref-type="bibr">MacNee 2000</xref>; <xref rid="b42-ehp0114-001013" ref-type="bibr">Romieu and Trenga 2001</xref>; <xref rid="b54-ehp0114-001013" ref-type="bibr">Viegi et al. 2001</xref>); and the population (1–2,499, 2,500–49,999, ≥ 50,000 persons, or
unknown) of the place where the deceased had lived
for most of his life as an indicator of urban or rural location. Smoking
status was coded as “never smoker” if the deceased
never smoked, as “current smoker at death” if the deceased
had smoked within the year of death, as “former smoker” if
the smoker had stopped smoking before the year of death, and
as “unknown smoker” if smoking information was not available. Average
number of cigarettes smoked and age started and stopped
smoking were used to calculate pack-years of cigarettes, and years
quit smoking was calculated for all former smokers by subtracting age
last smoked from age at death. Indicator variables were created for missing
information. Total years of retirement were calculated by subtracting
year of retirement from year of death and were used to account
for a potential healthy worker survivor effect, an effect where workers
who remain in the workplace tend to be healthier than those who leave (<xref rid="b3-ehp0114-001013" ref-type="bibr">Arrighi and Hertz-Picciotto 1993</xref>, <xref rid="b4-ehp0114-001013" ref-type="bibr">1994</xref>). Occupational categories with the potential for asbestos exposure were
available from the original case–control study (<xref rid="b22-ehp0114-001013" ref-type="bibr">Garshick et al. 1987a</xref>). Classification was based on the results of a survey of railroad employees (<xref rid="b23-ehp0114-001013" ref-type="bibr">Garshick et al. 1987b</xref>) and on a review of railroad, medical, and industrial literature and was
included as an indicator variable. Other exposures associated with
railroad work included silica as a result of track sanding operations (<xref rid="b56-ehp0114-001013" ref-type="bibr">Woskie et al. 1988a</xref>). However, specific information regarding this potential exposure was
not available to us.</p></sec><sec sec-type="methods"><title>Statistical analysis</title><p>We estimated the association between diesel-exhaust exposure and COPD mortality
using logistic regression and present odds ratios (OR) and 95% confidence
intervals (CIs). When appropriate, tests for trend
were performed using ordinal variables (0, 1, 2, 3) for increasing exposure
categories. All analyses were performed in SAS (version 8; SAS
Institute Inc., Cary, NC).</p></sec></sec><sec sec-type="results"><title>Results</title><p>Characteristics of the cases and controls are presented in <xref ref-type="table" rid="t1-ehp0114-001013">Table 1</xref>. Fifty-nine percent of the controls died from malignant neoplasms, and 23% from
nonexcluded cardiovascular causes. Cases were more likely
to have been smokers, to smoke more cigarettes on average, to be
Caucasian, and to have died outside of the Northeast.</p><p>Logistic regression results are presented in <xref ref-type="table" rid="t2-ehp0114-001013">Table 2</xref>, categorizing diesel exposure based on years of work in the engineer or
conductor job group starting in 1959, and shop exposure as years of
work in a shop job starting in 1959. In the unadjusted model, there was
no consistent association between work in a diesel-exposed job and COPD
mortality. There was also no consistent association with work in a
shop job. After adjustment for age, there was a trend apparent only in
the diesel-exposed group. After adjustment for age at death, race, and
the healthy worker survivor effect, the highest ORs were for workers
with ≥ 16 years of work in both the diesel-exposed and shop
categories. However, the <italic>p</italic>-value for trend was again significant only for the diesel-exposed group. Adjustment
for smoking (smoking status, pack-years, and years quit) attenuated
the ORs in the diesel-exposed group, but the <italic>p</italic>-value for trend remained significant. The smoking-adjusted OR for each
additional year of work as an engineer/conductor was 1.02 (95% CI, 1.01–1.04). After smoking adjustment, there was less evidence
of a relationship between years of work and COPD mortality in the
shop workers, but risks were generally elevated. Additional models using
smoking duration and average amount smoked per day to adjust for
smoking gave similar results. Adjustment for other possible confounders (Census
region of death, asbestos exposure, and vitamin C, A, and E
and multivitamin use) did not substantially change the risk estimates
and were not significant predictors of COPD mortality. Similar analyses
were conducted based on exposure starting in 1946. The ORs adjusted
for age, race, smoking, and healthy worker survivor effect for years of
work after 1946 in a diesel-exposed job were 0–20 years, OR = 0.60 (95% CI, 0.36–1.01); 21–25 years, OR = 1.08 (95% CI, 0.75–1.56); 26–30 years, OR = 1.58 (95% CI, 1.12–2.22); and ≥ 31 years, OR = 1.82 (95% CI, 1.13–2.94). For
the shop worker group, the ORs for years of work after 1946 ranged
from 1.17 in the group with 0–20 years of work after 1946 to 2.21 for
those with ≥ 30 years; no consistent pattern of increasing
risk with increasing years of work was observed. In analyses
where deaths from all cardiovascular causes were excluded from the control
series, similar results were found (data not shown).</p></sec><sec sec-type="discussion"><title>Discussion</title><p>In this study of diesel-exposed railroad workers, after adjustment for
active smoking and other covariates, work in diesel-exposed jobs was associated
with higher risks of COPD mortality compared with work in unexposed
jobs. These risks increased with increasing years of work. The
greatest risks were observed for those individuals with the longest work
on operating trains (multivariate-adjusted OR = 1.61; 95% CI, 1.12–2.30) for engineers/conductors with ≥ 16 years
of work starting in 1959 (<italic>p</italic>-value for trend = 0.02). The shop worker group, which likely includes
a mixture of both diesel-exposed and unexposed workers, did not
demonstrate a significant trend with increasing years of work after 1959, although
the ORs were elevated. In our retrospective cohort study
of 54,973 railroad workers, we also observed a similarly elevated relative
risk (RR) of 1.41 (95% CI, 1.27–1.55) for COPD
mortality between 1959 and 1996, comparing workers in the engineer and
conductor group in 1959 with workers in the unexposed group (<xref rid="b20-ehp0114-001013" ref-type="bibr">Garshick et al. 2004</xref>). In contrast to our current case–control study, however, we were
not able to directly adjust for smoking because this information was
not available.</p><p>The U.S. railroad industry converted from steam to diesel-powered locomotives
after World War II, with a rapid increase through the 1950s. First-generation
diesel locomotives introduced during the 1940s and 1950s
were described as “smokier” than locomotives introduced
later, although historical exposure measurements are not available (<xref rid="b19-ehp0114-001013" ref-type="bibr">Eschenroeder 2004</xref>; <xref rid="b56-ehp0114-001013" ref-type="bibr">Woskie et al. 1988a</xref>, <xref rid="b57-ehp0114-001013" ref-type="bibr">1988b</xref>). In addition, some railroads operated these locomotives with the cab
in the rear, a configuration that increased exposure because the exhaust
stack was located in front of the cab rather than behind it (<xref rid="b30-ehp0114-001013" ref-type="bibr">Liukonen et al. 2002</xref>; <xref rid="b52-ehp0114-001013" ref-type="bibr">Verma et al. 2003</xref>). Locomotives introduced in the 1960s and later in the 1980s had improved
emission characteristics. Our research group conducted an exposure
assessment in the early 1980s in four smaller railroads that used a combination
of first- and second-generation locomotives. No specific marker
of diesel exposure was measured, but workers on operating trains
had mean respirable PM levels adjusted for secondhand smoke approximately
two to three times that of unexposed clerical workers (71 μg/m<sup>3</sup> and 89 μg/m<sup>3</sup> of cigarette-adjusted PM vs. 33 μg/m<sup>3</sup>) (<xref rid="b56-ehp0114-001013" ref-type="bibr">Woskie et al. 1988a</xref>, <xref rid="b57-ehp0114-001013" ref-type="bibr">1988b</xref>). These results indicate that railroad workers who were on operating trains
and whose COPD mortality was assessed in this study were more exposed
to diesel exhaust than workers not on operating trains, and most
likely had greater exposures than contemporary workers.</p><p>In the present study, we considered diesel exposure to start in 1959, the
date that the railroad industry had largely converted from steam to
diesel-powered locomotives. However, the overall proportion of diesel
locomotives in service in large railroads was 27% in 1949 and 55% in 1952 (<xref rid="b50-ehp0114-001013" ref-type="bibr">U.S. Department of Labor Bureau of Labor Statistics 1972</xref>). Therefore, many of the workers in the study probably had between 5 and 10 years
or more of additional exposure to diesel exhaust before 1959, possibly
influencing the relationship between years of exposure and
COPD mortality. We attempted to account for work before 1959 by assessing
exposure starting in 1946. As in the analysis based on exposure
starting in 1959, workers with the greatest duration of work in jobs with
diesel exposure had the greater risk of COPD mortality. It was not
possible to account for historical differences in emission characteristics
of locomotives and work practices when calculating years of exposure. However, misclassification of exposure would be nondifferential
and thus would bias the results toward the null. Another possible source
of exposure misclassification is from PM attributable to steam combustion
products before the transition to diesel, and it is possible that
this also contributed to mortality. Although there is no study of the
PM size distribution from steam locomotives, studies of the PM emitted
from coal-fired boilers indicate that only a small percentage (4–6%) are
in the fine range and respirable, that is, < 2.5 μm
in diameter (<xref rid="b12-ehp0114-001013" ref-type="bibr">Chang et al. 2004</xref>). This suggests that diesel combustion PM is more likely to be inhaled
deeply into the lung and to more strongly contribute to the effects of
exposure on COPD mortality.</p><p>The shop worker group also included workers with diesel exposure, but also
with exposure to other dusts and fumes from locomotive and nonlocomotive
repair shop operations. After adjustment for cigarette smoking, there
was no relationship between years of work and COPD mortality, although
workers with the longest duration of work did have the highest
risk. It is possible that exposure to various dusts and fumes generated
in railroad shops before 1959 in addition to diesel exposure influenced
COPD mortality. It is also possible that the mix of shop workers
in diesel locomotive shops and in shops not responsible for locomotive
repair who were not exposed to operating trains contributed to the lack
of a dose response in that group, because there was no way to separate
workers with and without exposure using job titles. The results for
the shop workers may also be imprecise because of few subjects in the
two top cumulative exposure categories.</p><p>Our results are consistent with previous studies relating occupational
exposures to dusts and fumes to the development of COPD, and air pollution
studies where exposure to PM is associated with both hospitalizations
and COPD mortality (<xref rid="b13-ehp0114-001013" ref-type="bibr">Chew et al. 1999</xref>; <xref rid="b24-ehp0114-001013" ref-type="bibr">Harre et al. 1997</xref>; <xref rid="b47-ehp0114-001013" ref-type="bibr">Sunyer 2001</xref>). Occupational exposures to mineral dust, welding and metal fumes, inorganic
and organic dusts, and vehicle exhausts have been implicated as
potentially important risk factors for COPD, including in nonsmoking
individuals (<xref rid="b2-ehp0114-001013" ref-type="bibr">Anto et al. 2001</xref>; <xref rid="b5-ehp0114-001013" ref-type="bibr">Balmes et al. 2003</xref>; <xref rid="b6-ehp0114-001013" ref-type="bibr">Becklake 1989</xref>; <xref rid="b14-ehp0114-001013" ref-type="bibr">Christiani 2005</xref>; <xref rid="b17-ehp0114-001013" ref-type="bibr">Coggon and Newman Taylor 1998</xref>; <xref rid="b21-ehp0114-001013" ref-type="bibr">Garshick et al. 1996</xref>; <xref rid="b28-ehp0114-001013" ref-type="bibr">Korn et al. 1987</xref>; <xref rid="b33-ehp0114-001013" ref-type="bibr">Mastrangelo et al. 2003</xref>; <xref rid="b34-ehp0114-001013" ref-type="bibr">Meldrum et al. 2005</xref>; <xref rid="b36-ehp0114-001013" ref-type="bibr">Oxman et al. 1993</xref>; <xref rid="b48-ehp0114-001013" ref-type="bibr">Trupin et al. 2003</xref>; <xref rid="b49-ehp0114-001013" ref-type="bibr">Ulvestad et al. 2000</xref>; <xref rid="b53-ehp0114-001013" ref-type="bibr">Viegi and Di Pede 2002</xref>; <xref rid="b54-ehp0114-001013" ref-type="bibr">Viegi et al. 2001</xref>). Additionally, in the Third National Health and Nutrition Examination
Study, 19.2% of all U.S. COPD cases and 31.1% of nonsmoking
cases were attributed to work exposures (<xref rid="b25-ehp0114-001013" ref-type="bibr">Hnizdo et al. 2002</xref>).</p><p>Experimentally, exposure to the organic compounds found in diesel exhaust
and on the surface of the particle have also been linked to allergy, airway
inflammation, and changes in airway function (<xref rid="b39-ehp0114-001013" ref-type="bibr">Prieto et al. 2000</xref>; <xref rid="b40-ehp0114-001013" ref-type="bibr">Riedl and Diaz-Sanchez 2005</xref>; <xref rid="b43-ehp0114-001013" ref-type="bibr">Rudell et al. 1996</xref>; <xref rid="b44-ehp0114-001013" ref-type="bibr">Saxon and Diaz-Sanchez 2000</xref>). Air toxics and other polycyclic aromatic hydrocarbon compounds found
in diesel exhaust may be important in the induction of such airway inflammatory
changes and possibly oxidative stress in the lung (<xref rid="b31-ehp0114-001013" ref-type="bibr">Ma and Ma 2002</xref>). Taken together, these studies support the hypothesis that occupational
exposure to diesel exhaust can contribute to the occurrence of COPD
and COPD mortality even after exposure has ceased.</p><p>Since the 1950s, researchers have shown that acute air pollution exacerbates
existing COPD and asthma and also increases their incidence. This
research includes panel studies and time series studies of daily variation
in hospitalizations (<xref rid="b37-ehp0114-001013" ref-type="bibr">Pope 2000</xref>; <xref rid="b46-ehp0114-001013" ref-type="bibr">Schwartz et al. 1996</xref>). Furthermore, studies examining the long-term effects of air pollution
exposure have consistently found increased prevalence of symptoms or
diagnoses of emphysema or COPD for areas with higher levels of PM (<xref rid="b1-ehp0114-001013" ref-type="bibr">Abbey et al. 1995</xref>; <xref rid="b45-ehp0114-001013" ref-type="bibr">Schwartz 1993</xref>; <xref rid="b46-ehp0114-001013" ref-type="bibr">Schwartz et al. 1996</xref>; <xref rid="b47-ehp0114-001013" ref-type="bibr">Sunyer 2001</xref>). In residents living in areas of high air pollution, small airway fibrosis
and PM deposition were noted in small airways, suggesting that chronic
exposure to PM results in pathologic changes in the lung (<xref rid="b9-ehp0114-001013" ref-type="bibr">Brauer et al. 2001</xref>; <xref rid="b18-ehp0114-001013" ref-type="bibr">Dai et al. 2003</xref>). Similar findings have been noted in workers with occupational dust exposures (<xref rid="b15-ehp0114-001013" ref-type="bibr">Churg and Wright 1983</xref>, <xref rid="b16-ehp0114-001013" ref-type="bibr">1985</xref>). Therefore, our findings of an association of COPD mortality with occupational
exposure to diesel exhaust are consistent with observations
from the occupational health and air pollution literature.</p><p>There are several possible sources of limitation in this study, including
classification of outcome from death certificates, classification of
exposure based on job title and an industrial hygiene survey, and information
on confounders from next of kin. Death certificates have been
shown to underestimate the true number of workers with severe COPD at
death. In the Tucson Epidemiologic Study of Obstructive Airways Disease (<xref rid="b11-ehp0114-001013" ref-type="bibr">Camilli et al. 1991</xref>), 25% of deaths with clinically documented moderate to severe
obstructive lung disease were identified using underlying cause of death
only, whereas 81% had COPD noted as either underlying or contributing
cause on the death certificate. Similarly for asthma, in a
cohort of persons from Olmstead County, Minnesota (which excluded persons
with COPD based on detailed laboratory and clinical criteria), the
specificity was 99% but the sensitivity was only 42% for
detecting asthma based on death certificate diagnosis (<xref rid="b26-ehp0114-001013" ref-type="bibr">Hunt et al. 1993</xref>). Because it would be unlikely for a physician to report obstructive lung
disease based on diesel exposure category, misclassification would
be nondifferential and the observed ORs are likely attenuated. In models
considering COPD from both underlying and contributing causes, similar
results were observed to those using only underlying cause.</p><p>In any COPD study, cigarette smoking is an important potential confounder
of the exposure–disease association. Information on smoking
status was available only from next of kin. Proxy respondents have been
shown to accurately report cigarette smoking status, including duration
of smoking and amount smoked (<xref rid="b27-ehp0114-001013" ref-type="bibr">Kolonel et al. 1977</xref>; <xref rid="b41-ehp0114-001013" ref-type="bibr">Rogot and Reid 1975</xref>). There were, in fact, small differences in smoking rates by job title, with
a slightly higher rate of smoking among the diesel-exposed workers (<xref rid="b29-ehp0114-001013" ref-type="bibr">Larkin et al. 2000</xref>). We were able to account for these differences in the smoking-adjusted
analyses, and it is unlikely that residual confounding explains the
current associations.</p></sec><sec sec-type="conclusion"><title>Conclusion</title><p>In this case–control study of railroad workers, work in jobs with
exposure to diesel exhaust was associated with increased mortality
from COPD. These elevations persist after controlling for smoking and
increased with increasing years of work in exposed jobs. Further study
of incident cases in populations exposed to diesel exhaust are needed
to assess the robustness of this relationship and whether the relationship
is observed after exposure to exhaust from later-generation diesel
engines and with modern emission controls.</p></sec>
|
Diabetes, Obesity, and Hypertension May Enhance Associations between Air
Pollution and Markers of Systemic Inflammation
|
<p>Airborne particulate matter (PM) may lead to increased cardiac risk through
an inflammatory pathway. Therefore, we investigated associations
between ambient PM and markers of systemic inflammation among repeated
measures from 44 senior citizens (≥ 60 years of age) and examined
susceptibility by conditions linked to chronic inflammation. Mixed
models were used to identify associations between concentrations of
fine PM [aerodynamic diameter ≤ 2.5 μm (PM<sub>2.5</sub>)] averaged over 1–7 days and measures of C-reactive protein (CRP), interleukin-6 (IL-6), and white blood cells (WBCs). Effect
modification was investigated for diabetes, obesity, hypertension, and
elevated mean inflammatory markers. We found positive associations
between longer moving averages of PM<sub>2.5</sub> and WBCs across all participants, with a 5.5% [95% confidence
interval (CI), 0.10 to 11%] increase per
interquartile increase (5.4 μg/m<sup>3</sup>) of PM<sub>2.5</sub> averaged over the previous week. PM<sub>2.5</sub> and CRP also exhibited positive associations among all individuals for
averages longer than 1 day, with the largest associations for persons
with diabetes, obesity, and hypertension. For example, an interquartile
increase in the 5-day mean PM<sub>2.5</sub> (6.1 μg/m<sup>3</sup>) was associated with a 14% increase in CRP (95% CI, −5.4 to 37%) for all individuals and an 81% (95% CI, 21 to 172%) increase for persons with diabetes, obesity, and
hypertension. Persons with diabetes, obesity, and hypertension
also exhibited positive associations between PM<sub>2.5</sub> and IL-6. Individuals with elevated mean inflammatory markers exhibited
enhanced associations with CRP, IL-6, and WBCs. We found modest positive
associations between PM<sub>2.5</sub> and indicators of systemic inflammation, with larger associations suggested
for individuals with diabetes, obesity, hypertension, and elevated
mean inflammatory markers.</p>
|
<contrib contrib-type="author"><name><surname>Dubowsky</surname><given-names>Sara D.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000992">1</xref></contrib><contrib contrib-type="author"><name><surname>Suh</surname><given-names>Helen</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000992">1</xref></contrib><contrib contrib-type="author"><name><surname>Schwartz</surname><given-names>Joel</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000992">1</xref><xref ref-type="aff" rid="af2-ehp0114-000992">2</xref><xref ref-type="aff" rid="af3-ehp0114-000992">3</xref></contrib><contrib contrib-type="author"><name><surname>Coull</surname><given-names>Brent A.</given-names></name><xref ref-type="aff" rid="af4-ehp0114-000992">4</xref></contrib><contrib contrib-type="author"><name><surname>Gold</surname><given-names>Diane R.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000992">1</xref><xref ref-type="aff" rid="af3-ehp0114-000992">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>Substantial epidemiologic evidence links particulate air pollution to adverse
acute cardiovascular health effects. Associations are generally
consistent across studies and robust to adjustment by smoking, weather, and
seasonality (<xref rid="b4-ehp0114-000992" ref-type="bibr">Brook et al. 2004</xref>). Although the biologic mechanisms behind these associations remain uncertain, several
investigators have hypothesized that oxidative stress
in the lungs from inhaled particulate matter (PM) leads to a systemic
inflammatory cascade that can increase cardiovascular risk among susceptible
individuals. This elevated risk can occur via increased coagulability
of the blood (<xref rid="b34-ehp0114-000992" ref-type="bibr">Seaton et al. 1995</xref>) or development and destabilization of atherosclerotic plaques (<xref rid="b11-ehp0114-000992" ref-type="bibr">Donaldson et al. 2001</xref>).</p><p>The hypothesized role of inflammation in PM-mediated toxicity is well supported
by past findings. Researchers have associated PM with influxes
of inflammatory cells into the lungs (<xref rid="b13-ehp0114-000992" ref-type="bibr">Ghio et al. 2000</xref>), enhanced production of proinflammatory cytokines by alveolar macrophages (<xref rid="b39-ehp0114-000992" ref-type="bibr">van Eeden et al. 2001</xref>), elevated systemic blood viscosity (<xref rid="b25-ehp0114-000992" ref-type="bibr">Pekkanen et al. 2000</xref>; <xref rid="b27-ehp0114-000992" ref-type="bibr">Peters et al. 1997</xref>), and increased production of inflammatory cells by bone marrow (<xref rid="b37-ehp0114-000992" ref-type="bibr">Tan et al. 2000</xref>). Acute peripheral artery narrowing (<xref rid="b3-ehp0114-000992" ref-type="bibr">Brook et al. 2002</xref>), arterial reactivity (<xref rid="b23-ehp0114-000992" ref-type="bibr">O’Neill et al. 2005</xref>), extent of atherosclerotic lesions (<xref rid="b17-ehp0114-000992" ref-type="bibr">Kunzli et al. 2005</xref>; <xref rid="b36-ehp0114-000992" ref-type="bibr">Suwa et al. 2002</xref>), and elevated risk of myocardial infarctions (<xref rid="b26-ehp0114-000992" ref-type="bibr">Peters et al. 2001a</xref>; <xref rid="b42-ehp0114-000992" ref-type="bibr">Zanobetti and Schwartz 2005</xref>) also have been associated with PM. In addition, relationships have been
found between PM and C-reactive protein (CRP) (<xref rid="b28-ehp0114-000992" ref-type="bibr">Peters et al. 2001b</xref>; <xref rid="b29-ehp0114-000992" ref-type="bibr">Pope et al. 2004</xref>; <xref rid="b32-ehp0114-000992" ref-type="bibr">Riediker et al. 2004</xref>; <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. 1999</xref>), an inflammatory marker that has been shown to be predictive of cardiovascular
disease (<xref rid="b8-ehp0114-000992" ref-type="bibr">Danesh et al. 2004</xref>; <xref rid="b31-ehp0114-000992" ref-type="bibr">Ridker et al. 2000</xref>).</p><p>Past investigations also have demonstrated that the cardiovascular impacts
of air pollution are not the same for all individuals. Enhanced susceptibility
for air pollution–related cardiovascular events has
been shown for older individuals and persons with conditions associated
with chronic inflammation such as diabetes, coronary artery disease, and
past myocardial infarctions (<xref rid="b1-ehp0114-000992" ref-type="bibr">Bateson and Schwartz 2004</xref>; <xref rid="b14-ehp0114-000992" ref-type="bibr">Goldberg et al. 2001</xref>; <xref rid="b41-ehp0114-000992" ref-type="bibr">Zanobetti and Schwartz 2002</xref>). Based on these findings, it is conceivable that the short-term effects
of PM on inflammation also may be enhanced among individuals with existing
inflammation. Little has been published to answer this question, however. Therefore, the
goals of this analysis were to evaluate short-term
associations between ambient PM and markers of systemic inflammation
in older adults and to explore susceptibility by conditions linked
to chronic inflammation such as diabetes, obesity, and hypertension (<xref rid="b18-ehp0114-000992" ref-type="bibr">Libby 2002</xref>).</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Study population</title><p>Data were collected from 44 nonsmoking seniors (≥ 60 years of age) between
March and June of 2002 under the supervision of the Harvard
School of Public Health Human Subjects Committee. All participants were
independently mobile and lived in one of four independent senior residences
in suburban St. Louis, Missouri. Individuals with atrial flutter, atrial
fibrillation, and/or a paced rhythm were excluded from participation
because heart rate variability also was assessed during this
investigation. Similarly, participants with left bundle branch blocks
were selected only if their heart rate variability could be ascertained. Individuals
with unstable angina and persons who were unable to
provide informed written consent also were excluded from participation.</p></sec><sec sec-type="methods"><title>Study design</title><p>Data for this analysis were collected as part of a more comprehensive investigation
designed to examine the cardiovascular health effects of
traffic-related pollution. The main goal of this investigation, which
included continuous electrocardiogram measurements, was to evaluate if
cumulative exposures to fresh traffic-related pollution or even moment-to-moment
changes in traffic pollution would influence autonomic function
in the elderly. To meet this objective, we asked a group of senior
adults to participate in a series of four group trips into St. Louis
for a brief activity and lunch. All trips ranged from approximately 0945 hr
to 1430 hr and included two 1-hr rides aboard a diesel-powered
shuttle bus. Subjects participated approximately once per month along
with two to seven other individuals from their residence facility, resulting
in a total of 25 trips.</p><p>A second aim of this study was to examine associations between fresh traffic-related
pollution and ambient pollution on markers of inflammation. As
part of this exploratory investigation, we collected venous blood
samples at approximately 0900 hr on the morning after each trip. Questionnaires
regarding medication and vitamin use, food consumption, and
health status were also administered at that time. Blood pressure, height, and
weight data were collected before participation in any trips.</p></sec><sec><title>Blood analyses</title><p>Venous blood was collected per participant per trip. Each sample was analyzed
promptly for white blood cells (WBCs) at the Barnes Jewish Hospital
using a Coulter Ge-S system (Beckman Coulter Inc., Fullerton, CA). The
remaining plasma was extracted by the Washington University Core
Laboratory for Clinical Studies (St. Louis, MO), preserved at −80°C, and
shipped to the Clinical and Epidemiologic Research
Laboratory at Boston Children’s Hospital. Samples stored with
sodium citrate were analyzed for CRP using immunoturbidimetric assays
on the Hitachi 917 Chemistry Analyzer (Roche Diagnostics, Indianapolis, IN) with
reagents and calibrations from Denka Seiken (Niigata, Japan). Interleukin-6 (IL-6) was analyzed using enzyme-linked immunosorbent
assays from R&D Systems (Minneapolis, MN). Only samples with sufficient
blood for all three assays were included in our analysis.</p></sec><sec><title>Exposure measurements</title><p>Ambient PM data were obtained from the U.S. Environmental Protection Agency (EPA)–funded
Supersite in East St. Louis, Illinois. Concentrations
of PM with aerodynamic diameter ≤ 2.5 μm (PM<sub>2.5</sub>) were recorded using a continuous ambient mass monitor (Andersen Instruments; Smyrna, GA) with
a Nafion diffusion dryer (Perma Pure, Toms River, NJ). Ambient
black carbon (BC) was reported using a aethalometer (McGee
Scientific, Berkeley, CA).</p><p>Additional measurements of group-level PM<sub>2.5</sub> were collected during the 48 hr preceding each blood draw in order to
better capture participants’ true exposures. These samples were
collected continuously from participants’ microenvironments
using a portable cart that followed subjects from a centrally located
area in their living facility, onto the bus, to the activity and lunch, and
finally back to their housing facility again. Installed on these
carts were a DustTrak 8520 aerosol monitor (TSI Inc., Shoreview, MN) and
Nafion diffusion drier. A Harvard Impactor (Air Diagnostics Environmental
Inc., Harrison, ME) also was positioned on the carts as a means
to calibrate the DustTrak samples because the DustTrak has been shown
to overestimate concentrations despite being well correlated with several
reference methods (<xref rid="b6-ehp0114-000992" ref-type="bibr">Chang et al. 2001</xref>).</p><p>Gaseous criteria pollutant data were obtained from the Missouri Department
of Natural Resources and the Illinois Environmental Protection Agency
monitoring station, located immediately adjacent to the Supersite. These
ambient measurements were collected using a TEI 48 analyzer (Thermo
Environmental Instruments, Franklin, MA) for carbon monoxide, an
API 200A analyzer (Teledyne, San Diego, CA) for nitrogen dioxide, a Dasibi 4108 analyzer (Dasibi, Glendale, CA) for sulfur dioxide, and a Dasibi 1008RS
analyzer for ozone. Meteorologic parameters also were obtained
from the Missouri Department of Natural Resources station and used
to calculate ambient apparent temperature, a biologic weather stress
index (<xref rid="b24-ehp0114-000992" ref-type="bibr">O’Neill et al. 2003</xref>). Indoor apparent temperature was calculated using data from a HOBO data
logger (Onset Computer, Bourne, MA) in the participants’ microenvironments. Daily
mold and pollen data were obtained from the county
health department and examined as total counts.</p></sec><sec sec-type="methods"><title>Statistical analysis</title><p>We used linear mixed models (SAS, version 8.02; SAS Institute Inc., Cary, NC) to
evaluate relationships between air pollution and markers of
inflammation. To account for correlation among the multiple measurements
collected per person, our models included random intercepts for each
subject. Although autoregressive terms were evaluated, they were ultimately
not used because likelihood ratio tests indicated that they were
unnecessary. Random slopes were not considered because a maximum of
four samples was collected from any one individual.</p><p>Exposures were evaluated in our statistical models using ambient pollutant
concentrations averaged over the 1–7 days preceding each blood
draw. These averaging times were calculated using hourly pollution
data measured at the St. Louis Supersite and were selected based on
the findings of past investigations, which found associations with indicators
of inflammation on the order of days to 1 week (<xref rid="b28-ehp0114-000992" ref-type="bibr">Peters et al. 2001b</xref>; <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. 1999</xref>). Although our main analysis focused on moving averages, we also assessed
the impact of lagged ambient concentrations ranging from 1 to 7 days
to confirm our findings. In addition, models were run using microenvironmental
PM<sub>2.5</sub> measurements averaged over the 1 and 2 days before the health measurements
to examine the effects of measurement error on resulting effect estimates. (Longer
averaging times for microenvironmental exposures could
not be examined because they were measured only during the 2 days before
the blood draws.)</p><p>Before statistical modeling, all outcome variables were transformed logarithmically
because each was highly skewed. Univariate models were then
run to examine the impact of various individual characteristics on
the outcomes. Next, single-pollutant models for each moving average were
used to evaluate ambient PM<sub>2.5</sub> as a potential predictor of each inflammatory marker. Additional models
containing a moving average of ambient BC, NO<sub>2</sub>, SO<sub>2</sub>, O<sub>3</sub>, or microenvironmental PM<sub>2.5</sub> were run to confirm our ambient PM<sub>2.5</sub> results. Due to limited samples and our interest in effect modification, multipollutant
models were generally not employed because of power
constraints. Effect estimates from our models and their 95% confidence
intervals (CIs) were transformed into percent changes and reported
per interquartile range (IQR) of a pollutant.</p><p>All models were controlled for sex, obesity, diabetes, and smoking history (ever/never). Time-varying parameters considered as potential confounders
included apparent temperature, hour, day, trip (a proxy of activity
and season), residence, mold, pollen, illness, and juice intake. Medication
and vitamin consumption on the day of the blood draw was also
examined, with specific focus on medications that might influence
inflammation or oxidative stress (i.e., statins, inhaled steroids, aspirin, ibuprofen, and
vitamins). Respiratory medications other than inhaled
steroids were not examined because of insufficient daily variation. Of
all the potential confounders evaluated, we selected ambient and
microenvironmental apparent temperature, mold, pollen, vitamins, trip, and
hour for our models based on a significant relationship with one
of the outcomes at the 0.2 level among the unexposed (lower 50th percentile
by exposure) (<xref rid="b20-ehp0114-000992" ref-type="bibr">Mickey and Greenland 1989</xref>). Trip, hour, and vitamins were included in our models as categorical
terms. All other variables were modeled as linear except for mold, which
was treated as a linear spline with one knot for WBCs. These parameterizations
were selected using loess smoothing in S-Plus 2000 (MathSoft, Cambridge, MA) as
well as likelihood ratio tests and Akaike information
criterion comparisons in SAS. Residual checks confirmed the parameterization
of our models, and sensitivity analyses indicated that our
findings were qualitatively robust to confounder selection.</p><p>Effect modification by conditions linked to chronic inflammation was investigated
using interaction terms for diabetes, obesity, and hypertension. An
indicator for concurrent diabetes, obesity, and hypertension
was also created. For this analysis, we defined obesity by a body mass
index ≥ 30 kg/m<sup>2</sup> and diabetes by report of a doctor diagnosis or use of diabetes medications. Hypertension
was defined as a systolic blood pressure ≥140 mm
Hg, diastolic blood pressure ≥ 90 mm Hg, and/or taking
hypertensive medications.</p><p>A confirmatory analysis also was conducted to compare individuals with
elevated mean inflammatory markers throughout the study to the rest of
the population. Because baseline measurements were unavailable, individuals
were classified in the elevated marker group if their mean CRP, IL-6, or
WBC levels across all samples were within the top 25th percentile
of the study population. In absolute terms, these categories were
defined as having a mean CRP concentration > 5.7 mg/L, a mean IL-6 concentration > 4.6 pg/mL, or a mean WBC concentration > 7.7 × 10<sup>9</sup>/L. Indicators were created for each inflammatory outcome individually. Effect
modification by statin therapy also was examined.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Inflammatory markers</title><p>In total, 25 group trips were conducted over the duration of the study, with
most (35 of 44) subjects participating in each of their four trips. Of
the 158 completed person-trips, we obtained 133 samples with sufficient
volume for complete laboratory analysis. Samples were predominantly
missing because of the lack of a phlebotomist or insufficient blood
volume collected.</p><p>Overall summary statistics for the CRP, IL-6, and WBC levels of our samples
are presented by subject characteristics in <xref ref-type="table" rid="t1-ehp0114-000992">Table 1</xref>. Our 44 participants were predominantly white females with a median age
of 80 years. In general, higher median levels of inflammation across
our repeated samples were observed among individuals with diabetes, obesity, and/or
hypertension than those without these conditions, although
few of these differences were significant at the 95% confidence
level. Similarly, median levels of inflammatory markers were slightly
lower among those on statin therapy but not statistically different
than those who were not.</p></sec><sec><title>Exposure parameters</title><p><xref ref-type="table" rid="t2-ehp0114-000992">Table 2</xref> summarizes the mean air pollution, pollen, mold, and apparent temperature
levels for the day preceding each blood draw. Longer moving averages
exhibited similar means but smaller ranges and standard deviations. Mean
microenvironmental concentrations of PM<sub>2.5</sub> were generally lower than ambient levels, although the range of values
was somewhat comparable. Microenvironmental apparent temperature was
more moderate and less variable than the apparent temperature outdoors. All
pollutants were measured at levels well below current U.S. National
Ambient Air Quality Standards (<xref rid="b38-ehp0114-000992" ref-type="bibr">U.S. EPA 2006</xref>).</p><p>Daily concentrations of ambient PM<sub>2.5</sub> and BC were positively correlated in both spring (<italic>r</italic> = 0.6) and summer (<italic>r</italic> = 0.7), whereas microenvironmental PM<sub>2.5</sub> was only positively correlated with daily ambient PM<sub>2.5</sub> and BC during the summer (<italic>r</italic> = 0.73 and <italic>r</italic> = 0.30, respectively). Higher summertime correlations between
the ambient and microenvironmental measurements corresponded to an increased
frequency of open windows during the summer. Seasonal variation
also was observed for O<sub>3</sub>, a marker for regional pollution, which exhibited positive correlations
with daily ambient PM<sub>2.5</sub> during the summer (<italic>r</italic> = 0.2) but inverse correlations during the spring (<italic>r</italic> = −0.7).</p></sec><sec><title>Ambient PM<sub>2.5</sub> and inflammation</title><p>WBC counts were positively associated with ambient PM<sub>2.5</sub> across the whole population. These associations increased with longer
moving averages and reached statistical significance with the 7-day mean (<xref ref-type="fig" rid="f1-ehp0114-000992">Figure 1</xref>), where an IQR increase in PM<sub>2.5</sub> of 5.4 μg/m<sup>3</sup> was associated with a 5.5% increase in WBC counts (95% CI, 0.10–11). Associations with WBC counts remained significantly
elevated through the 14-day mean (data not shown) but declined with
longer moving averages. Ambient PM<sub>2.5</sub> also was associated with CRP across the entire population with positive
but nonsignificant relationships. These relationships were observed
for all moving averages longer than 1 day and peaked with the 5-day mean. For
IL-6, nonsignificant negative associations with ambient PM<sub>2.5</sub> were observed among the population as a whole. Analyses of pollution lagged
by 1–7 days were consistent with these findings. Similarly, associations
with microenvironmental PM<sub>2.5</sub> averaged over the previous 1 and 2 days demonstrated similar, although
often larger, associations with each of the outcomes. (Longer averaging
times for microenvironmental concentrations could not be examined, given
that they were only measured during the 2 days before the blood
draws.)</p></sec><sec><title>Effect modification by conditions linked to inflammation</title><p>Evidence of effect modification was most frequently observed with CRP. Associations
between PM<sub>2.5</sub> and CRP were consistently, and often significantly, elevated among the 8 individuals
with diabetes (26 repeated samples), 14 individuals with
obesity (41 repeated samples), and 4 individuals with concurrent diabetes, obesity, and
hypertension (14 repeated samples) (<xref ref-type="fig" rid="f2-ehp0114-000992">Figure 2</xref>). For example, an IQR (6.1 μg/m<sup>3</sup>) increase in the 5-day mean PM<sub>2.5</sub> was associated with a 48% increase (95% CI, 5.3–109) in
CRP for persons with obesity, a 74% increase (95% CI, 18–158) for persons with diabetes, and an 81% increase (95% CI, 21–172) in CRP for persons with diabetes, obesity, and
hypertension compared with a 12% increase (95% CI, –25 to 67) for individuals without any of these
conditions. Individuals with diabetes and those with concurrent diabetes, obesity, and
hypertension also demonstrated larger associations
between PM<sub>2.5</sub> and IL-6 (<xref ref-type="fig" rid="f3-ehp0114-000992">Figure 3</xref>). No evidence of effect modification was observed among the 36 individuals
with hypertension, nor did any of the conditions evaluated significantly
modify WBC counts (data not shown). Individuals within the top
quartile of mean circulating inflammatory markers also showed larger
increases in repeated measures of CRP, IL-6, and WBCs with increases in
ambient PM<sub>2.5</sub> (<xref ref-type="fig" rid="f4-ehp0114-000992">Figure 4</xref>). Identical results were produced if the median inflammatory level was
used to define individuals with elevated inflammatory levels. Conversely, the 10 individuals
on anti-inflammatory statin therapy demonstrated
nonsignificant reductions in the association between ambient PM<sub>2.5</sub> and CRP compared with those not on statin therapy.</p></sec><sec><title>Other ambient pollution metrics and inflammation</title><p>Individuals with concurrent diabetes, obesity, and hypertension exhibited
large positive associations between ambient BC, NO<sub>2</sub>, and O<sub>3</sub> and CRP and IL-6, as shown for the 5-day moving average concentrations
in <xref ref-type="table" rid="t3-ehp0114-000992">Table 3</xref>. These associations exhibited similar patterns to those for ambient PM<sub>2.5</sub> at all moving averages. WBC counts generally increased with IQR changes
in ambient BC, CO, and NO<sub>2</sub>, whereas inverse associations were observed between WBC counts and O<sub>3</sub>. Inconsistent findings were observed with SO<sub>2</sub>.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>In this investigation, we found evidence of positive associations between
air pollution and indicators of systemic inflammation (e.g., WBCs, CRP, and
IL-6) in older adults. These findings support the hypothesis
that systemic inflammation is a pathway through which airborne PM leads
to short-term increases in cardiac risk. We also found that the associations
with CRP and IL-6 were strongest and most consistent for the 8 individuals
with diabetes, 14 individuals with obesity, and 4 individuals
with concurrent diabetes, obesity, and hypertension, suggesting
that individuals with conditions often associated with both chronic inflammation
and increased cardiac risk (<xref rid="b18-ehp0114-000992" ref-type="bibr">Libby 2002</xref>) may be more vulnerable to the short-term proinflammatory effects of air
pollution. This hypothesis is supported by the finding that individuals
with the highest mean or median levels of inflammatory markers also
had larger associations between air pollution and CRP, IL-6, and WBC
counts.</p><p>For all three outcomes investigated, associations were strongest with PM<sub>2.5</sub>. Consistent associations were also generally seen for ambient BC and NO<sub>2</sub>, indicating that motor vehicles may be an important source for PM-mediated
inflammation. Although inconsistent associations were observed between
ambient CO (another marker of traffic pollution) and inflammation, this
may be due to measurement error because 95% of the hourly
CO concentrations were lower than the 1 ppm sensitivity of the reference
method (<xref rid="b7-ehp0114-000992" ref-type="bibr">Cogan and Lobert 1998</xref>). O<sub>3</sub> and/or regional pollution may also be partly responsible for the observed
relationships because positive associations were found between ambient
O<sub>3</sub> and CRP and IL-6. Because these relationships were generally not sensitive
to stratification by season, confounding by PM<sub>2.5</sub> is not likely. Associations between O<sub>3</sub> and WBC counts, on the other hand, did change sign by season, corresponding
to the seasonal change in the correlation between O<sub>3</sub> and PM<sub>2.5</sub>. This suggests that there may be confounding between PM<sub>2.5</sub> and O<sub>3</sub> with respect to WBCs.</p><p>Because our reported associations were predominantly for ambient concentrations
of air pollution, there may be some level of measurement error
that is inherent to this investigation. We do not believe that this
error is likely to be substantial, however, because moderately strong
correlations were observed between daily concentrations of PM<sub>2.5</sub> measured at ambient and microenvironmental monitors, and these correlations
should increase with longer averaging periods. Strong correlations
across space also have been previously reported in this region for
PM<sub>2.5</sub> from both sulfate and motor vehicles (<xref rid="b16-ehp0114-000992" ref-type="bibr">Kim et al. 2005</xref>). In addition, our results are supported by the finding of similar associations
with ambient concentrations and concentrations of PM<sub>2.5</sub> measured in participants’ microenvironments over the time periods
when both data were available.</p><p>Although our findings suggest that traffic may be an important source type
for the inflammatory effects of air pollution, it is unlikely that
our bus trips are responsible for the observed associations because the
critical averaging periods were on the order of days rather than hours. Because
bus periods lasted only 2 hr, the contribution of the bus
to exposures averaged over several days was small compared with that
of ambient pollution. In fact, when our analysis was limited to exposures
that occurred only on the bus, the effect estimates were similar in
direction to our main analysis but were smaller in magnitude, likely
due to greater measurement error. Nevertheless, we cannot exclude the
possibility that a specific inflammatory effect of the bus might have
occurred in addition to the overall cumulative effects of ambient pollution. Such
an effect might have been seen if we had measured markers
of inflammation immediately before and after each trip, but we were unable
to collect samples for reasons of feasibility and acceptability
to our participants. We do not believe this to be a critical flaw, however, because
we have less reason to suspect that pollution would have
near-immediate associations with our inflammatory markers, given that
other investigations of outcomes related to inflammation demonstrated
associations on the order of days to a week (<xref rid="b23-ehp0114-000992" ref-type="bibr">O’Neill et al. 2005</xref>; <xref rid="b28-ehp0114-000992" ref-type="bibr">Peters et al. 2001b</xref>; <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. 1999</xref>; <xref rid="b40-ehp0114-000992" ref-type="bibr">Zanobetti et al. 2004</xref>).</p><p>In addition to the timing of our associations, our results are generally
consistent with other investigations with respect to directionality
and magnitude of the associations for each of our inflammatory markers. For
example, several past investigations have illustrated positive associations
between ambient PM and CRP in the blood of older adults (<xref rid="b28-ehp0114-000992" ref-type="bibr">Peters et al. 2001b</xref>; <xref rid="b29-ehp0114-000992" ref-type="bibr">Pope et al. 2004</xref>; <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. 1999</xref>). <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. (1999)</xref> reported a 9.5% increase in CRP per 10 μg/m<sup>3</sup> in the 3-day mean ambient PM<sub>10</sub>. This was comparable with our findings of an 11% increase per 10μg/m<sup>3</sup> in the 5-day mean PM<sub>2.5</sub> for individuals without concurrent diabetes, obesity, and hypertension. Similarly, the
timing of the association between PM and CRP for our
study was consistent with past work as the Monitoring of Trends and Determinants
in Cardiovascular Disease (MONICA) study also reported its
maximum effect with the 5-day mean (<xref rid="b28-ehp0114-000992" ref-type="bibr">Peters et al. 2001b</xref>). Interestingly, our findings among seniors with low susceptibility also
were similar to associations reported for a cohort of young policemen
who exhibited a 21–32% change in CRP per 10 μg/m<sup>3</sup> increase in mean PM<sub>2.5</sub> over their 9-hr shift (<xref rid="b32-ehp0114-000992" ref-type="bibr">Riediker et al. 2004</xref>).</p><p>Only one study of 30 young Singaporean national guardsmen has reported
statistically significant associations between air pollution and IL-6 in
blood (<xref rid="b39-ehp0114-000992" ref-type="bibr">van Eeden et al. 2001</xref>). Other investigations have generally reported null associations between
air pollution and IL-6, in agreement with our study population as a
whole (<xref rid="b13-ehp0114-000992" ref-type="bibr">Ghio et al. 2000</xref>; <xref rid="b21-ehp0114-000992" ref-type="bibr">Nightingale et al. 2000</xref>). Despite this general lack of findings, elevated IL-6 has been found
in human sputum (<xref rid="b22-ehp0114-000992" ref-type="bibr">Nordenhall et al. 2000</xref>) and alveolar macrophages (<xref rid="b2-ehp0114-000992" ref-type="bibr">Becker et al. 1996</xref>; <xref rid="b39-ehp0114-000992" ref-type="bibr">van Eeden et al. 2001</xref>) after exposures to PM. In fact, PM from St. Louis was found to induce
IL-6 from human alveolar macrophages with greater potency than diesel, silicon
dioxide, and latex particles (<xref rid="b2-ehp0114-000992" ref-type="bibr">Becker et al. 1996</xref>). Our findings also are supported by the fact that our associations with
CRP are generally consistent with, and may lag those of, its precursor (IL-6) among
individuals with diabetes, obesity, and hypertension.</p><p>Inconclusive evidence links short-term changes in WBC counts to air pollution
exposures. Two investigations that found significant results for
WBC counts had opposite findings (<xref rid="b12-ehp0114-000992" ref-type="bibr">Ghio et al. 2003</xref>; <xref rid="b33-ehp0114-000992" ref-type="bibr">Schwartz 2001</xref>), and other investigations have reported null associations (<xref rid="b15-ehp0114-000992" ref-type="bibr">Holgate et al. 2003</xref>; <xref rid="b29-ehp0114-000992" ref-type="bibr">Pope et al. 2004</xref>; <xref rid="b35-ehp0114-000992" ref-type="bibr">Seaton et al. 1999</xref>). Because all of these investigations examined associations with pollution
for durations < 3 days, it is possible that associations would
become more consistent with longer averaging periods. Alternatively, as
a composite of various cell types, WBC counts may not perform well
as an indicator of the inflammatory effects of air pollution. Finally, it
is possible that the response of WBCs to PM may be best measured locally
in the lung rather than systemically.</p><p>Although our main results are consistent with other investigations, this
study is novel in that it suggests that individuals with conditions
associated with chronic inflammation may have an increased short-term
inflammatory response to air pollution. One important limitation of this
investigation, however, is the small number of individuals studied. In
our most susceptible group, only 4 individuals with 14 measurements
were classified as having concurrent diabetes, obesity, and hypertension. Despite
these low numbers, our findings withstood several sensitivity
checks. For example, our diabetes and obesity findings for 8 and 14 individuals, respectively, were robust to the exclusion of individuals
with concurrent conditions. Similarly, no one individual dominated
our results for persons with concurrent diabetes, obesity, and hypertension, and
individuals without any of these three conditions demonstrated
lower associations with air pollution. Our overall findings also
were robust to various modeling strategies, including the use of different
confounders and the use of fixed effects for each subject. In addition, confirmatory
analyses supported the suggestion of susceptibility
by inflammatory status, because individuals with higher levels of mean
inflammatory markers demonstrated elevated associations with pollution. Similarly, we
found a suggestion of reduced associations between
air pollution and CRP with use of anti-inflammatory statins. Despite
these assurances, future research is still needed to confirm our findings
of effect modification.</p><p>To our knowledge, few investigations have attempted to investigate effect
modification of short-term associations between air pollution and inflammation
by chronic inflammation directly. One study, conducted in
England, evaluated the influence of air pollution on the odds of myocardial
infarction or stroke based on baseline levels of fibrinogen. Their
findings implied that individuals with elevated inflammation were marginally
more susceptible than normal individuals on high-pollution days (<xref rid="b30-ehp0114-000992" ref-type="bibr">Prescott et al. 2000</xref>). Another study demonstrated that asthmatic children who were not on anti-inflammatory
medications exhibited stronger associations between respiratory
symptoms and PM<sub>10</sub> than did those on anti-inflammatory medications (<xref rid="b10-ehp0114-000992" ref-type="bibr">Delfino et al. 2002</xref>).</p><p>Although our investigation is suggestive that chronic inflammation leads
to an enhanced vulnerability to the short-term inflammatory effects
of ambient PM, other interconnected aspects of diabetes, obesity, and
hypertension may also contribute to the susceptibility of individuals
with these disease states. Possible alternative mechanisms include enhanced
insulin resistance, hyperglycemia, oxidative stress, and endothelial
dysfunction (<xref rid="b5-ehp0114-000992" ref-type="bibr">Brunner et al. 2005</xref>; <xref rid="b18-ehp0114-000992" ref-type="bibr">Libby 2002</xref>; <xref rid="b23-ehp0114-000992" ref-type="bibr">O’Neill et al. 2005</xref>). These factors may play a role in this study given that some but not
all of our diabetic, obese, and hypertensive participants had measurable
manifestations of a chronic inflammatory state.</p><p>Overall, the findings of this investigation may have important implications
for the biologic mechanism of air pollution and possibly its clinical
relevance, because inflammation plays an important role in atherosclerosis
and cardiovascular disease (<xref rid="b18-ehp0114-000992" ref-type="bibr">Libby 2002</xref>). Chronic levels of both CRP and IL-6 have been identified as important
risk factors for adverse cardiovascular outcomes (<xref rid="b8-ehp0114-000992" ref-type="bibr">Danesh et al. 2004</xref>; <xref rid="b19-ehp0114-000992" ref-type="bibr">Luc et al. 2003</xref>; <xref rid="b31-ehp0114-000992" ref-type="bibr">Ridker et al. 2000</xref>). Although short-term changes in inflammation are less well understood
and may differ from chronic changes, our findings still offer a potential
explanation as to why individuals with diabetes, coronary artery
disease, and past myocardial infarctions have elevated cardiovascular
risk after acute exposures to air pollution (<xref rid="b1-ehp0114-000992" ref-type="bibr">Bateson and Schwartz 2004</xref>; <xref rid="b14-ehp0114-000992" ref-type="bibr">Goldberg et al. 2001</xref>; <xref rid="b18-ehp0114-000992" ref-type="bibr">Libby 2002</xref>; <xref rid="b23-ehp0114-000992" ref-type="bibr">O’Neill et al. 2005</xref>; <xref rid="b41-ehp0114-000992" ref-type="bibr">Zanobetti and Schwartz 2002</xref>). It also raises the possibility that there is an interaction between
the short- and long-term effects of air pollution, given that previous
investigations have linked elevated exposures to PM to the development
of chronic inflammatory conditions such as asthma and atherosclerosis (<xref rid="b9-ehp0114-000992" ref-type="bibr">Delfino 2002</xref>; <xref rid="b36-ehp0114-000992" ref-type="bibr">Suwa et al. 2002</xref>).</p><p>In summary, our data suggest that increases in air pollution may be associated
with increases in systemic inflammation in older adults. Associations
between pollution and short-term increases in inflammatory markers
were the strongest for individuals with diabetes; obesity; and concurrent
diabetes, obesity, and hypertension. Elevated associations were
also found among persons with higher mean levels of inflammatory markers. Such
findings support the hypothesis that the cardiovascular effects
of air pollution are partially mediated via inflammation and suggest
that individuals with existing inflammation and existing cardiac
risk factors may be especially susceptible to the inflammatory effects
of air pollution.</p></sec><sec><title>C<sc>orrection</sc></title><p>In <xref ref-type="table" rid="t1-ehp0114-000992">Table 1</xref> of the manuscript published online, the ranges of CRP for both “Yes” and “No” under “Statin therapy” were
incorrect. These values have been corrected here.</p></sec>
|
The CHARGE Study: An Epidemiologic Investigation of Genetic and Environmental
Factors Contributing to Autism
|
<p>Causes and contributing factors for autism are poorly understood. Evidence
suggests that prevalence is rising, but the extent to which diagnostic
changes and improvements in ascertainment contribute to this increase
is unclear. Both genetic and environmental factors are likely to
contribute etiologically. Evidence from twin, family, and genetic studies
supports a role for an inherited predisposition to the development
of autism. Nonetheless, clinical, neuroanatomic, neurophysiologic, and
epidemiologic studies suggest that gene penetrance and expression may
be influenced, in some cases strongly, by the prenatal and early postnatal
environmental milieu. Sporadic studies link autism to xenobiotic
chemicals and/or viruses, but few methodologically rigorous investigations
have been undertaken. In light of major gaps in understanding of
autism, a large case–control investigation of underlying environmental
and genetic causes for autism and triggers of regression has
been launched. The CHARGE (Childhood Autism Risks from Genetics and
Environment) study will address a wide spectrum of chemical and biologic
exposures, susceptibility factors, and their interactions. Phenotypic
variation among children with autism will be explored, as will similarities
and differences with developmental delay. The CHARGE study infrastructure
includes detailed developmental assessments, medical information, questionnaire
data, and biologic specimens. The CHARGE study
is linked to University of California–Davis Center for Children’s
Environmental Health laboratories in immunology, xenobiotic
measurement, cell signaling, genomics, and proteomics. The goals, study
design, and data collection protocols are described, as well as preliminary
demographic data on study participants and on diagnoses of those
recruited through the California Department of Developmental Services
Regional Center System.</p>
|
<contrib contrib-type="author"><name><surname>Hertz-Picciotto</surname><given-names>Irva</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001119">1</xref><xref ref-type="aff" rid="af2-ehp0114-001119">2</xref></contrib><contrib contrib-type="author"><name><surname>Croen</surname><given-names>Lisa A.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001119">3</xref></contrib><contrib contrib-type="author"><name><surname>Hansen</surname><given-names>Robin</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001119">2</xref><xref ref-type="aff" rid="af4-ehp0114-001119">4</xref></contrib><contrib contrib-type="author"><name><surname>Jones</surname><given-names>Carrie R.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001119">1</xref><xref ref-type="aff" rid="af2-ehp0114-001119">2</xref></contrib><contrib contrib-type="author"><name><surname>van de Water</surname><given-names>Judy</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001119">2</xref><xref ref-type="aff" rid="af5-ehp0114-001119">5</xref></contrib><contrib contrib-type="author"><name><surname>Pessah</surname><given-names>Isaac N.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001119">2</xref><xref ref-type="aff" rid="af6-ehp0114-001119">6</xref></contrib>
|
Environmental Health Perspectives
|
<p>Autism is a serious neurodevelopmental disorder characterized by impairments
in social interaction, abnormalities in verbal and nonverbal communication, and
restricted, stereotyped interests and behaviors (<xref rid="b2-ehp0114-001119" ref-type="bibr">American Psychiatric Association 1994</xref>). Although a large proportion of individuals with autism manifest abnormal
development from birth, a subset of at least 20–30% experience
a regression with onset between 18 and 24 months of age
after a period of apparently normal development (<xref rid="b35-ehp0114-001119" ref-type="bibr">Lainhart et al. 2002</xref>). Autistic disorder is the most severe form of autism spectrum disorders (ASDs), which
include Asperger’s syndrome and pervasive developmental
disorders (PDDs) not otherwise specified. Approximately 70% of
individuals with autistic disorder have some degree of mental
retardation, and about half are nonverbal or have very impaired speech. Seizures
are present by adolescence in about 30% of children
with ASD, and between 5 and 10% of autism cases occur in
association with other serious medical conditions such as fragile X, tuberous
sclerosis, and Angelman’s syndrome (<xref rid="b26-ehp0114-001119" ref-type="bibr">Fombonne 2003</xref>). Gastrointestinal problems and sleep disturbances are also thought to
be common comorbidities; however, population-based prevalence estimates
for these conditions are currently lacking. Males are four times as
likely as females to have autism, but this ratio approaches one among
individuals with severe cognitive impairment (<xref rid="b27-ehp0114-001119" ref-type="bibr">Gillberg and Wing 1999</xref>). Most individuals with autism cannot live independently as adults (<xref rid="b50-ehp0114-001119" ref-type="bibr">Rapin 1999</xref>). Over the past 20 years, the prevalence of autism has reportedly risen, with
much public debate surrounding the reasons for this increase. Early
reports estimated prevalence at 4–5 per 10,000 births (<xref rid="b25-ehp0114-001119" ref-type="bibr">Fombonne 1999</xref>). Data published in the last few years suggest that autistic disorder
occurs in at least 1–2 per 1,000 births, and the prevalence of
the broader autism spectrum may be as high as 4–6 per 1,000 (<xref rid="b13-ehp0114-001119" ref-type="bibr">Chakrabarti and Fombonne 2005</xref>; <xref rid="b73-ehp0114-001119" ref-type="bibr">Yeargin-Allsopp et al. 2003</xref>).</p><p>The causes and contributing factors for autism are poorly understood. The
number of children with a diagnosis of autism as determined by the
California Department of Developmental Services (DDS) has been rising
continuously for over a decade (<xref rid="b9-ehp0114-001119" ref-type="bibr">California DDS 2003</xref>). Although diagnostic changes and improvements in detection probably contribute
to this increase (<xref rid="b13-ehp0114-001119" ref-type="bibr">Chakrabarti and Fombonne 2005</xref>; <xref rid="b19-ehp0114-001119" ref-type="bibr">Croen et al. 2002</xref>), a true rise in incidence may also be occurring (<xref rid="b7-ehp0114-001119" ref-type="bibr">Blaxill et al. 2003</xref>). Evidence for genetic causes is strong, yet concordance in monozygotic
twins suggests that a minimum of 40% of autism cases are likely
to have an environmental cause. No single gene has yet been specifically
linked to autism with replicability, but the disorder is believed
to be polygenic. A few specific environmental factors are associated
with autistic behaviors—prenatal exposures to thalidomide (<xref rid="b55-ehp0114-001119" ref-type="bibr">Rodier and Hyman 1998</xref>), valproic acid (<xref rid="b16-ehp0114-001119" ref-type="bibr">Christianson et al. 1994</xref>), or rubella (<xref rid="b15-ehp0114-001119" ref-type="bibr">Chess et al. 1978</xref>)—but these are likely to play a negligible role, if any, in incident
cases in Western countries over the last decade or so.</p><p>Mechanisms of pathogenesis have yet to be delineated. Contrary to early
beliefs that autism resulted from bad parent–child interactions (<xref rid="b6-ehp0114-001119" ref-type="bibr">Bettelheim 1967</xref>), it is now widely accepted that aberrant brain development underlies
autism pathogenesis (<xref rid="b5-ehp0114-001119" ref-type="bibr">Bauman and Kemper 2003</xref>; <xref rid="b18-ehp0114-001119" ref-type="bibr">Courchesne et al. 1988</xref>; <xref rid="b48-ehp0114-001119" ref-type="bibr">Piven et al. 1990</xref>; <xref rid="b56-ehp0114-001119" ref-type="bibr">Rodier et al. 1997</xref>). Autopsy studies demonstrate structural changes in the brain, and imaging
and electrophysiology investigations reveal neurophysiologic differences
in information processing between children with autism and those
with typical development (<xref rid="b21-ehp0114-001119" ref-type="bibr">Dawson et al. 2002</xref>; <xref rid="b42-ehp0114-001119" ref-type="bibr">Maziade et al. 2000</xref>; <xref rid="b43-ehp0114-001119" ref-type="bibr">McPartland et al. 2004</xref>; <xref rid="b51-ehp0114-001119" ref-type="bibr">Rapin and Dunn 2003</xref>; <xref rid="b57-ehp0114-001119" ref-type="bibr">Rosenhall et al. 2003</xref>). Neuroimmunomodulatory factors may also play a role (<xref rid="b62-ehp0114-001119" ref-type="bibr">Silva et al. 2004</xref>; <xref rid="b70-ehp0114-001119" ref-type="bibr">Vargas et al. 2005</xref>). Cytokine profiles, lymphocyte activation, and other immunologic parameters
differ between individuals with and without autism (<xref rid="b3-ehp0114-001119" ref-type="bibr">Ashwood and Van de Water 2004a</xref>, <xref rid="b4-ehp0114-001119" ref-type="bibr">2004b</xref>; <xref rid="b20-ehp0114-001119" ref-type="bibr">Croonenberghs et al. 2002</xref>). Distributions of neuropeptides and neurotrophins at birth appeared to
be altered among children who later developed autism (<xref rid="b45-ehp0114-001119" ref-type="bibr">Nelson et al. 2001</xref>).</p><p>Results from twin and family studies suggest a strong genetic contribution
to the etiology of autism. Beginning with the classic work by <xref rid="b23-ehp0114-001119" ref-type="bibr">Folstein and Rutter (1977)</xref>, data from three population-based twin studies have demonstrated a higher
concordance rate among monozygotic compared with dizygotic twins (<xref rid="b17-ehp0114-001119" ref-type="bibr">Cook 1998</xref>). Strong familial aggregation of autism has also been demonstrated. The
sibling recurrence risk (i.e., the probability of developing autism
given a person’s sibling is autistic) has been estimated at 2–14% (<xref rid="b33-ehp0114-001119" ref-type="bibr">Jorde et al. 1990</xref>; <xref rid="b53-ehp0114-001119" ref-type="bibr">Ritvo et al. 1989</xref>; <xref rid="b64-ehp0114-001119" ref-type="bibr">Smalley et al. 1988</xref>), a 10- to 20-fold increase over the general population prevalence. A
family history of social deficits, language abnormalities, and psychiatric
disorders has also been observed in case–control and clinic-based
studies (<xref rid="b24-ehp0114-001119" ref-type="bibr">Folstein and Rutter 1988</xref>; <xref rid="b49-ehp0114-001119" ref-type="bibr">Piven and Palmer 1999</xref>).</p><p>Autism co-occurs with several known genetic disorders, such as tuberous
sclerosis (<xref rid="b63-ehp0114-001119" ref-type="bibr">Smalley 1998</xref>), Angelman syndrome (<xref rid="b66-ehp0114-001119" ref-type="bibr">Steffenburg et al. 1996</xref>), phenylketonuria, Joubert syndrome (<xref rid="b46-ehp0114-001119" ref-type="bibr">Ozonoff et al. 1999</xref>), and Möbius syndrome (<xref rid="b32-ehp0114-001119" ref-type="bibr">Johansson et al. 2001</xref>), and chromosomal abnormalities such as fragile X syndrome (<xref rid="b52-ehp0114-001119" ref-type="bibr">Reiss and Freund 1990</xref>). More than 90% of autism cases, however, have none of the above
syndromes.</p><p>Linkage, association, and cytogenetic studies have been conducted. Numerous
candidate genes for autism have been suggested based on their functional
role, location within candidate chromosome regions, and positive
associations with the disease (<xref rid="b34-ehp0114-001119" ref-type="bibr">Korvatska et al. 2002</xref>). Replication of findings has been elusive (<xref rid="b71-ehp0114-001119" ref-type="bibr">Wassink et al. 2004</xref>), probably because of the polygenic etiology, heterogeneity of the phenotype, and, assuming
a role for gene–environment interaction, variation
in exposure distributions across populations. An epigenetic
mechanism related to Rett syndrome is also plausible (<xref rid="b60-ehp0114-001119" ref-type="bibr">Samaco et al. 2005</xref>). Genomewide scans to identify regions marked by differing gene expression
are considered key at this stage. One such scan hints at the possible
genetic basis for the well-established sex ratio of four males to
one female (<xref rid="b67-ehp0114-001119" ref-type="bibr">Stone et al. 2004</xref>). A comparison of tuberous sclerosis patients with and without autism
demonstrated 31 genes for which expression differed (<xref rid="b69-ehp0114-001119" ref-type="bibr">Tang et al. 2004</xref>); because both groups shared the tuberous sclerosis diagnosis, the differentially
expressed genes may be related to autism, although they are
not necessarily causal. It is plausible that a substantial proportion
of autism cases could be due to multiple genes interacting with one
or more environmental factors (<xref rid="b12-ehp0114-001119" ref-type="bibr">Cederlund and Gillberg 2004</xref>; <xref rid="b28-ehp0114-001119" ref-type="bibr">Glasson et al. 2004</xref>).</p><p>Neuroanatomic and epidemiologic investigations support a prenatal or early
postnatal origin. <xref rid="b18-ehp0114-001119" ref-type="bibr">Courchesne et al. (1988)</xref> observed cerebellar abnormalities consistent with abnormalities in cell
migration between the third and fifth month of gestation. Magnetic resonance
imaging studies point to migrational errors that result in disorganized
columns of the cerebral cortex (<xref rid="b11-ehp0114-001119" ref-type="bibr">Casanova et al. 2002</xref>). Anthropometric indicators, such as brain size and growth trajectory (<xref rid="b29-ehp0114-001119" ref-type="bibr">Herbert 2005</xref>), suggest overall cerebral volume to be larger in mid-childhood, with
growth that accelerates early and then decelerates, although this phenotype
may apply to only a subset of cases. Neuroimaging studies indicate
involvement of specific brain regions, including the amygdala, hippocampus, and
corpus callosum (<xref rid="b8-ehp0114-001119" ref-type="bibr">Brambilla et al. 2003</xref>; <xref rid="b61-ehp0114-001119" ref-type="bibr">Schumann et al. 2004</xref>).</p><p>Studies of environmental factors also relate to the prenatal origin of
autism. <xref rid="b15-ehp0114-001119" ref-type="bibr">Chess et al. (1978)</xref> reported that, within a cohort of about 250 children with congenital rubella, 7% were
later diagnosed with autism. A case–control
study using both maternal reports and medical records of illnesses
during pregnancy showed relative risks of 4.1 for influenza and 3.3 for
rubella (<xref rid="b22-ehp0114-001119" ref-type="bibr">Deykin and MacMahon 1979</xref>). Daily maternal smoking during early pregnancy was reported to be linked
to autism in a large case–control epidemiology study (odds
ratio = 1.4; 95% confidence interval, 1.1–1.8) (<xref rid="b30-ehp0114-001119" ref-type="bibr">Hultman et al. 2002</xref>), although in our estimation, these analyses may have inappropriately
adjusted for potentially intermediate variables. The link between autism
and early <italic>in utero</italic> exposure to thalidomide places the timing of the insult coincident with
neural tube closure in the fourth to fifth week of gestation (<xref rid="b55-ehp0114-001119" ref-type="bibr">Rodier and Hyman 1998</xref>). Case reports of autism in children gestationally exposed to valproic
acid (<xref rid="b16-ehp0114-001119" ref-type="bibr">Christianson et al. 1994</xref>; <xref rid="b56-ehp0114-001119" ref-type="bibr">Rodier et al. 1997</xref>; <xref rid="b72-ehp0114-001119" ref-type="bibr">Williams et al. 2001</xref>) are concordant with experimental animal studies (<xref rid="b31-ehp0114-001119" ref-type="bibr">Ingram et al. 2000</xref>). A small number of cases of autism after maternal infection with cytomegalovirus (<xref rid="b41-ehp0114-001119" ref-type="bibr">Markowitz 1983</xref>; <xref rid="b68-ehp0114-001119" ref-type="bibr">Stubbs et al. 1984</xref>), measles or mumps (<xref rid="b22-ehp0114-001119" ref-type="bibr">Deykin and MacMahon 1979</xref>), or herpes (<xref rid="b54-ehp0114-001119" ref-type="bibr">Ritvo et al. 1990</xref>) as well as one case each of syphilis and toxoplasmosis (<xref rid="b59-ehp0114-001119" ref-type="bibr">Rutter and Bartak 1971</xref>) have been reported.</p><p>Taken together, the literature suggests a prominent genetic component involving
multiple gene loci, but also a likely contribution from both
chemical and microbial agents. It is likely that further understanding
will require consideration of critical windows during gestation and possibly
early infancy, as well as interactions between genetic or epigenetic
predisposition and environmental factors.</p><sec><title>CHARGE Study Aims</title><p>In light of the enormous gap in our understanding of the causes of both
autism and developmental delay (DD), a large epidemiologic study was
initiated in 2002. The Childhood Autism Risk from Genetics and the Environment (CHARGE) study
is addressing a wide spectrum of environmental
exposures, endogenous susceptibility factors, and the interplay between
these two (<xref rid="b14-ehp0114-001119" ref-type="bibr">CHARGE 2006</xref>).</p><p>To structure the search for etiologic factors, we are beginning with known
neurodevelopmental toxicants and hints from the immunologic evidence. Additionally, physiologic differences that might provide clues about
susceptibility and mechanisms are being examined through characterization
of metabolic, immunologic, and gene expression profiles, as well
as genetic polymorphisms. <xref ref-type="fig" rid="f1-ehp0114-001119">Figure 1</xref> shows five broad classes of exposures of interest: pesticides, metals, persistent
pollutants with known or suspected neurodevelopmental or immunologic
toxicity, medications and other treatments, and infections. Exposures
from both the prenatal and early childhood periods are being
investigated, with data primarily from three sources: <italic>a</italic>) extensive interviews with parents; <italic>b</italic>) laboratory analysis of xenobiotics in blood, urine, and hair specimens; and <italic>c</italic>) prenatal, labor and delivery, neonatal, and pediatric medical records.</p><p>CHARGE study specimens are analyzed for immunologic, cell activation, xenobiotic, lipomic, and
genomic markers in laboratories of the University
of California–Davis (UC Davis) Center for Children’s
Environmental Health (CCEH) (<xref ref-type="table" rid="t1-ehp0114-001119">Table 1</xref>). Metals have been assayed in blood samples from > 300 index children, with
a focus on mercury, lead, arsenic, cadmium, and manganese. Immunologic
profiles are being characterized, including cellular responses
to bacterial antigenic stimulation, general immunoglobulins, and production
of chemokines and cytokines. Already, preliminary results have
demonstrated significant differences between children with autism and
children from the general population in leptin concentrations (Ashwood
P, Kwong C, Hansen R, Hertz-Picciotto I, Croen L, Krakowiak P, et al., unpublished
observations).</p><p>A detailed lipomics screen is being applied to the plasma from the first
few hundred children. Affymetrix GeneChip microarrays (Affymetrix, Santa
Clara, CA) have been generated from an initial sample of children
and analyzed to determine whether a genomic fingerprint for autism can
be identified; results will be replicated on a further set. Brominated
flame retardants are being measured in 80–100 children, and
metabolites of pyrethroid pesticides will be evaluated in urine specimens.</p><p>The CHARGE study also benefits from CCEH hypothesis-driven experimental
research on animal models for autism in mice and non-human primates and <italic>in vitro</italic> investigations of immune and neurogenic cells aimed at uncovering molecular
mechanisms. A common database coordinates the archival, retrieval, and
analysis of samples, and the combination of population-based epidemiology
with state-of-the-art molecular and cellular methods provides
a powerful basis for interdisciplinary collaborative research. With
future funding, the CHARGE study will undertake targeted evaluation of
candidate genes, such as those responsible for regulation of xenobiotic
metabolizing enzymes, cell signaling in both neurons and immune cells, and
immune cell activation.</p><p>Currently, the study is also characterizing phenotypic variation within
the autism case group and relating these phenotypes to the exposures
and physiologic profiles of interest. For example, we have begun to compare
immune function in regressive autism (children who have lost previously
acquired social or language skills) with those with early onset (children
who never acquired those skills). Other phenotypic subtypes
include, for example, high versus low cognitive function and presence
versus absence of gastrointestinal symptoms, macrocephaly, and sleep
disturbances.</p></sec><sec><title>Design and Subject Recruitment</title><p>The CHARGE study appears to be the first large-scale, population-based
epidemiologic investigation focusing primarily on environmental exposures, as
well as their interactions with genes, as underlying causes for
autism. It uses the case–control design, which provides the
most efficient sampling for studies of conditions that are rare or of
multifactorial etiology. A further advantage is the focus on a specific
outcome, which translates into close scrutiny of diagnoses and rigorous
measurement for the most highly suspect risk factors.</p><p>The CHARGE study population is sampled from three strata: children with
autism (full-syndrome autism, not those with a “spectrum disorder”), children
with DD but not autism, and children selected
from the general population without regard for developmental characteristics. All
participating children (currently > 500, with an ultimate
goal of between 1,000 and 2,000) meet the following criteria: <italic>a</italic>) between the ages of 24 and 60 months, <italic>b</italic>) living with at least one biologic parent, <italic>c</italic>) having a parent who speaks English or Spanish, <italic>d</italic>) born in California, and <italic>e</italic>) residing in the catchment areas of a specified list of regional centers (RCs) in
California. No further exclusions are made based on genetics
or family phenotype.</p><p>Children with autism and children with mental retardation or DD are identified
through RCs that contract with the California DDS to determine
eligibility and coordinate services for persons with developmental disabilities. Eligibility
in the DDS/RC system does not depend on citizenship
or financial status. Thus, the system is widely used across socioeconomic
levels and racial/ethnic groups. Referrals are from pediatricians, other
clinical providers, schools, friends, and family members.</p><p>The DDS/RC system is mandated to provide services for individuals with
autism, as well as for those with other PDDs who have mental retardation (IQ < 70) or
are substantially handicapped. One investigation estimated
that 75–80% of the total population of children
with an autism diagnosis in the state were enrolled in the DDS system (<xref rid="b19-ehp0114-001119" ref-type="bibr">Croen et al. 2002</xref>). Among preschoolers, the figure may be lower, with fewer mild cases. Additionally, this
proportion may decline with recent changes to eligibility
requirements that emphasize the extent of disability. Children
with Asperger’s or PDDs not otherwise specified without mental
retardation are not generally eligible for DDS/RC services and therefore
are not actively recruited into the CHARGE study.</p><p>Potential cases of autism for the CHARGE study are defined as those who
are eligible for services based on a DDS/RC diagnosis of autism. Families
with a child who has received a diagnosis but is not in the RC system
are also invited. The second study group, children with DD, is likewise
drawn from those determined eligible for services based on a diagnosis
of mental retardation or DD. Children 0–3 years of age
who are at risk for DD or disability can receive RC services under the
Early Start program and are also eligible to be in the second CHARGE
study group. The DD children must meet the above inclusion criteria but
are not age- or sex-matched to the children with autism.</p><p>Staff of the RCs contact parents of children with autism or DD, provide
them an information packet, and explain how they can participate in the
CHARGE study. For those who are interested, permission is obtained
for the study staff to telephone the families and schedule appointments. The
children then undergo further testing (see below) to confirm their
diagnoses.</p><p>The third group consists of children from the general population identified
from state birth files. Throughout the study, we generate random
samples of children meeting the study eligibility criteria according to
their birth certificate information. This group is frequency-matched
to the age, sex, and broad residential RC catchment area distribution
of the autism cases. Using names and social security numbers in birth
certificate files, study personnel attempt to locate current contact
information and then initiate a recruitment effort.</p><sec sec-type="methods"><title>Data collection protocols</title><p>Participation involves assessments of cognitive and social development, a
medical examination, biologic specimen collection, and completion of
an exposure interview and several self-administered questionnaires. Other
components include maternal and child medical records review and
abstractions. <xref ref-type="table" rid="t2-ehp0114-001119">Table 2</xref> summarizes the protocols, other than specimen and medical record collection.</p><p>CHARGE study children are assessed at the UC Davis Medical Investigations
of Neurodevelopmental Disorders (MIND) Institute; a small percentage
were seen at the UCLA Neuropsychiatric Institute. Standardized clinical
assessments are administered to confirm the child’s diagnostic
group. Autism cases are assessed using diagnostic tools widely accepted
for research: the Autism Diagnostic Interview–Revised (ADI-R) (<xref rid="b36-ehp0114-001119" ref-type="bibr">Le Couteur et al. 2003</xref>; <xref rid="b40-ehp0114-001119" ref-type="bibr">Lord et al. 1994</xref>, <xref rid="b37-ehp0114-001119" ref-type="bibr">1997</xref>) and the Autism Diagnostic Observation Schedules (ADOS) (<xref rid="b38-ehp0114-001119" ref-type="bibr">Lord et al. 2000</xref>, <xref rid="b39-ehp0114-001119" ref-type="bibr">2003</xref>). The ADI-R is a standardized, semistructured 2- to 3-hr interview with
caregivers of individuals with autism or PDDs. It yields summary scores
in the following domains: qualitative impairments in reciprocal social
interaction, communication, and repetitive behaviors and stereotyped
patterns. Published values for interrater reliability are good, with
kappa values ranging between 0.62 and 0.89 (<xref rid="b39-ehp0114-001119" ref-type="bibr">Lord et al. 2003</xref>).</p><p>The ADOS is a semistructured, standardized assessment of children in which
the examiner observes the social interaction, communication, play, and
imaginative use of materials. The ADOS requires approximately 30 min
and includes four possible modules; the examiner chooses the one that
best matches the expressive language level of the individual child
to prevent a relatively low level of language ability from impeding accurate
measurement. Diagnostic algorithms are available for autism or
for broader ASDs/PDDs (<xref rid="b39-ehp0114-001119" ref-type="bibr">Lord et al. 2003</xref>). The ADOS provides measures in the following domains: reciprocal social
interactions, communication, stereotyped behaviors and restricted interests, and
play. All kappa values for interrater reliability exceeded 0.60. All
CHARGE clinical assessment personnel are trained and have
attained research reliability on the ADI-R and the ADOS.</p><p>Cognitive function is measured in all children (those with autism or DD
and the general population controls) using the Mullen Scales of Early
Learning (MSEL) (<xref rid="b44-ehp0114-001119" ref-type="bibr">Mullen 1995</xref>). The MSEL is a standardized developmental test of children 3–60 months
of age. The MSEL consists of five subscales: gross motor, fine
motor, visual reception, expressive language, and receptive language. The
MSEL allows for separate standard verbal and nonverbal summary
scores to be constructed. The five MSEL scales demonstrate satisfactory
internal consistency (0.75–0.83), internal reliability (0.91), test–retest
reliability (0.71–0.96), and inter-rater
reliability (0.91–0.99) (<xref rid="b44-ehp0114-001119" ref-type="bibr">Mullen 1995</xref>).</p><p>Adaptive function is assessed by parental interview using the Vineland
Adaptive Behavior Scales (VABS) (<xref rid="b65-ehp0114-001119" ref-type="bibr">Sparrow et al. 1984</xref>). The VABS is the most widely used instrument for assessment of adaptive
behavior across the lifespan and covers the domains of socialization (interpersonal
relationships, play and leisure time, and coping skills), daily
living skills (personal, domestic, and community skills), motor
skills (gross and fine motor), and communication (receptive, expressive, and
written communication), with developmentally ordered skills
for each area. The scale is norm referenced, and recent supplemental
norms have been published for individuals with autism (<xref rid="b10-ehp0114-001119" ref-type="bibr">Carter et al. 1998</xref>). Psychometric properties of the instrument include excellent internal
consistency (0.90–0.98), test–retest reliability (<italic>r</italic> = 0.78–0.92), and interrater reliability (<italic>r</italic> = 0.87 for young children).</p><p>Before the clinic visit, the parent is mailed the consent form to review
and several self-administered forms to complete, including the Aberrant
Behavior Checklist, a standardized checklist constructed to rate inappropriate
and maladaptive behaviors in developmentally delayed individuals (<xref rid="b1-ehp0114-001119" ref-type="bibr">Aman and Singh 1994</xref>); Multiple Language Questionnaire to determine what languages are used
at home; Child Development Questionnaire (CDQ), consisting of 31 questions
regarding acquisition and loss of language and skills, a subset
of the Early Development Questionnaire (<xref rid="b47-ehp0114-001119" ref-type="bibr">Ozonoff et al. 2005</xref>) to examine loss of developmental skills; and structured questionnaires
about gastrointestinal symptoms and sleep habits of the child (developed <italic>de novo</italic>). Parents are also sent a list of autoimmune diseases with a description
of each, so that they can prepare to respond to questions about family
history of these disorders during the clinic visit. All instruments
and forms are administered in either English or Spanish, depending on
the language in which the parent or child feels most comfortable. The
CHARGE study employs trained bilingual/bicultural staff for every phase
of the study.</p><p>At the clinic, the psychometric assessments are administered, a family
medical history with an emphasis on mental health and autoimmune disorders
is taken, and a family characteristics questionnaire is used to document
developmental and other aspects of the broader phenotype in immediate
family members. Physical and neurologic exams are completed; dysmorphology
and growth or neurologic abnormalities are recorded. Finally, blood
specimens are collected at the end of the clinic visit. The
parent is asked to bring in urine specimens for the child and immediate
family members.</p><p>For families of children recruited from the nonautistic groups, the protocol
is essentially identical, except that the ADI-R and ADOS are not
routinely administered. The Social Communication Questionnaire (SCQ) was
developed from the ADI-R to screen children for evidence of features
of ASDs. If the score on the SCQ is above 15, the ADI-R and ADOS are
administered on a second visit.</p><p>Final autism case status is defined as meeting criteria on the communication, social, and
repetitive behavior domains of the ADI-R and scoring
at or above the total cutoff for autistic disorder on the ADOS module 1 or 2. Analyses
will be conducted for cases meeting criteria for autistic
disorder, as well as for a broader definition of impairment encompassing
ASDs. A similar approach will be used for mental retardation/DD: Children
obtaining an MSEL composite score of < 69 and a VABS
composite score of < 70 will be classified as meeting strict criteria
for DD.</p><p>Separate from the clinic visit, we conduct a telephone interview with the
primary caregiver regarding periconceptional, prenatal, and early childhood
exposures and experiences. The interview of approximately 1 hr 40 min
covers the following areas: demographics; mother’s medical
history; mother’s reproductive and pregnancy history; index
pregnancy, including use of reproductive technology for conception; maternal
illnesses and medications during index pregnancy; metals, diet, and
household product use; child’s illnesses and medications; maternal
lifestyle information; residential history; and occupational
history of the mother and father. An index time period is defined
as 3 months before pregnancy to the end of pregnancy or, if the child
was breast-fed, until weaning. Information on medications, metals, household
products, and the occupational and residential histories focuses
on this index period.</p><p>Blood and urine specimens are collected from the index child, parents, and
siblings. For any family member from whom blood is not obtained, an
attempt is made to collect buccal swabs for DNA extraction. Hair specimens
are collected from the index child and from the mother if her hair
is long enough to potentially contain information about exposures
during the pregnancy or lactation period. If the parent saved locks from
the child’s first haircut, we request a few strands. Additionally, neonatal
blood spots from the index child will be obtained from
the newborn screening specimen archive maintained by the Genetic Disease
Branch of the California Department of Health Services (Richmond, CA).</p><p>Medical records are procured and abstracted for information about procedures, medications
and other treatments, and conditions at birth of the
index child. Obstetric/gynecology/prenatal clinic and mental health
provider records are obtained for the mother. Similarly, labor and delivery, neonatal, pediatric, and specialty clinic medical records are procured. Dental
records are sought for the confirmation of mercury amalgams.</p><p>The study complies with all applicable requirements regarding human subjects
and is approved by the institutional review boards for the State
of California and the University of California. Informed consent is obtained
before collection of any data.</p></sec><sec sec-type="methods|subjects"><title>Preliminary data on participants and their diagnoses</title><p>Full recruitment into the CHARGE study began in late 2003. More than 520 children
and their families have enrolled in the CHARGE study at the
time of this writing. This includes > 360 recruited because of an
RC diagnosis of autism, > 50 with an RC diagnosis of DD (recruitment
began later for this group), and > 120 from the general population. By
the end of the first 5 years of funding, we expect to have a total
of approximately 650–700 children enrolled. Among contacted
families of children with autism, 20% were ineligible, 22% refused, and 58% agreed to participate. Among general population
families with whom we made contact, 22% were ineligible, 41% refused, and 36% agreed to join the study.</p><p>Among children with a diagnosis of autism recruited from RCs, after assessment
by CHARGE study personnel, 64% met criteria on both the
ADOS and ADI-R. Among those 3 or 4 years of age who are California DDS
eligible based on their diagnosis of autism, 64% meet criteria
on both instruments, another 9% meet criteria on the ADOS
alone, and a further 14% on the ADI-R alone, for a total of 87%. Additionally, among the remainder, 6% meet criteria
for ASD based on both examinations (scores at least 7 on ADOS module 1 or
at least 8 on ADOS module 2, and meets cutoff in ADI-R for section
D and either section A or B, and falls within 2 points on the other
section of A or B); another 5% meet criteria for ASD based on
either ADI-R alone or ADOS alone. Fewer than 2% would not be classified
as being on the spectrum.</p><p>Among those recruited through RCs with a diagnosis of DD, the percentage
that showed delay in both adaptive and cognitive domains was 64%, with
another 6% that met the cutoff on at least one of the
tests. Among those who entered the study with a diagnosis of DD, 3% met
criteria for autism and another 8% met criteria for
ASD.</p><p>Another phenotypic distinction we are investigating is early-onset versus
regressive autism, as defined by the language and social regression
questions on the ADI-R and the CDQ. Using a broad definition of regression
that includes loss of previously attained language and/or social
skills (<xref rid="b47-ehp0114-001119" ref-type="bibr">Ozonoff et al. 2005</xref>), close to 50% of the CHARGE children with a confirmed diagnosis
of autism had regression.</p><p>Finally, <xref ref-type="table" rid="t3-ehp0114-001119">Table 3</xref> provides basic demographic information about the CHARGE study sample, based
on data from the birth certificate. This table also provides comparison
information about the pool of births from which we recruit the
general population controls. Compared with this pool, mothers who participate
are older, more highly educated, and more likely to have private
health insurance. Participant mothers of general population controls
are also more likely to have been born in the United States. The children
were more likely to be twins. In further work, the autistic and
DD participants will be compared with their respective pools.</p></sec><sec><title>Community partnership</title><p>A community advisory council (CAC) was formed early in the development
of this project to maximize participation in the research by parents, clinicians, service
providers, advocacy organizations, and RC and DDS
staff. Parental suggestions regarding the collection of specimens and
information from younger siblings of affected children were incorporated
into the study design. The CAC meets regularly to hear updates on study
progress and provide input. CAC members have given critical advice
on data collection instruments, ways to make the clinical protocol as
child-friendly and special-needs–friendly as possible, and strategies
to enhance recruitment.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The CHARGE study is building an infrastructure that will support multiple
investigations of autism and related neurodevelopmental disorders. The
psychometric evaluations and clinical examinations combined with extensive
exposure information and biologic specimens represent rich resources
for research on etiology and phenotypic expression of these disorders
and make possible the comprehensive approach needed to advance
understanding of autism and DD. In our clinical assessments of > 300 children
identified with autism in the California DDS system, we have
confirmed the diagnosis in 87%, suggesting that the large increases
in DDS system clients with autism over the last decade or two
is unlikely to be due to overdiagnosis in younger cohorts.</p><p>Although several large birth cohort studies recently initiated or in progress
will be able to examine factors that predict autism, the number
of cases of autism in the CHARGE study may be comparable with what is
expected in birth cohorts of 100,000 (i.e., we have enrolled > 360 children
with autism and are continuing recruitment). In contrast with
large cohort studies with dispersed populations, we are able to confirm
diagnoses using standardized instruments administered by a small, well-trained
clinical assessment team. Additionally, in cohort studies
attempting to address a wide range of health and developmental outcomes, the
exposures and factors measured will not necessarily have been
chosen for relevance to autism.</p><p>The specimen bank is currently being used by several laboratories that
are part of the UC Davis CCEH. In this first stage, xenobiotic and biochemical
profiles of children with autism are being compared with those
of unaffected children, and comparisons are being made between different
autism phenotypes. As distinguishing features emerge, the second
stage will be to determine whether any differences in biomarkers were
present at birth, using the neonatal blood spots where possible. Data
and specimens will be made available to qualified researchers with targeted, worthwhile
hypotheses not being addressed by CCEH and CHARGE investigators.</p><p>Limitations of this study must be recognized. Much of the information will
be gathered retrospectively. The only biologic specimens prospectively
collected (i.e., before diagnosis) are the newborn blood spots and, for
some children, baby hair locks. Similarly, questionnaires on use
of pesticides and other household products will be retrospective and
hence subject to reporting/recall bias. Thus, the large birth cohort
studies under way or in preparation will complement the CHARGE study by
providing fully prospective data, although they are subject to the limitations
described above. Nevertheless, in the CHARGE study, medical
records will yield prospectively recorded data on treatments, illnesses, and
prescription medications. Other unbiased, relevant sources of
information on xenobiotics include blood measurements that represent cumulative
exposures for persistent compounds and California’s
Pesticide Use Reporting system, which documents commercial pesticide applications
that can be linked to participant residences during critical
time windows.</p></sec><sec sec-type="conclusion"><title>Conclusion</title><p>Although sporadic studies have identified specific environmental factors
that have been associated with autism, no previous effort has attempted
to address the broad spectrum of environmental factors that may, in
combination with genetic susceptibility, affect development and severity
of this condition in the population. The CHARGE study is charting
new territory in the investigation of etiologic factors for autism and
DD. The goal of the CHARGE study is to understand causes of autism and
DD, both genetic and environmental, in order to reduce their incidence
in the future. The design combines a large population-based sample
of children with different patterns of development; standardized diagnostic
assessments of autism, cognitive development, and adaptive behavior
by trained assessors; medical and neurologic examinations; detailed
reviews of medical records; and an extensive set of questionnaires
describing phenotypic characteristics and environmental exposures from
preconception through early childhood. Currently, it is unique in its
emphasis on environmental factors and its tight linkage with state-of-the-art
laboratories of the UC Davis CCEH that enable us to address biologic
markers of xenobiotic exposures, immunologic responses, and gene
expression. Other features include active community involvement, an
ethnically diverse pool of participants, and inclusion of developmentally
delayed children in addition to general population controls. Finally, the
collaboration by CHARGE study investigators with other population-based
autism epidemiologic efforts currently under way, such as the
national Centers for Autism and Developmental Disabilities Research
and Epidemiology (CADDRE) study, will create valuable opportunities for
replication and perhaps data pooling.</p></sec>
|
Intersubject Variability of Risk from Perchlorate in Community Water Supplies
|
<p>This article is a brief review and summary of the estimated incremental
risks (increases in hazard quotient or decreases in thyroid uptake of
iodine) to pregnant women (and hence their fetuses) associated with perchlorate
exposure in community water supplies (CWSs). The analysis draws
on the recent health effects review published in 2005 by the National
Research Council (NRC). We focus on the potential level of risk borne
by the NRC-identified most sensitive subpopulation (pregnant women
and hence their fetuses). Other members of the population should be
at a level of risk below that calculated here, and so protection of the
sensitive subpopulation would protect the general public health. The
analysis examines the intersubject distribution of risks to this sensitive
subpopulation at various potential drinking water concentrations
of perchlorate and also draws on estimates of the national occurrence
of perchlorate in U.S. CWSs to estimate the variability of risks under
defined regulatory scenarios. Results suggest that maximum contaminant
levels (MCLs) of up to 24.5 μg/L should pose little or no incremental
risk to the large majority of individuals in the most sensitive
subpopulations exposed in the United States at current levels of
perchlorate in water. The protectiveness of an MCL of 24.5 μg/L
depends, however, on whether the study subjects in the health effects
data used here may be assumed to have been exposed to background (non-drinking
water) contributions of perchlorate.</p>
|
<contrib contrib-type="author"><name><surname>Crawford-Brown</surname><given-names>Doug</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000975">1</xref></contrib><contrib contrib-type="author"><name><surname>Raucher</surname><given-names>Bob</given-names></name><xref ref-type="aff" rid="af2-ehp0114-000975">2</xref></contrib><contrib contrib-type="author"><name><surname>Harrod</surname><given-names>Megan</given-names></name><xref ref-type="aff" rid="af2-ehp0114-000975">2</xref></contrib>
|
Environmental Health Perspectives
|
<p>Perchlorate is an inorganic compound that has been manufactured and used
as a solid rocket fuel for several decades. Initial detection of perchlorate
in drinking waters was associated with proximity to military
and industrial sites where the compound was produced, stored, and/or used. More
recent data collection efforts suggest perchlorate is more widespread
than initially thought and in some locations may be associated
with sources other than military rocket fuels. In some locations, perchlorate
may be present from commonly used explosive devices (e.g., fireworks, road
blasting materials) and in other locations the compound
may be formed naturally under suitable atmospheric and soil conditions. For
example, some researchers hypothesize that lightning interactions
with desert soils containing certain salt compounds may be responsible
for perchlorate levels detected in western Texas (<xref rid="b2-ehp0114-000975" ref-type="bibr">Dasgupta et al. 2005</xref>). Similar natural forces may explain the presence of perchlorate in the
Atacama Desert region of Chile, and fertilizers mined from the Chilean
desert may contribute to perchlorate found in some areas of the United
States where those products were applied.</p><p>Perchlorate is among a class of goitergens that inhibit the uptake of iodide
by the thyroid and thereby cause goiter and related iodine deficiency
disorders (IDDs), including, in extreme cases, cretinism. IDD is
no longer considered a public health concern in the United States because
the large majority of Americans have ample iodide uptake through
their normal diet to prevent IDD. There is, however, a fraction of pregnant
women, between 10 and 15%, whose urinary excretion rates
are elevated (<xref rid="b4-ehp0114-000975" ref-type="bibr">Hollowell et al. 1998</xref>). If this increased urinary excretion rate is interpreted as indicating
a deficit of iodine uptake (this link is not established in the cited
report), these women are likely to be the sensitive subpopulation for
perchlorate exposures. Iodide intake is sufficient to typically enable
the thyroid to compensate and overcome any adverse effects from goitergen
exposure. It is important to note that the effects of perchlorate
are therefore dependent on the total pool of goitergens to which individuals
are exposed.</p><p>Goitergen exposure in humans is from a variety of routes, including both
water ingestion and consumption of food products found in the diet containing
those with relatively high levels of nitrate (fruits, vegetables, grains, drinking
water, and smoked meats), thiocyanates (broccoli, cabbage, corn, yams, sorghum, and milk), isoflavones (soy, beans, and
peas), bromide (drinking water), and disulfides (onions, garlic, and
peas). Goitergen intake from perchlorate exposure in water must be compared
against this background of exposure to other goitergens, with
risks from perchlorate resulting from the incremental effect of iodine
uptake inhibition above and beyond the inhibition caused by the intake
of other goitergens. Presently, the relative effectiveness of these
different routes of exposure at producing decreases in iodine uptake has
not been assessed, so it is not possible to specify the fraction of
total decrease due solely to perchlorate exposures.</p><p>Overall goitergen exposure would need to be quite high for iodide uptake
to be inhibited to a degree sufficient to elevate IDDs to a matter of
health concern, although again, this level of exposure is not known
at present and may be significantly lower for the sensitive subpopulation. The
National Research Council (NRC) examined the risks posed by perchlorate
ingestion (<xref rid="b5-ehp0114-000975" ref-type="bibr">NRC 2005</xref>) and indicated in their executive summary, “To cause declines
in thyroid hormone production that would have adverse health effects, iodide
uptake would most likely have to be reduced by at least 75% for
months or longer.” The mode of action for perchlorate
exposure and human health risk is summarized here in the <xref ref-type="app" rid="app1-ehp0114-000975">Appendix</xref>, based on the mode of action described in the <xref rid="b5-ehp0114-000975" ref-type="bibr">NRC (2005)</xref> report.</p><p>The NRC expert panel developed an oral reference dose (RfD) of 0.0007 mg
perchlorate per kilogram of body weight per day (mg/kg/day). This oral
RfD is intended to reflect a safe threshold dose at which no risk of
adverse health effect is anticipated for an iodide-deficient pregnant
woman and any developing fetus she might be carrying. As stated by the <xref rid="b5-ehp0114-000975" ref-type="bibr">NRC (2005)</xref>: “The committee concludes that an RfD of 0.0007 mg/kg per day
should protect the health of even the most sensitive populations.” This
RfD is based on observing no significant inhibition of thyroid
uptake of iodide at a perchlorate dose of 0.007 mg/kg/day in human
subjects (<xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. 2002</xref>). A total uncertainty factor of 10 then was applied to ensure protection
of the sensitive subpopulation: iodide-deficient pregnant women (and
their fetuses). Such a sub-population could be exposed to perchlorate
levels up to the RfD of 0.0007 mg/kg/day and not be expected to face
a significant risk of adverse health effect. Because this is the most
sensitive population, this RfD also would be protective of all other
exposed individuals (including infants).</p><p>The need for a larger uncertainty factor was precluded (according to the
NRC committee) by the use of a precursor to adverse effect (iodine uptake
inhibition) in establishing a threshold for exposure, which was
considered by the committee to represent a health-protective assumption
causing the recommended RfD to be based on a no observed effect level (NOEL) rather
than the more commonly used no observed adverse effects
level (NOAEL). A NOAEL is by definition an adverse effect equal to or
higher than a NOEL where the effect used to establish the NOEL is a
precursor to the adverse effect of interest in establishing a NOAEL.</p><p>A possible argument is that a larger uncertainty factor still is warranted
because we do not know the precise level at which a decrease in iodine
uptake becomes adverse, and so it is possible that even small decreases
may be adverse in the sense implied by the NOAEL and the lowest
observed adverse effect level (LOAEL). This would be true especially
in the case of women who already are iodine deficient. The authors of
the present article believe this confuses the concept of an uncertainty
factor as originally developed to argue for RfDs based on effects judged
adverse. The question is not whether a given decrease in iodine uptake
does or does not lead to adverse effects in some percentage of the
population but whether such a decrease in and of itself, absent any
sequellae, is to be taken as an adverse effect. Our position here is
that such a decrease is not adverse in and of itself and so does not warrant
the application of uncertainty factors developed originally to
reason from NOAELs and LOAELs. The NRC committee appears to agree, whether
explicitly or implicitly.</p><p>After the NRC report, the U.S. Environmental Protection Agency (<xref rid="b9-ehp0114-000975" ref-type="bibr">U.S. EPA 2005</xref>) issued a statement accepting the NRC’s RfD and announcing that
it had developed a drinking water equivalent level (DWEL). The DWEL
converts the RfD (represented in units of mg/kg/day) into an associated
concentration in drinking water (in units of micrograms per liter), taking
into account the relative source contribution (RSC) from water
versus other exposure routes. The DWEL was established by the EPA at 24.5 μg/L (<xref rid="b9-ehp0114-000975" ref-type="bibr">U.S. EPA 2005</xref>) and is derived assuming a 70-kg adult consuming 2 L of drinking water
per day. This gives an intake rate (of water) per unit body mass of 0.029 L/kg/day, which
is slightly above the mean value for women of child-bearing
age when both direct and indirect water ingestion are considered (<xref rid="b8-ehp0114-000975" ref-type="bibr">U.S. EPA 2004</xref>, table 6.1.A2). Hence, use of this value may be considered conservative (in
the sense of being health protective) for the sensitive subpopulation.</p><p>The present article places the NRC assessment into the framework of probabilistic
risk assessment. The question addressed here is what the distribution
of risks is in the sensitive subpopulation of pregnant women
in the United States resulting from exposure to perchlorate in water
from community water supplies (CWSs). The term “risk” in
this article has two metrics: a hazard quotient (HQ) and a percentage
reduction in iodide uptake. These risks then are examined using Monte
Carlo analysis to produce intersubject variability distributions under
a variety of scenarios of regulatory interest.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec><title>Exposure assessment</title><p>The occurrence of perchlorate in drinking waters has recently been reported
in a study sponsored by the American Water Works Association’s
Water Industry Technical Action Fund (<xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark 2004</xref>). The study relied principally on data collected under the U.S. EPA unregulated
contaminant monitoring rule (UCMR), supplemented with monitoring
data collected by the Massachusetts Department of Environmental Protection (MDEP), by
the California Department of Health Services (CalDHS), and
in Arizona and Texas (<xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark 2004</xref>). The results (summarized in <xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark 2004</xref>, <xref ref-type="table" rid="t2-ehp0114-000975">table 2</xref>.1) provide estimates of the percentage of CWSs exceeding a variety of
proposed maximum contaminant levels (MCLs) for perchlorate. For the U.S. EPA
sampling, the percentages of CWSs exceeding 2, 4, 6, 10, and 20 μg/L
were 4.1, 2.6, 1.6, 0.9, and 0.2%, respectively. For
the CalDHS sampling, the percentages of CWSs exceeding 2, 4, 6, 10, and 20 μg/L
were 10.5, 5.8, 3.2, 1.5, and 0.3%, respectively. For
the MDEP sampling, the percentages of CWSs exceeding 2, 4, 6, 10, and 20 μg/L were 1.1, 0.8, 0.6, 0.3, and 0.0%, respectively. Data
from Arizona and Texas are not included here because
they did not identify whether a water source was potable or nonpotable
or whether it was part of a water system. Unfortunately, the data
are insufficient at present to develop a fully probabilistic population-weighted
distribution of concentrations in CWSs, and so the present
analysis assumes no correlation between system size (and hence size
of population served) and perchlorate concentration.</p><p>The UCMR used analytic methods with a detection limit of 4 μg/L (micrograms
per liter are essentially the same as parts per billion), and
drew four quarterly samples from each entry point to the distribution
system (EPDS) for every CWS > 10,000 persons served in the United
States. Data also were collected for a sample of 771 smaller systems, but
this sample may be too small to provide a sound basis for statistical
inference. These data suggest a slightly higher concentration
in the smallest water supplies, and so the analysis of <xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark (2004)</xref> may underestimate exposures (by up to 20%) in the small percentage
of the population using these small systems serving fewer than 10,000 people. The
results reported by <xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark (2004)</xref> and used here reflect the UCMR database compiled as of August 2004, when
the database did not yet contain all the data from all quarters for
all EPDSs. Hence, the final UCMR data set may suggest results that differ
slightly from those discussed here.</p><p>The UCMR data reveal detectable amounts (≥ 4 μg/L) in 1.9% of
the samples taken. Because most CWSs have more than one
EPDS, and samples were taken for each EPDS, a higher percentage of CWSs (> 1.9%) were found to have at least one EPDS with detectable
levels of perchlorate. The UCMR data suggest that perchlorate occurs
in detectable amounts in at least one EPDS associated with 5.4% of
CWSs. In systems serving > 10,000 people, perchlorate was
detected in 6.1% of groundwater-based CWSs and in 4.9% of
the surface-water–fed systems.</p><p>Although > 5% of large CWSs in the UCMR database had some detectable
perchlorate in at least one of the EPDS-finished waters, the
levels observed were generally quite low. More than two-thirds (68%) of
the measurable perchlorate concentrations were in the 4–8 ppb
range, and 86% were < 12 μg/L. Only 2.6% of
the detected samples had concentrations > 24 μg/L (<xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark 2004</xref>), which is near the U.S. EPA-designated DWE of 24.5 μg/L (<xref rid="b9-ehp0114-000975" ref-type="bibr">U.S. EPA 2005</xref>). The highest observed level in the UCMR data was 420 μg/L.</p><p>In Massachusetts, samples were analyzed with a more sensitive detection
limit that yielded quantifiable results ≥ 1 μg/L and “trace” observations for levels < 1 μg/L. This
method revealed that 2.4% of treated drinking water samples
contained detectable levels of perchlorate. However, the vast majority
of the Massachusetts detections in treated waters were at or near the 1-μg/L
limit of detection: 66% of detects in treated
drinking water were at trace levels (≤ 1 μg/L), 83% of
detects were ≤ 2 μg/L, and 90% were ≤ 4 μg/L (<xref rid="b1-ehp0114-000975" ref-type="bibr">Brandhuber and Clark 2004</xref>).</p><p>The above data were fit by a lognormal distribution. The resulting distribution
is characterized by a median of 0.03 μg/L and a geometric
standard deviation (GSD) of 13. The assumption here is that the properties
of the distribution identified at the higher levels of exposure (≥ 1 μg/L) continue to apply in water supplies at
concentrations below the detection limit.</p><p>Based on the NRC review, potential for risk arises only if a person from
the sensitive subpopulation ingests perchlorate at an incremental rate (i.e., above
background) that exceeds the identified threshold for
effect. The average daily rate of intake (ADRI) for any individual is
based on how much tap water they consume, the concentration of perchlorate
in their tap water, and their body weight. These three factors vary
across the U.S. population of pregnant women. Using available data, the
distributions for these variables can be included in a Monte Carlo
analysis to develop a combined distribution of ADRI values across this
subpopulation. The distribution of water ingestion rates used here
is based on total CWS consumption values for adults established by the <xref rid="b8-ehp0114-000975" ref-type="bibr">U.S. EPA (2004)</xref>, which provides values associated with given percentiles of the variability
distribution.</p><p>Data on water ingestion for pregnant women were too limited to use reliably
in this analysis, but the existing data suggest that using the data
for U.S. adults does not understate exposures in pregnant women. As
demonstrated in the U.S EPA <italic>Exposure Factors Handbook</italic> (<xref rid="b6-ehp0114-000975" ref-type="bibr">U.S. EPA 1997</xref>), the difference in intake rates of tap water for the general population
of women of child-bearing age and pregnant women is small (mean of 1.16 vs. 1.19 L/day), and
so the former is assumed to approximate the
latter intake rates in this analysis. The distribution of body weight
for 25-year-old women (representing women 18–40 years of age, who
largely make up the child-bearing–age population) is taken
from the U.S. EPA <italic>Exposure Factors Handbook</italic> (<xref rid="b7-ehp0114-000975" ref-type="bibr">U.S. EPA 1999</xref>). The data on water ingestion rate per unit body weight described above
then were fitted by a lognormal distribution, with a best fit showing
a median of 0.0182 L/kg/day and a GSD of 1.8. This distribution is consistent
with the mean value assumed in regulatory calculations.</p><p>The U.S. EPA typically employs an RSC of drinking water, expressed as the
percentage of total contaminant dose that is provided by drinking water, to
estimate total risk from all routes of exposure (i.e., aggregate
risk). These RSCs for drinking water generally are in the range of 20–80%. The relevance of applying an RSC here depends
on how one interprets the human subject perchlorate study conducted by <xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. (2002)</xref> that forms the basis of the risk coefficients. An RSC is appropriate when
the study on which risk coefficients are based includes only exposures
through one route, whereas exposures through other routes will be
present in exposure situations envisioned in regulatory decisions (and
so must be factored in when regulating exposures by the first route). If
one assumes that the individuals in the study by <xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. (2002)</xref> were exposed to the same background levels of perchlorate as the rest
of the U.S. population (there is nothing in their diets or in the study
design to preclude this), then no further RSC adjustment is needed to
reflect total exposures via all routes because the risk coefficient
from the study already reflects the incremental risk from ingestion of
perchlorate in water above and beyond the contributions to perchlorate
exposure via the other routes. Similarly, because the study population
presumably was exposed to the complement of goitergens other than perchlorate, the
study by <xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. (2002)</xref> also reflects the incremental risk from ingestion of the goitergen perchlorate
above and beyond the contributions from these other goitergens. This
is the scenario we employ in our analysis. Unfortunately, adequate
data are not available at present to estimate the RSC for water exposures
reliably.</p></sec><sec><title>Risk characterization</title><p>A standard metric of potential health risk for threshold contaminants like
perchlorate is the HQ. The HQ is equal to the estimated ADRI (in units
of milligrams per kilogram per day) divided by the RfD. An HQ value
of 1.0 thus means that a person is receiving an ADRI equal to the RfD. Any
HQ value ≥ 1.0 indicates that exposure is at or below
the “no risk” threshold (the term “no risk” here
meaning a risk judged to be nonsignificant), and thus no significant
risk of adverse health effect is anticipated. An HQ value > 1.0 indicates
an ADRI above the RfD and suggests that there may be
some nonzero risk of adverse health effect (although, because of the uncertainty
factors in the RfD, which produce a margin of safety, the risk
may be zero even for exposures yielding HQ values < 1.0). In the
present article, we use the value of RfD suggested by the <xref rid="b5-ehp0114-000975" ref-type="bibr">NRC (2005)</xref>: 0.0007 mg/kg/day.</p><p>Another measure of effect used in this analysis is the estimated percent
decrease in iodide uptake by the thyroid (the critical effect used originally
to establish the RfD). This is estimated based on fitting a
dose–response curve to the data from <xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. (2002)</xref>, relating the ADRI to the percent decrease in iodide uptake. The resulting
curve is shown in <xref ref-type="fig" rid="f1-ehp0114-000975">Figure 1</xref>. The best model fit is as follows:</p><disp-formula><graphic xlink:href="ehp0114-000975e1.jpg" position="float" mimetype="image"/></disp-formula><p>where ADRI is in units of mg/kg/day. Note that this model suggests a threshold
at 0.005 mg/kg/day, which is slightly below the NOEL for the study
at 0.007 mg/kg/day. This is because there is a measured decrease
in iodine uptake (1.8%) even at the NOEL, although this decrease
is not statistically significant. A comparison point for risk here
is the <xref rid="b5-ehp0114-000975" ref-type="bibr">NRC (2005)</xref> observation that a 75% decrease in iodide uptake would be required
to initiate a potential health effect, although again, it must be
noted that the percent decrease required in the sensitive subpopulation
currently is unknown and is likely to be less than this value. As before, note
that our assumption here is that the dose–response
data from <xref rid="b3-ehp0114-000975" ref-type="bibr">Greer et al. (2002)</xref> reflect the incremental decrease in iodide uptake per unit incremental
increase in exposure to perchlorate through water alone, above and beyond
the modifying effects of the background perchlorate exposures through
other routes.</p></sec></sec><sec sec-type="results"><title>Results</title><p>The Monte Carlo assessment was conducted for hypothetical MCLs of 1, 2, 5, 6, 10, 20, 24.5, and 50 μg/L (values of 6 and 24.5 were included
to reflect potential limits by the California Environmental Protection
Agency and the U.S. EPA DWEL, respectively). The analysis was
conducted first using the national occurrence distribution to reflect
nationwide conditions. In this analysis the actual distribution of perchlorate
concentrations in CWSs is assumed (median of 0.03 μg/L
and GSD of 13), with systems above the MCL mitigated to exactly the
MCL (the nonexceeding systems remain at their current concentrations). From
this, the distribution of water concentrations in the United States
was established after the MCL is in place, and a value was selected
at random. An intake rate per unit body mass for an individual in the
sampled population (women of child-bearing age) then was selected at
random from the distribution described previously (median of 0.0182 L/kg/day
and a GSD of 1.8). The product of the perchlorate concentration
in water and the intake rate of water per unit body weight then equals
the ADRI for that sampled individual. The sampled ADRI was divided
by the RfD (0.007 mg/kg/day) to produce an estimate of the HQ, then the
ADRI was placed into the model in <xref ref-type="fig" rid="f1-ehp0114-000975">Figure 1</xref> to produce an estimate of percentage reduction in iodine uptake. The Monte
Carlo process was repeated for 10,000 individuals to generate intersubject
variability distributions for these two risk metrics. The value
of 10,000 was based on the goal of providing stability in the tails
of the distribution.</p><p>Then we focused on intersubject variability of doses and risk metrics for
people possibly exposed to water mitigated to exactly a potential MCL
to reflect risk distributions only within those CWSs that currently
have elevated perchlorate concentrations and might therefore be expected
to reduce concentrations down to the MCL. The Monte Carlo process
is the same as described previously (including the focus on the sensitive
subpopulation), with the exception that all individuals are exposed
at the same concentration of perchlorate in water, equal to the MCL. Both
sets of results are described below.</p><sec sec-type="results"><title>National occurrence results</title><p>The HQ results using the national occurrence analysis are summarized in <xref ref-type="table" rid="t1-ehp0114-000975">Table 1</xref>. For example, at the 95th percentile of the sensitive subpopulation, the
HQ value was 0.02 (i.e., dose was 2% of the RfD no risk threshold), even
at an MCL of 50 μg/L. Note from this same table
that the percent decrease in iodine uptake, using the model in <xref ref-type="fig" rid="f1-ehp0114-000975">Figure 1</xref>, is zero for all MCL values and percentiles examined because the ADRI
was below the threshold in the model.</p></sec><sec sec-type="results"><title>Results for systems at the MCL</title><p>In this second set of calculations, all individuals in the sensitive subpopulation
are assumed exposed at the concentration of a potential MCL. In
other words, in this analysis, we examine risks to the highly exposed
portion of the sensitive subpopulation after the potential MCL has
been established and all CWSs are mitigated down to that MCL.</p><p>For the at-the-MCL analysis, some HQ values do exceed 1.0. As shown in <xref ref-type="table" rid="t2-ehp0114-000975">Table 2</xref>, there were no HQ values > 1.0 at MCLs of ≤ 24.5 μg/L
for the percentiles of the cumulative distribution functions examined. In
systems with perchlorate concentrations of 50 μg/L, however, 28.6% of
the sensitive subpopulation had an HQ value exceeding 1.0 (an
HQ value of 1.0 was found at approximately the 71st percentile
of the variability distribution for this population). At the 90th
percentile, the HQ value at 50 μg/L exposure was 1.54, and
at the 95th percentile the HQ value was 1.89. There was, however, no
reduction in iodide uptake estimated from the model at any MCL because
all intake rates were below the threshold for the model in <xref ref-type="fig" rid="f1-ehp0114-000975">Figure 1</xref>.</p></sec><sec><title>Sensitivity analyses</title><p>The above-described analyses and results are based on several assumptions
that can be altered. We conducted several alternate Monte Carlo simulations
to reflect a mix of potential differences in selection of underlying
data or in how those data are interpreted. The goal here was to
determine an upper-bound estimate of the risks, and so more conservative
assumptions were used than was the case in <xref ref-type="table" rid="t1-ehp0114-000975">Tables 1</xref> and <xref ref-type="table" rid="t2-ehp0114-000975">2</xref>. Specifically, in this new analysis, the amount of water consumed was
increased to include total water intake (not just intake from CWSs), as
obtained from the U.S. EPA <italic>Exposure Factors Handbook</italic> (<xref rid="b7-ehp0114-000975" ref-type="bibr">U.S. EPA 1999</xref>). Using the national occurrence data for the concentration of perchlorate
in the drinking water (i.e., assuming the non-CWS concentration was
the same as that in the CWS for an individual), there is no appreciable
difference between the base case results from <xref ref-type="table" rid="t1-ehp0114-000975">Table 1</xref> and the “upper-end” values calculated here. Results in <xref ref-type="table" rid="t1-ehp0114-000975">Table 1</xref> may therefore be assumed to represent upper-end risks when all water consumption, and
not only drinking water, is considered in the exposure
assessment.</p><p>However, for the at-the-MCL analysis results (as shown in <xref ref-type="table" rid="t3-ehp0114-000975">Table 3</xref>, and equivalent to <xref ref-type="table" rid="t2-ehp0114-000975">Table 2</xref>), for those women consuming water with perchlorate at the potential MCLs, there
are some elevated HQ values compared with those in the base
analysis depicted in <xref ref-type="table" rid="t2-ehp0114-000975">Table 2</xref>. In particular, there are now HQ values > 1.0 at the 95th percentile
even at 20 μg/L.</p></sec></sec><sec><title>Discussion and Conclusions</title><p>Perchlorate in drinking water is more widespread than originally anticipated, with
perhaps 2% of sources showing detectable levels ≥ 4 μg/L. Combining the newly emerging risk and occurrence
information, we have modeled the percentage of the sensitive subpopulation (pregnant
women) that may face (or whose infants may face) a risk
of adverse health effects due to perchlorate in U.S. drinking waters. The
results indicate that for any population using a CWS with a perchlorate
concentration of 50 μg/L (i.e., slightly more than twice
the proposed U.S. EPA DWEL of 24.5 μg/L), there would be
an appreciable percentage of pregnant women who face a risk of adverse
effects in themselves or their fetuses because they would have an HQ
value > 1.0. When perchlorate concentrations are 50 μg/L, between 28.6% (if
only ingestion of drinking water is assumed) and 58.1% (if
all water ingestion is assumed, with the non-CWS
being similarly contaminated by perchlorate) of the sensitive subpopulation
might face a dose exceeding the RfD. Values of the HQ > 1.0 at
the 95th percentile of the intersubject variability distribution are
predicted at 20 μg/L perchlorate if ingestion of all water, and
not only drinking water, is included in the exposure assessment. The
results suggest that few women in the sensitive subpopulation would
face a significant perchlorate risk from drinking water at MCLs ≤ 24.5 μg/L
if only drinking water is considered, but that
the equivalent MCL would need to be slightly below 20 μg/L if
all water ingestion were considered.</p><p>We caution the reader on the interpretation of these results. The present
analysis falls within a framework of probabilistic risk assessment
that differs in significant ways from traditional approaches to determining
regulatory limits on exposure. In those traditional approaches, risks
are estimated to maximally exposed individuals within sensitive
subpopulations, and the concentration determined that produces an acceptable
level of risk in those individuals. This level is independent of
any consideration of the fraction of people in that subpopulation. The
question being addressed traditionally is to what extent a proposed
MCL will reduce the risk to an individual in this maximally exposed, sensitive
subpopulation.</p><p>Probabilistic risk assessment as conducted here, however, examines the
intersubject variability distribution of risks in this sub-population
and asks what fraction of people in an exposed population have a risk (HQ
or percentage decrease in iodide uptake) judged to be unacceptable. Such
probabilistic distributions form the basis of cost–risk–benefit
calculations, allowing society to determine how a given
mode of risk reduction (e.g., controls on perchlorate exposures) compares
against other modes of risk reduction. The goal then is to determine
the total burden of disease in a population and to use this estimate
of burden to determine whether the examined mode of risk reduction (here, control
on perchlorate exposures) represents an effective way
to allocate limited societal resources in improving the overall health
of the public. We have not attempted here to draw any conclusions in
that regard, but rather to present the probabilistic information on which
such cost–risk–benefit assessments might be based.</p></sec>
|
Isolation of Antibiotic-Resistant Bacteria from the Air Plume Downwind
of a Swine Confined or Concentrated Animal Feeding Operation
|
<sec><title>Objective</title><p>In this study we evaluated the levels of antibiotic- and multidrug-resistant
bacteria in bioaerosols upwind, within, and downwind at locations 25 m, 50 m, 100 m, and 150 m from a swine confined animal feeding operation.</p></sec><sec><title>Design</title><p>We used Andersen two-stage samplers to collect bacterial samples, the replicate
plate method to isolate organisms, and the Kirby-Bauer disk diffusion
method to determine antibiotic resistance.</p></sec><sec><title>Results</title><p>The percentage of organisms resistant to at least two antibiotic classes
and all four classes evaluated were, respectively, 2.1 and 3.0 times
higher inside (<italic>n</italic> = 69) than upwind (<italic>n</italic> = 59) of the facility. <italic>Staphylococcus aureus</italic> was the most prevalent organism recovered. Concentrations of antibiotic-resistant <italic>S. aureus</italic> decreased with increasing distance from the facility. Using Fisher’s
exact methods, the change in distribution of antibiotic resistance
profiles for each antibiotic was statistically significant (oxytetracycline, <italic>p</italic> = 0.010; tetracycline, <italic>p</italic> = 0.014; ampicillin, <italic>p</italic> = 0.007; erythromycin, <italic>p</italic> = 0.035); however, this relationship was not seen with lincomycin
and penicillin (<italic>p</italic> > 0.05). In addition, the levels of antibiotic-resistant <italic>S. aureus</italic> 25 m downwind were significantly greater than the levels from samples
taken upwind from the facility for the same four antibiotics (<italic>p</italic> < 0.05). The percentage of resistant group A streptococci and fecal
coliform increased within the facility compared with upwind values for
all antibiotics evaluated, except for lincomycin. The percentage of
resistant total coliform organisms increased within the facility compared
with upwind values for oxytetracycline and tetracycline.</p></sec><sec><title>Conclusions</title><p>Bacterial concentrations with multiple antibiotic resistances or multidrug
resistance were recovered inside and outside to (at least) 150 m downwind
of this facility at higher percentages than upwind. Bacterial
concentrations with multiple antibiotic resistances were found within
and downwind of the facility even after subtherapeutic antibiotics were
discontinued. This could pose a potential human health effect for those
who work within or live in close proximity to these facilities.</p></sec>
|
<contrib contrib-type="author"><name><surname>Gibbs</surname><given-names>Shawn G.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001032">1</xref></contrib><contrib contrib-type="author"><name><surname>Green</surname><given-names>Christopher F.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001032">2</xref></contrib><contrib contrib-type="author"><name><surname>Tarwater</surname><given-names>Patrick M.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001032">1</xref></contrib><contrib contrib-type="author"><name><surname>Mota</surname><given-names>Linda C.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001032">1</xref></contrib><contrib contrib-type="author"><name><surname>Mena</surname><given-names>Kristina D.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001032">1</xref></contrib><contrib contrib-type="author"><name><surname>Scarpino</surname><given-names>Pasquale V.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001032">2</xref></contrib>
|
Environmental Health Perspectives
|
<p>Modern animal husbandry in the United States and other parts of the world
has evolved the swine industry from one that was pasture based into
a system based predominantly upon confinement and concentration of animals (<xref rid="b19-ehp0114-001032" ref-type="bibr">Perez-Trallero and Zigorraga 1995</xref>; <xref rid="b21-ehp0114-001032" ref-type="bibr">Scarpino and Quinn 1998</xref>). Most current animal production facilities rely on confined animal feeding
operations (CAFOs) and the addition of subtherapeutic doses of broad-spectrum
antibiotics to swine feed as a cheaper way to prevent disease
and maintain production yields (<xref rid="b24-ehp0114-001032" ref-type="bibr">Witte 1998</xref>).</p><p>Antimicrobials are known to promote growth in swine and improve the efficiency
of feed conversion and can affect bacterial and fungal disease
prophylaxis among the confined animals (<xref rid="b4-ehp0114-001032" ref-type="bibr">Davies and Roberts 1999</xref>). These treated animals generally gained weight 4–5% faster
than other animals not given the antibiotics. Feeding animals antibiotics
is associated with the development of antibiotic-resistant bacteria
within these animals (<xref rid="b2-ehp0114-001032" ref-type="bibr">Burriel 1997</xref>; <xref rid="b18-ehp0114-001032" ref-type="bibr">Nijsten et al. 1996</xref>; <xref rid="b23-ehp0114-001032" ref-type="bibr">Threlfall et al. 1993</xref>). Antibiotic use within food production animals has raised concern among
public health authorities regarding the development of antibiotic-resistant
bacteria in dosed animals and the possible subsequent impact
on the health of farmworkers and others in proximity to the CAFOs (<xref rid="b24-ehp0114-001032" ref-type="bibr">Witte 1998</xref>).</p><p>Antibiotics are the leading treatment method for bacterial infectious diseases, which
remain the most common cause of death worldwide (<xref rid="b14-ehp0114-001032" ref-type="bibr">McGeer 1998</xref>). It is widely accepted that antibiotic-resistant pathogens make clinical
treatment more difficult (<xref rid="b22-ehp0114-001032" ref-type="bibr">Takafuji 1977</xref>). At local levels, areas surrounding swine production facilities might
notice a rise in the difficulty of treating human health problems (<xref rid="b8-ehp0114-001032" ref-type="bibr">Haglind and Rylander 1987</xref>). These health concerns include, but are not limited to, respiratory problems, infectious
disease, and hypersensitive reactions (<xref rid="b5-ehp0114-001032" ref-type="bibr">DuPont and Steele 1987</xref>). Those individuals who live or work in proximity to a facility spreading
antibiotic-resistant bacteria could face higher exposures to these
organisms (<xref rid="b21-ehp0114-001032" ref-type="bibr">Scarpino and Quinn 1998</xref>). <xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. (2004)</xref> recovered antibiotic-resistant organisms known to have adverse human health
effects both inside and downwind of the facility. It is an important
next step to begin evaluating the distance these organisms can travel
within bioaerosols to eventually address public health impact.</p><p>The CAFO evaluated in this study was not using subtherapeutic doses of
antibiotics at the time air sampling was conducted; however, the animals
had received subtherapeutic levels of antibiotics 4 weeks before sampling. The
primary objective of this study was to determine the levels
of antibiotic-resistant bacteria, including multidrug-resistant bacteria (those
resistant to at least two classes of antibiotics) found in
air plumes 25 m upwind and 25, 50, 100, and 150 m downwind from a CAFO. These
organisms could affect the health of those in proximity to the
facility, such as employees. We hypothesized that the quantity of antibiotic-resistant
bacteria would show a negative correlation with distance
from the CAFO facility, which would support previous research showing
that the animals within CAFOs are significant sources of antibiotic-resistant
organisms (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>).</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec><title>Sample collection</title><p>The sampling site, a 4-year-old facility that houses up to 1,000 sows for
reproduction purposes, has been described previously (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>; <xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al. 2006</xref>). The building is 12 m wide × 60 m long × 3 m high; its
sides are concrete to 1 m, with mesh above the concrete to allow air
exchange. Computer-controlled shades, located above the mesh, are adjusted
depending upon the facility’s internal temperature. The
facility employs a chimney ventilation system to draw air through the
sides of the building and up through the roof. This system, in conjunction
with the shades, cools the hogs and helps maintain the temperature
of the building. The facility has a grated floor that allows waste material
to fall through into a 1.3-m deep pit that runs the length and
width of the facility. Subtherapeutic levels of oxytetracycline were
administered to hogs for 2 weeks. The animals were not being given subtherapeutic
levels of antibiotics during the study period and had not
been exposed to subtherapeutic levels of antibiotics for 4 weeks. The
waste material was removed from the pit twice a year and injected into
the cropland surrounding all sides of the confinement facility as a source
of nutrients; however, at the time of sampling for this study, the
injection of waste material had not been done in > 4 months.</p><p>The site was sampled four times at different times of the day, with sampling
location sampled simultaneously on 16 June 2003 (in the afternoon), 14 July 2003 (in
the afternoon), 21 July 2003 (in the morning), and 28 July 2003 (in
the evening). The sampling was done at different times
of day to accommodate the needs of the facility operator. Methods
were adapted from previous studies (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>; <xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al. 2006</xref>). All sampling material that could be autoclaved was autoclaved for 15 min
at 15 psi and 121°C. Andersen two-stage samplers were sterilized
after each use, washed, and then sterilized again before their
next use. All other items were disinfected with a 70% ethanol
solution after each sampling trip and before the next sampling trip.</p><p>We used Andersen two-stage samplers to collect all bacterial samples from
the animal confinement facilities. The Andersen two-stage sampler is
a cascade impactor that contains 200 orifices for each of the two stages, which
separate particles according to their size. The sampler was
loaded with plates of tryptic soy agar (TSA; Difco Laboratories, Detroit, MI), an
excellent general agar known to have the ability to culture
a variety of bacterial microorganisms. The nonrespirable particles
approximately 8 μm or larger were deposited on the first petri
dish, and the respirable particles of 8 μm down to 0.8 μm
were deposited on the second petri dish.</p><p>During sampling, the wind direction and wind speed were determined (Davis
Vantage Pro weather station; Davis Instruments Corp., Hayward, CA). Air
samples were taken immediately upwind of the facility, inside the
facility, immediately downwind, and 25, 50, 100, and 150 m downwind. Triplicate
samples were taken at each location for quality control. Each
sample was taken from the top of a tripod 1.3 m above the ground or
floor to simulate the height of the average person. Separate equipment, including
a pump (Gast Oil-less Pressure/Vacuum Pump; Gast Manufacturing, Inc., Benton
Harbor, MI) and an Andersen two-stage sampler were
used for each location on the site. The pump was calibrated to 28.1 L/min
before each sampling event. Sampling time varied between 15 sec and 5min, depending
on the site’s proximity to the facility, to
provide a countable number of colony-forming units (cfu) per plate; samples
were taken in triplicate. We followed this procedure for each of
the sampling locations. The plates were always handled using aseptic
technique to ensure that the air sample was not contaminated and were
returned to the laboratory for analysis within 12 hr. In the laboratory, the
plates were placed in an inverted position in an incubator at 35°C. The
colonies that developed were counted after 24 and 48 hr
to determine if the plates were overgrown. After 48 hr of incubation, the
plates were inverted and refrigerated at 4°C until they
were ready to be used for the replica plate method (<xref rid="b11-ehp0114-001032" ref-type="bibr">Lederberg and Lederberg 1952</xref>).</p></sec><sec><title>Isolation and speciation</title><p>We used the replica plate method to identify recovered aerosolized bacteria
by transferring the bacterial colonies onto a selective medium (<xref rid="b11-ehp0114-001032" ref-type="bibr">Lederberg and Lederberg 1952</xref>). The replica plate method was conducted using mannitol salt agar for <italic>Staphylococcus</italic> spp., MacConkey agar for coliforms, fecal coliform agar for fecal coliforms, and
selective <italic>Streptococcus</italic> agar for isolation group A streptococci (Difco Laboratories). We investigated <italic>Staphylococcus</italic> spp. and coliforms because previous studies had found them in abundance
inside CAFOs (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>; <xref rid="b12-ehp0114-001032" ref-type="bibr">Lenhart 1982</xref>; <xref rid="b21-ehp0114-001032" ref-type="bibr">Scarpino and Quinn 1998</xref>). After pressing of the selective media, TSA was used as a final control
for the method, being pressed first and last to ensure that the organisms
were being completely transferred to all plates. All plates were
incubated at 35°C and counted at 24 and 48 hr. We further confirmed
the presence of <italic>Staphylococcus aureus</italic> using Bacto coagulase plasma (Fisher Scientific, Houston, TX). We performed
the replica plate method using aseptic techniques. After counting, the
plates were refrigerated in an inverted position at 4°C
until they were ready to be transferred onto TSA slants to be used for
the Kirby-Bauer disk diffusion method.</p></sec><sec><title>Antimicrobial susceptibility testing</title><p>We used the Kirby-Bauer disk diffusion method to determine the antibiotic-resistant
characteristics of the recovered organisms (<xref rid="b1-ehp0114-001032" ref-type="bibr">Bauer et al. 1966</xref>). Three Mueller-Hinton agar plates and three TSA plates were brought to
room temperature and dried for each microorganism to be tested for antibiotic
resistance. The TSA plates were used to ensure purity of the
micro-organisms. A sterile cotton swab was used to transfer several colonies
of the microorganism from the slant to a sterile saline tube until
the tube was the same turbidity as the 0.5 McFarland standard under
examination. This gave an estimated 10<sup>8</sup> cfu/mL. The Kirby-Bauer disk diffusion method was then performed with
aseptic techniques. The plates were checked for susceptibility after 24 hr. The
zones of inhibition were recorded for all of the plates and
then compared with the standard [<xref rid="b15-ehp0114-001032" ref-type="bibr">National Committee for Clinical Laboratory Standards (NCCLS) 1997</xref>]. We then determined whether the microorganism was susceptible, intermediately
resistant, or resistant to each antibiotic evaluated. <xref ref-type="table" rid="t1-ehp0114-001032">Table 1</xref> provides the specific NCCLS zone diameters used to categorize <italic>S. aureus</italic>, group A streptococci, fecal coliforms, and total coliforms as susceptible, intermediate, or
resistant.</p><p>Six types of antibiotic susceptibility test disks (Difco Laboratories) were
used in the Kirby-Bauer method. All six drugs (20 μg oxytetracycline, 30 μg
tetracycline, 15 μg erythromycin, 10 μg
ampicillin, 10 μg penicillin, and 2 μg lincomycin) are
commonly used in both animal agriculture and human medicine. These
six antibiotics represent four distinct classes of antibiotics. Ampicillin
and penicillin are both penicillins, tetracycline and
oxytetracycline are both tetracyclines, lincomycin is a lincosamide, and
erythromycin is a macrolide. Multidrug resistance is defined as resistance
to at least two different classes of antibiotics.</p><p>Control organisms were obtained from cultures in the environmental microbiology
laboratory at the Shriner’s Burn Center (Cincinnati, OH). Control
organisms (<italic>Escherichia coli</italic>, ATCC #25922; <italic>Klebsiella pneumoniae</italic>, ATCC #31488; <italic>S. aureus</italic>, ATCC #29213; <italic>Streptococcus pneumoniae</italic>, ATCC #49619; American Type Culture Collection, Manassas, VA) were
used to test both the quality of the antibiotics and the media used. The
control organisms were applied to the selective media to ensure
that it would be able to culture the selected organism. The control
organisms were also put through the Kirby-Bauer method to ensure that
the antibiotics used would inhibit growth of a nonresistant culture.</p></sec><sec sec-type="methods"><title>Statistical analysis</title><p>In primary analyses we used contingency table methods (3 × 4) to
analyze the change in frequency of resistance, if any, associated with
distance downwind from the facility. That is, the frequency distribution
for the three categories of the resistance profile was compared
across the four distances downwind from the facility (25, 50, 100, and 150 m). Comparisons
were made regarding resistance to each antibiotic
in each organism. A nonsignificant result implies that distributions
of frequencies were relatively constant as distance changed. For secondary
analyses, contingency table methods (2 × 2) were also used
to compare frequencies at each distance downwind to the 25 m upwind
frequencies (e.g., 25 m upwind vs. 25 m downwind, 25 m upwind vs. 50 m
downwind). A nonsignificant result from these tests implies that frequencies
of organisms at a downwind location were not different from upwind. All <italic>p</italic>-values were calculated using Fisher’s exact methods because many
cell counts were zero and expected frequencies were < 5. Even though
the Andersen two-stage samplers separate particles according to their
size (non-respirable and respirable), the analyses were performed
only for total organisms because of the low numbers of some selected
organisms.</p></sec></sec><sec sec-type="results"><title>Results</title><p>The summary results of the sampling are presented in <xref ref-type="table" rid="t2-ehp0114-001032">Table 2</xref>. The total number of organisms found within the facility was 287 times
higher than the number recovered upwind of the facility. This number
decreased downwind of the facility as far as 150 m downwind (the farthest
downwind sampling distance in this study), where the number of organisms
was only 2.2 times higher than that recovered upwind of the facility. The
percentage of organisms resistant to at least two classes of
the antibiotics was 2.1 times higher inside of the facility than upwind
of the facility. This percentage decreased slightly downwind of the
facility; however, none of the percentages of resistance downwind of
the facility was statistically different from any other (<italic>p</italic> > 0.05). This indicated that out to 150 m downwind, the percentage
of organisms resistant to at least two classes of the antibiotics did
not change. The percentage of organisms resistant to all four classes
of the antibiotics evaluated was three times higher inside the facility
than upwind of the facility. This immediately decreased downwind of
the facility to a percentage similar to the upwind value, and none of
the percentages of resistance downwind of the facility was statistically
different from any other (<italic>p</italic> > 0.05) (<xref ref-type="table" rid="t2-ehp0114-001032">Table 2</xref>). <xref ref-type="fig" rid="f1-ehp0114-001032">Figure 1</xref> shows the logarithmic decrease in multidrug-resistant bacteria downwind
of the CAFO.</p><p>As previously reported by <xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al. (2006)</xref>, <italic>S. aureus</italic> was the most prevalent organism sampled, accounting for 76% (1.4 × 10<sup>4</sup> cfu/m<sup>3</sup>; SD, 8.9 × 10<sup>3</sup> cfu/m<sup>3</sup>) of the bacteria recovered inside of the CAFO. The percent resistant organisms
increased from upwind values inside of the facilities for all
antibiotics evaluated with the exception of ampicillin, which did not
change. <italic>S. aureus</italic> was the only organism evaluated for which the decreased concentrations
with increased distance downwind of the facility were statistically significant (<xref ref-type="table" rid="t3-ehp0114-001032">Table 3</xref>). <italic>S. aureus</italic> showed this statistically significant relationship with distance from
the facility and resistance profile for four of the antibiotics evaluated: oxytetracycline (<italic>p</italic> = 0.010), tetracycline (<italic>p</italic> = 0.014), ampicillin (<italic>p</italic> = 0.007), and erythromycin (<italic>p</italic> = 0.035); however, this relationship was not seen with lincomycin
or penicillin (<italic>p</italic> > 0.05). Secondary analysis of <italic>S. aureus</italic> also showed a difference in resistant bacteria between upwind values and
those for immediately downwind (25 m) for resistance to oxytetracycline, tetracycline, ampicillin, and erythromycin (<italic>p</italic> > 0.05); however, this relationship was not observed with lincomycin
or penicillin (<italic>p</italic> > 0.05).</p><p>The percentage of resistant group A streptococci increased within the facility
compared with upwind values for all antibiotics evaluated except
lincomycin (<xref ref-type="table" rid="t4-ehp0114-001032">Table 4</xref>). The percentage of resistant group A streptococci was not statistically
different at any of the downwind distances (<italic>p</italic> > 0.05), and all downwind values were similar to the upwind values (<italic>p</italic> > 0.05) (<xref ref-type="table" rid="t4-ehp0114-001032">Table 4</xref>).</p><p>The percentage of resistant fecal coliform organisms increased within the
facility compared with upwind values for all antibiotics evaluated
except lincomycin (<xref ref-type="table" rid="t5-ehp0114-001032">Table 5</xref>). The percentage of resistant fecal coliform organisms was not statistically
different for any of the downwind distances (<italic>p</italic> > 0.05), and all downwind values were similar to the upwind values
for all antibiotics except lincomycin (<italic>p</italic> = 0.011) (<xref ref-type="table" rid="t5-ehp0114-001032">Table 5</xref>).</p><p>The percentage of resistant total coliform organisms increased within the
facility compared with upwind values only for oxytetra-cycline and
tetracycline (<xref ref-type="table" rid="t6-ehp0114-001032">Table 6</xref>). The percentage of resistant total coliform organisms was not statistically
different for any of the downwind distances (<italic>p</italic> > 0.05), and all downwind values were similar to the upwind values
for all antibiotics, with the exceptions of lincomycin and penicillin, which
could not be evaluated statistically (<xref ref-type="table" rid="t6-ehp0114-001032">Table 6</xref>).</p></sec><sec sec-type="discussion"><title>Discussion</title><p>This study was conducted over a month during the summer of 2003 in the
American Midwest in conjunction with a previously published study (<xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al. 2006</xref>). In the present study, we consistently found bacteria that exhibited
multiple antibiotic resistances to at least two classes of the study antibiotics. In
a previous study (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>), we demonstrated that the animals within the CAFO were responsible for
the density of organisms released from the facility and the source of
the antibiotic-resistant organisms. We also checked for patterns in
multiple antibiotic resistances for all strains of bacteria isolated. In
the present study, we found multiple antibiotic resistance present
out to 150 m from the CAFO; these percentages were significantly higher
than those recovered upwind of the facility (<xref ref-type="table" rid="t2-ehp0114-001032">Table 2</xref>) and could affect employee health. It is important to note that in the
previously published study the animals were currently receiving subtherapeutic
antibiotics (<xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>), whereas in this study the animals had received nontherapeutic doses
of antibiotics 4 weeks before sampling. This would seem to indicate that
antibiotic-resistant bacteria have been selected as a result of the
use of the nontherapeutic levels of oxy-tetracycline and are persisting
in the swine environment even after use has ceased. This is in agreement
with the findings of <xref rid="b13-ehp0114-001032" ref-type="bibr">Manson et al. (2004)</xref> and <xref rid="b10-ehp0114-001032" ref-type="bibr">Johnsen et al. (2005)</xref>.</p><p>As in previous studies (<xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. 2005</xref>; <xref rid="b6-ehp0114-001032" ref-type="bibr">Gibbs et al. 2004</xref>; <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. 2002</xref>), <italic>Staphylococcus</italic> was one of the most prevalent culturable genera of bacteria recovered
from swine CAFOs, and it exhibited multiple antibiotic resistances. <italic>S. aureus</italic> in the present study had multiple antibiotic resistances throughout the
distances examined (<xref ref-type="table" rid="t3-ehp0114-001032">Table 3</xref>). <xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. (2005)</xref> found that <italic>Staphylococcus</italic> spp. accounted for 32% of the organisms they recovered. This is
significantly less than the 76% recovered in this study and 84.1% recovered
by <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. (2002)</xref>. However, this difference could be due to the different collection methods: we
and <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. (2002)</xref> used impaction methods, whereas <xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. (2005)</xref> used all-glass impingers. It is possible that the all-glass impingers
provided better collection of other organisms or less collection of <italic>Staphylococcus</italic> spp. compared with the impaction collectors (<xref rid="b9-ehp0114-001032" ref-type="bibr">Jensen et al. 1992</xref>). However, the impinger collectors did not provide size differentiation. Both <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. (2002)</xref> and <xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. (2005)</xref> used media other than TSA: <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. (2002)</xref> placed R2A agar in Andersen samplers; and <xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. (2005)</xref> used mE agar for the isolation of <italic>Enterococcus</italic> isolates and tested each isolate for the production of catalase in the
presence of 3% hydrogen peroxide. Catalase-positive isolates
were then identified as <italic>Staphylococcus</italic> species. The differences in collection media used could also account for
the variability in <italic>Staphylococcus</italic> spp. recovery. The continued recovery of large densities of <italic>S. aureus</italic> from the bioaerosols indicates that future research should focus more
effort on culturable and nonculturable <italic>S. aureus</italic>, as well as other important human pathogens. This study and the others
discussed (<xref rid="b3-ehp0114-001032" ref-type="bibr">Chapin et al. 2005</xref>; <xref rid="b20-ehp0114-001032" ref-type="bibr">Predicala et al. 2002</xref>) examined only culturable bacterial organisms; the lack of examination
of other nonculturable bacteria is a limitation of the studies. The inclusion
of nonculturable bacteria may change the levels of multiple antibiotic
resistances, which will have to be examined in a separate study.</p><p>In the present study, an estimated 17,000 of the 18,000 cfu/m<sup>3</sup> released from the CAFO were defined as multidrug-resistant or multiple-antibiotic–resistant organisms because they were resistant to
at least two classes of antibiotics. By comparison, the air located upwind
of the CAFO contained an estimated 28 cfu/m<sup>3</sup> that were multidrug resistant. Approximately 8,200 cfu/m<sup>3</sup> recovered from inside the CAFO were resistant to all four classes of antibiotics
evaluated, whereas 8.8 cfu/m<sup>3</sup> recovered upwind showed the same level of resistance. This shows that
individuals who work inside the facility or live in proximity downwind
of the facility face a greater exposure to multidrug-resistant organisms, which
could potentially affect human health.</p><p><xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al. (2006)</xref> estimated that the bacterial concentration downwind of the facility would
equal the upwind concentration at approximately 175 m from the facility. Similar
predictions can be made with the multidrug-resistant bacterial
concentration. <xref ref-type="fig" rid="f1-ehp0114-001032">Figure 1</xref> shows the logarithmic decrease in multidrug-resistant bacteria downwind
of the CAFO. This indicates that those within 175 m downwind and inside
the facility receive a greater exposure to multidrug-resistant organisms
than those upwind of the facility.</p><p>Both the increase in percentage and quantity of multidrug-resistant bacteria
inside and downwind of the facility support <xref rid="b7-ehp0114-001032" ref-type="bibr">Green et al.’s (2006)</xref> statement that these facilities could pose a hazard to persons in direct
proximity to them. This would include those employed at the facility
and those who live in close proximity to the facility. This potential
health hazard exists independently of a halt in subtherapeutic treatment.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Bacterial concentrations with multiple antibiotic resistances or multidrug
resistances were routinely recovered inside and up to 150 m downwind
of this facility at higher percentages than upwind of the facility. Subsequent
numbers of multiple-antibiotic–resistant bacteria
are almost three orders of magnitude higher inside the facility compared
with upwind. These elevated concentrations persist to (at least) 150 m
downwind of the facility. Our findings indicate that bacterial concentrations
with multiple antibiotic resistances are found within and
downwind of CAFOs even after subtherapeutic doses of antibiotics are removed
from the animal feed. Those working at or inside the facility and
those living in close proximity downwind of the facility could be at
risk for adverse human health effects associated with exposure to large
numbers of multidrug-resistant organisms.</p></sec>
|
Evaluation of Serum Immunoglobulins among Individuals Living Near Six Superfund
Sites
|
<p>Residents living in communities near Superfund sites have expressed concern
that releases from these facilities affect their health, including
adverse effects on their immune systems. We used data from six cross-sectional
studies to evaluate whether people who live near several Superfund
sites are more likely to have individual immunoglobulin test results (IgA, IgG, and
IgM) below or above the reference range than those
who live in comparison areas with no Superfund site. Study participants
consisted of target-area residents who lived close to a Superfund
site and comparison-area residents who were not located near any Superfund
or hazardous waste sites. A consistent modeling strategy was used
across studies to assess the magnitude of the relationship between area
of residence and immunoglobulin test results, adjusting for potential
confounders and effect modifiers. In all study areas, the results
suggest that people who live near a Superfund site may have been more
likely to have IgA test results above the reference range than comparison
areas residents regardless of modeling strategy employed. The effect
measures were larger for residents who lived in communities near military
bases with groundwater contamination. For all analyses the wide
confidence intervals reflect uncertainty in the magnitude of these effects. To
adequately address the question of whether the immune system
is affected by low-level exposures to hazardous substances, we recommend
that more functional immunotoxicity tests be conducted in human populations
where individual exposure information is available or when it
can be reasonably estimated from environmental exposure measurements.</p>
|
<contrib contrib-type="author"><name><surname>Williamson</surname><given-names>Dhelia M.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001065">1</xref><xref ref-type="aff" rid="af2-ehp0114-001065">2</xref></contrib><contrib contrib-type="author"><name><surname>White</surname><given-names>Mary C.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001065">3</xref></contrib><contrib contrib-type="author"><name><surname>Poole</surname><given-names>Charles</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001065">2</xref></contrib><contrib contrib-type="author"><name><surname>Kleinbaum</surname><given-names>David</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001065">4</xref></contrib><contrib contrib-type="author"><name><surname>Vogt</surname><given-names>Robert</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001065">3</xref></contrib><contrib contrib-type="author"><name><surname>North</surname><given-names>Kari</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001065">2</xref></contrib>
|
Environmental Health Perspectives
|
<p>It is estimated that more than 14.5 million people in the contiguous United
States live within 1 mi of at least one Superfund site (<xref rid="b16-ehp0114-001065" ref-type="bibr">Heitgerd and Lee 2003</xref>). Common contaminants at these sites include heavy metals, volatile organic
compounds, and chlorinated hydrocarbons. Many of these substances
are present at elevated levels, with the potential for on-site and off-site
human contact (<xref rid="b16-ehp0114-001065" ref-type="bibr">Heitgerd and Lee 2003</xref>). Residents living in communities near Superfund sites have expressed
concern that these releases affect their health, including adverse effects
on their immune systems. These concerns have been difficult to address
because the available epidemiologic studies regarding immunologic
end points are based on occupationally exposed workers or accidentally
exposed cohorts exposed to high levels of environmental contaminants (<xref rid="b24-ehp0114-001065" ref-type="bibr">Tryphonas 2001</xref>).</p><p>To evaluate the potential effect of environmental toxicants on the immune
system of residents in communities located near hazardous waste sites, the
U.S. Agency for Toxic Substances and Disease Registry (ATSDR) developed
an immune test battery for inclusion in community health studies (<xref rid="b1-ehp0114-001065" ref-type="bibr">ATSDR 1994a</xref>). This test battery was proposed as a general evaluation of immune status
to be used when there was no clear indication of particular health
effects or well-defined exposures (<xref rid="b26-ehp0114-001065" ref-type="bibr">Vogt 1991</xref>). The basic immune test battery consists of lymphocyte flow cytometric
immunophenotyping with specific cluster designation antibody reagents
to identify the major types of lymphocytes (CD4 lymphocyte count, CD8 lymphocyte
count, CD4:CD8 ratio, T cells, and B cells) and quantitative
levels of immunoglobulins (IgA, IgG, and IgM). The tests included in
this panel were chosen to assess immune dysfunction in conjunction with
self-reported symptoms and illnesses relating to these conditions.</p><p>The biomarker panel was applied in several ATSDR cross-sectional health
investigations. However, the interpretation of the immunoglobulin values
was difficult because of the wide biological variability within populations, nonspecific
nature of the tests, and lack of established reference
ranges for these tests. In 1998, the Foundation for Blood Research
in Scarborough, Maine, provided age- and sex-specific reference limits
for IgA, IgG, and IgM (<xref rid="b21-ehp0114-001065" ref-type="bibr">Ritchie et al. 1998</xref>). These reference limits were based on automated immunoassay values from 115,017 serum
samples, which represents the largest study population
in North America obtained within a single laboratory.</p><p>The purpose of this study was to reevaluate immunoglobulin levels collected
over several investigations by using a consistent approach to data
analysis to determine whether individuals who live near several Superfund
sites are more likely to have test results below or above the reference
range than individuals who live in comparison areas with no Superfund
site. Other factors that may be potential confounders or modifiers
of these associations are examined as well.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Study areas</title><p>We used data from six cross-sectional studies conducted by the ATSDR between 1991 and 1994 for
this analysis. These studies were conducted in
Kentucky (<xref rid="b4-ehp0114-001065" ref-type="bibr">ATSDR 1995b</xref>), Texas (<xref rid="b6-ehp0114-001065" ref-type="bibr">ATSDR 1995d</xref>), California (<xref rid="b8-ehp0114-001065" ref-type="bibr">ATSDR 1996b</xref>), Nebraska (<xref rid="b9-ehp0114-001065" ref-type="bibr">ATSDR 1996c</xref>), Massachusetts (<xref rid="b10-ehp0114-001065" ref-type="bibr">ATSDR 1998a</xref>), and North Carolina (<xref rid="b11-ehp0114-001065" ref-type="bibr">ATSDR 1998b</xref>). <xref ref-type="table" rid="t1-ehp0114-001065">Table 1</xref> lists the study areas, types of facilities, potential exposure pathways, and
contaminants of concern. All six studies were approved by the Centers
for Disease Control and Prevention Institutional Review Board. Study
participants in all areas gave informed consent before participating.</p><p>Four additional cross-sectional studies conducted by ATSDR during this
time period were not included in this analysis because they did not include
comparison populations (<xref rid="b3-ehp0114-001065" ref-type="bibr">ATSDR 1995a</xref>, <xref rid="b7-ehp0114-001065" ref-type="bibr">1996a</xref>) or did not use questionnaires similar to those used in the other studies (<xref rid="b2-ehp0114-001065" ref-type="bibr">ATSDR 1994b</xref>, <xref rid="b5-ehp0114-001065" ref-type="bibr">1995c</xref>).</p></sec><sec sec-type="methods"><title>Study population</title><p>Each of the studies included randomly selected community residents. We
conducted a census of each community to create the sampling frame. Target
area populations consisted of residents living in well-defined areas
located close to a Superfund site. Participation rates ranged from 48 to 86% across
the six sites and were generally higher among
target area residents. We selected each target area based on environmental
sampling data that identified contaminated soil, groundwater, surface
water, or sediment. Individual exposure data typically were not
available. We selected comparison area communities on the basis of demographics
and socioeconomic status similar to those of the target area
community. Some of the socioeconomic factors considered were style and
age of housing, household income, and degree of urbanization. The comparison
areas were > 5 miles from the sites of interest and were not
located near any other Superfund or hazardous waste sites.</p></sec><sec sec-type="methods"><title>Data collection</title><p>We asked study participants to be interviewed and provide a blood sample. The
interview collected information on sociodemographic characteristics (age, race, sex, educational level, years of residence), history
of chronic diseases (arthritis, rheumatism, chronic bronchitis, asthma, cancer, multiple
sclerosis, lupus), history of specific symptoms (skin
rashes, eczema, asthma, bronchitis, allergies), smoking status of the
study participant and other household residents, and rating of general
health (excellent, good, fair, poor). At three of the study sites (Kentucky, Nebraska, and
North Carolina) we collected information about
sources of heat for the home (coal stove, fireplace, kerosene or gas
heater, or wood stove) and occupational exposures to chemicals (solvents, cleaning
agents, dust, insulating materials, paints, gasoline, or
kerosene).</p></sec><sec><title>Determination of serum immunoglobulins</title><p>Sera were separated by centrifugation at the phlebotomy site and shipped
to the Foundation for Blood Research in Scarborough, Maine, where they
were refrigerated and assayed within 3 working days. The immunoglobulins
measured corresponded to those recommended in the test battery (<xref rid="b1-ehp0114-001065" ref-type="bibr">ATSDR 1994a</xref>), which did not include IgE. We tested samples using the immunoturbidimetry
method previously described (<xref rid="b17-ehp0114-001065" ref-type="bibr">Hudson et al. 1987</xref>). Because of variations among sex and age groups, reference distributions
for IgA, IgG, and IgM measurements are sex and age specific (<xref rid="b21-ehp0114-001065" ref-type="bibr">Ritchie et al. 1998</xref>).</p></sec><sec sec-type="methods"><title>Statistical analysis</title><p>We used polytomous logistic regression to examine the relationship between
area of residence and immunoglobulin test result. We used a three-category
classification to code the immunoglobulin test results into those
above (> 97.5th percentile), within, or below (< 2.5th percentile) the
reference range because increases and decreases of immunoglobulins
have been associated with adverse health outcomes (<xref rid="b15-ehp0114-001065" ref-type="bibr">Fischbach 2000</xref>). SAS statistical software (version 9.0; SAS Institute Inc., Cary, NC) was
used for data management and statistical analysis.</p><p>We analyzed the data from each of the six studies using the same modeling
strategies to assess the magnitude of the relationship between area
of residence and immunoglobulin test results, adjusting for factors that
may be potential confounders or modifiers of these associations. The
modeling strategies included conducting the following five logistic
regression analyses: model 1, no adjustment (crude); model 2, adjustment
for sociodemographic variables only; model 3, adjustment for sociodemographic
variables and other exposure information (i.e., smoking status
of study participant and other household residents; use of coal stove, fireplace, kerosene
or gas heater, or wood stove as source of heat
for the home; and occupational exposure to chemicals); model 4, adjustment
for sociodemographic variables, other exposure information, history
of specific symptoms and illness, and rating of general health; and
model 5, a backward elimination method described by <xref rid="b18-ehp0114-001065" ref-type="bibr">Kleinbaum (1994)</xref>. To be included in the regression analysis, studies had to have at least 10 individuals
with immunoglobulin test results either below or above
the reference range and at least three individuals in each the target
and comparison group.</p><p>Model 5 included assessing each of the six studies individually using the
following strategy. First, interactions between the exposure variable (residence
in the target vs. comparison group) and one covariate at
a time were examined. Factors that may affect immunologic assay results
were considered effect-measure modifiers if the Breslow-Day <italic>p</italic>-value was < 0.5 (<xref rid="b12-ehp0114-001065" ref-type="bibr">Breslow and Day 1980</xref>). Next, those covariates not found to be effect-measure modifiers were
assessed univariately as potential confounders. A covariate was deemed
a confounder if the absolute value of the natural logarithm of the ratio
of the unadjusted to adjusted odds ratio (OR) exceeded 0.10. All
variables considered effect-measure modifiers or confounders were included
in the full model. We assessed each interaction term one at a time
using the backward elimination method to eliminate insignificant variables
from the model. We assessed significance by comparing the change
in log likelihoods (α < 0.20).</p></sec></sec><sec sec-type="results"><title>Results</title><sec sec-type="intro"><title>Descriptive characteristics</title><p>Most study participants from each study area were white, were older than 30 years, and
had attained at least a high school education (<xref ref-type="table" rid="t2-ehp0114-001065">Table 2</xref>). The Texas study had a more diverse racial and ethnic composition. Years
of residence in the current home varied considerably among the study
sites; all of the study participants in Nebraska had lived in their
residences for at least 10 years, whereas nearly all study participants
in Texas had lived in their residences for < 10 years. Sample sizes
of the six studies ranged from 258 participants in the North Carolina
study to 912 participants in the Massachusetts study.</p><p>Most study participants in both the target and comparison groups for all
six areas were in good or excellent health; did not report having a
history of specific symptoms, illness, or allergies; did not currently
smoke; and did not live in a household in which someone else smoked (<xref ref-type="table" rid="t3-ehp0114-001065">Table 3</xref>). Only in North Carolina did most study participants report using a coal
stove, fireplace, kerosene or gas heater, or wood stove to heat their
house. Most study participants in both Kentucky and North Carolina
reported having occupational exposure to chemicals.</p><p>The proportion of study participants who had individual immunoglobulin
test results (IgA, IgG, and IgM) either below or above the reference range
varied by specific immunoglobulin and study site (<xref ref-type="table" rid="t4-ehp0114-001065">Table 4</xref>). Overall, the percentage of study participants in both target and comparison
groups with an immunoglobulin test result above the reference
range was generally higher than the percentage of study participants with
an immunoglobulin test result below the reference range.</p></sec><sec><title>Multivariate analyses</title><p><xref ref-type="table" rid="t5-ehp0114-001065">Tables 5</xref>–<xref ref-type="table" rid="t7-ehp0114-001065">7</xref> show the ORs and 95% confidence intervals (CIs) examining the
relationship between area of residence (target vs. comparison) and having
an immunoglobulin (IgA, IgG, IgM) test result below or above the reference
range in six geographic areas using the different modeling strategies
described above. Results from the backward elimination modeling
strategy (model 5), which included interaction terms, are discussed
individually.</p></sec><sec><title>Immunoglobulin A</title><sec sec-type="results"><title>Results below the reference range</title><p>Target area residents in four study areas (California, Kentucky, North
Carolina, and Texas) had an increased prevalence of having an IgA test
result below the reference range compared with comparison area residents, whereas
target area residents in two study areas (Massachusetts and
Nebraska) had a decreased prevalence of having an IgA test result below
the reference range (<xref ref-type="table" rid="t5-ehp0114-001065">Table 5</xref>). OR estimates for Texas fell on both sides of the null, depending on
which modeling strategy was used. Data were too sparse for North Carolina
to generate adjusted OR estimates. All estimates were imprecise [upper-to-lower
confidence limit ratio (CLR) > 4] (<xref rid="b20-ehp0114-001065" ref-type="bibr">Poole 2001</xref>).</p><p>Using the backward elimination strategy (model 5), the models for California
and Nebraska included interaction terms. In California, the odds
of women living in the target area having an IgA test result below the
reference range were 2.66 times those of women living in the comparison
area. In contrast, the odds of males living in the target area having
an IgA test result below the reference range were 1.16 times those
of males living in the comparison area. In Nebraska, individuals who
reported having allergies and who lived in the target area were 1.29 times
more likely to have an IgA test result below the reference range
than those individuals who reported having allergies and who lived in
the comparison area. Individuals who reported not having allergies and
who lived in the target area in Nebraska were 0.23 times less likely
to have an IgA test result below the reference range than those living
in the comparison area who did not report having allergies.</p></sec><sec sec-type="results"><title>Results above the reference range</title><p>Target area residents in all study areas except North Carolina had an increased
prevalence of having IgA test results above the reference range
than comparison area residents (data were too sparse for North Carolina
to generate OR estimates). Adjusted OR estimates for Texas fell on
both sides of the unadjusted estimate, depending on which modeling strategy
was used. All estimates were imprecise (CLR > 4).</p><p>Using the backward elimination strategy (model 5), two interaction terms
were included in the Massachusetts model. The odds of women living in
the target area having an IgA test result above the reference range
were 11.3 times those of women living in the comparison area, whereas
the odds of males living in the target area were 1.66 times those living
in the comparison area. For smokers, the odds of having an IgA test
result above the reference range were 0.14 times lower among those living
in the target area than among those living in the comparison area, whereas
nonsmokers were 1.66 times more likely to have an IgA test result
above the reference range than nonsmokers living in the comparison
area.</p></sec></sec><sec><title>Immunoglobulin G</title><sec sec-type="results"><title>Results below the reference range</title><p><xref ref-type="table" rid="t6-ehp0114-001065">Table 6</xref> shows that target area residents in Nebraska and Texas had an increased
prevalence of having an IgG test result below the reference range compared
with comparison area residents, whereas target area residents in
Massachusetts had a decreased prevalence. Data were too sparse in California, Kentucky, and
North Carolina to generate OR estimates. Using
the backward elimination strategy (model 5), the OR for Texas was substantially
lower than the estimates generated from the other models. All
estimates were imprecise (CLR ≥ 4).</p></sec><sec sec-type="results"><title>Results above the reference range</title><p>When examining the relationship between area of residence and having an
IgG test result above the reference range, target area residents in four
study areas (Kentucky, Massachusetts, Nebraska, and Texas) had an
increased prevalence of IgG test results above the reference range compared
with comparison area residents, whereas target area residents in
one area (California) had a decreased prevalence. Data were too sparse
for North Carolina to generate OR estimates. In Massachusetts, adjustment
for additional covariates resulted in the OR estimates falling on
the opposite side of the null from the unadjusted OR. In Nebraska, the
OR generated from model 4 was approximately the null value. All OR
estimates were imprecise (CLR ≥4).</p></sec></sec><sec><title>Immunoglobulin M</title><sec sec-type="results"><title>Results below the reference range</title><p>None of the six studies had a total of 10 individuals with IgM test results
below the reference range or had at least three individuals in both
the target and comparison areas, so no analyses were conducted (<xref ref-type="table" rid="t7-ehp0114-001065">Table 7</xref>).</p></sec><sec sec-type="results"><title>Results above the reference range</title><p><xref ref-type="table" rid="t7-ehp0114-001065">Table 7</xref> shows that target area residents in two study areas (Kentucky and North
Carolina) had a modest increased prevalence of having IgM test results
above the reference range compared with comparison area residents, whereas
target area residents in three study areas (California, Massachusetts, and
Texas) had a decreased prevalence. The OR estimates for Nebraska
fell on both sides of the null depending on which modeling strategy
was used. Data were too sparse for North Carolina to generate adjusted
OR estimates. All estimates were imprecise (CLR ≥4).</p></sec></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Evidence from both human and animal studies suggests that a variety of
chemicals, including volatile organic compounds and metals, are able to
adversely affect the immune system (<xref rid="b1-ehp0114-001065" ref-type="bibr">ATSDR 1994a</xref>, <xref rid="b11-ehp0114-001065" ref-type="bibr">1998b</xref>; <xref rid="b13-ehp0114-001065" ref-type="bibr">Burns 1996</xref>; <xref rid="b19-ehp0114-001065" ref-type="bibr">National Research Council 1992</xref>; <xref rid="b23-ehp0114-001065" ref-type="bibr">Snyder 1994</xref>). Xenobiotic toxicants have been shown to either augment the normal immune
response, resulting in hypersensitivity, or suppress the immune responses, resulting
in immune deficiency. The consequences of immunosuppression
may include respiratory infections, opportunistic infections, and
cancer (<xref rid="b14-ehp0114-001065" ref-type="bibr">Descotes and Choquet-Kastylevsky 2001</xref>). The consequences of immunoenhancement are less well established but
include influenza-like reactions such as chills, malaise, and hypotension, as
well as exacerbation of chronic infections, psoriasis, Crohn disease, and
autoimmune diseases (<xref rid="b14-ehp0114-001065" ref-type="bibr">Descotes and Choquet-Kastylevsky 2001</xref>).</p><p>Because of the development of standardized reference ranges for IgA, IgG, and
IgM, we were able to explore a question that is often raised but
rarely investigated: whether individuals living near Superfund sites
are more likely to experience changes in immune status than individuals
living in areas with no nearby Superfund sites. We examined immunoglobulin
test results from six cross-sectional studies conducted in different
geographic areas using standardized reference ranges and consistent
modeling strategies. Our results suggest that there is variability
among the OR estimates generated when examining the relationship between
area of residence and having an immunoglobulin test result below
or above the reference range. The only consistent pattern observed in
all study areas was that individuals who live near a Superfund site were
more likely to have IgA test results above the reference range than
comparison area residents regardless of modeling strategy employed. However, the
wide CI values reflect large uncertainty in the magnitude
of these effects.</p><p>The effect measures for IgA were consistently larger (OR > 1.5) for
residents who lived in communities in Massachusetts and Nebraska near
military bases with only groundwater contamination. Although the estimates
were imprecise, our results also suggest that individuals living
closer to these military bases were less likely to have IgA test results
below the reference range than individuals who lived in the comparison
neighborhood. In addition, residents who lived near an industrial
complex in Kentucky with potential ambient air exposure to heavy metals
and other chemicals were more likely to have IgG test results above
the reference range than comparison area residents. Because the reference
ranges used for this analysis were age and sex adjusted, the observed
variability in immunoglobulin results is unlikely to be due to residual
confounding by age and sex.</p><p>Previous studies have also shown increased IgA levels among individuals
living near Superfund sites. ATSDR examined the association between biologic
markers of immune system impairment and environmental exposure
to cadmium and lead among children and adults living in communities contaminated
by mining and smelting operations at four Superfund sites (<xref rid="b22-ehp0114-001065" ref-type="bibr">Sarasua et al. 2000</xref>). For children 6–35 months of age, an association was found between
increased blood lead levels and increased serum IgA, IgG, and IgM. In
adults, urine cadmium was associated with higher levels of IgA after
adjustment for age, sex, and other confounders. Additionally, researchers
at the University of North Carolina at Chapel Hill examined the
effects on the immune system among residents living near the Pesticides
Dump Site in Aberdeen, North Carolina, who were potentially exposed
to 1,1-dichloro-2,2-bis(<italic>p</italic>-chlorophenyl)ethylene (DDE). The researchers found modestly increased
mean IgA levels with increased DDE levels (<xref rid="b25-ehp0114-001065" ref-type="bibr">Vine et al. 2001</xref>). Both these studies examined differences in mean IgA levels because standardized
reference ranges for immunoglobulins were not available.</p><p>The main strength of our study was the unique nature of the data. We used
questionnaire and biological data from nearly 4,000 individuals living
in six different geographic areas in the United States. The studies
used in this analysis were also standardized: they were conducted by
the same government agency during a relatively short time period and
used the same study design, questionnaire, and immune biomarker test battery. These
similarities allowed us to examine patterns of immunoglobulin
test results using standardized reference ranges across the different
study areas.</p><p>A major limitation of this study is the lack of exposure characterization. Individual
exposure data were not available, so area of residence
was used as a surrogate of exposure. Another limitation was the small
number of study participants in some locations, which precluded our ability
to have the statistical power to measure stable effect estimates. Also, the
information used in this study relied upon self-reported behaviors
and risk factors ascertained through interviewer-administered
questionnaires. Therefore, there is the possibility of misclassification
bias in that people in the target area might have been able to recall
symptoms or illnesses to a different extent than people in the comparison
areas. This bias could have limited our ability to adjust for
these potential confounders in the multivariate models. A final limitation
is that serum immunoglobulins are considered a fairly insensitive
indicator of immune functions</p><p>Because the immune system is a target for adverse effects from exposure
to hazardous substances, laboratory tests to measure immune status were
included in several ATSDR community health studies. It was thought
that the tests would provide quantification of effects given that values
outside established reference ranges are generally associated with
adverse health outcomes. However, to adequately address the question of
whether the immune system is affected by low-level exposures to hazardous
substances, we recommend that more functional immunotoxicity tests
be conducted in communities located near hazardous waste sites when
adequate sample size and individual exposure information are available
or when they can be reasonably estimated from environmental exposure
measurements. Tests of immune status could be included in such studies
but should be tailored for specific types of contaminants or health
end points. IgE should be included in the immune status tests because
individuals with allergic diseases have been shown to exhibit increased
IgE levels whereas decreased levels of IgE are found in cases of autoimmune
and other diseases. Information regarding potential confounders
should be collected through questionnaires or other mechanisms. Finally, the
use of a single reference laboratory would be ideal for quality
control and for comparison of results across studies.</p></sec>
|
Potential Immunotoxic Effect of Thimerosal: Compound Alters Dendritic Cell
Response <italic>in Vitro</italic>
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Barrett</surname><given-names>Julia R.</given-names></name></contrib>
|
Environmental Health Perspectives
|
<p>Thimerosal, an ethylmercury-based compound used for decades as a vaccine
preservative, has previously been linked to neurotoxic effects. New
research reveals that it may also affect the immune system by altering
how dendritic cells respond to biochemical signals <bold>[<italic>EHP</italic> 114:1083–1091; Goth et al.]</bold>.</p><p>Dendritic cells are influential primary actors in the immune system’s
response to infectious invasion of the body. Once activated, a
single dendritic cell can direct hundreds of T cells against an infectious
agent. This ability, however, depends on the dendritic cell responding
appropriately to signals.</p><p>Previous studies by other researchers have indicated that thimerosal is
an immunotoxicant, but its specific targets were unknown. Hypothesizing
that dendritic cells might be sensitive targets, the researchers cultured
bone marrow–derived dendritic cells from mice and assayed
how both mature and immature cells responded to activation following
treatment with thimerosal. They especially focused on the responses
of inositol 1,4,5-trisphosphate and ryanodine receptors (IP<sub>3</sub>R and RyR, respectively), which are known thimerosal targets. These gatekeepers
of intracellular calcium stores are essential for signaling activities
affecting dendritic cell function and maturation.</p><p>The team showed for the first time that both mature and immature dendritic
cells express isoforms of these receptors, IP<sub>3</sub>R1 and RyR1. Upon activation with the cellular energy source adenosine
triphosphate, immature control cells responded with a measurable rise
and fall in intracellular calcium concentration that involved RyR1 building
upon the initial IP<sub>3</sub>R1-controlled calcium release and afterward working with IP<sub>3</sub>R1 to bring calcium down to resting levels.</p><p>Exposure to thimerosal at concentrations as low as 20 ppb altered the time
course of these responses, however, and prolonged the length of time
that intracellular calcium levels remained elevated. One possible consequence
of these sustained calcium levels is a change in the rate and
timing of dendritic cells’ secretion of interleukin-6, a chemical
that triggers further immune system action. Exposure to thimerosal
at concentrations above 200 ppb caused immature dendritic cells to
die.</p><p>The continuing use of thimerosal in some vaccines and other products warrants
further investigation of possible immunotoxic effects of this compound
and its constituent ethylmercury. The researchers also note that
the human <italic>RyR1</italic> gene is highly polymorphic, an observation that raises several questions
about the role of RyR1 in the immune system’s genetic vulnerability
to mercury.</p>
|
Uncoupling of ATP-Mediated Calcium Signaling and Dysregulated Interleukin-6 Secretion
in Dendritic Cells by Nanomolar Thimerosal
|
<p>Dendritic cells (DCs), a rare cell type widely distributed in the soma, are
potent antigen-presenting cells that initiate primary immune responses. DCs
rely on intracellular redox state and calcium (Ca<sup>2+</sup>) signals for proper development and function, but the relationship between
these two signaling systems is unclear. Thimerosal (THI) is a mercurial
used to preserve vaccines and consumer products, and is used experimentally
to induce Ca<sup>2+</sup> release from microsomal stores. We tested adenosine triphosphate (ATP)-mediated
Ca<sup>2+</sup> responses of DCs transiently exposed to nanomolar THI. Transcriptional
and immunocytochemical analyses show that murine myeloid immature DCs (IDCs) and
mature DCs (MDCs) express inositol 1,4,5-trisphosphate receptor (IP<sub>3</sub>R) and ryanodine receptor (RyR) Ca<sup>2+</sup> channels, known targets of THI. IDCs express the RyR1 isoform in a punctate
distribution that is densest near plasma membranes and within dendritic
processes, whereas IP<sub>3</sub>Rs are more generally distributed. RyR1 positively and negatively regulates
purinergic signaling because ryanodine (Ry) blockade <italic>a</italic>) recruited 80% more ATP responders, <italic>b</italic>) shortened ATP-mediated Ca<sup>2+</sup> transients > 2-fold, and <italic>c</italic>) produced a delayed and persistent rise (≥ 2-fold) in baseline
Ca<sup>2+</sup>. THI (100 nM, 5 min) recruited more ATP responders, shortened the ATP-mediated
Ca<sup>2+</sup> transient (≥ 1.4-fold), and produced a delayed rise (≥ 3-fold) in
the Ca<sup>2+</sup> baseline, mimicking Ry. THI and Ry, in combination, produced additive
effects leading to uncoupling of IP<sub>3</sub>R and RyR1 signals. THI altered ATP-mediated interleukin-6 secretion, initially
enhancing the rate of cytokine secretion but suppressing cytokine
secretion overall in DCs. DCs are exquisitely sensitive to THI, with
one mechanism involving the uncoupling of positive and negative regulation
of Ca<sup>2+</sup> signals contributed by RyR1.</p>
|
<contrib contrib-type="author"><name><surname>Goth</surname><given-names>Samuel R.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001083">1</xref><xref ref-type="aff" rid="af2-ehp0114-001083">2</xref></contrib><contrib contrib-type="author"><name><surname>Chu</surname><given-names>Ruth A.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001083">2</xref></contrib><contrib contrib-type="author"><name><surname>Gregg</surname><given-names>Jeffrey P.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001083">1</xref><xref ref-type="aff" rid="af3-ehp0114-001083">3</xref><xref ref-type="aff" rid="af4-ehp0114-001083">4</xref></contrib><contrib contrib-type="author"><name><surname>Cherednichenko</surname><given-names>Gennady</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001083">2</xref></contrib><contrib contrib-type="author"><name><surname>Pessah</surname><given-names>Isaac N.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001083">1</xref><xref ref-type="aff" rid="af2-ehp0114-001083">2</xref><xref ref-type="aff" rid="af4-ehp0114-001083">4</xref></contrib>
|
Environmental Health Perspectives
|
<p>Recent animal and human studies have underscored the strong influence of
genetic, epigenetic, and physiologic factors in defining susceptibility
of the immune system to methylmercury (MeHg) and ethylmercury (EtHg) (<xref rid="b23-ehp0114-001083" ref-type="bibr">Havarinasab and Hultman 2005</xref>; <xref rid="b32-ehp0114-001083" ref-type="bibr">Lawler et al. 2004</xref>; <xref rid="b50-ehp0114-001083" ref-type="bibr">Silbergeld et al. 2005</xref>). Immune dysregulation triggered by organic mercury can include suppression, stimulation, loss
of tolerance, and generation of auto-antibodies. Therefore, the
pattern of immunotoxicity induced by organic mercury
is likely to depend not only on the chemical form, timing, and dose
to which an individual is exposed but also on susceptibility factors that
are poorly understood at present. Thus, significant attention is currently
focused on identifying which types of immune cells and biomolecules
are critical targets of low-level organic mercury and their functional
consequences on overall immune status.</p><p>Sodium ethylmercurithiosalicylate (thimerosal; THI) is an EtHg-containing
compound used to preserve cosmetics, blood products, and vaccines and
is also used experimentally to induce calcium (Ca<sup>2+</sup>) release from microsomal [endoplasmic reticulum/sarcoplasmic reticulum (ER/SR)] stores in intact cells. THI toxicity is due
to the EtHg moiety. THI and EtHg toxicity in humans consist of a few cases
of accidental high-dose poisoning (<xref rid="b9-ehp0114-001083" ref-type="bibr">Cinca et al. 1980</xref>; <xref rid="b11-ehp0114-001083" ref-type="bibr">Damluji 1962</xref>; <xref rid="b54-ehp0114-001083" ref-type="bibr">Zhang 1984</xref>). Attention has been focused on THI in vaccines, where it is used as a
preservative for multiuse formulations. THI was withdrawn from pediatric
vaccines starting in 1999 (<xref rid="b8-ehp0114-001083" ref-type="bibr">Centers for Disease Control and Prevention 1999</xref>) over concerns that organic mercury is a known neurodevelopmental toxicant. Nevertheless, THI
is still used in influenza, diphtheria toxoid, diphtheria
toxoid and acellular pertussis (DTaP), and tetanus toxoid
vaccines. The hypothesis that THI can cause neurodevelopmental disorders
was tested by injecting THI and THI-containing vaccines into inbred
strains of young mice (<xref rid="b26-ehp0114-001083" ref-type="bibr">Hornig et al. 2004</xref>). Growth, behavioral, and histologic abnormalities in the brains of the
autoimmune susceptible strain (SJL) were recorded after administration
of THI or THI plus vaccine. Autoimmune-resistant strains (C57BL/6, BALB/c) did
not display any of the abnormalities, suggesting a strong
influence of inherent immune status and the neurodevelopmental toxicity
of THI.</p><p>We hypothesized that especially sensitive targets of THI-mediated immune
dysregulation are dendritic cells (DCs), whose function is to acquire
antigens derived from self or nonself sources and efficiently present
them to naive and resting T cells (<xref rid="b4-ehp0114-001083" ref-type="bibr">Banchereau and Steinman 1998</xref>). This hypothesis stems from the fact that ambient oxygen (O<sub>2</sub>) tension or thiol concentration directly influences DC secretion of interferon-γ (IFN-γ) and interleukin-12 (IL-12) (<xref rid="b37-ehp0114-001083" ref-type="bibr">Murata et al. 2002</xref>), enhances expression of FcɛR1, the high affinity receptor for
IgE (<xref rid="b38-ehp0114-001083" ref-type="bibr">Novak et al. 2002</xref>), and regulates surface class II major histo-compatibility complex (MHC) expression (<xref rid="b20-ehp0114-001083" ref-type="bibr">Goth et al. 2006</xref>) <italic>in vitro</italic>. In this regard, Ca<sup>2+</sup> contributes essential signals for DC function and maturation. Differentiation (<xref rid="b3-ehp0114-001083" ref-type="bibr">Bagley et al. 2004</xref>), pro-inflammatory cytokine secretion (<xref rid="b18-ehp0114-001083" ref-type="bibr">Gardella et al. 2000</xref>), apoptotic cell phagocytosis (<xref rid="b43-ehp0114-001083" ref-type="bibr">Poggi et al. 1998</xref>), and migrational responsiveness to purine nucleotides or chemokines (<xref rid="b40-ehp0114-001083" ref-type="bibr">Partida-Sanchez et al. 2004</xref>; <xref rid="b46-ehp0114-001083" ref-type="bibr">Scandella et al. 2004</xref>) are Ca<sup>2 +</sup>-dependent processes. DCs rely on changes in intracellular redox state
and Ca<sup>2+</sup> signals for proper development and function, but the relationship between
these signaling systems in DCs is unclear.</p><p>THI contains an oxidized mercury atom (Hg<sup>2+</sup>) whose redox properties can enhance the activity of the inositol 1,4,5-trisphosphate
receptor (IP<sub>3</sub>R) and ryanodine receptor (RyRs), both intracellular Ca<sup>2+</sup> channels (<xref rid="b29-ehp0114-001083" ref-type="bibr">Kaplin et al. 1994</xref>; <xref rid="b41-ehp0114-001083" ref-type="bibr">Pessah et al. 2002</xref>). THI elicits Ca<sup>2+</sup> release from ER/SR stores in lymphocytes (<xref rid="b7-ehp0114-001083" ref-type="bibr">Bultynck et al. 2004</xref>) and ER/SR microsomes by targeting the IP<sub>3</sub>R and RyR (<xref rid="b1-ehp0114-001083" ref-type="bibr">Abramson et al. 1995</xref>; <xref rid="b29-ehp0114-001083" ref-type="bibr">Kaplin et al. 1994</xref>). THI-treated, monocyte-derived DCs failed to or only minimally phosphorylated
STAT (signal transducer and activator of transcription) proteins 1, 3, 4, and 6, implying that the JAK (janus kinase) signaling pathway
and, by extension, cytokine receptors are bypassed in the sensitization
phase induced by THI (<xref rid="b52-ehp0114-001083" ref-type="bibr">Valk et al. 2002</xref>). DCs express several classes of Ca<sup>2+</sup> channel proteins that mediate Ca<sup>2+</sup> signals. DCs express store-operated Ca<sup>2+</sup> channels (<xref rid="b27-ehp0114-001083" ref-type="bibr">Hsu et al. 2001</xref>) and IP<sub>3</sub>Rs that regulate release of Ca<sup>2+</sup> from ER/SR stores in response to adenosine triphosphate (ATP) (<xref rid="b48-ehp0114-001083" ref-type="bibr">Schnurr et al. 2003</xref>) and chemokines (<xref rid="b46-ehp0114-001083" ref-type="bibr">Scandella et al. 2004</xref>). Immature DCs (IDCs) express message for one of three genetic forms of
the RyR, the RyR1 (<xref rid="b27-ehp0114-001083" ref-type="bibr">Hsu et al. 2001</xref>; <xref rid="b39-ehp0114-001083" ref-type="bibr">O’Connell et al. 2002</xref>). The influence of THI and its metabolites EtHg and thiosalicylic acid (TSA) on
Ca<sup>2+</sup> signaling and activation of DCs remain unexplored.</p><p>To study how THI and EtHg influence Ca<sup>2+</sup>-dependent DC functions, we generated and tested murine DCs under normoxia (5% O<sub>2</sub> vol/vol) and omitted 2-mercaptoethanol (2-Me) from the culture medium. In
this article we report that DCs primarily express the type 1 isoform
of the IP<sub>3</sub>R and RyR ER/SR Ca<sup>2+</sup> channels, known targets of THI. THI and ryanodine (Ry) each block early
positive contributions of the RyR1 to ATP-induced Ca<sup>2+</sup> transients and uncouple inhibitory feedback, indicating a common mechanism. The
consequences of THI upon ATP-induced IL-6 production, a Ca<sup>2+</sup>-dependent process, were examined. THI initially enhanced the IL-6 secretion
rate, but ultimately suppressed its accumulation. DCs are exquisitely
sensitive to THI, with one prominent mechanism involving the uncoupling
of positive and negative regulation of Ca<sup>2+</sup> signals contributed by RyR1.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec><title>Chemicals and antibodies</title><p>THI (USP grade) and its metabolite TSA, propidium iodide (PI), diethylpyrocarbonate, and
Na<sub>2</sub>ATP were purchased from Sigma (St. Louis, MO). We purchased fibronectin (bovine
plasma) from Calbiochem (San Diego, CA) and ethyl-mercuric chloride (EtHgCl) from
ICN (Costa Mesa, CA). Recombinant murine granulocyte/macrophage
colony stimulating factor (GM-CSF) was purchased from Sigma
or R&D Systems (Minneapolis, MN), as were murine IL-6 ELISA kits. Antibodies (BD-Pharmingen, San Diego, CA) are as follows (clone name): class
II MHC-biotin (2G9), CD11c-APC (HL3), CD16/32 (2.4G2), and
hamster immunoglobulin. Anti-RyR monoclonal antibody 34C (recognizes
types 1 and 3) was purchased from the Developmental Studies Hybridoma
Bank (Iowa City, IA). Anti-IP<sub>3</sub>R and anti-IP<sub>3</sub>R1 polyclonal antibodies were purchased from Chemicon (Temecula, CA). Prolong
Antifade, Fura-2 AM, Fluo-4 AM, Alexa 488–conjugated goat
anti-mouse IgG antibody, Alexa 488–conjugated goat anti-rabbit
IgG antibody, and Alexa 647–conjugated goat anti-rabbit IgG
antibody were purchased from Invitrogen (Carlsbad, CA). 7-Aminoactinomycin
D (7AAD) was purchased from Calbiochem. A fluorescent terminal
deoxynucleotidyl transferase (TdT) labeling kit was purchased from Promega (Madison, WI).</p></sec><sec><title>Cell culture</title><p>Female C57BL/6J mice 6–8 weeks of age were purchased from JAX West, Inc. (Davis, CA), treated humanely and with regard for alleviation
of suffering, and euthanized in accordance with a protocol approved
by the University of California–Davis Animal Resources Service. Bone-marrow–derived DCs were generated by modifying a protocol (<xref rid="b35-ehp0114-001083" ref-type="bibr">Lutz et al. 1999</xref>), using normoxia (5% O<sub>2</sub> vol/vol), and omitting 50 μM 2-Me from the culture medium (<xref rid="b20-ehp0114-001083" ref-type="bibr">Goth et al. 2006</xref>). R10 medium was RPMI 1640 (Invitrogen) with 10% fetal bovine
serum (FBS; Hyclone, Logan, UT), 2 mM <sc>l</sc>-glutamine, 2 mM sodium pyruvate, 100 IU/mL penicillin, and 10 μg/mL
streptomycin. Cultures were maintained at 37°C in a Thermo
Forma model 3130 incubator (Thermo Forma, Marietta, OH) equipped with
a CO<sub>2</sub> and fuel-cell O<sub>2</sub> monitor and N<sub>2</sub> and CO<sub>2</sub> gas supplies. CO<sub>2</sub> was set to 5% vol/vol, and O<sub>2</sub> to 5% vol/vol, and were periodically verified using Fyrite gas
analyzers (Bacharach Inc., New Kensington, PA).</p></sec><sec sec-type="methods"><title>DC cytometric flow sorting and analysis</title><p>Cells were flow sorted between culture days 6 and 10. A detailed description
of our cell preparation for flow cytometry has been published (<xref rid="b20-ehp0114-001083" ref-type="bibr">Goth et al. 2006</xref>). Briefly, nonadherent cells were preblocked with 2.4G2 monoclonal antibody (1.0 μg/mL) and hamster IgG (0.25 μg/mL) for 10 min. Fluorescent
anti-class II MHC (0.1 μg/mL) and anti-CD11c (0.3 μg/mL) monoclonal antibodies were added and allowed to bind
for 15 min. After washing with 2% FBS in phosphate-buffered
saline (PBS), cells were aseptically sorted on a MoFlo cytometer (Cytometric, Fort
Collins, CO). Single-stained and unstained controls were
used to define sorting gates and to adjust compensation. CD11c-positive
cells were considered DCs and graded as IDCs or mature DCs (MDCs) depending
on their class II MHC expression. We routinely obtained ≥ 85% purities
of sorted IDC and MDC subsets. PI was added
to a final concentration of 0.5 μg/mL before sorting or before
analysis on a FACScan flow cytometric analyzer (Becton Dickinson, Palo
Alto, CA) to detect dead cells.</p></sec><sec><title>DC treatment</title><p>THI, EtHgCl, and TSA solutions were dissolved in sodium carbonate using
borosilicate glass pipettes and tubes. Dilutions were made in R10 and
used within 1 hr. DCs (1–2 × 10<sup>6</sup> cells/mL) were aliquoted into perfluorocarbon tissue culture vials or 96-well
format plates (Savillex, Minnetonka, MN). R10 or medium containing
THI, EtHgCl, TSA, or lipopolysaccharide (LPS) (to 1 μg/mL) was
added, and cells were placed in a 37°C incubator.</p></sec><sec sec-type="methods"><title>DC transcriptome analysis</title><p>Total RNA was isolated from sorted DCs using Trizol Reagent (Molecular
Research Center, Cincinnati, OH) according to the manufacturer’s
recommended procedure. DCs were resuspended to 1 × 106 cell/mL
in R10 media, plated in per-fluorocarbon containers, and incubated
for 20 hr. Biotinylated cRNA was synthesized from 5 μg of total
cellular RNA according to the protocol published by Affymetrix Inc. (<xref rid="b2-ehp0114-001083" ref-type="bibr">Affymetrix 2004</xref>). Fragmented, labeled cRNA was hybridized onto Affymetrix mouse 430A or 430 2.0 GeneChip
arrays. Microarrays were hybridized 16 hr at 45°C, stained, and
washed according to an Affymetrix protocol (EukGE-WS2v4; <xref rid="b2-ehp0114-001083" ref-type="bibr">Affymetrix 2004</xref>). Fluorescence intensity was measured with a scanner equipped with Affymetrix
Microarray Analysis Suite version 5.0. The average intensity for
each array was normalized by scaling to a target intensity value of 125, allowing
comparison between arrays. Individual transcripts are represented
by perfect-match probes in conjunction with a corresponding
set of mismatch probes. A transcript is called present if the average
intensity value of perfect-match cells is ≥ 1.5 times greater
than the average intensity of mismatch cells, and the average intensity
difference between perfect-match and mismatch cells is four or more
times the experimental noise. Poorly performing probes (where the ratio
between the average intensity of mismatch cells and perfect-match cells
is four or more times the experimental noise) were not included in
the analysis. RNA from three independent cultures was analyzed on GeneChips (i.e., three
GeneChips per treatment were analyzed).</p></sec><sec><title>Immunocytofluorescence of calcium channels</title><p>DCs were washed in 1% bovine serum albumin (BSA) in PBS, centrifuged
onto glass slides using a Cytofuge2 (Statspin, Norwood, MA), air
dried, fixed with 4% paraformaldehyde in PBS for 20 min at 4°C, and
then permeabilized with three washes of 0.2% Tween-20 in
PBS (TPBS). Nonspecific binding was blocked with goat IgG (50 μg/mL). Cells
were incubated with primary antibody (dilutions
were 1:20 34C and 1:100 anti-IP<sub>3</sub>R and IP<sub>3</sub>R1 polyclonal antibodies in TPBS) for 1 hr. Blocking peptide for the IP<sub>3</sub>R1 antibody was used at the manufacturer’s suggested concentration. After
washing, Alexa 488– or Alexa 647–conjugated
anti-mouse and anti-rabbit secondary antibodies were diluted 1:1,000 in
TPBS and allowed to bind 1 hr. Cells were washed with TPBS and then
with PBS. After mounting with Prolong Antifade plus 40 μg/mL 7AAD, DCs
were visualized for immunofluorescence using an MRC 600 laser
scanning confocal microscope (Bio-Rad, Richmond, CA). Confocal immunopheno-typing
was performed on two separate cultures.</p></sec><sec><title>TdT assay</title><p>IDCs were treated with 500 nM TSA, THI, or medium for 20 hr as described
above. Cells were then washed twice with ice-cold PBS and fixed with 4% paraformaldehyde in PBS for 20 min on ice. Cells were washed
in 1% BSA in PBS and resuspended to 0.5 × 106 cells/mL
in 1% BSA in PBS; cell aliquots were then spun onto microscope
slides. Slides were air dried at least 2 hr and then immersed in 4% diethylpyrocarbonate:ethanol prechilled to –20°C
for 30 min to stop endogenous nuclease activity. After washing twice
with PBS, cells were processed for DNA strand end-labeling according
to the protocol supplied by the kit manufacturer (Promega). After
TdT labeling, nuclei were counterstained with 1 μg/mL PI in water
for 15 min before mounting in Prolong Antifade for confocal imaging.</p></sec><sec sec-type="methods"><title>[<sup>3</sup>H]Ryanodine binding analysis</title><p>High-affinity binding of [<sup>3</sup>H]ryanodine ([<sup>3</sup>H]Ry; 56 or 50 Ci/mmol; PerkinElmer, Boston, MA) to rabbit skeletal
microsomes enriched in RyR1 was performed as previously described (<xref rid="b42-ehp0114-001083" ref-type="bibr">Pessah et al. 1987</xref>). Nonspecific binding was determined by including 1,000-fold unlabeled
Ry. Data were reported in picomoles of bound Ry per milligram of protein.</p></sec><sec><title>IL-6 assays</title><p>IDCs were pulsed with 100 nM THI or TSA for 20 min, pelleted, supernatant
aspirated, and resuspended to 2 × 10<sup>7</sup>/mL; 0.05 mL of the cell suspension was aliquoted per well into a perfluorocarbon 96-well
plate, and 0.05 mL ATP (0, 0.2, 2, or 20 μM
final) was added per well. Medium and LPS (1 μg/mL final) cells
received no pre-treatment. Supernatants for IL-6 ELISA were collected 20 hr
later from the top portion of the cultures. IL-6 concentrations
were interpolated from the linear response range of the cytokine standard; minimum
sensitivity was 7 pg/mL.</p></sec><sec><title>Calcium imaging</title><p>IDCs were resuspended in R10 supplemented with 5 ng/mL recombinant murine
GM-CSF to 0.5 × 10<sup>6</sup>/mL; 0.5 mL was plated overnight onto fibronectin-coated glass coverslips. The
next day, cells were labeled with 5 μM Fura-2 AM or Fluo-4 AM
for 20 min. Cells were washed with bathing solution (130 mM NaCl, 4 mM
KCl, 10 mM HEPES, 10 mM glucose, 2 mM CaCl<sub>2</sub>, 2 mM MgSO<sub>4</sub>, pH 7.3, with NaOH) and imaged within 1 hr. Changes in cytoplasmic Ca<sup>2+</sup> were measured by emission at 510 nm (Fluo-4) or ratioing emission at 510 nm
with excitation pair 340/380 nm (Fura-2) (<xref rid="b17-ehp0114-001083" ref-type="bibr">Fessenden et al. 2003</xref>). Rapid perfusion of ATP and caffeine was accomplished by a micropipette
above the cells being imaged (Automate, Oakland, CA).</p></sec><sec sec-type="methods"><title>Data analysis</title><p>Nonlinear curve fitting analysis and one-way analysis of variance were
performed using Origin 6.0 (OriginLab Corporation, Northampton, MA) software
to test for statistical significance.</p></sec></sec><sec sec-type="results"><title>Results</title><p>Culturing murine bone marrow at physiologic O<sub>2</sub> tension (5% vol/vol) without 2-Me supplementation generated myeloid
IDCs and MDCs with a similar yield, leukocyte marker expression, and
allostimulatory capacity as DCs produced from cultures using ambient (20%) O<sub>2</sub> and 50 μM 2-Me (<xref rid="b20-ehp0114-001083" ref-type="bibr">Goth et al. 2006</xref>). A representative flow cytometric dot plot showing sorting gates for
IDCs and MDCs is shown in <xref ref-type="supplementary-material" rid="SD1">Supplemental Material</xref>, <xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1A</xref> (available online at <ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8881/suppl.pdf">http://www.ehponline.org/docs/2006/8881/suppl.pdf</ext-link>). RNAs extracted from sorted IDCs and MDCs were analyzed using gene probe
arrays. IDCs and MDCs have a common lineage, and a small number of
changes in gene expression occurred during maturation. Of this limited
set of genes, class II MHC mRNA expression is down-regulated with maturation, whereas
CD86 message was up-regulated with maturation (<xref ref-type="table" rid="t1-ehp0114-001083">Table 1</xref>). Glyceraldehyde-3-phosphate dehydrogenase (GADPH) was expressed at similar
levels in IDCs and MDCs (<xref ref-type="table" rid="t1-ehp0114-001083">Table 1</xref>). Both DC subsets express two Ca<sup>2+</sup> channel types that are targets of THI. Of the three IP<sub>3</sub>R isoforms, IDCs and MDCs express message for ITPR1 and ITPR3, the latter
down-regulated with maturation (<xref ref-type="table" rid="t1-ehp0114-001083">Table 1</xref>). Of the three RyR isoforms, RyR1 mRNA is present in both DC subsets, RyR3 is
expressed upon maturation, and RyR2 is not expressed (<xref ref-type="table" rid="t1-ehp0114-001083">Table 1</xref>). Because the Ca<sup>2+</sup> channel genes detected encode targets of THI, we explored the distribution
of the proteins in IDCs using confocal microscopy.</p><p>IDCs express IP<sub>3</sub>R1 in a dense granular distribution in a pattern consistent with targeting
to ER membranes (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1A,E</xref>). In contrast, intense RyR1 staining localizes near the plasma membrane, and
foci of protein extend from the base into dendrites (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1C,E,F</xref>). IDCs lack detectable RyR1 within perinuclear regions (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1C,E,F</xref>). Cells stained with secondary antibody alone or with antibody-blocking
peptide (IP<sub>3</sub>R1) had no detectable signal (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1B,D</xref>). The codistribution of RyR1 and IP<sub>3</sub>R was further examined using an anti-RyR1 monoclonal antibody and a pan
anti-IP<sub>3</sub>R polyclonal antibody. The RyR1 is densely localized within a narrow band
at the cell periphery and extends within dendrites (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1E,F</xref>), whereas IP<sub>3</sub>R extends to regions lacking RyR1 (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1E</xref>). The granular appearance and distribution of IP<sub>3</sub>R protein in IDCs did not change when visualized with an isoform specific (IP<sub>3</sub>R1) or pan-IP<sub>3</sub>R antibody (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1A,E</xref>). The distribution of RyR proteins in IDCs and MDCs (using a monoclonal
antibody specific for RyR1 and 3) were similar (<xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1G,H</xref>) despite the additional expression of RyR3 transcripts in MDC (<xref ref-type="table" rid="t1-ehp0114-001083">Table 1</xref>).</p><p>Ca<sup>2+</sup> responses of IDCs to the RyR agonist caffeine were tested. Given the expression
of RyR1 within IDCs, and that MDCs respond to caffeine (<xref rid="b48-ehp0114-001083" ref-type="bibr">Schnurr et al. 2003</xref>), we expected IDCs would respond to caffeine with a rise in cytoplasmic
Ca<sup>2+</sup>. Approximately 50% of IDCs responded to 20 mM caffeine (<xref ref-type="fig" rid="f2-ehp0114-001083">Figure 2A</xref>). Because extracellular ATP induces maturity in IDCs (<xref rid="b30-ehp0114-001083" ref-type="bibr">la Sala et al. 2001</xref>, <xref rid="b31-ehp0114-001083" ref-type="bibr">2002</xref>; <xref rid="b47-ehp0114-001083" ref-type="bibr">Schnurr et al. 2001</xref>) and intracellular Ca<sup>2+</sup> contributes an essential DC maturation signal (<xref rid="b3-ehp0114-001083" ref-type="bibr">Bagley et al. 2004</xref>), we decided to test the ability of our IDCs to respond to extra-cellular
ATP. IDCs responded vigorously to a brief (5 sec) application of ATP, the
response amplitude dose-dependent between 0.2 and 20 μM (<xref ref-type="fig" rid="f2-ehp0114-001083">Figure 2B</xref>). Therefore, ATP potently elicits Ca<sup>2 +</sup> transients, likely mediated through agonist actions on P2Y purinergic
nucleotide receptors expressed in IDCs (<xref rid="b28-ehp0114-001083" ref-type="bibr">Idzko et al. 2002</xref>). We next generated viability–dose survival curves for DCs exposed
to THI and its metabolites EtHg (from EtHgCl) and TSA [chemical
structures are shown in <xref ref-type="supplementary-material" rid="SD1">Supplemental Material</xref>, <xref ref-type="fig" rid="f1-ehp0114-001083">Figure 1B</xref> (available online at <ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8881/suppl.pdf">http://www.ehponline.org/docs/2006/8881/suppl.pdf</ext-link>)].</p><p>Range-finding experiments determined that a 20 hr exposure to 10 μM
THI consistently killed > 90% of DCs using flow cytometry, judged
by their increased permeability to PI and decreased cell size [forward
light scatter (FSC); data not shown]. THI, EtHgCl, and
TSA were titrated from 50 nM to 10 μM, IDC and MDC
subsets were treated for 20 hr, and cell PI permeability was measured
by flow cytometry. <xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3A and B</xref> shows that THI and EtHgCl caused dose-dependent decreases in DC viability
compared with the TSA control. The viability dose–response
curves for THI and EtHgCl were the same within each DC subset and were
similar between the cell subsets. IC<sub>50</sub> (concentration that inhibits 50%) values for the DC subsets treated
with THI, TSA, and EtHgCl are shown in <xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3A and B</xref>. DC death triggered by THI could be mediated by apoptosis or primary necrosis, because
the FSC and PI uptake data acquired 20 hr posttreatment
cannot distinguish between the two possibilities. IDCs were treated
with medium alone, 500 nM TSA, or 500 nM THI for 20 hr and probed for
DNA strand breakage, an indicator of apoptosis (<xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3C,D,E</xref>, respectively). Only THI-treated cells demonstrate both increased 2′-deoxyuridine 5′-triphosphate-fluorescein isothiocyanate
labeling and a corresponding shrinkage in nuclear size (<xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3E</xref>); death from THI exposure is therefore likely to be an apoptotic outcome. The
action of THI and EtHgCl upon RyR1 channels was further explored
by measuring [<sup>3</sup>H]Ry binding to ER/SR membranes enriched in RyR1. THI and EtHgCl
inhibited the specific binding of [<sup>3</sup>H]Ry to RyR1 (<xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3F</xref>; IC<sub>50</sub> = 562 and 501 nM).</p><p>Lymphocyte tolerance or immunity to an antigen can be driven by IDCs or
MDCs, respectively (<xref rid="b51-ehp0114-001083" ref-type="bibr">Steinman et al. 2005</xref>), and we focused on characterizing the effects of THI upon ATP-mediated
Ca<sup>2+</sup> signaling (a maturation signal) in IDCs. THI can release Ca<sup>2+</sup> from ER/SR stores by selectively enhancing the activity of IP<sub>3</sub>R and RyR, and we examined its actions on ATP-mediated Ca<sup>2+</sup> signaling in IDC. ATP is an efficacious activator of DCs through its agonist
actions on P2Y purinergic nucleotide receptors (<xref rid="b28-ehp0114-001083" ref-type="bibr">Idzko et al. 2002</xref>). P2Y receptors couple to Gαq protein that initiates signal transduction
events leading to the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP<sub>2</sub>) to IP<sub>3</sub> and diacylglycerol. IP<sub>3</sub> in turn activates IP<sub>3</sub>R that mobilizes Ca<sup>2+</sup> from ER stores (<xref rid="b12-ehp0114-001083" ref-type="bibr">Di Virgilio et al. 2001a</xref>, <xref rid="b13-ehp0114-001083" ref-type="bibr">2001b</xref>). IDCs challenged with two pulses of 20 μM ATP (5 sec each) 30 min
apart (with constant perfusion) produced Ca<sup>2+</sup> transients whose peak heights, baseline to peak rates, or peak to baseline
decays were not significantly different (<xref ref-type="fig" rid="f4-ehp0114-001083">Figure 4A</xref>). This indicates that the concentration, duration, and frequency of the
ATP challenges did not desensitize the cell’s capacity to respond, nor
were they additive.</p><p>Perfusion of 50 nM THI after an initial ATP test pulse (<xref ref-type="fig" rid="f4-ehp0114-001083">Figure 4B</xref>) did not induce a rise in baseline Ca<sup>2+</sup> in the 30 min of perfusion that preceded the second ATP challenge. In
the presence of THI, the second ATP pulse produced a baseline to Ca<sup>2+</sup> peak height comparable with that elicited by the first ATP challenge. However, the
rate of decay of the response peak was significantly slowed
by THI compared with control cells. THI also induced a sustained elevation
in the 8 min of trace recording in intracellular Ca<sup>2+</sup> after ATP withdrawal. DCs exposed to 100 nM THI (<xref ref-type="fig" rid="f4-ehp0114-001083">Figure 4C</xref>) showed a gradual rise in resting Ca<sup>2+</sup>. Although cells responded to a second ATP challenge, decay of the response
was significantly slower and incomplete compared with DCs exposed
to 50 nM THI. After withdrawal of ATP from the perfusion medium, a more
pronounced slow rise of intracellular Ca<sup>2+</sup> concentration was seen and remained elevated for the duration of the measurement. Some
DCs tested with 20 μM ATP failed to show detectable
responses. However these “ATP-resistant” cells
invariably responded vigorously to ATP after exposure to 50 nM THI for 30 min (data
not shown). We next dissected the ATP-mediated calcium wave
in DCs using pharmacologic agents known to modify components of Ca<sup>2+</sup> signaling.</p><p>ATP generated a stereotyped Ca<sup>2+</sup> transient having a time to peak of 16.1 ± 2.6 sec and a decay
time of 106 ± 2.6 sec (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, top trace). To test the RyR1’s contribution to ATP-mediated Ca<sup>2+</sup> transients, IDCs were pre-treated 4 hr with 100 μM Ry, which irreversibly
locks the RyR in a nonconducting conformation (<xref rid="b6-ehp0114-001083" ref-type="bibr">Buck et al. 1992</xref>; <xref rid="b55-ehp0114-001083" ref-type="bibr">Zimanyi et al. 1992</xref>). ATP-challenged Ry-treated cells produced transients whose time to peak
did not significantly differ from control (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, second trace; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5B</xref>). However, after ATP withdrawal, the recovery time to baseline was > 2-fold (<italic>p</italic> < 0.01) faster (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, second trace; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5C</xref>) and showed a delayed and persistent rise (<italic>p</italic> = 0.036) in baseline Ca<sup>2+</sup> after triggering (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, second trace; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5D</xref>). A 5 min THI (100 nM) pre-exposure mimicked the effects of Ry. When challenged
with ATP, the rise time remained unchanged (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5B</xref>), but the Ca<sup>2+</sup> transient was shortened > 1.5-fold (<italic>p</italic> = 0.05; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5C</xref>) and the delayed rise in Ca<sup>2+</sup> baseline was prominent (<italic>p</italic> = 0.02) (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, third trace; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5D</xref>).</p><p>The effects of THI and Ry in combination were nearly additive (<xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5A</xref>, fourth trace; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5C,D</xref>), suggesting a common mechanism targeting RyR1 function. THI and Ry alone
or combined increased the number of IDCs responding to ATP 1.5- to 1.8-fold
compared with control (<italic>p</italic> < 0.05; <xref ref-type="fig" rid="f5-ehp0114-001083">Figure 5E</xref>). THI’s actions on IDCs Ca<sup>2+</sup> signaling were not seen with TSA; therefore, they were mediated by organic
mercury (data not shown). These results (<xref ref-type="fig" rid="f4-ehp0114-001083">Figures 4</xref>, <xref ref-type="fig" rid="f5-ehp0114-001083">5</xref>) indicate that as little as 50–100 nM THI in a short time span (5–30 min) potentiates agonist-mediated Ca<sup>2+</sup> signaling events in DCs that primarily manifest as a prolonged elevation
in intracellular Ca<sup>2+</sup>. DCs are arguably the most sensitive target cell for THI identified to
date, and this sensitivity to oxidative insult may reflect a unique manner
in which they generate and use Ca<sup>2+</sup> signals in response to a changing redox environment.</p><p>Increased intracellular Ca<sup>2+</sup> induced by inhibitors of the ER/SR Ca<sup>2+</sup>-ATPase is associated with the rapid secretion of macrophage IL-6 (<xref rid="b5-ehp0114-001083" ref-type="bibr">Bost and Mason 1995</xref>). DCs produce IL-6 in response to IL-1β, tumor necrosis factor-α, or
LPS, and other myeloid cells secrete IL-6 in response to ATP (<xref rid="b49-ehp0114-001083" ref-type="bibr">Shigemoto-Mogami et al. 2001</xref>). We hypothesized that THI-induced uncoupling of ATP-mediated Ca<sup>2+</sup> signaling would disrupt IL-6 secretion. IDCs were pretreated with 100 nM
THI or TSA and challenged with graded concentrations of ATP, and secreted
IL-6 was measured. <xref ref-type="fig" rid="f6-ehp0114-001083">Figure 6A</xref> shows that IDCs pretreated with THI or TSA alone secreted low levels of
IL-6 that did not significantly differ from the medium control. LPS, which
induces IL-6 synthesis by non-Ca<sup>2+</sup>-dependent pathways, induced a large increase in IL-6. All three ATP concentrations
induced comparable amounts of IL-6 by the TSA-pretreated
DCs. By 20 hr, these IL-6 concentrations were equal to the LPS-treated
control. Pretreating IDCs with THI and challenging with ATP attenuated
IL-6 secretion compared with TSA-pretreated controls, reaching significance
at the lowest (0.2 μM) ATP dose. This lowered IL-6 secretion
was not due to cell death because this was overcome by 2 and 20 μM
ATP. Next, we determined the kinetics of ATP-mediated IL-6 secretion
in THI-treated DCs to see if cytokine production was attenuated
at earlier time points.</p><p><xref ref-type="fig" rid="f6-ehp0114-001083">Figure 6B</xref> shows that 2 μM ATP induced IL-6 by 4 hr, consistent with its
rapid induction in other myeloid cells. THI pretreatment accelerated IL-6 secretion
compared with TSA controls, indicating that THI sensitized
DCs to ATP. Maximal IL-6 was secreted by THI-treated DCs by 8 hr, whereas
controls needed an additional 12 hr. A concentration of 20 μM
ATP was more potent than a 2 μM concentration, eliciting
IL-6 by 2 hr and near-maximal levels at 8 hr in both THI- and TSA-pretreated
control IDCs (<xref ref-type="fig" rid="f6-ehp0114-001083">Figure 6C</xref>). The strong accelerating effect of THI versus TSA on IL-6 secretion induced
by 2 μM ATP was not as pronounced with 20 μM ATP, generating
a small but significant increase 4 hr after its application (<xref ref-type="fig" rid="f6-ehp0114-001083">Figure 6C</xref>).</p></sec><sec sec-type="discussion"><title>Discussion</title><p>DC activation and associated immune functions are subject to regulation
by their redox environment (<xref rid="b3-ehp0114-001083" ref-type="bibr">Bagley et al. 2004</xref>; <xref rid="b18-ehp0114-001083" ref-type="bibr">Gardella et al. 2000</xref>; <xref rid="b20-ehp0114-001083" ref-type="bibr">Goth et al. 2006</xref>; <xref rid="b37-ehp0114-001083" ref-type="bibr">Murata et al. 2002</xref>; <xref rid="b38-ehp0114-001083" ref-type="bibr">Novak et al. 2002</xref>). We generated DCs under tightly regulated O<sub>2</sub> and without 2-Me to provide a more physiologic baseline to study the mechanism
of redox active environmental triggers such as THI and EtHg in
regulating DC activation <italic>in vitro</italic>. For DCs, Ca<sup>2+</sup> signaling events provide an essential “upstream” component, engaging
immediate events such as cytokine production and secretion (<xref rid="b16-ehp0114-001083" ref-type="bibr">Ferrari et al. 2000</xref>; <xref rid="b31-ehp0114-001083" ref-type="bibr">la Sala et al. 2002</xref>) and long-term (e.g., maturational) responses. Importantly, microsomal
IP<sub>3</sub>R and RyR Ca<sup>2+</sup> channel functions are tightly regulated by changes in local redox state (<xref rid="b7-ehp0114-001083" ref-type="bibr">Bultynck et al. 2004</xref>; <xref rid="b29-ehp0114-001083" ref-type="bibr">Kaplin et al. 1994</xref>; <xref rid="b41-ehp0114-001083" ref-type="bibr">Pessah et al. 2002</xref>).</p><p>We show for the first time the expression and distribution of two major
Ca<sup>2+</sup> channel proteins expressed in DCs. We have provided transcriptomal and
direct immunocytochemical evidence that DCs express specific isoforms
of the IP<sub>3</sub>R and RyR Ca2+ channels. RyR and IP<sub>3</sub>R proteins distinctly distribute in DC subsets. In IDCs, RyR1 and IP<sub>3</sub>R1 localize below the plasma membrane at the base of the dendrites and
into the processes. A similar distribution of the RyR was seen in MDCs. However
IP<sub>3</sub>Rs extend to reticular regions where the RyR1 is either absent or at very
low levels. In human monocytes, IP<sub>3</sub>R and RyR localization patterns similar to our murine DCs were found (<xref rid="b10-ehp0114-001083" ref-type="bibr">Clark and Petty 2005</xref>).</p><p>ATP-triggered Ca<sup>2+</sup> transients in IDCs have three components that engage cross-talk between
IP<sub>3</sub>Rs and RyRs. The initial rate and amplitude of the Ca<sup>2+</sup> transient (phase 1; <xref ref-type="fig" rid="f7-ehp0114-001083">Figure 7</xref>) is largely dependent upon IP<sub>3</sub>R activation. It is unlikely the RyR1 has a significant influence on phase 1 because
blocking RyR1 channels with Ry has no measurable consequence
on these parameters. The transient decay rate depends on both IP<sub>3</sub>R and RyR1 because blocking RyR1 channels with Ry significantly enhances
this rate back to baseline (phase 2; <xref ref-type="fig" rid="f7-ehp0114-001083">Figure 7</xref>). Ca<sup>2+</sup>-induced Ca<sup>2+</sup> release, presumably mediated by activation of RyR1, must therefore contribute
to slowing the decay rate.</p><p>However, RyR1 activity in IDCs also contributes to the negative regulation, reestablishing
a stable resting Ca<sup>2+</sup> level near that before activation of purinergic signaling (phase 3; <xref ref-type="fig" rid="f7-ehp0114-001083">Figure 7</xref>). Phase 3 was unmasked when Ry or THI modified RyR1 conformation. Phases 2 and 3 are
coupled events that rely on the conformational state of
the RyR1. In this regard, both micromolar Ry and nanomolar THI uncouple
functional cross-talk between IP<sub>3</sub>R and RyR1 channels normally regulating purinergic signaling in IDCs.</p><p>Collectively, these data show that RyR1 channels are closely coupled to
IP<sub>3</sub>-induced Ca<sup>2+</sup> release and contribute to the temporal properties of the transient by
prolonging the decay and restoring the original resting Ca<sup>2+</sup> level. RyR1 therefore contributes both positive and negative regulation
to purinergic signaling in IDCs. THI, like Ry, appears to uncouple RyR1 functions
from phosphoinositide signaling in IDCs in a concentration- and
time-dependent manner.</p><p>Nanomolar THI deregulated ATP-mediated signaling by a mechanism that uncoupled
phases 2 and 3 of the Ca<sup>2+</sup> transient. THI did not appear to influence the initial response to ATP
by activation of the IP<sub>3</sub>R (phase 1) but enhanced the transient decay rate after agonist withdrawal (phase 2) and
elicited a persistent rise in intracellular Ca<sup>2+</sup> (phase 3). THI’s actions on IP<sub>3</sub>R-mediated signals have not been previously explored in DCs. However, evidence
indicates that IP<sub>3</sub>R1 is a target of THI. Using triple-IP<sub>3</sub>R-knockout R23-11 cells derived from DT40 chicken B lymphoma cells, THI (1–100 nM) potentiated IP<sub>3</sub>-induced Ca2+ release when IP<sub>3</sub>R1, but not IP<sub>3</sub>R3, is expressed (<xref rid="b7-ehp0114-001083" ref-type="bibr">Bultynck et al. 2004</xref>).</p><p>RyR1 is also a sensitive target of THI (<xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3E</xref>) (<xref rid="b1-ehp0114-001083" ref-type="bibr">Abramson et al. 1995</xref>). We show that ATP triggers a Ca<sup>2+</sup> transient in IDCs whose temporal property relies on functional coupling
of IP<sub>3</sub>R and RyR1 channels that is extremely sensitive to THI (≤ 100 nM). How
THI sensitizes RyR1 to activation by Ca<sup>2+</sup> involves release of the inhibitory actions of Mg<sup>2+</sup>, an important physiologic modulator (<xref rid="b15-ehp0114-001083" ref-type="bibr">Donoso et al. 2000</xref>; <xref rid="b44-ehp0114-001083" ref-type="bibr">Sanchez et al. 2003</xref>) and may be mediated by hyperreactive cysteines within the RyR1 channel
complex (<xref rid="b34-ehp0114-001083" ref-type="bibr">Liu and Pessah 1994</xref>; <xref rid="b53-ehp0114-001083" ref-type="bibr">Voss et al. 2004</xref>). Exactly how THI alters ATP-dependent and -independent Ca<sup>2+</sup> signals may involve multiple molecular mechanisms involving the IP<sub>3</sub>R1, RyR1, and possibly other channels. Nevertheless, the present results
reveal that IDCs are a particularly sensitive target of THI, with as
little as 50 nM within 30 min uncoupling ATP-induced Ca<sup>2+</sup> transients.</p><p>DC sensitivity to their oxidative environment may reflect how they generate
and use Ca<sup>2+</sup> signals in response to a changing redox environment. Redox modulation
of DC function is underscored by our finding that THI modulates IL-6 synthesis
elicited by exogenous ATP. Myeloid DC IL-6 strongly influences
mucosal T-cell and gut B-cell responses. Lung DCs with intrinsic T<sub>H</sub>2-polarizing activities generated T<sub>H</sub>1 responses from naive CD4+ T cells in the presence of anti–IL-6 neutralizing
antibody (<xref rid="b14-ehp0114-001083" ref-type="bibr">Dodge et al. 2003</xref>). Peyer’s patch B cells were induced to secrete IgA by IL-6 elaborated
by local CD11b-positive DCs; IgA induction was reduced by anti–IL-6 antibodies (<xref rid="b45-ehp0114-001083" ref-type="bibr">Sato et al. 2003</xref>).</p><p><italic>In vivo</italic> ATP steady-state levels are at nanomolar and low micromolar (1–25 μM) concentrations in bulk fluids and at the cell surface, respectively (<xref rid="b33-ehp0114-001083" ref-type="bibr">Lazarowski et al. 2003</xref>). At these concentrations, metabotropic G-protein–coupled P2Y
receptors are engaged. In THI-treated DCs, IL-6 production kinetics are
enhanced by 2 μM and 20 μM ATP, and IL-6 secretion is
significantly suppressed by 0.2 μM ATP (<xref ref-type="fig" rid="f6-ehp0114-001083">Figure 6A</xref>). THI-enhanced IL-6 secretion may be a consequence of prolonged calcium
signals resulting from the uncoupling effects of THI toward IP<sub>3</sub>R- and RyR1-mediated signals. Increased intracellu-lar Ca<sup>2+</sup> is associated with IL-6 RNA stabilization and rapid IL-6 secretion (<xref rid="b5-ehp0114-001083" ref-type="bibr">Bost and Mason 1995</xref>). THI-mediated attenuation of IL-6 secretion at 0.2 μM ATP was
not unexpected because the suboptimal calcium signals induced by 0.2 μM
ATP may not have been sufficient to promote maximal rates of
IL-6 synthesis. Ca<sup>2+</sup>-stimulated RyR1 channels are initially activated and then inactivated
by THI (<xref ref-type="fig" rid="f3-ehp0114-001083">Figure 3F</xref>) (<xref rid="b1-ehp0114-001083" ref-type="bibr">Abramson et al. 1995</xref>). Over the 20 hr assay used in this study, the uncoupling of IP<sub>3</sub>R–RyR1 functions is likely to have broad effects upon Ca<sup>2+</sup>-dependent processes.</p><p>A practical implication of the present findings has relevance to the commercial
uses of THI as an antimicrobial agent in vaccines and consumer
products because they identify DCs as sensitive targets for THI- and
EtHg-mediated dysfunction. Given the importance of DCs as a front line
in regulating lymphocyte-mediated immunity and tolerance, altering DC
functions by forms of EtHg should be considered when assessing contributions
to altered immune function. In studies using the autoimmune-susceptible
A.SW (H-2<sup>s</sup>) mouse strain, THI induces a syndrome that is stronger and more generally
manifested than those produced by methylmercury (<xref rid="b25-ehp0114-001083" ref-type="bibr">Havarinasab et al. 2004</xref>), and development of autoimmunity in H-2<sup>s</sup> mice is dependent on cellular (T cell) and soluble (IFN-γ and
IL-6) factors (<xref rid="b22-ehp0114-001083" ref-type="bibr">Havarinasab et al. 2005</xref>). Onset of spontaneous systemic autoimmune disease symptoms (proteinuria, anti-DNA
antibodies) in (NZBxNZW)F<sub>1</sub> mice is hastened by THI (<xref rid="b24-ehp0114-001083" ref-type="bibr">Havarinasab and Hultman 2006</xref>). Interestingly, disease morbidity and mortality in these F<sub>1</sub> mice are dramatically reduced by neutralizing anti–IL-6 receptor
antibody (<xref rid="b36-ehp0114-001083" ref-type="bibr">Mihara et al. 1998</xref>), an effect associated with reduced IgG (auto-) antibody production.</p><p>The human <italic>RyR1</italic> gene is highly polymorphic. More than 60 single missense or deletional
mutations have been closely linked to the pharmacogenetic disorders malignant
hyperthermia and central core disease (<xref rid="b21-ehp0114-001083" ref-type="bibr">Gronert et al. 2004</xref>). Our findings that DCs primarily express the RyR1 channel complex and
that this complex is uncoupled by very low levels of THI with dysregulated
IL-6 secretion raise intriguing questions about a molecular basis
for immune dyregulation and the possible role of the RyR1 complex in
genetic susceptibility of the immune system to mercury.</p></sec>
|
Satratoxin G from the Black Mold <italic>Stachybotrys chartarum</italic> Evokes Olfactory Sensory Neuron Loss and Inflammation in the Murine Nose
and Brain
|
<p>Satratoxin G (SG) is a macrocyclic trichothecene mycotoxin produced by <italic>Stachybotrys chartarum</italic>, the “black mold” suggested to contribute etiologically
to illnesses associated with water-damaged buildings. Using an intranasal
instillation model in mice, we found that acute SG exposure specifically
induced apoptosis of olfactory sensory neurons (OSNs) in the
olfactory epithelium. Dose–response analysis revealed that the
no-effect and lowest-effect levels at 24 hr postinstillation (PI) were 5 and 25 μg/kg
body weight (bw) SG, respectively, with severity
increasing with dose. Apoptosis of OSNs was identified using immunohistochemistry
for caspase-3 expression, electron microscopy for ultrastructural
cellular morphology, and real-time polymerase chain reaction
for elevated expression of the proapoptotic genes <italic>Fas</italic>, <italic>FasL</italic>, <italic>p75NGFR</italic>, <italic>p53</italic>, <italic>Bax</italic>, caspase-3, and <italic>CAD</italic>. Time-course studies with a single instillation of SG (500 μg/kg
bw) indicated that maximum atrophy of the olfactory epithelium occurred
at 3 days PI. Exposure to lower doses (100 μg/kg bw) for 5 consecutive
days resulted in similar atrophy and apoptosis, suggesting
that in the short term, these effects are cumulative. SG also induced
an acute, neutrophilic rhinitis as early as 24 hr PI. Elevated mRNA
expression for the proinflammatory cytokines tumor necrosis factor-α, interleukin-6 (IL-6), and IL-1 and the chemokine macrophage-inflammatory
protein-2 (MIP-2) were detected at 24 hr PI in both the ethmoid
turbinates of the nasal airways and the adjacent olfactory bulb of
the brain. Marked atrophy of the olfactory nerve and glomerular layers
of the olfactory bulb was also detectable by 7 days PI along with mild
neutrophilic encephalitis. These findings suggest that neurotoxicity
and inflammation within the nose and brain are potential adverse health
effects of exposure to satratoxins and <italic>Stachybotrys</italic> in the indoor air of water-damaged buildings.</p>
|
<contrib contrib-type="author"><name><surname>Islam</surname><given-names>Zahidul</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001099">1</xref><xref ref-type="aff" rid="af2-ehp0114-001099">2</xref><xref ref-type="aff" rid="af3-ehp0114-001099">3</xref></contrib><contrib contrib-type="author"><name><surname>Harkema</surname><given-names>Jack R.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001099">1</xref><xref ref-type="aff" rid="af4-ehp0114-001099">4</xref></contrib><contrib contrib-type="author"><name><surname>Pestka</surname><given-names>James J.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001099">1</xref><xref ref-type="aff" rid="af2-ehp0114-001099">2</xref><xref ref-type="aff" rid="af3-ehp0114-001099">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>Numerous adverse human health effects have been attributed to damp indoor
air environments generated by aberrant water exposure due to excessive
condensation and failure of water-use devices, as well as building
envelope breach during heavy rains or flooding, as occurred during Hurricanes
Katrina and Rita on the Gulf Coast of the United States. An Institute
of Medicine (IOM) expert panel concluded that an association
exists between damp buildings and upper respiratory tract symptoms, wheeze, cough, and
exacerbation of chronic lung diseases such as asthma, whereas
supportive data for other reported outcomes such as neurocognitive
dysfunction, mucous membrane irritation, fatigue, fever, and immune
disorders are lacking (<xref rid="b24-ehp0114-001099" ref-type="bibr">IOM 2004</xref>). Building-related illnesses are often linked to dampness-promoted growth
of fungi (<xref rid="b11-ehp0114-001099" ref-type="bibr">Fog Nielsen 2003</xref>) and, most notably, <italic>Stachybotrys chartarum</italic>, a saprophytic “black mold” that grows on cellulosic materials, including
wall-board, ceiling tiles, and cardboard (<xref rid="b19-ehp0114-001099" ref-type="bibr">Hossain et al. 2004</xref>). Incidences of indoor <italic>S. chartarum</italic> contamination often generate costly litigation and remediation, are extensively
reported by the media, and have evoked intense public and scientific
controversy (<xref rid="b16-ehp0114-001099" ref-type="bibr">Hardin et al. 2003</xref>). The IOM panel suggested that although <italic>in vitro</italic> and <italic>in vivo</italic> research on <italic>S. chartarum</italic> and its mycotoxins suggests that adverse effects in humans are indeed “biologically
plausible,” their association with building-related
illnesses requires rigorous validation from the perspectives
of mechanisms, dose response, and exposure assessment (<xref rid="b24-ehp0114-001099" ref-type="bibr">IOM 2004</xref>).</p><p>The satratoxins, macrocyclic trichothecenes produced by <italic>S. chartarum</italic>, are potent inhibitors of protein translation that initiate both inflammatory
gene expression and apoptosis <italic>in vitro</italic> after upstream activation of mitogen-activated protein kinases (MAPKs) (<xref rid="b6-ehp0114-001099" ref-type="bibr">Chung et al. 2003</xref>; <xref rid="b38-ehp0114-001099" ref-type="bibr">Yang et al. 2000</xref>). Satratoxin equivalent airborne concentrations ranging from 2 to 34 ng/m<sup>3</sup> (<xref rid="b40-ehp0114-001099" ref-type="bibr">Yike et al. 1999</xref>) and from 54 to 330 ng/m<sup>3</sup> (<xref rid="b36-ehp0114-001099" ref-type="bibr">Vesper et al. 2000</xref>) have been previously estimated, by a translational bioassay, to occur
in rooms of water-damaged homes heavily contaminated with <italic>Stachybotrys</italic>. These water-soluble mycotoxins occur in the outer plasmalemma surface
and the inner wall layers of conidiospores (<xref rid="b15-ehp0114-001099" ref-type="bibr">Gregory et al. 2004</xref>) as well as in nonviable airborne particulates (<xref rid="b2-ehp0114-001099" ref-type="bibr">Brasel et al. 2005</xref>), which could facilitate entry and release into respiratory airway tissue. Indeed, pulmonary
toxicity of the spores of <italic>S. chartarum</italic> and associated trichothecenes has been demonstrated in animal studies
using intranasally or intratracheally exposed laboratory rodents (<xref rid="b41-ehp0114-001099" ref-type="bibr">Yike and Dearborn 2004</xref>; <xref rid="b42-ehp0114-001099" ref-type="bibr">Yike et al. 2005</xref>).</p><p>Under normal resting, nonexercising conditions, the nose functions to filter, warm, and
humidify inhaled air before it enters more delicate airway
and alveolar tissues in the distal lung (<xref rid="b7-ehp0114-001099" ref-type="bibr">Cole 1993</xref>). Nasal passages serve as “scrubbing towers” for the respiratory
tract by efficiently <italic>a</italic>) absorbing water-soluble and reactive gases and vapors, <italic>b</italic>) trapping inhaled particles, and <italic>c</italic>) metabolizing airborne xenobiotics (<xref rid="b17-ehp0114-001099" ref-type="bibr">Harkema 1991</xref>). Given these air conditioning and defensive roles, we hypothesized that
the nasal airways are another critical site for interaction with <italic>S. chartarum</italic> mycotoxins. To test this hypothesis, we employed a murine intranasal instillation
model previously used by our laboratory and others to study
the adverse effects of harmful toxic agents (<xref rid="b13-ehp0114-001099" ref-type="bibr">Giannetti et al. 2004</xref>), allergens (<xref rid="b10-ehp0114-001099" ref-type="bibr">Farraj et al. 2004</xref>), and pathogens (<xref rid="b37-ehp0114-001099" ref-type="bibr">Wiley et al. 2001</xref>) to investigate potential nasal toxicity of satratoxin G (SG), one of
the most potent trichothecenes produced by <italic>S. chartarum</italic> (<xref rid="b38-ehp0114-001099" ref-type="bibr">Yang et al. 2000</xref>).</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec><title>Toxins</title><p>SG and isosatratoxin F (ISF) were purified from <italic>S. chartarum</italic> cultures and kindly provided by B. Jarvis (University of Maryland, College
Park, MD). SG and ISF yielded a single peak at 254 nm by the HPLC
method of <xref rid="b18-ehp0114-001099" ref-type="bibr">Hinkley and Jarvis (2001)</xref>. SG and ISF identities were further confirmed by electrospray ionization/collision-induced
dissociation (ESI-CID) tandem mass spectroscopy at
the Michigan Sate University mass spectrometry facility by a modification
of a published method (<xref rid="b34-ehp0114-001099" ref-type="bibr">Tuomi et al. 1998</xref>) using a LCQ-DECA device (Finnigan, San Jose, CA) fitted with an ESI probe. Deoxynivalenol, T-2 toxin, and verrucarin A (Sigma Chemical Co., St. Louis, MO) had
reported purities of > 98%, 98%, and 95%, respectively.</p></sec><sec><title>Laboratory animals and intranasal instillation</title><p>Mice were maintained under humane conditions according to National Institutes
of Health guidelines (<xref rid="b23-ehp0114-001099" ref-type="bibr">Institute of Laboratory Animal Resources 1996</xref>) as overseen by the All University Committee on Animal Use and Care at
Michigan State University. Pathogen-free female C57Bl/6 mice (7–8 weeks
of age; Charles River, Portage, MI) were randomly assigned
to experimental groups (<italic>n</italic> = 5–6) and housed in polycarbonate cages containing Cell-Sorb
Plus bedding (A & W Products, Cincinnati, OH) covered with
filter bonnets and provided free access to food and water. Room lights
were set on a 12-hr light/dark cycle, and temperature and relative humidity
were maintained between 21 and 24°C and 40 and 55%, respectively. For
each experiment, mice were anesthetized with 4% halothane
and 96% oxygen and then instilled intranasally
at 50 μL/mouse with SG or other trichothecenes dissolved
in a vehicle of pyrogen-free saline (Abbott Laboratories, Abbott Park, IL) or
with the vehicle alone.</p></sec><sec><title>Animal necropsies and tissue processing for light microscopic examination</title><p>For histopathology and morphometry, mice were deeply anesthetized via intraperitoneal (ip) injection of 0.1 mL of 12% sodium pentobarbital
in saline at designated times post-instillation (PI), from 6 hr
to 28 days, and killed via exsanguination by cutting the abdominal aorta
or renal arteries. Heads from each mouse were immediately removed, and 1 mL
of 10% neutral buffered formalin (Fisher Scientific Co., Fairlawn, NJ) was
flushed retrograde through the nasopharyngeal meatus. After
the lower jaw, skin, muscles, eyes, and dorsal cranium were
removed, the head with brain intact was immersed and stored in a large
volume of the fixative for at least 48 hr before further tissue processing. Lungs
were also removed and intratracheally perfused with formalin
fixative at a constant pressure of 30 cm of water for approximately 1 hr
and then similarly immersed and stored for a minimum 48 hr.</p><p>After fixation, transverse tissue blocks from the head and left lung lobe
of these mice were selected for light microscopy as previously described (<xref rid="b33-ehp0114-001099" ref-type="bibr">Steiger et al. 1995</xref>). Before sectioning, the heads were decalcified in 13% formic
acid for 7 days and then rinsed in tap water for at least 4 hr. The nasal
cavity of each mouse was transversely sectioned at four specific anatomic
locations, designated T1–T4 (<xref rid="b27-ehp0114-001099" ref-type="bibr">Mery et al. 1994</xref>; <xref rid="b43-ehp0114-001099" ref-type="bibr">Young 1981</xref>). The most proximal nasal section was taken immediately posterior to the
upper incisor teeth (proximal, T1); the middle section was taken at
the level of the incisive papilla of the hard palate (middle, T2); the
third nasal section was taken at the level of the second palatal ridge (T3); and
the most distal nasal section (T4) was taken at the level
of the intersection of the hard and soft palate and through the proximal
portion of the olfactory bulb (OB) of the brain (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1</xref>). In addition, two transverse tissue blocks from the left lung lobe were
also taken for microscopic examination at the level of airway generation 5 (proximal) and
generation 11 (distal) along the main axial airway. Tissue
blocks were embedded in paraffin, and the anterior face of
each block was sectioned at a thickness of 5 μm, and stained
with hematoxylin and eosin (H&E).</p></sec><sec><title>Immunohistochemistry</title><p>Unstained and hydrated paraffin sections from nasal blocks T3 and T4 were
incubated first with a nonspecific protein-blocking solution containing
normal sera (Vector Laboratories Inc., Burlingame, CA) and then with
specific dilutions of primary polyclonal antibodies directed against
activated caspase-3 (1:100, rabbit anti-caspase-3 antibody; Abcam, Inc., Cambridge, MA), olfactory marker protein (OMP; 1:4,000, goat anti-OMP
antibody, provided by F. Margolis, University of Maryland), or infiltrating
neutrophils (1:600, rabbit anti-rat neutrophil antibody, provided
by R. Roth, Michigan State University). Tissue sections used for
caspase-3 or OMP detection were pretreated before the blocking solution
with 3% hydrogen peroxide in methanol to destroy endogenous
peroxidase. With these tissue sections, the primary antibody was followed
by the secondary antibody, biotinylated anti-species IgG. Immunoreactivity
of caspase-3 and OMP was visualized with Vector R.T.U. Elite
ABC-Peroxidase Reagent followed by Nova Red (Vector Laboratories Inc.) as
the chromagen. Anti-neutrophil antibody treatment was followed
by biotinylated anti-rabbit IgG, and then streptavidin-phosphatase complex (KPL
laboratories, Gaithersburg, MD) and Vector red as the chromagen. After
immunohistochemistry, slides were lightly counterstained with
hematoxylin.</p></sec><sec><title>Semiquantitative scoring of nasal histopathology</title><p>Nasal sections from mice that received a single instillation of SG at various
doses and were sacrificed 24 hr PI were scored for the amount of
toxin-induced, light microscopic lesions in the olfactory epithelium (OE). A
veterinary pathologist, without previous knowledge of exposure
history of the individual mice, ranked severity of SG-induced OE apoptosis
with atrophy in the examined nasal tissue sections (T1–T4) using
the following histopathologic numeric scores: 0, no SG-induced
nasal lesions in OE; 1 (minimal), 25% of OE with lesions; 2 (mild), 25–50% of OE with lesions; 3 (moderate), 50–75% of
OE with lesions; or 4 (marked), ≥ 75% of
OE with lesions.</p></sec><sec><title>Light microscopic morphometry</title><p>Thickness of the OE lining the medial surface of the second ethmoid turbinates (2E) in
T3 (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1</xref>) was morphometrically evaluated as previously described for airway epithelium (<xref rid="b22-ehp0114-001099" ref-type="bibr">Hyde et al. 1990</xref>, <xref rid="b21-ehp0114-001099" ref-type="bibr">1991</xref>; <xref rid="b30-ehp0114-001099" ref-type="bibr">Plopper et al. 1994</xref>). Measurements were conducted at a final magnification of 3,540× using
a light microscope (Olympus BX40; Olympus America Inc., Melville, NY) coupled
to a 3.3-megapixel digital color camera (Q-Color 3 Camera; Quantitative
Imaging Corp., Burnaby, British Columbia, Canada), and
a personal computer (Dimension 8200; Dell, Austin, TX). The morphometric
analyses were performed using a cycloid grid overlay and software
for counting points and intercepts (Stereology Toolbox; Morphometrix, Davis, CA) (<xref rid="b22-ehp0114-001099" ref-type="bibr">Hyde et al. 1990</xref>, <xref rid="b21-ehp0114-001099" ref-type="bibr">1991</xref>). The percentage volume density, <italic>V</italic><italic><sub>v</sub></italic>, the proportion of the epithelium composed of cytoplasm, nuclei, or apoptotic
nuclear fragments, was determined by point counting and calculated
using the following formula:</p><disp-formula><graphic xlink:href="ehp0114-001099e1.jpg" position="float" mimetype="image"/></disp-formula><p>where <italic>P</italic><italic><sub>p</sub></italic> is the point fraction of <italic>P</italic><italic><sub>n</sub></italic>, the number of test points hitting the structure of interest, divided
by <italic>P</italic><italic><sub>t</sub></italic>, the total points hitting the reference space (OE). The volume of the
epithelial component of interest (e.g., apoptotic nuclei) per unit of
basement membrane (<italic>S</italic><italic><sub>v</sub></italic>) was determined by point and intercept counting and was calculated using
the following formula:</p><disp-formula><graphic xlink:href="ehp0114-001099e2.jpg" position="float" mimetype="image"/></disp-formula><p>where <italic>I</italic><italic><sub>o</sub></italic> is the number of intercepts with the object (epithelial basal lamina) and <italic>L</italic><italic><sub>r</sub></italic> is the length of test line in the reference volume (epithelium). To determine
thickness of the OE, a volume per unit area of basal lamina (cubic
micrometers per square micrometer) was then calculated using the
following formula for arithmetic mean thickness (τ):</p><disp-formula><graphic xlink:href="ehp0114-001099e3.jpg" position="float" mimetype="image"/></disp-formula><p>Other standard morphometric and image analysis techniques were used to
determine the numeric cell density of mature olfactory sensory neurons (OSNs) in
OE. Morphometric estimates of the numeric cell density of OSNs
immunohistochemically reactive for OMP (protein indicator of mature
OSNs) were determined via light microscopy (790× final magnification) by
counting the number of nuclear profiles of these immunoreactive
neuroepithelial cells in the OE lining the medial surface of 2E
in T3 (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1</xref>) and dividing by the length of the underlying basal lamina. The length
of the basal lamina was determined from the contour length of a computerized
digital image of the basal lamina using the Scion Image program (Scion
Corporation, Frederick, MD). All numeric cell density data were
expressed as the number of OSN nuclei per millimeter of basal lamina.</p></sec><sec><title>Ultrastructural examination of the olfactory mucosa and OB via transmission
electron microscopy</title><p>Mice designated for transmission electron microscopy (TEM) analysis were
anesthetized with an ip injection of 0.1 mL 12% pentobarbital
containing 1 IU heparin. Immediately after anesthesia, the whole body
received an intravascular perfusion via the left heart with a saline
solution containing 10 IU heparin for 2–3 min, followed by a 7–10 min
perfusion with 4% glutaraldehyde fixative solution (Ted
Pella, Inc., Redding, CA). The nasal cavity and brain were
then removed and stored in the fixative until TEM processing after the
nasal cavity was decalcified with 10% EDTA for 3–4 weeks; selected
tissues from ethmoid turbinates and OB were postfixed in 1% phosphate-buffered osmium tetroxide, dehydrated through a
graded series of ethanol and propylene oxide, and embedded in Poly/Bed-Araldite
resin (Polysciences, Inc., Warrington, PA). Sections (1 μm) were
cut and stained with toluidine blue for light microscopic
identification of tissue sites for TEM. Ultrathin tissue sections for
TEM were cut at approximately 75 nm with a diamond knife, mounted on
copper grids, and stained with lead citrate and uranyl acetate. Sectioning
was done with an LKB Ultratome III (LKB Instruments, Inc., Rockville, MD). Ultrastructural
tissue examination and photography were performed
with a JEOL JEM 100CXII electron microscope (JEOL Ltd., Tokyo, Japan).</p></sec><sec><title>Real-time polymerase chain reaction</title><p>Mice used for polymerase chain reaction (PCR) analyses of nasal and brain
tissues were anesthetized and killed at designated times after SG instillation
as described above. Immediately after death, the head of each
mouse was removed from the carcass; after the skin, muscles, eyes, and
lower jaw were removed from the head, the nasal airways were opened
by splitting the nose in a sagittal plane adjacent to the midline. The
nasal septum was removed, thereby exposing the nasal turbinates projecting
from the lateral wall of each nasal passage (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1</xref>). Using a dissecting microscope and ophthalmic surgical instruments, all
ethmoid turbinates and the OB were dissected from both nasal passages
and brain, respectively. These excised tissues were stored in RNAlater (Ambion
Inc., Austin, TX) within 5 min, and RNA was isolated using
RNeasy Protect Mini kit (Qiagen Inc., Valencia, CA) within 7 days. Absence
of RNA degradation was routinely verified by agarose electrophoresis; real-time
PCR for apoptosis-related genes [<italic>Fas</italic>, <italic>FasL</italic>, p75-nerve growth factor receptor (<italic>p75NGFR</italic>), <italic>p53</italic>, <italic>Bax</italic>, <italic>Bcl-2</italic>, <italic>Apaf-1</italic>, caspase-3, and caspase-activated DNase (<italic>CAD</italic>)], cytokine genes [interleukin-1 (<italic>IL-1</italic>), tumor necrosis factor-α (<italic>TNF-</italic>α), <italic>IL-6</italic>], and the chemokine gene <italic>MIP-2</italic> (macrophage-inflammatory protein-2) were performed on an ABI PRISM 7900HT
Sequence Detection System using Taqman One-Step RT-PCR (reverse-transcriptase-PCR) Master
Mix and Assays-on-Demand primer/probe gene expression
products according to the manufacturer’s protocols (Applied
Biosystems, Foster City, CA). Relative quantification of apoptotic
and cytokine gene expression was carried out using an 18S RNA as the
loading control and an arithmetic formula method (<xref rid="b1-ehp0114-001099" ref-type="bibr">Audige et al. 2003</xref>).</p></sec><sec sec-type="results"><title>Statistics</title><p>All data were analyzed with SigmaStat (version 3.1; Jandel Scientific, San
Rafael, CA) with the criterion for significance set at <italic>p</italic> < 0.05. Morphometric and RT-PCR data were statistically analyzed using
one-way analysis of variance with Student-Newman-Keuls posttest. Data
from histopathologic severity scores of SG-induced lesions were analyzed
using the Mann Whitney rank sum test (nonparametric test) with
Bonferroni correction for multiple comparisons.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>OE targeted by nasal SG exposure</title><p>Light microscopic evaluation of four specific anatomical sites (T1–T4) revealed
that mice exposed to SG [500 μg/kg body
weight (bw)] and sacrificed at 1, 3, or 7 days PI had conspicuous
nasal epithelial and inflammatory lesions in the dorsocaudal half
of the nasal passages that is normally lined by OE. These lesions were
not apparent in the nasal cavity of saline vehicle–treated
controls (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1A</xref>). SG-related alterations were neither present in regions of the nasal
airways lined by other nasal epithelial types, including respiratory, transitional, or
squamous epithelium, nor found in the lungs of exposed
mice. In addition, mice that were sacrificed only 6 hr PI had no exposure-related
lesions in their nasal cavities. SG-induced OE lesions at 1 day
PI consisted of numerous individual epithelial cells with morphologic
features characteristic of apoptosis in the OE lining all the
ethmoid turbinates and the adjacent lateral walls that border the lateral
meatus in the distal regions of both nasal passages (T3 and T4), with
the most dorsolateral ethmoid turbinates (1E and 2E) being most severely
affected (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1B</xref>). SG-induced apoptosis was also present in the OE of the mid and ventral
septum lining the middle medial meatus in the distal nasal passages (T3 and
T4). In the middle of the nasal passages (T2), before the distal
regions containing the ethmoid turbinates (T3 and T4), SG-induced
apoptotic lesions in the OE were detected only in a small mucosal region
of the lateral walls and septum lining the middle medial meatus where
the OE meets the respiratory epithelium. SG-induced nasal epithelial
lesions were undetectable in the most proximal regions of the nasal
passages (T1). Interestingly, the OE lining the dorsal medial meatus throughout
the nasal passages (T1–T4) had no microscopic evidence
of SG-induced apoptosis or any other epithelial alterations.</p></sec><sec><title>Apoptosis induction in OE</title><p>Dose–response analysis of SG-induced apoptotic lesions indicated
that the no-effect level was 5 μg/kg bw (80 ng/mouse) and the
lowest effect level was 25 μg/kg bw (400 ng/mouse; <xref ref-type="fig" rid="f2-ehp0114-001099">Figure 2</xref>). SG-induced apoptosis was defined by condensation and shrinkage of individual
epithelial cells; clumping, fragmentation, and margination of
nuclear chromatin; and numerous widely scattered cellular fragments (apoptotic
bodies) (<xref ref-type="fig" rid="f3-ehp0114-001099">Figure 3A</xref>–C). Apoptosis was restricted to OSNs whose cell bodies and nuclei
reside in the middle nuclear layers of the OE below the distinct apical
row of sustentacular (support) cell nuclei and above the basal cell
nuclei near the basal lamina. Shrunken OSNs and associated apoptotic
bodies expressed the inducible apoptotic marker protein caspase-3 at 1 day
PI (<xref ref-type="fig" rid="f3-ehp0114-001099">Figure 3B</xref>). Prominent anti-caspase-3 staining was also present in olfactory nerve
bundles (axons of OSNs) in the underlying lamina propria of the mice
instilled with SG. Real-time PCR analysis of microdissected OE-lined
ethmoid turbinates from SG-treated mice demonstrated a marked up-regulation
of the proapoptotic genes <italic>Fas</italic>, <italic>FasL</italic>, <italic>p75NGFR</italic>, <italic>p53</italic>, <italic>Bax</italic>, caspase-3, and <italic>CAD</italic>, whereas expression of <italic>Apaf-1</italic> and the anti-apoptotic gene <italic>Bcl-2</italic> was unchanged (<xref ref-type="fig" rid="f3-ehp0114-001099">Figure 3D</xref>).</p></sec><sec><title>SG-mediated OE atrophy</title><p>Concurrent with SG-induced OSN apoptosis, OE atrophy was detectable at 1 day
PI (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4A</xref>). SG-induced OE lesions were similarly apparent at 3 and 7 days PI along
with increased atrophy. Morphometry of OE lining 2E (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4B</xref>) confirmed that greater atrophy occurred in mice sacrificed 3 and 7 days
PI than in mice sacrificed 1 day PI, with an approximately 50% reduction
in epithelial thickness compared with control mice (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4C</xref>). Restoration of the normal thickness of the OE compared with that of
control was still not complete at 28 days PI (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4C</xref>). Compared with SG-exposed OE at 1 day PI, volume density of apoptotic
nuclei within the OE was remarkably reduced at 3 days PI and absent at 7 and 28 days
PI (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4D</xref>). Exposure to a lower daily dose of SG (100 μg/kg bw or 1.6 μg/mouse) for 5 consecutive days resulted in similar atrophy and
apoptosis of OE compared with those alterations caused by the single 500 μg/kg
bw SG (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4C</xref>,D), suggesting that, in the short term, these effects were cumulative.</p></sec><sec><title>Role of trichothecene structure in OE atrophy</title><p>The nasal effects of trichothecenes not associated with <italic>Stachybotrys</italic> were also assessed. Mice intranasally exposed to deoxynivalenol, T-2, and
verrucarin A, which are type A, type B, and macrocyclic trichothecenes, respectively, using
doses of equivalent to one-third to one-fifth
of LD<sub>50</sub> values (doses lethal in 50% of test animals) (<xref rid="b35-ehp0114-001099" ref-type="bibr">Ueno 1984</xref>). These trichothecenes had no effect on OE compared with the <italic>Stachybotrys</italic> mycotoxins SG and ISF, which exhibited robust toxicity (<xref ref-type="fig" rid="f5-ehp0114-001099">Figure 5A</xref>). Thus, the trichothecene nucleus with characteristic 12–13 epoxide
found in trichothecenes was insufficient to induce OE atrophy in
mice; rather, the effect appeared to be dependent on satratoxin structure (<xref ref-type="fig" rid="f5-ehp0114-001099">Figure 5B</xref>).</p></sec><sec><title>Selective apoptosis induction in OSNs</title><p>OMP, a specific peptide found only in mature OSNs (<xref rid="b26-ehp0114-001099" ref-type="bibr">Kream and Margolis 1984</xref>), was markedly reduced in the OE of SG-instilled mice compared with saline-instilled
control mice (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6A–C</xref>). Morphometric analysis revealed a 67, 94, and 81% loss in OSNs
per millimeter of OE in the nasal mucosa lining the dorsolateral meatus
at 1, 3, and 7 days PI, respectively, compared with the identical
region in vehicle-instilled mice (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6D</xref>). Consistent with partial recovery of OE thickness at 28 days PI (<xref ref-type="fig" rid="f4-ehp0114-001099">Figure 4C</xref>), there was also incomplete restoration of the numbers of OMP-positive
OSNs (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6D</xref>) in the nasal mucosa lining the dorsolateral meatuses of OSNs and OE thickness. OSNs, unlike
neurons in other parts of the body, have the ability
to regenerate from OE basal cells and restore their synaptic connections
in the OB (<xref rid="b14-ehp0114-001099" ref-type="bibr">Graziadei and Graziadei 1978</xref>). Dosing with five consecutive daily SG instillations (100 μg/kg
bw) resulted in an 87% loss of OSNs, again suggesting that
neuronal effects were cumulative. At the ultrastructural level, SG-induced
atrophic OE had a conspicuous loss of nuclear and cytoplasmic profiles
of OSNs and their ciliated dendritic knobs that contain the animal’s
odorant receptors and normally project above the microvillar
apical surfaces of the sustentacular cells (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6E–G</xref>). Consistent with OSN loss, both ultrastructural and immunohistochemical
examination demonstrated a marked reduction of the normally dense mat
of cilia projecting from the dendritic knobs and lining the surface
of the nasal airway lumen (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6B,G</xref>).</p><p>Concurrent with the initial loss of OMP-positive OSNs, there was also noticeable
atrophy of the OMP-positive olfactory nerve bundles located
in the lamina propria underlying the atrophic OE (<xref ref-type="fig" rid="f6-ehp0114-001099">Figure 6B</xref>). This corresponded to marked bilateral atrophy of the olfactory nerve
layer and adjacent glomerular layer comprising the outer two tissue layers
of the OB in SG-instilled mice (<xref ref-type="fig" rid="f7-ehp0114-001099">Figure 7A,B</xref>). Loss of OMP-stained olfactory nerves was most marked in the lateral
and medial aspects of each of the OBs (<xref ref-type="fig" rid="f7-ehp0114-001099">Figure 7C</xref>).</p></sec><sec sec-type="background"><title>Inflammatory gene up-regulation and neutrophil infiltration in OE and OB</title><p>At 1 day PI, SG also induced conspicuous accumulations of exfoliated and
degenerating cellular debris from the dendritic portions of the apoptotic
OSNs in the nasal airways along the luminal surfaces of the atrophying
OE. With secondary degeneration of these exfoliated dendritic fragments, there
was accompanying infiltration of numerous phagocytic cells
consisting mainly of polymorphonuclear leukocytes (neutrophils) and
only occasional mononuclear cells (monocytes and macrophages). Many
of the luminal neutrophils had engulfed apoptotic cellular fragments (<xref ref-type="fig" rid="f3-ehp0114-001099">Figures 3C</xref>, <xref ref-type="fig" rid="f6-ehp0114-001099">6F</xref>). Phagocytosis of apoptotic bodies by sustentacular cells within the OE
was also evident by ultrastructural examination. Lesser numbers of infiltrating
neutrophils were also widely scattered in lamina propria of
the SG-altered olfactory mucosa. Consistent with leukocyte influx was
a markedly increased expression of mRNAs for the cytokines TNF-α, IL-1α, IL-1β, and IL-6 as well as the chemokine MIP-2 associated
with acute inflammatory cell infiltration (<xref ref-type="fig" rid="f8-ehp0114-001099">Figure 8A</xref>). Elevated <italic>MIP-2</italic>, <italic>TNF-</italic>α, and <italic>IL-</italic>6 mRNA expression was also observed in the OB (<xref ref-type="fig" rid="f8-ehp0114-001099">Figure 8B</xref>).</p><p>Slightly decreased severity of SG-induced neutrophilic rhinitis corresponded
with time-dependent disappearance of epithelial apoptosis and development
of epithelial atrophy. A mild-to-moderate influx of neutrophils
persisted in the lamina propria of the affected nasal mucosa underlying
the atrophic OE even at 7 days PI (<xref ref-type="fig" rid="f8-ehp0114-001099">Figure 8C</xref>). Remarkably, there were mild, widely scattered infiltrations of neutrophils
in the remaining olfactory nerve bundles penetrating the bony cribiform
plate that separates the nasal cavity and OB of the brain (<xref ref-type="fig" rid="f8-ehp0114-001099">Figure 8D</xref>). These infiltrations extended bilaterally into the atrophic olfactory
nerve and glomerular layers of the OB at 7 days PI (<xref ref-type="fig" rid="f8-ehp0114-001099">Figure 8E</xref>). Ultrastructural examination revealed that the infiltrating neutrophils
were closely associated with focal areas of degeneration and loss of
OSN axons in both of these outer layers of the OB (<xref ref-type="fig" rid="f9-ehp0114-001099">Figure 9A,B</xref>). A few isolated neutrophils were even detected in the deeper external
plexiform layer, although no neuronal damage was evident in this or other
areas of the OB of SG-exposed mice.</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The causes of damp-building syndrome are likely to be multifactorial and
involve toxic, inflammatory, and allergic responses to microbes and
their products; however, the underlying mechanisms, relative contributions
of individual organisms, and potential for interactions remain poorly
understood (<xref rid="b24-ehp0114-001099" ref-type="bibr">IOM 2004</xref>). Although exposure to either <italic>S. chartarum</italic> spores or associated satratoxins has been previously shown to initiate
acute inflammatory responses in the rodent lung (<xref rid="b41-ehp0114-001099" ref-type="bibr">Yike and Dearborn 2004</xref>), our observations that very low doses of SG are directly toxic to OSNs
and initiate an inflammatory response in the nose (rhinitis) that extends
into the brain (mild focal encephalitis) (<xref ref-type="fig" rid="f10-ehp0114-001099">Figure 10</xref>) are heretofore unreported. These findings raise significant new questions
about hazards associated with indoor exposure to this fungus in water-damaged
buildings. Several aerosol studies have demonstrated that
there is substantial deposition (> 50%) of either very large (> 5 μm
in particle diameter; e.g., a fungal spore) and
very small particles (< 10 nm in diameter; nanoparticles) in the nasal
airways of humans and laboratory animals when inhaled through the
nose (<xref rid="b3-ehp0114-001099" ref-type="bibr">Cheng et al. 1990</xref>, <xref rid="b5-ehp0114-001099" ref-type="bibr">1991</xref>, <xref rid="b4-ehp0114-001099" ref-type="bibr">1996</xref>; <xref rid="b39-ehp0114-001099" ref-type="bibr">Yeh et al. 1997</xref>). Therefore, it is very likely that the nasal airways will filter out
the inhaled spores or extremely small fragments emitting from the mold, preventing
deposition in the lower respiratory tract, including the
lungs.</p><p>Like other epithelial cells in the body, but unlike most neuronal cell
populations in the mammalian nervous system, OSNs undergo apoptosis and
genesis throughout the life of the animal as part of the normal turnover
of mature OE. OSNs are unique in that they have relatively short
life spans compared with other neurons and are continuously being replaced
through basal cell proliferation and differentiation (neuronal regeneration) (<xref rid="b14-ehp0114-001099" ref-type="bibr">Graziadei and Graziadei 1978</xref>). Most OSNs live for 30–40 days, but some cells have life spans
of 3 months or even longer. OSNs of laboratory animals may be induced
to die <italic>in vivo</italic> by experimentally manipulative methods that include olfactory bulbectomy, transection
of the olfactory nerve at the cribiform plate, and intranasal
exposure to chemicals known to be toxic to the OE, such as zinc
sulfate and methyl bromide (<xref rid="b8-ehp0114-001099" ref-type="bibr">Cowan and Roskams 2002</xref>). Exposures to most olfactory chemical toxins result in necrosis (oncosis) of
the OSNs along with other epithelial cells in the OE, unlike the
selective cell death of OSNs by apoptosis observed in the present study. Recently, however, exposure of mice to some chemotherapeutic agents, such
as vincristine, was found to induce marked apoptosis of OSNs
with subsequent OE atrophy that resembles SG-induced lesions described
herein but without obvious nasal inflammation (<xref rid="b25-ehp0114-001099" ref-type="bibr">Kai et al. 2004</xref>). In contrast to our study, mice in these previous studies were given
the chemical agents systemically and at much higher doses relative to
body weight (milligrams per kilogram vs. micrograms per kilogram).</p><p>SG might drive both extrinsic (death receptor–mediated) and intrinsic (mitochondrial-mediated) apoptotic pathways in OSNs. The trichothecenes
induce gene expression and apoptosis via a ribotoxic stress response
that involves MAPKs (<xref rid="b32-ehp0114-001099" ref-type="bibr">Shifrin and Anderson 1999</xref>; <xref rid="b38-ehp0114-001099" ref-type="bibr">Yang et al. 2000</xref>) and is mediated upstream by double-stranded RNA–activated protein
kinase (<xref rid="b46-ehp0114-001099" ref-type="bibr">Zhou et al. 2003</xref>) and Src-family kinases (<xref rid="b45-ehp0114-001099" ref-type="bibr">Zhou et al. 2005b</xref>). Notably, SG-induced genes that have previously been associated with
death receptor–mediated OSN apoptosis include <italic>TNF-</italic>α, <italic>Fas</italic>, <italic>FasL</italic>, and <italic>p75NGFR</italic> (<xref rid="b8-ehp0114-001099" ref-type="bibr">Cowan and Roskams 2002</xref>), as well as the downstream apoptotic genes <italic>p53</italic> (<xref rid="b20-ehp0114-001099" ref-type="bibr">Huang et al. 1995</xref>), <italic>Bax</italic> (<xref rid="b12-ehp0114-001099" ref-type="bibr">Ge et al. 2002</xref>), and caspase-3 (<xref rid="b9-ehp0114-001099" ref-type="bibr">Cowan and Roskams 2004</xref>). Relative to the intrinsic pathway, trichothecene deoxynivalenol induces
p38-mediated mitochondrial-dependent caspase-3 activation and apoptosis
in cloned macrophages (<xref rid="b44-ehp0114-001099" ref-type="bibr">Zhou et al. 2005a</xref>). Furthermore, satratoxin H–induced caspase-3 activation and apoptosis
in the PC12 neural cell model have recently been reported to
be both p38 and JNK dependent (<xref rid="b29-ehp0114-001099" ref-type="bibr">Nusuetrong et al. 2005</xref>).</p><p>It is unclear why SG specifically targeted OSNs when nasal respiratory
epithelium and other cell types in the OE were unaffected. OSN sensitivity
to SG might relate to longer regional exposure to epithelial cells
in OE compared with the exposure to cells in respiratory epithelium. This
is possibly due to a much slower rate of mucociliary clearance of
inhaled agents from OE-lined ethmoid turbinates, which are covered by
immotile cilia, compared with other parts of the nasal cavity that are
lined by respiratory epithelium containing motile cilia with high ciliary
beat frequencies. This latter movement generates rapid regional
flows of mucus out of the nasal cavity and through the nasopharynx into
the upper digestive tract (<xref rid="b28-ehp0114-001099" ref-type="bibr">Morgan et al. 1984</xref>). A slower rate of intranasal SG clearance from OE compared with respiratory
epithelium may also be due to differences in other factors known
to affect the clearance of chemicals from the nasal airway, such as
mucosal metabolism or blood flow.</p><p>Alternatively, based on our observations that satratoxin-induced OE atrophy
is highly dependent on chemical structure (<xref ref-type="fig" rid="f5-ehp0114-001099">Figure 5</xref>) and that one region of the nasal cavity lined by OE (dorsomedial meatus) was
consistently spared from toxicant-induced injury (<xref ref-type="fig" rid="f1-ehp0114-001099">Figure 1</xref>), it is tempting to speculate that these trichothecenes or as yet unidentified
metabolites bind to specific OSN receptors, thus facilitating
uptake and resultant toxicity. In support of this contention, populations
of distinct odorant receptors can be divided into four specific topographical
regions of the OE, one of which lines the dorsomedial meatus (<xref rid="b31-ehp0114-001099" ref-type="bibr">Ressler et al. 1993</xref>).</p><p>Taken together, our observations that the OE and OB are targets of SG and
ISF should be a critical consideration in future studies of damp-building–related
illnesses and the potential etiologic role of <italic>S. chartarum</italic>. The profile of induced cytokines and <italic>MIP-2</italic> is likely to contribute to OSN apoptosis as well as accompanying rhinitis
and mild focal encephalitis observed in the present study. In the
future, it will be necessary to ascertain the dose–response effects
and latency of recovery in nasal tissue after chronic exposure to
satratoxins alone, as well as the contributions of spore matrix, or
coexposures to other indoor air contaminants such as endotoxin. Particularly
intriguing will be understanding the basis for OSN specificity
and the role of toxin metabolism. Of further critical importance will
be the extent to which toxicant-induced inflammation and neuronal injury
occur in other parts of the brain along the olfactory pathway and whether
this contributes to neurocognitive dysfunction. Ultimately, all
such information must be framed against accurate quantitative assessments
of human exposure to satratoxins using both state-of-the-art sampling
and analytical methods and relevant biomarkers.</p></sec>
|
Integrating Research, Surveillance, and Practice in Environmental Public
Health Tracking
|
<p>The Centers for Disease Control and Prevention in the U.S. Department of
Health and Human Services is working with selected state and local health
departments, academic centers, and others to develop an environmental
public health tracking initiative to improve geographic and temporal
surveillance of environmental hazards, exposures, and related health
outcomes. The objective is to support policy strategies and interventions
for disease prevention by communities and environmental health
agencies at the federal, state, and local levels. The first 3 years of
the initiative focused on supporting states and cities in developing
capacity, information technology infrastructure, and pilot projects to
demonstrate electronic linkage of environmental hazard or exposure data
and disease data. The next phase requires implementation across states. This
transition could provide opportunities to further integrate
research, surveillance, and practice through attention to four areas. The
first is to develop a shared and transparent knowledge base that
draws on environmental health research and substantiates decisions about
what to track and the interpretation of results. The second is to identify
and address information needs of policy and stakeholder audiences
in environmental health. The third is to adopt mechanisms for coordination, decision
making, and governance that can incorporate and support
the major entities involved. The fourth is to promote disease prevention
by systematically identifying and addressing population-level
environmental determinants of health and disease.</p>
|
<contrib contrib-type="author"><name><surname>Kyle</surname><given-names>Amy D.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000980">1</xref></contrib><contrib contrib-type="author"><name><surname>Balmes</surname><given-names>John R.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000980">1</xref><xref ref-type="aff" rid="af2-ehp0114-000980">2</xref></contrib><contrib contrib-type="author"><name><surname>Buffler</surname><given-names>Patricia A.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000980">1</xref></contrib><contrib contrib-type="author"><name><surname>Lee</surname><given-names>Philip R.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-000980">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>Environmental conditions and exposures affect health and contribute to
chronic disease morbidity and mortality of importance in the United States [<xref rid="b21-ehp0114-000980" ref-type="bibr">Institute of Medicine (IOM) 1988</xref>, <xref rid="b22-ehp0114-000980" ref-type="bibr">2003</xref>; <xref rid="b24-ehp0114-000980" ref-type="bibr">Kindig et al. 2003</xref>; <xref rid="b27-ehp0114-000980" ref-type="bibr">Lee and Paxman 1997</xref>; <xref rid="b34-ehp0114-000980" ref-type="bibr">McGinnis et al. 2002</xref>; <xref rid="b41-ehp0114-000980" ref-type="bibr">Rosenstock 2003</xref>]. Monitoring of environmental factors is usually directed toward
assessing compliance with regulatory mandates [<xref rid="b44-ehp0114-000980" ref-type="bibr">U.S. Environmental Protection Agency (EPA) 2003a</xref>] and not focused on assessing health impacts. Surveillance of
noncommunicable, environmentally mediated diseases is limited. In 2000, the
Pew Environmental Health Commission (PEHC) recommended development
of a system to track environmental agents, exposures, and related diseases (<xref rid="b37-ehp0114-000980" ref-type="bibr">PEHC 2000a</xref>, <xref rid="b38-ehp0114-000980" ref-type="bibr">2000b</xref>, <xref rid="b39-ehp0114-000980" ref-type="bibr">2000c</xref>).</p><p>In 2001, the U.S. Congress appropriated $17.5 million to the Centers
for Disease Control and Prevention (CDC) to develop environmental
public health tracking (EPHT). The CDC selected competitively and funded 24 state
and local health departments and three schools of public
health to participate in this initiative (<xref rid="b32-ehp0114-000980" ref-type="bibr">McGeehin et al. 2004</xref>). Priority environmental factors initially identified by the CDC include
criteria and hazardous air pollutants, drinking water contaminants, persistent
pollutants, and lead (<xref rid="b8-ehp0114-000980" ref-type="bibr">CDC 2003a</xref>). Diseases identified as priorities are respiratory diseases including
asthma, birth defects, cancers, and neurologic disorders (<xref rid="b28-ehp0114-000980" ref-type="bibr">Litt et al. 2004</xref>).</p><p>Like public health surveillance, EPHT seeks to estimate the magnitude of
health problems in populations, detect outbreaks or elevated rates, understand
the natural history of diseases, and evaluate control strategies (<xref rid="b42-ehp0114-000980" ref-type="bibr">Teutsch 2000</xref>). However, tracking of environmental hazards and exposures and related
health outcomes differs from infectious disease surveillance (<xref rid="b40-ehp0114-000980" ref-type="bibr">Ritz et al. 2005</xref>). Occupational health surveillance offers a more relevant model. Both
occupational health surveillance and EPHT must address chemical agents; long
latency of many relevant diseases; multiplicity of exposures; and
the need to control economic and institutional behavior, rather than
individual actions, to prevent disease. The U.S. Congress identified
these concerns (<xref rid="b19-ehp0114-000980" ref-type="bibr">House Committee on Government Operations 1984</xref>) and the need for a national reporting system for occupational health (<xref rid="b20-ehp0114-000980" ref-type="bibr">House Committee on Government Operations 1986</xref>) in the 1980s. The National Institute for Occupational Safety and Health (NIOSH) provides
funding to some states for a Sentinel Event Notification
System for Occupational Risks to recognize, report, and prevent
certain disorders, including work-related asthma, silicosis, and acute
pesticide illness (<xref rid="b2-ehp0114-000980" ref-type="bibr">Baker 1989</xref>). This does not provide a comprehensive picture of occupational disease, because
geographic areas and disorders included are limited. Even when
additional data sources are used, current surveillance does not fully
ascertain the extent of workplace-related disease in the United States (<xref rid="b1-ehp0114-000980" ref-type="bibr">Azaroff et al. 2002</xref>). This experience suggests potential obstacles.</p><p>In EPHT to date, the CDC has emphasized pilot projects to electronically
link data and development of specifications for improved systems for
the electronic communication and use of data, consistent with broader
efforts to modernize public health information systems (<xref rid="b22-ehp0114-000980" ref-type="bibr">IOM 2003</xref>; <xref rid="b25-ehp0114-000980" ref-type="bibr">Kufafka et al. 2001</xref>; <xref rid="b29-ehp0114-000980" ref-type="bibr">Lumpkin 2001</xref>; <xref rid="b48-ehp0114-000980" ref-type="bibr">Yasnoff et al. 2004</xref>). The CDC has also funded planning and capacity building, review of data
sources, and assessment of indicators (<xref rid="b6-ehp0114-000980" ref-type="bibr">CDC 2002a</xref>, <xref rid="b7-ehp0114-000980" ref-type="bibr">2002b</xref>)</p><p>In defining a conceptual approach for EPHT, the CDC began with a model
including the three elements: hazards, exposures, and diseases (<xref rid="b43-ehp0114-000980" ref-type="bibr">Thacker et al. 1996</xref>). This model defines hazard surveillance as “assessment of the
occurrence of, distribution of, and secular trends in levels of hazards (toxic
chemical agents, physical agents, biomechanical stressors, as
well as biological agents) responsible for disease and injury.” It
defines exposure surveillance as the “monitoring of individual
members of the population for the presence of an environmental
agent or its clinically inapparent (i.e., subclinical or preclinical) effects” (<xref rid="b43-ehp0114-000980" ref-type="bibr">Thacker et al. 1996</xref>).</p><p>The CDC augmented the model by proposing to link data about hazards, exposures, and
diseases and to look for possible associations as part of
the surveillance system (<xref rid="b6-ehp0114-000980" ref-type="bibr">CDC 2002a</xref>, <xref rid="b7-ehp0114-000980" ref-type="bibr">2002b</xref>, <xref rid="b8-ehp0114-000980" ref-type="bibr">2003a</xref>, <xref rid="b9-ehp0114-000980" ref-type="bibr">2003b</xref>). Such data linkages would be accomplished through use of common geographic
and temporal identifiers to overlay or combine data over common
areas and time frames (<xref rid="b10-ehp0114-000980" ref-type="bibr">CDC 2004a</xref>). Most pilot projects funded through EPHT demonstrate data linkages (<xref rid="b11-ehp0114-000980" ref-type="bibr">CDC 2004b</xref>). The CDC notes that</p><disp-quote><p>A key distinction between EPHT and traditional surveillance is the emphasis
on data integration across health, human exposure, and hazard information
systems . . . that includes linkage of these data as part of
regular surveillance activities . . . . This system will be used to identify
potential relationships between exposure and health conditions
that either indicate the need for additional research or require intervention
to prevent disease, disability and injury.” (<xref rid="b32-ehp0114-000980" ref-type="bibr">McGeehin et al. 2004</xref>)</p></disp-quote><p>In 2005, the CDC selected four academic centers to participate in the next
phase of the EPHT initiative and plans to competitively select state
and local health departments for the next phase in 2006. The transition
from the first to the second phase provides an opportunity to build
on existing work and enhance EPHT by more closely integrating research, surveillance, and
practice.</p><p>In this article we address four topics relevant to further development
of EPHT. The first is to develop a shared and transparent knowledge base
that draws on environmental health research and substantiates decisions
about what to track and the interpretation of results. The second
is to identify and address information needs of policy and stakeholder
audiences in environmental health. The third is to adopt mechanisms
for coordination, decision making, and governance that can incorporate
and support the major entities involved. The fourth is to promote disease
prevention by systematically identifying and addressing population-level
environmental determinants of health and disease.</p><sec><title>Integrating Research, Surveillance, and Practice</title><p>A fundamental tenet of public health is that surveillance should be conducted
only when there is “some reasonable expectation of intervention,” i.e., actions to reduce disease or improve health (<xref rid="b42-ehp0114-000980" ref-type="bibr">Teutsch 2000</xref>). By integrating knowledge of the environmental factors that contribute
to health and disease into research, surveillance, and practice, EPHT
can contribute to disease prevention.</p><sec><title>Building a shared base of knowledge to support environmental public health</title><p>What to track, how to present and interpret results, and what to recommend
about possible interventions are important decisions. Further development
of deliberative processes that inform and support such decisions
would strengthen EPHT. An initial step would be to begin to define
the knowledge base for these decisions. Observations and conclusions supported
by research findings and informed by environmental monitoring
and public health surveillance might contribute to such a knowledge base.</p><p>Developing a knowledge base for EPHT is consistent with the IOM review
of the capabilities and needs of the public health system (<xref rid="b22-ehp0114-000980" ref-type="bibr">IOM 2003</xref>). The review distinguished between data, information, and knowledge. Data
are measurements and facts about individuals, environments, or communities. Information
is what is generated when data are placed in context
through analysis. Knowledge is what results from an understanding
and interpretation of the information. The IOM viewed the CDC as the
holder of a “research base that produces the scientific evidence
needed to support the regulations in health-related areas that other
federal agencies use” (<xref rid="b22-ehp0114-000980" ref-type="bibr">IOM 2003</xref>).</p><p>Use of a knowledge base could provide a substantiated and transparent basis
for the selection of targets for EPHT, methods used, and interpretation
of data collected. This could increase accountability by allowing
stakeholders to understand rationales for selecting targets and methods. It
could also better connect EPHT to the research community.</p><p>Models emerging internationally may be useful to consider. In Europe, an
environmental health initiative with purposes similar to EPHT emphasizes
the relationship between collection/analysis of data and policy making
and public access to information [<xref rid="b46-ehp0114-000980" ref-type="bibr">World Health Organization (WHO) European Region 2002</xref>, <xref rid="b47-ehp0114-000980" ref-type="bibr">2003</xref>]. The approach envisions a knowledge base about relationships
between environmental factors and health outcomes that exists apart from
the linkage of data in an electronic information system. Such a knowledge
base is seen as the venue for a common understanding of what is
known or suspected to be true about how environmental factors are related
to health effects or diseases.</p><p>Consideration of a knowledge base could affect the methods used for tracking
and decisions about whether linking data is the most appropriate
approach. It is relevant to consider how thoroughly relationships between
environmental hazards or exposures and health outcomes have been
investigated and the strength of any association. Some hazards/exposures
and diseases have been well studied. In such cases, linkage of surveillance
data may not provide new scientific insights unless it offers
methodologic innovations or increased power over previous efforts or
can contribute to determining the causal nature of an observed association. For
cases where associations have been established, it may be more
relevant to focus on tracking environmental determinants of disease. It
is also relevant to consider whether data or methodologic limitations
may cause data linkages to fail to find associations observed in
research studies with greater power or ability to control confounders.</p><p>In cases where relationships have not been investigated, data linkages
may generate hypotheses and lead to important results. In a classic ecologic
study, Goldberger found pellagra to be associated with low income
and later determined it was related to diet (Mullan 1989). Linkage
studies with sufficient power and ability to control for confounding factors
may contribute new scientific findings about associations or the
causal nature of these relationships. In other cases, targeted research
with adequate attention to design issues may be more informative. Issues
with the use of ecologic approaches have been reviewed (<xref rid="b31-ehp0114-000980" ref-type="bibr">Mather et al. 2004</xref>).</p><p>Consideration of the knowledge base is also relevant to interpretation
for policy contexts. It would not be appropriate to view data linkage
as a necessary prerequisite for interventions in situations where the
potential for harm is established. For example, exposure to lead measured
in blood has been conclusively associated with diminution of cognitive
abilities in children (<xref rid="b35-ehp0114-000980" ref-type="bibr">Needleman et al. 1990</xref>). It is not necessary to conduct data linkages to demonstrate this association
before taking action to reduce lead exposures. Moreover, to make
a case for action, communities may not accept the burden of demonstrating, at
a local level, exposure–disease relationships that
have been established through research (<xref rid="b36-ehp0114-000980" ref-type="bibr">National Environmental Justice Advisory Committee 2004</xref>). Work in California with community organizations suggests that the linking
of data may be less important to communities than readily understandable
presentations of information (<xref rid="b5-ehp0114-000980" ref-type="bibr">California Policy Research Center 2004</xref>).</p><p>EPHT may achieve important advances by developing or identifying new data
sources or taking steps that increase comparability of data across
large areas or populations. Regional efforts to coordinate data collection
and analysis across states could improve data and thus lead to new
findings. A biomonitoring and tracking collaborative group currently
underway in the western states may be a structure that could support
such advances. The Public Health Air Surveillance Evaluation—an
interagency project to produce geographically resolved predictions of
particulate matter and ozone concentrations—provides another
example. The project uses models that incorporate both ambient measurements
and satellite data to produce estimates of ambient concentrations
expected to be more accurate than monitoring data alone (<xref rid="b12-ehp0114-000980" ref-type="bibr">CDC 2004c</xref>). The air surveillance project team is exploring uses of these data to
assess relationships with acute health outcomes.</p><p>Incorporating a knowledge base into EPHT could increase integration between
research and surveillance.</p></sec><sec><title>Meeting information needs of policy and stakeholder audiences</title><p>The ultimate goal of EPHT is to improve health and reduce disease. This
requires actions by a variety of entities with capabilities and responsibilities
related to a wide array of environmental factors. Knowledge
must be conveyed to many parties, including local, state, and federal
health and environmental officials; elected officials; leaders of business, civic, and
health organizations; and stakeholders in discussions
about environmental health policy.</p><p>Many agencies have relevant responsibilities (<xref rid="b30-ehp0114-000980" ref-type="bibr">Lurie 2002</xref>). At the federal level, these include the Department of Health and Human
Services (including the U.S. Food and Drug Administration, CDC, and
NIOSH), U.S. EPA, Department of Agriculture, Department of Housing and
Urban Development, Department of Labor (including the Occupational Safety
and Health Administration), Department of Transportation, Department
of Defense, and Department of Energy (<xref rid="b16-ehp0114-000980" ref-type="bibr">Gostin 2000</xref>).</p><p>States have principal authority for public health actions as well as jurisdiction
over considerable health data. State environmental agencies
conduct a great deal of environmental monitoring, often using standard
protocols developed with U.S. EPA. Local agencies have varying degrees
of authority and capacity for assessments and actions related to environment
and health.</p><p>Communities also represent an important audience for EPHT, particularly
with regard to environmental issues at the local level. Stakeholders
can influence policy makers, especially elected officials. Community needs
may best be met by blending technical aspects of environmental health
sciences with health promotion (<xref rid="b23-ehp0114-000980" ref-type="bibr">Kegler and Miner 2004</xref>).</p><p>EPHT has engaged a wide variety of partners. An important next step is
to look carefully at the information needs of all partners, particularly
for policy and stakeholder audiences. Knowledge about environmental
health and the significance of the results of tracking activities must
be translated into information that is useful and compelling in policy
contexts.</p><p>The types of information of greatest use to support policies to protect
public health cannot be systematically identified from the current research
literature (<xref rid="b15-ehp0114-000980" ref-type="bibr">Goodman et al. 2000</xref>; <xref rid="b24-ehp0114-000980" ref-type="bibr">Kindig et al. 2003</xref>). It is fair to say, overall, that policy audiences seek information in
a more distilled and succinct form than researchers do. Policy makers
interested in gaining knowledge and information may lack time, expertise, or
interest to review and interpret data (<xref rid="b14-ehp0114-000980" ref-type="bibr">Fox et al. 2003</xref>). The IOM assessment of the public health system concluded that public
health officials must serve as educators for those in policy positions, noting
that the public health system must be supported by “political
will, i.e., the commitment of elected officials who direct resources
and influence based on evidence” (<xref rid="b22-ehp0114-000980" ref-type="bibr">IOM 2003</xref>). Another analysis notes that public health policies require “leadership
that informs and motivates, economic incentives that encourage
and facilitate change, and science that moves the frontiers. Leading
change requires facility in brokering partnerships and blending science
and community action” (<xref rid="b34-ehp0114-000980" ref-type="bibr">McGinnis et al. 2002</xref>).</p><p>To achieve this leadership requires approaches that are effective for the
intended audiences. Although EPHT is widely viewed as largely synonymous
with data linkage, other approaches to representing and explaining
impacts of environmental factors on health are likely to be useful.</p><p>The technical complexity of relationships between environmental factors
and health may require the use of multiple approaches. For example, factors
that contribute to disease vary by life stage (<xref rid="b13-ehp0114-000980" ref-type="bibr">Daston et al. 2004</xref>), which can be explained conceptually through synthesis of research findings
but would be difficult to demonstrate by linkage of data. In another
example, multiple relationships between environmental factors and
health outcomes often exist, as multiple environmental factors can contribute
to a single health outcome. Conversely, a single environmental
factor may contribute to many diseases. <xref ref-type="fig" rid="f1-ehp0114-000980">Figures 1</xref> and <xref ref-type="fig" rid="f2-ehp0114-000980">2</xref> show examples related to air pollution and respiratory effects. These
relationships can be explained but are not readily illustrated by data
linkages. A model that incorporates the idea of multiple exposures and
multiple effects is being used increasingly (<xref rid="b4-ehp0114-000980" ref-type="bibr">Briggs 2003</xref>). In another example, although it is well known that environmental factors
interact with genetic, behavioral, and social factors to affect health, these
relationships require interpretation not readily evident
by data linkage.</p><p>The cost of data linkage can be high in both time and money. The experience
of states participating in EPHT is that the effort and resources
required to obtain access to data and prepare it for linkage is greater
than initially anticipated (unpublished report of the meeting of the
western tracking states, February 2005, Berkeley Center for</p><p>Environmental Public Health Tracking). Environmental health indicators
or measures that summarize technical information in ways relevant to particular
audiences may be useful for EPHT. Widely used indicators include
the air quality index, reported in many newspapers in the United
States, which reflects air quality on a daily basis. Indicators or measures
can be scientifically based but portray data about important parameters
in ways that are more readily interpretable than the data themselves
might be, particularly for policy and stakeholder audiences. Relationships
between measures that present data about hazards, exposures, and
outcomes may be explained without linking all of the data (<xref rid="b26-ehp0114-000980" ref-type="bibr">Kyle et al. 2006</xref>). Such approaches could be included within the conceptual framework for
EPHT (Kyle AD, unpublished observations). The review of potential environmental
health indicators has been included in the scope of work for
EPHT, but the use of such indicators has not yet been integrated into
the conceptual approach.</p><p>Interest in showing results for certain kinds of governmental actions has
increased. One example is the development of goals under the <xref rid="b18-ehp0114-000980" ref-type="bibr">Governmental Performance Results Act of 1993</xref>. This has resulted in increased demand for demonstration of the health
benefits projected for environmental regulations, such as reductions
in cases of disease (such as asthma) associated with reductions in air
pollution (<xref rid="b45-ehp0114-000980" ref-type="bibr">U.S. EPA 2003b</xref>). Some regard EPHT as a way to document such demonstrations. With sufficient
funding of data collection and analyses, in some instances, particularly
for acute health effects that vary with short-term changes in
environmental conditions, it may be possible to demonstrate such improvements. However, EPHT
programs and public health communities need to
carefully assess and clearly articulate the circumstances under which
such demonstrations can be expected and the resources required to accomplish
them.</p><p>The next phase of EPHT could better integrate research and surveillance
with practice by identifying relevant audiences and developing methods
to meet the information needs of these groups in their efforts to promote
health and prevent disease.</p></sec><sec><title>Investigating governance structures to support partnership</title><p>The EPHT initiative is complex. Many decisions must be made about what
to do, how to do it, who controls what, and how to explain and disseminate
results. The EPHT network will need to be able to identify needs
for decisions, develop and vet proposals, make decisions and commitments, and
keep track of what has been done and needs to be done. A model
for decision making must incorporate shared expertise, joint priority
setting, defined responsibility, and accountability. Approaches to governance, priority
setting, and the commitment of resources that facilitate
partnership between the environmental and health sectors and among
federal, state, and local agencies are sorely needed. Development of
a structure of governance to support these needs will be an important
challenge for the next phase.</p><p>A successful approach would support participation by a wide array of entities. A
successful nationwide and sustainable EPHT program requires
the long-term participation and stable funding of all states.</p><p>Differences in the types of legal authorities available to the public health
and environmental protection sectors are relevant. Since the 1970s, legal
authority to control environmental factors of health consequence
at the federal and state level has been largely vested in environmental
agencies. The U.S. EPA plays a lead role and has the authority
to formally delegate responsibilities under many statutes to states. There
is no analogous authority in the public health sector. The CDC has
limited legal authority outside the area of communicable disease. Public
health law is widely recognized to be outdated and in need of significant
overhaul (<xref rid="b17-ehp0114-000980" ref-type="bibr">Gostin et al. 2003</xref>). EPHT program direction has been defined largely in funding agreements. Further
definition of a federal role in environmental health among
the public health agencies may be worth considering. At a minimum, the
implications of these differences need to be further addressed.</p><p>Governance structures that can integrate partners engaged in both surveillance
and practice and provide a transparent way of making, documenting, and
communicating decisions would be valuable.</p></sec><sec><title>Addressing environmental determinants of health and disease for populations</title><p>Preventing disease associated with environmental hazards/exposures requires
reduction or control of the hazards or exposures. The impact of EPHT
would be enhanced with greater emphasis on environmental determinants
of disease relevant at the population level. This would be consistent
with increased emphasis on determinants of health, which include the
physical environment (both natural and built), genetic factors, behavior, and
the social environment (<xref rid="b3-ehp0114-000980" ref-type="bibr">Boufford and Lee 2001</xref>; <xref rid="b21-ehp0114-000980" ref-type="bibr">IOM 1988</xref>, <xref rid="b22-ehp0114-000980" ref-type="bibr">2003</xref>; <xref rid="b27-ehp0114-000980" ref-type="bibr">Lee and Paxman 1997</xref>; <xref rid="b33-ehp0114-000980" ref-type="bibr">McGinnis and Foege 1993</xref>). Systematic approaches to identify and track known or suspected environmental
determinants are an important component of a modern public health
system (<xref rid="b30-ehp0114-000980" ref-type="bibr">Lurie 2002</xref>). EPHT could provide an opportunity for systematic evaluation of negative
and positive determinants stemming from the physical environment and
implementation of methods to track them.</p><p>Such an approach may require clearer delineation of the various elements
that constitute “hazards.” Hazards as defined in EPHT
include four conceptually distinct elements: sources of environmental
agents, emissions of agents, concentrations of such agents in environmental
media (such as lakes or streams), and concentrations in exposure
media (such as drinking water). These imply different types of data. The
term “hazard” also implies a judgment that these
elements pose harm. An approach that can accurately identify, measure, and
ultimately influence environmental determinants of health requires
more systematic assessment. EPHT represents an opportunity to identify
and address such determinants for the environment.</p></sec></sec><sec sec-type="conclusions"><title>Conclusion</title><p>The EPHT initiative offers an important opportunity to improve data collection
and analysis to generate and synthesize knowledge about environmental
determinants of population health. The opportunity also exists
to increase collaboration and reduce fragmentation between public health
and environmental agencies at all levels and to create a technical
and organizational foundation for improved environmental public health
policy. The goals are ambitious and current resources are insufficient. Further
attention to critical needs of the overall program could strengthen
it and increase the likelihood of success.</p></sec>
|
Validation and Calibration of a Model Used to Reconstruct Historical Exposure
to Polycyclic Aromatic Hydrocarbons for Use in Epidemiologic Studies
|
<sec><title>Objectives</title><p>We previously developed a historical reconstruction model to estimate exposure
to airborne polycyclic aromatic hydrocarbons (PAHs) from traffic
back to 1960 for use in case–control studies of breast cancer
risk. Here we report the results of four exercises to validate and
calibrate the model.</p></sec><sec sec-type="methods"><title>Methods</title><p>Model predictions of benzo[<italic>a</italic>]pyrene (BaP) concentration in soil and carpet dust were tested
against measurements collected at subjects’ homes at interview. In
addition, predictions of air intake of BaP were compared with blood
PAH–DNA adducts. These same soil, carpet, and blood measurements
were used for model optimization. In a separate test of the meteorological
dispersion part of the model, predictions of hourly concentrations
of carbon monoxide from traffic were compared with data collected
at a U.S. Environmental Protection Agency monitoring station.</p></sec><sec><title>Results</title><p>The data for soil, PAH–DNA adducts, and carbon monoxide concentrations
were all consistent with model predictions. The carpet dust data
were inconsistent, suggesting possible spatial confounding with PAH-containing
contamination tracked in from outdoors or unmodeled cooking
sources. BaP was found proportional to other PAHs in our soil and dust
data, making it reasonable to use BaP historical data as a surrogate
for other PAHs. Road intersections contributed 40–80% of
both total emissions and average exposures, suggesting that the repertoire
of simple markers of exposure, such as traffic counts and/or
distance to nearest road, needs to be expanded to include distance to
nearest intersection.</p></sec>
|
<contrib contrib-type="author"><name><surname>Beyea</surname><given-names>Jan</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001053">1</xref></contrib><contrib contrib-type="author"><name><surname>Hatch</surname><given-names>Maureen</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001053">2</xref></contrib><contrib contrib-type="author"><name><surname>Stellman</surname><given-names>Steven D.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001053">3</xref></contrib><contrib contrib-type="author"><name><surname>Santella</surname><given-names>Regina M.</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001053">4</xref></contrib><contrib contrib-type="author"><name><surname>Teitelbaum</surname><given-names>Susan L.</given-names></name><xref ref-type="aff" rid="af5-ehp0114-001053">5</xref></contrib><contrib contrib-type="author"><name><surname>Prokopczyk</surname><given-names>Bogdan</given-names></name><xref ref-type="aff" rid="af6-ehp0114-001053">6</xref></contrib><contrib contrib-type="author"><name><surname>Camann</surname><given-names>David</given-names></name><xref ref-type="aff" rid="af7-ehp0114-001053">7</xref></contrib><contrib contrib-type="author"><name><surname>Gammon</surname><given-names>Marilie D.</given-names></name><xref ref-type="aff" rid="af8-ehp0114-001053">8</xref></contrib>
|
Environmental Health Perspectives
|
<p>Geographic modeling, using emissions data and transport models, strives
to create the equivalent of a hypothetical, ideal monitoring system that
would have measured the concentration of pollutants at all locations
and times in the medium and domain under study (<xref rid="b2-ehp0114-001053" ref-type="bibr">Beyea and Hatch 1999</xref>). Such models, which are becoming increasingly more common in environmental
epidemiology (<xref rid="b20-ehp0114-001053" ref-type="bibr">Nuckols et al. 2004</xref>), represent a relatively new method for moving beyond the ecological studies
that have dominated past work. Once validated, these models can
reduce exposure misclassification by allowing the assignment of individualized, rather
than average, exposures to study subjects.</p><p>We have constructed a geographic model for airborne polycyclic aromatic
hydrocarbons (PAHs) from traffic that is being used in a population-based, case–control
epidemiologic study involving about 3,000 women
on Long Island, New York, known as the Long Island Breast Cancer
Study Project (LIBCSP; <xref rid="b11-ehp0114-001053" ref-type="bibr">Gammon et al. 2002a</xref>). The study area and surrounding traffic network are shown in <xref ref-type="fig" rid="f1-ehp0114-001053">Figure 1</xref>. The model is also being used in a similar study in Buffalo, New York (<xref rid="b19-ehp0114-001053" ref-type="bibr">Nie et al. 2005</xref>).</p><p>Just as with a real monitoring system, it is possible to both validate
and calibrate a geographic model. For this purpose we used samples collected
in the LIBCSP from subsets of subjects: <italic>a</italic>) soil PAHs at residence, <italic>b</italic>) carpet PAH, and <italic>c</italic>) PAH–DNA adducts assessed in peripheral blood. Details of the
measurements have been reported previously (<xref rid="b11-ehp0114-001053" ref-type="bibr">Gammon et al. 2002a</xref>; <xref rid="b24-ehp0114-001053" ref-type="bibr">Shantakumar et al. 2005</xref>). Additional information is provided in the accompanying online <xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>). Details of and default parameters for the geographic model are also
available (<xref rid="b3-ehp0114-001053" ref-type="bibr">Beyea et al. 2005</xref>; Beyea J, Hatch M, Stellman SD, Gammon MD, unpublished data). We refer
to a model before calibration as a “default” model and
a model after calibration as an “optimized” model.</p><p>In addition to the samples collected as part of the LIBCSP, data collected
by the U.S. Environmental Protection Agency (EPA) on concentrations
of carbon monoxide were used as a test of the basic meteorological dispersion
component of the model. A comparison of the historical emissions
data used in the model has been made to sediment PAH concentrations
and air measurements; the results will be reported elsewhere (Beyea
J, Hatch M, Stellman SD, Gammon MD, unpublished data).</p><sec sec-type="materials|methods"><title>Materials and Methods</title><p>Individual exposure estimates were generated using a meteorological dispersion
model (<xref rid="b3-ehp0114-001053" ref-type="bibr">Beyea et al. 2005</xref>) applied to estimates of PAHs emitted along hundreds of thousands of street
segments (in units of nanograms per kilometer). Emission data per
street segment were derived from historical data obtained for tailpipe
emissions and number of vehicles on roads. Receptor locations localized
to the street level were obtained by geocoding residence addresses
obtained at interview. The model has two distinct components related
to the temperature of the engines of emitting vehicles. “Warm-engine” emissions
occur throughout the traffic network, whereas “cold-engine” emissions occur only for a relatively
short distance from the vehicle starting point (default value, 1 km). Cold-engine
emissions differ from warm-engine emissions in magnitude, in
geographic location, and by time of day. For both warm- and cold-engine
conditions, emissions are restricted in the model to times when
the vehicles are traveling on major roads.</p><p>Vehicle emissions of PAHs are known to vary based on acceleration/deceleration
conditions and the engine temperature, although the magnitude
of the intersection contribution has not been quantified. Intersection
emissions were taken proportional to warm-engine or cold-engine emissions
on a particular street but restricted to a parameterized intersection
distance, initially 100 m. One proportionality factor was taken for
all warm-engine emissions and one for cold-engine emissions. Emissions
could be graded further within one-half and one-quarter of the intersection
distance.</p><p>Total emissions were written as the sum of five terms: (warm-engine emissions) + <italic>A</italic> × (warm-engine intersection emissions) + <italic>B</italic> × (cold-engine emissions) + <italic>C</italic> × (cold-engine intersection emissions) + <italic>D</italic> × background. The parameters <italic>A</italic>, <italic>B</italic>, <italic>C</italic>, and <italic>D</italic>, which are defined relative to the first term in the summation, were determined
from fits to either the soil or DNA adduct data that minimized
chi squared (<xref rid="b22-ehp0114-001053" ref-type="bibr">Press et al. 1992</xref>). This chi-square minimization process was carried out while simultaneously
varying, and thereby optimizing, a range of other model parameters
such as washout rate, particle deposition rates, photo decay rates, and
intersection distance.</p><sec sec-type="methods"><title>Validation data</title><sec sec-type="methods"><title>PAH soil data</title><p>Soil measurements were chosen as a potential validation and calibration
opportunity for the geographic model because deposition of PAHs is proportional
to airborne concentrations above the soil (<xref rid="b21-ehp0114-001053" ref-type="bibr">Odabasi et al. 1999</xref>) and because respirable particles in outdoor air are known to penetrate
indoors efficiently and have been found to dominate indoor respirable
PAH concentrations in a number of studies (<xref rid="b9-ehp0114-001053" ref-type="bibr">Dubowsky et al. 1999</xref>; <xref rid="b25-ehp0114-001053" ref-type="bibr">Sheldon et al. 1992</xref>, <xref rid="b26-ehp0114-001053" ref-type="bibr">1993</xref>). Soil measurements are easier to make than air measurements and retain
historical information (<xref rid="b14-ehp0114-001053" ref-type="bibr">Jones 1991</xref>).</p></sec><sec><title>PAH–DNA adducts</title><p>Airborne PAHs can enter the blood through the respiratory pathway, where
they can be metabolized and form PAH–DNA adducts. If our model
is valid, its predictions of recent airborne concentrations should
be correlated with PAH–DNA adduct levels in study subjects, provided
the traffic contribution is large enough to be detected. Previous
studies have demonstrated that DNA adduct levels in white blood cells
reflect short-term environmental exposure, if exposures are high enough (<xref rid="b10-ehp0114-001053" ref-type="bibr">Eder 1999</xref>). For example, in a study by <xref rid="b4-ehp0114-001053" ref-type="bibr">Binkova et al. (1995)</xref>, a scattergram of exposure accumulated on personal dosimeters versus adduct
levels showed a clear trend with only 21 subjects. Although the
ambient concentrations were perhaps 10-fold higher than current U.S. levels, ranging
from 1.6 to 2.9 ng/m<sup>3</sup> of benzo[<italic>a</italic>]pyrene (BaP), we have the advantage of being able to work with
many more subjects. We measured PAH–DNA adducts in the peripheral
blood of 999 study subjects, using the competitive ELISA method, as
described by <xref rid="b12-ehp0114-001053" ref-type="bibr">Gammon et al. (2002b)</xref>.</p><p>About 72% of women had detectable levels of adducts. We analyzed
detects and non-detects separately because descriptive statistics indicated
the existence of a bimodal distribution for adduct level. The
data show a normal distribution with a large spike at the origin that
is well separated from, and not part of, the normal distribution. The
nondetect spike contains 28% of the women in the sample.</p><p>Previous work with this study population has found that the likelihood
of having detectable adducts is elevated among past and current smokers, inversely
associated with increased BaP levels (nanograms per gram) in
dust in the home but positively associated with BaP levels in soil
outside of the house, although confidence intervals were large (<xref rid="b24-ehp0114-001053" ref-type="bibr">Shantakumar et al. 2005</xref>). The study authors did not find any consistent associations between the
odds of having detectable PAH–DNA adducts and various dietary
sources of PAH, including smoked and grilled foods eaten in the most
recent decade of life and a BaP food index assessed from responses to
a food frequency questionnaire (<xref rid="b24-ehp0114-001053" ref-type="bibr">Shantakumar et al. 2005</xref>). As suggested previously (<xref rid="b8-ehp0114-001053" ref-type="bibr">Dickey et al. 1997</xref>), persons with and without detectable adduct levels may represent two
different groups of individuals in their response to PAH exposure. Two
distinct populations responding to PAHs in diet have also been reported (<xref rid="b15-ehp0114-001053" ref-type="bibr">Kang et al. 1995</xref>). The first group of women, those with detectable adducts, are the focus
of this report. It is easier to model the number of adducts in detects
as a function of airborne PAH exposure than to predict the shift from
nondetects to detects. The model reported in this article is not able
to predict the odds ratio of having detectable adducts (<xref rid="b24-ehp0114-001053" ref-type="bibr">Shantakumar et al. 2005</xref>). Regardless of whether the distribution in adduct levels reflects a bimodal
biologic response or a bimodal exposure distribution, levels of
DNA adducts reflect DNA damage and therefore serve as a measure of the
effective biologic dose of PAHs (<xref rid="b4-ehp0114-001053" ref-type="bibr">Binkova et al. 1995</xref>; <xref rid="b18-ehp0114-001053" ref-type="bibr">Nesnow et al. 1993</xref>).</p></sec><sec><title>PAHs in carpet dust</title><p>PAHs in carpet dust come from three sources: ambient outdoor air PAHs that
penetrate indoors and deposit on carpet, indoor-generated PAHs, and
dirt-containing PAHs that are tracked from the outside. The geographic
model should be able to predict the variation in the amount of ambient
PAH deposition per square meter for wall-to-wall carpeting. For rugs
and other carpet that do not extend to the walls, an ambiguity arises
about what denominator to use, for example, carpet area or floor area
because carpet may act as a “sink” for household dust
deposited on the uncarpeted floor. Nevertheless, unless there is strong
spatial confounding with carpet size, indoor-generated sources of
PAHs (e.g., cooking), and/or dust track-in, we expect that the ambient
signal should be detectable in the carpet PAH measurements that were
collected as part of the LIBCSP to provide an exposure marker inclusive
of indoor-generated PAHs.</p></sec><sec><title>CO air concentrations</title><p>Modeling CO air concentration offers a good test of a PAH dispersion and
traffic model for a number of reasons. First, traffic is known to dominate
CO emissions, accounting for as much as 95% of emissions
in cities (<xref rid="b27-ehp0114-001053" ref-type="bibr">U.S. EPA 2001</xref>). Therefore, both CO and traffic PAH emissions will increase and decrease
with traffic density. Second, CO and PAHs are both associated with
incomplete combustion (<xref rid="b1-ehp0114-001053" ref-type="bibr">An and Ross 1996</xref>; <xref rid="b5-ehp0114-001053" ref-type="bibr">Bostrom et al. 2002</xref>), so relative CO emissions should rise and fall during the driving cycle
in a pattern that is similar to that of PAH. [As referenced
in the <xref ref-type="supplementary-material" rid="SD1">Supplemental Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>), hourly patterns of PAH and CO air concentrations have been found in
other studies to be similar indoors and out, with <italic>R</italic><sup>2</sup> coefficients ranging from 0.5 to 0.8.] Third, to model relative
hourly CO air concentrations in any single year, all the modeler has
to do is turn off all depletion phenomena, because deposition, washout, and
photo decay are negligible in the case of CO.</p><p>CO data are widely available for locations around the United States through
requests to the U.S. EPA. In our study area, hourly CO data have
been collected since 1974 at Eisenhower Park in Nassau County (<xref rid="b28-ehp0114-001053" ref-type="bibr">U.S. EPA 2005</xref>). We averaged hourly data for 1975, 1985, and 1995 and regressed the results
against comparable model predictions.</p></sec></sec><sec sec-type="methods"><title>Statistical methods</title><p>Multiple (linear) regression was used to assess and optimize the relationship
between model predictions and CO data, as well as to compare adduct
data with soil and dust data. In model fitting to soil, dust, and
detectable adduct levels, the variables and the model predictions were
all log-transformed to bring them to normal form. Because the model
parameters to be determined appeared inside the transforming logarithm, nonlinear
regression was required. When covariates, such as cooking
sources or BaP in diet, were controlled for, a combination of nonlinear
and multiple regression methods was used as discussed in the <xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>). Fits to the adduct and dust data were made with and without current
smokers included.</p><p>Significance values for bimodal distributions of PAH–DNA adducts
were handled using the method of Simes and Hochberg, as described by <xref rid="b16-ehp0114-001053" ref-type="bibr">Levin (1996)</xref>. According to this method, to obtain an overall significance of 95%, a <italic>p</italic>-value < 0.025 is required for one of the modes whenever a <italic>p</italic>-value for the other mode is > 0.05 (<xref rid="b16-ehp0114-001053" ref-type="bibr">Levin 1996</xref>). Because all <italic>p</italic>-values for adduct nondetects in this study are greater than 0.05, we must
always look for a <italic>p</italic>-value ≤ 0.025, for correlations using detectable adduct values
alone.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Model validation and calibration</title><sec sec-type="methods"><title>CO data</title><p>When normalized, hourly CO data in 1975 were virtually identical to the
hourly data in 1995, despite a 4-fold drop in absolute levels. Because
model regressions for 1975, 1985, and 1995 were similar, we discuss
only the 1995 results.</p><p>With the default model parameter values chosen before optimization, there
is a reasonable fit to the 1995 hourly CO data for the Nassau County
monitor (<italic>r</italic><sup>2</sup> = 52%). This rough agreement is no trivial result because
the <italic>r</italic><sup>2</sup> for the correlation between hourly CO emissions and concentrations was
only 0.033. This weak correlation arises because CO emissions are very
low in the early morning hours in contrast to the measured CO concentrations, which
are relatively high at this time, reaching half the maximum
daytime value. The delay time involved in distant CO reaching a
receptor explains the result. CO measured at 0200 hr was actually emitted
miles away during the tail of the evening rush hour.</p><p>Although the default model accounts for delay in CO arrival, the default
model over predicts in the early evening hours. The over prediction
remains despite the type of dispersion parameters used (urban or rural), despite
a switch to a different year’s meteorological data, and
despite the method chosen to convert the actual dispersion values
from raw meteorological data. However, optimization of the fit to the
CO data, allowing the relative strength of the intersection emissions
and the contribution from distant sources (background term) to increase
over the default value, eliminated the overprediction. The model parameters
determined from the CO optimization were qualitatively similar
to those determined from the soil data but differed from the parameters
determined with PAH–DNA adduct data. The results of the fit
to the CO data after optimization (<italic>r</italic><sup>2</sup> = 63%) are shown in <xref ref-type="fig" rid="f2-ehp0114-001053">Figure 2</xref>.</p><p>The fact that the optimized CO regression results for 1975 and 1985 were
similar to the 1995 results suggests that emissions at intersections
have been dominant over the entire period for which data are available.</p></sec><sec sec-type="methods"><title>Soil data</title><p>There was a high degree of correlation between BaP data and other PAHs [<xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>)]. The mean soil level was 2,300 ng/g of BaP, which is approximately
twice that reported in the only comparable data set we could find
in the Northeast, namely, average values for Boston, Providence, and
Springfield, Massachusetts, as reported by <xref rid="b6-ehp0114-001053" ref-type="bibr">Bradley et al. (1994)</xref>. The difference in mean soil levels may be attributable to differences
in traffic density or differences in the depth of the samples collected
in the two locations. The depth in the New England study ranged from 0 to 10 cm
rather than the average 2-cm depth used in our study.</p></sec><sec sec-type="methods"><title>Spatial variation of the soil data</title><p>Aggregated soil levels within geographic zones were found to decrease with
distance along the axis of Long Island away from urbanization (and
hence from pollution sources), as shown in <xref ref-type="table" rid="t1-ehp0114-001053">Table 1</xref> and <xref ref-type="fig" rid="f3-ehp0114-001053">Figure 3</xref>. The 10-fold decline in geometric mean values is consistent with other
studies (<xref rid="b13-ehp0114-001053" ref-type="bibr">Grass et al. 2000</xref>; <xref rid="b29-ehp0114-001053" ref-type="bibr">Wagrowski and Hites 1997</xref>). Although there is considerable individual variation between prediction
and soil data, the fit is highly significant [<italic>p</italic> < 0.0001, as discussed in the <xref ref-type="supplementary-material" rid="SD1">Supplemental Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>)]. The fit to the soil data when the model predictions are grouped
into 20 quantiles is shown in <xref ref-type="fig" rid="f4-ehp0114-001053">Figure 4</xref>. Further details about the soil regressions are given in the <xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>) along with values for the correlation coefficients between aggregated
soil, dust, and adduct data. In addition to allowing us to calibrate
the model, the usefulness of the results obtained from the soil fits demonstrates
the feasibility of using interviewer-phlebotomists on a tight
schedule to gather environmental samples in the context of a large-scale
case–control study.</p><p>The optimized parameter set determined from the soil data is shown in <xref ref-type="table" rid="t2-ehp0114-001053">Table 2</xref>. Intersection emissions contribute 80% of average emissions and
exposures when calculated with the optimized model.</p></sec><sec sec-type="methods"><title>Adduct data</title><p>Aggregated adduct levels within geographic zones were found to decrease
with distance along the axis of Long Island away from urbanization (and
hence from pollution sources), as shown in <xref ref-type="table" rid="t1-ehp0114-001053">Table 1</xref> and <xref ref-type="fig" rid="f5-ehp0114-001053">Figure 5</xref>. The number of detects and nondetects for PAH–DNA adducts varies
with 16-km zone along the length of Long Island [<xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref>, Table S-1 (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>)]. The odds ratio of having detectable adducts does not differ
significantly by zone (<italic>p</italic> = 0.23).</p><p>The results of optimization of the model to the DNA adduct data are shown
in <xref ref-type="table" rid="t2-ehp0114-001053">Table 2</xref>. Optimization produced a large coefficient for the cold-engine component, with
the coefficient for warm emissions negligible. This is the reverse
of the results for the soil-optimized model. On the other hand, when
it came to the importance of intersection emissions, the results
of fits to the adduct data were consistent with the fits to the soil data
in predicting a major role for enhanced emissions at intersections. Intersections
accounted on average for 40% of total cold-engine
exposures. Deposition velocity, rain washout rate, and photo decay
rate were all optimized at zero, which meant that the optimized adduct
model did not contain any depletion. The parameter values were not changed
significantly upon removing from the regressions women who smoked
within the last year, nor were they changed significantly when we controlled
for a BaP food index, assessed from responses to a food frequency
questionnaire, and the number of smoked and grilled foods eaten
in the most recent decade of life, also obtained from a questionnaire.</p><p>The fit of the adduct-optimized model to the adduct data is shown in <xref ref-type="fig" rid="f6-ehp0114-001053">Figure 6</xref>.The fit to the ungrouped data points can be found in the <xref ref-type="supplementary-material" rid="SD1">Supplemental
Material</xref> (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>; <italic>p</italic> = 0.02 before optimization). The results are not as good as in
the soil case, with the cold-engine version tracking the adduct data
less well than the warm-engine version tracks the soil data. The <italic>r</italic><sup>2</sup> for the grouped data is lower, at 58%.</p><p>Of interest is the fact that the soil data were comparable with the geographic
model in predicting PAH–DNA adduct levels in individual
women with detectable adducts (<italic>p</italic> = 0.004). Although soil data appear to be a simpler indicator
of airborne exposure than the geographic model, soil data are not available
for all women, nor are they available historically.</p></sec><sec sec-type="methods"><title>Dust data</title><p>Values for both PAHs per gram of dust and PAHs per square meter of carpet
vacuumed were available. The Pearson correlation coefficients between
BaP and the other two PAHs measured, dibenz[<italic>a</italic>,<italic>h</italic>]anthracene and benz[<italic>a</italic>]anthracene, were > 0.95. The trend in carpet BaP/m<sup>2</sup> shows a peak in the center of Long Island (<xref ref-type="table" rid="t1-ehp0114-001053">Table 1</xref>), indicating that some source of PAHs other than outside air is dominating
BaP in carpets. Differences by zone were statistically significant. The
same pattern is seen for BaP/g. In contrast, the total grams of
dust per square meter behaves as expected, with high values closer to
the urbanized portion of Long Island [<xref ref-type="supplementary-material" rid="SD1">Supplemental Material</xref>, Tables
S-2, S-3 (<ext-link ext-link-type="uri" xlink:href="http://www.ehponline.org/docs/2006/8659/suppl.pdf">http://www.ehponline.org/docs/2006/8659/suppl.pdf</ext-link>)]. Possible candidates for nontraffic PAHs are indoor sources
such as cooking, and track-in of PAHs from outdoors. This is the one validation
exercise that contradicts the model.</p><p>We found no explanation for this unexpected behavior of carpet dust with
distance. Also anomalous was the correlation between BaP per square
meter and the model’s prediction of deposition of ambient BaP
onto carpet. In fact, the regression slope was negative and remained so
even when potential confounders were included in the regression. Potential
confounders considered were years in residence, work status, age, use
of wood in stove/fireplace, number of children younger than 20 years
at time of data collection, season, number of adults in the home, number
of hours worked away from home, religion, education, income, smoking
status, and the number of times a study subject consumed grilled
meat or fish in the previous decade. This latter variable was the most
relevant surrogate we had for cooking intensity. Excluding homes of
study subjects who smoked within the last year did not reverse the negative
correlation.</p><p>At the individual level, BaP in carpet, whether measured in units of nanograms
per gram or nanograms per square meter, was not correlated with
the level of measurable PAH–DNA adducts in study subjects (<italic>p</italic> > 0.46). As previously reported, a negative correlation between nanograms
per gram of BaP in carpet dust was found in this population for
the odds ratio of a woman having detectable adducts (<xref rid="b24-ehp0114-001053" ref-type="bibr">Shantakumar et al. 2005</xref>). Clearly, much remains to be learned about the origins of PAHs in carpet
dust.</p><p>The fact that the model correlated with the number of PAH–DNA adducts
in women with detectable adducts, whereas the carpet dust data
did not, suggests that the cause of the discrepancy with dust is unlikely
to be connected with indoor sources of PAHs in the respirable range. Nevertheless, regardless of the size distribution, cooking is a likely
source of PAHs in carpet dust that might be confounding the correlation
with the geographic model. We only have a limited surrogate to use
in controlling for cooking PAH.</p><p>Another potential contributor is track-in of PAHs from outdoors, which
is a sequential process progressing from street, driveway, or attached
garage to entryway and then to carpet. Such a pathway could include contamination
at various points from vehicle oil drips, which could contribute
track-in of PAHs distinct from blown soil dust.</p><p>Track-in can be a significant source of PAHs in carpet dust (<xref rid="b7-ehp0114-001053" ref-type="bibr">Chuang et al. 1995</xref>). However, we have no explanation of why track-in patterns would vary
spatially with distance along Long Island according to the pattern we
found. It is thus not possible to rule out the possibility that some source
of indoor PAH, such as cooking, is overwhelming any traffic contribution
in carpets, particularly if the particle sizes are outside the
respirable range.</p></sec><sec sec-type="background"><title>Background model</title><p>Two background models were tested. Compared with a constant background
term, a better fit to the data was found in calibrations using the soil
data for a term that was proportional to exposures calculated from the
more distant counties (all but Nassau, Suffolk, and Queens counties
in New York State). The fit to the DNA adduct data was best with a constant
background term.</p></sec></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>The optimized model parameters can be compared with default values. For
the fits to the soil data, the optimized parameters are within a factor
of 2 of the default values for those dispersion parameters that are
widely reported in the literature, namely, deposition velocity and washout
rate (<xref rid="b17-ehp0114-001053" ref-type="bibr">National Council on Radiation Protection and Measurement 1993</xref>; <xref rid="b23-ehp0114-001053" ref-type="bibr">Ramsdell et al. 1994</xref>). For those parameters for which no strong guidance as to default values
was available in the literature, for example, photo decay rates and
acceleration/deceleration distances, the optimized values turned out
to differ by more than a factor of 2 from the values we chose as defaults.</p><p>The most striking result to come out of the fits to the adduct data is
the removal of depletion phenomena from the optimized cold-engine version
of the model, whether it be dry deposition, wet deposition, or photo
decay of PAHs. This artificial result is an indication that to optimize
the fit, the model needs contributions from more distant sources
than would normally be expected. Perhaps the best explanation is that, unlike
soil receptors, human receptors are mobile. Or, perhaps indoor
sources such as emissions of cooking PAHs for which we have not controlled, are
confounding the results. Despite the apparent differences in
the parameter values determined for the soil- and adduct-optimized models, the
correlation between their predictions of PAH exposure for women
in the study is quite high (<italic>r</italic><sup>2</sup> = 0.79–0.86), possibly because the difference in optimized
parameters compensates for differences in spatial patterns that would
otherwise result.</p></sec><sec sec-type="conclusion"><title>Conclusion</title><p>This study indicates that in developing inhalation exposure estimates it
is necessary to account for emissions at intersections to fully determine
the spatial distribution of PAH exposure. Three of four validation
exercises were consistent with model predictions. The unexpected geographic
pattern of carpet PAHs, which does not match the falloff with
distance from urbanization predicted by the geographic model, is the
only result we found in our validation exercises that calls into question
the relevance of our model. In contrast, the model predictions for
soil PAH data and hourly CO concentrations were very consistent with
the data, favoring a warm-engine emission model. PAH adduct levels for
women with detectable adducts were also consistent with model predictions
and favored a cold-engine emission model. Although we found a high
degree of correlation between the predictions of the warm-engine version
of the model and the cold-engine version, it will be prudent to use
both the warm-engine and cold-engine versions when evaluating the effects
of exposure on health outcomes.</p><p>Whatever model is used, the ability to make individualized exposure estimates
has the potential to reduce exposure misclassification that can
arise in environmental epidemiology studies from assigning group level
exposure based on interpolation from sparse environmental monitoring
data or from surrogate measures of exposure based on simple distance
from nearest major road and/or traffic density. Although geographic models
are complex, comparison of their output with field data helps to
build confidence in them.</p></sec>
|
The Effects of Air Pollution on Hospitalizations for Cardiovascular Disease
in Elderly People in Australian and New Zealand Cities
|
<sec><title>Objective</title><p>The goal of this study was to estimate the associations between outdoor
air pollution and cardiovascular hospital admissions for the elderly</p></sec><sec><title>Design</title><p>Associations were assessed using the case–crossover method for
seven cities: Auckland and Christchurch, New Zealand; and Brisbane, Canberra, Melbourne, Perth, and Sydney Australia. Results were combined
across cities using a random-effects meta-analysis and stratified for
two adult age groups: 15–64 years and ≥ 65 years of age (elderly). Pollutants
considered were nitrogen dioxide, carbon monoxide, daily
measures of particulate matter (PM) and ozone. Where multiple
pollutant associations were found, a matched case–control analysis
was used to identify the most consistent association.</p></sec><sec><title>Results</title><p>In the elderly, all pollutants except O<sub>3</sub> were significantly associated with five categories of cardiovascular disease
admissions. No associations were found for arrhythmia and stroke. For
a 0.9-ppm increase in CO, there were significant increases in elderly
hospital admissions for total cardiovascular disease (2.2%), all
cardiac disease (2.8%), cardiac failure (6.0%), ischemic
heart disease (2.3%), and myocardial infarction (2.9%). There
was some heterogeneity between cities, possibly due
to differences in humidity and the percentage of elderly people. In
matched analyses, CO had the most consistent association.</p></sec><sec><title>Conclusions</title><p>The results suggest that air pollution arising from common emission sources
for CO, NO<sub>2</sub>, and PM (e.g., motor vehicle exhausts) has significant associations with
adult cardiovascular hospital admissions, especially in the elderly, at
air pollution concentrations below normal health guidelines.</p></sec><sec><title>Relevance to clinical and professional practice</title><p>Elderly populations in Australia need to be protected from air pollution
arising from outdoor sources to reduce cardiovascular disease.</p></sec>
|
<contrib contrib-type="author"><name><surname>Barnett</surname><given-names>Adrian G.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001018">1</xref></contrib><contrib contrib-type="author"><name><surname>Williams</surname><given-names>Gail M.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001018">1</xref></contrib><contrib contrib-type="author"><name><surname>Schwartz</surname><given-names>Joel</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001018">2</xref></contrib><contrib contrib-type="author"><name><surname>Best</surname><given-names>Trudi L.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001018">3</xref></contrib><contrib contrib-type="author"><name><surname>Neller</surname><given-names>Anne H.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001018">3</xref></contrib><contrib contrib-type="author"><name><surname>Petroeschevsky</surname><given-names>Anna L.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001018">3</xref></contrib><contrib contrib-type="author"><name><surname>Simpson</surname><given-names>Rod W.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001018">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>There have been several studies on the short-term effects of air pollution
on hospital admissions (<xref rid="b3-ehp0114-001018" ref-type="bibr">Burnett et al. 1997a</xref>, <xref rid="b4-ehp0114-001018" ref-type="bibr">1997b</xref>; <xref rid="b11-ehp0114-001018" ref-type="bibr">Le Tertre et al. 2002</xref>; <xref rid="b15-ehp0114-001018" ref-type="bibr">Pope 2000</xref>; <xref rid="b16-ehp0114-001018" ref-type="bibr">Samet et al. 2000</xref>), but most have examined single cities. Such single-city studies have
been criticized for being applicable only to the city under study and
for using different modeling approaches. These comments have led to multicity
meta-analyses where the results are pooled—for example, the
National Morbidity, Mortality, and Air Pollution Study (NMMAPS) conducted
on behalf of the Health Effects Institute in the United States
and the APHEA (Air Pollution and Health: A European Approach) studies
in Europe. NMMAPS examined the associations between daily hospital
counts for cardiovascular admissions in the elderly and air pollutants
in 14 cities in different regions of the United States (<xref rid="b8-ehp0114-001018" ref-type="bibr">Dominici et al. 2002b</xref>; <xref rid="b16-ehp0114-001018" ref-type="bibr">Samet et al. 2000</xref>). The APHEA studies have taken place in two stages, and the latest (APHEA2) comprised
eight European cities in the investigation of associations
of air pollution on daily cardiovascular admissions (<xref rid="b11-ehp0114-001018" ref-type="bibr">Le Tertre et al. 2002</xref>). Multicity studies have also been conducted in Canada (<xref rid="b3-ehp0114-001018" ref-type="bibr">Burnett et al. 1997a</xref>, <xref rid="b4-ehp0114-001018" ref-type="bibr">1997b</xref>).</p><p>Despite these studies, the strength of the association between outdoor
air pollution and health effects is still unclear because of the complexity
of the time-series modeling. In addition, when multiple pollutants
have been examined, the independent effects of each pollutant are usually
addressed in multipollutant models, but these are sensitive to
the modeling assumptions. If the association with one pollutant is nonlinear
or varies by season, then a two-pollutant model assuming a linear
relationship with each pollutant might not give the independent effect
of the second pollutant. Therefore, the case–crossover design (<xref rid="b12-ehp0114-001018" ref-type="bibr">Maclure 1991</xref>), which is less sensitive to model assumptions, is more appropriate. This
method investigates the effects of acute exposures and can also examine
both multiple exposures and interactions between exposures. It has
been applied to the analysis of the acute effects of environmental
exposures, especially air pollution (<xref rid="b21-ehp0114-001018" ref-type="bibr">Sunyer et al. 2000</xref>). The method matches case days to nearby control days and hence controls
for covariates that change slowly over time (e.g., age, smoking behavior, and
usual diet). Such matching also controls for seasonal variation
and time trends in the health event (<xref rid="b2-ehp0114-001018" ref-type="bibr">Bateson and Schwartz 2001</xref>).</p><p>In this study we aimed to find associations between outdoor air pollutant
and cardiovascular disease (as measured by counts of hospital admissions) in
cities in Australia and New Zealand. The study used two age
groups, ≥ 65 years of age (elderly) and 15–64 years of
age, although the focus here is on the elderly. The study also examined
differences in the associations between cities.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Data collection</title><p>Daily hospital and pollution data were collected for the years 1998 through 2001 in
five of the largest cities in Australia (Brisbane, Canberra, Melbourne, Perth, Sydney) and two cities in New Zealand (Auckland, Christchurch). In 2001, these cities covered 53% of the Australian
population and 44% of the New Zealand population.</p></sec><sec sec-type="methods"><title>Cardiovascular health data and air pollution data</title><p>Health data were collected for all cardiovascular emergency hospital admissions
from state government health departments in Australia and the
New Zealand Health Information Service (Ministry of Health). The cardiovascular
disease categories used in the study are shown in <xref ref-type="table" rid="t1-ehp0114-001018">Table 1</xref>, and summary statistics and demography for each city are shown in <xref ref-type="table" rid="t2-ehp0114-001018">Table 2</xref>.</p><p>The pollutants considered were particulate matter < 2.5 μm in
diameter (PM<sub>2.5</sub>) and <10 μm in diameter (PM<sub>10</sub>) in micrograms per cubic meter; nitrogen dioxide in parts per billion; carbon
monoxide in parts per million; and ozone in parts per billion. Tapered
element oscillating microbalance (TEOM) air samplers provided
the PM data. Daily pollutant levels were calculated by averaging over
a network of monitors in each city. The summary statistics for air pollutants
and weather are shown in <xref ref-type="table" rid="t3-ehp0114-001018">Table 3</xref>.</p><p>CO and NO<sub>2</sub> were the only pollutants monitored in all seven cities on a daily basis. For
PM<sub>2.5</sub>, daily measurements were available in four of the Australian cities: Brisbane, Melbourne, Perth, and Sydney. PM<sub>10</sub> was measured on a daily basis in these four cities and in Christchurch.</p></sec><sec sec-type="methods"><title>Statistical methods</title><p>We used the time-stratified case–crossover method to find associations
between pollutants and daily counts of hospital admissions (<xref rid="b10-ehp0114-001018" ref-type="bibr">Janes et al. 2005</xref>). Controls were chosen from strata of length 28 days; days either side
of the case day were excluded to reduce the correlation between case
and control exposure. The method controlled for long-term trend, seasonal
changes, and respiratory epidemics by design. Using covariates, there
were additional controls for temperature, current minus previous day’s
temperature, relative humidity, pressure, extremes of hot
and cold (coldest and warmest 1% of days), day of the week, public
holiday (yes/no), and day after a public holiday(s) (yes/no). Rainfall
was also included in some investigational models.</p><p>The pollutant exposure was the average of the current and previous day. Changes
in admissions are shown for a one interquartile range (IQR) increase
in pollutant, using the mean IQR across cities. This makes the
increases from different pollutants more comparable. An IQR increase
can be thought of as the difference between a moderately good day and
a moderately bad day. The IQRs were 3.8 μg/m<sup>3</sup> for 24-hr PM<sub>2.5</sub>, 7.5 μg/m3 for 24-hr PM<sub>10</sub>, 5.1 ppb for 24-hr NO<sub>2</sub>, 0.9 ppm for 8-hr CO, and 8.8 ppb for 8-hr O<sub>3</sub>.</p><p>To estimate the average effect for all cities, we combined the estimates
across cities using a random effects meta-analysis (<xref rid="b13-ehp0114-001018" ref-type="bibr">Normand 1999</xref>) and quantified the differences (heterogeneity) between cities using the <italic>I</italic><sup>2</sup> statistic (<xref rid="b9-ehp0114-001018" ref-type="bibr">Higgins and Thompson 2002</xref>). <italic>I</italic>
<sup>2</sup> values > 80% indicate that differences between cities are high; > 50%, notable; > 20%, mild; and < 20%, small. To
test whether one city had an undue influence on the
meta-analysis, we used a leave-one-city-out sensitivity analysis (<xref rid="b13-ehp0114-001018" ref-type="bibr">Normand 1999</xref>).</p><p>We examined differences in the increases between cities using a hierarchical
model to incorporate variables that differ between cities and therefore
could modify the results (effect modifiers) (<xref rid="b7-ehp0114-001018" ref-type="bibr">Dominici et al. 2002a</xref>). The increases in admissions in each city were regressed against potential
city-level effect modifiers such as average pollutant level, temperature, and
percentage of the population ≥ 65 years of age. Differences
were examined only where there was notable heterogeneity (defined
by <italic>I</italic><sup>2</sup> > 50%).</p><p>When a health outcome showed a significant association with more than one
pollutant, we ran a multipollutant model using a matched case–crossover
approach (<xref rid="b20-ehp0114-001018" ref-type="bibr">Schwartz 2004</xref>). Matching is a traditional approach to control for potential confounding
in epidemiology. With control days that are both close in time to
the case day and also matched on the level of another pollutant, the effect
estimate cannot be confounded by the other pollutant. Matched control
days were defined as 24-hr PM<sub>2.5</sub> within 2 μg/m<sup>3</sup>, 24-hr PM<sub>10</sub> within 3 μg/m3, 24-hr NO<sub>2</sub> within 1 ppb, 24-hr CO within 0.5 ppm, and temperature within 1°C.</p><p>All analyses were conducted using SAS software (<xref rid="b19-ehp0114-001018" ref-type="bibr">SAS Institute Inc. 2001</xref>).</p><p>In the absence of an <italic>a priori</italic> opinion of which pollutants were important to health, we used a statistical
significance level of 5%, with no correction for multiple
comparisons. Although this increased the chances of finding spurious
associations, it reduced the chances of missing any important associations
during this early stage of investigation of the effects of air pollution
in Australia and New Zealand.</p><p>In this study we used monitoring data provided by the relevant monitoring
agency in each city. The data sets have been used without extensive
analysis or corrections beyond the basic quality control needed to ensure
data validity for the case–crossover analysis. Some data
sets were not fully used (e.g., the PM<sub>10</sub> data from Auckland) because they did not fully meet the strict requirements
of the study but are still regarded as valid data sets for the purposes
for which they were gathered.</p></sec></sec><sec sec-type="results"><title>Results</title><p>The associations between pollutants and cardiovascular hospital admissions
are shown in <xref ref-type="table" rid="t4-ehp0114-001018">Table 4</xref>. In the elderly, significant associations were found between the pollutants
CO, NO<sub>2</sub>, and PM and five categories of cardiovascular disease admissions. Arrhythmia
showed no associations in the elderly but did in the 15- to 64-year
age group. Stroke was the only disease category to show no associations
in either age group. O<sub>3</sub> was the only pollutant to show no associations.</p><p>In elderly admissions, the two largest statistically significant increases
were for cardiac failure, with a 6.9% increase for a 5.1-ppb
unit increase in NO<sub>2</sub> and a 6.0% increase for a 0.9-ppm increase in CO.</p><p>For the elderly age group, the relative risks for all cardiac admissions
associated with CO, NO<sub>2</sub>, PM<sub>2.5</sub>, and PM<sub>10</sub> are shown for each city and the meta-analysis in <xref ref-type="fig" rid="f1-ehp0114-001018">Figure 1</xref>, which highlights some of the differences in risk among the cities. This
heterogeneity is quantified by the <italic>I</italic><sup>2</sup> statistics in <xref ref-type="table" rid="t4-ehp0114-001018">Table 4</xref>. The <italic>I</italic><sup>2</sup> statistics indicate that more than half of the results had small heterogeneity. Notable
heterogeneity was more often observed in the elderly
group.</p><p><xref ref-type="table" rid="t5-ehp0114-001018">Table 5</xref> shows a much reduced <italic>I</italic><sup>2</sup> when Sydney was left out for the association between CO and cardiac admissions. <xref ref-type="fig" rid="f1-ehp0114-001018">Figure 1A</xref> shows that the association in Sydney was much larger than in the other
cities. The association was also larger in Perth, but the confidence
intervals (CIs) were wider. <xref ref-type="table" rid="t5-ehp0114-001018">Table 5</xref> and <xref ref-type="fig" rid="f1-ehp0114-001018">Figure 1B</xref> show that when Christchurch was left out, the association between NO<sub>2</sub> and cardiac admissions was similar for the remaining cities.</p><p>Statistically significant effect modifiers were found only for associations
with PM<sub>2.5</sub>. For cardiac admissions, there was a greater association with PM<sub>2.5</sub> in cities with less humidity. For cardiac failure, there was a greater
association with PM<sub>2.5</sub> in cities with higher pressure and a greater percentage of elderly.</p><p>Multipollutant results using a matched case–crossover analysis
are shown in <xref ref-type="table" rid="t6-ehp0114-001018">Table 6</xref>. None of the estimated increases changed greatly when cases and controls
were matched on temperature. The estimated increase due to NO<sub>2</sub> fell greatly when cases and controls were matched on CO.</p></sec><sec sec-type="discussion"><title>Discussion</title><sec><title>Cardiovascular admissions in the elderly</title><p>This study found many associations between air pollution and cardiovascular
admissions in cities in Australia and New Zealand. For every condition
but arrhythmia, the increases in hospital admissions were greater
in the elderly than in the younger age group (<xref ref-type="table" rid="t4-ehp0114-001018">Table 4</xref>), most likely because the elderly are a frailer population with probable
preexisting heart problems. The frailty of the elderly is also the
most likely reason that they did not show increases in arrhythmia. Arrhythmia
and cardiac failure are related conditions because atrial fibrillation
is a type of arrhythmia and may precipitate cardiac failure in
elderly people (<xref rid="b6-ehp0114-001018" ref-type="bibr">Cowie et al. 1999</xref>). Hence, exposure to NO<sub>2</sub> and CO that led to arrhythmia in the younger age group led to the more
serious condition of cardiac failure in the elderly.</p><p>We found associations at concentrations below normal air quality health
guidelines (<xref ref-type="table" rid="t3-ehp0114-001018">Table 3</xref>). This suggests that current air pollution guidelines need to be revised. There
is good reason to believe that lowering air pollution levels
would lead to improvements in cardiovascular health.</p><p>This results presented here are based on statistically significant findings. Although
this is not ideal practice, there is limited space in this
article; a complete set of results will be available in a forthcoming
report (Expansion of the Multi-City Mortality and Morbidity Study, National
Environment Protection Council). A non-statistically significant
association does not, of course, mean that a relationship does not
exist. This is particularly important for those admissions with smaller
numbers of events and hence less power (e.g., stroke in the younger
age group).</p></sec><sec><title>Differences among cities</title><p>Differences in the associations between cities in this study were mostly
not notable (<italic>I</italic><sup>2</sup> < 50%). This suggests that the relationship between exposure
and disease was often similar. There was more notable heterogeneity
in the elderly population, which is partly due to the greater size of
the associations in this age group.</p><p>In an attempt to explain the notable heterogeneity, we used effect-modifier
analyses. However, we found effect modifiers only for associations
with PM<sub>2.5</sub>. The effect of PM<sub>2.5</sub> in Australian cities depended on the percentage of the elderly and average
weather conditions. Less average humidity and higher average pressure
led to a greater association. To investigate these modifications
further, we reran the case–crossover models in each city including
an interaction term for 24-hr PM<sub>2.5</sub> and rainfall (results not shown). Higher rainfall led to a smaller association
between cardiovascular admissions and PM<sub>2.5</sub> in all four cities. This is not surprising, considering that rainfall
is a primary removal mechanism for PM<sub>10</sub> and PM<sub>2.5</sub>, but less so for gaseous pollutants.</p></sec><sec><title>Comparison with three other large studies</title><p>The aim and design of this study were similar to those of three other large
studies: the APHEA2 study of eight cities in Europe (<xref rid="b11-ehp0114-001018" ref-type="bibr">Le Tertre et al. 2002</xref>), NMMAPS with 14 U.S. cities (<xref rid="b16-ehp0114-001018" ref-type="bibr">Samet et al. 2000</xref>), and a Canadian study of 10 cities (<xref rid="b4-ehp0114-001018" ref-type="bibr">Burnett et al. 1997b</xref>). The results here for PM pollution in terms of elderly cardiac admissions
are similar to those found in APHEA2 and NMMAPS, and congestive heart
failure in the Canadian study. For example, we found the mean increase
for cardiac admissions for the 15–64-year age group to be
less than half that in the older group, a result similar to that of
the APHEA2 study. The APHEA2 study also found that the heterogeneity in
total cardiac admissions (all ages) was related to the percentage of
elderly. However, the results for the confounding effects on PM associations
by including CO are different (but PM was not monitored in every
city here).</p><p>A difference between this study and the three large multicity studies is
that emission sources for PM, and therefore the PM composition, differ. For
example, <xref rid="b5-ehp0114-001018" ref-type="bibr">Chan et al. (1999)</xref> found significantly higher contributions from sea salt and nonanthropogenic
crustal sources for both PM<sub>2.5</sub> and PM<sub>10</sub> in Brisbane than in overseas cities.</p><p>Another important difference from the other studies is in the statistical
methods used here. The NMMAPS and APHEA2 studies used generalized additive
models, in which confounding was estimated by adding the copollutant
into the model. In the APHEA2 study, the PM<sub>10</sub> associations were significantly reduced in the multipollutant models by
the inclusion of NO<sub>2</sub> (as found here) and slightly (but becoming statistically insignificant) for
CO. However, using black smoke to estimate the PM impacts showed
no confounding by CO and much less by NO<sub>2</sub>. There was no significant confounding of the PM associations by CO or
NO<sub>2</sub> found in NMMAPS.</p></sec><sec><title>Addressing confounding between pollutants</title><p>Instead of using a multipollutant model, we controlled for confounding
by matching in the case–crossover analysis. For elderly hospital
admissions, the CO associations remained of a similar size when matched
with NO<sub>2</sub>. Conversely, the NO<sub>2</sub> became smaller when matched with CO (<xref ref-type="table" rid="t6-ehp0114-001018">Table 6</xref>).</p><p>Matching was also used to control for the important confounder of temperature. The
results changed little when matched on temperature, strongly
suggesting that the association between air pollution and cardiovascular
disease is not confounded by temperature.</p></sec><sec><title>Is outdoor air pollution a good indicator of exposure?</title><p>A problem in interpreting the results from this study is that it used outdoor
air pollution concentrations measured at fixed-point monitors (ambient
concentrations), whereas people spend most of their time indoors. Recent
studies in Baltimore, Maryland (<xref rid="b18-ehp0114-001018" ref-type="bibr">Sarnat et al. 2001</xref>) and Boston, Massachusetts (<xref rid="b17-ehp0114-001018" ref-type="bibr">Sarnat et al. 2005</xref>) indicate that such ambient concentrations may be poor surrogates for
actual exposure to air pollution, especially in winter when buildings
are more sealed. However, winters in Australia are mild, meaning that
people will likely spend more time outdoors and that houses are designed
to lose heat rather than trap it. Hence, exposure to the air may be
high all year round in Australia (winter exposure in New Zealand may
be more similar to that in Baltimore). A similar conclusion was drawn
by a study of the effects of cold temperatures on cardiovascular disease (<xref rid="b1-ehp0114-001018" ref-type="bibr">Barnett et al. 2005</xref>). In that study, regions with mild winters showed greater increases in
cold-related cardiovascular events than did regions with usually cold
winters.</p><p>The study in Baltimore also found that ambient concentrations for CO and
NO<sub>2</sub> were often better surrogates for actual exposure to PM than to CO and
NO<sub>2</sub>, especially in summer (<xref rid="b18-ehp0114-001018" ref-type="bibr">Sarnat et al. 2001</xref>). However, the more recent Boston study did note that there were some
correlations between ambient concentrations and actual exposure to these
gases in summer (<xref rid="b17-ehp0114-001018" ref-type="bibr">Sarnat et al. 2005</xref>). Outdoor concentrations for pollutants such as NO<sub>2</sub>, CO, and PM often arise from the same combustion emissions sources, such
as motor exhausts.</p></sec><sec><title>CO as a marker for pollution sources</title><p>There is evidence that air pollutants (NO<sub>2</sub>, CO) may trigger fibrillation in people with a history of serious arrhythmia (<xref rid="b14-ehp0114-001018" ref-type="bibr">Peters et al. 2000</xref>). The effect of CO on cardiovascular disease is well known, with CO replacing
oxygen in the blood stream, but at the low CO concentrations prevailing
in the cities under study, it cannot be simply assumed that
if CO is the “cause” of any effects found here, it is
due to this mechanism. The associations found for CO, NO<sub>2</sub>, and PM are not additive, but probably refer to the impacts of a similar
pollutant “mix.” Given that the CO associations show
the least change when matched with the other pollutants (<xref ref-type="table" rid="t6-ehp0114-001018">Table 6</xref>), this indicates that the air pollutant mixture arising from emission
sources dominating the CO emissions (usually human combustion sources) is
the primary cause of the association, not the effect of CO itself.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>For both Australian and New Zealand cities, the results show that increases
in outdoor concentrations of CO, NO<sub>2</sub>, and PM have significant associations with increases in cardiovascular
admissions for adults, especially the elderly (≥ 65 years of
age). Associations were found at concentrations below normal air quality
health guidelines. There were significant associations between air
pollution and arrhythmia admissions in the younger age group, which were
not apparent for the elderly. For the elderly, there were significant
associations between air pollution increases and increases in hospital
admissions for ischemic heart disease and myocardial infarction, and
these were not apparent for the younger group. Atrial fibrillation
can precipitate cardiac failure, especially in the elderly, and a significant
relationship has been identified here in the adult age group (15–64 years) between increases in hospital admissions for arrhythmia
and increases in air pollution.</p><p>The associations for NO<sub>2</sub> appear to be stronger in Australian than in New Zealand cities, whereas
those of CO are similar for cities in both countries. In Australian
cities, PM<sub>10</sub> and PM<sub>2.5</sub> had a similar association, apart from that for arrhythmia. These PM<sub>2.5</sub> associations differed among cities due to different climate conditions
for humidity (the lower the humidity, the greater the association).</p><p>It is difficult to separate the associations for different pollutants because
there are common emission sources for CO, NO<sub>2</sub>, and PM (e.g., motor vehicle exhausts). Also, outdoor concentrations are
often not good surrogates for actual exposure, with outdoor levels
for the gases sometimes being good surrogates for actual exposure to PM, especially
in summer.</p></sec>
|
Meeting Report: National Workshops for the Communication of Air Pollution
and Health Information: Summary of Four Workshops in Different Regions
of Europe
|
<p>AIRNET was a thematic network project (2002–2004) initiated to
stimulate the interaction between researchers in air pollution and health
in Europe. As part of AIRNET’s communication strategy, a standardized
workshop model was developed to organize national meetings
on air pollution and health (AIRNET network days). Emphasis was given
to tailor the national workshop information and related activities to
the specific needs of a wider range of stakeholders (e.g., policy makers, nongovernmental
organizations, industry representatives). In this
report we present an overview of the results of four workshops held in
western, northern, central/eastern, and southern regions of Europe in 2004. Overall, workshop
experiences indicated that by actively involving
participants in the planning of each meeting, AIRNET helped create
an event that addressed participants’ needs and interests. A
wide range of communication formats used to discuss air pollution and
health also helped stimulate active interaction among participants. Overall, the
national workshops held by AIRNET offered a way to improve
communication among the different stakeholders. Because a broad stakeholder
involvement in decision making can positively affect the development
of widely supported policies, such meetings should be continued
for Europe and elsewhere.</p>
|
<contrib contrib-type="author"><name><surname>Sanderson</surname><given-names>Eric Gordon</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001108">1</xref></contrib><contrib contrib-type="author"><name><surname>Fudge</surname><given-names>Nina</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001108">2</xref></contrib><contrib contrib-type="author"><name><surname>Totlandsdal</surname><given-names>Annike Irene</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001108">2</xref></contrib><contrib contrib-type="author"><name><surname>Hovelynck</surname><given-names>Ingrid</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001108">3</xref></contrib><contrib contrib-type="author"><name><surname>Korbee</surname><given-names>Herbert</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001108">3</xref></contrib><contrib contrib-type="author"><name><surname>Rameckers</surname><given-names>Edith</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001108">4</xref></contrib><contrib contrib-type="author"><name><surname>Brunekreef</surname><given-names>Bert</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001108">1</xref></contrib><contrib contrib-type="author"><name><surname>van Bree</surname><given-names>Leendert</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001108">2</xref></contrib>
|
Environmental Health Perspectives
|
<p>A well-established body of evidence now shows that increasing levels of
air pollution are linked with more illness, higher use of health services, and
earlier death among the exposed population groups (<xref rid="b9-ehp0114-001108" ref-type="bibr">Brunekreef and Holgate 2002</xref>). Recently, five disciplinary reports by AIRNET (Thematic Network on Air
Pollution and Health) have addressed the evidence in the European Union (EU) from
a variety of scientific perspectives, including epidemiology, toxicology, exposure assessment, health impact assessment, and
the science–policy interface (<xref rid="b2-ehp0114-001108" ref-type="bibr">AIRNET 2005a</xref>, <xref rid="b3-ehp0114-001108" ref-type="bibr">2005b</xref>, <xref rid="b4-ehp0114-001108" ref-type="bibr">2005c</xref>, <xref rid="b5-ehp0114-001108" ref-type="bibr">2005d</xref>, <xref rid="b6-ehp0114-001108" ref-type="bibr">2005e</xref>). Overall, these reports indicate that European research has significantly
contributed to the better understanding of air pollution health effects.</p><p>AIRNET was a thematic network project (2002–2004) initiated to
stimulate the interaction between air pollution and health researchers
in Europe (<xref rid="b1-ehp0114-001108" ref-type="bibr">AIRNET 2002</xref>). AIRNET collected, interpreted, and disseminated information from individual (EU-funded) projects to strengthen the science–policy
interface and to draw policy-relevant recommendations. The objective of
this stakeholder network was to create a widely supported basis for
public health policy related to improving air quality in Europe—for
instance, the communication of scientific findings for policy use
and the identification of important gaps in the research. Overall, 23 project
partners were initially brought into AIRNET, representing the
scientific community and a variety of other stakeholders with an interest
in air pollution and health.</p><p>Several reports stress the importance of stakeholder involvement in understanding
the science at all stages of the decision-making process (<xref rid="b7-ehp0114-001108" ref-type="bibr">Beierle 2002</xref>; <xref rid="b15-ehp0114-001108" ref-type="bibr">Maynard et al. 2003</xref>; <xref rid="b19-ehp0114-001108" ref-type="bibr">Tamburlini and Ebi 2002</xref>). Realizing the need for more stakeholder input, AIRNET strived to increase
the number and diversity of participating stakeholders with varied
interests deriving from a local, national, or regional perspective. To
make the wealth of gathered and interpreted information available
to a broader spectrum of stakeholders, two things were considered paramount: first, a
fine-tuning of the information required to meet the needs
of different stakeholders; and second, a well-focused effort undertaken
to actively involve more stakeholders, including those who previously
might not have had any contact with AIRNET.</p><p>Therefore, a major goal of the activities of AIRNET in its final year was
to help bridge the gap between scientists, policy makers, and other
relevant stakeholders. To this end, the communication strategy focused
on the concept of national workshops (AIRNET network days). The workshop
model gave the participants an opportunity to influence the planning
of the meeting in line with their interests and needs. Ideally, such
an approach should produce an atmosphere where stakeholders can comfortably
create, broaden, and intensify their own personal network and
can share their knowledge and questions. In this report we present the
findings from four workshops organized to communicate and discuss air
pollution and health issues specific to western, northern, central/eastern, and
southern regions of Europe.</p><sec sec-type="methods"><title>Methods</title><p>Four countries (the Netherlands, Sweden, Hungary, and Spain) representing
different European regions (western, northern, central/ eastern and
southern) were selected to address region-specific air pollution and
health issues in a standardized workshop format developed by AIRNET’s
communication firm (Korbee & Hovelynck BV). Although the underlying
approach used to organize the national workshops is a traditional
management strategy, it is a little-used strategy in many scientific
areas, especially for air pollution and health.</p><p>As illustrated in <xref ref-type="fig" rid="f1-ehp0114-001108">Figure 1</xref>, the first step was for a national AIRNET coordinator (i.e., a scientist
or a representative from a government agency) to select a local communication
agency that could perform a stakeholder analysis to identify
relevant target groups according to their interests in air pollution
and health. The communication agency chosen was either a commercial public
relations firm or a professional conference management firm with
suitable experience in the field of public relations. Once the stakeholder
list was compiled and preliminary invitations to the workshop were
sent out, focus group discussions or interviews were held with representative
stakeholders. The local communication agency organized these
sessions and aimed to have stakeholder input from each stakeholder group
identified in the stakeholder analysis.</p><p>The goal of the focus group discussions and interviews before the workshop
was to understand what the stakeholders needed, how they could contribute
to the meeting, and what the preferred means were for communicating
and exchanging knowledge and opinions. There was also an opportunity
to widen participation by asking stakeholders for the names of other
interested parties who may have been missed in the initial stakeholder
analysis. As the national workshops were held at different times
throughout the year, we were able to build on the experiences and results
of previous workshops to help develop subsequent events.</p><p>All the authors have been involved in the planning and participation of
one or more of the workshops. For the overall descriptions, discussion, and
evaluation of the workshops, the authors draw on their experiences
and observations as well as any informal discussion with the participants.</p></sec><sec><title>Results and Discussion</title><sec><title>Stakeholder participation</title><p>Participants at the workshops were classified into several stakeholder
categories (<xref ref-type="table" rid="t1-ehp0114-001108">Table 1</xref>): scientists (i.e., air quality, health) who perform research, policy
makers (local, regional, national), industry representatives (i.e., automobile, oil, and
gas), and nongovernmental organizations (NGOs) (e.g., patient
rights, public health, and the environment). Additional stakeholders
included participants who represented public transportation
operators and clean fuel companies, and NGOs that advocated for public
transport and cycling. Except at the Netherlands workshop, researchers
were represented in the highest percentage. In general, policy makers
were second, followed by NGOs and industry representatives—all
of whom use research findings. Overall, AIRNET appears to have achieved
a wide range of stakeholder participation at the national workshops
between producers and users of research.</p><p>In general, the policy makers who attended the workshops represented national
ministries dealing with the environment, meteorology and climatology, health
care, and traffic. Other than those from a few of the municipalities
where the meetings took place, few regional or municipal
policy makers were present at the workshops. On the whole, the NGOs participating
in the workshops represented a wide range of stakeholders, from
consumer groups, environmental protection and management and environmental
law, to environmental health advocacy for susceptible individuals (i.e., asthmatics
and children). These NGO groups were typically
functioning at the national interest level, sometimes under the umbrella
of an international or pan-European parent organization.</p><p>Some stakeholder groups mentioned that international conferences (objectives, themes, content) are often biased toward researchers, making it
less attractive for nonscientists to participate. Furthermore, some stakeholders
groups (e.g., NGOs and local policy makers) find it difficult
to attend international conferences because of budgetary and time
constraints. This was demonstrated in the yearly AIRNET conferences, which
were attended predominantly by scientists (<xref ref-type="table" rid="t1-ehp0114-001108">Table 1</xref>). However, by offering local events where the attendee is involved in
the design and setup, AIRNET has shown a way to increase the diversity
of participation (see also <xref rid="b11-ehp0114-001108" ref-type="bibr">Huntington et al. 2002</xref>). We are confident that workshops tailored to the participants’ interests
increased the level of participation from all stakeholders, thereby
demonstrating their potential usefulness as a medium through
which to help develop consensus on research or policy.</p></sec><sec><title>Workshop communication formats</title><p>The available work formats varied little from country to country, despite
the fact that stakeholders were encouraged to indicate their preferred
methods of communication. As indicated by the preworkshop focus group
discussions and stakeholder interviews, the use of conventional presentation
formats (seminar presentations, poster presentations) and roundtable
discussions were favored. At three of the workshops, nonconventional
activities (silent wall discussions, speaker’s corner, literature
table, events calendar, contact board) were also used to
stimulate stakeholder participation. Overall, every effort was made to
ensure that the messages were relevant and easily understood to help
stimulate stakeholder dialogue.</p><p>By using the proper meeting format, sharing knowledge can become more effective, widespread, and
fine-tuned to meet different stakeholder needs (<xref rid="b11-ehp0114-001108" ref-type="bibr">Huntington et al. 2002</xref>). However, the importance of selecting the most appropriate communication
or work format for the intended audience is often overlooked by meeting
organizers. In some instances, the use of conventional formats may
be too passive to promote discussion. However, combined with more interactive
methods, conventional formats can provide the information needed
to fuel conversation. For example, silent wall discussions (reacting
in writing to a statement on a blank poster) or a speaker’s
corner presentation (analogous to a soapbox speech in London’s
Hyde Park) can encourage people to become more active and allow alternative
ways to participate.</p><p>Because of potential differences in air pollution and health issues and
the communication styles among European regions, emphasis was given to
tailoring information and work formats to the needs of the target groups. The
workshops were also held in the national language, effectively
removing potential barriers to communication caused by language. AIRNET’s
experiences at each national workshop suggested that the
selected work formats, room setups, chosen moderator, and rules for
roundtable discussions were of prime importance in helping the stakeholder
feel comfortable in contributing to the discussion. Overall, participants
at the workshops reaffirmed the need to encourage successful
two-way dialogue between stakeholders through both conventional and nonconventional
communication methods.</p></sec><sec><title>Major themes of stakeholder interest</title><p>Interviews and focus group discussions can provide better insight on stakeholder
questions (<xref rid="b14-ehp0114-001108" ref-type="bibr">Lion et al. 2002</xref>). Our focus group discussions and interviews produced a list of themes
for workshop agendas that varied slightly by country (<xref ref-type="table" rid="t2-ehp0114-001108">Table 2</xref>). The most prominent theme for all workshops was traffic-related air pollution
and human health. Except for Spain, air quality standards were
also of major interest. In addition, issues on asthma and allergy, as
well as child/infant health, were a major focus for three of the five
workshops. Except for Hungary, policy options aimed at air pollution
and health were included in the program.</p><p>These initial themes were used to help promote attendance at each workshop
but not necessarily to drive the direction of the discussion among
participants. In the end, the Netherlands and Hungary roundtable discussions
focused on the need to <italic>a</italic>) increase and improve public transportation, and <italic>b</italic>) encourage the public to take environmentally friendly steps to reduce
the volume of traffic. Similarly, participants from the Swedish workshop
indicated that health-orientated decision making would benefit from <italic>a</italic>) the development of traffic-related indicators of air quality, <italic>b</italic>) acute and chronic health effect studies for traffic, and <italic>c</italic>) integration of traffic and health policy with policies for air pollution
reduction. For more detailed summaries of each workshop, see the AIRNET
Web site (<xref rid="b1-ehp0114-001108" ref-type="bibr">AIRNET 2002</xref>).</p><p>Although the workshop themes varied little among countries (<xref ref-type="table" rid="t2-ehp0114-001108">Table 2</xref>), key messages emerging from each workshop were different in scope (<xref ref-type="app" rid="app1-ehp0114-001108">Appendix 1</xref>). For instance, the effects of wood burning and spring dust were important
topics in Sweden. In Hungary, many of the discussions focused on
the health effects of ragweed exposure. Participants at the Netherlands
workshop placed a greater focus on actions for the government and policy
makers. In Spain, a dry climate resulting in dust production was
an issue of importance among the attendees.</p></sec><sec><title>European scope of the national workshops</title><p>A geographical spread (northern, western, central/eastern, and southern
Europe) of national workshops allowed discussion of air pollution and
health issues specific to each region. This aided in attracting policy
makers and other stakeholders (e.g., environmental and health organizations) working
at the local, regional, or national level (<xref ref-type="table" rid="t1-ehp0114-001108">Table 1</xref>). A broad diversity of stakeholder perspectives helps improve decisions
over the status quo by adding new information and ideas while ensuring
adequate access to resources (<xref rid="b7-ehp0114-001108" ref-type="bibr">Beierle 2002</xref>).</p><p>As a converging point for the national workshop activities on a European
and regional level, answers to several questions posed during the parallel
breakout sessions at the Third AIRNET conference are listed in <xref ref-type="app" rid="app2-ehp0114-001108">Appendix 2</xref>. At this pan-European meeting, discussions relating to policy and decision-making
priorities and the value of national or regional meetings
reinforced AIRNET network activities (workgroup meetings, conferences, AIRNET
network days). Overall, for all regions of Europe, improved communication
between scientists, decision makers, and stakeholders was
seen by the participants as highly desirable to increase the effectiveness
of decision-making processes for environmental health improvement.</p></sec><sec><title>Participant feedback</title><p>Participant feedback from the workshops was positive. On a scale of 1 to 10 (where 1 is
bad and 10 is good), the overall ratings by participants
in the Netherlands, Sweden, and Spain were 7.9, 7.2, and 8.1, respectively. (No
rating was available for Hungary.) Most participants at
each workshop felt that the objectives of the day, to exchange knowledge
and strengthen personal networks, were well achieved. Moreover, participants
were positive about the work formats used (specifically the
roundtable discussions), despite not having worked in such formats before. Overall, most
participants felt that it would be valuable to hold
events of this type in the future, providing valuable feedback for the
organizers.</p><p>The feedback sheets from the workshops contained numerous suggestions and
ideas for the future, some of which are summarized below:</p><list list-type="bullet"><list-item><p>Events should be longer and contain more scientific lectures and discussions.</p></list-item><list-item><p>Circle of participants should be wider.</p></list-item><list-item><p>Such events should be continued, where different sectors, interest groups, and
stakeholders communicate with each other and with the public.</p></list-item><list-item><p>Students dealing with health issues and protection of the environment should
participate.</p></list-item></list></sec><sec><title>Usefulness of national workshops</title><p>A dynamic science–stakeholder–policy interplay is needed
to achieve successful air pollution abatement measures to decrease health
risks. This interplay is important when developing sustainable policies
that are transparent and sound and that carry the support of the
policy makers, researchers, other stakeholders (industry and NGOs), and
the general public. AIRNET attempted to incorporate the needs and
views of all players involved during its 3-year existence. To help get
this interplay running, the national workshops were used to improve
communication between all players and to better understand each other’s
needs.</p><p>Traditionally, expert workshops in Europe and internationally have been
used to discuss the scientific and policy issues related to ambient air
quality and human health [<xref rid="b8-ehp0114-001108" ref-type="bibr">Bell et al. 2002</xref>; <xref rid="b16-ehp0114-001108" ref-type="bibr">O’Neill et al. 2003</xref>; <xref rid="b20-ehp0114-001108" ref-type="bibr">World Health Organization (WHO) 2000</xref>, <xref rid="b21-ehp0114-001108" ref-type="bibr">2001</xref>, <xref rid="b22-ehp0114-001108" ref-type="bibr">2003</xref>]. Agencies, and the WHO in particular, regularly perform expert
reviews of air pollution and health information (<xref rid="b20-ehp0114-001108" ref-type="bibr">WHO 2000</xref>, <xref rid="b21-ehp0114-001108" ref-type="bibr">2001</xref>, <xref rid="b22-ehp0114-001108" ref-type="bibr">2003</xref>). In comparison, relatively few national workshops are organized by involving
the stakeholders in the planning of the meeting itself. Moreover, there
is a scarcity of publications offering clear guidance and suggestions
on how to organize and conduct multiparticipatory workshops
for a heterogeneous group of individuals. The workshops organized by AIRNET
included a wide range of stakeholders. Crucially, a representative
subset of the invited stakeholders was involved in the development
of the meeting itself.</p><p>By organizing events aimed at bringing a diverse audience together, AIRNET
gave stakeholders a responsibility to create and broaden their own
personal network, and to share their knowledge and questions. The national
workshops were a good opportunity for stakeholders to try out different
modes of communication. Overall, we believe that the workshops
held by AIRNET were a step forward in stakeholder engagement in the field
of air pollution and health and a clear response to discussions from
previous meetings on the use of scientific knowledge in decision making (<xref rid="b10-ehp0114-001108" ref-type="bibr">Ginsburg and Cowling 2003</xref>; <xref rid="b17-ehp0114-001108" ref-type="bibr">Samet and Lee 2001</xref>).</p></sec></sec><sec><title>Concluding Remarks</title><p>How will such workshop activities continue? Who has the time and inclination
to organize them? How is discussion translated into action?</p><p>Answers to the above questions are challenging, given AIRNET’s
initial focus and the relatively short mandate given by the EU (3 years). Nonetheless, these
issues were dominant at the final AIRNET conference, where
attendees clearly wanted events that enhanced participant
interaction (<xref ref-type="app" rid="app2-ehp0114-001108">Appendix 2</xref>). However, since AIRNET officially ceased to exist in 2005, other supporters (EU, national
governments, NGOs, industry, etc.) will need to take
the initiative and time to organize, sponsor, and promote similar
events. Doing so may help ensure that future actions are taken by maintaining
a direct link with those who will be making the decisions. Although
the issues are often big and the meeting times relatively short, events
such as national workshops can be seen as a beginning or continuation
of the existing dialogue and debate.</p><p>A limitation of the workshops (which were not initially part of AIRNET’s
planned activities) is that they did not examine the steps needed
to achieve subsequent actions resulting from the workshops. However, a
legacy of these activities is the creation of a stakeholder network
that will continue to interact (at some level) through other means. For
example, many of the participants of the workshop in Spain decided
that they would continue to communicate and work toward collectively
feeding into policy debate in their country. Reducing environmental
exposures may also require substantial financial investment, where broad
support by a variety of stakeholders can be achieved through meaningful, relevant, and
understandable communication. For a variety of reasons, this
publication of the meetings summary is a step toward action
rather than just keeping ideas merely at the level of discussion. This
is important, if not critical, for comprehensive and sound management
of any real or perceived risk to the human health (<xref rid="b12-ehp0114-001108" ref-type="bibr">Jardine et al. 2003</xref>).</p><p>Well-planned and -moderated workshops can enhance communication and knowledge
sharing among individuals who do not know each other well (or at
all) and have different levels of understanding (<xref rid="b11-ehp0114-001108" ref-type="bibr">Huntington et al. 2002</xref>; <xref rid="b13-ehp0114-001108" ref-type="bibr">Kontic et al. 2006</xref>). Through activities such as the AIRNET national workshops, we believe
that a substantial contribution to research planning or influencing policy
can be achieved by ensuring that</p><list list-type="bullet"><list-item><p>Stakeholders are familiar with the extent of the knowledge base (and its
limitations or gaps) and how to gain access to this information.</p></list-item><list-item><p>Stakeholders are able to use the information (available in a suitable format) for
practical application in their own fields of specialty (Sanderson
et al., in press).</p></list-item><list-item><p>Stakeholders know whom to turn to with specific questions and will do so
actively.</p></list-item><list-item><p>Stakeholders know with whom they can share their acquired knowledge to
maximize the impact of their efforts and help others in their pursuits.</p></list-item><list-item><p>Stakeholders have a sufficient understanding of the subject matter under
policy scrutiny to make a constructive and positive contribution to
the decision-making process.</p></list-item></list><p>In conclusion, we feel that the national workshops were highly valuable
in promoting participant interaction and improving communication among
a wide range of stakeholders. Herein, active participation is key to
enable a two-way flow of information. By bringing together the relevant
stakeholders, well-planned workshops can empower a group of individuals
who share a common interest or vision to participate collectively
in the policy debate.</p></sec>
|
Coal Home Heating and Environmental Tobacco Smoke in Relation to Lower
Respiratory Illness in Czech Children, from Birth to 3 Years of Age
|
<sec><title>Objective</title><p>The objective of this study was to evaluate how indoor pollution from tobacco
and home heating may adversely affect respiratory health in young
children.</p></sec><sec><title>Design</title><p>A birth cohort was followed longitudinally for 3 years to determine incidence
of lower respiratory illness (LRI).</p></sec><sec><title>Participants</title><p>A total of 452 children born 1994–1996 in two districts in the
Czech Republic participated.</p></sec><sec><title>Evaluations</title><p>Indoor combustion exposures were home heating and cooking fuel, mother’s
smoking during pregnancy, and other adult smokers in the household. Diagnoses
of LRI (primarily acute bronchitis) from birth to 3 years
of age were abstracted from pediatric records. Questionnaires completed
at delivery and at 3-year follow-up provided covariate information. LRI
incidence rates were modeled with generalized linear models
adjusting for repeated measures and for numerous potential confounders.</p></sec><sec><title>Results</title><p>LRI diagnoses occurred more frequently in children from homes heated by
coal [vs. other energy sources or distant furnaces; rate ratio (RR) = 1.45; 95% confidence interval (CI), 1.07–1.97]. Maternal prenatal smoking and other adult smokers also
increased LRI rates (respectively: RR = 1.48; 95% CI, 1.10–2.01; and
RR = 1.29; 95% CI, 1.01–1.65). Cooking
fuels (primarily electricity, natural gas, or propane) were
not associated with LRI incidence. For children never breast-fed, coal
home heating and mother’s smoking conferred substantially
greater risks: RR = 2.77 (95% CI, 1.45–5.27) and
RR = 2.52 (95% CI, 1.31–4.85), respectively.</p></sec><sec><title>Conclusions</title><p>Maternal smoking and coal home heating increased risk for LRI in the first 3 years
of life, particularly in children not breast-fed.</p></sec><sec><title>Relevance</title><p>Few studies have described effects of coal heating fuel on children’s
health in a Western country. Breast-feeding may attenuate adverse
effects of prenatal and childhood exposures to combustion products.</p></sec>
|
<contrib contrib-type="author"><name><surname>Baker (posthumous)</surname><given-names>Rebecca J.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001126">1</xref></contrib><contrib contrib-type="author"><name><surname>Hertz-Picciotto</surname><given-names>Irva</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001126">2</xref></contrib><contrib contrib-type="author"><name><surname>Dostál</surname><given-names>Miroslav</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001126">3</xref></contrib><contrib contrib-type="author"><name><surname>Keller</surname><given-names>Jean A.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001126">1</xref></contrib><contrib contrib-type="author"><name><surname>Nožička</surname><given-names>Jiři</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001126">4</xref></contrib><contrib contrib-type="author"><name><surname>Kotìšovec</surname><given-names>František</given-names></name><xref ref-type="aff" rid="af5-ehp0114-001126">5</xref></contrib><contrib contrib-type="author"><name><surname>Dejmek (posthumous)</surname><given-names>Jan</given-names></name></contrib><contrib contrib-type="author"><name><surname>Loomis</surname><given-names>Dana</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001126">1</xref><xref ref-type="aff" rid="af6-ehp0114-001126">6</xref></contrib><contrib contrib-type="author"><name><surname>Šrám</surname><given-names>Radim J.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001126">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>Children < 3 years of age, and especially in their first year, are at
greatest risk of serious respiratory illnesses (<xref rid="b35-ehp0114-001126" ref-type="bibr">Phelan et al. 1994</xref>). Studies from industrialized areas indicate that the rate of childhood
lower respiratory illness (LRI) peaks during the first year of life. Globally, particularly
in developing countries, acute lower respiratory
infections are the most important cause of death among children < 5 years
of age, accounting for about two million deaths per year (<xref rid="b6-ehp0114-001126" ref-type="bibr">Bruce et al. 2000</xref>).</p><p>Recently, increasing attention has focused on effects of indoor air exposures
on respiratory health of children (<xref rid="b7-ehp0114-001126" ref-type="bibr">Bruce et al. 2002</xref>; <xref rid="b10-ehp0114-001126" ref-type="bibr">Chen et al. 1990</xref>; <xref rid="b16-ehp0114-001126" ref-type="bibr">Ezzati and Kammen 2002</xref>; <xref rid="b25-ehp0114-001126" ref-type="bibr">Honicky and Osborne 1991</xref>; <xref rid="b31-ehp0114-001126" ref-type="bibr">Marbury et al. 1996</xref>; <xref rid="b42-ehp0114-001126" ref-type="bibr">Smith et al. 2000</xref>). These exposures include combustion products; semivolatile and volatile
organic compounds released by furnishings, building materials, and
chemical products; pollutants from volatilization of chemicals in soil; and
pollutants generated by decomposition of organic matter (<xref rid="b42-ehp0114-001126" ref-type="bibr">Smith et al. 2000</xref>). The main indoor pollutants from combustion are carbon monoxide, nitrogen
oxides, sulfurous oxides, particles, and volatile organics. In developing
countries, early childhood respiratory illness has been associated
with use of biomass or coal as fuel for heating and cooking (<xref rid="b7-ehp0114-001126" ref-type="bibr">Bruce et al. 2002</xref>; <xref rid="b16-ehp0114-001126" ref-type="bibr">Ezzati and Kammen 2002</xref>; <xref rid="b42-ehp0114-001126" ref-type="bibr">Smith et al. 2000</xref>). In more developed countries, reports on respiratory health effects of
combustion emissions from wood (<xref rid="b25-ehp0114-001126" ref-type="bibr">Honicky and Osborne 1991</xref>; <xref rid="b24-ehp0114-001126" ref-type="bibr">Honicky et al. 1983</xref>; <xref rid="b48-ehp0114-001126" ref-type="bibr">Tuthill 1984</xref>) and natural gas (<xref rid="b4-ehp0114-001126" ref-type="bibr">Braun-Fahrlander et al. 1992</xref>; <xref rid="b17-ehp0114-001126" ref-type="bibr">Farrow et al. 1997</xref>; <xref rid="b41-ehp0114-001126" ref-type="bibr">Samet et al. 1993</xref>) in the home have been mixed, whereas recent studies examining coal home
heating have found increased reports of respiratory symptoms such as
phlegm and cough in adults (<xref rid="b37-ehp0114-001126" ref-type="bibr">Pope and Xu 1993</xref>) and school children (<xref rid="b26-ehp0114-001126" ref-type="bibr">Jedrychowski et al. 1998</xref>; <xref rid="b38-ehp0114-001126" ref-type="bibr">Qian et al. 2004a</xref>, <xref rid="b39-ehp0114-001126" ref-type="bibr">2004b</xref>). Few studies, if any, have been conducted in Western countries using
physician diagnoses for respiratory illnesses (other than asthma) in early
childhood. Associations between environmental tobacco smoke (ETS) and
children’s respiratory illnesses are well established (<xref rid="b29-ehp0114-001126" ref-type="bibr">Li et al. 1999</xref>; <xref rid="b47-ehp0114-001126" ref-type="bibr">Strachan and Cook 1998</xref>).</p><p>Heightened susceptibility of infants and young children to environmental
toxicants has been suggested. Inhaled pollutant dose per unit body weight
is likely to be greater, because their body weight is smaller and
their respiratory rate higher than in adults (<xref rid="b9-ehp0114-001126" ref-type="bibr">Cerna et al. 1998</xref>; <xref rid="b18-ehp0114-001126" ref-type="bibr">Gilliland et al. 1999</xref>). Infants’ developing lungs and immune system put them at greater
risk for respiratory infections and may make them particularly vulnerable
to inhaled pollutants’ irritative and immune effects (<xref rid="b27-ehp0114-001126" ref-type="bibr">Koenig 2000</xref>; <xref rid="b35-ehp0114-001126" ref-type="bibr">Phelan et al. 1994</xref>), especially in the first year of life while the lungs are still maturing (<xref rid="b3-ehp0114-001126" ref-type="bibr">Braga et al. 2001</xref>). In air pollution studies, infant mortality is associated with ambient
particle concentrations (<xref rid="b2-ehp0114-001126" ref-type="bibr">Bobak and Leon 1999</xref>; <xref rid="b30-ehp0114-001126" ref-type="bibr">Loomis et al. 1999</xref>; <xref rid="b34-ehp0114-001126" ref-type="bibr">Penna and Duchiade 1991</xref>; <xref rid="b40-ehp0114-001126" ref-type="bibr">Saldiva et al. 1994</xref>; <xref rid="b51-ehp0114-001126" ref-type="bibr">Woodruff et al. 1997</xref>). Breast-feeding protects against respiratory infections, particularly
in the first year of life (<xref rid="b21-ehp0114-001126" ref-type="bibr">Heinig 2001</xref>; <xref rid="b35-ehp0114-001126" ref-type="bibr">Phelan et al. 1994</xref>). Breast-feeding also may modify adverse effects of ETS exposure on respiratory
illnesses (<xref rid="b11-ehp0114-001126" ref-type="bibr">Chulada et al. 2003</xref>; <xref rid="b33-ehp0114-001126" ref-type="bibr">Nafstad and Jaakkola 2003</xref>; <xref rid="b32-ehp0114-001126" ref-type="bibr">Nafstad et al. 1996</xref>; <xref rid="b52-ehp0114-001126" ref-type="bibr">Woodward et al. 1990</xref>).</p><p>This study, part of a large program of research on air pollution health
effects in the Czech Republic (Teplice Program), focused on respiratory
illnesses in young children in relation to indoor combustion of cigarettes
and of coal, wood, natural gas, and propane for heating or cooking. A
high proportion of adults smoke in the Czech Republic, and coal
is still used for heating of some homes; hence, this setting is advantageous
for studying indoor air exposures.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Study population</title><p>We followed a birth cohort to 3 years of age in two districts in the Czech
Republic: Teplice in the northwest, and Prachatice in the southwest. The
Teplice district, with a population of about 130,000, is infamous
for high air pollution due to power plants and home heating with coal. The
Prachatice district has a population of about 50,000 and no power
plants. Children born between May 1994 and December 1996 in the two
districts were eligible for this study.</p><p>The evolution of our study sample is shown in <xref ref-type="fig" rid="f1-ehp0114-001126">Figure 1</xref>. As part of the Teplice program (<xref rid="b43-ehp0114-001126" ref-type="bibr">Šrám et al. 1996</xref>), <xref rid="b13-ehp0114-001126" ref-type="bibr">Dejmek et al. (1999</xref>, <xref rid="b12-ehp0114-001126" ref-type="bibr">2000a</xref>, <xref rid="b14-ehp0114-001126" ref-type="bibr">2000b</xref>, <xref rid="b15-ehp0114-001126" ref-type="bibr">2002</xref>) enrolled 4,339 mother–infant pairs at birth into a study of pregnancy
outcomes and air pollution, the pregnancy outcome study (POS). Only
singleton births were included. A probability sample of that group, with
higher fractions of low-birth-weight (< 2,500 g) and preterm (< 37 completed
weeks of gestation) infants, was enrolled in the
immune biomarker study (IBS; <italic>n</italic> = 615) (<xref rid="b22-ehp0114-001126" ref-type="bibr">Hertz-Picciotto et al. 2002</xref>). Initially, investigators randomly sampled every fifth normal-birth-weight
infant and all low-birth-weight infants; however, sampling fractions
were increased in Prachatice because of a lower birth rate than expected, and
in Teplice to enroll more children starting in January 1996 during
a meteorologic inversion. All sampling was random within strata. Infants
born on Fridays and Saturdays were ineligible for the IBS, because
immunologic analyses had to be performed within 24 hr of blood
draw.</p><p>For this investigation, we recontacted the IBS participants at 3 years
of age. Of the 615 IBS children, 50 were ineligible for follow-up (32 families
had moved to another district, 11 children were adopted or in
social care, and 7 children died). Of the remaining 565, pediatric data
were abstracted for 548 (97%): Nine families were not located
for follow-up, and eight mothers denied permission to review medical
records. Of the 548 with pediatric data available, 452 (82.5%) of
the mothers completed a 3-year follow-up questionnaire. Their children
were the subjects of this analysis. This study was approved by
the institutional committees on human subjects at the Institute for Experimental
Medicine in Prague, the University of North Carolina, Chapel
Hill, and the University of California, Davis.</p></sec><sec sec-type="methods"><title>Data collection</title><p>Trained nurses administered questionnaires at the birth hospital within 3 days
of delivery and elicited reproductive and medical history; medications; demographic
information; smoking, alcohol, and other lifestyle
factors; and work histories and occupational exposures. For the 3-year
follow-up, pediatric nurses contacted families to invite their participation
and provided them a new questionnaire regarding breast-feeding, child
care attendance, child’s and family members’ allergies, indoor
heating and cooking fuel sources, and information
about household members’ age, relation to child, and smoking behaviors. Informed
consent was administered at birth and again at follow-up, before
collection of data.</p><p>Physicians or nurses abstracted medical data from birth and pediatric records. At
birth, information collected included mother’s obstetric, labor, and
delivery complications; newborn sex, birth weight, Apgar
scores, and congenital anomalies; and clinicians’ estimate
of gestational age. When the child was 3 years of age, health providers
abstracted dates of all visits, diagnoses using the <italic>International Classification of Diseases, 10th Revision</italic> (ICD-10) codes (<xref rid="b53-ehp0114-001126" ref-type="bibr">World Health Organization 1993</xref>), and any treatments or medications prescribed. The use of a standard
pediatric medical record form throughout the country facilitated transfer
of information onto the study forms. Records of visits to specialists, hospitals, or
emergency services are forwarded to the primary physician, so
these reports were also abstracted and included in the analyses. When
an ICD-10 code was not provided on our study form, medically
trained study personnel coded the diagnosis using the medical chart
notes. Because medical care is free and universally available, pediatrician
utilization is high, and most families stay with the same pediatrician
until their children reach maturity.</p></sec><sec><title>LRI events</title><p>The 452 children in our study sample experienced 1,049 LRI events based
on ICD-10 codes J20, J21, J40, and J44, of which 98% were acute
bronchitis (J20). Pneumonia diagnoses (<italic>n</italic> = 70) were excluded from this analysis. The date of pediatrician’s
diagnosis served as a proxy for date of illness occurrence. To
avoid counting the same illness twice, we did not include diagnoses
occurring within 30 days after a previous LRI diagnosis, leaving 893 incident
LRI events.</p></sec><sec><title>Indoor combustion emissions exposures</title><p>Mothers reported indoor heating and cooking fuel combustion sources, such
as gas-, wood-, or coal-burning devices in the home on the 3-year follow-up
questionnaire. We initially analyzed incidence of LRI within
categories of heating and cooking fuel and by presence of gas-using appliances (stove, oven, and
water heater) in the home. Because most fuels
for heating conferred no increase in LRI rates or were used by very
few households, in final models we compared coal home heating with all
other heating sources. Direct measurements of organic carbon, elemental
carbon, or other elements were not made in homes.</p><p>Information on ETS exposure was available from both the birth and 3-year
questionnaires. We assessed mother’s smoking during pregnancy (yes/no) and
adult smokers in the household at 3-year follow-up (yes/no).</p></sec><sec><title>Covariates</title><p>Covariates of interest were selected <italic>a priori</italic> based on a review of the literature and preliminary bivariate analyses. Time-independent
covariates included district of residence (Teplice
vs. Prachatice), child’s sex, maternal ethnicity [either
Central European or Rom (“Gypsy”) based on self-report], and
maternal education (categorized as 6–10 years, 11 years, and ≥ 12 years for tabular presentation, and coded
as a continuous variable in final models).</p><p>On the 3-year follow-up questionnaire, mothers were asked until what age
their child was “fully” and “partially” breast-fed; we
used the maximum of these two responses to construct
three time-varying categories: currently breast-feeding, ever breast-fed, and
never breast-fed. We also constructed categories for time since
cessation of breast-feeding, to capture possible protection beyond
the weaning period: currently breast-feeding, breast-fed in the last
month, last breast-fed 1–6 months ago, last breast-fed > 6 months
ago, and never breast-fed. For analyses of breast-feeding as a
modifier of the effects of indoor air exposures, we dichotomized children
as ever or never breast-fed.</p><p>Other time-dependent variables included child’s age in years, current
child care attendance (yes/no, based on maternal report of the
ages at which the child attended child care with other children), number
of other household members ≤ 14 years of age (0, 1, and ≥ 2), and
household density (number of adults and children living
in the household divided by the number of rooms). The seasons were winter (December–February), spring (March–May), summer (June–August), and
autumn (September–November). This categorization
captured school vacation (June–August) as a distinct
period. Days of the week were grouped based on similarity in LRI rates: Saturday–Sunday, Monday, and Tuesday–Friday.</p><p>We examined temperature and relative humidity using 24-hr means from which
we calculated moving averages for 3, 7, 14, and 30 days before (and
including) the day of diagnosis. Of these averaging periods, 14-day
average temperature showed the strongest association with LRI and therefore
was selected. The log-odds of LRI decreased linearly with temperature, permitting
a continuous term for temperature. Relative humidity
was not associated with LRI diagnosis.</p></sec><sec sec-type="methods"><title>Statistical methods</title><p>We used generalized estimating equations (GEEs) to model the associations
with LRI diagnoses (<xref rid="b54-ehp0114-001126" ref-type="bibr">Zeger and Liang 1986</xref>; <xref rid="b55-ehp0114-001126" ref-type="bibr">Zeger et al. 1988</xref>). This method provides robust variance estimates, which account for the
correlation among repeated observations in the same individuals (<xref rid="b45-ehp0114-001126" ref-type="bibr">Stokes et al. 2000</xref>). GEEs also allow characterization of effects from time-varying factors (listed
above). Data were structured in a child-day file, where each
row corresponded to one observation-day for one child. A child followed
for exactly 3 years would contribute 1,095 –30<italic>n</italic> observation-days, where <italic>n</italic> is the number of LRI events separated by at least 30 days. On a given
day of life, occurrence of an LRI was coded 1; nonevent days were coded 0.</p><p>We used a logit link, that is, a logistic model, of the binary outcome (event
day: yes/ no) and specified an exchangeable correlation structure
to account for greater within-child than between-children homogeneity. The
probability (rate per child-day) of an LRI event occurring on
any given child-day was small (0.002). Odds of an event are therefore
arithmetically very close to the per child-day rate. The rate ratio (RR) was
estimated by <italic>e</italic><sup>β</sup> for each variable, or category, and corresponding 95% confidence
intervals (CIs) were calculated.</p><p>Because the children in this follow-up study were not a simple random sample
of births in Teplice and Prachatice districts, we also accounted
for the unequal sampling probabilities of normal-birth-weight/full-term
and low-birth-weight/preterm infants, which differed by district and
by year of study. Inverse sampling probabilities were used as weights
with a design of stratified sampling without replacement. Model fitting
was conducted using SUDAAN software (version 8.0; Research Triangle
Institute, Research Triangle Park, NC, USA).</p><p>Initially, we determined “crude” rates of LRI (LRI event
counts per child-time at risk) within categories of predictor variables. RRs
and 95% CIs were then estimated using GEE models adjusted
for sampling design, but not for other covariates. Next, covariates
associated (defined loosely as Wald chi-squared <italic>p</italic>-value of < 0.15) with LRI in bivariable analyses were entered into
a full model, and less influential covariates were removed, one by one, beginning
with the least influential (greatest <italic>p</italic>-value). Covariates for which removal changed the point estimates for coal
heat or ETS by ≥ 10% were retained as confounders. Final
models included coal heating, mother’s smoking during pregnancy, presence
of other adult smokers in the household, mother’s
age and education, child’s sex and year of life, breast-feeding, child
care attendance, number of other children living in the
home, season, day of the week, and 14-day moving-average temperature.</p><p>We examined potential modification of the indoor pollutant–LRI
associations by year of life, breast-feeding, preterm birth and/or low
birth weight, child’s sex, child care attendance, temperature, and
season by inclusion of product terms with ETS or coal home heating, one
at a time. We also explored variability of effects by year of
life in three separate models, to evaluate more closely factors that changed
considerably over the first 3 years (e.g., breast-feeding and child
care attendance).</p></sec></sec><sec sec-type="results"><title>Results</title><p>Fifty percent of homes were heated primarily by distant heating (heat from
a remote source outside the home), 23% by natural gas, 13% by
coal, and 6% by wood (<xref ref-type="table" rid="t1-ehp0114-001126">Table 1</xref>). Natural gas, propane, or electricity accounted for 97% of primary
cooking fuels used; only 3% of families reported cooking
primarily with wood or coal. Twenty-four percent of mothers reported
smoking during pregnancy, and 50% of fathers smoked at time of
delivery. At 3-year follow-up, 35% of mothers reported smoking; almost 60% of
children lived in homes with at least one smoker; and 28% of
homes had two or more smokers.</p><p>Eight percent of mothers were of Rom ethnicity, and 87% breast-fed
the children, 23% for > 6 months. Twenty-one percent of
children attended child care with other children at some time during
the first 3 years of life, and about three-fifths lived in households
with another child.</p><p>Coal heat, mother’s smoking during pregnancy, and ETS during the
first 3 years of life were all associated with greater incidence of
LRI (<xref ref-type="table" rid="t2-ehp0114-001126">Table 2</xref>). Use of a wood- or gas-burning stove or other appliance in the home did
not increase LRI rates significantly. Other factors associated with 40–50% higher rates of LRI were Rom ethnicity, current
attendance at child care with other children, and presence of other
children in the household. Mondays and winter months were also associated
with greater numbers of LRI diagnoses. Protective factors included
increasing maternal age, maternal education, and temperature (each with
a monotonic relationship), as well as breast-feeding. Low birth weight
or preterm birth did not affect LRI rates.</p><p>The observed associations between indoor combustion exposures and LRI incidence
persisted after multivariate control for mother’s age
and education, child’s sex and year of life, breast-feeding, current
attendance at child care, other children living in the household, season, day
of the week, and 14-day moving-average temperature (<xref ref-type="table" rid="t3-ehp0114-001126">Table 3</xref>). Children living in homes where coal was used as the primary heating
fuel experienced 45% greater LRI incidence (RR = 1.45; 95% CI, 1.07–1.97) compared with children whose homes
were heated by natural gas, propane, electricity, wood, or a distant (outside
the home) source. Mother’s smoking during pregnancy increased
her child’s LRI incidence over the next 3 years by 48% (RR = 1.48; 95% CI, 1.10–2.01), and
an adult other than the mother smoking in the household independently
increased child’s LRI incidence 29% (RR = 1.29; 95% CI, 1.01–1.65). Among children never breast-fed, the
effects of both coal home heating (RR = 2.77; 95% CI, 1.45–5.27) and mother’s smoking (RR = 2.52; 95% CI, 1.31–4.85) on LRI incidence were greater than
among children who were breast-fed (<xref ref-type="table" rid="t4-ehp0114-001126">Table 4</xref>). The never–breast-fed group included more Romany, more mothers
with higher education, and more smokers. There was some suggestion that
the coal heating effect was primarily in the first 2 years of life (data
not shown), but no other effect modification was as striking as
that of breast-feeding.</p></sec><sec sec-type="discussion"><title>Discussion</title><p>We found that children exposed to indoor coal combustion experience a greater
incidence of pediatrician-diagnosed LRI during the first 3 years
of life. We did not observe associations between LRI incidence and use
of natural gas or propane as heating fuel. The coefficient for wood
as a heating source was elevated, but the CIs were wide, largely because
wood was used by very few households. Coal, wood, and propane for
cooking showed elevated but imprecise relative risks, and both coal and
wood were rarely used for cooking. Fewer than half of families with
gas heating had a furnace inside the flat, suggesting that exposures such
as nitric oxide and nitrogen dioxide would occur only in some of the
homes. Nevertheless, previous studies of infants also indicate no increases
in LRI from exposures to nitrogen dioxide (<xref rid="b41-ehp0114-001126" ref-type="bibr">Samet et al. 1993</xref>). Thus, our data were consistent with findings of no increased risk associated
with emissions from gas cooking and heating.</p><p>Indoor measurements were not available. The coal burning devices are stand-alone
units located inside the living spaces of the homes, which either
directly heat the air or heat water that circulates through a radiator. Furnaces
located outside the homes were designated as “distant
sources” and included in the reference group. Although
the venting is always to the outside, indoor air becomes polluted with
dust and gases when coal is added or the unit is cleaned by removal
of the ashes. Nearby outdoor air is also polluted by normal operation
of these units. Studies of emissions sources conducted in Teplice and
Prachatice in the early 1990s showed that home heating with lignite coal
contributed measurably to outdoor concentrations of ambient organic
and elemental carbon, sulfur, potassium, iron, zinc, lead, bromine, and
other elements (<xref rid="b36-ehp0114-001126" ref-type="bibr">Pinto et al. 1998</xref>). Moreover, it was shown that during the winter, the ratio of benzo[<italic>a</italic>]pyrene to lead in ambient outdoor air increased 5–15 times
over the ratio observed in summer months, which was attributed to
emissions from residential home heating by coal combustion (<xref rid="b44-ehp0114-001126" ref-type="bibr">Stevens et al. 1997</xref>). In houses occupied by nonsmokers, major sources of indoor pollution
are nearby homes that use coal for fuel (<xref rid="b1-ehp0114-001126" ref-type="bibr">Benes et al. 2003</xref>). Thus, emissions vented outside may make their way back into the flat
through windows and doors. Given the climate in these districts (mean
daily temperatures < 10°C, or 50°F, for more than
half the year), homes would be heated on a daily basis for not less than 6 months
a year. Indoor environments are likely to be heavily polluted
in homes where coal is used for fuel.</p><p>Higher vulnerability of very young children to indoor pollutants may occur
for several reasons. Infants and toddlers spend more time indoors
at home than do school-age children or adults (<xref rid="b5-ehp0114-001126" ref-type="bibr">Brinkman et al. 1999</xref>; <xref rid="b49-ehp0114-001126" ref-type="bibr">U.S. Environmental Protection Agency 2002</xref>; <xref rid="b50-ehp0114-001126" ref-type="bibr">Wiley et al. 1991</xref>). Indoor air exposures can be more concentrated than outdoor exposures, especially
when sources are indoors and buildings are closed, as in
the winter. Additionally, the respiratory and immune systems are undergoing
development during early life: In this study population, district
of residence, season, and prenatal exposures to ambient pollution were
related to altered distributions of lymphocyte immunophenotypes and
IgE in cord blood (<xref rid="b22-ehp0114-001126" ref-type="bibr">Hertz-Picciotto et al. 2002</xref>), with some data implicating specific vulnerable time periods (Herr CEW, unpublished
data; <xref rid="b23-ehp0114-001126" ref-type="bibr">Hertz-Picciotto et al. 2005</xref>).</p><p>Our findings for both maternal smoking during pregnancy and postnatal ETS
from other household smokers are similar to results of meta-analyses (<xref rid="b29-ehp0114-001126" ref-type="bibr">Li et al. 1999</xref>; <xref rid="b46-ehp0114-001126" ref-type="bibr">Strachan and Cook 1997</xref>). The risk of developing an acute LRI in the first 3 years of life is
increased about 60% if either parent smokes, 70% if the
mother smokes, and about 30% if the mother does not smoke but
other household members do (<xref rid="b46-ehp0114-001126" ref-type="bibr">Strachan and Cook 1997</xref>). Moreover, we found independent effects on young children’s LRI
risk from mothers’ smoking during pregnancy and from postnatal
ETS due to other adults’ smoking. Previous work suggests that
ETS carries the greatest risk to children < 3 years of age (<xref rid="b28-ehp0114-001126" ref-type="bibr">Kontos et al. 1999</xref>). Because of the strong correlation between maternal smoking during and
after pregnancy, we could not evaluate which time period contributed
most to the effects we observed.</p><p>Among children who had never breast-fed, associations with both coal combustion
and mother’s smoking were substantially greater, with
LRI rates elevated close to 3-fold and 2.5-fold, respectively. Breast-feeding
has been found to protect against the effects of maternal smoking
on LRI in infants (<xref rid="b32-ehp0114-001126" ref-type="bibr">Nafstad et al. 1996</xref>; <xref rid="b52-ehp0114-001126" ref-type="bibr">Woodward et al. 1990</xref>) and older children (<xref rid="b11-ehp0114-001126" ref-type="bibr">Chulada et al. 2003</xref>). Coal heating emissions and ETS share similar pollutant constituents, so
it was not surprising to find that breast-feeding protected children
from the adverse effects of coal home heating. Immunity conferred by
breast milk is understood to be both anti-infectious and anti-inflammatory (<xref rid="b19-ehp0114-001126" ref-type="bibr">Hanson et al. 2001</xref>, <xref rid="b20-ehp0114-001126" ref-type="bibr">2002</xref>). Breast-feeding may play a direct protective role against indoor combustion
exposures, and it may foster key immunologic responses, thereby
reducing susceptibility to infection.</p><p>Our study relied on home environmental data collected retrospectively. Some
error may be introduced when mothers are asked to recall events over
the past 3 years (e.g., when she stopped breast-feeding or when the
child began child care). These errors seem unlikely to differ by exposure, but
could possibly mask covariate–respiratory illness associations, which
may limit the ability to control for these covariates
as potential confounders of pollutant–respiratory illness associations.</p><p>Another limitation was the lack of information on changes in some home
environmental characteristics between birth and 3 years of age. At the
time of our study, a national policy was in effect to replace coal in
home heating with natural gas, with the goal to decrease outdoor air
pollution. Despite this, about 13% of the study children’s
homes were still heated primarily by coal at follow-up (1997–1999). A
family that switched from coal to gas would have reported
use of gas at 3-year follow-up, resulting in misclassification of heating
type for some part of the 3-year follow-up. Generally, this might
have attenuated the results, with some of the “unexposed” children
actually exposed to coal. If switching were related to other
factors, the direction of bias would be difficult to predict.</p><p>This study used pediatrician-reported LRI. Capture of an LRI diagnosis
depends on health service utilization, as well as the ability of the pediatrician
to reliably record the type of illness. Our data demonstrate
extensive utilization of health care services by this population of
Czech mothers. Despite some changes in the provision of health care in
the Czech Republic over the past decade, including the emergence of
private health insurance, mostly provided by employers, all residents
are entitled to health care, and consumer cost is relatively low. An indication
of the high utilization of physicians is the completeness of
the legally mandated 18-and 36-month pediatric well-child visits, which
in our study sample was > 96% and > 98%, respectively. Rates
of complete series of immunizations for diphtheria, pertussis, and
tetanus (DPT); polio; and measles were high, far exceeding
U.S. 1997 rates for children of the same age (<xref rid="b8-ehp0114-001126" ref-type="bibr">Centers for Disease Control and Prevention 2001</xref>). For example, 98% of children in our cohort received a complete
series of four DPT immunizations, compared with 71% of children
this age in the United States in 1997 (<xref rid="b8-ehp0114-001126" ref-type="bibr">Centers for Disease Control and Prevention 2001</xref>). With regard to reliable recording of events, we surveyed most of the
pediatricians who participated in this study about how they coded children
with specific sets of symptoms and found that their responses were
highly consistent and similar across the two districts, and matched
expected ICD-10 codes for overall LRI. Because virtually all previous
studies of coal home heating used parental reports of illnesses or symptoms, sometimes
for retrospective recall over long periods, the present
study represents a considerable improvement in the quality of health
outcome data.</p></sec><sec sec-type="conclusions"><title>Conclusions</title><p>In summary, we found exposure to coal home heating and ETS increase young
children’s LRI rates during the first 3 years of life. These
effects were attenuated by breast-feeding. Although ETS has been well
studied, residential coal combustion in economically developed countries
has not; these findings demonstrate that both sources of indoor air
combustion pollutants present a hazard to respiratory health in infancy
and early childhood. Efforts to reduce these emissions would benefit
infants and young children perhaps especially in the Czech setting, where
coal is still commonly used in home heating and smoking rates
are relatively high.</p></sec>
|
Benchmark Dose for Cadmium-Induced Renal Effects in Humans
|
<sec><title>Objectives</title><p>Our goal in this study was to explore the use of a hybrid approach to calculate
benchmark doses (BMDs) and their 95% lower confidence
bounds (BMDLs) for renal effects of cadmium in a population with low
environmental exposure.</p></sec><sec sec-type="methods"><title>Methods</title><p>Morning urine and blood samples were collected from 820 Swedish women 53–64 years
of age. We measured urinary cadmium (U-Cd) and tubular
effect markers [<italic>N</italic>-acetyl- β-D-glucosaminidase (NAG) and human complex-forming protein (protein
HC)] in 790 women and estimated glomerular filtration
rate (GFR; based on serum cystatin C) in 700 women. Age, body mass
index, use of nonsteroidal anti-inflammatory drugs, and blood lead
levels were used as covariates for estimated GFR. BMDs/BMDLs corresponding
to an additional risk (benchmark response) of 5 or 10% were
calculated (the background risk at zero exposure was set to 5%). The
results were compared with the estimated critical concentrations
obtained by applying logistic models used in previous studies on
the present data.</p></sec><sec><title>Results</title><p>For both NAG and protein HC, the BMDs (BMDLs) of U-Cd were 0.5–1.1 (0.4–0.8) μg/L (adjusted for specific gravity of 1.015 g/mL) and 0.6–1.1 (0.5–0.8) μg/g creatinine. For
estimated GFR, the BMDs (BMDLs) were 0.8–1.3 (0.5–0.9) μg/L
adjusted for specific gravity and 1.1–1.8 (0.7–1.2) μg/g creatinine.</p></sec><sec><title>Conclusion</title><p>The obtained benchmark doses of U-Cd were lower than the critical concentrations
previously reported. The critical dose level for glomerular
effects was only slightly higher than that for tubular effects. We suggest
that the hybrid approach is more appropriate for estimation of the
critical U-Cd concentration, because the choice of cutoff values in
logistic models largely influenced the obtained critical U-Cd.</p></sec>
|
<contrib contrib-type="author"><name><surname>Suwazono</surname><given-names>Yasushi</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001072">1</xref><xref ref-type="aff" rid="af2-ehp0114-001072">2</xref></contrib><contrib contrib-type="author"><name><surname>Sand</surname><given-names>Salomon</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001072">1</xref></contrib><contrib contrib-type="author"><name><surname>Vahter</surname><given-names>Marie</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001072">1</xref></contrib><contrib contrib-type="author"><name><surname>Filipsson</surname><given-names>Agneta Falk</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001072">1</xref><xref ref-type="aff" rid="af3-ehp0114-001072">3</xref></contrib><contrib contrib-type="author"><name><surname>Skerfving</surname><given-names>Staffan</given-names></name><xref ref-type="aff" rid="af4-ehp0114-001072">4</xref></contrib><contrib contrib-type="author"><name><surname>Lidfeldt</surname><given-names>Jonas</given-names></name><xref ref-type="aff" rid="af5-ehp0114-001072">5</xref></contrib><contrib contrib-type="author"><name><surname>Åkesson</surname><given-names>Agneta</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001072">1</xref></contrib>
|
Environmental Health Perspectives
|
<p>People are exposed to cadmium—a widespread nephrotoxic pollutant—via
food and tobacco smoking. The first sign of renal effects
is tubular damage, characterized by increased urinary excretion of low-molecular-weight
proteins or intracellular tubular enzymes. More important, in
succession to the tubular effects, Cd may affect glomerular
function (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>; <xref rid="b2-ehp0114-001072" ref-type="bibr">Bernard et al. 1992</xref>; <xref rid="b11-ehp0114-001072" ref-type="bibr">Friberg 1950</xref>; <xref rid="b16-ehp0114-001072" ref-type="bibr">Järup et al. 1995</xref>; <xref rid="b25-ehp0114-001072" ref-type="bibr">Nogawa 1984</xref>; <xref rid="b35-ehp0114-001072" ref-type="bibr">World Health Organization 1992</xref>). To protect people from Cd-induced health effects, it is crucial to determine
the critical exposure, that is, the concentration of urinary
Cd (U-Cd) below which the probability of adverse health effects is low. Attempts
to estimate this limit for tubular effects have so far displayed
large variations in critical U-Cd levels (1–10 μg
U-Cd/g creatinine) (<xref rid="b4-ehp0114-001072" ref-type="bibr">Buchet et al. 1980</xref>, <xref rid="b3-ehp0114-001072" ref-type="bibr">1990</xref>; <xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>; <xref rid="b15-ehp0114-001072" ref-type="bibr">Järup et al. 2000</xref>; <xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>; <xref rid="b21-ehp0114-001072" ref-type="bibr">Lauwerys et al. 1979</xref>).</p><p>The benchmark dose (BMD) method is increasingly used in the health risk
assessment of environmental contaminants [<xref rid="b8-ehp0114-001072" ref-type="bibr">Crump 1984</xref>; <xref rid="b10-ehp0114-001072" ref-type="bibr">Filipsson et al. 2003</xref>; <xref rid="b33-ehp0114-001072" ref-type="bibr">U.S. Environmental Protection Agency (EPA) 1995</xref>]. Only in a few cases has the BMD method been used for people
environmentally exposed to Cd (<xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>; <xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>; <xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>). The BMD can be defined as the exposure that corresponds to a certain
change in response compared with the background. The lower 95% confidence
bound of the BMD (BMDL) has been suggested to replace the
no observed adverse effect level (NOAEL) (<xref rid="b8-ehp0114-001072" ref-type="bibr">Crump 1984</xref>; <xref rid="b33-ehp0114-001072" ref-type="bibr">U.S. EPA 1995</xref>). One major advantage of the BMD/BMDL approach is that it uses the whole
dose–response curve (<xref rid="b33-ehp0114-001072" ref-type="bibr">U.S. EPA 1995</xref>). Thus, the BMD/BMDL is based on more information than the NOAEL. By using
a so-called hybrid approach, the concept of risk can be used for
a continuous outcome (effect variable). In that way, the limitations associated
with categorization of data can be avoided (<xref rid="b6-ehp0114-001072" ref-type="bibr">Crump 1995</xref>; <xref rid="b12-ehp0114-001072" ref-type="bibr">Gaylor and Slikker 1990</xref>; <xref rid="b19-ehp0114-001072" ref-type="bibr">Kodell and West 1993</xref>; <xref rid="b26-ehp0114-001072" ref-type="bibr">Ragland 1992</xref>).</p><p>Our aim in the present study was to determine the BMDs of U-Cd for Cd-induced
tubular and glomerular effects in an environmentally exposed population, using
the hybrid approach. To evaluate the unique feature of
the hybrid approach, the obtained BMDs/BMDLs were compared with the critical
concentrations obtained by the traditionally used procedures.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Study population and measurement</title><p>Within the population-based Women’s Health in the Lund Area (WHILA) study (<xref rid="b22-ehp0114-001072" ref-type="bibr">Lidfeldt et al. 2001</xref>), conducted in an area with no particular industrial emission, we assessed
health effects of Cd in 820 women 53–64 years of age (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>). Subjects with renal cancer and lithium treatment were excluded (<italic>n</italic> = 4). In addition, because of effect modification (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>), insulin-treated subjects with diabetes were excluded from calculation
of the BMD for tubular (<italic>n</italic> = 14) but not glomerular effects.</p><p>According to a questionnaire, 45% of the included women had ever
smoked (ever-smokers). In addition, nonsteroidal anti-inflammatory drugs (NSAIDs) were
regularly used by 6% of the women.</p><p>We used U-Cd as the measure of long-term Cd exposure, urinary <italic>N</italic>-acetyl-β-<sc>d</sc>-glu-cosaminidase (NAG) and human complex-forming protein (protein HC) as
markers of tubular effects, and estimated glomerular filtration rate (GFR) based
on cystatin C in serum (estimated GFR = 77.24 × cystatin
C<sup>–1.2623</sup>) (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>; <xref rid="b20-ehp0114-001072" ref-type="bibr">Larsson et al. 2004</xref>) as a marker of glomerular effect (<xref ref-type="table" rid="t1-ehp0114-001072">Table 1</xref>). Urinary analytes were adjusted to a specific gravity of 1.015 g/mL, because
creatinine may not adjust for all dilution-related variation of
U-Cd (<xref rid="b30-ehp0114-001072" ref-type="bibr">Suwazono et al. 2005</xref>). However, because creatinine adjustment is more commonly used, these
values are given for comparison. The ethics committee at Lund University
approved the WHILA study, and oral informed consent was obtained from
each participant.</p></sec><sec><title>Model fitting</title><p>We used the maximum likelihood approach to fit the dose–response
curve to the data (<xref rid="b6-ehp0114-001072" ref-type="bibr">Crump 1995</xref>). For normally distributed data with constant variance, the log-likelihood
function, log <italic>L</italic>, is given by</p><disp-formula><graphic xlink:href="ehp0114-001072e1.jpg" position="float" mimetype="image"/></disp-formula><p>where <italic>n</italic> is the number of subjects, d<italic><sub>i</sub></italic> is the dose for <italic>i</italic>th individual, σ<sup>2</sup> is the variance, μ (<italic>d</italic><italic><sub>i</sub></italic>) is the dose–response model for the mean response, and <italic>y</italic><italic><sub>i</sub></italic> is the response in the <italic>i</italic>th individual. To obtain a symmetrical distribution, data on NAG and protein
HC were log-transformed. Data on estimated GFR did not have to be
log-transformed. The model for the mean response, μ(<italic>d</italic><italic><sub>i</sub></italic>), was assumed to be linear:</p><disp-formula><graphic xlink:href="ehp0114-001072e2.jpg" position="float" mimetype="image"/></disp-formula><p>We found significant covariates only for estimated GFR (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>). Age, body mass index, use of NSAIDs, and blood lead levels (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>) were included in the model to control for possible confounding of estimated
GFR; NAG and protein HC displayed no such associations. Smoking (pack-years) was
not associated with any of the kidney effect markers.</p></sec><sec><title>Calculation of BMDs</title><p>BMDs were calculated using a hybrid approach, which allows for calculation
of risk for continuous data without dichotomizing the outcome (<xref rid="b7-ehp0114-001072" ref-type="bibr">Crump 2002</xref>; <xref rid="b12-ehp0114-001072" ref-type="bibr">Gaylor and Slikker 1990</xref>; <xref rid="b28-ehp0114-001072" ref-type="bibr">Sand et al. 2004</xref>). The benchmark response (BMR), corresponding to the BMD, was defined
as an additional prespecified increase in the probability of adverse response. For
positive associations between exposure (U-Cd) and effects (NAG
and protein HC), the effect level associated with a certain BMR
equals.</p><disp-formula><graphic xlink:href="ehp0114-001072e3.jpg" position="float" mimetype="image"/></disp-formula><p>where Φ<sup>–1</sup> is the inverse of the standard normal cumulative distribution function
and <italic>P</italic>(0) is the cutoff level for adverse response defined in terms of a specified
tail proportion of a “hypothetical” control distribution (at
U-Cd = 0), equivalent to the background probability
of adverse response. The cutoff, <italic>c</italic>, for the effect markers is given by</p><disp-formula><graphic xlink:href="ehp0114-001072e4.jpg" position="float" mimetype="image"/></disp-formula><p>The BMD is obtained by combining the equation for μ(BMD) with that
for the dose–response (Equation 1):</p><disp-formula><graphic xlink:href="ehp0114-001072e5.jpg" position="float" mimetype="image"/></disp-formula><p>For negative associations between exposure and effects (β<sub>1</sub> < 0), such as that for the association between Cd and estimated GFR, calculations
were performed in a similar way, substituting the absolute
value of β<sub>1</sub> into Equation 2 (<xref rid="b29-ehp0114-001072" ref-type="bibr">Sand et al. 2003</xref>).</p><p>The BMDL was calculated using the profile likelihood method (<xref rid="b6-ehp0114-001072" ref-type="bibr">Crump 1995</xref>; <xref rid="b10-ehp0114-001072" ref-type="bibr">Filipsson et al. 2003</xref>). BMDs/BMDLs with the <italic>P</italic>(0) = 5% and BMR = 5 or 10% were calculated
for all renal effect markers as representative threshold levels. To
describe how the BMD/BMDL depends on the BMR and <italic>P</italic>(0), data on NAG adjusted for specific gravity were used as an example. We
calculated BMD/BMDL for three different BMRs (5, 10, or 20%) with
varying <italic>P</italic>(0) (1–20%).</p></sec><sec sec-type="methods"><title>Comparisons with previously used procedures</title><p>To compare the hybrid approach with the previously used procedures, in
which the outcome is dichotomized, we applied the procedures used in the
Cadmibel study (<xref rid="b3-ehp0114-001072" ref-type="bibr">Buchet et al. 1990</xref>) and the OSteoporosis, CAdmium as a Risk Factor (OSCAR) study (<xref rid="b15-ehp0114-001072" ref-type="bibr">Järup et al. 2000</xref>) on our data. In the Cadmibel study, the relationship between renal adverse
response and U-Cd was investigated in a Belgian population (<xref rid="b3-ehp0114-001072" ref-type="bibr">Buchet et al. 1990</xref>). The authors derived the cutoffs as the 95th percentile of the renal
tubular markers in a part of the study population that was considered
to be free from kidney disease. The U-Cd levels at which the probability
of having an adverse response was 10% were estimated using
a logistic model after exclusion of individuals with diabetes and regular
use of NSAIDs. In the OSCAR study (<xref rid="b15-ehp0114-001072" ref-type="bibr">Järup et al. 2000</xref>), the adverse response of protein HC in relation to U-Cd was investigated
in a Swedish population. The adverse response was defined as urinary
protein HC above the 95th percentile (5.3 mg/g creatinine; 0.6 mg/mmol
creatinine in women) from another Swedish reference population (<xref rid="b31-ehp0114-001072" ref-type="bibr">Tencer et al. 1996</xref>). The U-Cd level at 15% probability of an adverse response was
then estimated using parameters obtained by logistic regression model
in the OSCAR study.</p><p>We fitted our data to a logistic model. The probability of adverse response
at the dose <italic>d</italic><italic><sub>i</sub></italic> of U-Cd is given by</p><disp-formula><graphic xlink:href="ehp0114-001072e6.jpg" position="float" mimetype="image"/></disp-formula><p>where αis the log odds of adverse response at the U-Cd = 0, and βis
the slope for dose–log odds relationship. Then, d<italic><sub>i</sub></italic> is given by</p><disp-formula><graphic xlink:href="ehp0114-001072e7.jpg" position="float" mimetype="image"/></disp-formula><p>The background probabilities and the U-Cd levels at the 10% (Cadmibel) or 15% (OSCAR) probability of adverse response were estimated
by Equations 3 and 4 and compared with corresponding background
probability and U-Cd levels using the hybrid approach.</p></sec><sec><title>Software</title><p>We used SPSS (version 12.0.1; SPSS Inc., Chicago, IL, USA) and Microsoft
Excel (Microsoft Corp., Redmond, WA, USA) for analyses. These results
were verified to be identical to the results by MATLAB (version 7.0; MathWorks, Inc., Novi, MI, USA) used in our previous studies (<xref rid="b29-ehp0114-001072" ref-type="bibr">Sand et al. 2003</xref>, <xref rid="b28-ehp0114-001072" ref-type="bibr">2004</xref>).</p></sec></sec><sec sec-type="results"><title>Results</title><p>We found significant associations between U-Cd, on the one hand, and NAG, protein
HC (both positive associations), and estimated GFR (negative
association), on the other, based on the maximum likelihood model (data
not shown).</p><p><xref ref-type="table" rid="t2-ehp0114-001072">Table 2</xref> shows the BMDs and BMDLs of U-Cd using a cutoff, <italic>P</italic>(0), of 5% and a BMR of 5 or 10% for the renal effect markers. For
the tubular effects (both NAG and protein HC), the BMDLs of
U-Cd were 0.4–0.8 μg/L, corresponding to 0.5–0.8 μg/g
creatinine. For the glomerular effects (estimated GFR), the
BMDLs of U-Cd were 0.5–0.9 μg/L, corresponding
to 0.7–1.2 μg/g creatinine. We obtained essentially the
same BMD/BMDL if we used cystatin C instead of estimated GFR.</p><p>We evaluated the effect of the cutoff value [<italic>P</italic>(0)] and the response criteria (BMR) on the BMDs/BMDLs. As shown
in <xref ref-type="fig" rid="f1-ehp0114-001072">Figure 1</xref>, a larger BMR and a smaller <italic>P</italic>(0) yield larger BMD/BMDLs.</p><p>As shown in <xref ref-type="fig" rid="f2-ehp0114-001072">Figure 2A</xref>, the cutoff concentrations of NAG and protein HC obtained by our hybrid
approach modeling were lower than those obtained by employing the procedure
used in the Cadmibel study in our study. The opposite was observed
for the OSCAR procedure. In <xref ref-type="fig" rid="f2-ehp0114-001072">Figure 2B</xref>, the cutoffs from <xref ref-type="fig" rid="f2-ehp0114-001072">Figure 2A</xref> are presented in terms of different background probabilities of adverse
response [<italic>P</italic>(0)]. By using the predefined cutoff values of the Cadmibel and
OSCAR studies, we obtained a lower and a higher <italic>P</italic>(0), respectively (<xref ref-type="fig" rid="f2-ehp0114-001072">Figure 2B</xref>). When we compared the critical concentration of U-Cd obtained by the
hybrid approach, the U-Cd levels were lower than those obtained by applying
the Cadmibel procedure to our data. The opposite was observed for
the OSCAR procedure (<xref ref-type="fig" rid="f2-ehp0114-001072">Figure 2C</xref>).</p></sec><sec sec-type="discussion"><title>Discussion</title><p>To our knowledge, this is the first estimation of BMDs of Cd-induced renal
effects using the recently developed hybrid approach. The critical
concentration was estimated for both tubular and glomerular effects in
a population of upper middle-age women living in an area in southern
Sweden without particular industrial Cd emission. Generally, the critical
concentrations obtained by the hybrid method approach were lower
than those previously reported.</p><p>The present method has several methodologic advantages. First, the BMDs/BMDLs
were calculated based on a continuous outcome. Calculations of
BMD/BMDL for continuous outcomes using the hybrid approach has been developed
during the last few years (<xref rid="b6-ehp0114-001072" ref-type="bibr">Crump 1995</xref>; <xref rid="b28-ehp0114-001072" ref-type="bibr">Sand et al. 2004</xref>). The advantage with the hybrid approach is that the categorization of
subjects with respect to the outcome variables can be avoided. Accordingly, the
statistical validity and efficiency of the BMD is higher using
the hybrid approach, compared with methods involving dichotomization
of a continuous outcome (<xref rid="b7-ehp0114-001072" ref-type="bibr">Crump 2002</xref>; <xref rid="b34-ehp0114-001072" ref-type="bibr">West and Kodell 1999</xref>).</p><p>Second, we defined the cutoff for adverse effects as the 95th percentile, obtained
by the model at no Cd exposure (U-Cd = 0) in the population
under study, rather than as the 95th percentile of the effect
marker in an apparently low-exposed “reference” population, with
little information on the overall comparability. Further, by
estimating the cutoff for adverse response by the model at zero Cd
exposure, any impact of the exposure level in a reference group will
be minimized. The obtained critical U-Cd levels then corresponds to an
adverse response of 10% (5% additional probability of
adverse response; BMR = 5%) or 15% (10% additional
probability of adverse response; BMR = 10%).</p><p>Third, we were able to avoid categorization of the exposure variable. Except
for the fact that the number categories and the dose interval for
each category chosen may strongly affect the result, categorization
will decrease the detection power (<xref rid="b27-ehp0114-001072" ref-type="bibr">Royston et al. 2000</xref>).</p><p>Furthermore, we further improved the method by using a multivariate linear
regression model instead of a univariate model (<xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>; <xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>; <xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>) to enable the adjustment of BMD/ BMDL for potential confounders.</p><p>The BMDs for U-Cd obtained in the present study were generally lower than
the previously reported critical levels. For instance, the lowest observed
effect levels based on the same data (<xref rid="b1-ehp0114-001072" ref-type="bibr">Åkesson et al. 2005</xref>) were on average 10% higher than the present BMDs. In the Cadmibel
study (<xref rid="b3-ehp0114-001072" ref-type="bibr">Buchet et al. 1990</xref>), the U-Cd level corresponding to the 10% probability of adverse
response was 1.9 μg/24 hr (equivalent to about 2 μg/g
creatinine) for calciuria and 2.7 μg/24 hr (roughly equivalent
to 3 μg/g creatinine) for NAG. However, in the OSCAR study (<xref rid="b15-ehp0114-001072" ref-type="bibr">Järup et al. 2000</xref>), the U-Cd level corresponding to 15% probability of adverse response
was 1.0 μg/g creatinine, similar to that obtained in the
present study. Further, the BMDLs obtained in the present study were
clearly lower than the BMDLs of 4–12 μg Cd/g creatinine (5% additional
probability) obtained for various kidney effect
markers (NAG and isoform B), β<sub>2</sub>-microglobulin (β<sub>2</sub>-MG), retinol-binding protein, and urinary albumin in China (<xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>), and slightly lower than the 0.9–1.2 μg Cd/g creatinine (10% additional
probability) (<xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>) obtained in another Chinese population that was coexposed to arsenic. The
present BMDLs were, however, very similar to that obtained in Japanese
women 40–59 years of age in a Cd-nonpolluted area. The BMDLs (5% additional probability) for the kidney effects (total
protein, β<sub>2</sub>-MG, and NAG) were 0.6–1.8 μg/g creatinine (<xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>). The corresponding results for men were lower: 0.3–0.6 μg/g
creatinine.</p><p>All of these other studies defined the adverse response (cutoff) as the 95th
percentile in a reference population assumed to be non-exposed (<xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>; <xref rid="b15-ehp0114-001072" ref-type="bibr">Järup et al. 2000</xref>; <xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>) or in a part of the study population considered free from kidney disease (<xref rid="b3-ehp0114-001072" ref-type="bibr">Buchet et al. 1990</xref>; <xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>). The study subjects were then categorized (dichotomous) as to the outcome. Obviously, a
more Cd-exposed reference group (<xref rid="b17-ehp0114-001072" ref-type="bibr">Jin et al. 2002</xref>) showed a considerably higher critical concentration (<xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>), emphasizing the need for a better standardized method to obtain the
threshold for adverse response. In addition, all the previous studies
on U-Cd and BMD (<xref rid="b13-ehp0114-001072" ref-type="bibr">Hong et al. 2004</xref>; <xref rid="b18-ehp0114-001072" ref-type="bibr">Jin et al. 2004</xref>; <xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>) categorized the exposure into strata (Benchmark Dose Software; U.S. EPA, Washington, DC, USA). We consider the present continuous approach
of calculating the BMD/BMDL more accurate and more efficient in using
the information.</p><p>When we applied previously used methods to our data, the procedure used
in the Cadmibel study for defining the cutoff resulted in a background
probability of < 5%, whereas the procedure used in the OSCAR
study resulted in a background probability of > 5%. The
main reason for the higher <italic>P</italic>(0) in the latter study was that the reference population was, on average, 20 years
younger than the study population. As illustrated in Equation 4 for
the logistic regression, a low background probability of adverse
response <italic>P</italic>(0) yields a larger [1 – <italic>P</italic>(0)]/<italic>P</italic>(0), which may yield a larger critical U-Cd level (for a constantβ). Considered
together, this shows that the cutoff value has a strong
effect on the estimated critical level. Thus, the choice of reference
population for determination of the cutoff for adverse effects (95th
percentile) may have large impact on the critical concentrations. The
advantage of the hybrid approach is that it allows for estimation of
the cutoff at zero exposure in the population under study.</p><p>Although, compared with other methods, the hybrid approach seemed to be
better in terms of the obtained critical concentration, it is still, as
for the logistic model, influenced by the actual value of the background
probability of adverse response, <italic>P</italic>(0). On the other hand, by using the hybrid approach, it is always possible
to set a defined <italic>P</italic>(0), which allows for interpopulation comparison of critical concentrations
under the same conditions. This is important because a lower <italic>P</italic>(0) leads to larger BMD/BMDL, as shown in <xref ref-type="fig" rid="f1-ehp0114-001072">Figure 1</xref>. The reason for this relates to the characteristics of the normal distribution. The
absolute distance between two points on the distribution
axis that bracket, for example, a 5% probability (i.e., a BMR = 5%) becomes higher in the extreme tail region compared
with in a more central part of the distribution. Thus, the lower <italic>P</italic>(0) becomes, the greater the distance between the two points corresponding
to <italic>P</italic>(0) and <italic>P</italic>(0) + BMR, which translates to a higher dose (BMD) being required
to produce the desired change in probability (BMR). Furthermore, the
impact of <italic>P</italic>(0) on BMD was more pronounced at lower than at higher values of <italic>P</italic>(0). To our knowledge, such importance of background probability has not
previously been evaluated in detail for either the hybrid approach or
the logistic model.</p><p>In several previous studies, a <italic>P</italic>(0) of 5% has been used as a standard for the hybrid approach (<xref rid="b5-ehp0114-001072" ref-type="bibr">Budtz-Jørgensen et al. 2000</xref>; <xref rid="b9-ehp0114-001072" ref-type="bibr">Crump et al. 2000</xref>; <xref rid="b14-ehp0114-001072" ref-type="bibr">Jacobson et al. 2002</xref>; <xref rid="b23-ehp0114-001072" ref-type="bibr">Murata et al. 2002</xref>, <xref rid="b24-ehp0114-001072" ref-type="bibr">2004</xref>), in accordance with the usual definition of clinical reference intervals. The
adopted BMR levels in the present study are in line with those
used in other recent epidemio-logic studies: 5% (<xref rid="b5-ehp0114-001072" ref-type="bibr">Budtz-Jørgensen et al. 2000</xref>; <xref rid="b23-ehp0114-001072" ref-type="bibr">Murata et al. 2002</xref>, <xref rid="b24-ehp0114-001072" ref-type="bibr">2004</xref>) or 10% (<xref rid="b5-ehp0114-001072" ref-type="bibr">Budtz-Jørgensen et al. 2000</xref>; <xref rid="b9-ehp0114-001072" ref-type="bibr">Crump et al. 2000</xref>). Obviously, other <italic>P</italic>(0) values and BMRs can be chosen, depending on the severity of the effects (<xref rid="b14-ehp0114-001072" ref-type="bibr">Jacobson et al. 2002</xref>).</p><p>The population-based design and the rather high participation rate advocate
generalization of the results to other female populations in the
same age interval. However, we cannot exclude gender difference in BMDLs
for kidney effects (<xref rid="b32-ehp0114-001072" ref-type="bibr">Uno et al. 2005</xref>).</p><p>In conclusion, the present BMDLs for tubular effects, using a cutoff <italic>P</italic>(0) of 5%, were 0.4 μg/L (0.5 μg/g creatinine) at
a BMR of 5% and 0.7 μg/L (0.8 μg/g creatinine) using
a BMR of 10%. The corresponding BMDLs for the glomerular
effect were 0.5 μg/L (0.7 μg/g creatinine) and 0.9 μg/L (1.2 μg/g creatinine) for BMRs of 5 and 10%, respectively. This
critical U-Cd level for glomerular effects was
lower and closer to the critical levels for tubular effects than expected
from previous studies.</p></sec>
|
Occupational Exposure to <italic>Pfiesteria</italic> Species in Estuarine Waters Is Not a Risk Factor for Illness
|
<sec><title>Background</title><p>Exposure to the dinoflagellate <italic>Pfiesteria</italic> has, under certain circumstances, been associated with deficits in human
learning and memory. However, uncertainties remain about the health
risk of chronic, low-level exposures (as seen among occupationally exposed
commercial fishermen), particularly in light of studies suggesting
that <italic>Pfiesteria</italic> strains are widespread in the estuarine environment in the U.S. mid-Atlantic
region.</p></sec><sec sec-type="methods"><title>Methods</title><p>We selected an initial cohort of 152 persons, including 123 persons with
regular, occupational exposure to the Chesapeake Bay; 107 of the cohort
members were followed for the full four summer “seasons” of
the study. Cohort members were questioned biweekly about symptoms, and
data were collected about the areas of the bay in which they
worked. These latter data were matched with data on the presence or
absence of <italic>Pfiesteria</italic> in each area, based on polymerase chain reaction analysis of > 3,500 water
samples. Cohort members underwent neuropsychological testing at
the beginning and end of each summer season.</p></sec><sec><title>Results</title><p>No correlation was found between work in an area where <italic>Pfiesteria</italic> was identified and specific symptomatology or changes on neuropsychological
tests.</p></sec><sec><title>Conclusions</title><p>Although high-level or outbreak-associated exposure to <italic>Pfiesteria</italic> species (or specific strains within a species) may have an effect on health, routine
occupational exposure to estuarine environments in which
these organisms are present does not appear to pose a significant health
risk.</p></sec>
|
<contrib contrib-type="author"><name><surname>Morris</surname><given-names>J. Glenn</given-names><suffix>Jr.</suffix></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref></contrib><contrib contrib-type="author"><name><surname>Grattan</surname><given-names>Lynn M.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-001038">2</xref></contrib><contrib contrib-type="author"><name><surname>Wilson</surname><given-names>Leslie A.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref></contrib><contrib contrib-type="author"><name><surname>Meyer</surname><given-names>Walter A.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref></contrib><contrib contrib-type="author"><name><surname>McCarter</surname><given-names>Robert</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref></contrib><contrib contrib-type="author"><name><surname>Bowers</surname><given-names>Holly A.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001038">3</xref></contrib><contrib contrib-type="author"><name><surname>Hebel</surname><given-names>J. Richard</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref></contrib><contrib contrib-type="author"><name><surname>Matuszak</surname><given-names>Diane L.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-001038">1</xref><xref ref-type="aff" rid="af4-ehp0114-001038">4</xref></contrib><contrib contrib-type="author"><name><surname>Oldach</surname><given-names>David W.</given-names></name><xref ref-type="aff" rid="af3-ehp0114-001038">3</xref></contrib>
|
Environmental Health Perspectives
|
<p>In the summer of 1997, a group of watermen (commercial fishermen) working
on the Pocomoke River on the eastern shore of the Chesapeake Bay in
Maryland developed a pattern of neuropsychological deficits marked by
difficulties in learning and memory (<xref rid="b19-ehp0114-001038" ref-type="bibr">Grattan et al. 1998</xref>). Initial deficits were profound, with affected individuals scoring 2–3 standard
deviations below national norms on standardized tests, but
resolved within 3–6 months of cessation of exposure to
the river. Affected watermen had had constant, high-level occupational
exposure to areas of the Pocomoke River where the dinoflagellate <italic>Pfiesteria</italic> (<xref rid="b10-ehp0114-001038" ref-type="bibr">Burkholder et al. 1992</xref>) had been identified in association with several fish-kill events, and
it was hypothesized that these river exposures contributed to the observed
health effects (<xref rid="b19-ehp0114-001038" ref-type="bibr">Grattan et al. 1998</xref>; <xref rid="b33-ehp0114-001038" ref-type="bibr">Morris 2001</xref>). There were also suggestions that isolated, acute exposure to affected
waterways in the midst of a fish-kill event could elicit a flu-like
syndrome, albeit without accompanying changes in neurocognitive test performance (<xref rid="b20-ehp0114-001038" ref-type="bibr">Haselow et al. 2001</xref>). Neuropsychological deficits similar to those seen among persons with
constant, high-level exposure to the Pocomoke River had been previously
observed after laboratory exposure to <italic>Pfiesteria</italic> (<xref rid="b17-ehp0114-001038" ref-type="bibr">Glasgow et al. 1995</xref>; <xref rid="b40-ehp0114-001038" ref-type="bibr">Schmechel and Koltai 2001</xref>) among persons working in the laboratory of J. Burkholder, who had initially
described the organism (<xref rid="b10-ehp0114-001038" ref-type="bibr">Burkholder et al. 1992</xref>). Studies by Levin and colleagues involving parenteral inoculation of
rats with material from <italic>Pfiesteria</italic> cultures provided further support for the idea that exposure to <italic>Pfiesteria</italic> resulted in deficits in new learning and memory (<xref rid="b24-ehp0114-001038" ref-type="bibr">Levin 2001</xref>; <xref rid="b25-ehp0114-001038" ref-type="bibr">Levin et al. 2003</xref>).</p><p>With the subsequent demonstration that <italic>Pfiesteria</italic> is a common inhabitant of estuarine waters in the mid-Atlantic region
and beyond (<xref rid="b21-ehp0114-001038" ref-type="bibr">Jakobsen et al. 2002</xref>; <xref rid="b39-ehp0114-001038" ref-type="bibr">Rublee et al. 2001</xref>), concerns arose about the possible health impact of chronic occupational
exposure to <italic>Pfiesteria</italic> species. Although there are anecdotal reports that watermen working in
areas where <italic>Pfiesteria</italic> were known to be present had non-specific health complaints, there has
not been clear, objective documentation of the presence or absence of
health effects associated with regular occupational exposure. In a small
case–control study (22 cases, 21 controls) from North Carolina, no
association was found between exposure and health effects, except
possibly for a deficit in visual contrast sensitivity (<xref rid="b45-ehp0114-001038" ref-type="bibr">Swinker et al. 2001</xref>); however, the study relied on fish health as a marker for the presence
of <italic>Pfiesteria</italic>. Larger cohort studies funded by the Centers for Disease Control and Prevention (CDC) have
been conducted in North Carolina and Virginia (<xref rid="b31-ehp0114-001038" ref-type="bibr">Moe et al. 2001</xref>); the North Carolina study again used fish health as its marker for exposure, whereas
the Virginia study used a combination of fish health and
molecular data. We report here the results of a 4-year study (1999–2002) of
a cohort of “high-risk” watermen and
community controls in Maryland, in which symptoms and neuropsychological
changes were linked with environmental exposure to <italic>Pfiesteria</italic> species as assessed by molecular methods.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec sec-type="methods"><title>Recruitment methods</title><p>Initial recruitment was based on a random selection of candidates from
the 1997 Maryland Department of Natural Resources (DNR) Commercial Fisheries
Licensure list, as stratified by age and ZIP code; recruitment
was restricted to watermen living in counties/ZIP codes along the eastern
shore of the Chesapeake Bay. Each participant had to average ≥10 hr/week
on Maryland Chesapeake waters or tributaries. Each had
to be healthy, with no self-reported history of past head injury, stroke, dementia, or
drug or alcohol abuse. When difficulties were encountered
in reaching the desired number of participants, this approach was
modified to a “semi-open” recruitment process, to include
referrals by previously enrolled participants. Initial efforts were
also made to recruit community control participants (who had minimal
contact with estuarine waters) using drivers’ license records
to match to enrolled watermen by ZIP code, age, and education. All applicable
human volunteer requirements were followed; the study was approved
by the institutional review board (IRB) at the University of Maryland, Baltimore
and the Maryland Department of Health and Mental Hygiene. All
participants gave written informed consent before the study.</p><p>We enrolled 123 watermen and 29 controls, for a total of 152 participants. The
average age was 47 years (range, 19–74 years); all participants
but one were male. Forty-five (30%; 35 watermen, 10 controls) of
the 152 were lost to attrition during the 4-year time period
of the study. Of the 45 who dropped out of the study, 26 (58%) cited
as their primary reason the inconvenience of the testing and
paperwork required by the study; 4 (9%) moved out of the area, and 2 died. One
hundred seven watermen participants who enrolled in
the study in year 1 completed the full 4 years of follow-up. Among watermen, those
who dropped out were slightly younger at the time of enrollment
than those who stayed in the study (42.2 years vs. 48.8 years; <italic>p</italic> = 0.02, chi-square); otherwise, there were no significant differences
between those who dropped out and those retained in the study. The
two deaths were in the exposed group; in both instances, deaths were
from causes that were independent of the exposures being evaluated
in this study.</p></sec><sec sec-type="methods"><title>Study design</title><p>Data collection centered around the four summer “seasons” in 1999, 2000, 2001, and 2002. This reflects the patterns of occupational
exposure of the watermen (whose on-the-water work year is generally
restricted by weather to spring, summer, and fall) and is in keeping
with our environmental surveys, which have shown that <italic>Pfiesteria</italic> demonstrate a clear seasonality, with organisms detected with increasing
frequency in the water column during the summer and early fall, and
then “disappearing” as winter begins (<xref ref-type="fig" rid="f1-ehp0114-001038">Figure 1</xref>). For the purposes of this study, each calendar year was broken into pre-season (February
through April), active season (May through October), and
postseason (November through January).</p><p>Biweekly monitoring was accomplished by self-reported logs or diaries. Every 2 weeks
throughout the year, each participant was sent a log covering
the previous 2-week period. The participant was asked to answer
questions dealing with how many days he/she worked, where the work occurred, what
type of fishing was engaged in, whether any type of fish “event” was
witnessed, any symptoms experienced (see <xref ref-type="app" rid="app1-ehp0114-001038">Appendix</xref>), and whether he/she was exposed to any type of known chemical toxicants. Symptom
lists were based on symptoms reported in the 1997 Pocomoke
River outbreak (<xref rid="b19-ehp0114-001038" ref-type="bibr">Grattan et al. 1998</xref>), persons with exposure in the Burkholder laboratory (<xref rid="b17-ehp0114-001038" ref-type="bibr">Glasgow et al. 1995</xref>; <xref rid="b40-ehp0114-001038" ref-type="bibr">Schmechel and Koltai 2001</xref>), and the CDC working definition of “possible estuary-associated
syndrome” (<xref rid="b11-ehp0114-001038" ref-type="bibr">CDC 1999</xref>). Fishing areas were divided into grids on standardized maps that were
given to the watermen, with map grid locations used to define where waterman
had worked during the period covered by the log report. Participants
received $50 compensation per quarter if they completed
a minimum of five of six logs; the overall completion rate for the logs
was 87.3%.</p><p>With the exception of 2001 (when post-season testing was delayed by IRB
issues), participants received a neuropsychological screening battery
preseason and postseason for 4 years. The neuropsychological screening
battery was approximately 2 hr in length and was designed to assess
a wide variety of cognitive functions that could potentially be altered
by exposure. Measures of mood, effort, and other personality or psychiatric
factors that could potentially interfere with cognitive performance
were also included. Participants received $100 compensation
for each testing session. The battery included the following components:</p><list list-type="bullet"><list-item><p>Sensory and motor: Snellen test, Functional Acuity Contrast Test (<xref rid="b14-ehp0114-001038" ref-type="bibr">Ginsburg 1993</xref>), smell test, and Lafayette Grooved Pegboard (Lafayette Instruments, Lafayette, IN)</p></list-item><list-item><p>Attention, divided attention, and concentration: Wechsler Adult Intelligence
Scale-III (WAIS-III) Digit Span (<xref rid="b51-ehp0114-001038" ref-type="bibr">Wechsler 1997</xref>), Symbol Digit Modalities Test (<xref rid="b42-ehp0114-001038" ref-type="bibr">Smith 1982</xref>), Trail-Making Test (<xref rid="b37-ehp0114-001038" ref-type="bibr">Reitan 1992</xref>), Stroop Color-Word Test (<xref rid="b18-ehp0114-001038" ref-type="bibr">Golden 1978</xref>), and WAIS-III Letter-Number Sequencing (<xref rid="b51-ehp0114-001038" ref-type="bibr">Wechsler 1997</xref>)</p></list-item><list-item><p>Memory: Rey Auditory Verbal Learning Test (<xref rid="b41-ehp0114-001038" ref-type="bibr">Schmidt 1996</xref>), Rey-Osterrieth Complex Figure Test, Rey 15-Item Memory Test (<xref rid="b38-ehp0114-001038" ref-type="bibr">Rey 1964</xref>), and recall (<xref rid="b30-ehp0114-001038" ref-type="bibr">Meyers and Meyers 1995</xref>)</p></list-item><list-item><p>Visual spatial and constructional: Rey-Osterreith Complex Figure, copy (<xref rid="b30-ehp0114-001038" ref-type="bibr">Meyers and Meyers 1995</xref>), WAIS-III, Block Design (<xref rid="b51-ehp0114-001038" ref-type="bibr">Wechsler 1997</xref>)</p></list-item><list-item><p>Verbal fluency: Controlled Oral Word Association (<xref rid="b3-ehp0114-001038" ref-type="bibr">Benton et al. 1994</xref>)</p></list-item><list-item><p>Effort: Portland Digit Recognition Test (<xref rid="b5-ehp0114-001038" ref-type="bibr">Binder 1993</xref>).</p></list-item></list><p>Measures of general intellectual functioning (Raven’s Standard
Progressive Matrices; <xref rid="b36-ehp0114-001038" ref-type="bibr">Raven et al. 1992</xref>) and reading proficiency (Wide Range Achievement Tests-3; <xref rid="b52-ehp0114-001038" ref-type="bibr">Wilkinson 1993</xref>) were taken at the time of the first (baseline visit). Personality and
mood were screened with the Profile of Mood States (<xref rid="b28-ehp0114-001038" ref-type="bibr">McNair et al. 1971</xref>, <xref rid="b29-ehp0114-001038" ref-type="bibr">1992</xref>), Beck Depression Inventory II (<xref rid="b2-ehp0114-001038" ref-type="bibr">Beck 1996</xref>), and the State Trait Anxiety Inventory (<xref rid="b43-ehp0114-001038" ref-type="bibr">Spielberger 1983</xref>). The Brief Michigan Alcoholism Screening Test (<xref rid="b35-ehp0114-001038" ref-type="bibr">Pokorny et al. 1972</xref>) and Alcohol Use Disorders Identification Test (AUDIT; <xref rid="b1-ehp0114-001038" ref-type="bibr">Babor et al. 1989</xref>) were used to determine alcohol use and history, and a blood alcohol content
screen was conducted via breathalyzer at the time of the exam. Finally, educational, occupational, neurological, psychiatric, and exposure
histories were obtained via standardized interview.</p></sec><sec><title>Environmental sampling</title><p>Water samples were collected for polymerase chain reaction (PCR)-based
monitoring for the presence of <italic>Pfiesteria</italic> and other harmful algal bloom species during the period of this study (1999–2002) as
part of an ongoing monitoring program by the Maryland
DNR (<xref ref-type="table" rid="t1-ehp0114-001038">Table 1</xref>). Samples collected by DNR staff during 1999 (<italic>n</italic> = 228), 2000 (<italic>n</italic> = 381), 2001 (<italic>n</italic> = 438), and 2002 (<italic>n</italic> = 387) were obtained from the lower eastern shore tributaries
where the enrolled watermen worked.</p><p>The overlapping study participant work-area grids and Maryland DNR <italic>Pfiesteria</italic> sampling grids provided an opportunity to analyze study outcomes (reported
symptoms and test results) with work in areas where <italic>Pfiesteria</italic> species were detected in the water column but did not provide certainty
regarding the temporal overlap of work exposure and <italic>Pfiesteria</italic> detection. We therefore engaged willing watermen in a sampling protocol
in the final two seasons of the study to further refine our correlation
of exposure estimation with outcome measures. Under the waterman sampling
protocol, potentially exposed cohort members from three general
areas (Smith Island, mainland Somerset county, and Dorchester county) took
water samples before departing their work area at the end of the
day. In 2001, watermen collected samples on a biweekly basis (<italic>n</italic> = 426), and in 2002, on a weekly basis (<italic>n</italic> = 1,677).</p><p>Samples were collected in either 50-mL tubes (watermen) or 500-mL bottles (Maryland
DNR) and fixed with 1% acidic Lugol’s solution (<xref rid="b50-ehp0114-001038" ref-type="bibr">Vollenweider 1974</xref>). Once received in the laboratory, a 50-mL aliquot was centrifuged to
pellet cells. Supernatant was removed, and total DNA was extracted using
the Puregene Genomic DNA Isolation Kit (Gentra Systems, Minneapolis, MN). Real-time
PCR assays specific for <italic>Pfiesteria piscicida</italic> and <italic>Pfiesteria shumwayae</italic> [recently renamed <italic>Pseudopfiesteria shumwayae</italic> (<xref rid="b27-ehp0114-001038" ref-type="bibr">Litaker et al. 2005</xref>)] were performed as previously described (<xref rid="b7-ehp0114-001038" ref-type="bibr">Bowers et al. 2000</xref>). We have previously demonstrated that Lugol’s-fixed specimens
can be used to detect these organisms in environmental water samples
without loss of assay sensitivity due to variations in sample processing
and delivery times (<xref rid="b7-ehp0114-001038" ref-type="bibr">Bowers et al. 2000</xref>).</p></sec><sec sec-type="methods"><title>Statistical analysis</title><p>Over the course of the study, any given waterman participant could be exposed
to <italic>Pfiesteria</italic> in one year and not exposed in another year. For this reason, we performed
a preliminary statistical analysis for each study year separately
before analyzing the pooled data. Statistical analyses for any given
period of time involved comparisons among the following groups: watermen
exposed to <italic>Pfiesteria</italic> (“exposed watermen”), watermen not exposed to <italic>Pfiesteria</italic> (“unexposed watermen”), and nonwaterman community residents (“community
controls”). The watermen were classified
as exposed or not exposed according to whether they spent time in
waters that tested positive for <italic>Pfiesteria</italic> during the in-season period. Additional analyses were also conducted with
watermen classified into four levels of exposure based on the amount
of time spent working in waters that tested positive for <italic>Pfiesteria</italic>: no exposure, low exposure, moderate exposure, and high exposure.</p><p>For purposes of analysis, symptoms were grouped into five major categories (see <xref ref-type="app" rid="app1-ehp0114-001038">Appendix</xref>): cognitive, gastrointestinal, irritation, pain, and respiratory. To determine
whether <italic>Pfiesteria</italic> exposure was associated with symptoms of various kinds, between-category
comparisons of symptom rates were made during each of the three periods
described above. A longitudinal (three time points) Poisson regression
model was fitted for each symptom category using the generalized
estimating equation (GEE) method (<xref rid="b26-ehp0114-001038" ref-type="bibr">Liang and Zeger 1986</xref>) with an offset corresponding to the time in each period covered by the
subject’s symptom logs. The dependent variable was determined
as the cumulative number of symptom episodes of a specific kind (skin, respiratory, etc.) reported by the subject in the period. The independent
variables included indicator variables for the comparison groups, for
the periods and for the period × group interactions. Relative
risks (RRs) for <italic>Pfiesteria</italic> exposure and 95% confidence intervals (CIs) were determined from
the estimated parameters of the model and their SEs. These analyses
were conducted for each year separately and then for the 4 study years
combined. Analyses were also conducted with <italic>Pfiesteria</italic> exposure as a binary (exposed/not exposed) variable and then with four
levels of exposure.</p><p>To determine whether <italic>Pfiesteria</italic> exposure influenced neurocognitive test performance, between-group comparisons
of test score means were made for the preseason and the postseason
periods (as noted above, testing was not typically performed during
the in-season period). A longitudinal (two time points) regression
model was fitted for each neuro-cognitive test using the GEE method. The
dependent variable was the <italic>z</italic>-score for a given test (standardization based upon initial test means
and SDs for the nonwaterman controls). The independent variables included
indicator variables for the comparison groups, for the periods, and
for the period × group interactions, as well as the covariates
age and education. Between-group standardized differences and 95% CIs
were determined from the estimated parameters of the model
and their SEs. As with the symptom data, the neuro-cognitive test data
were analyzed with <italic>Pfiesteria</italic> exposure treated as a binary variable and then as levels of exposure.</p></sec></sec><sec sec-type="results"><title>Results</title><sec><title>Environmental sampling</title><p>We analyzed water samples collected in the study region by the Maryland
DNR throughout the study period (<xref ref-type="table" rid="t1-ehp0114-001038">Table 1</xref>). In 1999, 4.8% of 228 DNR-collected samples were positive for <italic>P. piscicida</italic>, whereas in 2000, 1.8% of 381 DNR-collected samples were positive. Rivers
where <italic>P. piscicida</italic> was detected in the water column (<xref ref-type="fig" rid="f2-ehp0114-001038">Figure 2</xref>) included the Chicamacomico (10 of 11 samples in 1999 and 4 of 7 in 2000), the
Pocomoke (1 sample in 1999, 2 samples in 2000), and the Big Annemessex (1 sample
in 2000). <italic>P. shumwayae</italic> was not detected in either year. During 2001, <italic>P. piscicida</italic> was detected more frequently in DNR samples from the region (3.4%) than
in waterman samples (0.9%) This degree of variation
is partly influenced by more extensive DNR sampling of a particular watershed (the
Chicamacomico and Transquaking rivers) where 10 of the 15 positive
samples were collected during that year. Watermen tended not
to work in this watershed, and few waterman-submitted samples were received
from this area. Other rivers in which <italic>P. piscicida</italic> was detected in 2001 (DNR and waterman samples) included the Big Annemessex (<italic>n</italic> = 2), Manokin (<italic>n</italic> = 3), Choptank (<italic>n</italic> = 1), and Pocomoke (<italic>n</italic> = 1) rivers and Tangier Sound mainstem (<italic>n</italic> = 2). In 2001, <italic>P. shumwayae</italic> was not detected in waterman-submitted samples, but it was detected in 0.7% of
DNR samples, all from the Pocomoke River. In 2002, <italic>P. piscicida</italic> was detected in 7% of samples collected by DNR and in 2.6% of
waterman-submitted samples, a variation that we again attribute
to heavier sampling in the Chicamacomico/ Transquaking region by DNR (19 of
a total of 27 positive water samples were obtained in this watershed; six
waterman samples collected in this system were positive). Other
rivers in which <italic>P. piscicida</italic> was detected in 2002 (DNR and waterman samples) include the Tangier Sound
region (<italic>n</italic> = 15), Chesapeake Bay main-stem (<italic>n</italic> = 4), and Honga (<italic>n</italic> = 3), Nanticoke (<italic>n</italic> = 3), Choptank (<italic>n</italic> = 2), Little Choptank (<italic>n</italic> = 2), Manokin (<italic>n</italic> = 4), Pocomoke (<italic>n</italic> = 11), and Wicomico (<italic>n</italic> = 1) rivers. In 2002, <italic>P. shumwayae</italic> was detected only in water specimens submitted by watermen (0.2%), and
the organism was detected in the Pocomoke (<italic>n</italic> = 2), Honga (<italic>n</italic> = 1), and Little Choptank (<italic>n</italic> = 1) rivers and the mainstem of the bay (<italic>n</italic> = 1).</p><p>Through the course of the 4-year study, we observed a seasonal rhythm in
the presence of detectable <italic>Pfiesteria</italic> zoospores in the water column (<xref ref-type="fig" rid="f1-ehp0114-001038">Figure 1</xref>). Many dinoflagellate species in the Chesapeake have characteristic bloom
dynamics (e.g., <italic>Prorocentrum minimum</italic>; <xref rid="b48-ehp0114-001038" ref-type="bibr">Tyler and Seliger 1978</xref>), and for <italic>Pfiesteria</italic> species, it appears that the organism is most prevalent in the water column
during the late summer and early fall. In other investigations, we
have demonstrated that <italic>Pfiesteria</italic> can be detected in sediments (presumptively as cysts) throughout the year (Bowers
HB, Oldach DW, unpublished data).</p></sec><sec><title>Symptom reporting</title><p>Relative frequencies of reporting of each symptom category among watermen
and community control persons are shown in <xref ref-type="fig" rid="f3-ehp0114-001038">Figure 3</xref>. Watermen as a group did not report symptoms with significantly greater
frequency than did community controls, even after stratification by
season (preseason/active season/postseason) and year. Similarly, there
was no significant increase in the frequency of any symptom category, by
season and year, when exposed watermen (those who had worked in areas
where <italic>Pfiesteria</italic> had been detected) were compared with community controls.</p><p>When exposed watermen were compared with unexposed watermen (who had not
worked in areas where <italic>Pfiesteria</italic> had been detected), there was greater variability in the results, with
isolated increases in RR for specific symptom categories. For example, there
was a significant increase in cognitive symptoms among exposed
watermen during the active season (RR = 1.83; 95% CI, 1.22–2.70; <italic>p</italic> = 0.003) and postseason (RR = 4.63; 95% CI, 2.68–8.05; <italic>p</italic> < 0.001) in 2000; gastrointestinal symptoms, in contrast, showed an
increase in the preseason (RR = 3.86; 95% CI, 1.67–8.70; <italic>p</italic> = 0.001) and active season (RR = 1.90; 95% CI, 1.10–3.18; <italic>p</italic> = 0.016) in 2000, but no increase in the postseason. Irritation
symptoms showed an isolated increase in the preseason (RR = 2.45; 95% CI, 1.36–4.27; <italic>p</italic> = 0.002), and respiratory symptoms showed a slight increase through
all three seasons: preseason (RR = 2.16; 95% CI, 1.40–3.26; <italic>p</italic> < 0.001), active season (RR = 1.82; 95% CI, 1.36–2.42; <italic>p</italic> < 0. 001), and postseason (RR = 1.96; 95% CI, 1.43–2.66; <italic>p</italic> < 0.001).</p><p>In no instance was there a consistent pattern of increases in multiple
symptom categories in a single year, nor did we see a pattern of increase
in any one symptom during the active season (when exposures to <italic>Pfiesteria</italic> would have been most likely to have occurred) across multiple years. In
addition, no such patterns were seen regardless of whether watermen
were compared with community controls, exposed watermen were compared
with unexposed watermen, or four levels of exposure among watermen were
compared. Although sample sizes were small, results were unaffected
when the analysis was restricted to waterman-collected water samples or
when data for <italic>P. shumwayae</italic> were included.</p></sec><sec><title>Neuropsychological testing</title><p>Neuropsycho-logical data indicated no significant baseline differences
between exposed and nonexposed watermen for age (<italic>t</italic> = 1.64, <italic>p</italic> = 0.11), years of education (<italic>t</italic> = 1.08, <italic>p</italic> = 0.28), or general intellectual functioning (Raven’s
Standard Progressive Matrices total score; <italic>t</italic> = 0.72, <italic>p</italic> = 0.48). There were also no differences between the groups on
measures of mood (Profile of Mood States total mood disturbance; <italic>t</italic> = 0.08, <italic>p</italic> = 0.94), anxiety (State Trait Anxiety Inventory; <italic>t</italic> = 0.76, <italic>p</italic> = 0.45), alcohol use (AUDIT; <italic>t</italic> = 0.39, <italic>p</italic> = 0.70), or malingering [Rey 15-Item Memory Test (<xref rid="b38-ehp0114-001038" ref-type="bibr">Rey 1964</xref>); <italic>t</italic> = 0.50, <italic>p</italic> = 0.62].</p><p>When exposed and unexposed watermen were compared over time, we observed
no significant differences in test performance between the groups on
the Rey Auditory Verbal Learning Test or the Controlled Oral Word Association. In
fact, 1999 was the only year in which significant differences
were present between exposed and unexposed watermen on any of the
key measures in the neuropsychological test battery. In 1999, there was
a small but statistically significant increase in performance among
exposed watermen in the Trail-Making Test, part B, in both pre-season (score
difference = 0.62; 95% CI, 0.09–1.15) and
postseason (score difference = 0.81; 95% CI, 0.26–1.37). Also
in 1999, exposed watermen scored significantly higher
on the Lafayette Grooved Pegboard dominant hand on preseason testing (score
difference = 0.26; 95% CI, 0.01–0.51).</p><p>In no instance did exposed watermen show a pattern of neuropsychological
decline in the postseason testing compared with controls. There were
no alterations in psychological, psychiatric, or cognitive status. This
finding was true for all tests, for all years, and in comparisons with
both possible control groups (community controls and unexposed watermen).</p></sec></sec><sec sec-type="discussion"><title>Discussion</title><p>Dinoflagellates in the genus <italic>Pfiesteria</italic> were first identified in the early 1990s by <xref rid="b10-ehp0114-001038" ref-type="bibr">Burkholder et al. (1992)</xref> at North Carolina State University (Raleigh, NC) in association with fish
kills in the Pamlico and Neuse estuaries. Two <italic>Pfiesteria</italic> species, <italic>P. piscicida</italic> and <italic>P. shumwayae</italic>, have been identified (<xref rid="b15-ehp0114-001038" ref-type="bibr">Glasgow et al. 2001a</xref>, <xref rid="b16-ehp0114-001038" ref-type="bibr">2001b</xref>), although recent studies indicate that <italic>P. shumwayae</italic> should be placed in a separate genus, <italic>Pseudopfiesteria</italic> (<xref rid="b27-ehp0114-001038" ref-type="bibr">Litaker et al. 2005</xref>). Numerous other “<italic>Pfiesteria</italic>-like organisms” have been characterized by both classical taxonomic
and molecular methodologies (<xref rid="b44-ehp0114-001038" ref-type="bibr">Steidinger et al. 2001</xref>).</p><p>Early studies from the Burkholder laboratory suggested that characteristic “punched
out” skin lesions in fish were attributable
to exposure to toxic forms of <italic>P. piscicida</italic> (<xref rid="b10-ehp0114-001038" ref-type="bibr">Burkholder et al. 1992</xref>; <xref rid="b15-ehp0114-001038" ref-type="bibr">Glasgow et al. 2001a</xref>, <xref rid="b16-ehp0114-001038" ref-type="bibr">2001b</xref>). However, the linkage of <italic>Pfiesteria</italic> species to lesioned fish has been highly controversial (<xref rid="b22-ehp0114-001038" ref-type="bibr">Kane et al. 2000</xref>; <xref rid="b23-ehp0114-001038" ref-type="bibr">Law 2001</xref>; <xref rid="b49-ehp0114-001038" ref-type="bibr">Vogelbein et al. 2002</xref>). Current data indicate that most ulcerated lesions in menhaden (the fish
most commonly affected) from the Chesapeake are due to a highly invasive
fungal species, <italic>Aphanomyces invadeans</italic> (<xref rid="b6-ehp0114-001038" ref-type="bibr">Blaser et al. 1999</xref>; <xref rid="b49-ehp0114-001038" ref-type="bibr">Vogelbein et al. 2002</xref>). It has been hypothesized that fish-kill events attributed to <italic>Pfiesteria</italic>, and human health effects attributed to exposure to <italic>Pfiesteria</italic> blooms or laboratory cultures, are mediated by production of a toxic moiety
by the organism. This has also been a matter of substantial scientific
controversy, for full characterization of the presumptive hydrophilic
toxin (tentatively named PfTx) has not been achieved (<xref rid="b4-ehp0114-001038" ref-type="bibr">Berry et al. 2002</xref>; <xref rid="b9-ehp0114-001038" ref-type="bibr">Burkholder et al. 2005</xref>; <xref rid="b13-ehp0114-001038" ref-type="bibr">Fairey et al. 1999</xref>; <xref rid="b25-ehp0114-001038" ref-type="bibr">Levin et al. 2003</xref>; <xref rid="b32-ehp0114-001038" ref-type="bibr">Moeller et al. 2001</xref>; <xref rid="b49-ehp0114-001038" ref-type="bibr">Vogelbein et al. 2002</xref>). This is an ongoing area of investigation for multiple laboratories, and
confirmation of the existence of a <italic>Pfiesteria</italic> toxin falls outside the scope of the present epidemiologic investigation. Of
note, an assay to detect putative <italic>Pfiesteria</italic> toxins in environmental samples is not available and was not available
during the course of these studies.</p><p>Given uncertainty about the causal relationship between <italic>Pfiesteria</italic> species and fish lesions, the lack of specificity of both fish lesions
and fish kills as a marker for the organism, and the absence of an assay
for detection of the toxin in environmental samples, we elected to
use molecular methods (PCR) developed in our laboratories (<xref rid="b7-ehp0114-001038" ref-type="bibr">Bowers et al. 2000</xref>; <xref rid="b34-ehp0114-001038" ref-type="bibr">Oldach et al. 2000</xref>; <xref rid="b39-ehp0114-001038" ref-type="bibr">Rublee et al. 2001</xref>) to determine whether <italic>Pfiesteria</italic> was present in estuarine waters to which our cohort members were exposed. These
assays have been validated by a number of investigators and
have proven to be effective for monitoring <italic>Pfiesteria</italic> in the environment (<xref rid="b21-ehp0114-001038" ref-type="bibr">Jakobsen et al. 2002</xref>; <xref rid="b39-ehp0114-001038" ref-type="bibr">Rublee et al. 2001</xref>). We have previously demonstrated that the assay used for detection of <italic>P. piscicida</italic> in this study is equally effective for detection of <italic>Pfiesteria</italic> strains believed to have toxic and nontoxic phenotypes (in fact, these
organisms have identical 18S ribosomal DNA sequences) (<xref rid="b46-ehp0114-001038" ref-type="bibr">Tengs et al. 2003</xref>). Thus, the strategy we adopted for this 4-year field study to detect
human health effects of recurring occupational exposure to an organism
that may or may not actually make a toxin (that may or may not affect
humans), and that may or may not cause fish kills and fish lesions, was
to simply monitor for the organism itself, with an assay proven to
be able to detect “toxic” strains. The assay we used was
not quantitative, but we did not feel that adequate quantitative methods
were available at the time we were doing the study. Although we
cannot exclude the possibility that subtle differences were missed by
reliance on qualitative data, use of quantitative data would have generated
a number of additional uncertainties that would have made interpretation
of this complex data set even more difficult.</p><p>Through the first 2 years of the study, screening was restricted to water
samples collected by the Maryland DNR at designated sampling stations. When
additional funding became available during the final 2 years
of the study (2001 and 2002), we recruited watermen to collect specimens
from their work sites as described above. For the first 2 years, the
relatively large area encompassed by each of our map grid areas, coupled
with a sampling frequency of once each 2 weeks, may not have optimally
reflected the exposure of cohort members, despite the fact that
this represented a substantial improvement over previously available methodologies (lesioned
fish exposure and “<italic>Pfiesteria</italic>-like organism” counts on plankton microscopy). With the inclusion
of waterman-collected samples from 2001 and 2002, our findings remained
negative, with the added assurance that water sample data reflected
authentic “exposure” (time-and place-matched sampling). It
has been suggested that toxicity in <italic>Pfiesteria</italic> is restricted to a clonal subset of strains (<xref rid="b8-ehp0114-001038" ref-type="bibr">Burkholder et al. 2001</xref>, <xref rid="b9-ehp0114-001038" ref-type="bibr">2005</xref>). Because our assay identified all organisms within the species, independent
of possible toxicity, we cannot rule out the possibility that our
failure to observe any human health effects resulted from the lack
of “toxic” <italic>Pfiesteria</italic> in our study area during the 1999–2002 summers.</p><p>Despite these potential problems, this study represents the first systematic, multiyear
effort to correlate human health effects with exposure
to waterways where presence of <italic>Pfiesteria</italic> has been clearly documented. Given the large number of variables that
were monitored, it is not surprising that we saw occasional differences
between exposed and control populations; however, in no instance was
there a consistent pattern of responses (either of reported symptoms
or from formal neuro-psychological testing) that would suggest a health
risk arising from occupational exposure to the estuarine environment. We
saw no alterations in psychological or psychiatric status, in keeping
with our observation that the initial group of persons exposed to
the Pocomoke River were psychologically healthy with high energy, enthusiasm, and
positive mood [i.e., there was no evidence that the
initial symptom complex was related to functional or psychiatric factors (<xref rid="b47-ehp0114-001038" ref-type="bibr">Tracy et al. 1998</xref>)]. Our exposure assessments, based as they were on molecular testing, were
highly specific but may have lacked sensitivity. As noted
above, the use of fish health (the occurrence of fish lesions or fish
kills) is of uncertain validity as a marker for the organism but may
highlight the presence of toxic strains in the environment, should such
strains exist. In this context, the previously cited North Carolina
studies (which used fish health as a marker for exposure) are reassuring
in that they also failed to find a correlation between exposure and
health (except for a possible correlation with reduced visual contrast
sensitivity in the initial occupational prevalence study) (<xref rid="b31-ehp0114-001038" ref-type="bibr">Moe et al. 2001</xref>; <xref rid="b45-ehp0114-001038" ref-type="bibr">Swinker et al. 2001</xref>).</p><p>There is no question that persons exposed to the Pocomoke River in the
summer of 1997 had profound, reversible (and well-documented) neuropsychological
deficits (<xref rid="b19-ehp0114-001038" ref-type="bibr">Grattan et al. 1998</xref>). Based on findings from laboratory-exposed persons (<xref rid="b40-ehp0114-001038" ref-type="bibr">Schmechel and Koltai 2001</xref>), results of ongoing animal studies (<xref rid="b24-ehp0114-001038" ref-type="bibr">Levin 2001</xref>; <xref rid="b25-ehp0114-001038" ref-type="bibr">Levin et al. 2003</xref>), and studies that have begun to implicate specific neuro-receptors in
the observed effect (<xref rid="b12-ehp0114-001038" ref-type="bibr">El-Nabawi et al. 2000</xref>), it is plausible that <italic>Pfiesteria</italic>, in unique, isolated instances and/or in association with specific, unusually
toxic strains, can cause human health effects. However, this study, in
conjunction with similar studies from North Carolina and Virginia (<xref rid="b31-ehp0114-001038" ref-type="bibr">Moe et al. 2001</xref>; <xref rid="b45-ehp0114-001038" ref-type="bibr">Swinker et al. 2001</xref>), provides reassurance that, in the absence of an outbreak situation or
the identification of a particularly toxic strain, the routine, occupational
exposure to estuarine waters in which <italic>Pfiesteria</italic> is known to be present does not represent a significant human health risk.</p></sec>
|
Organophosphate Pesticide Exposure and Work in Pome Fruit: Evidence for
the Take-Home Pesticide Pathway
|
<p>Organophosphate (OP) pesticides are commonly used in the United States, and
farmworkers are at risk for chronic exposure. Using a sample of 218 farmworkers
in 24 communities and labor camps in eastern Washington
State, we examined the association between agricultural crop and OP pesticide
metabolite concentrations in urine samples of adult farmworkers
and their children and OP pesticide residues in house and vehicle dust
samples. Commonly reported crops were apples (71.6%), cherries (59.6%), pears (37.2%), grapes (27.1%), hops (22.9%), and
peaches (12.4%). Crops were grouped into
two main categories: pome fruits (apples and pears) and non-pome fruits. Farmworkers
who worked in the pome fruits had significantly higher
concentrations of dimethyl pesticide metabolites in their urine and
elevated azinphos-methyl concentrations in their homes and vehicles
than workers who did not work in these crops. Among pome-fruit workers, those
who worked in both apples and pears had higher urinary metabolites
concentrations and pesticide residue concentrations in dust than
did those who worked in a single pome fruit. Children living in households
with pome-fruit workers were found to have higher concentrations
of urinary dimethyl metabolites than did children of non-pome-fruit workers. Adult
urinary concentrations showed significant correlations with
both the vehicle and house-dust azinphos-methyl concentrations, and
child urinary concentrations were correlated significantly with adult
urinary concentrations and with the house-dust azinphos-methyl concentration. The
results provide support for the take-home pathway of pesticide
exposure and show an association between measures of pesticide exposure
and the number of pome-fruit crops worked by farmworkers.</p>
|
<contrib contrib-type="author"><name><surname>Coronado</surname><given-names>Gloria D.</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000999">1</xref></contrib><contrib contrib-type="author"><name><surname>Vigoren</surname><given-names>Eric M.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-000999">2</xref></contrib><contrib contrib-type="author"><name><surname>Thompson</surname><given-names>Beti</given-names></name><xref ref-type="aff" rid="af1-ehp0114-000999">1</xref></contrib><contrib contrib-type="author"><name><surname>Griffith</surname><given-names>William C.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-000999">2</xref></contrib><contrib contrib-type="author"><name><surname>Faustman</surname><given-names>Elaine M.</given-names></name><xref ref-type="aff" rid="af2-ehp0114-000999">2</xref></contrib>
|
Environmental Health Perspectives
|
<p>Organophosphate (OP) pesticides are the most widely used pesticides in
the United States, and farmworkers are at high risk for exposure. The
health effects of acute exposure to pesticides are well characterized. Previous
investigations have examined the long-term health effects among
workers after an acute or chronic low-level exposure. Studies have
reported deficits in verbal and visual attention, motor dexterity, confusion, and
lapses in memory, among others (<xref rid="b11-ehp0114-000999" ref-type="bibr">Eskenazi and Maizlish 1988</xref>; <xref rid="b28-ehp0114-000999" ref-type="bibr">McConnell et al. 1994</xref>; <xref rid="b34-ehp0114-000999" ref-type="bibr">Rosenstock et al. 1991</xref>). Others have reported elevated risks for leukemia (<xref rid="b2-ehp0114-000999" ref-type="bibr">Beane Freeman et al. 2005</xref>), non-Hodgkin lymphoma (<xref rid="b17-ehp0114-000999" ref-type="bibr">Fritschi et al. 2005</xref>), and lung cancer (<xref rid="b2-ehp0114-000999" ref-type="bibr">Beane Freeman et al. 2005</xref>), although some studies have not found significant associations (<xref rid="b4-ehp0114-000999" ref-type="bibr">Burns 2005</xref>; <xref rid="b32-ehp0114-000999" ref-type="bibr">Reynolds et al. 2005</xref>).</p><p>The U.S. Environmental Protection Agency (EPA) has published tables of
transfer coefficients that estimate the amount of treated foliage that
a farmworker contacts while performing occupational tasks on various
crops (<xref rid="b35-ehp0114-000999" ref-type="bibr">Science Advisory Council for Exposure 2000</xref>). The estimates are based on standard assumptions about protective clothing
worn by workers and the absorption rates for pesticides through
the skin or inhalation. Higher transfer rates are estimated for workers
who thin than for workers who harvest, prune, weed, irrigate, or perform
other farm tasks. The extent to which farm task is related to levels
of worker pesticide exposure when cross-sectional data are examined
remains controversial (<xref rid="b6-ehp0114-000999" ref-type="bibr">Coronado et al. 2004a</xref>, <xref rid="b7-ehp0114-000999" ref-type="bibr">2004b</xref>, <xref rid="b8-ehp0114-000999" ref-type="bibr">2004c</xref>; <xref rid="b15-ehp0114-000999" ref-type="bibr">Fenske et al. 2004</xref>; <xref rid="b23-ehp0114-000999" ref-type="bibr">Krieger and Zhang 2004</xref>).</p><p>Items to consider in assessing the relationship of farm tasks and pesticide
exposure include transfer coefficients, total amount of pesticides
applied, time of pesticide application, and the multiple agricultural
crops that farmworkers may work with. Equivalent transfer coefficients
are assigned to various orchard crops, such as apples, pears, cherries, and
peaches. Data from the <xref rid="b40-ehp0114-000999" ref-type="bibr">U.S. Department of Agriculture (USDA) (2000)</xref> show that differing crops have varying amounts of pesticides applied. Farmworkers
generally work in a variety of crops during a given growing
season, and it remains unclear how work in multiple agricultural crops
influences overall worker exposure.</p><p>Growing research interest is seen in the levels and patterns of pesticide
exposure among children of farmworkers. This emphasis was driven by
a report of the <xref rid="b31-ehp0114-000999" ref-type="bibr">National Research Council (1993)</xref> expressing concern about pesticide residues on food. Some studies have
identified possible risks for the development of cancers, birth defects, and
abnormal reflexes among children and neurologic impairments among
adults and children (<xref rid="b3-ehp0114-000999" ref-type="bibr">Blain 2001</xref>; <xref rid="b19-ehp0114-000999" ref-type="bibr">Guillette et al. 1998</xref>; <xref rid="b21-ehp0114-000999" ref-type="bibr">Kirkhorn and Schenker 2002</xref>; <xref rid="b29-ehp0114-000999" ref-type="bibr">Mills and Yang 2003</xref>; <xref rid="b33-ehp0114-000999" ref-type="bibr">Rohlman et al. 2005</xref>; <xref rid="b44-ehp0114-000999" ref-type="bibr">U.S. General Accounting Offices 2000</xref>; <xref rid="b46-ehp0114-000999" ref-type="bibr">Young et al. 2005</xref>).</p><p>Pesticide exposure in children is of concern because of the way in which
exposure occurs. Exposure is not always direct; it is generally believed
to occur from pesticides brought to the home through the take-home
pathway of farmworkers (<xref rid="b39-ehp0114-000999" ref-type="bibr">Thompson et al. 2003</xref>). Such paraoccupational exposure is important because of children’s
unique behaviors, such as greater amounts of time spent on floors
where pesticides accumulate, increased likelihood of dermal exposure
from wearing minimal clothing during the summer spray season, and increased
likelihood of pesticide ingestion from hand-to-mouth behavior (<xref rid="b30-ehp0114-000999" ref-type="bibr">Mills and Zahm 2001</xref>).</p><p>The importance of agricultural task was shown by <xref rid="b14-ehp0114-000999" ref-type="bibr">Fenske et al. (2000)</xref>, who reported that children of pesticide applicators had higher urinary
concentrations of dialkylphosphates than do children of nonagricultural
workers living in the same community (based on creatinine-adjusted
spray-season estimates). <xref rid="b24-ehp0114-000999" ref-type="bibr">Lambert et al. (2005)</xref> have reported crop-specific information showing that children whose parents
worked in pear orchards had higher concentrations of pesticide metabolites
in their urine than did children whose parents worked in berries
or cherries. Other studies have shown that children of agricultural
workers have higher exposures than children of nonagricultural workers (<xref rid="b25-ehp0114-000999" ref-type="bibr">Loewenherz et al. 1997</xref>; <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. 2000</xref>) and that pesticide metabolite levels in children’s urine correlate
with pesticide metabolite levels in adults’ urine within
the same household (<xref rid="b9-ehp0114-000999" ref-type="bibr">Curl et al. 2002</xref>). A limited number of previous investigations have examined households
for the presence of pesticide residues in dust samples where children
are thought to be at risk of exposure (<xref rid="b8-ehp0114-000999" ref-type="bibr">Coronado et al. 2004c</xref>; <xref rid="b10-ehp0114-000999" ref-type="bibr">Curwin et al. 2005</xref>; <xref rid="b16-ehp0114-000999" ref-type="bibr">Fenske et al. 2002</xref>; <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. 2000</xref>; <xref rid="b27-ehp0114-000999" ref-type="bibr">McCauley et al. 2003</xref>; <xref rid="b36-ehp0114-000999" ref-type="bibr">Shalat et al. 2003</xref>).</p><p>Using a large sample of farmworkers from several agricultural communities
in eastern Washington State, we examined the association between work
in specific agricultural crops and levels of OP pesticide exposure
among adult workers and children living in the same household. We aimed
to test the hypothesis that the take-home pathway results in children’s
exposure to pesticides.</p><sec sec-type="materials|methods"><title>Materials and Methods</title><sec><title>Setting</title><p>A study that tested a culturally appropriate intervention to interrupt
the take-home pathway of pesticide exposure provides the data for this
report. The setting, study design, study participants, and survey procedures
have been described previously (<xref rid="b39-ehp0114-000999" ref-type="bibr">Thompson et al. 2003</xref>). Briefly, the study took place in the Yakima Valley of Washington State. An
estimated 50,000 people in the region work in agriculture; the
primary crops are apples, grapes, pears, cherries, hops, and peaches (<xref rid="b40-ehp0114-000999" ref-type="bibr">USDA 2000</xref>). Approximately 50% of the area population is Hispanic, and most
work in agriculture. For most crops cultivated in the Yakima Valley, fieldwork
is done by hand.</p><p>In Washington State during 1999 (the year in which these data were collected), 172,000 acres
of farmland were dedicated to apple production; substantial
acreage was dedicated to the production of pears (24,400 acres), cherries (18,000 acres), peaches (2,500 acres), hops (25,076 acres), and
grapes (41,000 acres) (<xref rid="b40-ehp0114-000999" ref-type="bibr">USDA 2000</xref>). Yakima County ranked first among Washington State counties in number
of acres dedicated to the production of apples (75,264 acres), pears (10,190 acres), cherries (6,129 acres), peaches (1,438 acres), and hops (20,061 acres) and
second in the acres dedicated to grape production (15,529 acres) (<xref rid="b43-ehp0114-000999" ref-type="bibr">USDA 2004</xref>).</p></sec><sec><title>OP pesticide use in Washington State</title><p>One of the most commonly used pesticides in the Yakima Valley is azinphos-methyl. This
is a broad-spectrum insecticide registered for use in
the control of many insect pests on a wide variety of fruit, vegetable, nut, and
field crops as well as on ornamental plants, tobacco, and forest
and shade trees. Azinphos-methyl is classified by the U.S. EPA as
having level I toxicity. It is reported to be highly toxic through inhalation, dermal
absorption, ingestion, and eye contact (<xref rid="b12-ehp0114-000999" ref-type="bibr">Extension Toxicology Network 1996</xref>). The current reentry interval for azinphos-methyl applied to apples, pears, and
peaches is 14 days and for cherries, 15 days (<xref rid="b42-ehp0114-000999" ref-type="bibr">USDA 2003</xref>), during which time, only workers wearing protective equipment are allowed
in the fields. In 1999, the reentry interval for azinphos-methyl
was extended from 48 hr to 14 days. It is unclear to what extent the shorter
reentry interval was in practice during the 1999 season. For each
annual crop season, applications are limited to 8 pounds per acre for
apples, 6 pounds per acre for pears, and 3 and 4.5 pounds per acre
for cherries and peaches, respectively (<xref rid="b1-ehp0114-000999" ref-type="bibr">Bayer CropScience LP 2003</xref>). Data from the voluntary Washington Agricultural Statistics survey show
that in 1999 an estimated 309,300 pounds of azinphos-methyl were applied
to Washington State apple orchards and 33,000 pounds, 17,500 pounds, and 2,000 pounds
were applied to pear, cherry, and peach orchards, respectively (<xref rid="b40-ehp0114-000999" ref-type="bibr">USDA 2000</xref>). Azinphos-methyl is generally not used in hop or grape production (<xref rid="b41-ehp0114-000999" ref-type="bibr">USDA 2001</xref>; <xref rid="b45-ehp0114-000999" ref-type="bibr">Washington Association of Wine Growers 2004</xref>).</p><p>Other OP pesticides were also used on Yakima Valley crops in 1999 (<xref rid="b40-ehp0114-000999" ref-type="bibr">USDA 2000</xref>): Phosmet was applied to Washington State apples and pears in estimated
quantities of 46,000 pounds and 20,600 pounds, respectively; malathion
was applied to apples and peaches in quantities of 22,300 pounds and 1,500 pounds, respectively; methyl-parathion was applied to apples and
pears in quantities of 17,100 pounds and 1,400 pounds, respectively; and
chlorpyrifos was applied to apples, pears, peaches, cherries, and
grapes in quantities of 250,900 pounds, 28,300 pounds, 1,300 pounds, 20,600 pounds, and 8,000 pounds, respectively.</p></sec><sec><title>Questionnaire</title><p>An in-person interview was conducted with the farmworkers. The main questionnaire
was a 73-item instrument that included nine sections. Workers
were asked whether or not they had worked with apples, pears, peaches, cherries, hops, or
grapes in the previous 3 months, and to name any
additional crops. Workers were also asked whether in the previous 3 months
they had performed the following agricultural job tasks: harvesting
or picking; pruning; loading; packing; sorting or grading plants, fruits, or
vegetables; planting or transplanting; weeding; thinning; irrigating; mixing
or loading farm chemicals; spraying or applying pesticides; or
other tasks. Before implementation, the questionnaire was
translated into Spanish, piloted among farmworkers, and reviewed and edited
by members of a community advisory board. The questionnaire and
all study procedures were reviewed and approved by the Fred Hutchinson
Cancer Research Center. All adult participants and parents of child participants
gave informed consent to participate in the study.</p></sec><sec sec-type="methods"><title>Survey procedures</title><p>Recruitment procedures have been described previously (<xref rid="b39-ehp0114-000999" ref-type="bibr">Thompson et al. 2003</xref>). Briefly, individuals who worked in agriculture were recruited using
an in-person survey of randomly selected households conducted for another
study. Additional workers were recruited from labor camps and from
areas in the community known to have a large concentration of agricultural
workers. A total of 571 households were surveyed; 218 of these households
had age-eligible children (2–6 years of age) and agreed
to enroll in the specimen collection aspect of the study. This subgroup
forms the sample basis for this report.</p><p>Among eligible households (those with a farmworker and age-eligible child), an
adult respondent and study child were identified (<xref rid="b39-ehp0114-000999" ref-type="bibr">Thompson et al. 2003</xref>). We collected urine samples from the adult farmworker and study child
as well as dust samples from selected areas of the home and the vehicle
used to commute to and from work. Samples were collected between June
and October 1999.</p></sec><sec sec-type="methods"><title>Specimen collection and laboratory analysis</title><p>Procedures for the urine and dust collection and laboratory analysis have
been described in detail elsewhere (<xref rid="b9-ehp0114-000999" ref-type="bibr">Curl et al. 2002</xref>). Two or three spot urine samples were collected. Each collection was
separated by a minimum of 3 days and collected within a 2-week period, with
the first collection occurring at the interview. For each individual, equal
volumes of each urine sample were combined before specimen
analysis. This provided an estimate of pesticide exposure within the 2-week
period of assessment.</p><p>Urine samples were analyzed using gas chromatographic procedures for the
presence of five dialkylphosphate compounds produced by the metabolism
of most OP pesticides: dimethylphosphate (DMP), dimethylthio-phosphate (DMTP), dimethyldithiophosphate (DMDTP), diethylphosphate, and diethyl-thiophosphate. The
limits of detection for these compounds were 7.2 μg/L, 1.1 μg/L, 0.65 μg/L, 2.9 μg/L, and 1.2 μg/L, respectively.</p><p>Dust samples were collected from homes and commuter vehicles using a Nilfisk
vacuum cleaner unit (model GS-80; Nilfisk of America, Malvern, PA), and
sampling took place within 4 weeks of the interview. Selection
of the area to be vacuumed was determined by asking the parent or adult
participant where the child played most frequently. The size of the
area vacuumed depended on the floor type and ranged from a 1 m × 1 m
area for plush carpets to a 2 m × 2 m area for hard or
smooth floors. Foot wells from both the front and back of cars (and only
the front of trucks) were vacuumed. Mats were not removed before vacuuming. Dust
samples were analyzed using gas chromatographic procedures
for the presence of four dimethyl OP pesticides (azinphos-methyl, malathion, methyl-parathion, and phosmet) and two diethyl OP pesticides (chlorpyrifos
and diazinon). The limits of detection for these pesticides
residues and the percentage of analyzed samples that contained detectable
levels are shown in <xref ref-type="table" rid="t1-ehp0114-000999">Table 1</xref>. For this report, we limit our analyses of the dust samples to the azinphos-methyl
residues because in a large number of samples other pesticides
were below detection levels. Because azinphos-methyl is a dimethyl
OP pesticide, we limited our analyses of the urine samples to the dimethyl
metabolites.</p></sec><sec sec-type="methods"><title>Statistical and classification methods</title><p>Apples and pears are classified as pome fruit: a fleshy fruit having several
seed chambers. Peaches and cherries are stone fruits, containing
a single seed or pit. Grapes and hops are vine/trellis and bunch/bundle
crops, respectively. We grouped workers into two categories based on
crop type. Pome-fruit workers were those who worked in apples and/or
pears and possibly peaches, cherries, grapes, hops, and other crops. Non-pome-fruit
workers worked only in peaches, cherries, grapes, hops, or
other crops. We grouped pears with apples because they have similar
harvest seasons, and both are chemically and hand-thinned during the
pesticide spray season (June through August). The rates and types of
pesticide application are similar for both crops.</p><p>We present the frequencies of detection and estimated geometric mean (GM) concentrations
and geometric standard deviations of adult and child
urinary dimethyl metabolites and azinphos-methyl residues in vehicle
and house-dust samples. Concentrations for the urinary samples are presented
in units of micrograms per liter and are not creatinine adjusted. Concentrations
for the dust samples are presented in units of micrograms
per gram.</p><p>Quantile–quantile plots (not shown) demonstrate that the concentrations
of the urinary metabolites and of pesticide residues in dust
were approximately normally distributed after a log-transformation. Because
some samples had values below the limits of detection, we modeled
the missing data in a multivariate normal hierarchical Bayesian simulation
model with conjugate noninformative priors using WinBUGS (Windows-based
Bayesian inference Using Gibbs Sampling) (<xref rid="b38-ehp0114-000999" ref-type="bibr">Spiegelhalter et al. 2003</xref> ). The data below the limits of detection were treated as left censored
data with an upper cutoff at the limit of detection (<xref rid="b18-ehp0114-000999" ref-type="bibr">Griffith et al. 2002</xref>). The statistical model assumed a common variance/covariance structure
while allowing for a shift in the means for different farmworker classifications. After
a burn-in simulation run, we performed 250,000 simulations. Confidence
limits for the log-normal parameters (the 95% posterior
predictive probability intervals) were estimated. To examine
differences in the means of farmworker groups, at each simulation
we compared the means to see which were larger. These comparisons were
recorded and summed, and we report the probability that one GM was greater
than another.</p></sec></sec><sec sec-type="results"><title>Results</title><p>In this study, 571 (89.6%) of the 627 farm-workers identified for
the study were interviewed. A small percentage of households (3.8%) could
not be contacted after at least five visits, and 6.6% of
eligible farmworkers refused to participate, giving an overall
response rate of 89.6%, or 93.1% of the known eligibles. Of
the 571 respondents, 231 households (40.5%) included
children 2–6 years of age. Of these, 218 households were available
for sample collection. The total number agreeing to provide urine
samples included 213 adult farmworkers (92.2% of those eligible) and 211 children (91.3% of those eligible). House-dust samples
were collected from 210 homes, and of the households with vehicles (<italic>n</italic> = 207), 205 (99.0%) allowed us to collect vehicle dust. For 54 homes
and 15 vehicles, insufficient masses of dust were collected
for analysis; thus, pesticide residue analysis was conducted on 156 house-dust
samples and 190 vehicle-dust samples.</p><p>Three-quarters of the farmworkers in our study sample reported working
in apples within the 3 months before being interviewed (<xref ref-type="table" rid="t2-ehp0114-000999">Table 2</xref>). More than one-third worked in pears, and nearly two-thirds worked in
cherries. Approximately one-tenth of workers worked in peaches and nearly
one-quarter worked in grapes or hops. Fewer than one-third of the
respondents worked in other crops, which included asparagus, apricots, plums, peppermint, corn, and onions.</p><p>When farmworkers reported working in multiple crops, apples were the crop
with the greatest overlap with other crops. All of the peach workers
and all but two pear workers had also worked in apples. The percentages
of cherry, grape, hop, and other crop workers who had also worked
in apples were 80.8, 69.5, 62.0, and 61.9%, respectively.</p><p>Those who worked in pome fruits were slightly older than those who worked
in non-pome fruits (<xref ref-type="table" rid="t3-ehp0114-000999">Table 3</xref>). Pome-fruit workers, on average, had fewer years of education and had
lower household incomes than did non-pome-fruit workers; those working
in both apples and pears on average had the lowest education and income
levels. The groups were similar in marital status. Birthplace differed
slightly by crop category. Approximately one-fifth of pome-fruit
workers reported having worked in agriculture for 20 or more years; the
proportion was about one-fourth for non-pome-fruit workers. Most adult
participants were male and completed the survey in Spanish.</p><p>Eight of the farmworkers had DMP concentrations that were orders of magnitude
higher than other study participants and ranged from 3,780 to 12,000 μg/mL (the
remaining study participants’ concentrations
ranged from less than the limit of detection to 100 μg/mL). All
eight farmworkers reported working in apples, and half had also
worked in pears; seven had worked as thinners. We performed our analysis
with and without the data from these farmworker households. The
two analyses showed similar relationships between urinary metabolite
and dust concentrations and crop category; however, those households were
influential in estimating the GMs and the geometric standard deviations. Including
them in the analysis made it unclear whether the crop
effect was primarily a result from these eight households, or whether
there was a crop effect among all who worked in pome fruit. By excluding
them, we could show that the association between crop category and
metabolite and dust concentrations was present among the rest of the
population. Therefore, measurements taken from these eight households
are not represented in <xref ref-type="table" rid="t4-ehp0114-000999">Tables 4</xref>–<xref ref-type="table" rid="t8-ehp0114-000999">8</xref>; the results reported in these tables are based on 210 farmworker households.</p><p>Examining the dimethyl urinary metabolites among workers who did or did
not work in pome fruits, we observed differences in the frequency of
detection and in the metabolite concentrations (<xref ref-type="table" rid="t4-ehp0114-000999">Table 4</xref>). Workers who reported working in pome fruit had higher concentrations
of dimethyl metabolites than did non-pome-fruit workers; GM concentrations
were 2.4-fold, 3.5-fold, and 2.9-fold higher for DMP, DMTP, and
DMDTP, respectively. Among pome-fruit workers, those who worked in both
apples and pears had the highest dimethyl metabolite concentrations.</p><p>Children in our study had patterns of exposure that were similar to those
in adults (<xref ref-type="table" rid="t4-ehp0114-000999">Table 4</xref>). Children who lived in households with a farmworker who worked in pome
fruit had greater frequency of detection of dimethyl metabolites than
did children living in households with a non-pome-fruit worker. The
frequency of detection for DMP, for example, was 7.1% for children
living in households with non-pome-fruit workers and 22.5% for
children living with pome-fruit workers. GM concentrations of DMP, DMTP, and
DMDTP were 2.6-fold, 1.7-fold, and 1.5-fold higher, respectively, for
children who lived in households with a pome-fruit worker. Among
children living with a pome-fruit worker, those who lived in households
with an apple and pear worker had higher dimethyl metabolite
concentrations than those who lived in households with an apple or pear
worker only.</p><p>Differences between the crop worker groups are seen in the concentrations
of azinphos-methyl residue in the dust samples (<xref ref-type="table" rid="t5-ehp0114-000999">Table 5</xref>). More than 90% of those who worked in pome fruit had detectable
azinphos-methyl in their vehicles and homes, compared with only slightly
more than 60% for the non-pome-fruit workers. Among pome-fruit
workers, those who worked in both apples and pears had the greatest
percentage detection and higher concentrations of azinphos-methyl
in their house and vehicle dust. Those who worked in pome fruit had GM
concentrations of azinphos-methyl in their vehicle and house dust that
were 6.8-fold and 4.6-fold greater, respectively, than those for farmworkers
who did not work in pome fruit.</p><p>The estimated correlations for the dimethyl urinary metabolite concentrations
and the azinphos-methyl concentrations in house and vehicle dust
that we observed are given in <xref ref-type="table" rid="t6-ehp0114-000999">Table 6</xref>. Within both the adult and the child urine samples, there were statistically
significant high positive correlations between the dimethyl metabolite
concentrations, particularly for DMTP and DMDTP concentrations. The
vehicle- and house-dust concentrations of azinphos-methyl were also
highly correlated. Urine samples of children and adults living in
homes with elevated levels of dust indicate exposure to higher levels
of pesticides; adult urine DMP and DMTP concentrations showed significant
correlations with both the vehicle- and house-dust azinphos-methyl
concentrations, and child urine DMP and DMTP concentrations were correlated
significantly with the house-dust azinphos-methyl concentration.</p><p>When we examined differences in urinary metabolite concentrations among
those who did and did not perform thinning within pome-fruit and non-pome-fruit
worker categories, we found no remarkable differences between
the groups (<xref ref-type="table" rid="t7-ehp0114-000999">Table 7</xref>). Similarly, there were no notable differences in concentrations of azinphos-methyl
and vehicle and dust samples between thinners and nonthinners
when categorized by work in pome fruit (<xref ref-type="table" rid="t8-ehp0114-000999">Table 8</xref>).</p></sec><sec sec-type="discussion"><title>Discussion</title><p>We assessed OP pesticide exposure among farmworkers based on whether on
not they worked in pome-fruit crops. Those who worked in pome fruit had
higher concentrations of OP pesticide metabolites in their urine and
higher concentrations of azinphos-methyl residues in dust collected
from their homes and vehicles. The increased presence of pesticide-laden
dust probably contributed to the pesticide exposure of children in
the household.</p><p>Previously, we reported higher proportions of urine samples with detectable
levels of the OP pesticide urinary metabolite DMTP from children
of farmworkers who reported thinning, compared with urine samples from
children of nonthinners. We also reported higher proportions of detectable
azinphos-methyl in vehicle and house dust among workers who reported
thinning (<xref rid="b8-ehp0114-000999" ref-type="bibr">Coronado et al. 2004c</xref>). The present findings suggest that agricultural crop was an important
factor that was not considered in our previous study; 91.4% of
thinners reported having worked in pome fruits in the 3 months before
the survey. After controlling for work in pome-fruit crops, our data
revealed no significant differences between workers who did or did not
report thinning, for both percent detection and concentration measurements.</p><p>The higher urinary metabolite levels found in pome-fruit workers matched
pesticide use patterns in Washington State. Data from the Washington
Agricultural Statistics survey show that in 1999 the rate of application
of azinphos-methyl was higher in apples (1.8 lb/acre) than in pears (1.4 lb/acre), cherries (1.0 lb/acre), or peaches (0.8 lb/acre) (<xref rid="b40-ehp0114-000999" ref-type="bibr">USDA 2000</xref>); thus, our data appear to show higher exposure levels among those who
worked in crops with the highest pesticide applications.</p><p>Particularly noteworthy is the distinct gradient we see in the azinphos-methyl
concentrations in vehicle and house dust. Samples collected from
households of farm-workers who worked in both apples and pears had
the higher pesticide concentrations than did samples collected from households
of farmworkers who worked in only one pome fruit (<xref ref-type="table" rid="t5-ehp0114-000999">Table 5</xref>). Samples collected from households of farmworkers who did not work with
pome fruit had the lowest concentrations. The association between higher
adult and child urinary OP pesticide metabolite concentrations and
work with increasing numbers of pome fruit was also observed (<xref ref-type="table" rid="t4-ehp0114-000999">Table 4</xref>). These findings suggest the potential for cumulative pesticide exposure
among workers performing tasks on multiple crops.</p><p>We found no association within worker groups between pesticide and metabolite
concentrations and self-reported information about the recent application
of pesticides in the workplace. This suggests that pesticides
that accumulate in the home environment may account in part for the
urinary metabolite levels of adults and children, and that pome-fruit
workers are more likely to track home pesticides. The high positive correlation
between azinphos-methyl levels in house and vehicle dust that
we observed along with the statistically significant positive correlations
between the dust concentrations and urinary metabolite concentrations
provides perhaps the strongest evidence for the take-home pathway (<xref ref-type="table" rid="t6-ehp0114-000999">Table 6</xref>).</p><p>The Centers for Disease Control and Prevention (<xref rid="b5-ehp0114-000999" ref-type="bibr">CDC 2005</xref>) reported that the GM concentration of the OP pesticide urinary metabolite
DMTP for adults 20–59 years of age in the general population
is 1.47 μg/L averaged across a year. We observed concentrations
three times higher in the spray season for farmworker adults of
similar ages who did not work in pome-fruit crops (GM = 4.4 μg/L) and
concentrations > 10 times higher for those who did (GM = 15.3 μg/L). Although the CDC report provides no
data for children in the same age group as our study, it does give a GM
urinary metabolite concentration of 2.95 μg/L for DMTP in children 6–11 years
of age. Our findings for children 2–6 years
of age are 1.2 times higher than this for those in households
of farmworkers who did not work in pome fruits (GM = 3.5 μg/L) and > 2.1 times higher for those in households of farmworkers
who did work in pome fruits (GM = 6.2 μg/L). Because
our children were younger, they may have higher concentrations of metabolites
than do older children, as has been reported in a limited number
of previous investigations on the topic (<xref rid="b25-ehp0114-000999" ref-type="bibr">Loewenherz et al. 1997</xref>; <xref rid="b36-ehp0114-000999" ref-type="bibr">Shalat et al. 2003</xref>).</p><p>Few previous pesticide exposure investigations have examined household
dust. We found higher concentrations of azinphos-methyl in house and vehicle
dust among workers who reported having worked in the pome fruits
versus farmworkers who did not report working in these fruits. Our GM
concentration of azinphos-methyl in house dust (0.79 μg/g) among
pome-fruit workers is comparable with that reported for agricultural
workers by <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. (2000)</xref> (median concentration = 1.0 μg/g combined value for pesticide
applicators and farmworkers in pome-fruit–growing region
in Washington State) and <xref rid="b36-ehp0114-000999" ref-type="bibr">Shalat et al. (2003)</xref> (median concentration = 0.51 μg/g for homes in an agricultural
community near the U.S.–Mexico border).</p><p>Our analyses support the notion that children of farmworkers are exposed
through pesticides that are tracked into homes. Exposures among adult
workers and children living in the same household varied with the agricultural
crop in which the adult worked. This finding would not be expected
if dietary ingestion of pesticides or home use of pesticides were
primary sources of exposure because these factors are unlikely to
be related to work in given agricultural crops. Our analyses, however, indicate
that pesticide residues found in house and vehicle dust were
significantly greater for workers who worked in pome fruit than for those
who did not, providing evidence for increased exposure to pesticides
that are subsequently tracked from the fields to worker vehicles and
homes. Moreover, among the pome-fruit workers, we observed higher concentrations
of azinphos-methyl in vehicle dust than in house dust, consistent
with expectations that pesticides are carried from work-places
to vehicles and eventually homes on workers’ clothing, hats, and
boots.</p><p>Analyses of urine from children of farm-workers reported by other studies
also provide support for a take-home pathway. <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. (2000)</xref> showed that median DMTP concentrations in agricultural children were four
times higher than those of nonagricultural children. A study by <xref rid="b22-ehp0114-000999" ref-type="bibr">Koch et al. (2002)</xref> collected urine samples from 44 children on a biweekly basis over a period
of 21 months. Data from the study show that children had higher GM
concentrations of combined dialkylphosphates during the spray months
compared with the nonspray months. These findings support the take-home
pathway but do not discount the contributions of other pathways. Our
data set was collected shortly after the spray season. In a previous
analysis of this data, <xref rid="b9-ehp0114-000999" ref-type="bibr">Curl et al. (2002)</xref> demonstrated a strong correlation between urinary dimethyl metabolite
concentrations of adult farmworkers and children living in the same household, a
finding that would be expected with the take-home pathway. Studies
that examined pesticide residues in dust samples also provide
support for the take-home pathway. Separate studies conducted by <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. (2000)</xref> and <xref rid="b14-ehp0114-000999" ref-type="bibr">Fenske et al. (2000)</xref> report higher median house-dust concentrations of azinphos-methyl in homes
of agricultural families compared with nonagricultural families, and <xref rid="b27-ehp0114-000999" ref-type="bibr">McCauley et al. (2003)</xref> showed that house-dust concentrations rose with increasing numbers of
agricultural workers in a household.</p><sec><title>Limitations</title><p>This study has some limitations. It is possible that some urine and dust
samples were collected relatively late in the spray season. Although
we report higher urinary metabolite concentrations for those who worked
in crops where higher amounts of pesticides were used, the late-season
collection of urine samples may have increased the likelihood of detecting
metabolites from pesticides used on crops that are harvested
late in the season (e.g., apples and pears). Nevertheless, we relied on
local knowledge and previous research conducted in the area to determine
when to begin collection (<xref rid="b37-ehp0114-000999" ref-type="bibr">Simcox et al. 1999</xref>).</p><p>Apart from the possibility that the late-season collection may have attenuated
our reported exposure estimates, it is likely that our urinary
metabolite concentrations underestimate levels during peak exposures. Information
from the Washington State Tree Fruit Research Commission
shows that temperature readings for the spring of 1999 were much lower
than normal and resulted in the recommendation to growers to cancel or
delay the first several sprayings of the season (<xref rid="b22-ehp0114-000999" ref-type="bibr">Koch et al. 2002</xref>). The first application of OP pesticide for the 1999 season is reported
to have been in April, and the number of applications in that season
was reduced; thus, the levels of OP pesticide urinary metabolites in
our samples may have been lower than for other years. Our urinary metabolite
estimates may have been further attenuated because peak excretion
of metabolites occurs relatively quickly (24–48 hr after exposure) (<xref rid="b13-ehp0114-000999" ref-type="bibr">Feldmann and Maibach 1974</xref>) and we did not time our collections according to spray events. Despite
these potential limitations, we were able to find significant differences (based
on crop category) both in urinary OP metabolite concentrations
in adults and children and in pesticide concentrations in home and
vehicle dust.</p><p>Questions asked about work in the past 3 months resulted in a high percentage
of workers reporting work in multiple crops; thus, our ability
to discern differences in exposure levels associated with given crops
was limited. An additional limitation of our assessment of occupational
pesticide exposure is that we did not include number of hours worked
per week. This factor could account for a substantial fraction of pesticide
exposures and could explain differences in urinary pesticide metabolite
concentrations. Although we do not believe this limits the conclusions
we can draw from our results, further investigations on this
topic would benefit by collection of this variable and its inclusion
in analysis.</p><p>Farmworkers as a group are difficult populations to assess (<xref rid="b20-ehp0114-000999" ref-type="bibr">Kamel et al. 2001</xref>; <xref rid="b47-ehp0114-000999" ref-type="bibr">Zahm and Blair 2001</xref>). The strength of this study is the large sample size and the substantial
variation in types of agricultural crops reported in our sample. Moreover, our
data reflect the real-life experience of farmworkers in that
most work in multiple crops. Two unique features of our statistical
analyses techniques were strengths. First, we were able to concurrently
examine the relationships between adult and child urinary OP pesticide
concentrations and pesticide residues in house and vehicle dust based
on crop and job task categories, whereas previous investigations
have generally considered only two-way comparisons. Second, we used a
modeling technique to estimate values that were below the limit of detection (<xref rid="b18-ehp0114-000999" ref-type="bibr">Griffith et al. 2002</xref>); this offered greater predictive power compared with some previous investigations
that have assigned a single value to data points that are
below the limit of detection or have considered only data points above
the limit of detection. Previous investigations have recruited farmworkers
from a select number of farms or community organizations (<xref rid="b14-ehp0114-000999" ref-type="bibr">Fenske et al. 2000</xref>; <xref rid="b22-ehp0114-000999" ref-type="bibr">Koch et al. 2002</xref>; <xref rid="b25-ehp0114-000999" ref-type="bibr">Loewenherz et al. 1997</xref>; <xref rid="b26-ehp0114-000999" ref-type="bibr">Lu et al. 2000</xref>; <xref rid="b30-ehp0114-000999" ref-type="bibr">Mills and Zahm 2001</xref>; <xref rid="b37-ehp0114-000999" ref-type="bibr">Simcox et al. 1999</xref>). Growers who volunteer their farms for participation in studies may promote
more protective work practices or apply fewer applications of pesticides, thus
potentially biasing collected data. Our household recruitment
process attempted to minimize this bias.</p></sec></sec><sec sec-type="conclusions"><title>Conclusions</title><p>Two general conclusions can be drawn from our findings. First, our findings
demonstrate the potential for increased pesticide exposure among
workers performing tasks on multiple crops. This is shown by the elevated
urinary OP metabolite concentrations (in adults and children) and
pesticide concentrations in dust samples in those who worked in both applies
and pears, compared with those who worked in a single pome fruit. Second, our
findings support the notion that pesticides are tracked
into homes of workers, where children are exposed. This is demonstrated
by the correlation in the quantity and type of pesticides found in
the home and accompanying vehicle, the finding that adults had higher
urinary OP pesticide metabolites than did children, and the direct correlation
between the concentration of pesticides in house dust and the
concentrations of urinary OP metabolites in both adults and children.</p></sec>
|
Dpb11, the budding yeast homolog of TopBP1, functions with the checkpoint clamp in recombination repair
|
<p>Dpb11 is required for the loading of DNA polymerases α and ɛ on to DNA in chromosomal DNA replication and interacts with the DNA damage checkpoint protein Ddc1 in <italic>Saccharomyces cerevisiae</italic>. The interaction between the homologs of Dpb11 and Ddc1 in human cells and fission yeast is thought to reflect their involvement in the checkpoint response. Here we show that <italic>dpb11-1</italic> cells, carrying a mutated Dpb11 that cannot interact with Ddc1, are defective in the repair of methyl methanesulfonate (MMS)-induced DNA damage but not in the DNA damage checkpoint at the permissive temperature. Epistatic analyses suggested that Dpb11 is involved in the Rad51/Rad52-dependent recombination pathway. Ddc1 as well as Dpb11 were required for homologous recombination induced by MMS. Moreover, we found the <italic>in vivo</italic> association of Dpb11 and Ddc1 with not only the HO-induced double-strand break (DSB) site at <italic>MAT</italic> locus but also the donor sequence <italic>HML</italic> during homologous recombination between <italic>MAT</italic> and <italic>HML</italic>. Rad51 was required for their association with the <italic>HML</italic> donor locus, but not with DSB site at the <italic>MAT</italic> locus. In addition, the association of Dpb11 with the <italic>MAT</italic> and <italic>HML</italic> locus after induction of HO-induced DSB was dependent on Ddc1. These results indicate that, besides the involvement in the replication and checkpoint, Dpb11 functions with Ddc1 in the recombination repair process itself.</p>
|
<contrib contrib-type="author"><name><surname>Ogiwara</surname><given-names>Hideaki</given-names></name></contrib><contrib contrib-type="author"><name><surname>Ui</surname><given-names>Ayako</given-names></name></contrib><contrib contrib-type="author"><name><surname>Onoda</surname><given-names>Fumitoshi</given-names></name></contrib><contrib contrib-type="author"><name><surname>Tada</surname><given-names>Shusuke</given-names></name></contrib><contrib contrib-type="author"><name><surname>Enomoto</surname><given-names>Takemi</given-names></name><xref rid="au1" ref-type="aff">1</xref></contrib><contrib contrib-type="author"><name><surname>Seki</surname><given-names>Masayuki</given-names></name><xref ref-type="corresp" rid="cor1">*</xref></contrib><aff><institution>Molecular Cell Biology Laboratory, Graduate School of Pharmaceutical Sciences, Tohoku University</institution><addr-line>Aoba 6-3, Aramaki, Aoba-ku, Sendai 980-8578, Japan</addr-line></aff><aff id="au1"><sup>1</sup><institution>Tohoku University 21st Century COE Program ‘Comprehensive Research and Education Center for Planning of Drug development and Clinical Evaluation’</institution><addr-line>Sendai, Miyagi 980-88578, Japan</addr-line></aff>
|
Nucleic Acids Research
|
<sec><title>INTRODUCTION</title><p>DNA replication in eukaryotic cells is carried out by three essential DNA polymerases, Polα, Polδ and Polɛ, which are loaded onto chromatin at replication origins by a number of loading factors, including Dpb11 (<xref ref-type="bibr" rid="b1">1</xref>). <italic>DPB11</italic> was first cloned as a multi-copy suppressor of a temperature-sensitive mutant of the DNA polymerase ɛ subunit, Dpb2, and acts as a suppressor of certain mutants of the catalytic subunit (Pol2) of Polɛ, which is essential for DNA replication in eukaryotic cells (<xref ref-type="bibr" rid="b2">2</xref>). One of the initial steps in eukaryotic chromosomal DNA replication is the assembly of minichromosome maintenance (MCM) proteins into pre-replicative complexes (pre-RCs) at the end of mitosis (<xref ref-type="bibr" rid="b3">3</xref>). The association of Dpb11 with origins of replication is dependent on the presence of certain components of the MCM complex, RPA, and Dpb2, but not Polα-primase. Dpb11 co-immunoprecipitates with Polɛ from S-phase cell extracts, but only weakly from G<sub>1</sub> cell extracts (<xref ref-type="bibr" rid="b4">4</xref>). After replication origin firing, the MCM complex dissociates from origins and translocates to neighboring DNA sites with kinetics that resemble those of replication fork progression (<xref ref-type="bibr" rid="b4">4</xref>). In one temperature sensitive <italic>dpb11</italic> mutant, <italic>dpb11-1</italic>, the MCM complex binds to origin DNA but is unable to dissociate at the restrictive temperature. RPA binds to origin DNA independently of Dpb11 function, but neither Polɛ nor Polα is loaded on to origin DNA in <italic>dpb11-1</italic> mutants at the non-permissive temperature (<xref ref-type="bibr" rid="b4">4</xref>). Thus, although pre-RCs containing MCM and RPA are formed in <italic>dpb11-1</italic> mutants, DNA synthesis is not initiated in these cells.</p><p>Dpb11 is associated with Sld2, which is a substrate for the cyclin-dependent kinase (CDK1) in <italic>Saccharomyces cerevisiae</italic> (<xref ref-type="bibr" rid="b5">5</xref>,<xref ref-type="bibr" rid="b6">6</xref>). The phosphorylation of multiple S/T residues in Sld2 at the onset of replication initiation increases Sld2 affinity for Dpb11 and is necessary for the initiation of DNA replication (<xref ref-type="bibr" rid="b6">6</xref>). Thus, Dpb11 performs its DNA polymerase loading function at origins by forming a complex with Sld2.</p><p>Besides its function in the initiation of DNA replication, Dpb11 also plays a role in the DNA replication checkpoint (<xref ref-type="bibr" rid="b2">2</xref>,<xref ref-type="bibr" rid="b4">4</xref>). In addition, Dpb11 interacts with Ddc1 of budding yeast (<xref ref-type="bibr" rid="b7">7</xref>), which is one of the components of the Ddc1–Rad17–Mec3 complex involved in the DNA damage checkpoint (<xref ref-type="bibr" rid="b8">8</xref>). The Ddc1–Rad17–Mec3 complex is structurally related to proliferating cell nuclear antigen (PCNA), ‘sliding clamp’, whose homotrimeric structure allows it to encircle DNA (<xref ref-type="bibr" rid="b9">9</xref>,<xref ref-type="bibr" rid="b10">10</xref>). Rad24–Rfc2-5, which is also required for the DNA damage checkpoint (<xref ref-type="bibr" rid="b11">11</xref>), loads the Ddc1–Mec3–Rad17 complex onto damaged sites on DNA in a manner analogous to that of RFC-mediated loading of PCNA onto sites of DNA replication (<xref ref-type="bibr" rid="b12">12</xref>–<xref ref-type="bibr" rid="b14">14</xref>). Interestingly, fission yeast Rad9 and human RAD9, which are the homologs of budding yeast Ddc1, interact with the Dpb11 homologs, fission yeast Cut5 and human TopBP1, respectively (<xref ref-type="bibr" rid="b15">15</xref>,<xref ref-type="bibr" rid="b16">16</xref>). Thus, this interaction has been conserved throughout evolution. The interaction is dependent on the phosphorylation of Rad9 in fission yeast, and is required for both checkpoint signaling and DNA metabolism (<xref ref-type="bibr" rid="b15">15</xref>). Similar to its yeast counterpart, human TopBP1 is required for cell survival, DNA replication, DNA damage checkpoint and transcriptional regulation (<xref ref-type="bibr" rid="b17">17</xref>). However, the meaning of the interaction between Dpb11 and Ddc1 has remained obscure in budding yeast.</p><p>The <italic>dpb11-1</italic> mutant is sensitive to methyl methanesulfonate (MMS) and ultraviolet (UV) light even at the permissive temperature (<xref ref-type="bibr" rid="b2">2</xref>), although the mutant is proficient in the G<sub>2</sub>/M DNA damage checkpoint following UV irradiation at the non-permissive temperature (<xref ref-type="bibr" rid="b7">7</xref>). This suggests that Dpb11 is involved in the DNA repair process itself, although little is known about what this function might be. To elucidate the repair function of Dpb11, we performed a number of experiments with the <italic>dpb11-1</italic> mutant of <italic>S.cerevisiae</italic>, whose gene product fails to interact with Ddc1 (<xref ref-type="bibr" rid="b7">7</xref>).</p></sec><sec sec-type="materials|methods"><title>MATERIALS AND METHODS</title><sec><title>Yeast strains</title><p>Yeast strains used in this study were listed in Supplementary Table 1. The <italic>dpb11-1</italic> mutant was constructed by two-step gene replacement using the <italic>Pst</italic>I fragment of the plasmid YIplac121dpb11-1ΔN, which was provided by Dr H. Araki. Complementation of the thermosensitivity of the <italic>dpb11-1</italic> mutant was confirmed by transforming a single-copy plasmid carrying the wild-type <italic>DPB11</italic> gene, which was also provided by Dr H. Araki. The <italic>apn1::hisG-URA3-hisG</italic> was constructed by one-step gene replacement using the <italic>Bam</italic>HI–<italic>Eco</italic>RI fragments of plasmids pSCP108 (these plasmids were from Dr B. Demple). Complete null mutants, Myc-tagged and HA-tagged alleles were made by standard PCR-based gene disruption and insertion methods (<xref ref-type="bibr" rid="b18">18</xref>–<xref ref-type="bibr" rid="b20">20</xref>). Gene disruption was confirmed by genomic PCR. Some double-mutant strains were isolated by tetrad dissection. Complete sequences of the primers used for the construction of DNA used to disrupt an appropriate gene or to check a disruption will be provided upon request, along with the details of the strains. All analyses were performed at 23°C (i.e. at the permissive temperature of the <italic>dpb11-1</italic> mutant) unless otherwise indicated.</p></sec><sec><title>Analysis of MMS and UV sensitivity</title><p>Strains with the MR93-28c background were constructed. Logarithmically growing cells were serially diluted by 10-fold with distilled water and spotted onto YPAD plates or YPAD plates containing the indicated concentrations of MMS, or irradiated with the indicated doses of UV light. The plates were incubated for 4 days and photographed. Alternatively, logarithmically growing cells were diluted to 10<sup>7</sup> cells/ml and cultured in the presence of 15 μg/ml nocodazole for 200 min, which synchronized them in the G<sub>2</sub>/M phase. The G<sub>2</sub>/M-arrested cells were exposed to 0.1% MMS for 1 h, washed to remove MMS, and then cultured in YPAD medium with or without nocodazole. At specific intervals, the cells were washed to remove nocodazole, and then diluted and inoculated onto YPAD plates.</p></sec><sec><title>Assay of the frequency of interchromosomal homologous recombination between heteroalleles</title><p>Strains with the MR101 background were constructed such that recombination between the heteroalleles <italic>his1-1</italic> and <italic>his1-7</italic> in a diploid could be detected by the restoration of histidine prototrophy. As described previously (<xref ref-type="bibr" rid="b21">21</xref>), logarithmically growing cells were inoculated onto SC-His plates containing various concentrations of MMS and on to YPAD plates to evaluate survival. When UV-induced recombination was assayed, cells were inoculated onto SC-His plates and onto YPAD plates, and irradiated with given doses of UV. After 4 days incubation, colonies were enumerated. The numbers of His<sup>+</sup> colonies per 10<sup>6</sup> survivors are indicated. The data show a typical result of at least three independent experiments, and line bars indicate the standard deviation.</p></sec><sec><title>Pulsed-field gel electrophoresis</title><p>Logarithmically growing cells were diluted to 10<sup>7</sup> cells/ml and cultured in the presence of 15 μg/ml nocodazole for 200 min, which synchronized them in the G<sub>2</sub>/M phase. Cells arrested in the G<sub>2</sub>/M phase were exposed to 0.1% MMS for 1 h and were then washed to remove MMS and cultured in YPAD medium containing nocodazole for specific periods. Agarose plugs of chromosomal DNA were prepared following the protocol described previously (<xref ref-type="bibr" rid="b22">22</xref>). After electrophoresis, the gel was stained with 0.5 μg/ml ethidium bromide for 30 min, destained in deionized water for 20 min, and photographed.</p></sec><sec><title>G<sub>2</sub>/M phase DNA damage checkpoint</title><p>Logarithmically growing cells were arrested with nocodazole and treated with 0.3% MMS for 30 min. At the indicated times after release from nocodazole treatment, the percentage of uninucleate large budded cells was scored by DAPI staining. To detect eventual modification of Rad53-13Myc and Chk1-13Myc in response to DNA damage, cells were arrested in the G<sub>2</sub>/M phase in YPAD medium containing nocodazole for 3 h, after which 0.2% MMS was added. After 2.5 h, the cells were harvested for immunoblotting analysis. The cells were incubated at 30°C for 5 min in 100 mM K-EDTA and 10 mM DTT, followed by incubation at 30°C for 5 min in 0.6 M sorbitol, 50 mM KH<sub>2</sub>PO<sub>4</sub>/K<sub>2</sub>HPO<sub>4</sub>, 10 mM DTT and 2.5 mg/ml zymolyase 100T. The cells were washed twice with wash buffer (0.4 M sorbitol, 100 mM KCl, 2.5 mM MgCl, 50 mM HEPES–KOH, 1 mM phenylmethlysulfonyl fluoride and protease inhibitors). A total of 50 μl of the cell pellet was resuspended in 900 μl of wash buffer and the cells were lysed by the addition of 1 vol of lysis buffer (wash buffer with 2% Triton X-100). Lysis was performed on ice for 10 min, after which SDS–polyacrylamide sample buffer was added. For western blot analysis, equivalent volumes of whole cell extract (WCE) were loaded. The proteins were detected by immunoblotting with anti-Myc.</p></sec><sec><title>Chromatin immunoprecipitation (ChIP)</title><p>ChIP was carried out as described previously with minor modifications (<xref ref-type="bibr" rid="b23">23</xref>,<xref ref-type="bibr" rid="b24">24</xref>). Asynchronous cultures were grown overnight at 30°C in YP medium containing 2% raffinose. When cultures reached logarithmic phase, expression of HO endonuclease was induced by adding 2% galactose. The cultures were washed to remove galactose 1 h after the addition of galactose and treated with 2% glucose to repress the expression of HO endonuclease. Cells were harvested, and proteins were cross-linked to DNA with 1% formaldehyde for 20 min, followed by quenching with glycine at a final concentration of 125 mM for 5 min. The cells were lysed with glass beads, and the extracts were sonicated to shear DNA to an average size of 1 kb. Extracts were then divided into immunoprecipitate (IP) and input samples (20:1 ratio). Immunoprecipitation of cross-linked DNA was carried out with a monoclonal anti-Myc antibody (9E10) (Santa Cruz Biotechnology, Inc.) and Dynabeads Protein G (Dynal Biotech) for 4 h at 4°C. After a series of wash of the protein-bound beads, proteins were released from beads by reversal of cross-linking (6 h at 65°C). The samples were treated with proteinase K, and DNA for PCR was prepared by phenol extraction and ethanol precipitation.</p></sec><sec><title>PCR amplification</title><p>To amplify immunoprecipitated DNA, primers, P1 (5′-TCCCCATCGTCTTGCTCT-3′) and P2 (5′-GCATGGGCAGTTTACCTTTAC-3′), which span the HO recognition site, or P1 (5′-TCCCCATCGTCTTGCTCT-3′) and P3 (5′-CCCAAGGCTTAGTATACACATCC-3′), which span the <italic>HML</italic>α locus were used in combination with primers, SMC2-FW1 (5′-GACGACCTTGTAACAGTCCAGACAG-3′) and SMC2-RV1 (5′-GGCGAATTCCATCACATTATACTAACTACGG-3′), which anneal to the <italic>SMC2</italic> locus used as the control. PCR products were separated by agarose gel electrophoresis, and the amount of the product was determined using NIH Image. Since the IP efficiencies of different loci were often very different, the signal from the interested locus was first normalized with the signal from an independent locus (<italic>SMC2</italic>) on chromosome VI, namely, by dividing each <italic>MAT</italic> or <italic>HML</italic> IP signal by the corresponding <italic>SMC2</italic> IP signal. Then, <italic>MAT</italic> or <italic>HML</italic> IP signals at later time points were normalized to the IP signal at 0 h as 1. ChIP data are representative of at least three independent experiments.</p></sec><sec><title>Detection of DSB induction and strand invasion</title><p>Double-strand break (DSB) induction (<italic>MAT</italic><bold>a</bold> locus) and strand invasion were detected by PCR on input DNA. The primers used for detection of DSB and strand invasion were 5′-CTTTTAGTTTCAGCTTTCCG-3′ (pI)/5′-ACTCTATAAGGCCAAATGTACAAAC-3′ (pJ) and 5′-GCAGCACGGAATATGGGACT-3′ (pA)/5′-ATGTGAACCGCATGGGCAGT-3′ (pB), respectively. PCR products were separated by agarose gel electrophoresis. For quantification, the signal from the interested locus was normalized with the signal from an independent locus (<italic>SMC2</italic>) on chromosome VI.</p></sec></sec><sec><title>RESULTS</title><sec><title>Dpb11 is required for DNA repair</title><p>The <italic>dpb11-1</italic> mutant is sensitive to MMS and UV light even at the permissive temperature (<xref ref-type="bibr" rid="b2">2</xref>), although the mutant is proficient in the G<sub>2</sub>/M DNA damage checkpoint following UV irradiation at the non-permissive temperature (<xref ref-type="bibr" rid="b7">7</xref>). This suggests that Dpb11 is involved in the DNA repair process itself, although little is known about what this function might be.</p><p>To elucidate the repair function of Dpb11, we performed a number of experiments with the <italic>dpb11-1</italic> mutant. <italic>DPB11</italic> is essential for cell growth and a mutant allele, <italic>dpb11-1</italic>, causes cell death at non-permissive temperatures (<xref ref-type="bibr" rid="b2">2</xref>). All assays in this study were performed at the permissive temperature since <italic>dpb11-1</italic> shows an S-phase defect at non-permissive temperatures (<xref ref-type="bibr" rid="b2">2</xref>). We first examined whether Dpb11 is required for the repair of MMS-induced DNA damage in G<sub>2</sub>/M-arrested cells to avoid any complications due to the S-phase defect of the <italic>dpb11-1</italic> mutant. When cells arrested in the G<sub>2</sub>/M phase by nocodazole were treated with MMS and then cultured in MMS-free medium containing nocodazole to maintain their G<sub>2</sub>/M-arrested state, the distinctive bands corresponding to each chromosome disappeared upon exposure to MMS in both wild-type and <italic>dpb11-1</italic> cells and a low-molecular weight smear due to damaged DNA appeared instead (<xref ref-type="fig" rid="fig1">Figure 1A</xref>). The restoration of chromosome-sized DNA bands occurred in both wild-type and <italic>dpb11-1</italic> cells after culture in MMS-free medium; however, the restoration occurred less efficiently in <italic>dpb11-1</italic> cells than in wild-type cells (<xref ref-type="fig" rid="fig1">Figure 1A</xref>). Indeed, <italic>dpb11-1</italic> cells arrested in the G<sub>2</sub>/M phase showed higher sensitivity to MMS than wild-type cells (<xref ref-type="fig" rid="fig1">Figure 1B</xref>). We clarified whether the G<sub>2</sub>/M DNA damage checkpoint is intact in <italic>dpb11-1</italic> cells by monitoring their mitotic division following exposure to MMS. As shown in <xref ref-type="fig" rid="fig1">Figure 1C</xref>, wild-type and <italic>dpb11-1</italic> cells showed delayed nuclear division while the <italic>rad24</italic> mutant, which has a defective DNA damage checkpoint, proceeded through mitosis faster than wild-type cells. Thus, the checkpoint function of <italic>dpb11-1</italic> cells remains intact upon exposure to MMS (<xref ref-type="fig" rid="fig1">Figure 1C</xref>), as it does upon exposure to UV (<xref ref-type="bibr" rid="b7">7</xref>). To confirm this, we examined the phosphorylation of Rad53 and Chk1, which are involved in the damage checkpoint and are phosphorylated during its activation (<xref ref-type="bibr" rid="b25">25</xref>). Phosphorylated forms of Rad53 and Chk1 appeared in G<sub>2</sub>/M-arrested <italic>dpb11-1</italic> cells as well as in wild-type cells upon exposure to MMS (<xref ref-type="fig" rid="fig1">Figure 1D</xref>). These results indicate that Dpb11 is itself required for DNA repair.</p></sec><sec><title>Dpb11 is involved in MMS-induced recombination repair</title><p>We next performed an epistatic analysis of <italic>dpb11-1</italic> and mutants of various repair pathways involved in base excision repair (BER: <italic>apn1</italic>), nucleotide excision repair (NER: <italic>rad1</italic>), post-replication-repair (PRR: <italic>rad18</italic>) and homologous recombination repair (HR: <italic>rad52</italic>). The double mutants <italic>apn1 dpb11-1, rad1 dpb11-1</italic> and <italic>rad18 dpb11-1</italic> showed an additive or synergistic sensitivity to MMS and UV compared with the corresponding single mutants (<xref ref-type="fig" rid="fig2">Figure 2A</xref>), suggesting that Dpb11 functions in a pathway other than BER, NER and PRR. However, the sensitivity to MMS of <italic>rad52 dpb11-1</italic> cells was not additive to that of <italic>rad52</italic> cells, while their sensitivity to UV was greater than that of either single mutant. Although the above results seem to indicate the involvement of Dpb11 in the Rad52 repair pathway, this notion is not firmly supported by the low sensitivity of <italic>dpb11-1</italic> cells to the concentration of MMS employed. To confirm that Dpb11 actually functions in the Rad52-dependent recombination repair, we examined whether Dpb11 is involved in MMS-induced interchromosomal recombination between the heteroallelic loci <italic>his1-1</italic> and <italic>his1-7</italic> in diploid <italic>dpb11-1</italic> cells. In wild-type cells, the recombination frequency was significantly increased by exposure to MMS or UV. The induction of UV-induced recombination in <italic>dpb11-1</italic> cells is at the same level as that of wild-type cells (<xref ref-type="fig" rid="fig2">Figure 2C</xref>). However, the induction level of recombination in <italic>dpb11-1</italic> cells upon exposure to MMS was low compared with that of wild-type cells (<xref ref-type="fig" rid="fig2">Figure 2B</xref>). Taken together, it seems likely that Dpb11 is required for MMS-induced interchromosomal recombination.</p><p>Homologous recombination mediated by Rad52, as well as by Rad51, Rad54, Rad55 and Rad57, is the dominant pathway for the repair of DSBs in <italic>S.cerevisiae.</italic> On the other hand, DSBs can be also repaired by Rad52-dependent recombination pathway requiring Rad59 function (<xref ref-type="bibr" rid="b26">26</xref>–<xref ref-type="bibr" rid="b29">29</xref>). We, therefore, carried out another epistatic analysis with <italic>dpb11-1</italic> and <italic>rad51</italic> or <italic>rad59</italic> in cells exposed to MMS. The <italic>dpb11-1</italic> mutation increased the MMS sensitivity of the <italic>rad59</italic> mutant (<xref ref-type="fig" rid="fig3">Figure 3A</xref>) but not of the <italic>rad51</italic> mutant (<xref ref-type="fig" rid="fig3">Figure 3B</xref>). We further examined the relationship between Dpb11 and Rad51, Rad52, or Rad59 in MMS-induced interchromosomal recombination. The frequency of MMS-induced recombination in the <italic>dpb11-1 rad59</italic> mutant was lower than those of the single mutants, unlike <italic>dpb11-1 rad51</italic> and <italic>dpb11-1 rad52</italic> mutants (<xref ref-type="fig" rid="fig3">Figure 3C</xref>). These results suggest that Dpb11 plays a role in recombination repair and is involved in the Rad52/Rad51-dependent homologous recombination pathway upon exposure to MMS.</p></sec><sec><title>Ddc1, which interacts with Dpb11, is also required for the recombination repair</title><p>Dpb11 physically interacts with Ddc1 (<xref ref-type="bibr" rid="b7">7</xref>), which is involved in the damage checkpoint (<xref ref-type="bibr" rid="b8">8</xref>). However, <italic>dpb11-1</italic> encodes a protein that is unable to interact with Ddc1 (<xref ref-type="bibr" rid="b7">7</xref>). The biological significance of the interaction between Dpb11 and Dcd1 is not clear at present but it is worth noting that the fission yeast and human homologs of Dpb11 (Cut5 and TopBP1, respectively) also interact with their Ddc1 counterpart (Rad9) (<xref ref-type="bibr" rid="b15">15</xref>,<xref ref-type="bibr" rid="b16">16</xref>). To examine whether Ddc1 mediates the repair function of Dpb11, we monitored DNA repair in G<sub>2</sub>/M phase-arrested <italic>ddc1</italic> cells (to avoid the deficient of checkpoint function) as well as the frequency of MMS-induced interchromosomal recombination. As in <italic>dpb11-1</italic> cells, the repair of DNA damages induced by MMS was considerably impaired in <italic>ddc1</italic> cells compared with wild-type cells (<xref ref-type="fig" rid="fig4">Figure 4A</xref>), suggesting that Ddc1 is also required for the repair of MMS-induced DNA damages.</p><p>This reveals that the Ddc1–Mec3–Rad17 checkpoint clamp is directly involved in DSB repair. We also examined whether Ddc1 is required for recombination repair as well as Dpb11. As shown in <xref ref-type="fig" rid="fig4">Figure 4B</xref>, MMS-induced recombination was severely affected in <italic>ddc1</italic> cells compared with wild-type cells. It is known that the Ddc1–Mec3–Rad17 and Rad24–RFC2-5 (RAD24–RFC) complexes function in a DNA damage checkpoint pathway (<xref ref-type="bibr" rid="b9">9</xref>,<xref ref-type="bibr" rid="b10">10</xref>). Moreover, it is known that Ddc1–Mec3–Rad17 shares sequence and structural similarities with the DNA polymerase sliding clamp, PCNA while Rad24–RFC shares sequence and structural similarities with the PCNA clamp loader (replication factor C complex RFC1-5) (<xref ref-type="bibr" rid="b30">30</xref>). Therefore, Ddc1–Mec3–Rad17 and Rad24–RFC are referred to as the checkpoint clamp and checkpoint clamp loader, respectively. Similar to the <italic>ddc1</italic> mutant, the <italic>rad24</italic> mutant showed less efficient DNA repair (<xref ref-type="fig" rid="fig4">Figure 4A</xref>) and a defect in the induction of homologous recombination upon exposure to MMS (<xref ref-type="fig" rid="fig4">Figure 4B</xref>). Similar results were obtained with the <italic>mec3</italic> and <italic>rad17</italic> mutants (data not shown). To elucidate the relationship between Ddc1 and Rad24 in MMS-induced interchromosomal recombination, we measured the recombination frequencies in various mutants. The recombination frequency in <italic>ddc1 rad24</italic> double mutants was almost the same as that in <italic>ddc1</italic> or <italic>rad24</italic> single mutants (<xref ref-type="fig" rid="fig4">Figure 4C</xref>). These results indicated that the Ddc1–Mec3–Rad17 complex functions in the same pathway as Rad24–RFC2-5 (RAD24–RFC) in the recombination repair as well as the checkpoint response.</p></sec><sec><title>Ddc1 functions in the same recombination pathway as Dpb11</title><p>Dpb11 and Ddc1 physically interact with each other (<xref ref-type="bibr" rid="b7">7</xref>). We showed each of Dpb11 and Ddc1 is involved in recombination repair (<xref ref-type="fig" rid="fig2">Figures 2B</xref> and <xref ref-type="fig" rid="fig4">4B</xref>). We further confirmed the relationship of Dpb11 and Ddc1 in recombination repair. <italic>dpb11-1</italic> and <italic>ddc1</italic> cells showed a reduced frequency of recombination compared with wild-type cells, but the frequency of MMS-induced recombination in <italic>ddc1 dpb11-1</italic> cells was reduced to the same level as that in <italic>ddc1</italic> cells (<xref ref-type="fig" rid="fig4">Figure 4D</xref>), indicating that <italic>DPB11</italic> is epistatic with <italic>DDC1</italic> in the MMS-induced recombination repair pathway. These results indicated that the Dpb11-interacting protein Ddc1 plays a role in facilitating recombination repair in addition to its DNA damage checkpoint function.</p></sec><sec><title>Dpb11 and Ddc1 are recruited onto not only the recipient <italic>MAT</italic> locus but also the donor <italic>HML</italic> locus upon induction of DSB by HO endonuclease</title><p>One of the best studied homologous recombination events is the HO endonuclease-induced <italic>MAT</italic> switching, using <italic>HML</italic> as the donor template (<xref ref-type="bibr" rid="b31">31</xref>). Introduction of a galactose-inducible <italic>GAL::HO</italic> gene into cells provides the means to induce DSB synchronously in all cells of the population. It has been shown that the DSB end of <italic>MAT</italic> DNA produced by HO endonuclease is resected by 5′−3′ exonucleases (<xref ref-type="bibr" rid="b32">32</xref>) to make a 3′ ended ssDNA, which recruits the Rad51-strand exchange protein, and the DNA–protein complex invades into donor DNA. Thus, Rad51 associates not only with the HO-cut <italic>MAT</italic> locus but also the <italic>HML</italic> donor strand (<xref ref-type="bibr" rid="b23">23</xref>,<xref ref-type="bibr" rid="b24">24</xref>,<xref ref-type="bibr" rid="b33">33</xref>,<xref ref-type="bibr" rid="b34">34</xref>). To confirm the involvement of Dpb11 and Ddc1 in recombination repair, we used this system. DSB formation at the HO recognition site and invasion of the <italic>MAT</italic> recipient strand into the donor strand (<italic>HML</italic>) were monitored by PCR using the primers shown in <xref ref-type="fig" rid="fig5">Figure 5A</xref>. The amount of PCR product of <italic>MAT</italic><bold>a</bold> locus was considerably decreased 1 h after induction of endonuclease, i.e. cleavage at the <italic>MAT</italic> locus occurred within 1 h with high efficiency, and strand invasion was detected 2 and 4 h after induction (<xref ref-type="fig" rid="fig5">Figure 5B</xref>). Association of Ddc1 and Dpb11 with the <italic>MAT</italic> locus and the <italic>HML</italic> donor strand was monitored by ChIP assay using primers shown in <xref ref-type="fig" rid="fig5">Figure 5C</xref>. As reported previously (<xref ref-type="bibr" rid="b35">35</xref>,<xref ref-type="bibr" rid="b36">36</xref>), the association of Ddc1 to the <italic>MAT</italic> locus was observed 1 h after induction of HO endonucelease when DSB was formed (<xref ref-type="fig" rid="fig5">Figure 5B</xref> and D). Importantly, Ddc1 also associated with the <italic>HML</italic> donor strand (<xref ref-type="fig" rid="fig5">Figure 5D</xref>). Similar results were obtained with Dpb11, showing the association of Dpb11 with the <italic>MAT</italic> locus and the <italic>HML</italic> donor strand (<xref ref-type="fig" rid="fig5">Figure 5E</xref>).</p></sec><sec><title>Rad51 is required for the association of Dpb11 and Ddc1 with the donor <italic>HML</italic> locus but not the DSB site at <italic>MAT</italic> locus</title><p>Dpb11 and Ddc1 were required for recombination repair (<xref ref-type="fig" rid="fig3">Figures 3</xref> and <xref ref-type="fig" rid="fig4">4</xref>) and associated with the DSB site and the donor strand for recombination repair after induction of DSB (<xref ref-type="fig" rid="fig5">Figure 5</xref>). Next, we examined whether the association of Dpb11 and Ddc1 onto the DSB site and the donor strand is dependent on the recombination protein Rad51. In <italic>rad51</italic> null mutant cells, the cleavage at the <italic>MAT</italic> locus occurred within 1 h with high efficiency, but strand invasion was not detected even 4 h after induction of HO endonuclease (<xref ref-type="fig" rid="fig5">Figure 5B</xref>). The association of Ddc1 and Dpb11 with the DSB site at <italic>MAT</italic> locus was not affected in <italic>rad51</italic> mutant cells (<xref ref-type="fig" rid="fig5">Figure 5D</xref> and E). However, Dpb11 and Ddc1 did not associate with the <italic>HML</italic> donor strand in <italic>rad51</italic> mutant cells (<xref ref-type="fig" rid="fig5">Figure 5D</xref> and E). These results suggest that Dpb11 and Ddc1 are recruited to the <italic>HML</italic> donor strand following Rad51-mediated strand invasion.</p></sec><sec><title>Ddc1 is required for the association of Dpb11 with not only the DSB site but also the donor strand</title><p>We next examined whether the association of Dpb11 with the damaged DNA is dependent on Ddc1. In <italic>ddc1</italic> null mutant cells, the cleavage at the <italic>MAT</italic> locus occurred within 1 h with high efficiency and strand invasion occurred with similar kinetics to those of wild-type cells (<xref ref-type="fig" rid="fig5">Figure 5B</xref>). However, in <italic>ddc1</italic> mutant cells, the increase of association of Dpb11 with the <italic>MAT</italic> and <italic>HML</italic> locus was not observed after induction of HO endonuclease (<xref ref-type="fig" rid="fig5">Figure 5E</xref>). These results indicate that Dpb11 is recruited on to the DSB site and the donor site by Ddc1.</p></sec></sec><sec><title>DISCUSSION</title><p>In budding yeast, Dpb11 is involved in the initiation of DNA replication and replication checkpoint (<xref ref-type="bibr" rid="b2">2</xref>,<xref ref-type="bibr" rid="b4">4</xref>). The Ddc1–Rad17–Mec3 complex is involved in the activation of DNA damage checkpoint pathway as a damage sensor (<xref ref-type="bibr" rid="b8">8</xref>). The budding yeast Dpb11 interacts with Ddc1, while the meaning of the interaction has remained obscure. Here we show that both Dpb11 and Ddc1 function in the same pathway of MMS-induced recombination repair.</p><p>Using the <italic>dpb11-1</italic> mutant that expresses the protein truncated the C-terminal region to be incapable of interaction with Ddc1 (<xref ref-type="bibr" rid="b2">2</xref>,<xref ref-type="bibr" rid="b7">7</xref>), we showed that <italic>dpb11-1</italic> cells are not defective in the function of DNA damage checkpoint unlike <italic>ddc1</italic> cells (<xref ref-type="bibr" rid="b7">7</xref>,<xref ref-type="bibr" rid="b8">8</xref>) but both <italic>dpb11-1</italic> and <italic>ddc1</italic> cells are defective in MMS-induced recombination (<xref ref-type="fig" rid="fig2">Figures 2B</xref> and <xref ref-type="fig" rid="fig4">4B</xref>). After the induction of HO endonuclease, Dpb11 and Ddc1 associated with not only the <italic>MAT</italic>/DSB site but also the <italic>HML</italic> donor region (<xref ref-type="fig" rid="fig5">Figure 5D</xref> and E). Importantly, the association of Dpb11 and Ddc1 with the <italic>HML</italic> donor region did not occur in <italic>rad51</italic> mutants (<xref ref-type="fig" rid="fig5">Figure 5B</xref>). These results suggest that Dpb11 and Ddc1 function in some step of recombination repair after strand invasion mediated by Rad51 in the DSB-induced recombination repair process. Thus, we propose that besides the involvement in checkpoint responses, Dpb11 and Ddc1 are also involved in the recombination repair process itself.</p><p>How do the checkpoint clamp and the checkpoint clamp loader participate in DSB-induced recombination? In the models of DSB-induced recombination repair, the two ends of DSB are resected by 5′–3′ exonucleases, which allows the 3′ ends to invade an intact homologous template. This process involves DNA-strand exchange proteins such as Rad52 and Rad51, before new DNA synthesis initiates from 3′ ends at 3′-primer/template junctions (<xref ref-type="bibr" rid="b37">37</xref>). <italic>In vitro</italic> study showed that the checkpoint clamp loader has been shown to load the checkpoint clamp onto DNA substrates, including 3′-recessed primer templates in an ATP-dependent manner (<xref ref-type="bibr" rid="b14">14</xref>). The checkpoint clamp loader Rad24–RFC loads the checkpoint clamp Ddc1–Mec3–Rad17 at DSB site created in the <italic>MAT</italic> locus by HO endonuclease (<xref ref-type="bibr" rid="b35">35</xref>,<xref ref-type="bibr" rid="b36">36</xref>), that is thought to be resulted in the activation of checkpoint response. Interestingly, Lisby <italic>et al</italic>. (<xref ref-type="bibr" rid="b38">38</xref>) reported that Ddc1 foci appear almost simultaneously with Mre11 foci after gamma irradiation. However, in contrast to Mre11 foci they last much longer and the turnover kinetics of Ddc1 foci are similar to the turnover of Rad52 foci. Importantly, we observed that Ddc1 bound to the homologous donor strand depending on Rad51 after induction of DSB (<xref ref-type="fig" rid="fig5">Figure 5D</xref>). These results seem to support our notion that Ddc1 as well as Dpb11 are involved in the later step of homologous recombination.</p><p>How does Dpb11 participate in recombination? It has been reported that DNA polymerase ɛ and δ are redundantly involved in recombination repair (<xref ref-type="bibr" rid="b39">39</xref>). Dpb11 loads DNA polymerase ɛ at origins of DNA replication (<xref ref-type="bibr" rid="b4">4</xref>). The CDK1 is important for a latter step in homologous recombination that occurs after strand invasion but before the initiation of DNA synthesis (<xref ref-type="bibr" rid="b40">40</xref>). In this context, it is worth noting that Dpb11 forms a complex with the CDK1 phosphorylation substrate, Sld2, and that the Dpb11–Sld2 complex is essential for chromosomal DNA replication (<xref ref-type="bibr" rid="b5">5</xref>,<xref ref-type="bibr" rid="b6">6</xref>). We show here that Dpb11 associates with the homologous donor strand depending on Rad51. In addition, the association of Dpb11 with the homologous donor strand is also dependent on Ddc1 (<xref ref-type="fig" rid="fig5">Figure 5E</xref>). These results suggest that Dpb11 functions in the later step of recombination repair by the aid of Ddc1.</p><p>Considering these observations together with our results, in the later step of recombination repair, i.e. after the strand invasion by Rad51, Ddc1–Mec3–Rad17 checkpoint clamp associates with 3′-primer/template junctions of the homologous donor strand, and recruits Dpb11. And then, these proteins may recruit DNA polymerases to the recombination intermediates for DNA polymerization at damaged site. Interestingly, the interaction between Dpb11 and Ddc1 has been conserved throughout evolution. Indeed, fission yeast Rad9 and human RAD9, which are homologs of budding yeast Ddc1, interact with the Dpb11 homologs, fission yeast Cut5 and human TopBP1, respectively (<xref ref-type="bibr" rid="b15">15</xref>,<xref ref-type="bibr" rid="b16">16</xref>). The interaction with Cut5 is dependent on the phosphorylation of Rad9 in fission yeast, and is required for both checkpoint signaling and DNA metabolism (<xref ref-type="bibr" rid="b15">15</xref>). Thus, to date, the interaction between the homologs of Dpb11 and Ddc1 has been thought to reflect their involvement in the checkpoint response (<xref ref-type="bibr" rid="b41">41</xref>–<xref ref-type="bibr" rid="b43">43</xref>). The results presented here suggest that the interaction between Dpb11 and Ddc1 is required for recombination repair process itself. It is known that efficient homologous recombination contributes to maintenance of genome stability. Recent analysis of physiological function of human TopBP1, which is the homolog of yeast Dpb11, in S-phase by using small interfering RNA, revealed that TopBP1 seems to monitor ongoing DNA replication and play a critical role in the maintenance of genomic stability during normal S-phase as well as following genotoxic stress (<xref ref-type="bibr" rid="b44">44</xref>). Taking into account our current data concerning the function of yeast Dpb11, it is interesting to examine whether the human counterpart, TopBP1, is also involved in damage-induced homologous recombination in the future.</p><p>Thus, on a more general level, our observations provide a novel insight about genome stability into how proteins involved in replication, repair and checkpoint engage in more intimate crosstalk than previously held up until now.</p></sec>
|
Allelic drop-out may occur with a primer binding site polymorphism for the commonly used RFLP assay for the -1131T>C polymorphism of the <italic>Apolipoprotein AV </italic>gene
|
<p><italic>Apolipoprotein AV (ApoAV) </italic>gene variant, -1131T>C, is associated with increased triglyceride concentrations in all ethnic groups studied. An <italic>MseI </italic>based RFLP analysis is the most commonly used method for genotyping this SNP. We genotyped a large cohort comprising 1185 Asian Indians and 173 UK Caucasians for -1131T>C using an ARMS-PCR based tetra-primer method. For quality control, we re-genotyped approximately 10% random samples from this cohort utilizing the <italic>MseI </italic>RFLP, which showed a 2.9% (3/102) genotyping error rate between the two methods. To investigate further, we sequenced the 900 bp region around the -1131T>C polymorphism in 25 Asian Indians and 15 UK Caucasians and found a number of polymorphisms including the -987C>T polymorphism. Further analysis of the -987C>T SNP showed a higher rare allele frequency of 0.23 in Asian Indians (n = 158) compared to 0.09 in the UK Caucasians (n = 157). This SNP is located 4 bp from the 3' end of the RFLP forward primer and is in weak linkage disequilibrium with -1131T>C variant (r<sup>2 </sup>= 0.084 and D' = 1). Repeated RFLP analysis of seven subjects heterozygous for -987C>T (seven times), showed discordant results with the sequence at -1131T>C SNP nearly one third (15/49) of the time. We conclude that presence of -987C>T polymorphism in the forward primer of the <italic>MseI </italic>RFLP assay may lead to allelic drop-out and generate unforeseen errors in genotyping the -1131T>C polymorphism. Our results also emphasise the need for careful quality control in all molecular genetic studies, particularly while transferring genotyping methods between various ethnic groups.</p>
|
<contrib id="A1" contrib-type="author"><name><surname>Ward</surname><given-names>Kirsten J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Ellard</surname><given-names>Sian</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Yajnik</surname><given-names>Chittaranjan S</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Frayling</surname><given-names>Timothy M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Hattersley</surname><given-names>Andrew T</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Venigalla</surname><given-names>Prathyusha NS</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Chandak</surname><given-names>Giriraj R</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
Lipids in Health and Disease
|
<sec><title>Introduction</title><p>Apolipoprotein AV is an important regulator of triglycerides concentrations. The <italic>ApoAV </italic>gene was identified adjacent to the Apolipoprotein cluster (consisting of <italic>ApoAI</italic>, <italic>ApoAIV </italic>and <italic>ApoCIII </italic>genes) [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The gene encodes a protein that is thought to be fundamental in the transport of triglyceride rich lipoproteins from the liver [<xref ref-type="bibr" rid="B3">3</xref>]. Its hepatic expression has been shown to influence post-prandial triglyceride concentrations [<xref ref-type="bibr" rid="B3">3</xref>] and has the potential to act as an intracellular break in the very low density lipoprotein (VLDL) assembly [<xref ref-type="bibr" rid="B4">4</xref>]. In addition, ApoAV also enhances VLDL metabolism through physical interaction with lipoprotein lipase [<xref ref-type="bibr" rid="B5">5</xref>].</p><p>Polymorphisms in <italic>APOAV </italic>are known to influence the circulating triglyceride concentrations. These variants are split into 3 common haplotypes, of which 2 are associated with increased triglyceride concentrations irrespective of age, gender, ethnicity, pregnancy, lipid profile, and dietary fat intake and are tagged by two SNPs, -1131T>C and S19W [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. In keeping with this, the prevalence of the rare allele for these polymorphisms is higher among subjects with diseases associated with a raised triglyceride concentration such as Familial Hypercholesterolemia, Familial Combined Hyperlipidaemia and Hypertriglyceridaemia [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B15">15</xref>]. The rare allele at -1131T>C polymorphism is also associated with increased risk of coronary heart disease [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B18">18</xref>].</p><p>The rare allele frequency varies between different ethnic groups and populations of Chinese and Japanese descent have a higher prevalence of the -1131C allele and a lower prevalence of the W19 allele when compared to Caucasian populations [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>].</p><p>This paper concentrates on a potential pitfall in the most widely used method (>65% publications) for genotyping the -1131T>C polymorphism using the abolition of a restriction site for <italic>MseI </italic>enzyme [<xref ref-type="bibr" rid="B1">1</xref>]. We detected an unacceptable error rate using our standard quality control checks (a random 10% of samples re-genotyped using a second method). Investigation of the discordant results between the RFLP and the tetra-primer methods for genotyping -1131T>C polymorphism led to identification of a polymorphism within the primer binding site of the RFLP forward primer resulting in allelic drop-out. To alleviate these problems, we propose alternative methods for genotyping -1131T>C polymorphism in the <italic>APOAV </italic>gene.</p></sec><sec><title>Results</title><sec><title>Discordant results in samples typed by two methods</title><p>One thousand, one hundred and eighty five Asian Indian and 173 UK Caucasian subjects were initially genotyped for the -1131T>C polymorphism using the tetra-primer ARMS-PCR method. In keeping with our standard laboratory protocol, 102 of these subjects were re-genotyped using the <italic>MseI </italic>RFLP as a quality control check. In 2.9% (3/102) of samples the results for the two methods were discordant. This high rate of discordant samples could not be explained by sample mix-up or by failed or partial digestion. Sequencing suggested that the erroneous results were due to the RFLP method.</p></sec><sec><title>Polymorphisms present in the primer binding sites</title><p>As the discordant results could most plausibly be explained by allelic drop-out, we sequenced the 900 bp region covering the primer sequences for both the methods in 40 subjects (25 Asian Indians and 15 Caucasians). Sequence analysis identified several polymorphisms (Fig. <xref ref-type="fig" rid="F1">1</xref>) of which three were located in the primer binding sites and may potentially alter primer binding. The -987C>T (rs17120035) polymorphism was situated 4 bp from the 3' end of the forward primer for RFLP analysis (Fig. <xref ref-type="fig" rid="F2">2</xref>). Two other polymorphisms were identified in the tetra-primer inner primer regions; -1108C>G (rs1729411) was 5 bp from the 5' region of the forward primer and a novel SNP -1148G>A that was located 9 bp from the 5' end of the reverse primer (Fig. <xref ref-type="fig" rid="F1">1</xref>).</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Schematic of the <italic>ApoAV </italic>gene, its variants and primer positions.</p></caption><graphic xlink:href="1476-511X-5-11-1"/></fig><fig position="float" id="F2"><label>Figure 2</label><caption><p>This shows the -987C>T polymorphism which potentially alters primer binding for the RFLP method of genotyping (primer sequence is placed within the boxed area, with the site of the polymorphism in capital letter).</p></caption><graphic xlink:href="1476-511X-5-11-2"/></fig><p>Since no polymorphisms were identified in the region of the outer primers used for the tetra-primer assay, we used them for sequencing to assess the prevalence of the identified polymorphisms in 158 unrelated Asian Indian adults and 157 unrelated UK Caucasian adults (table <xref ref-type="table" rid="T1">1</xref>). The -987C>T polymorphism located within the RFLP forward primer was fairly common with a rare allele frequency of 0.23 in Asian Indian subjects and 0.09 within UK Caucasian subjects. While the -1108C>T SNP was not observed in the Asian Indian subjects and only in the UK Caucasian subjects (rare allele frequency = 0.10), the -1148C>T polymorphism situated within the tetra-primer inner reverse primer was observed in the Asian Indian samples only (rare allele frequency = 0.01). The polymorphisms studied were in Hardy Weinberg Equilibrium (HWE) and linkage disequilibrium was observed between -987C>T and -1131T>C both in Asian Indians and UK Caucasians (D' = 1.00, r<sup>2 </sup>= 0.084 and D' = 0.998, r<sup>2 </sup>= 0.001, respectively).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Rare allele frequencies for the polymorphisms in the <italic>ApoAV </italic>gene found through initial sequencing</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Polymorphism</td><td align="center">Refseq number</td><td align="center">Asian Indian</td><td align="center">UK Caucasian</td></tr></thead><tbody><tr><td align="center">-798G>A</td><td align="center">rs10750097</td><td align="center">0.27</td><td align="center">0.27</td></tr><tr><td align="center">-872A>G</td><td align="center">rs7103224</td><td align="center">0.20</td><td align="center">0.07</td></tr><tr><td align="center">-987C>T</td><td align="center">rs17120035</td><td align="center">0.23</td><td align="center">0.09</td></tr><tr><td align="center">-1026G>A</td><td align="center">Not on databases</td><td align="center">0.35</td><td align="center">0.00</td></tr><tr><td align="center">-1108C>G</td><td align="center">rs4938312</td><td align="center">0.00</td><td align="center">0.10</td></tr><tr><td align="center">-1131T>C</td><td align="center">rs662799</td><td align="center">0.22</td><td align="center">0.05</td></tr><tr><td align="center">-1148C>T</td><td align="center">Not on databases</td><td align="center">0.01</td><td align="center">0.00</td></tr><tr><td align="center">-1163C>T</td><td align="center">rs1729411</td><td align="center">0.05</td><td align="center">0.02</td></tr><tr><td align="center">-1242C>T</td><td align="center">Not on databases</td><td align="center">0.004</td><td align="center">0.01</td></tr></tbody></table></table-wrap></sec><sec><title>Heterozygosity in the primer polymorphism -987C>T is associated with errors in genotyping -1131T>C using the RFLP assay</title><p>We performed the <italic>MseI </italic>RFLP assay for -1131T>C on two separate occasions in 82 Asian Indians of known genotype (by sequencing the SNP). This showed at least one discordant result between the sequencing and the RFLP in 13 subjects (14%), with three being discordant on both occasions. Eight of the 13 (62%) discordant samples were heterozygous for both the -1131T>C and the -987C>T polymorphisms compared to 1 of the 69 (1.4%) who did not show discordant results (p < 0.001). As heterozygosity at the -987C>T polymorphism was associated with discordant results, this suggests that this polymorphisms altered primer binding resulting in allelic drop-out on RFLP assay for -1131T>C. To substantiate the role of -987C>T SNP further, we identified 7 subjects heterozygous for both -1131T>C and -987C>T and re-genotyped them using the RFLP assay 7 times. Results for the RFLP differed with sequencing (gold standard) results for 15 out of a total of 49 genotypes (31%) with each subject being discordant on at least one occasion. No individual showed a genotype that was different from that determined by sequencing on all 7 occasions (range 1/7 – 4/7). These results suggest that -987C>T SNP causes allelic drop-out on some but not all occasions.</p></sec></sec><sec><title>Discussion</title><p>Our study shows that presence of a polymorphism-987C>T in the <italic>ApoAV </italic>gene can produce allelic drop-out resulting in an incorrect genotype at SNP -1131T>C when genotyping by the widely used <italic>MseI </italic>RFLP assay. The error occurs in the background of heterozygosity both at -987C>T and -1131T>C and only on some occasions on the same sample. This potential source of error was probably detected in an Asian Indian cohort because both -987C>T and -1131T>C are over 3 times more prevalent than UK Caucasians. There is evidence in the literature that common polymorphisms within primer binding sites [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B23">23</xref>] can result in allelic drop-out which may be due to preferential amplification of one allele or through other mechanisms [<xref ref-type="bibr" rid="B24">24</xref>]. The degree of allelic drop-out may also result in a relative lack of heterozygous subjects within a population and deviation from Hardy Weinberg equilibrium [<xref ref-type="bibr" rid="B23">23</xref>]. Many such polymorphisms are known to be race-specific [<xref ref-type="bibr" rid="B25">25</xref>]. As the majority of data on -1131T>C exists on European Caucasian subjects with very limited data available on other ethnic groups such as Asian Indians and Sub-Saharan Africans, our observations prompt for more caution while initiating similar analysis in new populations.</p><p>The -987C>T polymorphism is likely to contribute to genotyping error especially in Asian Indian samples if the previously described primers are used. Therefore it is wise to use another assay to define the genotype at -1131T>C such as the ARMS-PCR based tetra-primer assay we used [<xref ref-type="bibr" rid="B9">9</xref>]. This is based on a method first described by Ye et al, in 2001 [<xref ref-type="bibr" rid="B26">26</xref>] with primers designed using the Tetra-primer ARMS-PCR website[<xref ref-type="bibr" rid="B27">27</xref>]. Although we did find SNPs that are present in primers used in this method, these are considerably less common, are not close to the 3' end of the primer and no errors were seen compared to sequencing. Therefore the tetra-primer method is our preferred alternative.</p><p>We have developed a new restriction assay for the -1131T>C polymorphism that avoids all known polymorphisms in primer binding sites. This uses the tetra-primer outer forward primer (5'-CAAGGTGACAGACAACTGGTGCAATGAT-3') and the RFLP reverse primer (5'- CCCCAGGAACTGGAGCGAAATT-3') to amplify a fragment of 239 bp. In the presence of the rare C allele, <italic>MseI </italic>digestion will result in fragments of 217 bp and 22 bp. The PCR was performed in 10 μl reactions with standard reagents using 2.5 mM MgCl<sub>2 </sub>and 0.5U of Amplitaq Gold (Applied Biosystems, UK) using cycling conditions as initial denaturation at 94°C for 12 minutes followed by 40 cycles of denaturation at 94°C for 45 seconds, annealing at 63°C for 45 seconds and extension at 72°C for 45 seconds, with a final extension of 72°C for 10 minutes. The <italic>MseI </italic>digestion was performed as previously described by Pennacchio et al, 2001 [<xref ref-type="bibr" rid="B1">1</xref>]. This new PCR RFLP assay worked in 96 samples with no discrepancies and no errors when compared to the original genotyping results for the tetra-primer assay.</p></sec><sec><title>Conclusion</title><p>In conclusion, the -987C>T polymorphism is present in UK Caucasians and Asian Indians and can result in allelic drop-out within the widely used <italic>MseI </italic>RFLP assay for genotyping the triglyceride associated polymorphism -1131T>C. This is especially important in the Asian Indian population as both -987C>T and -1131T>C polymorphisms are more prevalent. Our result stresses the importance of checking and repeatedly checking SNP databases for polymorphisms in the primer binding sites (the -987C>T polymorphism was not reported when we started out studies on the -1131T>C polymorphism). It also emphasises the need to use a second genotyping method as a quality control. This type of check is particularly important when studying new populations as allele frequencies of the same polymorphism can vary greatly between populations meaning a primer binding site polymorphism may be of much greater significance in some populations than others.</p></sec><sec sec-type="materials|methods"><title>Materials and methods</title><p>Genomic DNA samples from 1185 Asian Indian subjects from 395 Asian Indian family trios (mother, father and child) from the Pune Children Study [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B29">29</xref>] and 173 unrelated UK Caucasians from the EFS and Plymouth EarlyBird study [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B30">30</xref>] were initially genotyped for -1131T>C variant using an ARMS based tetra-primer method. The design of primers would amplify a control product of 404 bp while the PCR products of 250 bp and 242 bp would identify T and C alleles respectively.</p><p>The primer sequences used are as follows:</p><p>Tetra-primer Outer Forward: 5'-CAAGGTGACAGACAACTGGTGCAATGAT-3'</p><p>Tetra-primer Outer Reverse: 5'-AGCCCCTGAAAGCTTCACTACAGGTTCC-3'</p><p>Tetra-primer Inner Forward: 5'-TTCAGCTTTTCCTCATGGGGCAAATATC-3'</p><p>Tetra-primer Inner Reverse: 5'-GAGCCCCAGGAACTGGAGCGAAATTA-3'</p><p>The PCR was performed using cycling conditions of initial denaturation at 95°C for 2 min, followed by 35 cycles of denaturation at 95°C for 1 min, annealing at 59°C for 1 min and extension at 72°C for 1 min, with a final 2 min extension at 72°C. PCR products were run on a 3% agarose gel at 200 V for 90 minutes and scored by two independent workers (KW and GRC). Subsequently, ~10% of these samples (n = 102) were analysed by PCR-RFLP using 2U of <italic>MseI </italic>enzyme [<xref ref-type="bibr" rid="B1">1</xref>] for the quality control check.</p><p>In an attempt to investigate the discordant results between the RFLP based and tetra-primer based genotyping, primers were designed to amplify a 900 bp region encompassing both tetra-primers and RFLP primers (Fig. <xref ref-type="fig" rid="F1">1</xref>). Finally, tetra-primer outer primer pair was utilised as the gold standard for confirmation of individual sequence since this region did not show any polymorphism on sequence analysis of the 900 bp region. All the PCR products were purified using post-PCR purification plates from Millipore and sequenced individually on both the strands using Big Dye terminator cycle sequencing ready kit (Version 1.1, Applied Biosystems, Warrington, UK) on ABI 3100 Genetic Analyzer (Applied Biosystems).</p></sec><sec><title>Abbreviations</title><p>ApoAV – Apolipoprotein AV</p><p>ApoAI – Apolipoprotein AI</p><p>ApoAIV – Apolipoprotein AIV</p><p>ApoCIII – Apolipoprotein CIII</p><p>VLDL – Very Low Density Lipoprotein</p><p>SNPs – Single Nucleotide Polymorphisms</p><p>RFLP – Restriction Fragment Length Polymorphism</p><p>PCS – Pune Children Study</p><p>EFS – Exeter Family Study of Childhood Health</p><p>PEB – Plymouth EarlyBird study</p><p>HWE – Hardy Weinberg Equilibrium</p><p>LD – Linkage Disequilibrium</p><p>PCR – Polymerase Chain Reaction</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they do not have any competing interests.</p></sec><sec><title>Authors' contributions</title><p>KJW carried out the molecular genetic studies, performed statistical analysis and drafted the manuscript. SE advised on methodologies. CSY designed the studies of the phenotype of patients and supervised the isolation of DNA. TMF and ATH helped advise on design of molecular studies. GRC designed and carried out molecular genetic studies. PNSV performed the molecular genetic studies especially the sequencing. All authors read and approved the final manuscript.</p></sec>
|
Assessment of the potential impact of a reminder system on the reduction of diagnostic errors: a quasi-experimental study
|
<sec><title>Background</title><p>Computerized decision support systems (DSS) have mainly focused on improving clinicians' diagnostic accuracy in unusual and challenging cases. However, since diagnostic omission errors may predominantly result from incomplete workup in routine clinical practice, the provision of appropriate patient- and context-specific reminders may result in greater impact on patient safety. In this experimental study, a mix of easy and difficult simulated cases were used to assess the impact of a novel diagnostic reminder system (ISABEL) on the quality of clinical decisions made by various grades of clinicians during acute assessment.</p></sec><sec sec-type="methods"><title>Methods</title><p>Subjects of different grades (consultants, registrars, senior house officers and medical students), assessed a balanced set of 24 simulated cases on a trial website. Subjects recorded their clinical decisions for the cases (differential diagnosis, test-ordering and treatment), before and after system consultation. A panel of two pediatric consultants independently provided gold standard responses for each case, against which subjects' quality of decisions was measured. The primary outcome measure was change in the count of diagnostic errors of omission (DEO). A more sensitive assessment of the system's impact was achieved using specific quality scores; additional consultation time resulting from DSS use was also calculated.</p></sec><sec><title>Results</title><p>76 subjects (18 consultants, 24 registrars, 19 senior house officers and 15 students) completed a total of 751 case episodes. The mean count of DEO fell from 5.5 to 5.0 across all subjects (repeated measures ANOVA, p < 0.001); no significant interaction was seen with subject grade. Mean diagnostic quality score increased after system consultation (0.044; 95% confidence interval 0.032, 0.054). ISABEL reminded subjects to consider at least one clinically important diagnosis in 1 in 8 case episodes, and prompted them to order an important test in 1 in 10 case episodes. Median extra time taken for DSS consultation was 1 min (IQR: 30 sec to 2 min).</p></sec><sec><title>Conclusion</title><p>The provision of patient- and context-specific reminders has the potential to reduce diagnostic omissions across all subject grades for a range of cases. This study suggests a promising role for the use of future reminder-based DSS in the reduction of diagnostic error.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Ramnarayan</surname><given-names>Padmanabhan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Roberts</surname><given-names>Graham C</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Coren</surname><given-names>Michael</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Nanduri</surname><given-names>Vasantha</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Tomlinson</surname><given-names>Amanda</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Taylor</surname><given-names>Paul M</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Wyatt</surname><given-names>Jeremy C</given-names></name><xref ref-type="aff" rid="I7">7</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Britto</surname><given-names>Joseph F</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib>
|
BMC Medical Informatics and Decision Making
|
<sec><title>Background</title><p>A recent Institute of Medicine report has brought the problem of medical error under intense scrutiny[<xref ref-type="bibr" rid="B1">1</xref>]. While the use of computerized prescription software has been shown to substantially reduce the incidence of medication-related error [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>], few solutions have demonstrated a similar impact on diagnostic error. Diagnostic errors impose a significant burden on modern healthcare: they account for a large proportion of medical adverse events in general [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B6">6</xref>], and form the second leading cause for malpractice suits against hospitals [<xref ref-type="bibr" rid="B7">7</xref>]. In particular, diagnostic errors of omission (DEO) during acute medical assessment, resulting from cognitive biases such as 'premature closure' and 'confirmation bias', lead to incomplete diagnostic workup and 'missed diagnoses' [<xref ref-type="bibr" rid="B8">8</xref>]. This is especially relevant in settings such as family practice [<xref ref-type="bibr" rid="B9">9</xref>], as well as hospital areas such as the emergency room and critical care [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>], 20% of patients discharged from emergency rooms raised concerns in a recent survey that their clinical assessment had been complicated by diagnostic error [<xref ref-type="bibr" rid="B12">12</xref>]. The use of clinical decision-support systems (DSS) has been one of many strategies proposed for the reduction of diagnostic errors in practice [<xref ref-type="bibr" rid="B13">13</xref>]. Consequently, a number of DSS have been developed over the past few years to assist clinicians during the process of medical diagnosis [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Even though studies of several diagnostic DSS have demonstrated improved physician performance in simulated (and rarely real) patient encounters [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>], two specific characteristics may have contributed to their infrequent use in routine practice: intended purpose and design. Many general diagnostic DSS were built as 'expert systems' to solve diagnostic conundrums and provide the correct diagnosis during a 'clinical dead-end' [<xref ref-type="bibr" rid="B19">19</xref>]. Since true diagnostic dilemmas are rare in practice [<xref ref-type="bibr" rid="B20">20</xref>], and the initiative for DSS use had to originate from the physician, diagnostic advice was not sought routinely, particularly since clinicians prefer to store the patterns needed to solve medical problems in their heads [<xref ref-type="bibr" rid="B21">21</xref>]. There is, however, evidence that clinicians frequently underestimate their need for diagnostic assistance, and that the perception of diagnostic difficulty does not correlate with their clinical performance [<xref ref-type="bibr" rid="B22">22</xref>]. In addition, due to the demands of the information era [<xref ref-type="bibr" rid="B23">23</xref>]. diagnostic errors may not be restricted to cases perceived as being difficult, and might occur even when dealing with common problems in a stressful environment under time pressure [<xref ref-type="bibr" rid="B24">24</xref>]. Further, most 'expert systems' utilized a design in which clinical data entry was achieved through a controlled vocabulary specific to each DSS. This process frequently took > 15 minutes, contributing to infrequent use in a busy clinical environment [<xref ref-type="bibr" rid="B25">25</xref>]. These 'expert systems' also provided between 20 and 30 diagnostic possibilities [<xref ref-type="bibr" rid="B26">26</xref>], with detailed explanations, leading to a lengthy DSS consultation process.</p><p>In order to significantly affect the occurrence of diagnostic error, it seems reasonable to conclude that DSS advice must therefore be readily available, and sought, during most clinical encounters, even if the perceived need for diagnostic assistance is minor. Ideally, real-time advice for diagnosis can be actively provided by integrating a diagnostic DSS into an existing electronic medical record (EMR), as has been attempted in the past [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. However, the limited uptake of EMRs capable of recording sufficient narrative clinical detail currently in clinical practice indicates that a stand-alone system may prove much more practical in the medium term [<xref ref-type="bibr" rid="B29">29</xref>]. The key characteristic of a successful system would be the ability to deliver reliable diagnostic reminders rapidly following a brief data entry process in most clinical situations. ISABEL (ISABEL Healthcare, UK) is a novel Web-based pediatric diagnostic reminder system that suggests important diagnoses during clinical assessment [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. The development of the system and its underlying structure have been described in detail previously [<xref ref-type="bibr" rid="B32">32</xref>,<xref ref-type="bibr" rid="B33">33</xref>]. The main hypotheses underlying the development of ISABEL were that the provision of diagnostic reminders generated following a brief data entry session in free text would promote user uptake, and lead to improvement in the quality of diagnostic decision making in acute medical settings. The reminders provided (a set of 10 in the first instance) aimed to remind clinicians of important diagnoses that they might have missed in the workup. Data entry is by means of natural language descriptions of the patient's clinical features, including any combination of symptoms, signs and test results. The system's knowledge base consists of natural language text descriptions of > 5000 diseases, in contrast to most 'expert systems' that use complex disease databases [<xref ref-type="bibr" rid="B34">34</xref>-<xref ref-type="bibr" rid="B36">36</xref>]. The advantages and trade-offs of these differences in system design have been discussed in detail elsewhere [<xref ref-type="bibr" rid="B37">37</xref>]. In summary, although the ability to rapidly enter patient features in natural language to derive a short-list of diagnostic suggestions may allow frequent use by clinicians during most patient encounters, variability resulting from the use of natural language for data entry, and the absence of probability ranking, may compromise the accuracy and usefulness of the diagnostic suggestions.</p><p>The overall evaluation of the ISABEL system was planned in systematic fashion in a series of consecutive studies [<xref ref-type="bibr" rid="B38">38</xref>].</p><p>a) An initial clinical performance evaluation: This would evaluate the feasibility of providing relevant diagnostic suggestions for a range of cases when data is entered in natural language. System accuracy, speed and relevance of suggestions were studied.</p><p>b) An assessment of the impact of the system in a quasi-experimental setting: This would examine the effects of diagnostic decision support on subjects using simulated cases.</p><p>c) An assessment of the impact of the system in a real life setting: This would examine the effects of diagnostic advice on clinicians in real patients in their natural environment.</p><p>In the initial performance evaluation, the ISABEL system formed the unit of intervention, and the quality of its diagnostic suggestions was validated against data drawn from 99 hypothetical cases and 100 real patients. Key findings from cases were entered into the system in free text by one of the developers. The system included the final diagnosis in 95% of the cases [<xref ref-type="bibr" rid="B39">39</xref>]. This design was similar to early evaluations of a number of other individual diagnostic DSS [<xref ref-type="bibr" rid="B40">40</xref>-<xref ref-type="bibr" rid="B42">42</xref>], as well as a large study assessing the performance characteristics of four expert diagnostic systems [<xref ref-type="bibr" rid="B43">43</xref>]. Since this step studied ISABEL in isolation, and did not include users uninvolved in the development of the system, it was vital to examine DSS <italic>impact </italic>on decision making by demonstrating in the subsequent step that the clinician-DSS combination functioned better than either the clinician or the system working in isolation [<xref ref-type="bibr" rid="B44">44</xref>,<xref ref-type="bibr" rid="B45">45</xref>]. Evaluation of impact is especially relevant to ISABEL: despite good system performance when tested in isolation, clinicians may not benefit from its advice either due to variability associated with user data entry leading to poor results, or the inability to distinguish between diagnostic suggestions due to the lack of ranking [<xref ref-type="bibr" rid="B46">46</xref>]. A previous evaluation of Quick Medical Reference (QMR) assessed a group of clinicians working a set of difficult cases, and suggested that the extent of benefit gained by different users varied with their level of experience [<xref ref-type="bibr" rid="B47">47</xref>].</p><p>In this study, we aimed to perform an impact evaluation of ISABEL in a quasi-experimental setting in order to quantify the effects of diagnostic advice on the quality of clinical decisions made by various grades of clinicians during acute assessment, using a mix of easy and difficult simulated cases drawn from all pediatric sub-specialties. Study design was based on an earlier evaluation of the impact of ILIAD and QMR on diagnostic reasoning in a simulated environment [<xref ref-type="bibr" rid="B48">48</xref>]. Our key outcome measure focused on appropriateness of decisions during diagnostic workup rather than accuracy in identifying the correct diagnosis. The validity of textual case simulations has previously been demonstrated in medical education exercises [<xref ref-type="bibr" rid="B49">49</xref>], and during the assessment of mock clinical decision making [<xref ref-type="bibr" rid="B50">50</xref>,<xref ref-type="bibr" rid="B51">51</xref>].</p></sec><sec sec-type="methods"><title>Methods</title><p>The simulated field study involved recording subjects' clinical decisions regarding diagnoses, test-ordering and treatment for a set of simulated cases, both before and immediately after DSS consultation. The impact of diagnostic reminders was determined by measuring changes in the quality of decisions made by subjects. In this study, the quality of ISABEL's diagnostic suggestion list <italic>per se </italic>was not examined. The study was coordinated at Imperial College School of Medicine, St Mary's Hospital, London, UK between February and August 2002. The study was approved by the Local Research Ethics Committee.</p><sec><title>Subjects</title><p>A convenience sample consisting of pediatricians of different grades (senior house officers [interns], registrars [residents] and consultants [attending physicians] from different geographical locations across the UK), and final year medical students, was enrolled for the study. All students were drawn from one medical school (Imperial College School of Medicine, London, UK). Clinicians were recruited by invitation from the ISABEL registered user database which consisted of a mixture of regular users as well as pediatricians who had never used the system after registration. After a short explanation of the study procedure, all subjects who consented for the study were included within the sample.</p></sec><sec><title>Cases</title><p>Cases were drawn from a pool of 72 textual case simulations, constructed by one investigator, based on case histories of real children presenting to emergency departments (data collected during earlier evaluation). Each case was limited to between 150 and 200 words, and only described the initial presenting symptoms, clinical signs and basic laboratory test results in separate sections. Since the clinical data were collected from pediatric emergency rooms, the amount of clinical information available at assessment was limited but typical for this setting. Ample negative features were included in order to prevent the reader from picking up positive cues from the text. These cases were then classified into one of 12 different pediatric sub-specialties (e.g. cardiology, respiratory) and to one of 3 case difficulty levels within each specialty (1-unusual, 2-not unusual, and 3-common clinical presentation, with reference to UK general hospital pediatric practice) by the author. This allocation process was duplicated by a pediatric consultant working independently. Both investigators assigned 57 cases to the same sub-specialty and 42 cases to both the same sub-specialty and the same level of difficulty (raw agreement 0.79 and 0.58 respectively). From the 42 cases in which both investigators agreed regarding the allocation of both specialty and level of difficulty, 24 cases were drawn such that a pair of cases per sub-specialty representing two different levels of difficulty (level 1 & 2, 1 & 3 or 2 & 3) was chosen for the final case mix. This process ensured a balanced set of cases representing all sub-specialties and comprising easy as well as difficult cases.</p></sec><sec><title>Data collection website</title><p>A customized, password protected version of ISABEL was used to collect data during the study. This differed from the main website in that it automatically displayed the study cases to each subject in sequence, assigned each case episode a unique study number, and recorded time data in addition to displaying ten diagnostic suggestions. Three separate text boxes were provided to record subjects' clinical decisions (diagnoses, tests and treatment) pre- and post-DSS consultation. The use of the customized trial website ensured that subjects proceeded from one step to the next without being able to skip steps or revise clinical decisions already submitted.</p></sec><sec><title>Training</title><p>Training was intended only to familiarize subjects with the trial website. During training, all subjects were assigned unique log-in and passwords, and one sample case as practice material. Practice sessions involving medical students were supervised by one investigator in group sessions of 2–3 subjects each. Pediatricians (being from geographically disparate locations) were not supervised during training, but received detailed instructions regarding the use of the trial website by email. Context-specific help was provided at each step on the website for assistance during the practice session. All subjects completed their assigned practice case, and were recruited for the study.</p></sec><sec><title>Study procedure</title><p>Subjects were allowed to complete their assessments of the simulated cases from any computer connected to the Internet at any time (i.e. they were not supervised). After logging into the trial website, subjects were presented with text from a case simulation. They assessed the case, abstracted the salient clinical features according to their own interpretation of the case, and entered them into the designated search query box in free text. Following this, they keyed in their decisions regarding diagnostic workup, test-ordering and treatment into the designated textboxes. These constituted pre-DSS clinical decisions. See figure <xref ref-type="fig" rid="F1">1</xref> for an illustration of this step of the study procedure. On submitting this information, a list of diagnostic suggestions was instantly presented to the subject based on the abstracted clinical features. The subjects could not read the case text again at this stage, preventing them from processing the case a second time, thus avoiding 'second-look' bias. Diagnostic suggestions were different for different users since the search query was unique for each subject, depending on their understanding of the case and how they expressed it in natural language. On the basis of the diagnostic suggestions, subjects could modify their pre-DSS clinical decisions by adding or deleting items: these constituted post-DSS clinical decisions. All clinical decisions, and the time taken to complete each step, were recorded automatically. See figure <xref ref-type="fig" rid="F2">2</xref> for an illustration of this step of the study procedure. The text from one case, and the variability associated with its interpretation during the study, is depicted in figure <xref ref-type="fig" rid="F3">3</xref>.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Screenshot of ISABEL simulated study procedure – step 1</bold>. This figure shows how one subject was presented with the text of a case simulation, how he could search ISABEL by using summary clinical features, and record his clinical decisions prior to viewing ISABEL's results. For this case, clinically important diagnoses provided by the expert panel are: nasopharyngitis (OR viral upper respiratory tract infection) and meningitis/encephalitis. This subject has committed a DEO (failed to include both clinically important diagnoses in his diagnostic workup).</p></caption><graphic xlink:href="1472-6947-6-22-1"/></fig><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Screenshot of ISABEL simulated study procedure – step 2</bold>. This figure shows how the same subject was provided the results of ISABEL's search in one click, and how he was provided the opportunity to modify the clinical decisions made in step 1. It was not possible to go back from this step to step 1 to modify the clinical decisions made earlier. Notice that the subject has not identified meningitis/encephalitis from the ISABEL suggestions as clinically important. He has made no changes to his workup, and has committed a DEO despite system advice.</p></caption><graphic xlink:href="1472-6947-6-22-2"/></fig><fig position="float" id="F3"><label>Figure 3</label><caption><p>Example of one simulated case used in study*, the variability in clinical features as abstracted by five different users (verbatim), and clinically important diagnoses as judged by panel.</p></caption><graphic xlink:href="1472-6947-6-22-3"/></fig><p>Each subject was presented with 12 cases such that one of the pair drawn from each sub-specialty was displayed. Cases were presented in random order (in no particular order of sub-specialty). Subjects could terminate their session at any time and return to complete the remainder of cases. If a session was terminated midway through a case, that case was presented again on the subject's return. If the website detected no activity for > 2 hours, the subject was automatically logged off, and the session was continued on their return. All subjects had 3 weeks to complete their assigned 12 cases. Since each case was used more than once, by different subjects, we termed each attempt by a subject at a case as a 'case episode'.</p></sec><sec><title>Scoring metrics</title><p>We aimed to assess if the provision of key diagnostic reminders would reduce errors of omission in the simulated environment. For the purposes of this study, a subject was defined to have committed a DEO for a case episode if they failed to include <italic>all </italic>'clinically important diagnoses' in the diagnostic workup (rather than failing to include the 'correct diagnosis'). A diagnosis was judged 'clinically important' if an expert panel working the case independently decided that the particular diagnosis had to be included in the workup in order to ensure safe and appropriate clinical decision making, i.e. they would significantly affect patient management and/or course, and failure to do so would be construed clinically inadequate. The expert panel comprised two general pediatricians with > 3 years consultant level experience. 'Clinically important' diagnoses suggested by the panel thus included the 'most likely diagnosis/es' and other key diagnoses; they did not constitute a full differential containing all plausible diagnoses. This outcome variable was quantified by a binary measure (for each case episode, a subject either committed a DEO or not). Errors of omission were defined for tests and treatments in similar fashion.</p><p>We also sought a more sensitive assessment of changes in the quality of clinical decisions made by subjects in this study. Since an appropriate and validated instrument was essential for this purpose, a measurement study was first undertaken to develop and validate such an instrument. The measurement study, including the development and validation of a diagnostic quality score and a management plan quality score, has previously been reported in detail [<xref ref-type="bibr" rid="B52">52</xref>]. The scoring process, tested using a subset of cases worked on by clinicians during this study (190 case episodes), was reliable (intraclass correlation coefficient 0.79) and valid (face, construct and concurrent validity). During the scoring process, the expert panel was provided an aggregate list of decisions drawn from all subjects (pre- and post-DDSS consultation) for each case. They provided measurements of quality for each of the clinical decisions in addition to identifying 'clinically important' decisions for each case. Prior to scoring, one investigator (PR) mapped diagnoses proposed by subjects and the expert panel to the nearest equivalent diagnoses in the ISABEL database. Quality of each diagnostic decision was scored for the degree of plausibility, likelihood in the clinical setting, and its impact on further patient management. These measurements were used to derive scores for each set of subjects' clinical decisions (diagnostic workup, tests and treatments). As per the scoring system, subjects' decision plans were awarded the highest score (score range: 0 to 1) only if they were both comprehensive (contained <italic>all </italic>important clinical decisions), and focused (contained <italic>only </italic>important decisions). Scores were calculated for each subject's diagnostic, test-ordering and treatment plans both pre- and post-ISABEL consultation. Figure <xref ref-type="fig" rid="F4">4</xref> provides a schematic diagram of the complete scoring procedure.</p><fig position="float" id="F4"><label>Figure 4</label><caption><p>Schematic of scoring procedure.</p></caption><graphic xlink:href="1472-6947-6-22-4"/></fig></sec><sec><title>Primary outcome</title><p>1. Change in the number of diagnostic errors of omission among subjects.</p></sec><sec><title>Secondary outcomes</title><p>1. Mean change in subjects' diagnostic, test-ordering and treatment plan quality scores.</p><p>2. Change in the number of irrelevant diagnoses contained within the diagnostic workup.</p><p>3. Proportion of case episodes in which at least one additional 'important' diagnosis, test or treatment decision was considered by the subject <italic>after </italic>DSS consultation.</p><p>4. Additional time taken for DSS consultation.</p></sec><sec><title>Analysis</title><p>Subjects were used as the unit of analysis for the primary outcome measure. For each subject, the total number of DEOs was counted separately for pre- and post-DSS diagnostic workup plans; only subjects who had completed all assigned cases were included in this calculation. Statistically significant changes in DEO count following DDSS consultation and interaction with grade was assessed by two-way mixed-model analysis of variance (grade being between-subjects factor and time being within-subjects factor). Mean number of DEOs was calculated for each subject grade, and DEOs were additionally analyzed according to level of case difficulty. Statistical significance was set at a p value of 0.05.</p><p>Subjects were used as the unit of analysis for the change in mean quality scores (the development of quality scores and their validation has been previously described; however, the scores have never been used as an outcome measure prior to this evaluation). In the first step, subjects' quality score (pre- and post-DSS) was calculated for each case episode. For each subject, a mean quality score across all 12 cases was computed. Only case episodes from subjects who completed all 12 assigned cases were used during this calculation. A two-way mixed model ANOVA (grade as between-subjects factor; time as within-subjects factor) was used to examine statistically significant differences in quality scores. This analysis was performed for diagnostic quality scores as well as test ordering and treatment plan scores. Data from a pilot study suggested that data from 64 subjects were needed to demonstrate a mean diagnostic quality score change of 0.03 (standard deviation 0.06, power 80%, level of significance 5%).</p><p>Using subjects as the unit of analysis, the mean count of diagnoses (and irrelevant diagnoses) included in the workup was calculated pre- and post-DSS consultation for each subject as an average across all case attempts. Only subjects who attempted all assigned cases were included in this analysis. Using this data, a mean count for diagnoses (and irrelevant diagnoses) was calculated for each subject grade. A two-way mixed model ANOVA was used to assess statistically significant differences in this outcome with respect to grade as well as occasion. Using case episodes as the unit of analysis, the proportion of case episodes in which at least one additional 'important' diagnosis, test or treatment was prompted by ISABEL was determined. The proportion of case episodes in which at least one clinically significant decision was deleted, and at least one inappropriate decision was added, after system consultation, was also computed. All data were analyzed separately for the subjects' grades.</p><p>Two further analyses were conducted to enable the interpretation of our results. First, in order to provide a direct comparison of our results with other studies, we used case episodes as the unit of analysis and examined the presence of the 'most likely diagnosis' in the diagnostic workup. The 'most likely diagnosis' was part of the set of 'clinically important' diagnoses provided by the panel, and represented the closest match to a 'correct' diagnosis in our study design. This analysis was conducted separately for each grade. Second, since it was important to verify whether any reduction of omission errors was directly prompted by ISABEL, or simply by subjects re-thinking about the assigned cases, all case episodes in which at least one additional significant diagnosis was added by the user were examined. If the diagnostic suggestion added by the user had been displayed in the DSS list of suggestions, it strongly suggested that the system, rather than subjects' re-thinking, prompted these additions.</p></sec></sec><sec><title>Results</title><p>The characteristics of subjects, cases and case episodes are summarized in table <xref ref-type="table" rid="T1">1</xref>. Ninety seven subjects were invited for the study. Although all subjects consented and completed their training, only seventy six subjects attempted at least one case (attempters) during their allocated three weeks. This group consisted of 15 medical students, 19 SHOs, 24 registrars and 18 consultants. There was no significant difference between attempters and non-attempters with respect to grade (Chi square test, p 0.07). Only 6/76 subjects had used ISABEL regularly (at least once a week) prior to the study period (3 SHOs, 1 Registrar and 2 Consultants); all the others had registered for the service, but never used the DSS previously. A total of 751 case episodes were completed by the end of the study period. Fifty two subjects completed all assigned 12 cases to produce 624 case episodes (completers); 24 other subjects did not complete all their assigned cases (non-completers). Completers and non-completers did not differ significantly with respect to grade (Chi square test, p 0.06). However, more subjects were trained remotely in the non-completers group (Chi square test, p 0.003). The majority of non-completers had worked at least two cases (75%); slightly less than half (42%) had worked at least 6 cases. Forty-seven diagnoses were considered 'clinically important' by the panel across all 24 cases (average ~ 2 per case). For 21/24 cases, the panel had specified a single 'most likely diagnosis'; for 3 cases, two diagnoses were included in this definition.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Study participants, cases and case episodes*</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center" colspan="8"><bold>Grade of subject</bold></td></tr></thead><tbody><tr><td></td><td align="center">Consultant (%)</td><td align="center">Registrar (%)</td><td align="center">SHO (%)</td><td align="center">Student (%)</td><td align="center"><bold>Total</bold></td><td align="center"><bold>Case episodes</bold></td><td align="center"><bold>Cases</bold></td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left">Subjects invited to participate</td><td align="center">27 (27.8)</td><td align="center">33 (34)</td><td align="center">20 (20.6)</td><td align="center">17 (17.5)</td><td align="center"><bold>97</bold></td><td></td><td></td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left">Subjects who attempted at least one case (attempters)</td><td align="center">18 (23.7)</td><td align="center">24 (31.6)</td><td align="center">19 (25)</td><td align="center">15 (19.7)</td><td align="center"><bold>76</bold></td><td align="center"><bold>751</bold></td><td align="center"><bold>24</bold></td></tr><tr><td align="left">Subjects who attempted at least six cases</td><td align="center">16 (25.8)</td><td align="center">18 (29)</td><td align="center">15 (24.2)</td><td align="center">13 (20.9)</td><td align="center"><bold>62</bold></td><td align="center"><bold>715</bold></td><td align="center"><bold>24</bold></td></tr><tr><td align="left">Subjects who completed all 12 cases (completers)</td><td align="center">15 (28.8)</td><td align="center">14 (26.9)</td><td align="center">10 (19.2)</td><td align="center">13 (25)</td><td align="center"><bold>52</bold></td><td align="center"><bold>624</bold></td><td align="center"><bold>24</bold></td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left">Regular DSS users (usage > once/week)</td><td align="center">2</td><td align="center">1</td><td align="center">3</td><td align="center">0</td><td align="center"><bold>6</bold></td><td></td><td></td></tr></tbody></table><table-wrap-foot><p>* Since each case was assessed more than once, each attempt by a subject at a case was termed as a 'case episode'.</p></table-wrap-foot></table-wrap><sec><title>Diagnostic errors of omission</title><p>624 case episodes generated by 52 subjects were used to examine DEO. During the pre-DSS consultation phase, all subjects performed a DEO in at least one of their cases, and 21.1% (11/52) in more than half their cases. In the pre-ISABEL consultation stage, medical students and SHOs committed the most and least number of DEO respectively (6.6 vs. 4.4); this gradient was maintained post-ISABEL consultation (5.9 vs. 4.1). Overall, 5.5 DEO were noted per subject pre-DSS consultation; this reduced to 5.0 DEO after DSS advice (p < 0.001). No significant interaction was noticed with grade (F<sub>3, 48 </sub>= 0.71, p = 0.55). Reduction in DEO following DSS advice within each grade is shown in table <xref ref-type="table" rid="T2">2</xref>. Overall, more DEOs were noted for easy cases compared to difficult cases pre- and post-DSS advice (2.17 vs. 2.05 and 2.0 vs. 1.8); however, this was not true for medical students as a subgroup (2.5 vs. 2.9). Improvement following DSS advice seemed greater for difficult cases for all subjects, although this was not statistically significant. These findings are summarized in table <xref ref-type="table" rid="T3">3</xref>.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Mean count of diagnostic errors of omission (DEO) pre-ISABEL and post-ISABEL consultation</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Grade of subject</td><td align="center">DEO pre-ISABEL (SD)</td><td align="center">DEO post-ISABEL (SD)</td><td align="center">Reduction (SD)</td></tr></thead><tbody><tr><td align="left">Consultant</td><td align="center">5.13 (1.3)</td><td align="center">4.6 (1.4)</td><td align="center">0.53 (0.7)</td></tr><tr><td align="left">Registrar</td><td align="center">5.64 (1.5)</td><td align="center">5.14 (1.6)</td><td align="center">0.5 (0.5)</td></tr><tr><td align="left">SHO</td><td align="center">4.4 (1.6)</td><td align="center">4.1 (1.6)</td><td align="center">0.3 (0.5)</td></tr><tr><td align="left">Medical student</td><td align="center">6.61 (1.3)</td><td align="center">5.92 (1.4)</td><td align="center">0.69 (0.7)</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left"><bold>Mean DEO across all subjects (n = 52)*</bold></td><td align="center"><bold>5.50 (1.6)</bold></td><td align="center"><bold>4.98 (1.5)</bold></td><td align="center"><bold>0.52 (0.6)</bold></td></tr></tbody></table><table-wrap-foot><p>*Total number of subjects who completed all 12 assigned cases</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Mean DEO count analyzed by level of case and subject grade</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Grade</bold></td><td align="center" colspan="2"><bold>Difficult cases</bold></td><td align="center" colspan="2"><bold>Easy cases</bold></td></tr><tr><td></td><td colspan="4"><hr></hr></td></tr><tr><td></td><td align="center">Pre-DSS</td><td align="center">Post-DSS</td><td align="center">Pre-DSS</td><td align="center">Post-DSS</td></tr></thead><tbody><tr><td align="left">Consultant</td><td align="center">1.66</td><td align="center">1.47</td><td align="center">2.0</td><td align="center">1.87</td></tr><tr><td align="left">Registrar</td><td align="center">2.21</td><td align="center">1.93</td><td align="center">2.14</td><td align="center">1.92</td></tr><tr><td align="left">SHO</td><td align="center">1.3</td><td align="center">1.2</td><td align="center">2.0</td><td align="center">1.8</td></tr><tr><td align="left">Medical student</td><td align="center">2.92</td><td align="center">2.54</td><td align="center">2.54</td><td align="center">2.30</td></tr></tbody></table></table-wrap></sec><sec><title>Mean quality score changes</title><p>624 case episodes from 52 subjects who had completed all assigned 12 cases were used for this analysis. Table <xref ref-type="table" rid="T4">4</xref> summarizes mean diagnostic quality scores pre- and post-ISABEL consultation, and the change in mean quality score for diagnoses, for each grade of subject. There was a significant change in the weighted mean of the diagnostic quality score (0.044; 95% confidence interval: 0.032, 0.054; <italic>p </italic>< 0.001). No significant interaction between grade and occasion was demonstrated. In 9/52 subjects (17.3%), the pre-DSS score for diagnostic quality was higher than the post-DSS score, indicating that subjects had lengthened their diagnostic workup without substantially improving its quality. Overall, the mean score for test-ordering plans increased significantly from 0.345 to 0.364 (an increase of 0.019, 95% CI 0.011–0.027, t<sub>51 </sub>= 4.91, p < 0.001); this increase was smaller for treatment plans (0.01, 95% CI 0.007–0.012, t<sub>51 </sub>= 7.15, p < 0.001).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Mean quality scores for diagnoses broken down by grade of subject</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center">Mean pre-ISABEL score</td><td align="center">Mean post-ISABEL score</td><td align="center">Mean score change*</td></tr></thead><tbody><tr><td align="left">Consultant</td><td align="center">0.39</td><td align="center">0.43</td><td align="center">0.044</td></tr><tr><td align="left">Registrar</td><td align="center">0.40</td><td align="center">0.44</td><td align="center">0.038</td></tr><tr><td align="left">SHO</td><td align="center">0.45</td><td align="center">0.48</td><td align="center">0.032</td></tr><tr><td align="left">Medical student</td><td align="center">0.31</td><td align="center">0.37</td><td align="center">0.059</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left"><bold>Weighted average (all subjects)†</bold></td><td align="center"><bold>0.383</bold></td><td align="center"><bold>0.426</bold></td><td align="center"><bold>0.044</bold></td></tr></tbody></table><table-wrap-foot><p>* There was no significant difference between grades in terms of change in diagnosis quality score (one-way ANOVA p > 0.05)</p></table-wrap-foot></table-wrap></sec><sec><title>Number of irrelevant diagnoses</title><p>624 case episodes from 52 subjects were used for this analysis. The results are illustrated in table <xref ref-type="table" rid="T5">5</xref>. Overall, the mean count of diagnoses included by subjects in their workup pre-DSS advice was 3.9. This increased to 5.7 post-DSS consultation (an increase of 1.8 diagnoses). The increase was largest for medical students (a mean increase of 2.6 diagnoses) and least for consultants (1.4 diagnoses). The ANOVA showed significant interaction between grade and occasion (F<sub>3,58 </sub>= 3.14, p = 0.034). The number of irrelevant diagnoses in the workup changed from 0.7 pre-DSS to 1.4 post-DSS advice (an increase of 0.7 irrelevant diagnoses, 95% CI 0.5–0.75). There was a significant difference in this increase across grades (most for medical students and least for consultants; 1.1 vs. 0.3 irrelevant diagnoses, F<sub>3, 48 </sub>= 6.33, p < 0.01). The increase in irrelevant diagnoses did not result in a corresponding increase in the number of irrelevant or deleterious tests and treatments (an increase of 0.09 tests and 0.03 treatment decisions).</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Increase in the average number of diagnoses and irrelevant diagnoses before and after DSS advice, broken down by grade</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Grade</bold></td><td align="center" colspan="3"><bold>No. of diagnoses</bold></td><td align="center" colspan="3"><bold>No. of irrelevant diagnoses</bold></td></tr></thead><tbody><tr><td></td><td align="center">Pre-DSS</td><td align="center">Post-DSS</td><td align="center">Increase</td><td align="center">Pre-DSS</td><td align="center">Post-DSS</td><td align="center">Increase</td></tr><tr><td colspan="7"><hr></hr></td></tr><tr><td align="left">Consultant</td><td align="center">3.3</td><td align="center">4.6</td><td align="center">1.3</td><td align="center">0.4</td><td align="center">0.7</td><td align="center">0.3</td></tr><tr><td align="left">Registrar</td><td align="center">4.3</td><td align="center">5.9</td><td align="center">1.6</td><td align="center">0.8</td><td align="center">1.3</td><td align="center">0.5</td></tr><tr><td align="left">SHO</td><td align="center">4.4</td><td align="center">6.1</td><td align="center">1.7</td><td align="center">0.6</td><td align="center">1.4</td><td align="center">0.8</td></tr><tr><td align="left">Medical student</td><td align="center">4.0</td><td align="center">6.6</td><td align="center">2.6</td><td align="center">1.1</td><td align="center">2.2</td><td align="center">1.1</td></tr></tbody></table></table-wrap></sec><sec><title>Additional diagnoses, tests and treatment decisions</title><p>At least one 'clinically important' diagnosis was added by the subject to their differential diagnosis <italic>after </italic>ISABEL consultation in 94/751 case episodes (12.5%, 95% CI 10.1%-14.9%). 47/76 (61.8%) subjects added at least one 'clinically important' diagnosis to their diagnostic workup after consultation. Overall, 130 'clinically important' diagnoses were added after DSS advice during the experiment. In general, students were reminded to consider many more important diagnoses than consultants, although this was not statistically significant (44 vs. 26, Chi square p > 0.05); a similar gradient was seen for difficult cases, but DSS consultation seemed helpful even for easy cases. Similar proportions for tests and treatment items were smaller in magnitude (table <xref ref-type="table" rid="T6">6</xref>). No clinically significant diagnoses were deleted after consultation. Important tests included by subjects in their pre-DSS plan were sometimes deleted from the post-DSS plan (64 individual items from 44 case episodes). A similar effect was seen for treatment steps (34 individual items from 24 case episodes). An inappropriate test was added to the post-ISABEL list in 7/751 cases.</p><table-wrap position="float" id="T6"><label>Table 6</label><caption><p>Number of case episodes in which clinically 'important' decisions were prompted by ISABEL consultation</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Number of 'important' decisions prompted by ISABEL</td><td align="center">Diagnoses</td><td align="center">Tests</td><td align="center">Treatment steps</td></tr></thead><tbody><tr><td align="left"><bold>1</bold></td><td align="center">69</td><td align="center">56</td><td align="center">42</td></tr><tr><td align="left"><bold>2</bold></td><td align="center">19</td><td align="center">12</td><td align="center">5</td></tr><tr><td align="left"><bold>3</bold></td><td align="center">2</td><td align="center">2</td><td align="center">2</td></tr><tr><td align="left"><bold>4</bold></td><td align="center">3</td><td align="center">0</td><td align="center">0</td></tr><tr><td align="left"><bold>5</bold></td><td align="center">1</td><td align="center">0</td><td align="center">0</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left"><bold>None</bold></td><td align="center">657</td><td align="center">637</td><td align="center">678</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left">No. of case episodes in which at least ONE 'significant' decision was prompted by ISABEL</td><td align="center">94 (12.5%)</td><td align="center">70 (9.3%)</td><td align="center">49 (6.5%)</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left">Total number of individual 'significant' decisions prompted by ISABEL</td><td align="center">130</td><td align="center">86</td><td align="center">58</td></tr></tbody></table></table-wrap><p>751 case episodes were used to examine the presence of the 'most likely diagnosis'. Overall, the 'most likely diagnosis/es' were included in the pre-DSS diagnostic workup by subjects in 507/751 (67.5%) case episodes. This increased to 561/751 (74.7%) case episodes after DSS advice. The improvement was fully attributable to positive consultation effects (where the 'most likely diagnosis' was absent pre-DSS but was present post-DSS); no negative consultations were observed. Diagnostic accuracy pre-ISABEL was greatest for consultants (73%) and least for medical students (57%). Medical students gained the most after DSS advice (an absolute increase of 10%). Analysis performed to elucidate whether ISABEL was responsible for the changes seen in the rate of diagnostic error indicated that all additional diagnoses were indeed present in the system's list of diagnostic suggestions.</p></sec><sec><title>Time intervals</title><p>Reliable time data was available for 633/751 episodes (table <xref ref-type="table" rid="T7">7</xref>). Median time taken for subjects to abstract clinical features and record their initial clinical decisions on the trial website was 6 min (IQR 4–10 min); median time taken to examine ISABEL's suggestions and make changes to clinical decisions was 1 min (IQR 30 sec-2 min). Time taken for ISABEL to display its suggestions was less than 2 sec on all occasions.</p><table-wrap position="float" id="T7"><label>Table 7</label><caption><p>Time taken to process case simulations broken down by grade of subject</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>Median time pre-ISABEL</bold></td><td align="center"><bold>Median time post-ISABEL</bold></td></tr></thead><tbody><tr><td align="left">Consultant</td><td align="center">5 min 5 sec</td><td align="center">42 sec</td></tr><tr><td align="left">Registrar</td><td align="center">5 min 45 sec</td><td align="center">57 sec</td></tr><tr><td align="left">SHO</td><td align="center">5 min 54 sec</td><td align="center">53 sec</td></tr><tr><td align="left">Medical student</td><td align="center">8 min 36 sec</td><td align="center">3 min 42 sec</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="left">Overall</td><td align="center">6 min 2 sec (IQR: 4:03 – 9:47)</td><td align="center">1 min (IQR: 30 sec – 2:04)</td></tr></tbody></table></table-wrap></sec></sec><sec><title>Discussion</title><p>We have shown in this study that errors of omission occur frequently during diagnostic workup in an experimental setting, including in cases perceived as being common in routine practice. Such errors seem to occur in most subjects, irrespective of their level of experience. We have also demonstrated that it is possible to influence clinicians' diagnostic workup and reduce errors of omission using a stand-alone diagnostic reminder system. Following DSS consultation, the quality of diagnostic, test-ordering and treatment decisions made by various grades of clinicians improved for a range of cases, such that a clinically important alteration in diagnostic decision-making resulted in 12.5% of all consultations (1 in 8 episodes of system use).</p><p>In a previous study assessing the impact of ILIAD and QMR, in which only diagnostically challenging cases were used in an experimental setting, Friedman <italic>et al </italic>showed that the 'correct diagnosis' was prompted by DSS use in approximately 1 in 16 consultations [<xref ref-type="bibr" rid="B48">48</xref>]. Although we used a similar experimental design, we used a mix of easy as well as difficult cases to test the hypothesis that incomplete workup was encountered in diagnostic conundrums as well as routine clinical problems. Previous evaluations of expert systems used the presence of the 'correct' diagnosis as the main outcome. We focused on clinical safety as the key outcome, preferring to use the inclusion of all 'clinically important' diagnoses in the workup as the main variable of interest. In acute settings such as emergency rooms and primary care, where an incomplete and evolving clinical picture results in considerable diagnostic uncertainty at assessment, the ability to generate a focused and 'safe' workup is a more clinically relevant outcome, and one which accurately reflects the nature of decision making in this environment [<xref ref-type="bibr" rid="B53">53</xref>]. Consequently, we defined diagnostic errors of omission at assessment as the 'failure to consider all clinically important diagnoses (as judged by an expert panel working the same cases)'. This definition resulted in the 'correct' diagnosis, as well as other significant diagnoses, being included within the 'minimum' workup. Further, changes in test-ordering and treatment decisions were uniquely measured in this study as a more concrete marker of the impact of diagnostic decision support on the patient's clinical management; we were able to demonstrate an improvement in test-ordering in 1 in 10 system consultations, indicating that diagnostic DSS may strongly influence patient management, despite only offering diagnosis-related advice. Finally, the time expended during DSS consultation is an important aspect that has not been fully explored in previous studies. In our study, subjects spent a median of 6 minutes for clinical data entry (including typing in their unaided decisions), and a median of 1 minute to process the advice provided and make changes to their clinical decisions.</p><p>The research design employed in this study allowed us to confirm a number of observations previously reported, as well as to generate numerous unique ones. These findings relate to the operational consequences of providing diagnostic assistance in practice. In keeping with other DSS evaluations, different subject grades processed system advice in different ways, depending on their prior knowledge and clinical experience, leading to variable benefit. Since ISABEL merely offered diagnostic suggestions, and allowed the clinician to make the final decisions (acting as the 'learned intermediary') [<xref ref-type="bibr" rid="B54">54</xref>], in some cases, subjects ignored even important advice. In some other cases, they added irrelevant decisions or deleted important decisions after DSS consultation, leading to reduced net positive effect of the DDSS on decision making. For some subjects whose pre-DSS performance was high, a ceiling effect prevailed, and no further improvement could be demonstrated. These findings complement the results of our earlier system performance evaluation which solely focused on system accuracy and not on user interaction with DDSS. One of the main findings from this study was that consultants tended to generate shorter diagnostic workup lists containing the 'most likely' diagnoses, with a predilection to omit other 'important' diagnoses that might account for the patient's clinical features, resulting in a high incidence of DEO. Medical students generated long diagnostic workup lists, but missed many key diagnoses leading to a high DEO rate. Interestingly, all subject grades gained from the use of ISABEL in terms of a reduction in the number of DEO, although to varying degrees. Despite more DEOs occurring in cases considered to be routine in practice than in rare and difficult ones in the pre-DSS consultation phase, ISABEL advice seemed to mainly improve decision making for difficult cases, with a smaller effect on easy cases. The impact of DSS advice showed a decreasing level of beneficial effect from diagnostic to test-ordering to treatment decisions. Finally, although the time taken to process cases without DSS advice in this study compared favorably with the Friedman evaluation of QMR and ILIAD (6 min vs. 8 min), the time taken to generate a revised workup with DSS assistance was dramatically shorter (1 min vs. 22 min).</p><p>We propose a number of explanations for our findings. There is sufficient evidence to suggest that clinicians with more clinical experience resort to established pattern-recognition techniques and the use of heuristics while making diagnostic decisions [<xref ref-type="bibr" rid="B55">55</xref>]. While these shortcuts enable quick decision making in practice, and work successfully on most occasions, they involve a number of cognitive biases such as 'premature closure' and 'confirmation bias' that may lead to incomplete assessment on some occasions. On the other hand, medical students may not have developed adequate pattern-recognition techniques or acquired sufficient knowledge of heuristics to make sound diagnostic decisions. It may well be that grades at an intermediate level are able to process cases in an acute setting with a greater emphasis on clinical safety. This explanation may also account for the finding that subjects failed to include 'important' diagnoses during the assessment of easy cases. Recognition that a case was unusual may trigger a departure from the use of established pattern-recognition techniques and clinical shortcuts to a more considered cognitive assessment, leading to fewer DEO in these cases. We have shown that it is possible to reduce DEOs by the use of diagnostic reminders, including in easy cases, although subjects appeared to be more willing to revise their decisions for difficult cases on the basis of ISABEL suggestions. It is also possible that some subjects ignored relevant advice because the system's explanatory capacity was inadequate and did not allow subjects to sufficiently discriminate between the suggestions offered. User variability in summarizing cases may also explain why variable benefits were derived from ISABEL usage – subjects may have obtained different results depending on how they abstracted and entered clinical features. This user variability during clinical data entry has been demonstrated even with use of a controlled vocabulary in QMR [<xref ref-type="bibr" rid="B56">56</xref>]. We observed marked differences between users' search terms for the same textual case; however, diagnostic suggestions did not seem to vary noticeably. This observation could be partially explained by the enormous diversity associated with various natural language disease descriptions contained within the ISABEL database, as well as by the system's use of a thesaurus that converts medical slang into recognized medical terms.</p><p>The diminishing level of impact from diagnostic to test-ordering to treatment decisions may be a result of system design – ISABEL does not explicitly state which tests and treatments to perform for each of its diagnostic suggestions. This advice is usually embedded within the textual description of the disease provided to the user. Study and system design may both account for the differences in time taken to process the cases. In previous evaluations, subjects processed cases without using the DSS in the first instance; in a subsequent step, they used the DSS to enter clinical data, record their clinical decisions, and processed system advice to generate a second diagnostic hypothesis list. In our study, subjects processed the case and recorded their own clinical decisions while using the DSS for clinical data entry. The second stage of the procedure only involved processing ISABEL advice and modifying previous clinical decisions. As such, direct comparison between the studies can be made only by the total time involved per case (30 min vs. 7 min). This difference could be explained by features in the system's design that resulted in shorter times to enter clinical data and to easily process the advice provided.</p><p>The findings from this study have implications specifically for ISABEL as well as other diagnostic DSS design, evaluation and implementation. It is well recognized that the dynamic interaction between user and DSS plays a major role in their acceptance by physicians [<xref ref-type="bibr" rid="B57">57</xref>]. We feel that adoption of the ISABEL system during clinical assessment in real time is possible even with current computer infrastructure, providing an opportunity for reduction in DEO. Its integration into an EMR would allow further control on the quality of the clinical input data as well as provision of active decision support with minimum extra effort. Such an ISABEL interface has currently been developed and tested with four commercial EMRs [<xref ref-type="bibr" rid="B58">58</xref>]; this integration also facilitates iterative use of the system during the evolution of a patient's condition, leading to increasingly specific diagnostic advice. The reminder system model aims to enable clinicians to generate 'safe' diagnostic workups in busy environments at high risk for diagnostic errors. This model has been successfully used to alter physician behavior by reducing errors of omission in preventive care [<xref ref-type="bibr" rid="B59">59</xref>]. It is clear from recent studies that diagnostic errors occur in the emergency room for a number of reasons. Cognitive biases, of which 'premature closure' and faulty context generation are key examples, contribute significantly [<xref ref-type="bibr" rid="B60">60</xref>]. Use of a reminder system may minimize the impact of some of these cognitive biases. When combined with cognitive forcing strategies during decision making, DDSS may act as 'safety nets' to reduce the incidence of omission errors in practice [<xref ref-type="bibr" rid="B61">61</xref>]. Reminders to perform important tests and treatment steps may also allow a greater impact on patient outcome [<xref ref-type="bibr" rid="B62">62</xref>]. A Web-based system model in our study allowed users from disparate parts of the country to participate in this study without need for additional infrastructure or financial resources, an implementation model that would minimize the cost associated with deployment in practice. Finally, the role of DDSS in medical education and training needs formal evaluation. In our study, medical students gained significantly from the advice provided, suggesting that use of DDSS during specific diagnostic tasks (e.g. problem-based case exercises) might be a valuable adjunct to current educational strategies. Familiarity with DDSS will also predispose to greater adoption of computerized aids during future clinical decision making.</p><p>The limitations of this study stem mainly from its experimental design. The repeated measures design raises the possibility that some of the beneficial effects seen in the study are a result of subjects 'rethinking' the case, or the consequence of a reflective process [<xref ref-type="bibr" rid="B63">63</xref>]. Consequently, ISABEL's effects in practice could be related to the extra time taken by users in processing cases. We believe that any such effects are likely to be minimal since subjects did not actually process the cases twice during the study – a summary of the clinical features was generated by subjects when the case was displayed for the first time, and subjects could not review the cases while processing ISABEL suggestions in the next step. Subjects also spent negligible time between their first assessment of the cases and processing the diagnostic suggestions from the DSS. The repeated measures design provided the power to detect differences between users with minimal resources; a randomized design using study and control groups of subjects would have necessitated the involvement of over 200 subjects. The cases used in our study contained only basic clinical data gained at the time of acute assessment, and may have proved too concise or easy to process. However, this seems unlikely since subjects only took an average of 8 min to process even diagnostic conundrums prior to DSS use when 'expert systems' were tested. Our cases pertained to emergency assessments, making it difficult to generalize the results to other ambulatory settings. The ability to extract clinical features from textual cases may not accurately simulate a real patient encounter where missed data or 'red herrings' are quite common. The inherent complexity involved in patient assessment and summarizing clinical findings in words may lead to poorer performance of the ISABEL system in real life, since its diagnostic output depends on the quality of user input. As a corollary, some of our encouraging results may be explained by our choice of subjects: a few were already familiar with summarizing clinical features into the DSS. Subjects were not supervised during their case exercises since they may have performed differently under scrutiny, raising the prospect of a Hawthorne effect [<xref ref-type="bibr" rid="B64">64</xref>]. The use of a structured website to explicitly record clinical decisions may have invoked the check-list effect, as illustrated in the Leeds abdominal pain system study [<xref ref-type="bibr" rid="B65">65</xref>]. The check list effect might also be invoked during the process of summarizing clinical features for ISABEL input; this may have worked in conjunction with 'rethinking' to promote better decision making pre-ISABEL. We also measured decision making at a single point in time, making it difficult to assess the effects of iterative usage of the DSS on the same patient. Finally, our definition of diagnostic error aimed to identify inadequate diagnostic workup at initial assessment that might result in a poor patient outcome. We recognize the absence of an evidence-based link between omission errors and diagnostic adverse events in practice, although according to the Schiff model [<xref ref-type="bibr" rid="B53">53</xref>], it seems logical to assume that avoiding process errors will prevent actual errors at least in some instances. In the simulated setting, it was not possible to test whether inadequate diagnostic workup would directly lead to a diagnostic error and cause patient harm. Our planned clinical impact assessment in real life would help clarify many of the questions raised during this experimental study.</p></sec><sec><title>Conclusion</title><p>This experimental study demonstrates that diagnostic omission errors are common during the assessment of easy as well as difficult cases. The provision of patient- and context-specific diagnostic reminders has the potential to reduce these errors across all subject grades. Our study suggests a promising role for the use of future reminder-based DSS in the reduction of diagnostic error. An impact evaluation, utilizing a naturalistic design and conducted in real life clinical practice, is underway to verify the conclusions derived from this simulation.</p></sec><sec><title>Competing interests</title><p>This study was conducted when the ISABEL system was available free to users, funded by the ISABEL Medical Charity (2001–2002). Dr Ramnarayan performed this research as part of his MD thesis at Imperial College London. Dr Britto was a Trustee of the ISABEL medical charity (non-remunerative post). Ms Tomlinson was employed by the ISABEL Medical Charity as a Research Nurse.</p><p>Since June 2004, ISABEL is managed by a commercial subsidiary of the Medical Charity called ISABEL Healthcare. The system is now available only to subscribers. Dr Ramnarayan now advises ISABEL Healthcare on research activities on a part-time basis; Dr Britto is now Clinical Director of ISABEL Healthcare; Ms Tomlinson is responsible for Content Management within ISABEL Healthcare. All three hold stock options in ISABEL Healthcare. All other authors declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>PR conceived the study, contributed to the study design, analyzed the data and drafted the manuscript.</p><p>GR assisted with the design of the study and data analysis.</p><p>MC assisted with the study design, served as gold standard panel member, and revised the draft manuscript</p><p>VN assisted with study design, served as gold standard panel member, and helped with data analysis</p><p>MT assisted with data collection and analysis</p><p>PT assisted with study conception, study design and revised the draft manuscript</p><p>JW assisted with study conception, provided advice regarding study design, and revised the draft manuscript</p><p>JB assisted with study conception, study design and data analysis.</p><p>All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1472-6947/6/22/prepub"/></p></sec>
|
Evaluation and selection of tandem repeat loci for a <italic>Brucella </italic>MLVA typing assay
|
<sec><title>Background</title><p>The classification of <italic>Brucella </italic>into species and biovars relies on phenotypic characteristics and sometimes raises difficulties in the interpretation of the results due to an absence of standardization of the typing reagents. In addition, the resolution of this biotyping is moderate and requires the manipulation of the living agent. More efficient DNA-based methods are needed, and this work explores the suitability of multiple locus variable number tandem repeats analysis (MLVA) for both typing and species identification.</p></sec><sec><title>Results</title><p>Eighty tandem repeat loci predicted to be polymorphic by genome sequence analysis of three available <italic>Brucella </italic>genome sequences were tested for polymorphism by genotyping 21 <italic>Brucella </italic>strains (18 reference strains representing the six 'classical' species and all biovars as well as 3 marine mammal strains currently recognized as members of two new species). The MLVA data efficiently cluster the strains as expected according to their species and biovar. For practical use, a subset of 15 loci preserving this clustering was selected and applied to the typing of 236 isolates. Using this MLVA-15 assay, the clusters generated correspond to the classical biotyping scheme of <italic>Brucella </italic>spp. The 15 markers have been divided into two groups, one comprising 8 user-friendly minisatellite markers with a good species identification capability (panel 1) and another complementary group of 7 microsatellite markers with higher discriminatory power (panel 2).</p></sec><sec><title>Conclusion</title><p>The MLVA-15 assay can be applied to large collections of <italic>Brucella </italic>strains with automated or manual procedures, and can be proposed as a complement, or even a substitute, of classical biotyping methods. This is facilitated by the fact that MLVA is based on non-infectious material (DNA) whereas the biotyping procedure itself requires the manipulation of the living agent. The data produced can be queried on a dedicated MLVA web service site.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Le Flèche</surname><given-names>Philippe</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Jacques</surname><given-names>Isabelle</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Grayon</surname><given-names>Maggy</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Al Dahouk</surname><given-names>Sascha</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Bouchon</surname><given-names>Patrick</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Denoeud</surname><given-names>France</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Nöckler</surname><given-names>Karsten</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Neubauer</surname><given-names>Heinrich</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Guilloteau</surname><given-names>Laurence A</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A10" corresp="yes" contrib-type="author"><name><surname>Vergnaud</surname><given-names>Gilles</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Microbiology
|
<sec><title>Background</title><p>Brucellosis is a zoonosis affecting animals and humans worldwide. <italic>Brucella </italic>infections may result in significant economic losses due to abortion and slaughtering of infected animals. Humans are mainly infected through the consumption of contaminated dairy products or by direct contact with infected animals. In addition, certain <italic>Brucella </italic>spp have to be considered as potential biowarfare agents. Six species are currently recognized, <italic>B. abortus </italic>(8 biovars), <italic>B. melitensis </italic>(3 biovars), <italic>B. suis </italic>(5 biovars), <italic>B. ovis</italic>, <italic>B. canis </italic>and <italic>B. neotomae </italic>[<xref ref-type="bibr" rid="B1">1</xref>]. More recently, <italic>Brucella </italic>strains have been isolated from marine mammals [<xref ref-type="bibr" rid="B2">2</xref>], suggesting the existence of additional species [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>].</p><p>The genus <italic>Brucella </italic>is highly homogeneous (more than 90% DNA/DNA homology [<xref ref-type="bibr" rid="B5">5</xref>]). <italic>Brucella </italic>classification is mainly based on differences in pathogenicity, host preferences, and conventional microbiological tests used for phenotyping (biotyping) [<xref ref-type="bibr" rid="B6">6</xref>]. Routine identification of <italic>Brucella </italic>species and biovars still relies on biotyping (reviewed in [<xref ref-type="bibr" rid="B7">7</xref>]). Only a few tools exist for further molecular subtyping, of which none has proven to be fully satisfactory for epidemiologic investigations or tracing back strains to their origin. Tandem repeat (TR) sequences may be an interesting class of markers, since multiple alleles can be present at a single locus, and size differences are easily resolved by electrophoresis (reviewed by [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]). Tandem repeats are often classified as microsatellites (repeat units up to 8 bp) and minisatellites [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. Tandem repeat typing has proven to be highly appropriate for the typing of pathogenic bacterial species with a high genetic homogeneity, including the <italic>Mycobacterium tuberculosis </italic>complex, <italic>Bacillus anthracis</italic>, and <italic>Yersinia pestis </italic>[<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B15">15</xref>]. Recently, a family of tandem repeats located within a repeated sequence and present in multiple loci in the <italic>Brucella </italic>genome was used for strain typing [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. The proposed set of eight microsatellite loci is extremely discriminant and highly efficient to distinguish strains within a local outbreak, but is unable to correctly predict the biovar or even the species of an isolate. A possible reason for that is the high mutation rate of these loci. Consequently, this MLVA assay cannot replace classical biotyping methods.</p><p>The availability of the whole genome sequences of <italic>B. melitensis </italic>16 M, <italic>B. suis </italic>1330 and <italic>B. abortus </italic>strain 9–941 [<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref>] greatly facilitates the search for polymorphic DNA sequences [<xref ref-type="bibr" rid="B21">21</xref>]. In this report, we evaluated most tandem repeats showing at least two alleles among the three sequenced strains [<xref ref-type="bibr" rid="B22">22</xref>]. Eighteen reference strains and 3 strains isolated from marine mammals [<xref ref-type="bibr" rid="B23">23</xref>] were typed using these TR candidates to evaluate their associated polymorphism. For routine typing, a subset of 15 markers which enabled to cluster the isolates according to their biotype was selected. This set of markers was further evaluated on a collection of 236 isolates representing the major biovars affecting terrestrial mammals (Table <xref ref-type="table" rid="T1">1</xref>) to produce a first reference data set [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>] which can be queried via the internet [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B24">24</xref>].</p></sec><sec><title>Results</title><sec><title>Evaluation of tandem repeats polymorphism</title><p>Comparison of the three genome sequences [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>] identifies 107 TRs with a repeat unit larger than 5 bp and predicted to display size polymorphism. Eighty of them were evaluated for polymorphism among 21 reference and marine mammal strains (Table <xref ref-type="table" rid="T1">1</xref>). Twenty-two TRs (numbered Bruce01 to Bruce22 in Table <xref ref-type="table" rid="T2">2</xref>) have three predicted alleles. Twelve of the 22 are octamers, five of which have been previously characterized [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Typing was done by PCR using the set of primers listed in Table <xref ref-type="table" rid="T2">2</xref>, as described [<xref ref-type="bibr" rid="B13">13</xref>]. Six markers failed to amplify DNA satisfactorily, and were not included in the further study: they generated multiple band profiles (bruce20-BRU329_8bp_148bp_7u; bruce38-BRU1116_18bp_108bp_2u; bruce71-BRU337_12bp_394bp_3u), or lacked amplification using the selected primers (bruce79-BRU163_12bp_141bp_4u), or no appropriate primers could be designed targeting the flanking regions because of the presence of repeated elements (bruce76-BRU243_21bp_2u; bruce77-BRU195_21bp_2u, not listed in Table <xref ref-type="table" rid="T2">2</xref>).</p><p>Three markers (bruce44-BRU256_12bp_110bp_3u; bruce65-BRU824_41bp_182bp_2u; bruce69-BRU488_57bp_181bp_1u) turned out to be monomorphic for the 21 reference strains. The results of the clustering analysis using the 71 remaining markers fits very well with the current knowledge of the degree of relationship between <italic>Brucella </italic>species [<xref ref-type="bibr" rid="B25">25</xref>] (Figure <xref ref-type="fig" rid="F1">1</xref>). We then looked for a subset of markers providing a similar discriminative power as the whole set for the collection of reference strains evaluated. Although extremely informative, the family of octamers, which includes the eight tandem repeats previously investigated [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>], are not appropriate for species/biovar discrimination because of their hypervariability and more stable markers must be used. Among the other markers, a set of the ten most polymorphic loci clusters the different species as expected. Two of these ten markers display allele size ranges not appropriate for analysis on currently available automated DNA fragments sizing machines such as capillary electrophoresis sequencing machines (Bruce02 and Bruce15 have alleles up to 2 kb and 5 kb respectively). The amplification patterns of the 21 reference strains using the other eight TRs are shown in Figure <xref ref-type="fig" rid="F2">2</xref>. These 8 markers (Bruce06, 08, 11, 12, 42, 43, 45, 55) will subsequently be called MLVA typing panel 1. These are minisatellites loci with repeat units length above 9 bp [<xref ref-type="bibr" rid="B10">10</xref>]. In addition, 7 robust and highly polymorphic octamers (microsatellites) were selected to constitute MLVA typing panel 2. Panel 2 comprises Bruce04 (designated as TR6 in [<xref ref-type="bibr" rid="B16">16</xref>]), Bruce07, Bruce09 (TR8), Bruce16, Bruce18, Bruce21 and Bruce30 (TR2).</p></sec><sec><title>Evaluation of a MLVA assay comprising 15 markers</title><p>The set of 15 TR markers (panel 1 and 2, listed with one or two asterisk in Table <xref ref-type="table" rid="T2">2</xref>) was used for typing a larger collection of biotyped isolates including various species and biovars [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. Among the 257 strains, panel 1 alone resolves 51 genotypes. This panel does not distinguish <italic>B. suis </italic>biovar 4 and <italic>B. canis</italic>. All <italic>B. canis </italic>strains investigated share panel 1 genotype 2 with some of the <italic>B. suis </italic>biovar 4 strains (Figure <xref ref-type="fig" rid="F3">3</xref>). Similarly, most <italic>B. suis </italic>biovar 3 strains share panel 1 genotype 4 with <italic>B. suis </italic>biovar 1. Panel 2 alone discriminates 200 genotypes. However, the resulting clustering only approximately fits with the expected species and biovar assignment. When using panel 1 and panel 2 together (MLVA-15 assay), 204 genotypes can be differentiated. The clustering analysis is shown in Figure <xref ref-type="fig" rid="F3">3</xref>, <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F5">5</xref>. A number of major clusters weakly connected to each other can be identified: <italic>B. suis </italic>biovar 1 (Figure <xref ref-type="fig" rid="F3">3</xref>), <italic>B. suis </italic>biovar 2( Figure <xref ref-type="fig" rid="F3">3</xref> and figure <xref ref-type="fig" rid="F4">4</xref>), <italic>B. abortus </italic>(2 clusters, Figure <xref ref-type="fig" rid="F4">4</xref> and Figure <xref ref-type="fig" rid="F5">5</xref>), <italic>B. melitensis </italic>(3 clusters, figure <xref ref-type="fig" rid="F5">5</xref>), <italic>B. ovis</italic> (Figure <xref ref-type="fig" rid="F3">3</xref>). <italic>Brucella suis </italic>biovar 5, <italic>B. neotomae </italic>and the marine mammal strains are quite distinct from the closest strains (Figure <xref ref-type="fig" rid="F4">4</xref>). <italic>Brucella canis </italic>and <italic>B. suis </italic>biovar 4 are closely related and loosely connected to the <italic>B. suis </italic>biovar 1 cluster (Figure <xref ref-type="fig" rid="F3">3</xref>). The three <italic>B. melitensis </italic>clusters fit moderately with the biotyping results. Similarly, <italic>B. suis </italic>biovar 3 strains do not constitute a consistent group.</p></sec></sec><sec><title>Discussion</title><p>The genus <italic>Brucella </italic>has been divided into species and biovars for a long time, but this classification has been discussed controversially since DNA-DNA hybridization has been applied. The genus proved to be highly monomorphic with a level of relatedness among all species higher than 90% [<xref ref-type="bibr" rid="B5">5</xref>]. This homogeneity complicated the development of molecular assays able to efficiently recognise the species-specific entities. This finding led to the proposal of a monospecies genus, i.e. <italic>B. melitensis</italic>. The classical species would be considered as biovars only. However, most bacteriologists did not accept this concept which has recently been rejected by the subcommittee of taxonomy [<xref ref-type="bibr" rid="B26">26</xref>]. The purpose of the present study was firstly to investigate the polymorphism of tandem repeat loci predicted to be polymorphic by comparing the data of the three different <italic>Brucella </italic>strains already sequenced and secondly to evaluate to which extend tandem repeat typing and classical biotyping clustering fit together. We evaluated most of these loci with a repeat unit of 5 bp or more.</p><p>Polymorphism has been confirmed at 71 loci. DNA was amplified at every locus from all 21 reference strains, including the 3 marine mammal strains (except for Bruce04 in the <italic>B. melitensis </italic>bv 3 reference strain Ether and Bruce01 in the <italic>B.ovis</italic> reference strain BOW63/290) confirming the very high genetic homogeneity of the genus <italic>Brucella</italic>.</p><p>A MLVA typing assay depends on the selection of markers which individually would not provide a relevant clustering. Taken separately, the TR markers are either not informative enough, or too variable or show a high level of homoplasy. However, the combination of well selected independent loci may be highly discriminatory and to some extend phylogenetically relevant, as shown previously for other species [<xref ref-type="bibr" rid="B9">9</xref>], and demonstrated here for <italic>Brucella</italic>. We propose a selection of 15 markers to be used in a <italic>Brucella </italic>MLVA assay consisting of two complementary panels, panel 1 (8 markers) and panel 2 (7 markers). The fifteen markers are a combination of moderately variable (minisatellites, panel 1) and highly discriminant (microsatellites, panel 2) loci (Table <xref ref-type="table" rid="T2">2</xref>).</p><p>The strain clustering achieved is consistent with well-established phenotypic and molecular characteristics (Figure <xref ref-type="fig" rid="F3">3</xref>, <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F5">5</xref>). The biovars 1, 2 and 4 of <italic>B. abortus </italic>are gathered in agreement with (i) the sensitivity to thionin and (ii) the PCR-RFLP pattern of the <italic>omp2a </italic>genes specific for these biovars [<xref ref-type="bibr" rid="B27">27</xref>]. <italic>B. abortus </italic>biovar 3 strains are found in a separate group except for 2 strains originated from Africa (BCCN 93-26 and the reference strain Tulya). Strains isolated in Africa often show distinct phenotypes [<xref ref-type="bibr" rid="B28">28</xref>] and thus, it is not surprising to find these two strains separated. The two strains do not require CO<sub>2 </sub>for growing. Their MLVA closest neighbours are two <italic>B. abortus </italic>biovar 6 strains also isolated in Africa. Assignment to biovar 3 or 6 reflects the H<sub>2</sub>S production which is the unique phenotypical criteria to differentiate these two biovars. The MLVA assay confirms that some African strains significantly differ from isolates of other origin and that <italic>B. abortus </italic>biovar 3 is a heterogeneous group.</p><p>The <italic>B. melitensis </italic>group is very heterogeneous using either panel 1 or both panels (MLVA-15), and comprises four main subgroups. Biovar 2 and 3 strains are mixed in two groups, together with a few biovar 1 strains. The other biovar 1 isolates form 2 groups, one including the 16M reference strain, and the other (genotypes 173 and 174, Figure <xref ref-type="fig" rid="F5">5</xref>) comprising 3 isolates from the United Arab Emirates. <italic>B. melitensis </italic>BCCN 84-3 strain (MLVA-15 genotype 20) is an isolate from a dog in Costa Rica, which was biotyped as <italic>B. melitensis </italic>biovar 2, but appears to be distantly related to other <italic>B. melitensis </italic>strains. This strain is smooth as observed by the agglutination with anti-A serum, and the profile obtained in oxidative metabolism is typical of <italic>B. melitensis</italic>. Panel 1 analysis (not shown) does associate this strain with <italic>B. melitensis</italic>, but the full MLVA-15 analysis suggests a position closer to the <italic>B. canis </italic>group (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><p><italic>B. suis </italic>strains are clearly differentiated in three groups (Figures <xref ref-type="fig" rid="F3">3</xref> and <xref ref-type="fig" rid="F4">4</xref>). A first group includes all biovar 1, 3, and 4 strains, and a second group all biovar 2 strains. The two rare biovar 5 strains are very distantly related. The correlation with biovars is good with some interesting exceptions. The five <italic>B. suis </italic>biovar 3 isolates from Croatia have the same genotype (MLVA-15 genotype 36, Figure <xref ref-type="fig" rid="F3">3</xref> [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]), and cluster with <italic>B. suis </italic>biovar 1 strains but not with the reference <italic>B. suis </italic>biovar 3 strain. More <italic>B. suis </italic>strains phenotypically identified as biovar 3 from other geographic origins are required. This may suggest that the biovar 3 phenotype may have appeared independently more than once. Biovar 1 and biovar 3 strains are distinguished by sensitivity to fuchsine and ability to produce H<sub>2</sub>S. Atypical fuchsine-resistant biovar 1 strains have already been described [<xref ref-type="bibr" rid="B6">6</xref>], as well as atypical fuchsine-sensitive <italic>B. melitensis </italic>strains [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>]. So both the fuchsine sensitivity, and the H<sub>2</sub>S production (as suggested above for <italic>B. abortus</italic>) may appear to be phylogenetically weak markers with some degree of homoplasy. Among biovar 2, strains isolated from Spain and Portugal are related and can be distinguished from other European strains investigated. Biovar 4 strains can be found right beside <italic>B. canis</italic>. Meyer [<xref ref-type="bibr" rid="B31">31</xref>] has previously proposed a model for evolutionary derivation of <italic>Brucella </italic>organisms on the basis of phenotypic characteristics and proposed a close relationship between <italic>B. suis </italic>biovar 3/4, and <italic>B. canis</italic>. PCR-RFLP analyses of the porin genes are in agreement with this finding [<xref ref-type="bibr" rid="B27">27</xref>].</p><p>Three classical vaccine strains were included, Rev.1 (genotype 201), S19 (genotype 161) and RB51 (genotype 159). Six other isolates, from Israel, share genotype 201. These streptomycin resistant isolates were confirmed as Rev.1 vaccine strains using the previously described assay [<xref ref-type="bibr" rid="B32">32</xref>] (data not shown). This is not unexpected since vaccination is used in this country, and simply illustrates the stability of the MLVA assay in the present case.</p><p>Strains clustering together frequently have a close or identical geographic origin, e.g. MLVA-15 genotype 16 comprises 2 <italic>B. ovis </italic>isolates, coming from the same region of France "Provence-Côte d'Azur" (departments 06 and 13). In almost all such instances where the MLVA genotype of two isolates is identical, the available epidemiological data is indeed compatible with a common source of infection. The rare exceptions would then suggest that some strains travel efficiently. MLVA-15 genotype 132 was observed in Germany in 1972 and in the centre of France (department 87) in 1994. MLVA-15 genotype 1 (<italic>B. canis</italic>) was observed in Greece and Germany. More epidemiological data will be needed in order to draw precise conclusions on the circulation of the strains.</p><p>The MLVA-15 results support the current classification of the genus <italic>Brucella</italic>. In addition, differences found by phenotypic identification and/or by molecular studies are also detected by MLVA. One major advantage of MLVA is the ease of data exchanges. The data itself can be summarized by a very simple flat text file containing the repeat copy numbers for each locus and each strain. This data can also be made accessible and queried across the internet as shown [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B24">24</xref>].</p><p>Another advantage is that MLVA typing only depends on the measurement of DNA amplicon sizes, so that a number of electrophoretic techniques can be used, ranging from manual, low-cost, agarose gels, to high-throughput capillary electrophoresis sequencing machines.</p><p>In the near future, it is tempting to speculate that international databases containing MLVA data of thousands of strains will be produced, and MLVA will become a routine assay for any new isolate. We believe that the MLVA-15 assay will be one step in this direction. A first use of the assay for a clinical application was recently described [<xref ref-type="bibr" rid="B33">33</xref>].</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Bacterial strains</title><p>The 257 strains and isolates used for MLVA typing are listed or described globally in Table <xref ref-type="table" rid="T1">1</xref>. One hundred and seventeen <italic>B. suis</italic>, 43 <italic>B. melitensis</italic>, 52 <italic>B. abortus</italic>, 24 <italic>B. ovis</italic>, one <italic>B. neotomae</italic>, 17 <italic>B. canis </italic>and 3 strains isolated from marine mammals [<xref ref-type="bibr" rid="B2">2</xref>] were investigated. This collection includes the 18 classical reference strains representing the different species and biovars of <italic>Brucella</italic>. All strains were mainly isolated from animals and in a few cases from humans or unknown species (Figure <xref ref-type="fig" rid="F3">3</xref>, <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F5">5</xref>), and were identified by phenotypical tests based on agglutination with monospecific antisera (serotyping), phage typing, dye sensitivity, CO<sub>2 </sub>requirement and H<sub>2</sub>S production [<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec><title>Identification of variable number tandem repeats by genomic sequence comparison</title><p>The methods previously described [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>] and the genome sequence data for <italic>B. suis </italic>strain 1330, <italic>B. melitensis </italic>strain 16 M and <italic>B. abortus </italic>strain 9–941 [<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref>] were used to identify TRs that may help to differentiate closely related genomes.</p><p>The different TRs are designated by using the nomenclature previously described [<xref ref-type="bibr" rid="B13">13</xref>]. For instance BRU211_63bp_257bp_2u (bruce11) is a TR at position 211 kb in the <italic>B. melitensis </italic>16 M genome. Its common laboratory name (alias name) is Bruce11. It has a 63 bp motif, and a total PCR product length of 257 bp in the <italic>B. melitensis </italic>16 M strain when using the primer set indicated in Table <xref ref-type="table" rid="T2">2</xref>. This allele size corresponds to 2 units.</p></sec><sec><title>PCR amplification and genotyping</title><p><italic>Brucella </italic>DNA was prepared as previously described [<xref ref-type="bibr" rid="B27">27</xref>]. PCR amplification was performed in a total volume of 15 μl containing 1ng of DNA, 1× PCR Reaction Buffer, 1 U of <italic>Taq </italic>DNA polymerase (Qbiogen, Illkirch, France), 200 μM of each deoxynucleotide triphosphate, and 0.3 μM of each flanking primer as described previously [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>Amplifications were performed in a MJ Research PTC200 thermocycler. An initial denaturation step at 96°C for 5 minutes was followed by 30 cycles of denaturation at 96°C for 30 s, primer annealing at 60°C for 30 s, and elongation at 70°C for 1 min. The final extension step was performed at 70°C for 5 min.</p><p>Two to five microliters of the amplification product were loaded on a 3% standard agarose gel for analyzing tandem repeats with a unit length shorter than 10 bp and on a 2% standard agarose gel for all others, and run under a voltage of 8 V/cm until the bromophenol blue dye had reached the 20 cm position. Gels were stained with ethidium bromide, visualized under UV light, and photographed (Vilber Lourmat, Marnes-la-Vallée, France). A 100-bp and a 20-bp ladder (EZ Load 100 pb or 20 bp PCR Molecular Ruler, Biorad, Marnes-la-Coquette, France) were used as molecular size markers depending on the tandem repeat unit length. Gel images were managed using the Bionumerics software package (version 4.0, Applied-Maths, Belgium).</p></sec><sec><title>Data analysis</title><p>Band size estimates were converted to a number of units within a character dataset using Bionumerics version 4.0 (Applied-Maths, Belgium) [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. Clustering analyses used the categorical coefficient and UPGMA (unweighted pair group method using arithmetic averages). The use of the categorical parameter implies that the character states are considered unordered. The same weight is given to a large or a small number of differences in the number of repeats at each locus. Maximum parsimony was done using Bionumerics, running 200 bootstrap simulations and treating the data as categorical.</p></sec><sec><title>Polymorphism index</title><p>The Hunter Gaston diversity index [<xref ref-type="bibr" rid="B34">34</xref>] (HGDI) was used.</p></sec></sec><sec><title>Authors' contributions</title><p>MG, IJ, SAD, KN, HN were in charge of strain selection, collection and checking of related data, preparation and provision of DNAs. PLF did the MLVA genotyping work. GV was in charge of the Bionumerics database, error checking, clustering analyses. FD and PB did the genome sequence analyses for polymorphic tandem repeat searches and the genotyping page. GV wrote the report. IJ and MG helped to draft the manuscript. All authors read, commented and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>MLVA-15 data for each of the 204 genotypes. The first three columns from the left are genotype numbers obtained with the different panels. The subsequent columns are the typing data itself. The first 8 markers (headings, bruce06 to bruce55) constitute panel 1 (minisatellites, tandem repeat unit length above 9 bp). The last 7 columns (starting from bruce04) constitute panel 2 (microsatellites, tandem repeat unit length up to 8 base-pairs).</p></caption><media xlink:href="1471-2180-6-9-S1.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Genetic diversity of <italic>Clostridium perfringens </italic>type A isolates from animals, food poisoning outbreaks and sludge
|
<sec><title>Background</title><p><italic>Clostridium perfringens</italic>, a serious pathogen, causes enteric diseases in domestic animals and food poisoning in humans. The epidemiological relationship between <italic>C. perfringens </italic>isolates from the same source has previously been investigated chiefly by pulsed-field gel electrophoresis (PFGE). In this study the genetic diversity of <italic>C. perfringens </italic>isolated from various animals, from food poisoning outbreaks and from sludge was investigated.</p></sec><sec><title>Results</title><p>We used PFGE to examine the genetic diversity of 95 <italic>C. perfringens </italic>type A isolates from eight different sources. The isolates were also examined for the presence of the beta2 toxin gene (<italic>cpb2</italic>) and the enterotoxin gene (<italic>cpe</italic>). The <italic>cpb2 </italic>gene from the 28 <italic>cpb2</italic>-positive isolates was also partially sequenced (519 bp, corresponding to positions 188 to 706 in the consensus <italic>cpb2 </italic>sequence). The results of PFGE revealed a wide genetic diversity among the <italic>C. perfringens </italic>type A isolates. The genetic relatedness of the isolates ranged from 58 to 100% and 56 distinct PFGE types were identified. Almost all clusters with similar patterns comprised isolates with a known epidemiological correlation.</p><p>Most of the isolates from pig, horse and sheep carried the <italic>cpb2 </italic>gene. All isolates originating from food poisoning outbreaks carried the <italic>cpe </italic>gene and three of these also carried <italic>cpb2</italic>. Two evolutionary different populations were identified by sequence analysis of the partially sequenced <italic>cpb2 </italic>genes from our study and <italic>cpb2 </italic>sequences previously deposited in GenBank.</p></sec><sec><title>Conclusion</title><p>As revealed by PFGE, there was a wide genetic diversity among <italic>C. perfringens </italic>isolates from different sources. Epidemiologically related isolates showed a high genetic similarity, as expected, while isolates with no obvious epidemiological relationship expressed a lesser degree of genetic similarity. The wide diversity revealed by PFGE was not reflected in the 16S rRNA sequences, which had a considerable degree of sequence similarity. Sequence comparison of the partially sequenced <italic>cpb2 </italic>gene revealed two genetically different populations. This is to our knowledge the first study in which the genetic diversity of <italic>C. perfringens </italic>isolates both from different animals species, from food poisoning outbreaks and from sludge has been investigated.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Johansson</surname><given-names>Anders</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Aspan</surname><given-names>Anna</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Bagge</surname><given-names>Elisabeth</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Båverud</surname><given-names>Viveca</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Engström</surname><given-names>Björn E</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Johansson</surname><given-names>Karl-Erik</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Microbiology
|
<sec><title>Background</title><p><italic>Clostridium perfringens</italic>, an anaerobic Gram-positive bacterium known to be a common pathogen in humans, in domestic animals and in wildlife, is the primary cause of clostridial enteric disease in domestic animals. The complete genome sequence of <italic>C. perfringens </italic>has been published previously [<xref ref-type="bibr" rid="B1">1</xref>]. <italic>C. perfringens </italic>has 10 rRNA operons whose heterogeneity was investigated by Shimizu <italic>et al</italic>. [<xref ref-type="bibr" rid="B2">2</xref>]. They found 18 polymorphic sites among the 16S rRNA genes. A common feature of <italic>C. perfringens </italic>is the large number of exotoxins produced; 17 different exotoxins have been described in the literature [<xref ref-type="bibr" rid="B3">3</xref>]. In addition, <italic>C. perfringens </italic>produces an enterotoxin, CPE [<xref ref-type="bibr" rid="B4">4</xref>].</p><p><italic>Clostridium perfringens </italic>is subdivided into five toxinotypes (A – E) based on the production of the four major exotoxins (viz. alpha, beta, epsilon, and iota). The major toxins together with the enterotoxin and the beta2 toxin [<xref ref-type="bibr" rid="B5">5</xref>], play an important role in several serious diseases [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. CPE causes food-borne disease in humans, canine enteritis and porcine enteritis. Beta2 toxin, recently described [<xref ref-type="bibr" rid="B5">5</xref>], has been associated with enteric diseases in domestic animals, especially piglets [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>] and horses [<xref ref-type="bibr" rid="B10">10</xref>]. However, two recently published studies, by Jost <italic>et al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>] and Vilei <italic>et al</italic>. [<xref ref-type="bibr" rid="B12">12</xref>], demonstrated that beta2 toxin, encoded by <italic>cpb2</italic>, was expressed by most porcine <italic>C. perfringens </italic>isolates, but seldom by isolates of non-porcine origin. The results of those studies indicate that beta2 toxin does not cause enteritis in animal species other than pigs. Vilei <italic>et al</italic>. [<xref ref-type="bibr" rid="B12">12</xref>] reported that gentamicin and streptomycin induced expression of an atypical <italic>cpb2 </italic>gene in a non-porcine isolate. In a recent publication by Waters <italic>et al</italic>. [<xref ref-type="bibr" rid="B13">13</xref>], it was reported that <italic>cpb2 </italic>of <italic>C. perfringens </italic>from horses was transcriptionally active and that the levels of <italic>cpb2 </italic>mRNA were 35-fold lower than a high beta2 toxin producing pig isolate. Isolates originating from humans with gastrointestinal diseases carrying both <italic>cpb2 </italic>and <italic>cpe </italic>have recently been described [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>].</p><p>The epidemiological relationship between <italic>C. perfringens </italic>isolates has previously been investigated primarily by pulsed-field gel electrophoresis (PFGE) and in most of these studies a majority of isolates from food poisoning outbreaks were examined [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B21">21</xref>]. <italic>C. perfringens </italic>isolates originating from poultry have also been investigated previously by PFGE [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. The general conclusions drawn from the previously published articles, concerning both food poisoning outbreaks and animals, is that isolates from the same outbreak have very similar patterns while the genetic diversity is high in non-outbreak isolates and isolates selected randomly [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. The problem of DNA degradation of certain isolates due to endogenous bacterial nucleases, which are rather common among clostridial isolates, has been discussed elsewhere [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B26">26</xref>].</p><p>The purpose of this study was to compare the genetic relationships of <italic>C. perfringens </italic>type A from eight different sources by PFGE. A further aim was to investigate the distribution of the <italic>cpb2 </italic>and <italic>cpe </italic>genes. The <italic>cpb2 </italic>gene from all <italic>cpb2</italic>-positive isolates was also partially sequenced. In this study a generally wide genetic diversity of <italic>C. perfringens </italic>isolates from eight different sources was found. Furthermore, PFGE clearly distinguished between unrelated isolates of <italic>C.perfringens </italic>and supported a clonal relationship between related isolates. Sequence analysis of the partially sequenced <italic>cpb2 </italic>gene revealed two genetically different populations of the gene.</p></sec><sec><title>Results</title><sec><title>PCR</title><p>Multiplex PCR detected only the alpha-toxin gene (<italic>plc</italic>) and all isolates were therefore classified as <italic>C. perfringens </italic>type A. Altogether 28 isolates carried the <italic>cpb2 </italic>gene and 17 carried the <italic>cpe </italic>gene (Table <xref ref-type="table" rid="T1">1</xref>). The <italic>cpb2 </italic>gene was found in 6 of the 8 groups studied: pigs (83%), horses (60%), sheep (50%), food poisoning outbreaks (20%), sludge (21%) and poultry (10%). The <italic>cpb2 </italic>gene was not detected in any of the isolates from roedeer or wild birds. The <italic>cpe </italic>gene was found in all isolates originating from food poisoning outbreaks and in one isolate each from horse and roedeer. Both <italic>cpb2 </italic>and <italic>cpe </italic>were found in three isolates originating from outbreaks of food poisoning.</p></sec><sec><title>PFGE</title><p>Of the 101 isolates of <italic>C. perfringens </italic>examined in this study, 88 were successfully characterized by PFGE after genomic DNA digestion with <italic>Sma</italic>I (Table <xref ref-type="table" rid="T1">1</xref>, Figure <xref ref-type="fig" rid="F1">1</xref>). The <italic>Sma</italic>I PFGE patterns of some of these isolates are shown in Figure <xref ref-type="fig" rid="F2">2</xref>. The genetic relatedness of the isolates ranged from 57 to 100% and 56 distinct PFGE profiles were observed. Cluster analysis of PFGE data showed no apparent relationship between the source of the isolate and its PFGE profile. However, the isolates originating from food poisoning outbreak formed distinct clusters in the dendrogram (Figure <xref ref-type="fig" rid="F1">1</xref>). Altogether, 17 PFGE profiles with indistinguishable patterns were found. Most of the isolates having PFGE profiles with indistinguishable patterns were from the same outbreak of disease. All of the eight food isolates clustering with indistinguishable patterns originated from the same outbreak of food poisoning. The poultry and porcine isolates clustering with indistinguishable patterns were also isolated from the same outbreaks. Two of the PFGE profiles with indistinguishable patterns contained isolates with no obvious epidemiological correlation. Of the 28 <italic>cpb2</italic>-positive isolates, 22 were successfully characterized by PFGE. The <italic>Sma</italic>I PFGE patterns of these isolates are shown in Figure <xref ref-type="fig" rid="F3">3</xref>. The overall genetic relatedness of the <italic>cpb2</italic>-positive isolates was 63%. All the 10 isolates originating from pigs cluster together with a genetic relatedness of 72%. This cluster was subdivided into four subclusters with 100% genetic relatedness in each cluster.</p></sec><sec><title>Sequence data analysis</title><p>In the phylogenetic tree (Figure <xref ref-type="fig" rid="F4">4</xref>) two main groups of the partially sequenced <italic>cpb2 </italic>gene (the distance matrix comprised 425 nucleotide positions, corresponding to positions 282 to 706 in the consensus <italic>cpb2 </italic>sequence) were observed, I and II, each subdivided to three subclusters, a, b, c. A total of six groups could be identified (Figure <xref ref-type="fig" rid="F4">4</xref>, Table <xref ref-type="table" rid="T2">2</xref>). Groups I and II are mutually related, with a sequence similarity of 73.8%. All isolates in our study were found in group I. The sequence similarities within this group varied between 93.3% and 100%. Our <italic>cpb2 </italic>sequences from porcine (Ia) and from non-porcine (Ib) cluster with the <italic>cpb2 </italic>sequences deposited by Vilei <italic>et al </italic>[<xref ref-type="bibr" rid="B12">12</xref>]. The three <italic>cpb2 </italic>sequences from food and the horse isolate AN 5036/01 was found in group Ic. Most of the isolates in group II were isolated in the USA, while most of those in group I were isolated in Europe.</p><p>Sequencing of the 16S rRNA gene revealed a very high similarity among the 11 isolates analysed. As shown in Table <xref ref-type="table" rid="T3">3</xref>, altogether 18 polymorphic sites were detected in the 16S rRNA region between nucleotide no. 91 and 1401 (based on the consensus sequence of the 16S rRNA gene in the 10 rRNA operons of <italic>C. perfringens </italic>strain 13, <italic>rrnA </italic>– <italic>rrnI</italic>). Seven of the polymorphisms found in our isolates were also found in <italic>C. perfringens </italic>strain 13 [<xref ref-type="bibr" rid="B2">2</xref>]. At position 154 of the 16S rRNA genes, one nucleotide difference was observed; at this position, isolates from food poisoning outbreaks (AN 4279/01 455/99) had a T, whereas all others had a C. This was the only nucleotide difference involving all operons that was observed in the sequenced isolates.</p></sec></sec><sec><title>Discussion</title><sec><title>Prevalence of the <italic>cpb2 </italic>and <italic>cpe </italic>genes</title><p>The high prevalence of the <italic>cpb2 </italic>gene in isolates from pigs and horses is consistent with other studies reporting a high prevalence of <italic>cpb2 </italic>in pigs and horses suffering from gastrointestinal diseases [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. The distribution of <italic>cpe </italic>in isolates of animal origin is low, which also tallies with data from other studies [<xref ref-type="bibr" rid="B27">27</xref>-<xref ref-type="bibr" rid="B29">29</xref>]. Isolates from food poisoning outbreaks typically carry a chromosomal <italic>cpe </italic>gene [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]; in this study three of these isolates also carried the <italic>cpb2 </italic>gene. In this study three of the isolates from food poisoning outbreaks also carried the <italic>cpb2 </italic>gene. Isolates originating from humans with gastrointestinal diseases, carrying both <italic>cpb2 </italic>and <italic>cpe</italic>, have been described recently [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>].</p></sec><sec><title>PFGE analysis</title><p>The aim of this study was to elucidate the general genetic diversity of <italic>C. perfringens </italic>type A isolated from a variety of sources in Sweden and Norway. In the present study, any band difference between two PFGE types was considered sufficient to distinguish between two different PFGE types. The results obtained by PFGE reveal a wide genetic diversity and no definite relationship between the source of the isolate and the positions in the dendrogram can be established.</p><p>The wide genetic diversity revealed in this study was not unexpected, because of the wide diversity of the isolates analysed. Previous studies on <italic>C. perfringens </italic>by PFGE have shown that isolates from the same food poisoning outbreak have a very similar pattern [<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref>] and that <italic>C. perfringens </italic>isolated from retail food showed a high genetic variation [<xref ref-type="bibr" rid="B17">17</xref>]. This has also been observed in poultry where isolates from diseased birds showed a similar pattern, whereas the genetic diversity was considerable in isolates from healthy birds [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. Those observations are consistent with our results, where we can see that PFGE profiles with indistinguishable patterns in almost all cases contain isolates having a known epidemiological connection. Comparison of the PFGE patterns of the <italic>cpb2</italic>-positive isolates revealed a clonal relationship among the porcine isolates (group Ia), which cluster together in the dendrogram (Figure <xref ref-type="fig" rid="F3">3</xref>). However, no clonal relationship could be established for isolates belonging to groups Ib and Ic. The <italic>cpb2 </italic>gene is known to be located on several low copy number plasmids [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. The genetic diversity of <italic>cpb2</italic>-positive isolates is therefore, not surprising, as PFGE mainly reflects the chromosomal diversity.</p><p>In this study 13% of the isolates were non-typable due to DNA degradation. This problem has been reported previously [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. It is a disadvantage when PFGE is used as a subtyping method for clostridial species. However, problems with DNA degradation have not been reported previously when analysing poultry isolates by PFGE [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. As in this study none of the isolates from poultry, horses or pigs were found degraded, PFGE is a very suitable method for epidemiological investigation of enteric diseases caused by <italic>C. perfringens</italic>. The 16S rRNA gene was sequenced for three of the isolates that were found degraded. Neither DNA degradation nor the generally wide genetic diversity was reflected in the 16S rRNA sequences, which were very similar. The use of <italic>Salmonella enterica </italic>subsp. <italic>enterica </italic>serotype Braenderup, (<italic>Salmonella </italic>Braenderup H9812), digested with <italic>Xba</italic>I as a size marker made it easier to normalize tiff images, than by using a commercially available weight marker, thanks to stable levels of DNA in the prepared plugs [<xref ref-type="bibr" rid="B33">33</xref>].</p><p>In our opinion, PFGE is a reliable and robust method that can be used in combination with epidemiological data to establish <italic>C.perfringens </italic>as the etiological agent whether in food-borne outbreaks or in outbreaks of enteric disease in domestic animals.</p></sec><sec><title>Sequence analysis of the <italic>cpb2 </italic>gene</title><p>Sequence analysis of the <italic>cpb2 </italic>gene (Figure <xref ref-type="fig" rid="F4">4</xref>) indicated the existence of two evolutionary differing populations based on 425 nucleotide positions, corresponding to positions 282 to 706 in the consensus <italic>cpb2 </italic>sequence. A relatively high sequence difference of 26.2% showed that <italic>cpb2 </italic>evolved into two different variants (I and II). The isolates sequenced in this study all belonged to group I and were divided into three sub-clusters: porcine isolates (Ia), animal isolates of non-porcine origin (Ib), and isolates from food poisoning outbreaks (Ic). One of the horse isolates AN 5036/01 clustered together with the food isolates. Most of the group I <italic>cpb2</italic>-positive isolates were of European origin [<xref ref-type="bibr" rid="B12">12</xref>], while most <italic>cpb2</italic>-positive isolates from group II had an American origin [<xref ref-type="bibr" rid="B11">11</xref>]. However, isolates containing <italic>cpb2 </italic>isolated from humans with a gastrointestinal disease were distributed in both groups (Ic and IIa), irrespective of origin [<xref ref-type="bibr" rid="B15">15</xref>]. Groups I and II both contained isolates capable of expressing CPB2 and also isolates that express non-detectable levels of beta2 toxin [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. Jost <italic>et al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>] identified a <italic>cpb2 </italic>gene that was present in most non-porcine isolates. The <italic>cpb2 </italic>sequences deposited by Jost <italic>et al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>] were all affiliated to group II (IIa, IIb or IIc). Isolates from European animals carrying the <italic>cpb2 </italic>gene were all affiliated to group I. It is interesting that none of the <italic>cpb2</italic>-positive isolates from animals in Europe carried the <italic>cpb2 </italic>gene, which are found in group II.</p></sec><sec><title>Sequence analysis of the 16S rRNA gene</title><p>The reason for sequencing the 16S rRNA gene was to ascertain whether the diversity found by PFGE analysis was reflected in the 16S rRNA gene. Isolates that were degraded and those found in different clusters in the PFGE dendrogram showed a very high sequence similarity. However, we found 18 positions with polymorphisms among our investigated isolates (Table <xref ref-type="table" rid="T3">3</xref>). These polymorphisms are caused by the fact that <italic>C. perfringens </italic>carries 10 rRNA operons and sequence differences exist between them. Some of these polymorphisms were isolate specific. Seven of those found by us were also found by Shimizu <italic>et al</italic>. [<xref ref-type="bibr" rid="B2">2</xref>] who analysed all the 10 rRNA operons of <italic>C. perfringens </italic>strain 13. The polymorphisms found in the <italic>C. perfringens </italic>16S rRNA genes reflect certain diversity. In a few other bacterial species polymorphisms have been used for subtyping [<xref ref-type="bibr" rid="B34">34</xref>], but due to the 10 rRNA operons in <italic>C. perfringens</italic>, it is more difficult to use the polymorphism pattern for subtyping.</p></sec></sec><sec><title>Conclusion</title><p>In this study a wide genetic diversity of <italic>C. perfringens </italic>was found when isolates from eight different sources were analysed by PFGE. Degradation of DNA was observed in 13% of the isolates investigated. The considerable diversity found by PFGE was not reflected in the 16S rRNA sequences, which were very similar. As expected, <italic>C. perfringens </italic>isolates from the same outbreak seemed to be similar genetically, while isolates with no obvious epidemiological connection differed more noticeably. The groups with the highest genetic relatedness were isolates from food-borne outbreaks and pig isolates carrying the <italic>cpb2 </italic>gene. Isolates from the other sources were more widely dispersed in the dendrogram. Two genetically differing populations of the partially sequenced <italic>cpb2 </italic>gene were found by sequence analysis. Furthermore, isolates causing enteric diseases in humans and other animals seem not to have a strong genetic relatedness, based on PFGE analysis. We conclude that PFGE is a reliable and robust method for genotyping of <italic>C. perfringens </italic>isolates.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Bacterial isolates and growth conditions</title><p>Altogether 95 isolates of <italic>C. perfringens </italic>type A were obtained from eight different sources: poultry (<italic>n </italic>= 20), pigs (<italic>n </italic>= 12), horses (<italic>n </italic>= 10), sheep (<italic>n </italic>= 6), roedeer (<italic>n </italic>= 12), wild birds (<italic>n </italic>= 6), food poisoning outbreaks (<italic>n </italic>= 15) and environmental samples (<italic>n </italic>= 14) (Table <xref ref-type="table" rid="T1">1</xref>). The isolates were collected between 1994 to 2003 (Table <xref ref-type="table" rid="T1">1</xref>), all isolates from sludge were collected 2002. One <italic>C. perfringens </italic>type strain CCUG 1795 (A) and five reference strains, CIP 106526 (A), CCUG 2035 (B), CCUG 2036 (C), CCUG 2037 (D) and CCUG 44727 (E), were also included in the study. CCUG strains were obtained from the Culture Collection of the University of Gothenburg, Sweden; the CIP strain was obtained from the Culture Collection of Institute Pasteur, Paris, France. The <italic>C. perfringens </italic>isolates were identified as type A by applying standard biochemical tests and multiplex-PCR [<xref ref-type="bibr" rid="B22">22</xref>]. The isolates were stored at -70°C. Thawed isolates were grown on fastidious anaerobe agar (FAA) (LabM, Bury, Lancashire, England) with 10% defibrinated horse blood and incubated in anaerobic jars at 37°C.</p></sec><sec><title>PCR</title><p>The four major toxin genes <italic>plc</italic>, <italic>cpb1</italic>, <italic>iap </italic>and <italic>etx </italic>were detected by a modified version of the multiplex PCR assay of Engström <italic>et al</italic>. [<xref ref-type="bibr" rid="B22">22</xref>]. Template DNA (2 μl), prepared by the direct lysis method of Herholz <italic>et al</italic>. [<xref ref-type="bibr" rid="B10">10</xref>] was added to a 50 μl reaction mixture, with the following reagents: 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 2.5 mM MgCl<sub>2</sub>, 50 nM of each deoxynucleotide, 1 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, USA), 50 nM of each <italic>plc </italic>(α-toxin) primer, 25 nM of each <italic>cpb1 </italic>primer, 100 nM of each <italic>etx </italic>primer and 25 nM of each <italic>iap </italic>primer. The thermocycling (incubations for 1 min at 94°C, 55°C and 72°C, respectively, repeated 35 times) was preceded by incubation for 10 min at 94°C. The presence of the CPB2 gene (<italic>cpb2</italic>) and enterotoxin gene (<italic>cpe</italic>) was also determined. <italic>cpb2 </italic>primers (250 nM) from Herholz <italic>et al</italic>. [<xref ref-type="bibr" rid="B10">10</xref>] and <italic>cpe </italic>primers (50 nM) from Kadra <italic>et al</italic>. [<xref ref-type="bibr" rid="B35">35</xref>] were used in a duplex PCR. The conditions were as in the multiplex PCR, except for the annealing temperature, which was 59°C. The amplicons were analysed by electrophoresis on a 1.5% agarose gel according to standard procedures.</p></sec><sec><title>PFGE</title><p>The isolates analysed by PFGE are listed in Table <xref ref-type="table" rid="T1">1</xref>. The PFGE protocol of Lukinmaa <italic>et al</italic>. [<xref ref-type="bibr" rid="B20">20</xref>] was followed in all essentials. DNA was digested with <italic>Sma</italic>I. Electrophoresis was performed at 6 V/cm with 2.0% AgaroseNA agar (Amersham Biosciences, Uppsala, Sweden) by using the CHEF-DR II system (Bio-Rad Laboratories, Richmond, Calif, USA). Running conditions for <italic>Sma</italic>I-digested DNA were 0.5 to 40 s for 20 h. Lambda ladder with a size range of 0.13 to 194 kb (Low Range PFG Marker, New England Biolabs Inc.) and <italic>Salmonella </italic>Braenderup (H9812) digested with <italic>Xba</italic>I, which is described in greater detail by Hunter <italic>et al</italic>. [<xref ref-type="bibr" rid="B33">33</xref>], were used as molecular weight standards. The DNA bands were visualized on a UV transilluminator and Polaroid photographs of the gels were scanned and images in tiff file format were imported into GelCompar II (Applied Maths, Kortrijk, Belgium). Degrees of similarity between isolates were calculated with 1.4% tolerance and 0.5% optimization by applying the band-based Dice similarity coefficient. Clustering analysis was performed with the unweighted pair group method (UPGMA), by average linkages.</p></sec><sec><title>Sequence analysis of the 16S rRNA genes</title><p>The 16S rRNA genes of the isolates were amplified with primers [<xref ref-type="bibr" rid="B36">36</xref>] suitable for members of the phylum <italic>Firmicutes </italic>(Table <xref ref-type="table" rid="T4">4</xref>). The amplicons were used for cycle sequencing with labelled terminators (Big Dye; Applied Biosystems, Foster City, Calif, USA) and with the sequencing primers listed in Table <xref ref-type="table" rid="T4">4</xref>. The sequencing products were analysed with an ABI Prism 3100 genetic analyser (Applied Biosystems) and contigs were generated with the Contig Express program included in the Vector NTI Suite (InforMax, Bethesda, Md, USA). The contigs were edited manually with the Genetic Data Environment software [<xref ref-type="bibr" rid="B37">37</xref>] before further analysis and deposition in GenBank.</p></sec><sec><title>Sequence analysis of the <italic>cpb2 </italic>genes</title><p>The <italic>cpb2 </italic>gene of the 28 <italic>cpb2</italic>-positive isolates was amplified with the β2 primers (250 nM) from Herholz <italic>etal</italic>. [<xref ref-type="bibr" rid="B10">10</xref>] and these primers were also used in the cycle sequencing reaction, as described earlier. The <italic>cpb2 </italic>sequences determined in this work were aligned with sequences retrieved from GenBank by the AlignX program in the Vector NTI Suite. The alignment was checked with the Genetic Data Environment software [<xref ref-type="bibr" rid="B37">37</xref>]. A phylogenetic tree comprising 88 <italic>cpb2 </italic>sequences was constructed by the neighbour-joining method [<xref ref-type="bibr" rid="B38">38</xref>] from a distance matrix that was corrected for multiple substitutions at single locations by the two-parameter method [<xref ref-type="bibr" rid="B39">39</xref>]. A representative of cluster Ib was chosen as outgroup. Only one representative for each subgroup was included in the final tree. The distance matrix comprised 425 nucleotide positions, corresponding to positions 282 to 706 in the <italic>cpb2 </italic>sequence of <italic>C. perfringens</italic>, isolate P762/97 [Gen Bank:<ext-link ext-link-type="gen" xlink:href="AJ537531">AJ537531</ext-link>]. Consenus alignments are shown in <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>.</p></sec><sec><title>Nucleotide sequence accession numbers</title><p>The sequences of the <italic>cpb2 </italic>gene of the 28 <italic>cpb2</italic>-positive isolates have been deposited in GenBank (National Center for Biotechnology, Bethesda, Md, USA) under accession numbers [GenBank:<ext-link ext-link-type="gen" xlink:href="DQ201544">DQ201544</ext-link> – GenBank:<ext-link ext-link-type="gen" xlink:href="DQ201571">DQ201571</ext-link>]. The 16S rRNA sequences have also been deposited in GenBank under accession numbers [GenBank:<ext-link ext-link-type="gen" xlink:href="DQ196132">DQ196132</ext-link> – GenBank:<ext-link ext-link-type="gen" xlink:href="DQ196142">DQ196142</ext-link>].</p></sec></sec><sec><title>Authors' contributions</title><p>AJ, KEJ, AA, VB, BEE participitated in the discussions on the study design, the collection of isolates, analysis and interpretation of the data, and in the writing of the manuscript. AJ and AA carried out the analysis and interpretation of PFGE data. Analysis and interpretation of sequence data of the <italic>cpb2 </italic>gene and 16S rRNA gene were carried out by AJ, KEJ and EB. AJ was the principal author of the manuscript. All authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Consensus alignment of <italic>cpb2 </italic>clusters</p></caption><media xlink:href="1471-2180-6-47-S1.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Identification of DNA sequence variation in <italic>Campylobacter jejuni </italic>strains associated with the Guillain-Barré syndrome by high-throughput AFLP analysis
|
<sec><title>Background</title><p><italic>Campylobacter jejuni </italic>is the predominant cause of antecedent infection in post-infectious neuropathies such as the Guillain-Barré (GBS) and Miller Fisher syndromes (MFS). GBS and MFS are probably induced by molecular mimicry between human gangliosides and bacterial lipo-oligosaccharides (LOS). This study describes a new <italic>C. jejuni</italic>-specific high-throughput AFLP (htAFLP) approach for detection and identification of DNA polymorphism, in general, and of putative GBS/MFS-markers, in particular.</p></sec><sec><title>Results</title><p>We compared 6 different isolates of the "genome strain" NCTC 11168 obtained from different laboratories. HtAFLP analysis generated approximately 3000 markers per stain, 19 of which were polymorphic. The DNA polymorphisms could not be confirmed by PCR-RFLP analysis, suggesting a baseline level of 0.6% AFLP artefacts. Comparison of NCTC 11168 with 4 GBS-associated strains revealed 23 potentially GBS-specific markers, 17 of which were identified by DNA sequencing. A collection of 27 GBS/MFS-associated and 17 enteritis control strains was analyzed with PCR-RFLP tests based on 11 of these markers. We identified 3 markers, located in the LOS biosynthesis genes cj1136, cj1138 and cj1139c, that were significantly associated with GBS (P = 0.024, P = 0.047 and P < 0.001, respectively). HtAFLP analysis of 13 highly clonal South African GBS/MFS-associated and enteritis control strains did not reveal GBS-specific markers.</p></sec><sec><title>Conclusion</title><p>This study shows that bacterial GBS markers are limited in number and located in the LOS biosynthesis genes, which corroborates the current consensus that LOS mimicry may be the prime etiologic determinant of GBS. Furthermore, our results demonstrate that htAFLP, with its high reproducibility and resolution, is an effective technique for the detection and subsequent identification of putative bacterial disease markers.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Godschalk</surname><given-names>Peggy CR</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Bergman</surname><given-names>Mathijs P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Gorkink</surname><given-names>Raymond FJ</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Simons</surname><given-names>Guus</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>van den Braak</surname><given-names>Nicole</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Lastovica</surname><given-names>Albert J</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Endtz</surname><given-names>Hubert P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Verbrugh</surname><given-names>Henri A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>van Belkum</surname><given-names>Alex</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Microbiology
|
<sec><title>Background</title><p>The Guillain-Barré syndrome (GBS) is the most frequent form of acute immune-mediated neuropathy. The Miller Fisher syndrome (MFS) is a variant of GBS, affecting mainly the eye muscles [<xref ref-type="bibr" rid="B1">1</xref>]. A respiratory or gastro-intestinal infection preceding the neurological symptoms is reported by nearly two-thirds of all patients [<xref ref-type="bibr" rid="B2">2</xref>]. The most frequently identified infectious agent is <italic>Campylobacter jejuni</italic>, which is also the predominant cause of bacterial diarrhoea worldwide [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. The neuropathy is probably induced by molecular mimicry between gangliosides in nerve tissue and lipo-oligosaccharides (LOS) on the <italic>Campylobacter </italic>cell surface [<xref ref-type="bibr" rid="B5">5</xref>]. This structural resemblance leads to a cross-reactive immune response causing neurological damage. Biochemical and serological studies have revealed that many <italic>C. jejuni </italic>strains express ganglioside-like structures in their LOS [<xref ref-type="bibr" rid="B6">6</xref>]. However, not all strains expressing ganglioside mimics induce GBS. It is estimated that only 1 in every 1000–3000 <italic>C. jejuni </italic>infections is followed by GBS [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>], which suggests that additional bacterial determinants and/or host-related factors are important as well.</p><p>Many researchers have studied collections of GBS-associated and control "enteritis-only" strains in search of GBS-specific microbial features. Several reports describe an overrepresentation of specific Penner (heat stable, HS) serotypes among GBS-associated strains from certain geographical areas [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. The HS:19 and HS:41 serotypes are the predominant serotypes preceding GBS in Japan and South Africa, respectively [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Because HS:19 and HS:41 strains represent a clonal population [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>], the observed overrepresentation of these serotypes does not imply that the determinant of the Penner serotyping system, the capsular polysaccharide, is involved in the pathogenesis of GBS [<xref ref-type="bibr" rid="B13">13</xref>]. In addition, this phenomenon is not seen in other regions, where GBS-associated strains are genetically heterogeneous [<xref ref-type="bibr" rid="B14">14</xref>]. Various molecular typing techniques have been used in search of GBS-specific features in <italic>C. jejuni</italic>, such as <italic>flaA</italic>-PCR-restriction fragment length polymorphism (RFLP), pulsed field gel electrophoresis (PFGE), randomly amplified polymorphic DNA (RAPD) analysis, ribotyping, amplified fragment length polymorphism (AFLP) analysis and multi locus sequence typing (MLST), but none of these have identified GBS-specific markers [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B17">17</xref>]. Very recently, Leonard et al. also failed to detect GBS-specific features by the use of an open reading frame (ORF)-specific <italic>C. jejuni </italic>DNA microarray [<xref ref-type="bibr" rid="B18">18</xref>]. However, this array was based on the genome sequence of strain NCTC 11168 and ORFs that are not present in this strain will not be detected. In addition, possible GBS-factors other than those relating to presence or absence of certain genes will not be detected using this approach. Based on the molecular mimicry hypothesis, other researchers focused on genes involved in LOS biosynthesis and found significant associations with GBS [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B22">22</xref>]. However, these associations are not absolute and the question remains whether other GBS-specific microbial factors, either LOS-related or not, may exist.</p><p>Comparative genomics technology facilitates genetic marker identification but not all methods may be equally suited for high-throughput marker searches. Molecular typing techniques for <italic>Campylobacter </italic>strains differ in sensitivity and the overall coverage of genome regions screened. For the detection of specific disease markers it is desirable to use a technique that screens diversity in the overall genome with a very high resolution. MLST and <italic>flaA </italic>PCR-RFLP analyze restricted parts of the genome and are not suitable for the detection of additional GBS-markers [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref>]. PFGE is based on digestion of genomic DNA with a rare cutting restriction enzyme and only large insertions or deletions and mutations in the restriction sites will be detected [<xref ref-type="bibr" rid="B14">14</xref>]. PFGE patterns normally display between 10 and 20 fragments, which covers 120 nucleotides when a six nucleotide restriction enzyme recognition sequence is involved. AFLP analysis is considerably more sensitive than PFGE for the detection of DNA sequence polymorphism. In a conventional AFLP analysis, the use of two restriction enzymes and a primer pair with 1 or 2 selective nucleotides leads to a DNA fingerprint pattern consisting of approximately 50–80 fragments per strain. This approach physically covers in the order of 600–1000 nucleotides of the total genome for sequence polymorphism [<xref ref-type="bibr" rid="B25">25</xref>]. Even the use of two restriction enzymes in PFGE will not make up for the difference observed under a single AFLP reaction. As indicated above, DNA microarrays cover the full genome but will only detect differences in the presence of known genes. Recently, we described a new high throughput AFLP (htAFLP) approach for the identification of DNA polymorphism in <italic>Mycobacterium tuberculosis </italic>[<xref ref-type="bibr" rid="B26">26</xref>]. This method has the capacity to detect mutations in more than 30,000 nucleotides scattered throughout the genome, depending on the number of restriction enzymes and primer pairs used. The choice of primers and restriction enzymes is crucial and selection of these requires close attention. A wrong choice may lead to crowding of amplified fragments, caused by limiting (limited?) resolution of the gelsystem. Correct, computer-mediated comparison of AFLP fingerprints may then be compromised. However, especially the enhanced resolution makes htAFLP an excellent candidate technique for the detection of potential disease-associated markers.</p><p>The main objective of the current study was to search an elaborate collection of C. jejuni stains isolated from GBS patients for genetic markers associated with bacterial neuropathogenicity. To this aim, we developed htAFLP for <italic>C. jejuni</italic>. We analyzed six isolates of strain NCTC 11168, the "genome" strain, obtained from different laboratories, with the aim to detect base-level polymorphism introduced by sub-culturing or storage. In search of potential GBS-specific markers, we compared the NCTC 11168 AFLP patterns with those of four GBS-associated strains. In addition, we analyzed a collection of highly clonal GBS-associated and control strains from South Africa. Potential GBS-specific htAFLP markers were further identified by DNA sequencing and PCR-RFLP tests were developed. These PCR-RFLP tests were used to screen a larger collection of GBS-associated and control strains for confirmation of the potential GBS markers.</p></sec><sec><title>Results</title><sec><title>Detection and identification of DNA sequence polymorphism in NCTC 11168 strains of diverse origin</title><p>HtAFLP analysis of <italic>C. jejuni </italic>was performed with one enzyme combination. Genomic DNA was digested with MboI and DdeI and the restricted DNA was amplified by using all 64 possible combinations of +1/+2 selective primer pairs. This resulted in the generation of approximately 3000 fragments per strain. The average fragment size was 243 basepairs (bp), ranging from 46 to 613 bp. Comparative htAFLP analysis of the six NCTC 11168 isolates revealed 19 polymorphic fragments. After excision from the gel, these fragments were amplified and the DNA sequences were determined. BLAST analysis of the DNA sequences resulted in the identification of 13/19 polymorphisms, which were spread throughout the genome (Table <xref ref-type="table" rid="T1">1</xref>). For the other polymorphic fragments, repeated amplification and sequencing failed to generate DNA sequences of sufficient quality for BLAST analysis.</p></sec><sec><title>Validation of NCTC 11168 polymorphism with a PCR-RFLP approach</title><p>To verify whether the htAFLP-polymorphic fragments represent true DNA sequence polymorphism, we analyzed the six NCTC 11168 isolates by PCR-RFLP analysis. An AFLP polymorphism that is based on mutations in the restriction site will result in a polymorphic RFLP pattern, whereas insertions or deletions in the AFLP fragment will result in size differences between the PCR products. Based on the BLAST hit sequences (Table <xref ref-type="table" rid="T1">1</xref>), PCR tests for amplifying twelve marker fragments and their flanking regions were developed. Fragments of correct size were produced with all primer sets and for all isolates (results not shown). Next, PCR products of the six NCTC 11168 strains were digested in separate reactions with the AFLP restriction enzymes (results not shown). None of the digests showed polymorphic RFLP patterns. Thus, restriction site polymorphism or insertions/deletions could not be confirmed as cause of the observed AFLP polymorphisms. Nine out of twelve (75%) digests with <italic>DdeI </italic>and six of twelve (50%) digests with <italic>MboI </italic>resulted in RFLP patterns as expected based on the NCTC 11168 DNA sequence. An AFLP polymorphism can also be the result of a mutation in the nucleotides complementary to the selective primer nucleotides. Because all 64 possible combinations of the +1/+2 primer pairs were used in this htAFLP, such a mutation would be expected to result in an additional polymorphism, with the same fragment length and localization on the genome, in the AFLP pattern generated with a different primer pair. We did not detect such complementary polymorphisms in the NCTC 11168 comparison. In conclusion, the polymorphic AFLP bands observed in the NCTC 11168 comparison, representing approximately 0.6% of all bands, probably represent the low "background noise" of the htAFLP technique.</p></sec><sec><title>Comparison of NCTC 11168 with GBS-associated strains for the detection and identification of potential markers for GBS</title><p>For the detection of putative markers for the Guillain-Barré syndrome, we compared the NCTC 11168 isolates with strain GB11, which was isolated from a GBS patient. Strain NCTC 11168 was originally isolated from a patient with gastroenteritis without neurological symptoms. It had previously been shown that GB11 and NCTC 11168 are genetically closely related [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B27">27</xref>]. Because of this relatedness, these strains are very suitable for the detection of potential GBS-specific markers. HtAFLP analysis of NCTC 11168 and GB11 generated 241 putative GBS markers. Overall, 156 of 241 markers could be successfully identified with DNA sequencing and BLAST analysis. A proportion of the marker fragments that were excised from the gel could not be reliably reamplified and were excluded from the analysis. Although BLAST searches were conducted against all DNA sequences in the Pubmed database, the most significant homology for all AFLP DNA sequences was with <italic>C. jejuni </italic>DNA sequences. To further reduce this excessive number of putative GBS markers, we analyzed three additional GBS-associated strains, not related to the NCTC 11168 and GB11 strains (Cura7, Cura276 and 260.94; See <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>). This reduced the number of successfully sequenced putative GBS markers to 17 (Table <xref ref-type="table" rid="T2">2</xref>). Three of these markers were located in the LOS biosynthesis gene locus. Other genes encoded a putative periplasmic protein and proteins involved in signal transduction, metabolism, transport, binding, amino acid biosynthesis, fatty acid biosynthesis and DNA replication. Three genes were of unknown function. Markers 5 and 6 displayed distinct restriction site polymorphism concordant with the AFLP polymorphism. These markers contained largely overlapping DNA sequences and showed a complementary pattern of presence and absence in the GBS-associated and control strains (Table <xref ref-type="table" rid="T2">2</xref>). Comparison of the DNA sequences revealed that these markers were based on one DNA polymorphism: the presence of an additional restriction site in the GBS strains due to a point mutation (Figure <xref ref-type="fig" rid="F1">1</xref>).</p></sec><sec><title>Screening of a large strain collection for potential GBS markers by PCR-RFLP analysis</title><p>After htAFLP analysis of five strains to identify potential GBS markers, we developed PCR-RFLP tests to screen a large collection of 27 GBS/MFS-associated and 17 control enteritis strains for the presence of these markers (for a survey of these strains see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>). The strains used in the htAFLP analysis were also included in the PCR-RFLP analysis. One randomly selected NCTC 11168 isolate was included. Based on the BLAST hit sequences (Table <xref ref-type="table" rid="T2">2</xref>), PCR tests for amplifying 11/17 marker fragments and their flanking regions were developed (See <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). Because markers 5 and 6 represented the same DNA polymorphism, they were included in one PCR test. Fragments of correct, expected size were produced with all primer sets. For 5/11 markers a PCR product was absent in a variable proportion of strains (Table <xref ref-type="table" rid="T3">3</xref>), probably due to primer site sequence heterogeneity. For example, we have observed previously that gene <italic>cj1136</italic>, part of the LOS biosynthesis gene locus and containing marker 7, shows a large degree of DNA sequence heterogeneity between strains (P. Godschalk, unpublished results). This leads to primer mismatches and absence of PCR products in a proportion of strains. In 2/10 PCR tests (markers 7 and 8), the htAFLP GBS-associated strains could be distinguished from strain NCTC 11168 through the pattern of presence and absence of PCR products. PCR products for marker 7 were absent in the GBS-associated strains used in the htAFLP and present in NCTC 11168, which seemed to be in contrast with the observation that the original AFLP fragment of marker 7 was present in the GBS strains and absent in NCTC 11168. However, this apparent inconsistency can be explained by the fact that the primer sequences of the PCR test were based on the NCTC 11168 DNA sequence. For marker 7, a PCR product was seen significantly more frequently in control enteritis strains (5/27 (18.5%) GBS/MFS strains versus 9/17 (52.9%) control enteritis strains, <italic>P </italic>= 0.024). Next, we subjected the PCR products to a combined digestion with the AFLP restriction enzymes (Table <xref ref-type="table" rid="T3">3</xref>). In 4/10 PCR tests (markers 3, 5/6, 9 and 13), the RFLP types were concordant with the AFLP analysis i.e. the htAFLP GBS-associated strains shared the same RFLP type whereas NCTC 11168 displayed a different RFLP type. In 3/10 PCR tests (markers 11, 12 and 14), the htAFLP GBS-associated strains did not have identical RFLP types (and for marker 14 there was no PCR product in two GBS strains), but these RFLP types were also different from that of NCTC 11168. This is not necessarily in contrast with the htAFLP results, because different RFLP types among the htAFLP GBS-associated strains may be due to heterogeneity in the flanking regions of the AFLP fragment. For marker 2, the RFLP types of the htAFLP strains were not concordant with the htAFLP polymorphism: the NCTC 11168 and GB11 RFLP types were the same. RFLP types 3 and 4 of marker 9, located in gene cj1139c, were only detected in GBS/MFS-associated strains (RFLP type 3 or 4 present in 15/27 (55.6 %) GBS/MFS-associated strains versus 0/17 (0%) enteritis strains, P <0.0001). Although a PCR product for marker 8, located in gene cj1138, was absent in the majority of strains, RFLP type 1 was more frequently found in enteritis strains (5/15 enteritis strains versus 1/27 GBS-associated strains, P = 0.047).</p></sec><sec><title>Analysis of South-African GBS-associated and control HS:41 strains by htAFLP</title><p>In South Africa, serotype HS:41 is over-represented among GBS-associated strains [<xref ref-type="bibr" rid="B10">10</xref>]. Previous studies have shown that HS:41 strains, both GBS-associated and controls, are highly clonal [<xref ref-type="bibr" rid="B12">12</xref>]. As expected, htAFLP of six GBS-associated, one MFS-associated and six control HS:41 strains generated very homogeneous banding patterns (results not shown). A total of forty-five AFLP polymorphisms were detected, but there were no GBS-specific markers. Interestingly, 28 AFLP polymorphisms displayed a similar pattern: fragments were present in the MFS-associated strain and two or three control enteritis strains but absent in the other strains (Figure <xref ref-type="fig" rid="F2">2</xref>). These fragments were excised and DNA sequences were determined. BLAST analysis of five DNA sequences revealed homologies with various bacterial plasmidal DNA sequences (results not shown).</p></sec></sec><sec><title>Discussion</title><p>This study describes a high-throughput AFLP approach for the detection and identification of DNA polymorphism and putative GBS-markers in <italic>C. jejuni</italic>. Previously, we showed that htAFLP is an excellent tool for assessing the population structure and the expansion of pathogenic clones in <italic>Staphylococcus aureus </italic>and for identification of genetic polymorphism in the clonal microorganism <italic>Mycobacterium tuberculosis </italic>[<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. The optimal enzyme and selective primer pair combinations are determined by <italic>in silico </italic>calculations using the whole genome DNA sequence of the target microorganism. The optimal number of AFLP fragments to be generated depends on the aim of the study. For the detection and identification of potential disease markers, such as GBS-specific markers in <italic>C. jejuni </italic>in the current study, it is desirable to screen the genome with high resolution. For this, the generation of a large number of AFLP fragments per strain is needed. Such high resolution AFLP approach limits the number of strains that can be analyzed, but this can be overcome by the subsequent analysis of a large number of strains by PCR-RFLP tests, translated from the potential markers as detected by the preceding htAFLP analysis. It is, of course, possible that a disease-specific marker is not detected by htAFLP because this approach does not result in 100% genome coverage, which can only be reached with whole genome sequencing. For <italic>C. jejuni</italic>, one enzyme combination (<italic>Mbo</italic>I and <italic>Dde</italic>I) and 64 different +1/+2 selective primer pair combinations physically covered approximately 30,000 nucleotides per strain, which represents approximately 2% of the genome.</p><p>To find out whether subculturing or storage of <italic>C. jejuni </italic>strains leads to the emergence of DNA polymorphism, we compared 6 isolates of the "genome" strain NCTC 11168 obtained from different laboratories worldwide by htAFLP analysis. The observed AFLP polymorphisms, approximately 0.6% of the fragments per strain, could not be confirmed by PCR-RFLP analysis. This indicates that the AFLP polymorphisms probably represent the low background noise of the htAFLP technique and that true DNA polymorphism could not be detected in the six NCTC 11168 isolates. The observed background noise equals a reproducibility of approximately 99.6%, which is still very high when compared to other genotyping techniques [<xref ref-type="bibr" rid="B29">29</xref>].</p><p>Recently, two groups described differences in virulence properties, i.e. the ability to colonise chickens, between different NCTC 11168 isolates that were not included in the current study [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. Full transcriptional profiling revealed expression differences for several gene families in the NCTC 11168 strains. HtAFLP analysis of the two NCTC 11168 isolates that were studied by Carrillo et al. [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>] failed to identify polymorphisms responsible for the difference in virulence properties (P. Godschalk and C. Szymanski, unpublished data). It has to be emphasized that by htAFLP still only a random proportion of the genome is screened. DNA sequence variation (such as single-nucleotide polymorphisms, SNPs) leading to biological differences may therefore not be detected.</p><p>In search of GBS-specific markers, we first compared the NCTC 11168 patterns with the AFLP patterns of the genetically related GBS-associated strain GB11. However, although NCTC 11168 was originally isolated from the faeces of a patient with gastroenteritis, we cannot exclude that NCTC 11168 can induce GBS if a patient with the right host susceptibility factors becomes infected with this strain. The only substantial but probably very important difference between NCTC 11168 and GB11 that has been found so far, is that the LOS biosynthesis gene locus strongly diverges between these strains, probably as result of a horizontal exchange event [<xref ref-type="bibr" rid="B32">32</xref>]. Comparison of NCTC 11168 with GB11 led to the detection of more than two hundred possible GBS markers, which was substantially higher than the expected background noise of 0.6%, underscoring the phylogenetic relevance of the polymorphisms. The number of possible GBS markers was reduced to 23 after adding three additional GBS-associated strains. For 17 markers, the location on the genome could be identified after DNA sequence analysis. A relatively large proportion of potential GBS markers (3/17;18%) was located in the LOS biosynthesis gene locus, whereas this locus only comprises 1% of the <italic>C. jejuni </italic>genome (1.64 Mbp). Although this may represent a true pathogenic association with GBS, it is also possible that htAFLP preferentially picked up the LOS locus because it is a highly polymorphic region. However, analysis of a larger <italic>C. jejuni </italic>strain collection by PCR-RFLP analysis showed that the three LOS-specific markers were indeed associated with GBS (marker 7, <italic>P = </italic>0.024; marker 8, <italic>P = </italic>0.047; marker 9, <italic>P <</italic>0.001). These findings are concordant with the proposed pathogenic mechanism of GBS and with previous reports that certain genes involved in LOS biosynthesis or specific nucleotide sequences within these genes occur more frequently in GBS-associated <italic>C. jejuni </italic>strains [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B22">22</xref>].</p><p>There are several possible explanations for the fact that we did not find molecular markers that are 100% specific for the Guillain-Barré syndrome. First, it is possible that truly GBS-specific features do not exist in <italic>C. jejuni</italic>. There is a broad variability in the severity and spectrum of clinical symptoms in GBS patients [<xref ref-type="bibr" rid="B33">33</xref>]. Different ganglioside mimicking structures and anti-ganglioside antibody specificities have been associated with certain clinical presentations [<xref ref-type="bibr" rid="B34">34</xref>-<xref ref-type="bibr" rid="B36">36</xref>], and therefore, <italic>C. jejuni </italic>markers may be associated with a subset of various disease entities. Because of this heterogeneity and the presumed importance of host-related factors, the existence of features in <italic>C. jejuni </italic>that are specific for GBS may be questionable. Second, a certain combination of multiple <italic>C. jejuni </italic>genes may be required for the induction of GBS. Detection of such combinations of markers ("polygenic markers") is extremely labour-intensive and cannot be achieved with the current approach. Finally, it is possible that htAFLP failed to detect GBS-specific markers because htAFLP does not accomplish 100% genome coverage.</p><p>One of the three additional GBS-associated strains mentioned above was from a collection of South-African HS:41 strains. In South Africa, the HS:41 serotype is over-represented among GBS-associated strains [<xref ref-type="bibr" rid="B10">10</xref>]. A certain feature of these strains may be responsible for their capacity to trigger GBS. HS:41 strains were found to be indistinguishable by previous genotyping studies, indicating that HS:41 strains form a genetically stable clone [<xref ref-type="bibr" rid="B12">12</xref>]. It is important to note that the enteritis-only HS:41 strains may have the same GBS-inducing capacity as the GBS-associated strains, because host-related factors are also crucial for developing GBS. We analyzed both GBS-associated and control enteritis-only HS:41 strains, as well as an MFS-associated isolate by htAFLP. Although we did not detect GBS-specific bands, we found that the MFS-associated isolate and half of the enteritis-only strains contained several additional fragments that appeared to be linked. DNA sequences of these fragments showed homologies with plasmidal DNA sequences, indicating that a subset of the HS:41 strains contained a plasmid. To our knowledge, the South African HS:41 strains we used in this study have never been analyzed for the presence of plasmids. Whether the presence of a plasmid is of importance for the virulence or neuropathogenic potential of HS:41 strains currently remains unknown, but seems unlikely based on the distribution of plasmidal DNA in the tested strains.</p></sec><sec><title>Conclusion</title><p>Previous searches for <italic>C. jejuni </italic>markers for GBS-invoking potential were unsuccessful when performed with general genotyping techniques. Some studies that focussed at specific loci or sometimes even specific genes found potential, though not absolute, GBS markers within the LOS biosynthesis genes [<xref ref-type="bibr" rid="B19">19</xref>-<xref ref-type="bibr" rid="B22">22</xref>]. We have used a method, htAFLP, that detects sequence polymorphisms in a wide, non-gene dependent scale. Theoretically approximately 2% of the total genome is covered by this approach. However, we still conclude that bacterial GBS markers are not absolute, limited in number and located in the LOS biosynthesis gene locus. This corroborates the current consensus that LOS mimicry with human gangliosides may be the prime etiologic determinant of GBS. In addition to bacterial factors, host-related factors probably play an important role in the pathogenesis of GBS as well.</p><p>Furthermore, our results demonstrate that htAFLP, with its high reproducibility and resolution, is an adequate technique for the detection and subsequent identification of putative disease and epidemiological markers. Analysis of a limited number of strains in great detail by htAFLP and subsequent screening of a large collection of strains with simple PCR-RFLP tests combines high sensitivity with the possibility to screen large groups of strains. This allows for the identification of regions of genomic instability or variability. Finally, htAFLP does not require complete genome sequences and it is not influenced by the presence of sequences not present in the genome strain(s). As such, htAFLP is the second best option, after complete sequencing of the genome of multiple strains, for the unbiased detection of genome polymorphisms associated with pathogenicity or other features of bacterial isolates from diverse clinical and environmental origin.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Bacterial strains, culture conditions and DNA isolation</title><p>The <italic>C. jejuni </italic>strains used for htAFLP analysis are described in <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>. We collected 6 isolates of the "genome" strain NCTC 11168 strains from different labs worldwide [<xref ref-type="bibr" rid="B37">37</xref>]. For the detection of potential GBS markers, we included four <italic>C. jejuni </italic>strains isolated from the diarrhoeal stools of GBS patients from different geographical areas (The Netherlands, Curaçao, South Africa). In addition, we analyzed a collection of 6 GBS-associated, 1 MFS-associated and 6 enteritis-only HS:41 strains isolated from South African patients [<xref ref-type="bibr" rid="B12">12</xref>]. After identification of potential GBS markers by htAFLP analysis of these strains, we screened a larger collection of 27 GBS/MFS-associated and 17 control strains isolated from enteritis patients with PCR-RFLP tests for the presence of these markers (See <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>). <italic>C. jejuni </italic>strains were cultured for 24–48 hours on blood agar plates in a micro-aerobic atmosphere at 37°C. DNA was isolated using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI).</p></sec><sec><title>High-throughput AFLP</title><p>AFLP analysis was performed essentially as described by Vos et al. [<xref ref-type="bibr" rid="B38">38</xref>]. The optimal enzyme and primer combinations for <italic>C. jejuni </italic>were determined using the predictive software package REcomb [<xref ref-type="bibr" rid="B39">39</xref>]. Digestion with <italic>Mbo</italic>I and <italic>Dde</italic>I (Boehringer-Mannheim, Mannheim, Germany) was combined with the ligation of a specific linker oligonucleotide pair (<italic>Mbo</italic>I: 5'-CTCGTAGACTGCGTACC-3' and 5'-GATCGGTACGCAGTCTAC-3'; <italic>Dde</italic>I: 5'-GACGATGAGTCCTGAG-3' and 5'-TNACTCAGGACTCAT-3'). Subsequently, a non-selective pre-amplification was performed using the <italic>Mbo</italic>I primer (5'-GTAGACTGCGTACCGATC-3') and <italic>Dde</italic>I primer (5'-GACGATGAGTCCTGAGTNAG-3'). The selective amplifications were performed using different linker-specific primer combinations. The <sup>33</sup>P-labeled <italic>Mbo</italic>I primer was extended with a single nucleotide (+1), whereas the <italic>Dde</italic>I primer was equipped with a 3' terminal dinucleotide (+2). These nucleotides probe sequence variation beyond that present in the restriction site itself. All 64 possible extension combinations were used. Radioactive labelling was used to enable isolation of DNA fragments from gels for post-AFLP sequencing analysis. Amplified material was analyzed on 50x30 cm slabgels and the amplimers were visualized using phosphor-imaging. Post-AFLP, gels were fixed, dried and stored at ambient temperature.</p></sec><sec><title>Marker selection and identification</title><p>Marker bands were scored using the automated interpretation software package AFLP QuantarPro (Keygene NV, Wageningen, The Netherlands), resulting in a binary table scoring marker fragment absence (0) or presence (1). Polymorphic marker fragments were validated by visual inspection of the autoradiographs. Bands differing in signal intensity were not considered to be polymorphic. A potential marker for GBS was defined as an AFLP polymorphism that discerns the GBS-associated strains from the NCTC 11168 isolates. Potential GBS marker fragments can either be present or absent in GBS-associated strains as compared to NCTC 11168.</p><p>Relevant fragments were excised from the gels and re-amplified using their matching AFLP consensus primer set without restriction site-specific +1 and +2 extension sequences attached. The amplimers were subjected to DNA sequencing using a 96-well capillary sequencing machine (MegaBACE; Amersham). For fragment identification, the DNA sequences were subjected to BLASTn and BLASTx searches through the NCBI website [<xref ref-type="bibr" rid="B40">40</xref>]. BLAST results enable genomic localization and gene annotation for the polymorphic marker fragments.</p></sec><sec><title>Development of PCR-RFLP tests</title><p>PCR-RFLP tests were developed to confirm polymorphism in the different NCTC 11168 isolates and to screen a collection of <italic>C. jejuni </italic>GBS/MFS-associated and control strains. PCR-RFLP tests could only be developed for the markers of which the localization on the <italic>C. jejuni </italic>genome was identified. Forward and reverse primers were designed (Primer Designer 4, Sci Ed Software, North Carolina) and synthesized, located approximately 50–200 bp upstream or downstream of the homologous region, respectively (Table <xref ref-type="table" rid="T2">2</xref>). Because of the wide range of melting temperatures (Tm) of the primers and the sometimes considerable differences in Tm between primers within one PCR reaction, a touch-down PCR approach was applied. The program consisted of 15 cycles of 1 min 94°C, 1 min 70°C minus 1°C for each following cycle (lowest temperature 55°C), 1 min 72°C, followed by 25 cycles of 0.5 min 94°C, 1 min Tm – 5°C and 1 min at 72°C. Tm – 5°C represents the lowest melting temperature of the two primers used in the reaction minus 5°C. This resulted in the amplification of not only the AFLP fragment itself, but also of their flanking sequences. Next, 15 μl of each PCR product was subjected to a separate or combined digestion with the restriction enzymes (1 unit/reaction) used for the AFLP (<italic>MboI </italic>and <italic>DdeI</italic>). After overnight incubation at 37°C, the digests were analyzed on 2% agarose gels. The PCR-RFLP analysis will reveal whether or not the AFLP variability was due to variation in the restriction sites (different RFLP patterns) or to insertions or deletions within the AFLP fragment (size differences in PCR products). AFLP variation due to differences in the selective extension nucleotides of the AFLP primers will not be detected using this approach.</p></sec></sec><sec><title>Abbreviations</title><p>AFLP amplified fragment length polymorphism</p><p>GBS Guillain-Barré syndrome</p><p>HS heat stable</p><p>htAFLP high-throughput AFLP</p><p>LOS lipo-oligosaccharide</p><p>MFS Miller Fisher syndrome</p><p>MLST multi locus sequence typing</p><p>ORF open reading frame</p><p>PCR-RFLP polymerase chain reaction – restriction fragment length polymorphism</p><p>PFGE pulsed field gel electrophoresis</p></sec><sec><title>Authors' contributions</title><p>PCRG analyzed and interpreted the data and drafted the manuscript. MPB designed and carried out the PCR-RFLP marker analysis. RFJG performed the htAFLP analysis and DNA sequencing. GS participated in the design of the study and htAFLP setup. NVDB collected the strains, performed DNA extractions and participated in the marker identification. AJL provided the South African <italic>C. jejuni </italic>strains and participated in the design of the study. HPE participated in the design of the study and writing of the manuscript. HAV participated in the design of the study. AVB conceived and coordinated the study and participated in writing of the manuscript. All authors critically read the manuscript and approved the final version.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p><italic>C. jejuni </italic>strains used in this study Description of data: a summary of all strains used in this study is given in this table, with strain numbers, associated disease, origin of the strain and technique(s) that were used to analyse the strains in this study.</p></caption><media xlink:href="1471-2180-6-32-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2</title><p>Primer sequences used in the PCR-RFLP analysis for the validation of potential GBS markers</p><p>Description of data: (none, title sufficiently describes data)</p></caption><media xlink:href="1471-2180-6-32-S2.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
A coding polymorphism in matrix metalloproteinase 9 reduces risk of scarring sequelae of ocular <italic>Chlamydia trachomatis </italic>infection
|
<sec><title>Background</title><p>Trachoma, an infectious disease of the conjunctiva caused by <italic>Chlamydia trachomatis</italic>, is an important global cause of blindness. A dysregulated extracellular matrix (ECM) proteolysis during the processes of tissue repair following infection and inflammation are thought to play a key role in the development of fibrotic sequelae of infection, which ultimately leads to blindness. Expression and activity of matrix metalloproteinase 9 (MMP-9), a major effector of ECM turnover, is up-regulated in the inflamed conjunctiva of trachoma subjects. Genetic variation within the <italic>MMP9 </italic>gene affects <italic>in vitro MMP9 </italic>expression levels, enzymatic activity and susceptibility to various inflammatory and fibrotic conditions.</p></sec><sec sec-type="methods"><title>Methods</title><p>We genotyped 651 case-control pairs from trachoma endemic villages in The Gambia for coding single nucleotide polymorphisms (SNPs) in the <italic>MMP9 </italic>gene using the high-throughput Sequenom<sup>® </sup>system. Single marker and haplotype conditional logistic regression (CLR) analysis for disease association was performed.</p></sec><sec><title>Results</title><p>The Q279R mutation located in exon 6 of <italic>MMP9 </italic>was found to be associated with lower risk for severe disease sequelae of ocular <italic>Chlamydia trachomatis </italic>infection. This mutation, which leads to a nonsynonymous amino-acid change within the active site of the enzyme may reduce MMP-9-induced degradation of the structural components of the ECM during inflammatory episodes in trachoma and its associated fibrosis.</p></sec><sec><title>Conclusion</title><p>This work supports the hypothesis that MMP-9 has a role in the pathogenesis of blinding trachoma.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Natividad</surname><given-names>Angels</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Cooke</surname><given-names>Graham</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Holland</surname><given-names>Martin J</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Burton</surname><given-names>Matthew J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Joof</surname><given-names>Hassan M</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Rockett</surname><given-names>Kirk</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Kwiatkowski</surname><given-names>Dominic P</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Mabey</surname><given-names>David CW</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Bailey</surname><given-names>Robin L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Medical Genetics
|
<sec><title>Background</title><p>Trachoma, a chronic keratoconjunctivitis caused by <italic>Chlamydia trachomatis</italic>, is the commonest infectious cause of blindness. The blinding complications of trachoma are due to progressive scarring of the conjunctiva (trachomatous scarring) eventually leading to in-turning of eyelashes (trichiasis) and corneal opacification. Genital <italic>C. trachomatis </italic>infection causes similar lesions in the female genital tract contributing to ectopic pregnancy and infertility. Severe and persistent inflammation triggered by repeated conjunctival infections is believed to increase the risk of pathological scarring later in life[<xref ref-type="bibr" rid="B1">1</xref>].</p><p>The mechanisms of disease pathology are not completely understood. Some evidence suggests that the dysregulated ECM proteolysis seen during the processes of tissue repair following infection and inflammation [<xref ref-type="bibr" rid="B2">2</xref>] may play a key role in the development of fibrotic sequelae of chlamydial infection in humans. In support of this hypothesis, we have recently shown that ocular <italic>C. trachomatis </italic>infection upregulates the expression of MMP-9 in the human conjunctival epithelium [<xref ref-type="bibr" rid="B3">3</xref>]. MMP-9 activity has been detected in immune cells present in the inflammatory infiltrate in conjunctival biopsy specimens from individuals with active trachoma [<xref ref-type="bibr" rid="B4">4</xref>]. In addition recent comparative studies of the role of MMP-9 in genital <italic>Chlamydia muridarum </italic>(MoPn) infection found greater MMP-9 transcription and activity during infection in those mouse strains exhibiting increased susceptibility to fibrotic sequelae following infection [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><p>Matrix metalloproteinases (MMPs) are a tightly regulated family of zinc-dependent enzymes that degrade structural proteins of the ECM and basement membranes. Among them, MMP-9 is a major effector of ECM turnover during homeostasis and pathology [<xref ref-type="bibr" rid="B7">7</xref>]. <italic>MMP9 </italic>expression is regulated at the transcriptional level in response to pro-inflammatory cytokines such as tumor necrosis factor (TNF) and interleukin 1 beta (IL-1β) [<xref ref-type="bibr" rid="B8">8</xref>]. Post-transcriptional regulation also occurs by control of activation of the secreted pro-enzyme (proMMP-9), and inhibition of proMMP-9 and MMP-9 by tissue inhibitors (TIMPs) [<xref ref-type="bibr" rid="B7">7</xref>].</p><p>A number of SNPs have been identified in regulatory and coding regions of the <italic>MMP9 </italic>gene. Some of them have been reported to affect <italic>in vitro MMP9 </italic>expression levels, enzymatic activity and susceptibility to various inflammatory and fibrotic conditions [<xref ref-type="bibr" rid="B9">9</xref>]. We tested the hypothesis that genetic variation in coding regions of <italic>MMP9 </italic>affects the risk of scarring sequelae of trachoma.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Patients</title><p>One thousand three hundred and fifteen subjects identified by clinical examination using World Health Organization (WHO) criteria were recruited from trachoma endemic villages in The Gambia. They included 651 subjects with scarring trachoma (TS), of whom 307 additionally had trichiasis (TT), and pair-matched by sex, age, ethnic group and village of residence individuals with normal eyelids. The subjects were otherwise healthy. We have previously studied and reported polymorphism at the <italic>IFNγ </italic>and <italic>IL10 </italic>loci in these subjects [<xref ref-type="bibr" rid="B10">10</xref>]. The study and its procedures were approved by the Gambia Government/MRC Ethics Committee (SCC 729/857), the Ethics committees of the London School of Hygiene and Tropical Medicine and of Oxford University, and are in accordance with the Declaration of Helsinki. Subjects diagnosed with trichiasis were offered free corrective surgery.</p></sec><sec><title>DNA extraction and SNP genotyping</title><p>Genomic DNA was isolated from either venous blood in EDTA or buccal brush samples and genotypes were determined by the Sequenom<sup>® </sup>system as previously described [<xref ref-type="bibr" rid="B10">10</xref>]. Primer sequences are available on request.</p></sec><sec><title>Analytical methods</title><sec><title>Haplotype reconstruction and frequencies</title><p>Haplotypes were inferred from population genotype data and their frequencies were estimated as previously described [<xref ref-type="bibr" rid="B10">10</xref>].</p></sec><sec><title>Association analysis</title><p>A crude analysis (crude Mantel-Haenzsel) was first carried out using chi-square testing to test for differences in allele and haplotype frequencies between cases of TS and TT with their pair-matched controls. In addition, conditional logistic regression (CLR) analysis for disease association was performed. Conditional logistic regression is the analysis of choice for dichotomous paired data (in our case paired cases and controls) where the risk estimates associated with specific attributes (e.g genotype) are required to be adjusted for potential confounders. Reference genotypes were selected to be those that were common in our study population. A CLR test for trend in the odds ratios (OR) was performed to examine dose response effects in the relationship between genotype and disease. All analyses were performed using STATA (v8.0) software.</p></sec></sec></sec><sec><title>Results</title><sec><title>Single-marker analysis</title><p>A total of four exonic SNPs of population frequency > 10% were typed at the MMP9 locus (Figure <xref ref-type="fig" rid="F1">1</xref>). The distribution of all marker genotypes among cases and controls were in Hardy-Weinberg equilibrium (data not shown). Table <xref ref-type="table" rid="T1">1</xref> shows the genotype frequencies at each locus for each phenotype and the results of CLR association analysis of matched case-control pairs. The Q279R G allele frequency was marginally higher among controls than scarred subjects (0.31 vs 0.27, OR = 0.854, 95%CI = 0.714, 1.021, p = 0.083). The Q279R G allele was present in 45.1% of TS subjects against 51.6% of controls, and in 38.7% of TT subjects compared to 52.0% of controls. In adjusted analysis, presence of the Q279R G allele was associated with a reduced risk of TS (OR = 0.74, 95%CI = 0.59, 0.94, p = 0.012), and with a greater decrease in risk for the more severe TT phenotype (OR = 0.66, 95%CI = 0.46, 0.94, p = 0.021). Although there was some evidence for a trend towards decreasing risk of TS and TT with increasing number of Q279R G alleles, it did not reach statistical significance (OR = 0.864, 95%CI = 0.727, 1.027 and OR = 0.832, 95%CI = 0.644, 1.074 for TS and TT respectively). Heterozygotes Q279R AG were at lower risk of both TS and TT (OR = 0.69, 95%CI = 0.54, 0.81, p = 0.004 and OR = 0.57, 95%CI = 0.38, 0.85 p = 0.006 respectively) (data in Table <xref ref-type="table" rid="T1">1</xref>).</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Diagram showing genotyped SNPs labelled according to the amino acid position.</p></caption><graphic xlink:href="1471-2350-7-40-1"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Genotype frequencies for the polymorphisms at the <italic>MMP9 </italic>locus in the study population and risk estimates from CLR analysis of matched case-control pairs.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left" colspan="6">Polymorphisms at the MMP9 locus in the study population</td></tr></thead><tbody><tr><td align="left">Genotype</td><td align="center" colspan="2"># of genotype (freq)</td><td align="center">crude M-H OR(95%CI)</td><td align="center">CLR-OR (95%CI)</td><td align="center">P value</td></tr><tr><td></td><td align="center">Scarring cases (TS)</td><td align="center">controls</td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">Q279R AA</td><td align="center">330(0.55)</td><td align="center">311(0.48)</td><td align="center">1.18 (1.00, 1.38)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">Q279R AG</td><td align="center">209(0.35)</td><td align="center">270(0.42)</td><td align="center">0.80 (0.67, 0.96)</td><td align="center">0.70(0.54, 0.89)</td><td align="center">0.004</td></tr><tr><td align="left">Q279R GG</td><td align="center">62(0.10)</td><td align="center">61(0.10)</td><td align="center">1.07 (0.74, 1.55)</td><td align="center">0.95(0.64,1.41)</td><td align="center">0.805</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">R574P CC</td><td align="center">450(0.73)</td><td align="center">459(0.72)</td><td align="center">0.98 (0.86, 1.12)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">R574P CG</td><td align="center">155(0.25)</td><td align="center">161(0.25)</td><td align="center">0.96 (0.77, 1.20)</td><td align="center">0.97(0.74, 1.28)</td><td align="center">0.845</td></tr><tr><td align="left">R574P GG</td><td align="center">15(0.02)</td><td align="center">14(0.02)</td><td align="center">1.07 (0.52, 2.22)</td><td align="center">1.07(0.50, 2.28)</td><td align="center">0.857</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">G607G AA</td><td align="center">204(0.35)</td><td align="center">193(0.35)</td><td align="center">1.06 (0.87, 1.29)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">G607G AC</td><td align="center">265(0.46)</td><td align="center">268(0.48)</td><td align="center">0.99 (0.83, 1.17)</td><td align="center">0.91(0.68, 1.22)</td><td align="center">0.538</td></tr><tr><td align="left">G607G CC</td><td align="center">114(0.20)</td><td align="center">94(0.17)</td><td align="center">1.21 (0.92, 1.59)</td><td align="center">1.06(0.73, 1.55)</td><td align="center">0.746</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">V694V GA</td><td align="center">453(0.75)</td><td align="center">443(0.74)</td><td align="center">1.02 (0.90, 1.17)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">V694V GG</td><td align="center">138(0.23)</td><td align="center">145(0.24)</td><td align="center">0.95 (0.75, 1.20)</td><td align="center">0.94(0.71, 1.26)</td><td align="center">0.694</td></tr><tr><td align="left">V694V AA</td><td align="center">14(0.02)</td><td align="center">11(0.02)</td><td align="center">1.27 (0.58, 2.80)</td><td align="center">1.11 (0.49, 2.48)</td><td align="center">0.807</td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left" colspan="6">Risk estimates from CLR analysis of matched case-control pairs</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">Q279R AA</td><td align="center">166(0.61)</td><td align="center">131(0.48)</td><td align="center">1.28 (1.01, 1.59)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">Q279R AG</td><td align="center">75(0.28)</td><td align="center">117(0.43)</td><td align="center">0.64 (0.48, 0.86)</td><td align="center">0.57 (0.38, 0.85)</td><td align="center">0.006</td></tr><tr><td align="left">Q279R GG</td><td align="center">30(0.11)</td><td align="center">25(0.11)</td><td align="center">1.20 (0.71, 2.04)</td><td align="center">0.95 (0.54, 1.69)</td><td align="center">0.945</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">R574P CC</td><td align="center">218(0.74)</td><td align="center">213(0.71)</td><td align="center">1.02 (0.85, 1.24)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">R574P CG</td><td align="center">68(0.23)</td><td align="center">80(0.27)</td><td align="center">0.85 (0.62, 1.17)</td><td align="center">0.76(0.50, 1.14)</td><td align="center">0.181</td></tr><tr><td align="left">R574P GG</td><td align="center">7(0.02)</td><td align="center">7(0.02)</td><td align="center">1.00 (0.35, 2.85)</td><td align="center">1.06(0.31–3.69)</td><td align="center">0.927</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">G607G AA</td><td align="center">97(0.35)</td><td align="center">101(0.37)</td><td align="center">0.96 (0.73, 1.27)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">G607G AC</td><td align="center">127(0.46)</td><td align="center">132(0.48)</td><td align="center">0.96 (0.75, 1.23)</td><td align="center">0.96(0.63, 1.46)</td><td align="center">0.851</td></tr><tr><td align="left">G607G CC</td><td align="center">55(0.20)</td><td align="center">43(0.16)</td><td align="center">1.28 (0.86, 1.91)</td><td align="center">1.03(0.60, 1.78)</td><td align="center">0.908</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">V694V GA</td><td align="center">225(0.79)</td><td align="center">213(0.76)</td><td align="center">1.06 (0.88, 1.27)</td><td align="center">reference</td><td align="center">-</td></tr><tr><td align="left">V694V GG</td><td align="center">54(0.19)</td><td align="center">63(0.22)</td><td align="center">0.86 (0.60, 1.23)</td><td align="center">0.94(0.60, 1.48)</td><td align="center">0.371</td></tr><tr><td align="left">V694V AA</td><td align="center">5(0.02)</td><td align="center">5(0.02)</td><td align="center">1.00 (0.29, 3.45)</td><td align="center">0.79 (0.21, 2.95)</td><td align="center">0.725</td></tr></tbody></table></table-wrap></sec><sec><title>Haplotype analysis</title><p>Table <xref ref-type="table" rid="T2">2</xref> shows the estimated frequency and associated risks of TS and TT by multivariate CLR analysis for inferred haplotypes at the <italic>MMP9 </italic>locus spanning exons 6 to 13. The 4 SNP sites segregated into seven haplotypes. Four common haplotypes with population frequency > 10% accounted for more than 80% of total variation, suggesting high linkage disequilibrium (LD) between segregating sites.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Comparison of MMP9 haplotype frequency estimates in cases and controls and risk estimates from CLR analysis of matched case-control pairs. The haplotype configuration is as follows: Q279R, R574P, G607G and V694V.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">haplotype</td><td align="center"># controls</td><td align="center">freq</td><td align="center"># cases (TS)</td><td align="center">freq</td><td align="center">p-value</td><td align="center">CLR:OR (95%CI)</td><td align="center"># controls</td><td align="center">freq</td><td align="center"># cases(TT)</td><td align="center">freq</td><td align="center">p-value</td><td align="center">CLR:OR (95%CI)</td></tr></thead><tbody><tr><td align="center">ACCG</td><td align="center">450</td><td align="center">0.34</td><td align="center">466</td><td align="center">0.36</td><td align="center">-</td><td align="center">reference</td><td align="center">204</td><td align="center">0.32</td><td align="center">218</td><td align="center">0.36</td><td align="center">-</td><td align="center">reference</td></tr><tr><td align="center">ACAG</td><td align="center">298</td><td align="center">0.22</td><td align="center">303</td><td align="center">0.23</td><td align="center">0.845</td><td align="center">0.98(0.80, 1.21)</td><td align="center">150</td><td align="center">0.24</td><td align="center">167</td><td align="center">0.27</td><td align="center">0.514</td><td align="center">1.11 (0.82, 1.50)</td></tr><tr><td align="center">GCAA</td><td align="center">201</td><td align="center">0.15</td><td align="center">186</td><td align="center">0.14</td><td align="center">0.257</td><td align="center">0.87 (0.69, 1.11)</td><td align="center">91</td><td align="center">0.14</td><td align="center">75</td><td align="center">0.12</td><td align="center">0.224</td><td align="center">0.79 (0.54, 1.15)</td></tr><tr><td align="center">GCAG</td><td align="center">161</td><td align="center">0.12</td><td align="center">130</td><td align="center">0.10</td><td align="center">0.048</td><td align="center">0.77 (0.59, 0.99)</td><td align="center">80</td><td align="center">0.13</td><td align="center">55</td><td align="center">0.09</td><td align="center">0.021</td><td align="center">0.62 (0.41, 0.93)</td></tr><tr><td align="center">AGAG</td><td align="center">123</td><td align="center">0.09</td><td align="center">130</td><td align="center">0.10</td><td align="center">0.705</td><td align="center">1.06 (0.79,1.40)</td><td align="center">61</td><td align="center">0.10</td><td align="center">55</td><td align="center">0.09</td><td align="center">0.451</td><td align="center">0.85 (0.56, 1.30)</td></tr><tr><td align="center">AGCG</td><td align="center">66</td><td align="center">0.05</td><td align="center">55</td><td align="center">0.04</td><td align="center">0.204</td><td align="center">0.78 (0.53, 1.15)</td><td align="center">33</td><td align="center">0.05</td><td align="center">27</td><td align="center">0.04</td><td align="center">0.171</td><td align="center">0.66 (0.37, 1.20)</td></tr><tr><td align="center">GCCG</td><td align="center">30</td><td align="center">0.02</td><td align="center">32</td><td align="center">0.02</td><td align="center">0.977</td><td align="center">0.99 (0.58, 1.68)</td><td align="center">13</td><td align="center">0.02</td><td align="center">17</td><td align="center">0.03</td><td align="center">0.604</td><td align="center">1.24 (0.56, 2.75)</td></tr></tbody></table></table-wrap><p>The GCAG haplotype was associated with lower risk of TS and TT (Table <xref ref-type="table" rid="T2">2</xref>). The risk of both TS and TT decreased with the number of copies of the haplotype GCAG (test for trend OR = 0.81, 95%CI = 0.63, 1.02, p = 0.07 and OR = 0.67 95%CI = 0.46, 0.96, p = 0.03 for TS and TT respectively). The GCAG-haplotype effect on risk of TT was greater than that on risk of TS.</p><p>Haplotypes containing the Q279R G allele, itself associated with decreased risk, were more commonly seen among controls than cases, although the difference in haplotype frequency between cases and controls reach significance only for haplotype GCAG. Conversely, the commonest haplotypes ACCG and ACAG were more frequently seen among cases than controls, and both carry the Q279R A allele. The strength of the single marker association is very similar to, if not greater than that of the haplotype, which supports the suggestion that the Q279R SNP itself may be causal.</p></sec></sec><sec><title>Discussion</title><p>The present study found a coding SNP (Q279R in exon 6) within the <italic>MMP9 </italic>gene to be associated with a lowered risk of trachomatous scarring and trichiasis, which was more marked for trichiasis. This supports the validity of the association because trichiasis is a more advanced form of cicatricial trachoma and also less subject to clinical misclassification than scarring trachoma.</p><p>The Q279R mutation leads to the substitution of a positively charged amino-acid (arginine) by an uncharged amino acid (glutamine) at position 279 within the <italic>MMP9 </italic>active site. Although the functional impact of this polymorphism on the protein is unknown, this variant is noteworthy; it is located in the coding sequence of one of the highly conserved fibronectin type II-like repeats which confers MMP-9 with high affinity binding to type IV collagen, type I gelatin and elastin [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. The digestion of type IV collagen in the epithelial basement membrane has been suggested to be a key regulatory event in the initiation of fibrosis. A plausible explanation of these results is that Q279R represents a partial loss-of-function mutation within the proteinase whose presence reduces the development of fibrosis; these findings implicate MMP-9 as a key molecule in the pathogenesis of conjunctival scarring in trachoma.</p><p>The effect observed in this work was primarily associated with the Q279R GA heterozygous genotype. It is possible that the apparent lack of statistical effect of the GG homozygous genotype may have resulted from the modest sample size within the Q279R GG stratum. However we do not rule out the possibility of this finding being genuine and of biological significance; the increased risk of cicatricial trachoma associated with the Q279R AA homozygous genotype may be linked to the excessive MMP-9-driven proteolytic activation seen in the chronic inflammatory environment of the conjunctiva of subjects with trachoma[<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. Conversely, if Q279R GG homozygotes secrete only "sub-functional" copies of the enzyme, associated with low MMP-9 activity, this may result in disordered tissue remodelling because of a greater deposition of matrix, which has been found to correlate with hypertrophic scars[<xref ref-type="bibr" rid="B13">13</xref>]. In this setting, an advantage for Q279R GA heterozygotes could arise from the secretion of both native and defective forms of the enzyme at the site of inflammation upon cellular activation. This may lead to both a reduction of tissue proteolysis in chronic inflammation and protection against excessive tissue destruction and scarring.</p><p>MMP-9 expression and activity is up-regulated in the inflamed conjunctiva of trachoma subjects and increases with the severity of clinical inflammation [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. There is evidence to suggest that the elevated local levels of TNF and IL-1β associated with chlamydial infection and inflammation[<xref ref-type="bibr" rid="B3">3</xref>] act as an important trigger of matrix degradation by inducing MMP-9 activity in inflamed tissue. Destruction of the ECM may facilitate cell migration and result in the characteristic infiltrating leukocytes and stromal cells (e.g. macrophages and fibroblasts) seen in trachoma subjects[<xref ref-type="bibr" rid="B4">4</xref>]. The secretion of MMP-9, cytokines and chemokines by these cells may drive the inflammatory process into a positive feedback loop. MMP-9 can activate pro-inflammatory cytokines[<xref ref-type="bibr" rid="B14">14</xref>] which may further increase and perpetuate inflammation leading to persistent tissue damage. Later during the tissue repair phase, excessive MMP activities may contribute to contractile scarring, characteristic of TS and TT, through their role in ocular fibroblast-mediated matrix contraction [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. It is plausible that persistently increased MMP9 activity in the conjunctiva of subjects with trachoma may be a key event in the pathogenesis of conjunctival scarring through excessive degradation of ECM and fibrosis. The recent characterisation of MMP-9 expression and activity in mouse strains exhibiting variable susceptibility to sequelae of genital <italic>Chlamydia muridarum </italic>(MoPn) infection[<xref ref-type="bibr" rid="B5">5</xref>] supports this suggestion: C3H/HeN mice, which are particularly susceptible to fibrotic sequelae in this model[<xref ref-type="bibr" rid="B6">6</xref>], exhibited greater MMP-9 transcription and activity during infection.</p><p>The blinding lesion in trachoma is corneal opacity (CO), which is thought to result from the lashes rubbing against the eyeball (TT) and damaging the cornea. Epidemiological studies indicate that after successful TT surgery, CO can develop[<xref ref-type="bibr" rid="B17">17</xref>], and there is evidence that inflammatory episodes in TT subjects correlate with the presence of CO[<xref ref-type="bibr" rid="B15">15</xref>]. MMPs are expressed at sites of epithelial loss in ulcerated corneas and found in the tear film of patients with external ocular inflammatory disorders[<xref ref-type="bibr" rid="B18">18</xref>]. In the cornea, an imbalance in MMP9 production has been implicated in corneal scarring and loss of corneal transparency and visual function[<xref ref-type="bibr" rid="B7">7</xref>]. Proteolytic destruction of the glandular tissue contributes to reduced secretion of tears, and is associated with a number of corneal pathologies[<xref ref-type="bibr" rid="B7">7</xref>]. The constant exposure of the corneal surface to conjunctival and/or tear MMP-9 and inflammatory cytokines may contribute to progressive corneal opacification. CO may be a relevant phenotype for <italic>MMP9 </italic>genetic association analysis, but for practical reasons large numbers of patients with CO are difficult to come by.</p></sec><sec><title>Conclusion</title><p>This work supports the hypothesis that MMP-9 has a role in the pathogenesis of blinding trachoma. The risk reductions associated with the Q279R exonic mutation are modest: 26% for scarring and 34% for trichiasis. CLR analysis suggests that these effects are independent of those previously reported in the interleukin 10 (<italic>IL10</italic>) and interferon gamma (<italic>IFNγ</italic>) loci [<xref ref-type="bibr" rid="B10">10</xref>]. This is consistent with a model of cicatricial trachoma as a complex disorder with multiple genetic factors contributing to risk of scarring and trichiasis after ocular chlamydial infection. Further studies of MMP-9 in chlamydial infection and of the functional significance of genetic variation at the MMP-9 locus, are warranted.</p></sec><sec><title>Abbreviations</title><p>ECM -extracellular matrix, MMP -matrix metalloproteinase, SNP -single nucleotide polymorphisms, CLR -conditional logistic regression, TIMPs – tissue inhibitors of matrix metalloproteinases, WHO -World Health Organization, TS -scarring trachoma, TT -trichiasis, OR -odds ratio, CI -confidence interval, LD- linkage disequilibrium, CO -corneal opacity, TNF -tumor necrosis factor, IL-1β-interleukin 1 beta, IL10 -interleukin 10, IFNγ-interferon gamma.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>AN collected, edited, analysed the data and wrote the manuscript. RB directed the study, study design and co-wrote the manuscript. GC study design and contributed to the manuscript. DM, MH and MB directed the study, study design and contributed to the manuscript. DK and KR co-directed the study. OJ and HJ collected clinical data and provide clinical material</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2350/7/40/prepub"/></p></sec>
|
Screening chest radiography: results from a Greek cross-sectional survey
|
<sec><title>Background</title><p>Public health authorities worldwide discourage the use of chest radiography as a screening modality, as the diagnostic performance of chest radiography does not justify its application for screening and may even be harmful, since people with false positive results may experience anxiety and concern. Despite the accumulated evidence, various reports suggest that primary care physicians throughout the world still prescribe chest radiography for screening. We therefore set out to index the use of chest radiography for screening purposes among the healthy adult population and to analyze its relationship with possible trigger factors.</p></sec><sec sec-type="methods"><title>Methods</title><p>The study was designed as a cross-sectional survey. Five thousand four hundred and ninety-nine healthy adults, coming from 26 Greek provinces were surveyed for screening practice habits in the nationwide anticancer study. Data were obtained for the use of screening chest radiography. Impact of age, gender, tobacco exposure, family history positive for malignancies and professional-risk for lung diseases was further analyzed.</p></sec><sec><title>Results</title><p>we found that 20% (n = 1099) of the surveyed individuals underwent chest radiography for screening purposes for at least one time during the previous three years. Among those, 24% do so with a frequency equal or higher than once yearly, and 48% with a frequency equal or higher than every three years. Screening for chest radiography was more commonly adopted among males (OR 1.130, 95% CI 0.988–1.292), pensioners (OR 1.319, CI 1.093–1.593) and individuals with a positive family history for lung cancer (OR 1.251, CI 0.988–1.583). Multivariate analysis confirmed these results.</p></sec><sec><title>Conclusion</title><p>Despite formal recommendations, chest radiography for screening purposes was a common practice among the analyzed sample of Greek adults. This practice is of questionable value since the positive predictive value of chest radiography is low. The implementation of even a relatively inexpensive imaging study on a national scale would greatly burden health economics and the workload of radiology departments.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Kamposioras</surname><given-names>Konstantinos</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Casazza</surname><given-names>Giovanni</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" corresp="yes" contrib-type="author"><name><surname>Mauri</surname><given-names>Davide</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Lakiotis</surname><given-names>Velisarios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Cortinovis</surname><given-names>Ivan</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Xilomenos</surname><given-names>Apostolos</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Peponi</surname><given-names>Christina</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Golfinopoulos</surname><given-names>Vassilis</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Milousis</surname><given-names>Athanasios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A10" contrib-type="author"><name><surname>Kakaridis</surname><given-names>Dimitrios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A11" contrib-type="author"><name><surname>Zacharias</surname><given-names>Georgios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A12" contrib-type="author"><name><surname>Karathanasi</surname><given-names>Ioanna</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A13" contrib-type="author"><name><surname>Ferentinos</surname><given-names>Georgios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A14" contrib-type="author"><name><surname>Proiskos</surname><given-names>Anastasios</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>Chest radiography has a long tradition in medical care, however its prescription for screening purposes among healthy individuals is discouraged by public health authorities [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. In fact, due to the low prevalence of tuberculosis in developed countries and the incapability to modify lung cancer-specific mortality, the use of chest radiography as a screening tool is not effective.</p><p>The diagnostic performance of chest radiography does not justify its application for screening neither in the general population nor in "high risk" groups like smokers or people with a family history of lung cancer [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. Screening chest radiography is not considered effective, it does not have a high yield, and false positive exams result in additional and unnecessary medical tests, associated economic costs, and patient anxiety and stress [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>].</p><p>Despite the accumulated evidence and the clear guidelines, various reports suggest that primary care physicians throughout the world still prescribe chest radiography for screening both in the general population and in selected "high risk" subgroups [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. Consequently, screening chest radiography may represent a major problem that harms screenees' health, and burdens public-health economics and radiology departments' activities.</p><p>Nevertheless, since the proportion of physicians believing in and recommending a screening test may consistently differ from the proportion of healthy individuals undergoing the test (still dependent on patients' will), the negative impact of screening chest radiography on health and economics may be only speculated. Little is in fact known in peer-reviewed literature about how chest radiography for screening purposes is practiced among the general healthy adult population [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B18">18</xref>].</p><p>We therefore tried to evaluate the rate of screening chest radiography practice among a large sample of Greek healthy adults. Furthermore, we analyzed the resulting chest radiography screening practice for the impact of professional risk for lung diseases, family history of cancer and smoking practice.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>PACMeR_02 trial</title><p>This study is part of a large ongoing survey on cancer screening and preventive practice in Greece, which is organized by PACMeR (Panhellenic Association for continual Medical Research), and has the purpose to reveal the current rate of cancer screening among the Greek adult population, to evidence possible barriers to early diagnosis of cancer and to analyze over-practice events and possible sources of worthless costs. For the project, PACMeR physicians had dedicatedly prepared two medical questionnaires (one for male and one for female) for face-to-face interviews that were employed during the research program. The exact phrasing of the chest radiography questions used is provided in the supplementary note for the facilitation of the peer-review process.</p><p>The project was ethically approved by PACMeR's Scientific Committee (protocol number 08_020720) and conformed to the ethical guidelines of the 1975 Declaration of Helsinki. A written informed consent form was obtained from all the participants before completing the study questionnaire and the data retrieved were analyzed in anonymous and codified form.</p></sec><sec><title>Population and data extraction</title><p>The study population was composed of a nationwide convenience sample: adults bringing or visiting their relatives while getting healthcare in Hospitals and Health Centers of 26 Greek provinces (Fig. <xref ref-type="fig" rid="F1">1</xref>). Most populated Greek areas were involved in the study, including more than 80% of the Greek population. Five thousand four hundred and ninety-nine individuals (2948 female, 2551 male, age range 21–97) entered the study and answered the questionnaires during a face-to-face interview between 2000 and 2004.</p><p>Ninety-two physicians employed in primary care activities were involved in the study, 87 of them as interviewers, and five as data managers and quality control personnel. Data storing was assured by SESy, a dedicated database [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>] tailored to population-based cross-sectional surveys for cancer prevention and screening assessment.</p><p>Data were extracted for overall chest radiography practice. For each individual we retrieved the chronological period that elapsed from the last chest radiography and the cause for which chest radiography was performed. We further evaluated the proportion of individuals who assessed that they underwent chest radiogram last time for screening purposes. For people who performed it within three years we still analyzed the frequency by which they underwent the test.</p></sec><sec><title>Definition</title><p>Since the diagnostic performance of chest radiography does not justify its application in any screening setting, we considered chest radiography being done for screening purposes in any of the following situations: 1) periodic health examination (conducted at regular intervals, e.g. yearly); 2) check-up visit (requested by individuals who do not undergo health examination at regular intervals); 3) chest radiography in asymptomatic individuals due to patients' will; 4) regulatory reasons (driving license, health certificate etc.).</p></sec><sec><title>Subgroup analysis</title><p>We analyzed the rate of screening chest radiography by the following parameters: age (<45, 45–64, 65–74 and >75 years old), professional category (pensioners, professions at risk for lung diseases and other professions), cancer family history, smoking activity (no smokers, smokers, ex smokers), number of daily cigarettes smoked (<10, 10–20, 20–30, 30–40, >40, no smokers), duration of tobacco exposure (<10, 10–20, 20–30, 30–40, >40 years).</p></sec><sec><title>Statistical analysis</title><p>In order to evidence population subgroups at higher probability of undergoing screening chest radiography, we performed univariate and multivariate analysis. Only subjects for whom there were data about the time elapsed from last chest radiography and about the reason for which they underwent chest radiograms, were considered (n = 5282).</p><p>Individuals entering the analysis were therefore divided in:</p><p>(1) Subjects at higher probability of undergoing screening chest radiography: people who performed chest radiography for screening purposes (regulatory reason excluded) within the last three years (n = 1080);</p><p>(2) Subjects at lower probability of undergoing screening chest radiography exposure: individuals who underwent chest radiography for any other reason (than screening) at any time and those who underwent chest radiograms for screening purposes but more than 3 years had elapsed (n = 4202).</p><p>Univariate analysis was used in order to examine the association between over-practice and all subgroups previously defined. Multiple logistic regression analysis was performed to analyze the relationship between over-practice and some relevant covariates of interest: sex, age, professional category, smoke and family history of lung cancer. All independent variables were taken as categorical, dichotomized where appropriate. We used SAS statistical package, version 8.2 [<xref ref-type="bibr" rid="B21">21</xref>], for analyses at 95% confidence intervals.</p></sec></sec><sec><title>Results</title><sec><title>Population</title><p>The characteristics of the individuals involved in the study are reported in table <xref ref-type="table" rid="T1">1</xref>. The mean age of individuals who entered the study was 60.35 years. There was a significant difference in the age distribution of female and male individuals involved in the survey (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2458-6-113-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msubsup><mml:mi>χ</mml:mi><mml:mn>3</mml:mn><mml:mn>2</mml:mn></mml:msubsup></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaaiiGacqWFhpWydaqhaaWcbaGaeG4mamdabaGaeGOmaidaaaaa@307D@</mml:annotation></mml:semantics></mml:math></inline-formula> = 332.89 p < 0.001) largely driven by the higher proportion of women included in the age group < 45 years old (14.3% for women vs 2.3% for men). The mean age of male individuals involved in the study was consistently higher than the mean age of women: 63.4 years old (standard error 0.19) versus 57.7 years old (standard error 0.21) (Table <xref ref-type="table" rid="T1">1</xref>).</p></sec><sec><title>Chest radiography patterns</title><p>76.6% of the population analyzed (n = 4212) referred that they underwent at least one chest radiogram during their life; 29.5% (n = 1622) assessed that they underwent it last time for screening purposes; 43.4% (n = 2385) performed it in out-patient basis for medical reasons; and 3.7% (n = 205) underwent it in in-patient basis.</p><p>Among people who underwent chest radiography for screening purposes, (Table <xref ref-type="table" rid="T2">2</xref>) we found that 19.98% (1099 individuals; 537 males and 562 females) did so within a three-year period. Among those (information available for 936 out of 1099), 24.15% (n = 226) underwent it with a frequency equal or higher than once yearly and 47.97% (n = 449) with a frequency equal or higher than every three years. Details on frequencies are reported in table <xref ref-type="table" rid="T3">3</xref>.</p></sec><sec><title>Subgroup analysis</title><p>Univariate analyses evidenced that the risk of screening radiogram performance was statistically higher among pensioners (OR 1.319, 95%CI 1.093–1.593). Trends to higher chest X-ray practice were still found among individuals of male gender (OR 1.130, 95% CI 0.988–1.292) and those with a family history positive for lung cancer (OR 1.251, 95% CI 0.988–1.583), although these trends were not statistically significant. Interestingly people with professions at risk for lung diseases showed lower probability to undergo screening chest radiograms (OR 0.846, 95% CI 0.713–1.003). Chest radiography performance for screening purposes was not influenced by age, tobacco consumption, and family history positive for malignancies (Table <xref ref-type="table" rid="T4">4</xref>).</p><p>Multivariate analysis confirmed the results obtained by univariate analysis. Pensioners have a higher probability of undergoing a screening chest x-ray exam (OR 1.277, 95% CI 1.041–1.568); and trends for male gender (OR 1.188, 95% CI 1.013–1.393), and professions at risk for lung diseases (OR 0.808, 95% CI 0.675–0.968) became statistically significant (Table <xref ref-type="table" rid="T5">5</xref>).</p></sec></sec><sec><title>Discussion</title><p>Screening tests are generally harmful and only in selected cases their benefit outweighs potential harms [<xref ref-type="bibr" rid="B22">22</xref>]. In two systematic reviews of older randomized trials there was no evidence supporting the use of chest radiography for lung cancer screening [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. If anything, in these reviews screening with chest radiography was associated with increased lung cancer mortality [<xref ref-type="bibr" rid="B7">7</xref>], although this finding is consistent with over-diagnosis bias, given that overall mortality was not affected. Health hazards are not related to radiation exposure, since the delivered dose is very low [<xref ref-type="bibr" rid="B23">23</xref>]; they rather stem from the additional diagnostic and/or therapeutic interventions during further evaluation of false positive findings [<xref ref-type="bibr" rid="B6">6</xref>]. Indicatively, the proportion of abnormal chest x-ray findings ranges between 3–10% [<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B25">25</xref>] with a rate of false positive results ranging from 40–60% [<xref ref-type="bibr" rid="B7">7</xref>]. Thus, the implementation of even a relatively inexpensive imaging study on a massive scale would greatly burden health economics and the workload of radiology departments.</p><p>Despite the available evidence and recommendations, physicians throughout the world still prescribe chest radiography for screening purposes [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. Little is known in peer-reviewed literature about how chest radiography is practiced for screening purposes among the general healthy adult population. Based on our review of the current literature, only three studies have been published since 1995 [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B18">18</xref>], but all these studies present major limitations. In the study of Woodward (1996) the "perceptions of 452 Canadian physicians about the extent to which patients in their practices obtained screening chest radiography at regular intervals" were investigated [<xref ref-type="bibr" rid="B17">17</xref>]. In the study of Hutchison (1998) the proportion of chest radiograms recommended by 62 Canadian physicians during 246 unannounced "standard patients" was evaluated [<xref ref-type="bibr" rid="B18">18</xref>]. However no data had been reported in these two studies about the real application of the test among the underlying populations. In the third study (1995), 3281 patients' charts were audited from medical archives of 60 physicians, and data were further abstracted for screening chest radiography practices [<xref ref-type="bibr" rid="B16">16</xref>]. Still in this case the information should be considered incomplete since we do not know anything about the proportion of patients who performed the test due to their own will (opportunistic screening), or prescription by another physician.</p><p>This is therefore the first study indexing the impact of screening chest radiography habit among a population subgroup. Practice of chest radiograms for screening purposes was common among the examined sample of Greek adults: 20% underwent it for at least one time during the previous three years and among these, 48% declared to perform it with a frequency equal or higher than once every three years.</p><p>Interestingly, in logistic regression analyses the high-utilization rates were not strongly driven by smoking practice (smokers versus no smokers), as previously hypothesized [<xref ref-type="bibr" rid="B26">26</xref>]. Moreover, people with professions at risk for lung disease also showed lower high-utilization trends. Individuals at major risk for over-screening chest radiograms were male subjects, pensioners and individuals with a family history positive for lung cancer.</p><p>The retrieved rates of screening chest radiography should not surprise. In a recent Greek survey of 211 physicians, 88% declared to recommend chest radiography for early diagnosis procedures: 78% prescribed it during usual check-up visit, and 77% recommend it for cancer screening [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>High chest radiography prescription rates may still be explained by the absence of national guidelines and it might be guessed that the European Code Against Cancer recommendations [<xref ref-type="bibr" rid="B1">1</xref>] do not have any impact on prescription practices. Ignorance of the formal recommendations on the issue might be an explanation, especially in countries without a strong tradition in primary care medicine.</p><p>Some limitations should be discussed. First, despite the fact that screening chest radiography is being studied from the sixties and onwards, this is the first study analyzing its practice among the general population. Since the Greek primary care system based on specialized physicians is "newborn", it might be precarious to generalize these findings globally. Second, we analyzed only patients that underwent chest radiography within three years. This may under-estimate the proportion of individuals screened since many of them may have undergone screening chest radiograms in an antecedent date. Furthermore, data were derived from a cross-sectional study on a large convenience sample of the Greek healthy adult population. This design has limited internal validity and is sensitive to a variety of biases. Nevertheless, cross-sectional surveys are most commonly used, and are considered appropriate and easy to perform.</p></sec><sec><title>Conclusion</title><p>Chest radiography practice for screening purposes is an old habit that dies hard. More research should be conducted concerning the causes and possible remedies of this phenomenon.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>KK was the coordinator of the Greek branch of the study. He was still actively involved in the discussion of the project & study planning, and manuscript writing. GC: statistician, main co-operator from the University of Milan (Italy), dept. statistics, he was still actively involved in the discussion of the project, statistics and manuscript writing. DM: main coordinator of the Italian and Greek braches of the study. He was still actively involved in the discussion of the project, realization of the draft & study planning, review of data abstraction and manuscript writing. VL was involved in study planning and he was responsible for data collection in Peloponnesus and Cephalonia island. He was still actively involved in the discussion of the project and manuscript. IC: statistician, second co-operator from the University of Milan (Italy), dept. Statistics, he was still actively involved in the discussion of the project, statistics and manuscript discussion. AX: responsible for double-blind controls of data inserted by data-managers (thus allowing data-feasibility for both hard- and electronic-data). He was still actively involved in the discussion of the project & study planning, and manuscript writing. CP: main data-manager. She was involved in the study planning and was responsible for data collection in the north-western part of Greece. She was responsible of data entering in the peripheral units of SESy database. She was involved in manuscript discussion. VG was actively involved in the discussion of the project, realization of the draft, discussion of the outcomes, reviewing and formatting the manuscript. AM was involved in the study planning and was responsible for data collection in north-eastern Greece. He was still actively involved in the discussion of the project and manuscript. DK was involved in the study planning and was responsible for data collection in north-central Greece. He was still actively involved in the discussion of the project and manuscript. GZ was involved in the study planning and was responsible for data collection in the wide area of Attika. Hhe was still actively involved in the discussion of the project and manuscript. IK was involved in the study planning and was responsible for data collection in Athens area and the province of Kozani. She was still actively involved in the discussion of the project and manuscript. GF: PACMeR internal statistician. He was involved in study planning, definition of outcome, and draft writing. He constitutes a basic internal support (Greek branch of the study) that avoids possible miss-understanding in the international collaboration. He still participated in the discussion of the results. AP was involved in the study planning and was responsible for data collection in Piraeus area. He was still actively involved in the discussion (any study phase) since his area of expertising is respiratory-diseases. All authors read and approved the final manuscript.</p></sec><sec><title>Supplementary note</title><p>Exact phrasing of the tobacco and chest radiography related questions used for both males and females during the questionnaire-based interviews</p><p>Tobacco-related questions:</p><p>Are You a smoker? [No] [yes]</p><p>How old did you start smoking? []</p><p>How old did you stop smoking? []</p><p>How many cigarettes/tobacco do you daily smoke?......................................................</p><p>Chest radiography related questions:</p><p>When did you perform chest radiography last time?</p><p>ϒ never ϒ Within 1 year ϒ 2 years ϒ 3 years ϒ 5 years ϒ more than 5 years</p><p>For which reason did you do it?.....................................................................................</p><p>At what frequency do you undergo chest radiography?.................................................</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/113/prepub"/></p></sec>
|
Attribution of physical complaints to the air disaster in Amsterdam by exposed rescue workers: an epidemiological study using historic cohorts
|
<sec><title>Background</title><p>In 1992 a cargo aircraft crashed into a residential area of Amsterdam. A troublesome aftermath followed, with rumors on potential toxic exposures and health consequences. Health concerns remained even though no excess morbidity was predicted in retrospective risk evaluations. This study aimed to assess to what extent the rescue workers attribute long-term physical complaints to this disaster, including its aftermath, and to examine associations between such attribution and types of exposure and background variables.</p></sec><sec sec-type="methods"><title>Methods</title><p>Historic cohort study that collected questionnaire data on occupational disaster exposure, attribution of physical complaints, and background variables on average 8.5 years post-disaster. For the present study the workers who were exposed to the disaster were selected from the historic cohort, i.e. the professional firefighters (n = 334), police officers (n = 834), and accident and wreckage investigators (n = 241) who performed disaster-related tasks.</p></sec><sec><title>Results</title><p>Across the three occupational groups, a consistent percentage (ranging from 43% to 49%) of exposed workers with long-term physical complaints attributed these to the disaster, including its aftermath. Those with more physical complaints attributed these to a stronger degree. Multivariate logistic regression analyses showed that attribution was significantly more often reported by firefighters who rescued people, and by police officers who reported the identification and recovery of or search for victims and human remains, clean-up, or security and surveillance of the disaster area; who witnessed the immediate disaster scene; who had a close one affected by the disaster; and who perceived the disaster as the worst thing that ever happened to them. Age, sex and educational level were not significantly associated with attribution.</p></sec><sec><title>Conclusion</title><p>This study provides further cross-sectional evidence for the role of causal attribution in post-disaster subjective physical health problems. After on average 8.5 years, almost a third (32%) of all the exposed workers, and almost half (45%) of the exposed workers with physical complaints, attributed these complaints to the disaster, including its aftermath. The similarity of the results across the occupational groups suggests a general rather than an occupation-specific attribution process. Longitudinal studies are needed to determine whether causal disaster attribution leads to persistence of post-disaster complaints and health care utilization.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Slottje</surname><given-names>Pauline</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Smidt</surname><given-names>Nynke</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Twisk</surname><given-names>Jos WR</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Huizink</surname><given-names>Anja C</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Witteveen</surname><given-names>Anke B</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>van Mechelen</surname><given-names>Willem</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A7" corresp="yes" contrib-type="author"><name><surname>Smid</surname><given-names>Tjabe</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I7">7</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>In 1992 a cargo aircraft crashed into apartment buildings in a residential area of Amsterdam, killing 43 persons and destroying 266 apartments[<xref ref-type="bibr" rid="B1">1</xref>]. Through the years after the disaster, various alleged disaster-related exposures to hazardous materials and health consequences were reported in the media and publicly discussed [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]. No excess morbidity due to exposure to hazardous materials was predicted in retrospective risk evaluations [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>], but health concerns remained among some of the residents and workers who were involved in the disaster [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Therefore, the epidemiological study air disaster in Amsterdam (ESADA) was initiated in 2000, to assess the relationship between occupational disaster exposure and the long-term health of rescue workers [<xref ref-type="bibr" rid="B7">7</xref>]. Previous results showed increased physical symptom rates among exposed workers compared to their colleagues who were not exposed to this disaster [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. Although no clinical diagnostic data on these symptoms is available, extensive analysis of blood and urine samples did not reveal evidence for disaster-related pathological processes [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. Subjective physical health problems without sufficient clinical or toxicological explanation are commonly referred to as "unexplained physical symptoms" in the literature. They are probably ubiquitous in the general population, but have also been reported after disasters and (perceived) noxious exposures in civilian [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B14">14</xref>] and military populations [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Causal attribution could play a role in the reported post-disaster subjective health problems. In general, people strive to understand the bodily sensations they detect and the causal attributions they adopt have been characterized as normalizing (i.e. benign sensations) as opposed to somatizing and psychologizing (i.e. symptoms) [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>]. Further, a distinction is made between attributing symptoms to internal versus external (environmental) causes [<xref ref-type="bibr" rid="B20">20</xref>]. In times of stress and uncertainty about noxious exposures, people could be even more likely to attend to their bodies and detect bodily sensations (hypervigilance), and also to interpret the detected sensations as pathological (hypochondria) [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. It has further been postulated that people affected by a disaster who perceive bodily sensations as pathological, are likely to attribute them to the disaster experience and related exposures, whether this is realistic or not [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]. According to Vasterman et al "there is a strong relationship between the symptomatology seen in the aftermath of disasters and medically unexplained physical symptoms in the general population. The only difference is that, after a disaster, the symptoms are attributed to the event" [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>In addition to its potential role in post-disaster subjective health problems, causal attribution may also affect the course of physical symptoms. Strong attributions to somatic causes and environmental exposures, for example, have been shown to be associated with a worse prognosis in patients with chronic fatigue syndrome and in other patients with medically unexplained physical symptoms [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B30">30</xref>]. The medical expenses of these patients with sustained physical symptoms could be higher, because of their frequent visits to various medical specialists, who usually cannot confirm the patient's conviction [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>].</p><p>Thus, causal attributions could affect physical health perception, functioning, and health care utilization, both in general and after disasters. However, little is known about the causal attributions of those involved in disasters. The main aim of the present study is to assess the extent to which the rescue workers, who were occupationally exposed to the air disaster in Amsterdam, attribute their long-term physical complaints to this disaster, including its aftermath. Because those who attribute physical complaints to a disaster may need a different approach in after care programs, an attempt is also made to characterize those with such attribution. To this end, the present study also examines which factors are associated with attribution, in terms of types of exposure and background characteristics. These characteristics might help in targeting a tailored after care program to those who attribute physical complaints to the disaster.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Participants and data collection</title><p>The present study is part of the Epidemiological Study Air Disaster in Amsterdam (ESADA), of which the study design has been published previously [<xref ref-type="bibr" rid="B7">7</xref>]. The ESADA is a historic cohort study, using self-reported exposure status. The study population consists of a historic cohort of rescue workers, including both the workers who were occupationally exposed to this disaster (i.e. who reported one or more disaster-related tasks), and their colleagues who were not exposed to it. The overall participation rate was 70% of those traced and invited to join in the study (n = 3742). For the purpose of the present study the participating 1409 exposed workers were selected. This concerns three occupational groups: (1) professional firefighters (n = 334) working at the Amsterdam fire department at the time of the disaster; (2) police officers (n = 834) working at the Amsterdam-Amstelland regional police force at the time of the disaster and at the start of the study; and (3) accident and wreckage investigators (so-called hangar workers) (n = 241), who worked for one of the departments involved in the transport, security and sorting of the wreckage at Schiphol Airport at the time of the disaster.</p><p>The medical ethics committees of the medical centers involved in the ESADA approved the study protocol (i.e. the VU University Medical Center and the Onze Lieve Vrouwe Gasthuis in Amsterdam). Participants signed informed consent and participated voluntarily. Data were collected from January 2000 to March 2002, i.e. on average 8.5 years post-disaster. All the data used in the present study were assessed by means of questionnaires and were entered twice, after which inconsistencies were reviewed and any mistakes rectified.</p></sec><sec><title>Long-term physical complaints and attribution thereof</title><p>Workers were asked to indicate whether they currently had physical complaints on a four-point scale (very many, many, few, or none). Those who reported having physical complaints were subsequently asked to indicate to what extent they thought these were related to the air disaster, including its aftermath, on a four-point scale (a very strong, strong, weak, or no relationship). Attribution was defined as reporting a weak, strong or very strong relationship. Workers who attributed physical complaints, were asked to specify these complaints, i.e. skin, back, joints, shortness of breath or lung problems, fatigue, headaches, or other.</p><p>In addition, all workers were asked to indicate the current absence or presence of 34 physical symptoms (from a questionnaire that was constructed specifically for this study [<xref ref-type="bibr" rid="B33">33</xref>]), and to complete the 20-item Checklist Individual Strength [<xref ref-type="bibr" rid="B34">34</xref>] on fatigue-related symptoms, the presence of which was defined as a total score above 76 [<xref ref-type="bibr" rid="B35">35</xref>].</p></sec><sec><title>Type of exposure to the disaster</title><p>Type of exposure is characterized according to the following variables:</p><p>a. Disaster-related tasks: having performed the following tasks: rescuing people; identification and recovery of or search for victims and human remains; firefighting; clean-up of the disaster site; security and surveillance of the disaster area; supporting injured victims and workers; and sorting of the wreckage in a hangar at Schiphol Airport.</p><p>b. Witnessed the immediate disaster scene: having seen the disaster scene within the first hours after the crash, or when the wreckage was still there.</p><p>c. Having a close one affected by the disaster: having a close or beloved one (i.e. family members, relatives, friends or acquaintances) who was affected by the disaster (i.e. in life-threatening danger; injured; destroyed apartment; died; or affected in any other way).</p><p>d. Perceived severity of the disaster: workers were asked to complete the following statement with one of the offered answers: "The air disaster and its aftermath was...": (i) not bad; (ii) quite bad; (iii) terrible, but not the worst thing that ever happened to me (abbreviated to "terrible" from now on); or (iv) the worst thing that ever happened to me (abbreviated to "worst thing ever" from now on).</p></sec><sec><title>Background characteristics</title><p>Background characteristics were categorized as follows: age at time of assessment (young versus old [≥ median age of exposed workers with physical complaints per occupational group]); sex (male versus female); and highest level of completed education (low [no education, elementary school, lower vocational education, or lower general secondary education], intermediate [intermediate vocational education, higher general secondary education, pre-university education], versus high [higher vocational education, university]). Data on age and sex were complete. For level of education an additional missing category was used in the statistical analysis, to prevent excluding these workers (6%).</p></sec><sec><title>Statistical analysis</title><p>The following statistical analyses were performed among the workers with physical health complaints, using SPSS (version 10.1) and considering two-sided P-values less than 0.05 as statistically significant. Associations between the degree of physical complaints ([very] many versus few) and attribution thereof (a [very] strong, or weak versus no relationship) were analyzed by means of Pearson χ<sup>2 </sup>tests. Logistic regression was used to analyze associations between the specific types of physical symptoms (dependent variable: present versus absent) and the extent of attribution (categorical independent variable: [very] strong, weak versus no relationship).</p><p>Logistic regression was also used to analyze associations between attribution (dichotomized into yes [very strong, strong or weak relationship] and no [no relationship]) and the following independent variables: the applicable disaster-related tasks, having witnessed the immediate disaster scene, having a close one affected by the disaster (each coded as yes versus no), perceived severity of disaster experience (categorical with "worst thing ever" as reference category), and the background characteristics. In this logistic regression analysis, the independent variables were first introduced separately in "univariate" models, after which they were all introduced together and those with P > 0.10 were subsequently removed in a step-wise backward manner, until only those with P ≤ 0.10 were retained in the final "multivariate" model.</p></sec></sec><sec><title>Results</title><sec><title>Background characteristics of exposed workers</title><p>The firefighters, all male, had a mean age of 51.4 years (SD 5.9), and 59%, 28%, and 6% of them reported a low, intermediate, and high level of education, respectively (8% no data on education). The mean age of the police officers was 44.0 years (SD 6.2); 89% of them were male, and 21%, 53%, and 21% of them reported a low, intermediate, and high level of education, respectively (6% no data on education). The hangar workers, all male, had a mean age of 43.9 years (SD 7.8), and 43%, 44%, and 8% of them reported a low, intermediate, and high level of education, respectively (5% no data on education).</p></sec><sec><title>Prevalence of long-term physical complaints and attribution thereof</title><p>The prevalence of long-term physical complaints and attribution thereof was very similar across the three occupational groups (Table <xref ref-type="table" rid="T1">1</xref>). 72% of all exposed workers reported long-term physical complaints, of whom the majority reported to have few physical complaints. 45% of the workers with long-term physical complaints attributed these to the disaster to some extent. 23% of the workers who attributed physical symptoms to the disaster reported this to be a (very) strong relationship. Workers with more physical complaints were more likely to attribute these complains to a stronger degree to the disaster (P < 0.0001 within each occupational group).</p><p>The top three of types of physical complaints most frequently attributed to the disaster by firefighters, police officers and hangar workers were: skin complaints (58, 50% and 50%, respectively), fatigue (42%, 47% and 70%, respectively), and joint complaints (47%, 35% and 54%, respectively). The prevalence rates of these three types and the other types of physical symptoms were also positively associated with the extent of attribution of physical complaints to the disaster (Table <xref ref-type="table" rid="T2">2</xref>).</p></sec><sec><title>Factors associated with attribution of physical complaints</title><p>In the univariate analysis of firefighters with physical complaints, attribution was significantly associated with rescuing people, firefighting, supporting injured victims and workers, and having witnessed the immediate disaster site (Table <xref ref-type="table" rid="T3">3</xref>). However, only rescuing people remained significant in the multivariate analysis, while supporting injured victims and workers, and having witnessed the immediate disaster scene had P ≤ 0.10. The effect sizes of these types of exposure were similar, but tended to be somewhat lower in multivariate compared to univariate analyses. Background characteristics were not significantly associated with attribution.</p><p>With respect to the police officers, attribution was significantly associated with all the types of exposure both in univariate and in multivariate analyses, except for identification and recovery of or search for victims and human remains in univariate analysis, and rescuing and supporting people in multivariate analysis (Table <xref ref-type="table" rid="T4">4</xref>). Some multivariate odds ratio's were somewhat higher, while others were somewhat lower compared to the univariate ones. Background characteristics were not significantly associated with attribution among police officers.</p><p>Regarding the hangar workers, none of the types of exposure or background characteristics were significantly associated with attribution, although the analysis of education level indicated that hangar workers with intermediate and low levels of education were less likely to attribute physical complaints to the disaster than those with a high level of education (P = 0.05) (Table <xref ref-type="table" rid="T5">5</xref>).</p></sec></sec><sec><title>Discussion</title><p>The aim of the present study was primarily to assess the extent to which exposed workers attributed their long-term physical health complaints to the disaster, including its aftermath, and, secondary, to characterize those who did report such attribution. The results were remarkably similar across the three occupational groups despite their distinct occupational involvement in the disaster. The similarity of results concerned (a) the prevalence of long-term physical health complaints (varying from 70 to 79%, depending on the occupational group); (b) the proportion attributing these complaints to the disaster, including its aftermath, to a weak (32–38%), strong (7–9%), or very strong (1–2%) degree; (c) the types of physical complaints they attributed (the top three being skin complaints, fatigue, and joint pain in each group); (d) the positive associations between the extent of attribution and the severity of physical complaints as well as the prevalence of various physical symptoms, including the abovementioned top three of complaints.</p><p>A remaining intriguing question is to what specific aspects of the disaster and its aftermath the workers attributed their physical complaints, and to what extent these attributions are realistic. The finding that the majority of workers who attributed physical complaints to the disaster reported this to be a weak relationship might indicate that these workers could simply not exclude the possibility of such a relationship, rather than that they had explicit causal ideas about it. Moreover, the similarity of the results across the occupational groups could indicate that attributing physical complaints to the disaster depended on general factors rather than on occupation-specific factors.</p><p>There are three ways through which disasters may result in long-term health problems: direct physical harm (such as burns, injuries), exposure to psychotraumatic and otherwise stressful events, and exposure to hazardous materials. Long-term health consequences of acute physical harm seems unlikely in this case, because only 0.4% of the exposed police officers and none of the exposed firefighters and hangar workers reported to have been personally injured.</p><p>Regarding psychotraumatic events, the exposure types "rescuing people" and "identification and recovery of or search for victims and human remains" have previously been identified as potentially psychotraumatic by experts on posttraumatic stress disorder [<xref ref-type="bibr" rid="B7">7</xref>]. These tasks were statistically significantly associated with attribution among firefighters and police officers, respectively. Post-hoc analyses nevertheless showed that the long-term prevalence of high levels of posttraumatic stress symptoms was only about 6% among firefighters and police officers and that inclusion of these symptoms in the multivariate models did not essentially change the associations between attribution and types of exposure (data not shown). Thus, if any, psychotrauma probably only plays a minor role in the attribution process in this case.</p><p>With respect to exposure to hazardous materials, retrospective risk evaluations predicted no excess in chronic morbidity due to noxious exposures related to air disaster [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Moreover, previous comparisons of these exposed workers and their colleagues who were not exposed to the air disaster revealed no evidence for disaster-related pathological processes based on extensive clinical analysis of blood and urine samples [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. The comparison of exposed and nonexposed workers did show that exposed workers reported various physical symptoms statistically significantly more often [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. These previously reported results therefore resemble a phenomenon commonly referred to as "unexplained physical symptoms". Such phenomena have also been demonstrated in civilian and military populations after disastrous events and after perceived exposure to hazardous materials [<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. The absence of epidemiological evidence for disaster-related pathological processes that could explain the excess physical symptoms among exposed workers, however, does not imply that the worker's appraisal of a relationship between their physical complaints and the disaster is unjust or fictitious.</p><p>One likely candidate for a general factor affecting health perception and attribution is the complex aftermath of this disaster. The aftermath was characterized by numerous media reports and public discussions on alleged exposure to hazardous materials and health consequences [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]. A variety of hazardous exposures sequentially emerged in the public debate and were coined as explanation for any type of post-disaster health problem. The rumors were most probably also discussed among the exposed rescue workers and they could have affected health perception and attribution of health complaints among exposed workers in two ways. Firstly, they might have contributed to perceiving the disaster as a health threat. Previous studies have argued that considering an environmental factor as harmful to health is important for subjective health. For example, in a study after the Chernobyl disaster, risk perception was suggested to play a mediating role between (perceived) exposure and subjective health problems [<xref ref-type="bibr" rid="B36">36</xref>]. Also, in comparisons of residents of potentially contaminated and control areas, residents who consider the contamination as harmful to health are the once that most often report lower levels of subjective health [<xref ref-type="bibr" rid="B37">37</xref>,<xref ref-type="bibr" rid="B38">38</xref>].</p><p>Secondly, the rumors and speculations might have contributed to sustained uncertainty. In an attempt to reduce such uncertainty, workers may have been more inclined to refer to the actions of "similar others", i.e. symptomatic colleagues who attributed health problems to the disaster, to decide how to act themselves [<xref ref-type="bibr" rid="B14">14</xref>]. Vastermans et al previously suggested that "there is a strong relationship between the symptomatology seen in the aftermath of disasters and medically unexplained physical symptoms in the general population. The only difference is that, after a disaster, the symptoms are attributed to the event" [<xref ref-type="bibr" rid="B3">3</xref>]. As more people adopt that attribution, it might become socially accepted and spread across the affected population. Such spreading of attribution might be enhanced in cases of distrust in official information on noxious exposures, as was the case in the air disaster for some of the affected people. An interesting remaining question is whether awareness of the attributions of others could also lead to spreading of particular symptoms, i.e. that more people perceive the same symptoms if it becomes generally known that similar others have attributed those particular symptoms to the disaster. In this respect, a parallel can also be drawn with so-called Mass Psychogenic/Sociogenic Illness, in which a short-lived spreading of particular physical symptoms in communities has been postulated [<xref ref-type="bibr" rid="B39">39</xref>]. However, no evidence was found for spreading of particular physical symptoms, because the prevalence rates of a wide variety of physical symptoms was higher among exposed compared to nonexposed workers, and because the prevalence rates of various physical symptoms were also positively associated with the extent of attribution to the air disaster in Amsterdam.</p><p>Besides genuinely perceiving a relationship between physical health complaints and the disaster, some exposed workers might alternatively have been motivated to report such a relationship for reasons of secondary gain, such as recognition, attention, and financial compensation. Unfortunately, no data are available to further look into this matter.</p><p>In an attempt to characterize those who attributed physical complaints to the disaster, associations with types of exposure and background characteristics were examined. As discussed above, the similarity of the results across the different occupational groups suggest that general rather than occupation-specific factors contributed to attributing physical complaints to the disaster. Nevertheless, the multivariate logistic regression analyses showed that attribution was significantly associated with some types of exposure within a particular occupational group. This concerned rescuing people among firefighters and almost all types of exposure among police officers, i.e. three of the five tasks, having witnessed the immediate disaster scene, having a close one affected by the disaster, and perceiving the disaster and its aftermath as the worst thing that ever happened to them. Because most of the firefighters reported multiple tasks, whereas most of the police officers reported one of the specified tasks, it may have been easier to detect independent effects of particular tasks among police officers. The difference in sample size between the groups may also have contributed to this. No significant associations between attribution and types of exposure were found among hangar workers, but the analysis was limited to three variables: sorting the wreckage, witnessing the immediate disaster scene, and having a close one affected by the disaster. Only a few hangar workers reported the latter two items.</p><p>A positive association was also found between attribution and the perceived severity of the experience of the disaster, including its aftermath, which was significantly only in the largest of the studied groups; the police officers. It was decided to regard the perceived severity of the disaster primarily as an exposure variable, which putatively encompassed a general appraisal of the experience of their involvement in the disaster and its aftermath. It might alternatively have been regarded as an outcome variable, which would imply that the perceived severity depended on workers' appraisal of the health impact of this disaster. Both these interpretations could partly explain the positive associations seen between attribution and the severity of the disaster experience.</p><p>The analyses also revealed that attribution was not significantly associated with the background characteristics age, level of education, and sex. Sex could only be taken in consideration for police officers, because all the included firefighters and hangar workers were male. The results regarding age and sex are not in line with those of Stuart et al. (2003) who found that among veterans of the first Gulf War, females and those who were older (age 32 to 61 years) were more likely to report belief in exposure to terrorist agents (nerve or mustard gas) [<xref ref-type="bibr" rid="B40">40</xref>]. In that study, belief in exposure to terrorist agents was also associated with degree of exposure, i.e. reporting more exposures (non-nerve or mustard gas) to potentially toxic agents and traumatic events.</p><p>Donker et al. (2002) previously reported on a self-selected group (n = 553) of residents, rescue workers and others affected by the air disaster in Amsterdam [<xref ref-type="bibr" rid="B6">6</xref>]. On their own initiative, these individuals called a toll free call centre (in June-July 1998) to report the health complaints they attributed to this disaster. Of the three physical complaints that were most frequently attributed to the disaster in the present study, fatigue and dry skin were also in the top ten of spontaneously reported health complaints at the call centre (reported by 45% and 13% of the callers, respectively). For 3% and 15% of these two symptoms, respectively, a relationship between the disaster and these particular symptoms was considered to be realistic according to the general practitioners of the callers with these complaints.</p><p>In the present study, no clinical judgment of the perceived relationship between physical complaints and the air disaster is available. However, irrespective of the credibility from a clinical perspective, the causal attributions of exposed workers could affect the prognosis of post-disaster health complaints [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B30">30</xref>], functioning, and the utilization of health care [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. For example, the prevalence of role-limitations due to physical problems was positively and significantly associated with the strength of attribution to the disaster in each of the occupational groups (data not shown). It could thus be relevant to establish whether people involved in disasters (or other events with perceived exposure) attribute any health complaints to that disaster, because they may need a different approach in after care programs. Due to the cross-sectional nature of the present study, it was unfortunately not possible to assess the longitudinal course of health complaints, attribution thereof, and the health care used for them.</p><p>The strength of the present study is that it is based on a historically-defined study population consisting of all the rescue workers who were occupationally exposed to the disaster, irrespective of their health status. Therefore it tentatively provides a representative estimate of the prevalence of long-term physical complaints (72%) and attribution thereof among all the exposed professional firefighters, police officers and hangar workers. In total, 32% of all the exposed workers, and 45% of the exposed workers with physical complaints, attributed their complaints to the disaster, including its aftermath.</p><p>Some limitations should also be mentioned. First of all, as in all cross-sectional studies, the study design precludes drawing inferences on the direction or causality of the associations, for example regarding the positive associations between attribution and the severity of physical health complaints as well as the perceived severity of the disaster experience.</p><p>Secondly, recall or reporting bias may have biased the associations between the self-reported types of exposure and attribution. Furthermore, the types of exposure are likely to be partly inter-related, e.g. rescuers probably also witnessed the immediate disaster scene. Therefore, in addition to univariate associations with attribution, the types of exposure were entered in a multivariate model from which those with P > 0.10 were eliminated in a step-wise, backward manner. The associations that were found univariately remained essentially the same in the multivariate analysis, thus indicating independent associations between these types of exposure and attribution.</p><p>A final point of attention is that, although almost all data were virtually complete, the data on the item "which type of physical complaints" the workers attributed to the disaster was only available for 18%, 99.6% and 53% of the firefighters, police officers and hangar workers, respectively, who attributed any physical complaints to the disaster and its aftermath. This was due to administrative difficulties at the start of the data collection. While the data on this item might be biased by its incompleteness, there is no reason to believe that the workers who were assessed in the first part would have attributed a different type of physical symptom to the disaster than the workers who were assessed in the last part of the data collection period. In addition, the top three of most frequently attributed physical complaints was the same while the completion rate varied considerably across the three occupational groups.</p></sec><sec><title>Conclusion</title><p>In conclusion, this study provides further cross-sectional evidence for the role of causal attribution in post-disaster subjective physical health problems, which lack sufficient clinical or toxicological explanation. After on average 8.5 years, almost a third (32%) of all the exposed workers, and almost half (45%) of the exposed workers with physical complaints, attributed physical complaints to the disaster, including its aftermath. The similarity of the results across the occupational groups suggests that this attribution process was a general rather than an occupation-specific phenomenon, tentatively fed by the rumors on noxious exposures and health consequences in the aftermath of the disaster. Longitudinal studies are needed to determine whether attribution of post-disaster health complaints leads to persistence of health complaints and health care utilization.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>All authors participated in the multidisciplinary ESADA project team of the EMGO Institute, provided comments on the draft versions and approved the final manuscript. In addition, PS drafted the manuscript and performed the statistical analyses; NS supervised the second part of the ESADA; JT contributed to the design and supervised the statistical analyses; AH coordinated the acquisition of data and supervised the first part of the ESADA; AW contributed to the statistical analysis of posttraumatic stress symptoms and the Checklist Individual Strength; WM contributed to the design and is Vice-President of the ESADA project team; and TS conceived of the study, and participated in its design and coordination as President of the ESADA project team.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/142/prepub"/></p></sec>
|
Benchmarking pK<sub>a </sub>prediction
|
<sec><title>Background</title><p>pK<sub>a </sub>values are a measure of the protonation of ionizable groups in proteins. Ionizable groups are involved in intra-protein, protein-solvent and protein-ligand interactions as well as solubility, protein folding and catalytic activity. The pK<sub>a </sub>shift of a group from its intrinsic value is determined by the perturbation of the residue by the environment and can be calculated from three-dimensional structural data.</p></sec><sec><title>Results</title><p>Here we use a large dataset of experimentally-determined pK<sub>a</sub>s to analyse the performance of different prediction techniques. Our work provides a benchmark of available software implementations: MCCE, MEAD, PROPKA and UHBD. Combinatorial and regression analysis is also used in an attempt to find a consensus approach towards pK<sub>a </sub>prediction. The tendency of individual programs to over- or underpredict the pK<sub>a </sub>value is related to the underlying methodology of the individual programs.</p></sec><sec><title>Conclusion</title><p>Overall, PROPKA is more accurate than the other three programs. Key to developing accurate predictive software will be a complete sampling of conformations accessible to protein structures.</p></sec>
|
<contrib id="A1" equal-contrib="yes" corresp="yes" contrib-type="author"><name><surname>Davies</surname><given-names>Matthew N</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" equal-contrib="yes" contrib-type="author"><name><surname>Toseland</surname><given-names>Christopher P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Moss</surname><given-names>David S</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Flower</surname><given-names>Darren R</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Biochemistry
|
<sec><title>Background</title><p>A proper understanding of protein pK<sub>a </sub>values is essential to a proper understanding of pH-dependent characteristics of protein function. If the pK<sub>a </sub>of a particular group is known then one can determine its protonation state at a given pH, helping to determine several important properties including protein solubility, protein folding and catalytic activity. Knowledge of the pK<sub>a </sub>values of the residues of an active site can help to identify the reaction mechanism of an enzyme or aid in the interpretation of experimental results [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B4">4</xref>]. The pK<sub>a </sub>value is -log<sub>10</sub>(K<sub>a</sub>) where K<sub>a</sub>, is the ionization constant, a measure of a titratable group's ability to donate a proton:</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2091-7-18-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msub><mml:mi>K</mml:mi><mml:mi>a</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mo stretchy="false">[</mml:mo><mml:msup><mml:mi>H</mml:mi><mml:mo>+</mml:mo></mml:msup><mml:mo stretchy="false">]</mml:mo><mml:mo stretchy="false">[</mml:mo><mml:msup><mml:mi>A</mml:mi><mml:mo>−</mml:mo></mml:msup><mml:mo stretchy="false">]</mml:mo></mml:mrow><mml:mrow><mml:mo stretchy="false">[</mml:mo><mml:mi>H</mml:mi><mml:mi>A</mml:mi><mml:mo stretchy="false">]</mml:mo></mml:mrow></mml:mfrac><mml:mtext>     </mml:mtext><mml:mrow><mml:mo>(</mml:mo><mml:mn>1</mml:mn><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGlbWsdaWgaaWcbaGaemyyaegabeaakiabg2da9maalaaabaGaei4waSLaemisaG0aaWbaaSqabeaacqGHRaWkaaGccqGGDbqxcqGGBbWwcqWGbbqqdaahaaWcbeqaaiabgkHiTaaakiabc2faDbqaaiabcUfaBjabdIeaijabdgeabjabc2faDbaacaWLjaGaaCzcamaabmaabaGaeGymaedacaGLOaGaayzkaaaaaa@4224@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>The pK<sub>a </sub>value is therefore equal to the pH when there is an equal concentration of the protonated and deprotonated groups in solution. Each residue with a titratable group has a model or 'intrinsic' pK<sub>a </sub>value, defined as the pK<sub>a </sub>value when all the other groups are fixed in their neutral state. Ionizable groups may be divided into acidic, which are neutral in their protonated state, and basic, which are positively charged in their protonated state. The protonated and the non-protonated forms of a residue can be very different chemically. In the case of His, the protonated form is hydrophilic and positively charged while the non-protonated form has a hydrophobic and aromatic character. Consequently the nature of the interaction made by an ionizable group may differ significantly at a pH above or below the pK<sub>a</sub>.</p><p>Table <xref ref-type="table" rid="T1">1</xref> shows the intrinsic or 'model' pK<sub>a </sub>values for all protein titratable groups [<xref ref-type="bibr" rid="B5">5</xref>]. However, in real protein-solvent systems, interactions between a residue and its environment will cause the titratable group's pK<sub>a </sub>value to deviate from that of the model. Hence the intrinsic pK<sub>a </sub>value, pK<sub>Model</sub>, combined with the environmental perturbation, ΔpK<sub>a</sub>, describes the real pK<sub>a </sub>value of a group [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B9">9</xref>].</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Model pK<sub>a </sub>values for all protein basic and acidic titratable groups. See reference 5.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td></td><td align="center" colspan="2"><bold>No. in Study</bold></td><td></td></tr><tr><td align="center"><bold>Titratable Group</bold></td><td align="center"><bold>pK<sub><bold>model </bold></sub>Value</bold></td><td align="center"><bold>LARGE</bold></td><td align="center"><bold>SMALL</bold></td><td align="center">Mean pK<sub><bold>exp </bold></sub>Value</td></tr></thead><tbody><tr><td align="center">N-Termini</td><td align="center">7.5</td><td align="center">-</td><td align="center">-</td><td align="center">-</td></tr><tr><td align="center">C-Termini</td><td align="center">3.8</td><td align="center">-</td><td align="center">-</td><td align="center">-</td></tr><tr><td align="center">Arg</td><td align="center">12</td><td align="center">1</td><td align="center">1</td><td align="center">-</td></tr><tr><td align="center">Asp</td><td align="center">4</td><td align="center">143</td><td align="center">112</td><td align="center">3.5</td></tr><tr><td align="center">Cys</td><td align="center">9.5</td><td align="center">11</td><td align="center">4</td><td align="center">6.6</td></tr><tr><td align="center">Glu</td><td align="center">4.4</td><td align="center">126</td><td align="center">105</td><td align="center">4.3</td></tr><tr><td align="center">His</td><td align="center">6.3</td><td align="center">130</td><td align="center">24</td><td align="center">6.4</td></tr><tr><td align="center">Lys</td><td align="center">10.4</td><td align="center">57</td><td align="center">23</td><td align="center">9.6</td></tr><tr><td align="center">Tyr</td><td align="center">10</td><td align="center">26</td><td align="center">16</td><td align="center">9.5</td></tr></tbody></table></table-wrap><p><italic>pK</italic><sub><italic>a </italic></sub>= <italic>pK</italic><sub><italic>Model </italic></sub>+ Δ<italic>pK</italic><sub><italic>a </italic></sub>      (2)</p><p>The pK<sub>a </sub>shift caused by the environment is not easily quantified. This is especially true of ionizable residues within protein active sites as they often have markedly higher or lower values than the intrinsic pK<sub>a </sub>[<xref ref-type="bibr" rid="B5">5</xref>]. The three main factors that contribute towards environmental perturbation of the pK<sub>a </sub>value are inter-molecular hydrogen bonding, the desolvation effect and Coulombic interactions. Previous studies have identified hydrogen bonding as the most important determinant of pK<sub>a </sub>values [<xref ref-type="bibr" rid="B6">6</xref>]. The hydrogen bonding strength is both distance and angle dependent and therefore the extent of the perturbation is heavily dependent on the position of the interacting residues relative to each other. This is less of a factor with side chain hydrogen bonds than with main chain as the former is more flexible and therefore more likely to adopt an optimal orientation for hydrogen bond interactions. The desolvation effect is also important; this describes the energy that is required to move a group from a state of full solvation to a position within the folded protein. Desolvation effects within the protein interior preferentially increases the energies of the negatively-charged, base forms, which will increase the pK<sub>a </sub>value, while in the case of His, Lys and Arg, the desolvation preferentially increases the energy of the positively-charged, acid forms, which will decrease the pK<sub>a </sub>values. The extent of the shift is dependent on the degree to which the group is buried within the protein. The third of the major factors, which may cause a pK<sub>a </sub>shift, are Coulombic interactions between ionizable groups. The pair-wise interactions are dependent on the charges of the respective groups, but also on their location as only residues that are buried produce significant charge-charge interactions.</p><p>It is possible to predict the pK<sub>a </sub>value of a given protein residue from three-dimensional structural data. The pK<sub>a </sub>shift may be calculated from the difference in energy between the group's charged and neutral forms and added to the pK<sub>model </sub>value to estimate its true value. Several different algorithms have been developed to generate predicted pK<sub>a </sub>values based on structural data.</p><p>The majority of papers which have assessed the reliability of pK<sub>a </sub>predictive algorithms have only examined a limited number of proteins, making an evaluation of their accuracy very difficult [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B10">10</xref>-<xref ref-type="bibr" rid="B13">13</xref>]. The largest of these [<xref ref-type="bibr" rid="B13">13</xref>] looked at 260 experimental pK<sub>a </sub>values taken from 41 proteins. Here we use a large pK<sub>a </sub>dataset of 100 proteins, which is more than double that of the most extensive previous paper [<xref ref-type="bibr" rid="B13">13</xref>], to analyse the predictive capabilities of the MCCE [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>], MEAD [<xref ref-type="bibr" rid="B16">16</xref>], PROPKA [<xref ref-type="bibr" rid="B17">17</xref>] and UHBD [<xref ref-type="bibr" rid="B18">18</xref>] programs. The programs differ in their methodology and we assessed the merits of each. An enhanced approach to the problem of pK<sub>a </sub>prediction is proposed.</p></sec><sec><title>Results and discussion</title><p>Technical difficulties with the UHBD program, due to errors in the protonation of histidine residues, prevented the successful processing of all 100 proteins; in total only 43 could be completed. This lead to the creation of two separate datasets, the Large dataset (containing 492 residues and excluding the UHBD program) and the Small Dataset (containing 280 residues and including the UHBD program). Several of the programs produced outliers, in some cases outside of the physically possible pH 0–14 range. Outliers were removed from the dataset using a variety of different parameters to see the effect upon the overall accuracy of prediction Datasets were generated where all predicted values were lesser or greater than the intrinsic values by 3, 5, 7 and 10 pH units were removed as well a dataset where only physically possible values were included. This data is presented in Table <xref ref-type="table" rid="T2">2</xref> and Table <xref ref-type="table" rid="T3">3</xref>. It may been seen from the data that in general the best results were obtained by using values within a range of 5 pH units. Using those parameters, 89 residues were removed from the Large dataset leaving 403 and 39 residues were removed from the Small dataset leaving 241 residues. It is unlikely that the pKa of a titratable residues can deviate more than 5 pH units from the residue's intrinsic value (see Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Overview of the prediction accuracy of the Large Dataset (404 Residues) I. The table shows the RMSD values for each of the residues from the whole dataset and the dataset following removal of non-physical values and of all outliers outside of a range of 3, 5, 7 and 10 pH units. Figures marked in bold indicate significant results (<italic>P </italic>= 0.05).</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>AMBER COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td><td align="center"><bold>PARSE COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td></tr></thead><tbody><tr><td align="center"><bold>ARG</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">2.698</td><td align="center">1.928</td><td align="center">2.513</td><td align="center">1.495</td><td align="center">1.354</td><td align="center">1.273</td><td align="center">3.743</td><td align="center">1.729</td><td align="center">2.251</td><td align="center">1.837</td><td align="center">1.25</td><td align="center">1.106</td></tr><tr><td align="center"><bold>CYS</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.885</td><td align="center">1.466</td><td align="center">1.616</td><td align="center">1.62</td><td align="center">1.42</td><td align="center">1.026</td><td align="center">3.175</td><td align="center">1.48</td><td align="center">1.841</td><td align="center">1.681</td><td align="center">1.442</td><td align="center">1.034</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">3.691</td><td align="center">2.171</td><td align="center">3.016</td><td align="center">2.281</td><td align="center">2.022</td><td align="center">1.417</td><td align="center">3.691</td><td align="center">2.122</td><td align="center">2.997</td><td align="center">2.293</td><td align="center">1.993</td><td align="center">1.465</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">1.244</td><td align="center">1.263</td><td align="center">1.263</td><td align="center">1.263</td><td align="center">1.122</td><td align="center">0.84</td><td align="center">25.78</td><td align="center">1.162</td><td align="center">1.162</td><td align="center">1.162</td><td align="center">0.991</td><td align="center">0.741</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">2.739</td><td align="center">2.195</td><td align="center">2.195</td><td align="center">2.195</td><td align="center">1.939</td><td align="center">1.516</td><td align="center">3.163</td><td align="center">2.143</td><td align="center">2.143</td><td align="center">2.143</td><td align="center">1.882</td><td align="center">1.481</td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td></td><td align="center"><bold>MCCE COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td><td align="center"><bold>PROPKA COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td align="center"><bold>ARG</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">2.024</td><td align="center">1.774</td><td align="center">2.032</td><td align="center">1.79</td><td align="center">1.534</td><td align="center">1.242</td><td align="center">1.301</td><td align="center">1.041</td><td align="center">1.313</td><td align="center">1.301</td><td align="center"><bold>0.934</bold></td><td align="center">0.806</td></tr><tr><td align="center"><bold>CYS</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.642</td><td align="center">1.255</td><td align="center">1.335</td><td align="center">1.296</td><td align="center">1.259</td><td align="center">0.97</td><td align="center">1.011</td><td align="center">0.851</td><td align="center">0.994</td><td align="center">0.997</td><td align="center"><bold>0.849</bold></td><td align="center">0.611</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">4.503</td><td align="center">1.736</td><td align="center">2.598</td><td align="center">2.104</td><td align="center"><bold>1.522</bold></td><td align="center">1.718</td><td align="center">1.865</td><td align="center">1.586</td><td align="center">1.819</td><td align="center">1.631</td><td align="center"><bold>1.530</bold></td><td align="center">1.551</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">1.137</td><td align="center">1.125</td><td align="center">1.125</td><td align="center">1.125</td><td align="center">1.005</td><td align="center">1.129</td><td align="center">0.417</td><td align="center">0.412</td><td align="center">0.412</td><td align="center">0.412</td><td align="center"><bold>2.600</bold></td><td align="center">0.423</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">5.426</td><td align="center">2.643</td><td align="center">2.634</td><td align="center">2.643</td><td align="center">1.668</td><td align="center">1.593</td><td align="center">2.225</td><td align="center">1.551</td><td align="center">1.551</td><td align="center">1.551</td><td align="center"><bold>1.001</bold></td><td align="center">1.049</td></tr></tbody></table><table-wrap-foot><p>Outliers </p><p>COMPLETE 0 removed</p><p>TRUE 52 removed</p><p>WITHIN 10 23 removed</p><p>WITHIN 7 27 removed</p><p>WITHIN 5 89 removed</p><p>WITHIN 3 110 removed</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Overview of the prediction accuracy of the Small Dataset (242 Residues) I. The table shows the RMSD values for each of the residues from the whole dataset and the dataset following removal of non-physical values and of all outliers outside of a range of 3, 5, 7 and 10 pH units. Figures marked in bold indicate significant results (<italic>P = 0.05</italic>).</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>AMBER COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td><td align="center"><bold>PARSE </bold><bold>COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td></tr></thead><tbody><tr><td align="center"><bold>ARG</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">2.078</td><td align="center">1.787</td><td align="center">2.079</td><td align="center">2.077</td><td align="center">1.691</td><td align="center">1.781</td><td align="center">3.774</td><td align="center">1.582</td><td align="center">2.142</td><td align="center">1.912</td><td align="center">1.53</td><td align="center">1.267</td></tr><tr><td align="center"><bold>CYS</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.641</td><td align="center">1.416</td><td align="center">1.603</td><td align="center">1.603</td><td align="center">1.294</td><td align="center">1.047</td><td align="center">3.041</td><td align="center">1.474</td><td align="center">1.71</td><td align="center">1.71</td><td align="center">1.334</td><td align="center">1.051</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">2.786</td><td align="center">1.689</td><td align="center">1.689</td><td align="center">1.689</td><td align="center">1.347</td><td align="center">1.118</td><td align="center">2.736</td><td align="center">1.809</td><td align="center">1.81</td><td align="center">1.81</td><td align="center">1.488</td><td align="center">1.402</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">1.291</td><td align="center">1.29</td><td align="center">1.291</td><td align="center">1.291</td><td align="center">1.291</td><td align="center">1.291</td><td align="center">1.278</td><td align="center">1.278</td><td align="center">1.278</td><td align="center">1.278</td><td align="center">1.278</td><td align="center">1.278</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">2.06</td><td align="center">1.297</td><td align="center">1.297</td><td align="center">1.297</td><td align="center">1.933</td><td align="center">0.766</td><td align="center">2.368</td><td align="center">1.262</td><td align="center">1.262</td><td align="center">1.262</td><td align="center">1.871</td><td align="center">0.792</td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td></td><td align="center"><bold>MCCE COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td><td align="center"><bold>UHBD </bold><bold>COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td align="center"><bold>ARG</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">1.915</td><td align="center">1.735</td><td align="center">1.921</td><td align="center">1.731</td><td align="center">1.319</td><td align="center">1.419</td><td align="center">0.95</td><td align="center">0.824</td><td align="center">0.838</td><td align="center">0.842</td><td align="center">0.89</td><td align="center">0.641</td></tr><tr><td align="center"><bold>CYS</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.575</td><td align="center">1.185</td><td align="center">1.237</td><td align="center">1.237</td><td align="center">1.188</td><td align="center">0.893</td><td align="center">0.508</td><td align="center">0.478</td><td align="center">0.493</td><td align="center">0.493</td><td align="center"><bold>0.442</bold></td><td align="center">0.395</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">1.985</td><td align="center">1.584</td><td align="center">1.584</td><td align="center">1.584</td><td align="center">1.056</td><td align="center">1.593</td><td align="center">0.634</td><td align="center">0.453</td><td align="center">0.453</td><td align="center">0.453</td><td align="center"><bold>0.494</bold></td><td align="center">0.428</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">1.152</td><td align="center">1.152</td><td align="center">1.152</td><td align="center">1.152</td><td align="center">1.152</td><td align="center">1.152</td><td align="center">0.412</td><td align="center">0.412</td><td align="center">0.412</td><td align="center">0.412</td><td align="center">0.412</td><td align="center">0.412</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">6.027</td><td align="center">1.373</td><td align="center">1.373</td><td align="center">1.373</td><td align="center">1.456</td><td align="center">1.419</td><td align="center">0.687</td><td align="center">0.582</td><td align="center">0.582</td><td align="center">0.582</td><td align="center">0.61</td><td align="center">0.631</td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td></td><td align="center"><bold>PROPKA COMPLETE</bold></td><td align="center"><bold>TRUE</bold></td><td align="center"><bold>WITHIN10</bold></td><td align="center"><bold>WITHIN7</bold></td><td align="center"><bold>WITHIN5</bold></td><td align="center"><bold>WITHIN3</bold></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td colspan="13"><hr></hr></td></tr><tr><td align="center"><bold>ARG</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">1.827</td><td align="center">1.826</td><td align="center">1.806</td><td align="center">1.809</td><td align="center"><bold>0.879</bold></td><td align="center">0.745</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>CYS</bold></td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td align="center">*</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">0.987</td><td align="center">0.773</td><td align="center">0.959</td><td align="center">0.959</td><td align="center">0.781</td><td align="center">0.632</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">2.235</td><td align="center">2.11</td><td align="center">2.11</td><td align="center">2.11</td><td align="center">1.724</td><td align="center">2.172</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">0.394</td><td align="center">0.394</td><td align="center">0.394</td><td align="center">0.394</td><td align="center"><bold>0.394</bold></td><td align="center">0.394</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">1.533</td><td align="center">0.992</td><td align="center">0.991</td><td align="center">0.991</td><td align="center"><bold>0.572</bold></td><td align="center">1.011</td><td></td><td></td><td></td><td></td><td></td><td></td></tr></tbody></table><table-wrap-foot><p>Outliers </p><p>COMPLETE 0 removed</p><p>TRUE 25 removed</p><p>WITHIN 10 11 removed</p><p>WITHIN 7 13 removed</p><p>WITHIN 5 34 removed</p><p>WITHIN 3 54 removed</p></table-wrap-foot></table-wrap><p>The majority of the outliers in both datasets were generated by the MEAD program, particularly when the PARCE force field was used. Considerably more residues are present within the +/- 1 unit bands for MCCE, UHBD and PROPKA. Thus there is a clear division between the performance of MEAD and that of the other programs. The same trend may be seen in the Root Mean Squared Deviation (RMSD) values (Table <xref ref-type="table" rid="T2">2</xref>, <xref ref-type="table" rid="T3">3</xref>). PROPKA is more accurate for Asp, Glu, Lys and Tyr with RMSD values of 0.934, 0.849, 0.260 and 1.001 respectively. His is more accurately predicted by MCCE with an RMSD of 1.522. With respect to the Small dataset in Table <xref ref-type="table" rid="T3">3</xref>, PROPKA is the best predictor for all residues except Glu and His, where UHBD performs best: RMSD of 0.442 and 0.494 respectively. The overall accuracy of each program to a level of <0.5 pK<sub>a </sub>units is 27% AMBER, 34% PARSE, 42% MCCE, 40% UHBD (242 dataset) and 48% PROPKA. When the error range is increased to <1 unit, the difference between the programs is more distinct: 56% AMBER, 56% PARSE, 71% MCCE, 67% UHBD (Small dataset) and 81% PROPKA (Table <xref ref-type="table" rid="T4">4</xref>, <xref ref-type="table" rid="T5">5</xref>). Scatter plots for each program are shown in Figure <xref ref-type="fig" rid="F1">1</xref>.</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Overview of the prediction accuracy of the Large Dataset (404 Residues) II Three tables show the accuracy of the predictions to the measured pK<sub>exp </sub>within the ranges of <2 to <0.5. This is taken as the number of residues predicted within each range. Figure marked in bold indicate significant results (<italic>P = 0.05</italic>).</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>TOTAL (404)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr></thead><tbody><tr><td align="center"><2</td><td align="center">322</td><td align="center">80</td><td align="center">320</td><td align="center">79</td><td align="center">357</td><td align="center">88</td><td align="center">371</td><td align="center"><bold>92</bold></td></tr><tr><td align="center"><1.5</td><td align="center">283</td><td align="center">70</td><td align="center">282</td><td align="center">70</td><td align="center">328</td><td align="center">81</td><td align="center">354</td><td align="center"><bold>88</bold></td></tr><tr><td align="center"><1</td><td align="center">213</td><td align="center">53</td><td align="center">221</td><td align="center">55</td><td align="center">283</td><td align="center">70</td><td align="center">317</td><td align="center"><bold>78</bold></td></tr><tr><td align="center"><0.5</td><td align="center">108</td><td align="center">27</td><td align="center">139</td><td align="center">34</td><td align="center">168</td><td align="center">42</td><td align="center">195</td><td align="center"><bold>48</bold></td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center"><bold>SURFACE (337)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">285</td><td align="center">85</td><td align="center">284</td><td align="center">84</td><td align="center">311</td><td align="center">92</td><td align="center">329</td><td align="center"><bold>98</bold></td></tr><tr><td align="center"><1.5</td><td align="center">255</td><td align="center">76</td><td align="center">253</td><td align="center">75</td><td align="center">290</td><td align="center">86</td><td align="center">317</td><td align="center"><bold>94</bold></td></tr><tr><td align="center"><1</td><td align="center">196</td><td align="center">58</td><td align="center">200</td><td align="center">59</td><td align="center">248</td><td align="center">74</td><td align="center">285</td><td align="center"><bold>85</bold></td></tr><tr><td align="center"><0.5</td><td align="center">98</td><td align="center">29</td><td align="center">132</td><td align="center">39</td><td align="center">151</td><td align="center">45</td><td align="center">184</td><td align="center"><bold>55</bold></td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center"><bold>BURIED (66)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">37</td><td align="center">56</td><td align="center">36</td><td align="center">55</td><td align="center">46</td><td align="center"><bold>70</bold></td><td align="center">42</td><td align="center">64</td></tr><tr><td align="center"><1.5</td><td align="center">28</td><td align="center">42</td><td align="center">29</td><td align="center">44</td><td align="center">38</td><td align="center"><bold>58</bold></td><td align="center">37</td><td align="center">56</td></tr><tr><td align="center"><1</td><td align="center">17</td><td align="center">26</td><td align="center">21</td><td align="center">32</td><td align="center">35</td><td align="center"><bold>53</bold></td><td align="center">32</td><td align="center">48</td></tr><tr><td align="center"><0.5</td><td align="center">10</td><td align="center">15</td><td align="center">7</td><td align="center">11</td><td align="center">17</td><td align="center"><bold>26</bold></td><td align="center">11</td><td align="center">17</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Overview of the prediction accuracy of the Small Dataset (242 Residues) II. Three tables show the accuracy of the predictions to the measured pK<sub>exp</sub>. This is taken as the number of residues predicted within each range. Figures marked in bold indicate significant results (<italic>P = 0.05</italic>).</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>TOTAL (242)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr></thead><tbody><tr><td align="center"><2</td><td align="center">195</td><td align="center">81</td><td align="center">191</td><td align="center">79</td><td align="center">216</td><td align="center">89</td><td align="center">225</td><td align="center">93</td><td align="center">230</td><td align="center">95</td></tr><tr><td align="center"><1.5</td><td align="center">174</td><td align="center">72</td><td align="center">171</td><td align="center">71</td><td align="center">201</td><td align="center">83</td><td align="center">209</td><td align="center">86</td><td align="center">220</td><td align="center">91</td></tr><tr><td align="center"><1</td><td align="center">131</td><td align="center">54</td><td align="center">135</td><td align="center">56</td><td align="center">172</td><td align="center">71</td><td align="center">161</td><td align="center">67</td><td align="center">195</td><td align="center">81</td></tr><tr><td align="center"><0.5</td><td align="center">63</td><td align="center">26</td><td align="center">87</td><td align="center">36</td><td align="center">110</td><td align="center">45</td><td align="center">97</td><td align="center">40</td><td align="center">125</td><td align="center">52</td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><bold>SURFACE (209)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">179</td><td align="center">86</td><td align="center">176</td><td align="center">84</td><td align="center">193</td><td align="center">92</td><td align="center">199</td><td align="center">95</td><td align="center">205</td><td align="center"><bold>98</bold></td></tr><tr><td align="center"><1.5</td><td align="center">163</td><td align="center">78</td><td align="center">159</td><td align="center">76</td><td align="center">181</td><td align="center">87</td><td align="center">186</td><td align="center">89</td><td align="center">199</td><td align="center"><bold>95</bold></td></tr><tr><td align="center"><1</td><td align="center">126</td><td align="center">60</td><td align="center">125</td><td align="center">60</td><td align="center">156</td><td align="center">75</td><td align="center">152</td><td align="center">73</td><td align="center">181</td><td align="center"><bold>87</bold></td></tr><tr><td align="center"><0.5</td><td align="center">60</td><td align="center">29</td><td align="center">83</td><td align="center">40</td><td align="center">100</td><td align="center">48</td><td align="center">94</td><td align="center">45</td><td align="center">122</td><td align="center"><bold>58</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><bold>BURIED (33)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">16</td><td align="center">48</td><td align="center">15</td><td align="center">45</td><td align="center">23</td><td align="center">70</td><td align="center">26</td><td align="center"><bold>79</bold></td><td align="center">25</td><td align="center">76</td></tr><tr><td align="center"><1.5</td><td align="center">11</td><td align="center">33</td><td align="center">12</td><td align="center">36</td><td align="center">20</td><td align="center">61</td><td align="center">23</td><td align="center"><bold>70</bold></td><td align="center">21</td><td align="center">64</td></tr><tr><td align="center"><1</td><td align="center">5</td><td align="center">15</td><td align="center">10</td><td align="center">30</td><td align="center">16</td><td align="center"><bold>48</bold></td><td align="center">9</td><td align="center">27</td><td align="center">14</td><td align="center">42</td></tr><tr><td align="center"><0.5</td><td align="center">3</td><td align="center">9</td><td align="center">4</td><td align="center">12</td><td align="center">16</td><td align="center"><bold>30</bold></td><td align="center">3</td><td align="center">9</td><td align="center">3</td><td align="center">9</td></tr></tbody></table></table-wrap><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Correlation plots for the individual programs</bold>. The bold line indicates perfect prediction (pK<sub>pred </sub>= pK<sub>exp</sub>). The outer lines indicate +/- 1 unit from the pK<sub>exp</sub>.</p></caption><graphic xlink:href="1471-2091-7-18-1"/></fig><p>From a previous study [<xref ref-type="bibr" rid="B19">19</xref>], 39 carboxyl residues found within protein active sites were selected. These are shown in Table <xref ref-type="table" rid="T6">6</xref>[<xref ref-type="bibr" rid="B20">20</xref>-<xref ref-type="bibr" rid="B30">30</xref>]. The 27 Asp and 12 Glu residues have experimental values that differ from the model pK<sub>a </sub>value by at least 1 unit. Values for Asp and Glu range from 2.0 – 9.9 and 2.1 – 6.7 respectively. PROPKA and UHBD are distinct within the <0.5 and <1 unit error bands (Table <xref ref-type="table" rid="T7">7</xref>), with PROPKA performing best with an accuracy of 66.67% within the <1 unit band. However a large discrepancy exists between the <1 and <0.5 bands for all of the programs, with an approximate 50% drop in accuracy. For the Asp residues, PROPKA predicts far better than the other programs, with values of 37.04%, compared to 18.52% for MCCE and 0% for UHBD at the <0.5 level. However, for Glu residues program performance is closer, with values of 25 % for PARSE, MCCE and UHBD and 33% for AMBER and PROPKA at the <0.5 level. When the error level is extended to <1, PROPKA is far better, with a value of 83.33% compared to its nearest rivals UHBD, AMBER and PARSE with values of 50%. PROPKA shows an accuracy of 85% to within 1 pK<sub>a </sub>unit for surface residues, whereas the same accuracy is limited to 53% with the MCCE program for buried residues, a considerable reduction. The accuracy values obtained for MCCE were also comparable with those that recently appeared on the program's website, which show an RMSD of 0.77 and an accuracy range of 98% within the <2 pH unit range and 84% accuracy within the <1 pH unit range [<xref ref-type="bibr" rid="B31">31</xref>].</p><table-wrap position="float" id="T6"><label>Table 6</label><caption><p>Carboxyl sites of interest. B = Buried, S = Surface. Figures marked in bold indicate predictions >2 units from the pK<sub>exp</sub>.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>PDB</bold></td><td align="center"><bold>RESIDUE</bold></td><td align="center"><bold>LOCATION</bold></td><td align="center"><bold>AMBER</bold></td><td align="center"><bold>PARSE</bold></td><td align="center"><bold>MCCE</bold></td><td align="center"><bold>UHBD</bold></td><td align="center"><bold>PROPKA</bold></td><td align="center"><bold>pKexp</bold></td></tr></thead><tbody><tr><td align="center">1A2P<sup>20</sup></td><td align="center">ASP-101</td><td align="center">S</td><td align="center"><bold>6.19</bold></td><td align="center">3.56</td><td align="center">2.31</td><td align="center">3.75</td><td align="center">1.20</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-93</td><td align="center">S</td><td align="center"><bold>-0.64</bold></td><td align="center"><bold>-1.38</bold></td><td align="center">1.00</td><td align="center">3.92</td><td align="center">0.69</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-54</td><td align="center">S</td><td align="center">0.74</td><td align="center">0.82</td><td align="center">1.34</td><td align="center">3.57</td><td align="center">2.70</td><td align="center">2.00</td></tr><tr><td></td><td align="center">GLU-73</td><td align="center">B</td><td align="center"><bold>4.66</bold></td><td align="center">4.10</td><td align="center">2.37</td><td align="center"><bold>4.68</bold></td><td align="center">3.11</td><td align="center">2.10</td></tr><tr><td align="center">1A91<sup>21</sup></td><td align="center">ASP-7</td><td align="center">S</td><td align="center">3.99</td><td align="center"><bold>20.29</bold></td><td align="center">4.17</td><td align="center">4.04</td><td align="center">3.87</td><td align="center">5.60</td></tr><tr><td></td><td align="center">ASP-44</td><td align="center">S</td><td align="center">6.00</td><td align="center"><bold>20.22</bold></td><td align="center">5.55</td><td align="center">4.69</td><td align="center">4.19</td><td align="center">5.60</td></tr><tr><td></td><td align="center">ASP-61</td><td align="center">S</td><td align="center">5.52</td><td align="center"><bold>26.20</bold></td><td align="center">5.01</td><td align="center"><bold>4.33</bold></td><td align="center"><bold>4.01</bold></td><td align="center">7.00</td></tr><tr><td></td><td align="center">GLU-2</td><td align="center">S</td><td align="center">4.14</td><td align="center">7.45</td><td align="center">4.53</td><td align="center">4.45</td><td align="center">4.50</td><td align="center">5.50</td></tr><tr><td></td><td align="center">GLU-37</td><td align="center">S</td><td align="center">5.15</td><td align="center"><bold>-20.96</bold></td><td align="center">4.27</td><td align="center">4.66</td><td align="center">4.32</td><td align="center">5.50</td></tr><tr><td align="center">1BEO<sup>22</sup></td><td align="center">ASP-21</td><td align="center">S</td><td align="center"><bold>6.38</bold></td><td align="center"><bold>5.52</bold></td><td align="center">3.11</td><td align="center">3.75</td><td align="center">1.35</td><td align="center">2.50</td></tr><tr><td></td><td align="center">ASP-30</td><td align="center">S</td><td align="center">3.56</td><td align="center">3.80</td><td align="center">4.38</td><td align="center">4.00</td><td align="center">2.64</td><td align="center">2.50</td></tr><tr><td></td><td align="center">ASP-72</td><td align="center">S</td><td align="center">3.84</td><td align="center">3.95</td><td align="center">3.69</td><td align="center">4.34</td><td align="center">3.30</td><td align="center">2.60</td></tr><tr><td align="center">1DE3<sup>23</sup></td><td align="center">GLU-96</td><td align="center">B</td><td align="center"><bold>9.88</bold></td><td align="center"><bold>10.41</bold></td><td align="center">3.53</td><td align="center">5.70</td><td align="center">4.10</td><td align="center">5.10</td></tr><tr><td></td><td align="center">GLU-115</td><td align="center">S</td><td align="center">5.19</td><td align="center">5.23</td><td align="center">3.81</td><td align="center">4.45</td><td align="center">4.50</td><td align="center">4.90</td></tr><tr><td align="center">1KXI<sup>24</sup></td><td align="center">ASP-59</td><td align="center">S</td><td align="center">3.13</td><td align="center">3.90</td><td align="center">2.33</td><td align="center">4.20</td><td align="center">2.49</td><td align="center">2.30</td></tr><tr><td align="center">1LZ3<sup>25</sup></td><td align="center">ASP-18</td><td align="center">S</td><td align="center">3.79</td><td align="center">3.94</td><td align="center"><bold>6.77</bold></td><td align="center">4.05</td><td align="center">3.19</td><td align="center">2.70</td></tr><tr><td></td><td align="center">ASP-48</td><td align="center">S</td><td align="center">3.57</td><td align="center">3.17</td><td align="center"><bold>5.15</bold></td><td align="center">3.99</td><td align="center">2.51</td><td align="center">2.50</td></tr><tr><td></td><td align="center">ASP-66</td><td align="center">S</td><td align="center">0.23</td><td align="center"><bold>-0.31</bold></td><td align="center"><bold>12.38</bold></td><td align="center">3.07</td><td align="center">1.19</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-87</td><td align="center">S</td><td align="center">4.06</td><td align="center">3.96</td><td align="center"><bold>4.89</bold></td><td align="center">3.89</td><td align="center">2.17</td><td align="center">2.10</td></tr><tr><td></td><td align="center">GLU-7</td><td align="center">S</td><td align="center">3.92</td><td align="center">3.44</td><td align="center"><bold>4.73</bold></td><td align="center">4.36</td><td align="center">3.01</td><td align="center">2.70</td></tr><tr><td></td><td align="center">GLU-35</td><td align="center">B</td><td align="center">4.92</td><td align="center">5.23</td><td align="center">7.80</td><td align="center">4.78</td><td align="center">5.40</td><td align="center">6.10</td></tr><tr><td align="center">1RNZ<sup>26</sup></td><td align="center">ASP-14</td><td align="center">B</td><td align="center"><bold>9.08</bold></td><td align="center"><bold>7.86</bold></td><td align="center">3.51</td><td align="center"><bold>4.89</bold></td><td align="center"><bold>-0.62</bold></td><td align="center">2.00</td></tr><tr><td></td><td align="center">GLU-2</td><td align="center">B</td><td align="center"><bold>-1.52</bold></td><td align="center"><bold>-1.48</bold></td><td align="center">1.44</td><td align="center"><bold>5.03</bold></td><td align="center">2.66</td><td align="center">2.80</td></tr><tr><td align="center">1TRS<sup>27</sup></td><td align="center">ASP-26</td><td align="center">B</td><td align="center">6.18</td><td align="center"><bold>4.38</bold></td><td align="center">7.84</td><td align="center"><bold>4.64</bold></td><td align="center"><bold>4.96</bold></td><td align="center">8.10</td></tr><tr><td></td><td align="center">GLU-6</td><td align="center">S</td><td align="center">3.91</td><td align="center">3.92</td><td align="center">4.54</td><td align="center">4.44</td><td align="center">4.50</td><td align="center">4.90</td></tr><tr><td></td><td align="center">GLU-68</td><td align="center">S</td><td align="center">4.27</td><td align="center">4.24</td><td align="center">4.59</td><td align="center">4.55</td><td align="center">4.57</td><td align="center">5.10</td></tr><tr><td align="center">1TRW<sup>27</sup></td><td align="center">ASP-26</td><td align="center">B</td><td align="center">8.23</td><td align="center"><bold>5.24</bold></td><td align="center">8.63</td><td align="center"><bold>4.83</bold></td><td align="center"><bold>5.62</bold></td><td align="center">9.90</td></tr><tr><td></td><td align="center">GLU-68</td><td align="center">S</td><td align="center">5.07</td><td align="center">5.33</td><td align="center">3.55</td><td align="center">4.88</td><td align="center">4.34</td><td align="center">4.90</td></tr><tr><td align="center">1XNB<sup>28</sup></td><td align="center">ASP-11</td><td align="center">S</td><td align="center">1.83</td><td align="center">0.57</td><td align="center">3.44</td><td align="center">3.82</td><td align="center">1.99</td><td align="center">2.50</td></tr><tr><td></td><td align="center">ASP-83</td><td align="center">B</td><td align="center"><bold>6.29</bold></td><td align="center"><bold>7.89</bold></td><td align="center">6.35</td><td align="center"><bold>4.28</bold></td><td align="center">1.36</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-101</td><td align="center">B</td><td align="center"><bold>5.01</bold></td><td align="center">2.94</td><td align="center"><bold>9.96</bold></td><td align="center"><bold>4.28</bold></td><td align="center">1.50</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-106</td><td align="center">S</td><td align="center"><bold>8.72</bold></td><td align="center"><bold>8.94</bold></td><td align="center">3.18</td><td align="center"><bold>4.98</bold></td><td align="center">3.02</td><td align="center">2.70</td></tr><tr><td></td><td align="center">GLU-172</td><td align="center">B</td><td align="center">6.62</td><td align="center">6.42</td><td align="center">5.94</td><td align="center">5.22</td><td align="center">7.32</td><td align="center">6.70</td></tr><tr><td align="center">2OVO<sup>29</sup></td><td align="center">ASP-7</td><td align="center">S</td><td align="center">4.05</td><td align="center">4.01</td><td align="center"><bold>6.25</bold></td><td align="center">3.72</td><td align="center">2.51</td><td align="center">2.50</td></tr><tr><td></td><td align="center">ASP-27</td><td align="center">S</td><td align="center">2.08</td><td align="center">2.32</td><td align="center">2.77</td><td align="center">3.77</td><td align="center">2.39</td><td align="center">2.50</td></tr><tr><td align="center">2RN2<sup>30</sup></td><td align="center">ASP-10</td><td align="center">B</td><td align="center"><bold>4.01</bold></td><td align="center"><bold>3.38</bold></td><td align="center"><bold>8.47</bold></td><td align="center"><bold>3.83</bold></td><td align="center">6.99</td><td align="center">6.10</td></tr><tr><td></td><td align="center">ASP-70</td><td align="center">B</td><td align="center">4.11</td><td align="center">3.55</td><td align="center">3.15</td><td align="center">3.50</td><td align="center">4.10</td><td align="center">2.60</td></tr><tr><td></td><td align="center">ASP-102</td><td align="center">B</td><td align="center"><bold>7.33</bold></td><td align="center"><bold>6.77</bold></td><td align="center">3.00</td><td align="center">3.40</td><td align="center">0.13</td><td align="center">2.00</td></tr><tr><td></td><td align="center">ASP-148</td><td align="center">B</td><td align="center">-1.10</td><td align="center"><bold>-1.31</bold></td><td align="center">0.55</td><td align="center">3.79</td><td align="center"><bold>-0.79</bold></td><td align="center">2.00</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T7"><label>Table 7</label><caption><p>Accuracy of prediction for the carboxyl sites. The accuracy was tested to the <2 to <0.5 ranges. The individual accuracy of the residues is given in the bottom two tables. Figures marked in bold indicate the greatest accuracy.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>TOTAL (242)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr></thead><tbody><tr><td align="center"><2</td><td align="center">26</td><td align="center">66.67</td><td align="center">20</td><td align="center">51.28</td><td align="center">30</td><td align="center">76.92</td><td align="center">30</td><td align="center">76.92</td><td align="center">34</td><td align="center"><bold>87.18</bold></td></tr><tr><td align="center"><1.5</td><td align="center">19</td><td align="center">48.72</td><td align="center">15</td><td align="center">38.46</td><td align="center">25</td><td align="center">64.10</td><td align="center">20</td><td align="center">51.28</td><td align="center">32</td><td align="center"><bold>82.05</bold></td></tr><tr><td align="center"><1</td><td align="center">10</td><td align="center">25.64</td><td align="center">10</td><td align="center">25.64</td><td align="center">17</td><td align="center">43.59</td><td align="center">8</td><td align="center">20.51</td><td align="center">26</td><td align="center"><bold>66.67</bold></td></tr><tr><td align="center"><0.5</td><td align="center">6</td><td align="center">15.38</td><td align="center">4</td><td align="center">10.26</td><td align="center">8</td><td align="center">20.51</td><td align="center">3</td><td align="center">7.69</td><td align="center">14</td><td align="center"><bold>35.90</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><bold>SURFACE (209)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">17</td><td align="center">62.96</td><td align="center">13</td><td align="center">48.15</td><td align="center">10</td><td align="center">70.37</td><td align="center">19</td><td align="center">70.37</td><td align="center">22</td><td align="center"><bold>81.48</bold></td></tr><tr><td align="center"><1.5</td><td align="center">10</td><td align="center">37.04</td><td align="center">8</td><td align="center">29.63</td><td align="center">16</td><td align="center">59.26</td><td align="center">11</td><td align="center">40.74</td><td align="center">20</td><td align="center"><bold>74.07</bold></td></tr><tr><td align="center"><1</td><td align="center">4</td><td align="center">14.81</td><td align="center">4</td><td align="center">14.81</td><td align="center">12</td><td align="center">44.44</td><td align="center">2</td><td align="center">7.41</td><td align="center">16</td><td align="center"><bold>59.26</bold></td></tr><tr><td align="center"><0.5</td><td align="center">2</td><td align="center">7.41</td><td align="center">1</td><td align="center">3.70</td><td align="center">5</td><td align="center">18.52</td><td align="center">0</td><td align="center">0.00</td><td align="center">10</td><td align="center"><bold>37.04</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><bold>BURIED (33)</bold></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>AMBER</bold></td><td></td><td align="center"><bold>PARSE</bold></td><td></td><td align="center"><bold>MCCE</bold></td><td></td><td align="center"><bold>UHBD</bold></td><td></td><td align="center"><bold>PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td><td></td><td align="center"><bold>%</bold></td></tr><tr><td colspan="11"><hr></hr></td></tr><tr><td align="center"><2</td><td align="center">9</td><td align="center">75.00</td><td align="center">9</td><td align="center">75.00</td><td align="center">11</td><td align="center">91.67</td><td align="center">11</td><td align="center">91.67</td><td align="center">12</td><td align="center"><bold>100.00</bold></td></tr><tr><td align="center"><1.5</td><td align="center">9</td><td align="center">75.00</td><td align="center">7</td><td align="center">58.33</td><td align="center">9</td><td align="center">75.00</td><td align="center">9</td><td align="center">75.00</td><td align="center">12</td><td align="center"><bold>100.00</bold></td></tr><tr><td align="center"><1</td><td align="center">6</td><td align="center">50.00</td><td align="center">6</td><td align="center">50.00</td><td align="center">5</td><td align="center">41.67</td><td align="center">6</td><td align="center">50.00</td><td align="center">10</td><td align="center"><bold>83.33</bold></td></tr><tr><td align="center"><0.5</td><td align="center">4</td><td align="center"><bold>33.00</bold></td><td align="center">3</td><td align="center">25.00</td><td align="center">3</td><td align="center">25.00</td><td align="center">3</td><td align="center">25.00</td><td align="center">4</td><td align="center"><bold>33.33</bold></td></tr></tbody></table></table-wrap><p>Given the capacity of the predictive programs to under or over predict the true pK<sub>a </sub>value (see Discussion), the possibility of using a consensus approach to integrate the various programs was investigated. Using the Small dataset, combinations of the prediction values were calculated and the accuracy tested as before. From this dataset, 25 combinations (Table <xref ref-type="table" rid="T8">8</xref>) were tried. One, UHBD + PROPKA, leads to improvements in all residues except histidine. The His RMSD value of 0.955, from this combination, is far better than all of the programs except UHBD, which has a value of 0.494. The overall accuracy of this combination was not a surprise due to the individual performance of each program. A further attempt to integrate the programs was made by using Partial Least Squared (PLS) regression. The PLS model generated had a correlation coefficient (<italic>r</italic><sup><italic>2</italic></sup>) of 0.9 and a cross-validated correlation coefficient (<italic>q</italic><sup><italic>2</italic></sup>) of 0.89. The resulting equations were applied to the Small dataset and the accuracy results are shown in Table <xref ref-type="table" rid="T9">9</xref>. The accuracy improved greatly in the <0.5 range to almost 58%, significantly greater than either the other programs run individually or the combination method (UHBD + PROPKA). Again, a disparity between the predictive capabilities of MEAD and the other programs may be seen. AMBER and PARSE have coefficients of -0.03129 and -0.0001836 respectively, which are comparatively small compared to the other coefficients (0.239, 0.4282 and 0.4108 for MCCE, UHBD and PROPKA respectively). The results above would indicate a more significant relationship between the PROPKA and UHBD predictions and the experimental pK<sub>a </sub>values. This data fits with the trends seen in the other analysis. However, multiple linear regression is not an ideal way to increase predictive accuracy as combining programs will cause the propagation of experimental errors within a given dataset.</p><table-wrap position="float" id="T8"><label>Table 8</label><caption><p>Overview of the combination methods (242 Residues). The residue RMSD values are given for all of the 25 combinations consisting of AMBER (A), PARSE (P), MCCE (M), UHBD (U) and PROPKA (P). Figures marked in bold indicate an improvement while the asterisk indicates the best score.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>A + P</bold></td><td align="center"><bold>A +M</bold></td><td align="center"><bold>A + U</bold></td><td align="center"><bold>A + PR</bold></td><td align="center"><bold>P + M</bold></td></tr></thead><tbody><tr><td align="center"><bold>ASP</bold></td><td align="center">1.556</td><td align="center">1.245</td><td align="center">1.174</td><td align="center"><bold>0.837</bold></td><td align="center">1.161</td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.301</td><td align="center">1.012</td><td align="center">0.816</td><td align="center">0.676</td><td align="center">1.026</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">1.331</td><td align="center">0.631</td><td align="center">0.827</td><td align="center">1.000</td><td align="center">0.727</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">1.281</td><td align="center">0.763</td><td align="center">0.742</td><td align="center">0.646</td><td align="center">0.793</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">1.899</td><td align="center">1.512</td><td align="center">1.238</td><td align="center">1.028</td><td align="center">1.502</td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td></td><td align="center"><bold>P + U</bold></td><td align="center"><bold>P + PR</bold></td><td align="center"><bold>M + U</bold></td><td align="center"><bold>M + PR</bold></td><td align="center"><bold>U + PR</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">1.074</td><td align="center"><bold>0.786</bold></td><td align="center">0.910</td><td align="center"><bold>0.818</bold></td><td align="center"><bold>0.596</bold></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">0.834</td><td align="center">0.690</td><td align="center">0.660</td><td align="center">0.766</td><td align="center"><bold>0.393</bold></td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">0.812</td><td align="center">0.987</td><td align="center">0.566</td><td align="center">1.289</td><td align="center">0.955</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">0.742</td><td align="center">0.647</td><td align="center">0.713</td><td align="center">0.755</td><td align="center"><bold>0.366</bold></td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">1.212</td><td align="center">1.016</td><td align="center">0.974</td><td align="center">0.956</td><td align="center"><bold>0.466</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td></td><td align="center"><bold>A + P + M</bold></td><td align="center"><bold>A + P + U</bold></td><td align="center"><bold>A + P + PR</bold></td><td align="center"><bold>A + M + U</bold></td><td align="center"><bold>A + M + PR</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">1.262</td><td align="center">1.242</td><td align="center">1.000</td><td align="center">1.038</td><td align="center"><bold>0.837</bold></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">1.064</td><td align="center">0.973</td><td align="center">0.844</td><td align="center">0.761</td><td align="center">0.714</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">0.792</td><td align="center">0.954</td><td align="center">0.942</td><td align="center">0.520</td><td align="center">0.839</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">0.868</td><td align="center">0.913</td><td align="center">0.846</td><td align="center">0.615</td><td align="center">0.590</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">1.607</td><td align="center">1.446</td><td align="center">1.300</td><td align="center">1.195</td><td align="center">1.097</td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td></td><td align="center"><bold>A + U + PR</bold></td><td align="center"><bold>P + M + U</bold></td><td align="center"><bold>P + M + PR</bold></td><td align="center"><bold>M + U + PR</bold></td><td align="center"><bold>A + P + M + U</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center"><bold>0.773</bold></td><td align="center">0.972</td><td align="center"><bold>0.795</bold></td><td align="center"><bold>0.681</bold></td><td align="center">1.101</td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">0.539</td><td align="center">0.770</td><td align="center">0.720</td><td align="center">0.529</td><td align="center">0.870</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">0.771</td><td align="center">0.523</td><td align="center">0.840</td><td align="center">0.897</td><td align="center">0.646</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">0.518</td><td align="center">0.634</td><td align="center">0.610</td><td align="center">0.597</td><td align="center">0.718</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">0.869</td><td align="center">1.188</td><td align="center">1.098</td><td align="center">0.782</td><td align="center">1.345</td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td></td><td align="center"><bold>A + P + M + PR</bold></td><td align="center"><bold>A + P + U + PR</bold></td><td align="center"><bold>A + M + U + PR</bold></td><td align="center"><bold>P + M + U + PR</bold></td><td align="center"><bold>A + P + M + U + PR</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="center"><bold>ASP</bold></td><td align="center">0.929</td><td align="center">0.905</td><td align="center"><bold>0.779</bold></td><td align="center"><bold>0.738</bold></td><td align="center"><bold>0.866</bold></td></tr><tr><td align="center"><bold>GLU</bold></td><td align="center">0.798</td><td align="center">0.706</td><td align="center">0.588</td><td align="center">0.592</td><td align="center">0.690</td></tr><tr><td align="center"><bold>HIS</bold></td><td align="center">0.754</td><td align="center">0.771</td><td align="center">0.690</td><td align="center">0.668</td><td align="center">0.650</td></tr><tr><td align="center"><bold>LYS</bold></td><td align="center">0.683</td><td align="center">0.689</td><td align="center">0.525</td><td align="center">0.540</td><td align="center">0.605</td></tr><tr><td align="center"><bold>TYR</bold></td><td align="center">1.256</td><td align="center">1.112</td><td align="center">0.959</td><td align="center">0.959</td><td align="center">1.115</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T9"><label>Table 9</label><caption><p>Accuracy of the multiple regression. The accuracy is given as the number of predictions within a range of the pK<sub>exp</sub>. For comparison the UHBD + PROPKA combination is added. Figures marked in bold indicate improvements.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>TOTAL (242)</bold></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td align="center"><bold>REGRESSION</bold></td><td></td><td align="center"><bold>UHBD + PROPKA</bold></td><td></td></tr><tr><td></td><td></td><td align="center">%</td><td></td><td align="center">%</td></tr></thead><tbody><tr><td align="center"><2</td><td align="center">234</td><td align="center">96.69</td><td align="center">233</td><td align="center">96</td></tr><tr><td align="center"><1.5</td><td align="center">228</td><td align="center"><bold>94.21</bold></td><td align="center">226</td><td align="center">93</td></tr><tr><td align="center"><1</td><td align="center">205</td><td align="center"><bold>84.71</bold></td><td align="center">197</td><td align="center">81</td></tr><tr><td align="center"><0.5</td><td align="center">140</td><td align="center"><bold>57.85</bold></td><td align="center">124</td><td align="center">51</td></tr></tbody></table></table-wrap><p>The disparity between the predictive capabilities of the program relates to the algorithm that used for the calculation. MEAD, UHBD and MCCE are all based upon an electrostatic continuum model that solves the linearised Poisson-Boltzmann equation numerically [<xref ref-type="bibr" rid="B32">32</xref>,<xref ref-type="bibr" rid="B33">33</xref>]. The electrostatic potential φ(<italic>r</italic>)can be calculated by the Poisson-Boltzmann equation:</p><p>∇ ε (<italic>r</italic>)∇ φ (<italic>r</italic>) - κ<sup>2 </sup>(<italic>r</italic>)ε (<italic>r</italic>)φ (<italic>r</italic>) = -4πρ (<italic>r</italic>)       (3)</p><p>Whereε is the dielectric constant, <italic>r </italic>is the position vector, <italic>Φ </italic>is the electrostatic potential, ρ is the charge distribution and κ is a parameter that represents the effect of mobile ions in solution. All three programs work on the assumption that the major determinant of the pK<sub>a </sub>shift from the model values are the electrostatic effects of burying titratable groups in low dielectric medium. A model of the macromolecule-solvent system is used with dielectric constants of 80 for the solvent and 4 for the protein. The details of the atomic structure are incorporated into the placement of charges and dielectric boundaries. The calculation accounts for the desolvation energy, the titratable group's interaction with partial charges and the group's interaction with other titratable groups in the protein. MEAD consistently performs more poorly than the other two Poisson-Boltzmann-based programs. This may be because, in addition to the basic calculation, UHBD and MCCE also incorporate a Monte Carlo function to sample the multiple conformations of each titratable site. The Monte Carlo method achieves convergence by random sampling of side chain conformers. This allows the MCCE and UHBD programs to make a more realistic calculation of the charge-charge interactions than MEAD. The RMSD values of the two MEAD data sets – PARSE and AMBER – are comparable, both producing similar RMSD values and numbers of outliers. However, Table <xref ref-type="table" rid="T2">2</xref> shows that the PARSE force field generates outliers that deviate much further from the experimentally-determined values than those of AMBER. Although the parameters of the two force fields are similar; the atomic radii of the hydrogens for PARSE are slightly larger which may have created inaccurate charge-charge interactions that have increased the calculated pK<sub>a </sub>value (this would also account for the program's propensity to generate outliers). This is especially noticeable for Lys where the respective RMSD values of AMBER and PARSE are 1.2 and 25.8.</p><p>Although the PROPKA and MCCE programs are of comparable accuracy, the data suggests that the former tends to under-predict pK<sub>a </sub>values whilst the latter over-predicts them (Figure <xref ref-type="fig" rid="F1">1</xref>). This observation may reflect the different approaches towards the calculation of the pK<sub>a </sub>value in the two programs. PROPKA [<xref ref-type="bibr" rid="B17">17</xref>] takes a different approach to the other three programs, calculating the pK<sub>a </sub>shift by using empirical rules that incorporate effects from hydrogen bonds, desolvation and Coulombic interactions. The extent of the pK<sub>a </sub>shift caused by hydrogen bonding is proportional to the number of hydrogen bonds formed by the titratable group [<xref ref-type="bibr" rid="B19">19</xref>]. The desolvation effect is calculated from the solvent accessible surface and the 'depth of burial' (the distance of the group from the protein surface). Lastly, the strength of the Coulombic charge-charge interactions is dependent on the distance between the charges and on the state of the surrounding ionizable residues. This process, however, is only applied to buried pairs of ionizable residues. Therefore PROPKA's tendency to under-predict pK<sub>a </sub>values may be caused by the program's emphasis on the dominance of hydrogen bonding in determining the extent of the shift. Hydrogen bonds have the effect of lowering pK<sub>a </sub>values [<xref ref-type="bibr" rid="B34">34</xref>] and the predicted values may reflect that. Conversely, MCCE's tendency to over-predict may be the result of charge-charge interaction forcing an increase in the pK<sub>a </sub>value. The majority of the over-predicted values are surface residues and, unlike PROPKA, the MCCE program does not take into account the lessened effects of charge-charge interactions when the respective residues are not buried within the protein interior. Consequently, PROPKA and MCCE tend to be more accurate for surface and buried residues respectively.</p><p>Side chains located at active sites are of particular interest as they often have unusually high or low pK<sub>a </sub>values. In some instances, the electrostatic charge of the active site can be radically different from that the rest of the protein as a means to 'steer' a ligand towards the binding cleft [<xref ref-type="bibr" rid="B35">35</xref>]. For a program to work as an effective pK<sub>a </sub>prediction tool it must be able to predict unusual pK<sub>a </sub>values accurately. Generally speaking, the accuracy of prediction decreased the further the measured pK<sub>a </sub>value was from the side chain's intrinsic pK<sub>a</sub>. Again, PROPKA proved to be the most consistent of the programs. This is not surprising as the design of the model and assignment of parameters were based upon a large dataset of carboxyl pK<sub>a </sub>values. Overall, the active site data was encouraging at the <1 unit level. However, once reduced to <0.5, the accuracy of all of the programs decreased. This highlights a key area for the development of new models and programs.</p><p>An interesting correlation is seen with respect to the regression coefficients and the general performance of the programs. Coefficients are generally an indicator of the relative importance of the contributing terms in a regression equation. The comparative performance of PROPKA, the combination methods and the regression model is seen in Figure <xref ref-type="fig" rid="F2">2</xref>. PROPKA is equally effective as these additional methods, although the regression data performs better than the best combination. A Molecular Dynamics simulation of one of the proteins from the dataset (Barnase wild type ribonuclease (<italic>pdb </italic>code: 1A2P)) showed a standard deviation of ± 1.4 for the pK<sub>a </sub>value over a one-nanosecond period. This indicates that a dynamic structure has a large capacity for extreme pK<sub>a </sub>shifts. This suggests that any accurate prediction pK<sub>a </sub>method would need to incorporate conformational variability into the algorithm.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Comparative performance of the prediction methods</bold>. The accuracy ranges (0.5 – 2) apply to the deviation from the measured pK<sub>a </sub>value. The percentage score represents the number of residues predicted in each range.</p></caption><graphic xlink:href="1471-2091-7-18-2"/></fig></sec><sec><title>Conclusion</title><p>PROPKA is the most accurate method for all residues except Glu and His, where it is narrowly surpassed by UHBD and MCCE, respectively. Furthermore, the program also produces by far the best values for surface residues, most likely by taking sufficient account of hydrogen bonding. However, MCCE predicts buried residues far better than PROPKA, possibly by a more accurate evaluation of the charge-charge interaction with the conformers optimised by the Monte Carlo procedure. It must be noted that in all cases, the prediction of the buried residues is less accurate than for surface residues, indicating it is easier to calculate the interaction of a solvated or partially solvated residue than one densely packed within the protein interior. Overall, the best standalone program is PROPKA, which also produced the fewest outliers and is computationally much faster than the other programs. What the program lacks is a capacity to fully explore the conformational space available to the protein, which may ultimately limit its capacity to predict pK<sub>a </sub>value. The reliability of the predictive programs tends to vary with both the residue type and its spatial location. For glutamic acid residues, UHBD produced the best results while for Histidine and for all buried residues, MCCE performed well. The comparatively poor prediction of the 'unusual' pK<sub>a </sub>values by all of the programs was disappointing. Their ability to only predict a third of the residues to a high degree of accuracy highlights an area requiring further development. The variation in pK<sub>a </sub>values observed in our molecular dynamics simulation strongly suggests a complete sampling of conformations accessible to protein structures may be useful in creating accurate predictive software.</p></sec><sec sec-type="methods"><title>Methods</title><p>100 proteins for which pK<sub>a </sub>values had been determined experimentally were taken from PPD, a database of protein ionization constants [<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B37">37</xref>]. The full list of the <italic>pdb </italic>files comprising the dataset is included as an additional file [See PDB codes]. A wide range of both protein size and function was represented in the dataset. The protein structures were taken from the RCSB protein data bank [<xref ref-type="bibr" rid="B38">38</xref>]. In order to run the MEAD program, <italic>pdb </italic>files were protonated by using the <italic>leap </italic>program and the AMBER 94 force field (subsequent versions of the force field proved to be incompatible) and changed into <italic>pqr </italic>format using the online PDB2PQR converter [<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>]. Separate sets of files were created based on the AMBER99 and PARSE force fields. MEAD and UHBD were run on an IBM Blade Center Cluster, which consists of 5 Blade Centers containing 67 Dual Xeon (3.06Ghz, 1Gb) Blades. The MCCE calculations were carried out on an SG Octane. The majority of the <italic>pdb </italic>files did not need any modification. However, 1D3K, 1GU8, 1HRH and 1DRH were protonated with the <italic>leap </italic>program and the AMBER 03 force field in order to remove inconsistencies in the <italic>pdb </italic>files. Additionally, 1DUK, 1NFN and 2CI2 underwent minimization with <italic>sander </italic>using a steepest descent method that continued for 20,000 1 fs time steps or until the root mean square deviation between successive time-steps had fallen below 0.01Å in order to eliminate steric clashes. The PROPKA program was run online from its server [<xref ref-type="bibr" rid="B41">41</xref>]; no modification was required to run the files. Values for all Asp, Glu, His, Tyr, Lys residues were predicted. Arg was excluded from the calculation due to lack of experimental data. Arginines's high pK<sub>a </sub>precludes establishing a titratable curve as the protein denatures at high pH. Cys was also excluded from the calculations due to a lack of experimental data.</p><p>The resultant data was also analysed using the Partial Least Squares (PLS) method. PLS is an extension of Multiple Linear Regression (MLR) that where a set of coefficients are developed from dependent variables, in this case the pK<sub>a </sub>prediction values, by comparison with the independent variables, the experimental pK<sub>a </sub>values. The PLS analysis was performed using the program GOLPE (Generating Optimal Linear PLS Estimations)[<xref ref-type="bibr" rid="B42">42</xref>].</p></sec><sec><title>Authors' contributions</title><p>MND formatted the data carried out the calculations for all of the pK<sub>a </sub>programs mentioned. CPT assembled the data set and carried out statistical analysis on the output of the pK<sub>a </sub>programs. DSM supervised the pK<sub>a </sub>calculations using the MEAD and UHBD programs at Birkbeck College. DRF instigated and supervised the entire project. MND, CPT and DRF drafted the manuscript. All authors have read and accepted the manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>PDB codes, a full list of the pdb codes for the three-dimensional structures comprising the dataset.</p></caption><media xlink:href="1471-2091-7-18-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
A placebo-controlled pilot study of intensification of antiretroviral therapy with mycophenolate mofetil
|
<sec><title>Purpose</title><p>We studied the safety, tolerability, virologic, and immunologic effects of mycophenolate mofetil (MMF) added to a stable antiretroviral therapy (ART) in the setting of low-level viremia.</p></sec><sec sec-type="methods"><title>Methods</title><p>MMF 500 mg BID or placebo was given to patients thought to be adherent on stable ART with plasma viremia between 200 and 4000 copies/mL. At week 4 unblinding was performed and patients on placebo were offered open-label MMF.</p></sec><sec><title>Results</title><p>Six patients were enrolled. At entry mean plasma HIV-1 RNA (VL) was 2.98 log<sub>10 </sub>copies/mL; mean CD4 count was 523. All subjects randomized to placebo elected to cross over to open label MMF. No significant adverse events were observed during MMF therapy. Three patients on MMF achieved VL < 50 copies/mL by week 4; a fourth had VL decline of > 0.5 log. Two patients on placebo had declines of VL. One of these had further decline on open label MMF. Cell surface markers of apoptosis, activation, and proliferation on CD4+ and CD8+ cells declined modestly or remained low. CD4 counts were stable at week 24. All but one subject had rebound of viremia by week 24, universally associated with missed doses of medication by pill count.</p></sec><sec><title>Conclusion</title><p>MMF appears to be safe, and its administration lead to decreased T cell activation. During periods of adherence to therapy, the use of MMF was correlated with declines in viremia, but this small pilot study could not prove this association. Further study of MMF in patients with viremia should be considered for whom additional or alternative antiretrovirals are impractical.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Kaur</surname><given-names>Rupinderjeet</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Bedimo</surname><given-names>Roger</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Kvanli</surname><given-names>Mary Beth</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Turner</surname><given-names>Diana</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Shaw</surname><given-names>Leslie</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" corresp="yes" contrib-type="author"><name><surname>Margolis</surname><given-names>David</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
AIDS Research and Therapy
|
<sec><title>Introduction</title><p>The adjunctive use of inhibitors of nucleoside metabolism may exploit the reliance of HIV-1 on nucleoside pools for reverse transcription. Further, directly blunting host cell activation might have clinical benefits in HIV infection.</p><p>Mycophenolic acid (MPA) is a selective and reversible inhibitor of de novo synthesis of deoxyguanosine triphosphate (dGTP) [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. MPA's effects are selective for lymphocytes, and it suppresses HIV replication through guanine depletion [<xref ref-type="bibr" rid="B3">3</xref>], increasing the efficacy of several reverse transcriptase inhibitors in vitro [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B6">6</xref>] and in vivo [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>].</p><p>We hypothesized that MMF could improve virologic suppression in the setting of low-level viremia, preserving other antiretroviral agents for future use. We conducted a placebo-controlled pilot study to evaluate the safety, tolerability, and immunologic and virologic effects of the addition of MMF to an incompletely successful ART regimen. Volunteers with persistent viremia < 4000 but > 200 copies/ml were recruited. We found that MMF appears safe, and its use was associated with a decreased T cell activation as well as a short-term decline in plasma HIV-1 RNA. However, due to the confounding effect of non-adherence we could not irrefutably attribute the virologic effect seen to the activity of MMF.</p></sec><sec sec-type="methods"><title>Methods</title><p>HIV-infected patients gave IRB-approved consent and were medically stable at study entry, without history of opportunistic infection within 12 months. All were on stable antiretroviral therapy including tenofovir, abacavir, and/or didanosine (agents shown to be potentiated by MMY in vitro; ref. 6) for ≥ 12 weeks with plasma HIV-1 RNA between 200 and 4000 copies/mL. Patients were carefully interviewed and felt to be adherent to therapy at entry by their long-term medical providers. Due to the theoretical possibility of clinical antagonism between thymidine analogs and MMF [<xref ref-type="bibr" rid="B4">4</xref>], patients receiving zidovudine or stavudine allowed to enroll if at least three of the following mutations had been detected in HIV-1 reverse transcriptase at a prior genotype: M41L, D67N, K70R, V75T, L210W, T215F/Y, K219E/Q, K65R, L74V, Q151M.</p><p>Patients with AIDS Clinical Trials Group (ACTG) grade IV liver function test abnormalities, grade III or higher renal insufficiency, grade III or higher leucopenia, or dementia thought to impair adherence were excluded. Study subjects were prohibited from concurrent use of systemic corticosteroids, hydroxyurea, or other immunosuppressive medications, cholestyramine, oral contraceptives, and probenecid or other inhibitors of tubular secretion.</p><p>Patients were first randomized to the addition of MMF 500 mg BID (Arm A) or matched placebo (Arm B) to their antiretroviral regimen (Step 1). Provider interviews and review of medication refill records were used to assess patient adherence. After 4 weeks of study therapy, unblinding was performed and patients on placebo offered open-label MMF for the remainder of the 24-week follow-up (Step 2), if they maintained HIV-1 RNA measurements of < 4000 copies/ml. Virologic and immunologic responses, MPA levels, and clinical status were monitored. Subjects on MMF during Step 1, regardless of their response to blinded MMF, were given the option of continued open-label MMF therapy and follow-up, or study discontinuation.</p><p>At each study visit, patients underwent clinical evaluation, HIV-1 RNA level by Roche Amplicor PCR assay, CD4 lymphocyte counts, hematology, and clinical chemistry (including serum lactate levels and anion gap analysis). Blood was also collected for cell surface marker studies.</p><p>Flow cytometry was performed on a FACS-Calibur, and data was analyzed with Cellquest software (Becton Dickinson, San Jose, CA) to measure expression of CD4, CD8, Annexin V, CD69, CD38, CD25 and Ki67 on CD4+ and CD8+ T cells. Annexin V-FITC Apoptosis Detection Kit I (BD Pharmingen) was used for detection of apoptosis. One million lymphocytes were examined from each study subject before the therapy was initiated, at weeks 4, 8 and 12 and at week 24 at the end of the therapy.</p></sec><sec><title>Results</title><p>Six patients meeting the above criteria were enrolled. Baseline mean plasma HIV-1 RNA (VL) was 2.98 log<sub>10 </sub>copies/mL (range 1.9–3.9); and mean CD4 count was 523 (range 180–800). All subjects randomized to placebo elected to cross over to open label MMF. No significant adverse events were observed during MMF therapy. None of the patients experienced significant changes in blood hematocrit, metabolic profile, liver function tests or lipid profile during protocol therapy.</p><p>Four patients were randomized to receive MMF. Three of these achieved VL < 50 copies/mL by week 4, and elected to enter Step 2 of the study. One subject did not have a significant decline of plasma HIV-1 RNA on blinded MMF, and left the study after week 4. One patient on placebo had a significant decline of VL of > 0.5 log. This subject had a further VL decline of > 0.5 log copies/ml during Step II while receiving open-label MMF.</p><p>There was no significant change in mean CD4 count (422/mL at week 24) in subjects receiving MMF. All but one subject had rebound of viremia by week 24, universally associated with missed doses of medication by pill count. As observed in previous studies, average MPA AUC measured at week 4 was 19.40 (range 18.79–19.90) regardless of antiretroviral regimen [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B11">11</xref>].</p><p>The administration of low-dose MMF might decrease T cell activation, either by a direct immunomodulatory effect, or secondarily via an antiviral effect [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. However, MMF has also been reported to induce apoptosis in the setting of HIV infection [<xref ref-type="bibr" rid="B10">10</xref>], although this effect may only be seen in activated cells [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. Patient M4 received HAART and blinded MMF during Step 1 but did not display a virologic response. However, the Ki67, CD69, CD38 and CD25 levels declined while he was receiving MMF.</p><p>Subjects P1 and P2 received HAART and placebo during Step 1, and then HAART and open-label MMF during Step 2. These subjects had a decline in viral load during Step 1, presumably due to study-related improvements in adherence. While on open-label MMF, pill counts suggested non-adherence, correlated with a loss of virologic response. Levels of annexin and Ki67 decreased initially, but returned to baseline levels with the loss of virologic response. However, CD69 and CD25 levels declined somewhat and remained suppressed despite the loss of virologic response. In P1 the level of CD38 on CD8 cells also remained low despite viral rebound, whereas in P2 this marker increased after viral rebound.</p><p>Subject M1 displayed a gradual and persistent response to MMF during the course of the study. Annexin and Ki67 levels also declined during observation. However, activation markers increased at week 24. M3 and M2 received blinded MMF during the first 4 weeks of study, with declines in viremia. Annexin and Ki67 levels decreased initially, and surface levels of CD69, CD38 and CD25 remained low and stable. Both subjects lost virologic response; and non-adherence was simultaneously observed by pill count. During this time, while open-label MMF and HAART was prescribed but apparently taken irregularly, small but variable increases in Annexin and Ki67 were seen, as well as moderate increases in CD69, CD38 and CD25 levels in patient M2.</p></sec><sec><title>Discussion</title><p>In total therefore, MMF induced a persistent decrease in T cell activation in all but one patient (Fig <xref ref-type="fig" rid="F1">1A</xref> and <xref ref-type="fig" rid="F1">1B</xref>). Associated with suboptimal treatment adherence, virologic response was not durable in this patient population. The clinical scenario of partially successfully ART, similar to those screened for this study, is not uncommon. Treatment strategies for this group of patients are not well defined.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Immunologic and virologic effects of MMF intensification</bold>. Fig. 1A shows subjects initially assigned blinded placebo (light grey) who later elected to receive open-label MMF (grey). Fig. 1B shows those assigned blinded MMF (gray) who elected open-label MMF during weeks 4–24 (hatched grey). L.o.d.: limit of detection (< 50 HIV-1 RNA copies/ml).</p></caption><graphic xlink:href="1742-6405-3-16-1"/></fig><p>Continued therapy despite low-level viremia in the setting of drug resistance may be beneficial. Mutations conferring resistance to antiretroviral drugs commonly lower viral replicative capacity, and may blunt viremia In some treated patients [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. Clinical and immunologic benefits can be maintained in patients with partial virologic suppression [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. However, when partially effective treatment is continued, slow accumulation of resistance mutations may lead to increased viremia, and may jeopardize future treatment options. It is also reasonable to question whether continued exposure to the toxicity of multiple drugs is warranted in the face of limited virologic and immunologic response. A second approach in the face of partially effective antiretroviral therapy is the intensification of therapy. The risks of this approach include the development of resistance to the newly added antiretroviral(s), and cumulative toxicities.</p><p>A third option might be the use of an agent like MMF as a "stopgap" measure in patients without full virologic suppression who require antiviral therapy, but in whom active antivirals are not desired due to nonadherence, or not available due to drug resistance. Our data show a blunting of CD3+ cell activation in weeks 4 through 24, despite loss of virologic response in some patients. This, together with an absence in CD4 decline under MMF therapy is consistent with finding by other investigators in different settings [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>].</p><p>It is possible that viral resistance to MMF might develop over time. One mechanism for this might be a shift in viral replication away from activagted lymphocytes and monocytes to cell types with lower levels of dependence on IMPDH type I, e.g. resting CD4+ T cells. However, substantial levels of viral replication in such cell populations. Alternatively, HIV RT could evolve higher affinity for native dGTP substrates. Three subjects responding to MMF initiated at the time of antiretroviral optimization during during our initial study [<xref ref-type="bibr" rid="B8">8</xref>] elected to extend MMF therapy under IRB oversight. Continued response to salvage therapy that including MMF, as measured by at least 0.5 log<sub>10 </sub>suppression of viral load and CD4 cell count stability, was observed for 27, 30, and 33 months, respectively prior to the clinical need for re-optimization of therapy. Only one patient developed a single new RT mutation during this time, although several changes in protease were observed.</p><p>In summary, in a short-term evaluation, MMF appears to be safe its use was associated with decreased T cell activation but the effect on VL suppression was not clearly ascertained, due to intermittent non-adherence to therapy during this study. Consistent with previous reports [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>], we found no clinically significant cytopenias during MMF therapy. MMF has the potential to improve antiretroviral treatment response as well as delay virologic rebound. However, a comprehensive evaluation of the clinical efficacy of MMF will require a larger or a longer controlled study, due in part to the many factors which blunt treatment efficacy in patients with partially suppressed viremia.</p></sec>
|
Costing the distribution of insecticide-treated nets: a review of cost and cost-effectiveness studies to provide guidance on standardization of costing methodology
|
<sec><title>Background</title><p>Insecticide-treated nets (ITNs) are an effective and cost-effective means of malaria control. Scaling-up coverage of ITNs is challenging. It requires substantial resources and there are a number of strategies to choose from. Information on the cost of different strategies is still scarce. To guide the choice of a delivery strategy (or combination of strategies), reliable and standardized cost information for the different options is required.</p></sec><sec sec-type="methods"><title>Methods</title><p>The electronic online database PubMed was used for a systematic search of the published English literature on costing and economic evaluations of ITN distribution programmes. The keywords used were: net, bednet, insecticide, treated, ITN, cost, effectiveness, economic and evaluation. Identified papers were analysed to determine and evaluate the costing methods used. Methods were judged against existing standards of cost analysis to arrive at proposed standards for undertaking and presenting cost analyses.</p></sec><sec><title>Results</title><p>Cost estimates were often not readily comparable or could not be adjusted to a different context. This resulted from the wide range of methods applied and measures of output chosen. Most common shortcomings were the omission of certain costs and failure to adjust financial costs to generate economic costs. Generalisability was hampered by authors not reporting quantities and prices of resources separately and not examining the sensitivity of their results to variations in underlying assumptions.</p></sec><sec><title>Conclusion</title><p>The observed shortcomings have arisen despite the abundance of literature and guidelines on costing of health care interventions. This paper provides ITN specific recommendations in the hope that these will help to standardize future cost estimates.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Kolaczinski</surname><given-names>Jan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Hanson</surname><given-names>Kara</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
Malaria Journal
|
<sec><title>Introduction</title><p>Insecticide-treated nets (ITNs) are an effective tool for the prevention of morbidity and mortality caused by malaria [<xref ref-type="bibr" rid="B1">1</xref>] and other vector-borne diseases [<xref ref-type="bibr" rid="B2">2</xref>]. Cost-effectiveness of ITNs in the prevention of malaria has also been amply demonstrated in a variety of settings [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. The present challenge is to scale-up and sustain coverage with ITNs [<xref ref-type="bibr" rid="B6">6</xref>]. Many approaches to ITN delivery have evolved. These vary widely in scale, target populations and in the strategy used to provide nets and insecticide to the end user. Recent advances in the development of nets with long-lasting insecticide impregnations (LLINs) simplify strategies by reducing the need for re-treatment [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>].</p><p>To choose a delivery strategy (or combination of strategies), health planners and policy makers require reliable and standardised information on the cost per unit coverage [<xref ref-type="bibr" rid="B9">9</xref>]. Standardized costs are also needed for use in combination with effectiveness estimates to judge the efficiency of ITN programmes and the scope for their improvement. To generate data whereby alternative approaches can be compared a common set of techniques will need to be proposed and agreed upon [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. The present study aims at contributing towards this process by reviewing the existing literature on costing and cost-effectiveness of ITN distribution programmes, identifying short-comings and suggesting ways to improve on quality and comparability of costing methods.</p></sec><sec sec-type="materials|methods"><title>Materials and methods</title><p>A systematic search of the published English literature on costing and economic evaluations of ITN distribution programmes was conducted by means of the electronic online database PubMed (US National Library of Medicine, Bethesda, USA). The keywords used were: net, bednet, insecticide, treated, ITN, cost, effectiveness, economic and evaluation. These searches were supplemented by iterative reviews of reference lists of relevant published papers. Two consultancy reports were also included.</p><p>All papers that provided cost estimates for the delivery of mosquito nets and/or insecticide treatment were considered relevant, even if they did not fully satisfy the criteria of a costing study [<xref ref-type="bibr" rid="B11">11</xref>]. To fulfil the aim of the present review of providing comprehensive guidance on how to improve and standardise costing of ITN distribution programmes, it was considered important to use cost and costing in their broader sense.</p><p>Studies were examined by means of a checklist adapted from other publications [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B14">14</xref>]. A summary of the following information was prepared: delivery and financing mechanism used, type of study (costing or cost-effectiveness), country where the study was implemented, perspective from which the authors measured costs, costs included, whether or not guidelines were used to assist in the identification, measurement and/or valuation of inputs, year of prices and results obtained. During this process, all studies that had been referred to by the authors as addressing cost-effectiveness were allocated to this category, rather than attempting to reclassify them to more precise categories such as those proposed elsewhere [<xref ref-type="bibr" rid="B11">11</xref>].</p></sec><sec><title>Results</title><p>Twenty-six documents including two consultancy reports, published between 1989 and 2005, were identified as containing relevant cost estimates. Overall, 15 of the studies used an intermediate output (henceforth referred to as cost studies) whereas 11 related cost to a health outcome (cost-effectiveness studies) (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Summary of studies that provide cost estimates of the delivery of insecticide and/or nets</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Category</bold>*</td><td align="left"><bold>Distribution</bold></td><td align="center"><bold>Study</bold>**</td><td align="left"><bold>Country</bold></td><td align="left"><bold>Perspective</bold></td><td align="center" colspan="2"><bold>Cost included</bold></td><td align="left"><bold>Guideline used</bold>***</td><td align="left"><bold>Year of price</bold></td><td align="left"><bold>Currency</bold></td><td align="center" colspan="5"><bold>Cost</bold>****</td><td align="left"><bold>Source</bold></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td colspan="2"><hr></hr></td><td></td><td></td><td></td><td colspan="5"><hr></hr></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td align="left"><bold>Financial</bold></td><td align="left"><bold>Economic</bold></td><td></td><td></td><td></td><td align="left">Per net delivered</td><td align="left">Per treatment delivered</td><td align="left">Per ITN delivered</td><td align="left">Per treated net year</td><td align="left">Per person protected</td><td></td></tr></thead><tbody><tr><td align="left">Public/Public</td><td align="left">Campaign (Door-to-door)</td><td align="center">C</td><td align="left">Solomon Islands (Florida Islands)</td><td align="left">Provider</td><td align="left">Nets, insecticide, transport, fuel, repairs, wages, equipment, materials, facilities</td><td align="left">Time taken</td><td align="left">Yes</td><td align="left">1988/89</td><td align="left">SI$</td><td></td><td></td><td></td><td></td><td align="left">3.85 (= US$ 1.97)</td><td align="left">34<sup>1</sup></td></tr><tr><td></td><td align="left">Campaign (Door-to-door)</td><td align="center">C</td><td align="left">China (Napo County)</td><td align="left">Provider</td><td align="left">Insecticide</td><td></td><td align="left">No</td><td align="left">1990 – 92</td><td align="left">Yuan</td><td></td><td></td><td></td><td></td><td align="left">0.48<sup>F</sup></td><td align="left">35</td></tr><tr><td></td><td align="left">Campaign (Door-to-door)</td><td align="center">CE</td><td align="left">Northern Ghana (Kassena – Nankana district)</td><td align="left">Provider</td><td align="left">Nets, insecticide, wages, transport, supplies & services</td><td align="left">Adjusted financial costs plus time of volunteers</td><td align="left">Yes</td><td align="left">1993/94</td><td align="left">US$ (and Cedis)</td><td></td><td></td><td align="left">2.40</td><td></td><td></td><td align="left">20</td></tr><tr><td></td><td align="left">General health facilities</td><td align="center">CE</td><td align="left">The Gambia</td><td align="left">Provider</td><td align="left">Nets, insecticide, etc. (taken from Picard <italic>et al</italic>. 1993)</td><td align="left">Adjusted financial costs</td><td align="left">No</td><td align="left">1990</td><td align="left">US$</td><td></td><td></td><td align="left">6.24</td><td></td><td></td><td align="left">36</td></tr><tr><td></td><td align="left">Campaign (Door-to-door)</td><td align="center">C</td><td align="left">Tanzania</td><td align="left">Provider</td><td align="left">Nets, insecticide, wages, transport</td><td align="left">Cost of nets was annualised (but not discounted)</td><td align="left">No</td><td align="left">1996</td><td align="left">US$</td><td align="left">1.00<sup>F</sup></td><td align="left">0.46<sup>F</sup></td><td align="left">1.46<sup>F</sup></td><td></td><td></td><td align="left">37</td></tr><tr><td></td><td align="left">Campaign (Distrib. point not specified; assumed central)</td><td align="center">CE</td><td align="left">Thailand (Thai-Myanmar Border)</td><td align="left">Patients</td><td align="left">Direct medical costs, transportation, food</td><td align="left">Time absent from work</td><td align="left">Yes</td><td align="left">1994/95</td><td align="left">US$</td><td></td><td></td><td></td><td></td><td></td><td align="left">38</td></tr><tr><td></td><td align="left">Campaign (Door-to-door)</td><td align="center">C</td><td align="left">Pakistan (Afghan refugee camps)</td><td align="left">Provider</td><td align="left">Net, insecticide, operational costs</td><td align="left">Cost of net annualised (but not discounted)</td><td align="left">No</td><td align="left">1991 – 94</td><td align="left">US$</td><td></td><td></td><td></td><td></td><td align="left">1.51</td><td align="left">39</td></tr><tr><td></td><td align="left">Campaign (Fixed-site)</td><td align="center">CE</td><td align="left">South Africa (KwaZulu-Natal)</td><td align="left">Provider</td><td align="left">Nets, insecticide, wages, equipment, transport, storage space</td><td align="left">Adjusted financial cost plus donated insecticide</td><td align="left">Yes</td><td align="left">1999</td><td align="left">US$ (and Rand)</td><td></td><td></td><td></td><td></td><td align="left">3.82</td><td align="left">40</td></tr><tr><td></td><td align="left">Campaign (Distrib. point not specified; assumed to be central)</td><td align="center">CE</td><td align="left">Thailand (Thai-Myanmar border)</td><td align="left">Provider</td><td align="left">Wages, material, capital cost</td><td></td><td align="left">Yes</td><td align="left">1994</td><td align="left">US$</td><td></td><td></td><td></td><td></td><td align="left">1.30<sup>F</sup></td><td align="left">41</td></tr><tr><td></td><td align="left">Antenatal Clinics</td><td align="center">C</td><td align="left">Kenya</td><td align="left">Provider</td><td align="left">Net and insecticide, transport</td><td></td><td align="left">No</td><td align="left">2001</td><td align="left">US$</td><td></td><td></td><td align="left">3.81<sup>F </sup>5.26<sup>F, L</sup></td><td></td><td></td><td align="left">19</td></tr><tr><td colspan="16"><hr></hr></td></tr><tr><td align="left">Public/Public</td><td align="left">Campaign (Door-to-door)</td><td align="center">C</td><td align="left">Columbia</td><td align="left">Provider</td><td align="left">Wages, per diems, insecticide, transport</td><td></td><td align="left">Yes</td><td align="left">2001</td><td align="left">US$</td><td></td><td align="left">5.10<sup>F </sup>(near) 12.40<sup>F </sup>(far)</td><td></td><td></td><td></td><td align="left">42</td></tr><tr><td colspan="16"><hr></hr></td></tr><tr><td></td><td align="left">Campaign (Fixed-site)</td><td align="center">C</td><td align="left">Tropical Africa</td><td align="left">Provider</td><td align="left">Nets, wages, allowances, transport</td><td></td><td align="left">Yes</td><td align="left">2003</td><td align="left">US$</td><td align="left">2.40<sup>F</sup></td><td></td><td></td><td></td><td></td><td align="left">22</td></tr><tr><td></td><td align="left">Campaign (Fixed-site)</td><td align="center">CE</td><td align="left">Kenya</td><td align="left">Society</td><td align="left">Insecticide, nets, wages, supplies, transport, buildings, equipment, furniture, water, time of users</td><td align="left">Adjusted financial costs plus value of community labour</td><td align="left">Yes</td><td align="left">1996</td><td align="left">US$</td><td></td><td></td><td align="left">1.90<sup>NC </sup>2.20<sup>N</sup></td><td></td><td align="left">1.4<sup>NC </sup>1.6<sup>N</sup></td><td align="left">4</td></tr><tr><td></td><td align="left">Measles vaccination sites</td><td align="center">C</td><td align="left">Ghana</td><td align="left">Provider</td><td align="left">ITNs, training and supervision, transportation, community education</td><td></td><td align="left">No</td><td align="left">2002</td><td align="left">US$</td><td></td><td></td><td align="left">3.74<sup>F</sup></td><td></td><td></td><td align="left">43</td></tr><tr><td colspan="16"><hr></hr></td></tr><tr><td align="left">Public/Mixed</td><td align="left">Campaign – fixed site (Field trial using PHC workers & villagers)</td><td align="center">CE</td><td align="left">The Gambia</td><td align="left">Society</td><td align="left">Insecticide, wages, transport, equipment, treatment, funeral expenses</td><td align="left">Adjusted financial costs plus time of volunteers, carers, hours lost due to mourning</td><td align="left">Yes</td><td align="left">1990</td><td align="left">US$</td><td></td><td align="left">5.65</td><td></td><td></td><td></td><td align="left">44</td></tr><tr><td></td><td align="left">Local clinics Mobile teams CHW</td><td align="center">C</td><td align="left">Afghanistan</td><td align="left">Provider</td><td align="left">Nets, insecticide, monitoring, supervision, training, clinic overheads, wages, transport</td><td align="left">Cost per net was adjusted (but not of other capital items)</td><td align="left">Yes</td><td align="left">1995</td><td align="left">US$</td><td align="left">2.01 1.89</td><td align="left">1.08 0.49 1.87</td><td></td><td></td><td></td><td align="left">45</td></tr><tr><td></td><td align="left">Campaign (Door-to-door)</td><td align="center">CE</td><td align="left">The Gambia (National Programme)</td><td align="left">Society (Provider + Community)</td><td align="left">Insecticide, wages, supplies, services, transport, equipment, awareness campaign, community capital costs incl. nets</td><td align="left">Adjusted financial costs plus community time and water</td><td align="left">Yes</td><td align="left">1991 – 92</td><td align="left">US$ (and Dalasis)</td><td></td><td align="left">1.00<sup>N</sup></td><td align="left">3.30<sup>N</sup></td><td></td><td></td><td align="left">26</td></tr><tr><td></td><td align="left">Campaign (Fixed site)</td><td align="center">C</td><td align="left">Vietnam (Hoa Binh Province)</td><td align="left">Provider</td><td align="left">Insecticide, equipment, labour, nets and transport (purchased by community)</td><td align="left">Adjusted financial costs</td><td align="left">Yes</td><td align="left">1996</td><td align="left">US$</td><td align="left">0.58 – 0.61<sup>P</sup></td><td align="left">0.32<sup>P</sup></td><td></td><td></td><td align="left">0.90 – 0.93</td><td align="left">31</td></tr><tr><td colspan="16"><hr></hr></td></tr><tr><td align="left">Mixed/Mixed</td><td align="left">Public and private sales agents</td><td align="center">C</td><td align="left">Tanzania</td><td align="left">Project costs and users' contributions</td><td></td><td></td><td align="left">No</td><td align="left">1998 – 99</td><td align="left">US$</td><td align="left">6.14 – 6.87<sup>F</sup></td><td align="left">1.72 – 2.11<sup>F</sup></td><td></td><td></td><td></td><td align="left">46</td></tr><tr><td></td><td align="left">Public and private sales agents</td><td align="center">CE</td><td align="left">Tanzania</td><td align="left">Project implementation costs, user contributions, travel costs</td><td align="left">Adjusted financial costs plus time of users, in-kind community contributions, donated inputs</td><td align="left">Yes</td><td align="left">2000</td><td align="left">US$ (and Shillings)</td><td></td><td></td><td></td><td align="left">8.30<sup>F</sup></td><td align="left">13.38</td><td></td><td align="left">5</td></tr><tr><td></td><td align="left">General health facilities & commercial outlets</td><td align="center">CE</td><td align="left">Malawi</td><td align="left">Provider</td><td align="left">Capital and recurrent costs</td><td align="left">Adjusted capital costs (assumed life-span as in Hanson <italic>et al</italic>. 2003)</td><td align="left">Yes</td><td align="left">1998 – 2003</td><td align="left">US$</td><td></td><td></td><td align="left">2.63</td><td align="left">4.41</td><td></td><td align="left">27</td></tr><tr><td colspan="16"><hr></hr></td></tr><tr><td align="left">Private/Mixed</td><td align="left">Community (Community groups)</td><td align="center">C, CE</td><td align="left">Kenya (Western highlands)</td><td align="left">Provider & Society</td><td align="left">Wages, nets, insecticide, cost-recovery included and excluded in analysis</td><td align="left">Adjusted financial costs plus opportunity cost of MoH Government, community, and of using NGO truck</td><td align="left">Yes</td><td align="left">2000</td><td align="left">US$ (and Shillings)</td><td></td><td></td><td align="left">4.68 30.00<sup>O</sup></td><td></td><td></td><td align="left">47,48</td></tr><tr><td align="left">Private/Private</td><td align="left">Community</td><td align="center">C</td><td align="left">Benin (Savalou region)</td><td align="left">Provider</td><td align="left">Local mosquito net production, insecticide, transportation, wages</td><td></td><td align="left">No</td><td align="left">1993</td><td align="left">US$ (and CFA)</td><td></td><td></td><td align="left">10.50<sup>F</sup>(sales price incl. profit)</td><td></td><td></td><td align="left">49</td></tr><tr><td></td><td align="left">Community (Community groups)</td><td align="center">C</td><td align="left">Kenya</td><td align="left">Provider</td><td align="left">Nets, netting material, insecticide, project running cost, monitoring, awareness campaign, training, transport</td><td align="left">Some financial costs adjusted (but nets not included as capital items)</td><td align="left">No</td><td align="left">2002</td><td align="left">US$</td><td></td><td></td><td align="left">15.80<sup>F </sup>14.40</td><td></td><td></td><td align="left">9</td></tr><tr><td></td><td align="left">Formal sector retail outlet</td><td align="center">C</td><td align="left">The Gambia</td><td align="left">Provider</td><td align="left">Net (locally made), insecticide</td><td align="left">Cost of net annualised but not discounted</td><td align="left">No</td><td align="left">1987</td><td align="left">US$</td><td align="left">1.75</td><td align="left">0.30</td><td align="left">2.05</td><td></td><td></td><td align="left">50</td></tr></tbody></table><table-wrap-foot><p>* Delivery Point/Financing of net & insecticide (Public = Free, Mixed = Partially Subsidised, Private = Full Cost and includes payment by households)</p><p>** C = Study providing cost estimate; CE = Study carried out to establish cost-effectiveness</p><p>*** Guidelines include reference by the authors to relevant documents for guidance on costing</p><p>**** Economic cost is provided unless otherwise indicated (economic costs are derived from financial costs by excluding VAT, annualising capital costs and annualising programme development costs over expected useful life of net)</p><p><sup>C </sup>Community effect included; also called 'mass effect', see reference 51 and 52 for examples</p><p><sup>F </sup>Financial cost, rather than economic cost</p><p><sup>I </sup>Incremental cost over two year study period</p><p><sup>L </sup>Cost including leakage to non-target group</p><p><sup>N </sup>Net cost, i.e. total implementation cost minus resources saved</p><p><sup>O </sup>Cost estimate including overheads to support the NGO facilitating the ITN programme</p><p><sup>P </sup>Average cost per person protected, not per net; generally nets were single or double size</p><p><sup>1 </sup>see reference 23 for price in US$</p></table-wrap-foot></table-wrap><sec><title>Technical characteristics</title><sec><title>Viewpoint taken</title><p>A study's viewpoint determines the costs to be included in the analysis. Though the perspective was sometimes not explicitly stated, it was possible to infer it for all the papers after consideration of the costs included. The majority of studies were conducted from the perspective of the service provider (Table <xref ref-type="table" rid="T1">1</xref>). The use of a societal viewpoint was limited to five of the cost-effectiveness and one of the cost studies. Less than half of the cost-effectiveness studies included the societal viewpoint among the alternatives investigated. One study examined only the patients' viewpoint.</p></sec></sec><sec><title>Types and clarity of cost measures</title><p>Financial costs of the service provider were included in all but one study, which was costed from the patients' viewpoint. Considerable variation was observed between studies in the items identified and measured. No general template could be identified by which authors had decided which inputs to include. Few studies detailed all the inputs, their unit cost, quantities consumed and reasons for inclusion/exclusion in the analysis. This observed lack of consistency and often detail could not be attributed to authors not consulting the relevant health economics literature. More than half of the studies referred to papers on costing or to one of the leading texts in this field. Most frequently quoted were Phillips <italic>et al </italic>[<xref ref-type="bibr" rid="B15">15</xref>], Creese and Parker [<xref ref-type="bibr" rid="B16">16</xref>], the first and second editions of Drummond <italic>et al</italic>. [<xref ref-type="bibr" rid="B17">17</xref>] and Gold <italic>et al</italic>. [<xref ref-type="bibr" rid="B18">18</xref>].</p><p>Insufficient attention was given to the cost of reaching specific target groups; in the case of ITNs these are generally children below the age of five or pregnant women. Two papers explicitly recognized this cost. Guyatt <italic>et al</italic>. [<xref ref-type="bibr" rid="B19">19</xref>] calculated two costs, one for delivery of nets to all users and a higher one for delivery to pregnant women (i.e. accounting for the proportion of nets that had been taken-up by other users). An intervention trial in northern Ghana considered it necessary to distribute nets to all family members to ensure that children were protected and considered this in the cost estimate [<xref ref-type="bibr" rid="B20">20</xref>].</p><p>Studies varied greatly in the way in which financial costs were adjusted to obtain economic costs (i.e. annualized and discounted) and which other economic costs, such as donated items and time of volunteers, were included. Eight studies only included financial costs. Some studies quoted the cost net of any cost savings, and one study included the community effect, i.e. the protection gained by people without a mosquito net when sleeping near ITN users. Comparison of cost estimates between studies that had used the same output was therefore not always meaningful.</p></sec><sec><title>Sources of cost data</title><p>A number of studies were undertaken alongside the early efficacy trials for ITNs, which will have facilitated the collection of cost data. However, the resulting cost estimates may be an overestimate of the costs of a routine intervention because efficacy trials require close monitoring and supervision to provide reliable clinical results; in such circumstances effectiveness may also be overestimated compared to routine conditions [<xref ref-type="bibr" rid="B21">21</xref>]. On the opposite end of the spectrum, one study collected cost data for ITN distribution alongside measles vaccination, but did not fully account for the proportional use of campaign resources. Some studies did not collect primary data on expenditure. Instead, investigators either used results of other studies to recalculate cost estimates under different assumptions or a combination of results from other studies extrapolated to a generalized scenario [see [<xref ref-type="bibr" rid="B22">22</xref>]].</p></sec><sec><title>Issues of time</title><p>An ITN programme generally lasts for more than one year and the capital items purchased to implement it mostly have a life expectancy in excess of one year. It is therefore useful to express capital costs as an annual equivalent, which requires judgements on the average life expectancy of each type of item and choice of a discount rate. Capital inputs that were annualised included mosquito nets, vehicles and equipment. Assumptions on the life expectancy and discount rate varied considerably between studies. Mosquito nets were considered to last three to seven years, the need to re-treat them was estimated at once or twice a year and the number of people protected by each net ranged from one to more than three. Discount rates applied to capital items ranged from 3% to 10%.</p></sec><sec><title>Choice of output and outcome measure</title><p>Costs were generally quoted in US$ and in some cases also in the equivalent local currency. Two studies provided results only in local currency. Overall, five intermediate output measures were encountered: i) Cost per net delivered, ii) Cost per insecticide treatment delivered, iii) Cost per ITN delivered, iv) Cost per treated net year, and v) Cost per person protected. More than half of the studies provided only one of these; the preferred one being cost per ITN delivered. Programmes that did not distribute pre-treated nets could have distinguished the costs of delivering the net and insecticide, but often failed to report these separately. Outputs were thus not easily comparable between studies and the information provided was insufficient to calculate cost for a different output measure from that provided by the authors.</p><p>Most of the cost-effectiveness studies determined their own effectiveness data and provided one or more of a selection of outcome measures, such as cost per case averted, cost per death averted or cost per disability-adjusted life year (DALY) averted. As in case of the outputs reported, the selection of outcome measures varied between studies, with most authors reporting results for only one.</p></sec><sec><title>Sensitivity analysis</title><p>Cost estimates inevitably involve some assumptions and methodological controversy. To account for this, careful analysts need to identify critical assumptions and areas of uncertainty and then re-estimate the results using different assumptions to test the sensitivity of the results and conclusions to such change. Of the studies reviewed, approximately half provided some form of sensitivity analysis. Eight (i.e. 73%) of the cost-effectiveness analysis included this component, as opposed to four (29%) of the studies that provided cost estimates.</p></sec></sec><sec><title>Discussion</title><p>The present work adds to the literature on economic aspects of the use of ITNs. A previous review covered the costs and benefits provided in sixteen published and unpublished studies from 13 countries [<xref ref-type="bibr" rid="B23">23</xref>]. It is assumed that the "price of bednets" quoted in this earlier review refers to the financial or economic cost per net delivered, rather than the price nets were sold for in the market. Variation in the quoted cost of mosquito nets ranged from US$ 3.00 in China to US$ 72.00 in rural Cameroon, and the cost for insecticide treatment from US$ 0.10 in China to US$ 2.00 in urban Cameroon. The assumed and observed life expectancy of nets varied from one to six years [<xref ref-type="bibr" rid="B23">23</xref>].</p><p>More recent results are similar, though the sources of data are different. Only four of the previously reviewed studies were included in the present review, as they were published in the English scientific literature. Here the cost per net delivered ranged from US$ 0.58 for a public sector programme in northern Vietnam to US$ 6.87 for a social-marketing programme in four areas in Tanzania. Variation in the cost of insecticide treatment was more pronounced. A study from The Gambia reports cost per treatment to be US$ 0.30, whereas delivery of treatment in remote areas of Colombia was estimated at US$ 12.40. Unfortunately, these costs could not be directly compared, because costing methodology varied too widely. The broad definition of costing inevitably captured studies with methods that would not fulfil economic standards of cost analysis. Unexpectedly, however, even studies that claimed to be costing or cost-effectiveness studies varied widely in quality.</p><p>A typology for ITN distribution and payment mechanisms recently updated by J. Webster (pers. com.) was used to categorize the studies included in this review. This indicated that the largest number of cost estimates came from programmes that use public distribution mechanisms and provide nets free of charge. Correspondingly, this category also held the largest variation in quality of the costing presented, ranging from very detailed estimates calculated as a component of cost-effectiveness studies [e.g. [<xref ref-type="bibr" rid="B4">4</xref>]] to "back of the envelope" figures used for advocacy [e.g. [<xref ref-type="bibr" rid="B22">22</xref>]]. Costing of delivery strategies categorized as mixed or private has not received much attention to date. Further cost analyses and economic-evaluations of ITN programmes are thus necessary to close this gap and to provide much needed evidence as to which delivery and payment methods, and combinations, present good value for money and should be supported in order to achieve targets outlined in the Roll Back Malaria Abuja targets [<xref ref-type="bibr" rid="B26">26</xref>] and the Millennium Development Goals (http://www.developmentgoals.org).</p><p>The present review also highlights the narrow perspective frequently taken to arrive at cost estimates. Authors predominantly chose the provider perspective, failing to consider the costs attributed to ITN users and society. This lack of consideration is by no means limited to the costing of ITN programmes, but has been reported for economic evaluations of interventions to control communicable and parasitic diseases [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. However, user costs vary depending on the degree to which strategies shift costs from the provider to the user. For example, distribution from a central location may be relatively cheap for the programme but shifts costs to the users who need to travel to acquire ITNs. Community-based distribution systems will result in a different distribution of costs between the provider and the user. With ITN programmes generally aiming to reach certain target groups, such as the rural poor (which tend to be at highest risk of deleterious effects of malaria [e.g. [<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B25">25</xref>]]), shifting of costs to the users should not be overlooked.</p><p>A narrow perspective also pays insufficient attention to the wastage of resources, such as leakage of nets and insecticide to non-target groups. If nets or insecticide are taken up by non-target groups, but still put to their intended use, this could be considered as adequate use of the programmes resources, because it increases overall coverage. However, if nets are used for other purposes, such as fishing or bridal wear, and if insecticide is diverted to agricultural use, then this should certainly be reflected in the ITN delivery cost. Ideally, authors should document such events, provide details on the type and scale of leakage, and calculate cost estimates including and excluding it [e.g. [<xref ref-type="bibr" rid="B19">19</xref>]].</p><p>This type of transparency should in fact be applied to all aspects of the cost analysis, allowing the reader to judge the reliability of the final cost estimate and, ideally, to recalculate costs under different scenarios. Transparency starts from a clear statement of the perspective taken and should be followed through by citing the source of cost data, listing the costs identified, the amounts required and their value separately, rather than as composites and by making assumptions explicit. By not adequately describing their methods of cost analysis, authors evoke the suspicion of omitting certain costs to advocate for a particular intervention [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>Only few costing studies on ITN interventions have investigated the potential cost implications of scaling-up. Those studies that have investigated this issue have arrived at different conclusions. For example, Aikins <italic>et al</italic>. [<xref ref-type="bibr" rid="B26">26</xref>] noted a considerable difference in cost-effectiveness estimate for two types of implementation in the Gambia. When compared to a controlled intervention study, the National Impregnated Bednet Programme had difficulties controlling its field activities, which almost tripled the cost per death averted (US$ 220 <italic>versus </italic>US$ 620). A recent cost-effectiveness study on a nationwide ITN programme in Malawi showed that as the programme expanded and increased ITN sales over a 5-year period, costs per ITN distributed and per treated net year decreased from US$ 5.04 to US$ 1.92 and from US$ 7.69 to US$ 3.44, respectively [<xref ref-type="bibr" rid="B27">27</xref>]. As highlighted by the authors, further economic evaluations of other large-scale programmes using different ITN delivery methods will be required to assess under which conditions increasing returns to scale can be expected in other contexts. In any case, extrapolation from small-scale trials to the potential cost of an ITN intervention at national or international scale [<xref ref-type="bibr" rid="B22">22</xref>] is unlikely to be reliable, due to inevitable changes in returns to scale associated with scaling-up [<xref ref-type="bibr" rid="B28">28</xref>]. Authors that wish to explore the potential changes in cost when scaling-up an ITN intervention should consider the use of non-linear cost functions and other methods promoted by WHO-CHOICE [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>].</p><p>Costing of health interventions presents a challenge in the selection of appropriate methods, but it is not too technical a task for non-economists. What investigators require is a critical attitude towards their own work and an aspiration to produce results that are meaningful in a specific context and transferable to others. The following recommendations are meant to help address some of the methodological challenges specific to ITN distribution and to make the process more accessible. Following the suggestion of Gold <italic>et al</italic>. [<xref ref-type="bibr" rid="B18">18</xref>] a reference case scenario is provided, which aims at generating comparable results from future studies.</p></sec><sec><title>Recommendations</title><sec><title>Viewpoint</title><p>Cost analysis should be undertaken from the broadest perspective, that of society, to incorporate all costs regardless of whom incurs them [<xref ref-type="bibr" rid="B18">18</xref>]. Other perspectives, such as that of the provider, patient or Ministry of Health, can be included alongside this perspective.</p></sec><sec><title>Costing</title><p>General costing methods are outlined elsewhere and should be consulted prior to the study. To aid generalisability and to ensure transparency, the required resources, their quantities and their value should be reported separately. Specifics related to ITN programmes are as follows.</p></sec><sec><title>Resource identification</title><p>The costs of any activities undertaken to educate people to facilitate behavioural change should be accounted for, including any advertising that is part of the intervention. It should also be made clear whether an intervention provides mosquito nets or treatment only, or whether it delivers both components. For the latter, separate costs should be provided.</p><p>In areas where LLINs are being introduced alongside conventional nets the resources associated with this change, such as additional educational campaigns/materials and the amount of insecticide wasted by re-treating LLINs during net treatment campaigns should be included. For example, the malaria control programme in Uganda treats all nets while LLINs are being phased in. In 2004, an estimated 25% of all nets were LLINs, amounting to approximately 500,000 nets (A. Kilian, pers. com). In a recent campaign in 20 districts the treatment of LLINs brought for re-treatment resulted in a cost of approximately US$ 100,000 (cost includes insecticide, equipment and implementation).</p><p>For intervention trials, sensitivity analysis should consider the resources that may be required under routine programme conditions. Research costs should not be included in the intervention costs. However, it should be investigated whether the programme would operate as well if research were not carried out alongside it. If research staff are considered to be essential for the intervention, then this cost should be included [e.g. [<xref ref-type="bibr" rid="B5">5</xref>]]. Another parameter that should be subjected to sensitivity analysis is the proportion of shared resources in cases where distribution is carried out as a component of another programme/campaign. Here it is particularly important to be explicit about the criterion used to attribute resource use (e.g. % time, % value of resources), as the basis for the breakdown is always arguable. For collaboration between organisations (e.g. local NGO & international NGO), resources including overheads that are relevant to the programme need to be identified. The final cost estimate needs to reflect the fact that, for example, a programme was made feasible because an international NGO facilitated procurement, management, fund raising, etc.</p><p>Programme management resources need to be identified and clearly specified, even if the main purpose of the study is to compare two interventions provided by the same programme (e.g. ITN/IRS). To ensure generalisability it needs to be clear what capacity is assumed to exist and whether authors have calculated average costs for delivery of each intervention or the incremental cost of adding an intervention to an existing programme (e.g. ITN delivery alongside vaccination campaigns). In areas where a mix of locally made (cotton) and imported (polyester, polyethylene) nets is being impregnated, the extra resources required to treat the former need to be considered, as these nets take up at least twice the amount of insecticide [e.g. [<xref ref-type="bibr" rid="B31">31</xref>]].</p></sec><sec><title>Resource measurement</title><p>An attempt should be made to measure and report the leakage of resources, to be able to differentiate between measured outputs (e.g. nets delivered) and outputs that reach the intended target. To inform policy, it is essential to determine the degree to which different approaches to targeting achieve their aim and at what cost.</p><p>It should be established how long nets last under local conditions. Observations to date indicate that assumed life expectancies of more than three years are unrealistic (except in the case of polyethylene nets such as Olyset<sup>®</sup>). In general, a relevant timeframe needs to be used for the costing. As most projects last for more than one year there may be issues around price level adjustments, so that all the costs can be expressed in the same year, and because of lumped expenditure (i.e. high start-up costs, but falling over time). The general recommendation with regards to tracking of costs is to select a follow-up period that does not bias the analysis in favour of one intervention over another [<xref ref-type="bibr" rid="B17">17</xref>].</p></sec><sec><title>Resource valuation</title><p>Mosquito nets need to be treated as a capital item. Annualization based on a 3 years life span is recommended. Should surveys indicate otherwise, costs for observed and assumed 3 year duration should be provided. To all capital items a discount rate of 3% should be applied, to be consistent with the rate used by the World Bank [<xref ref-type="bibr" rid="B32">32</xref>]. Sensitivity analysis should also include 0%, 5% and 10%, to presents costs in their undiscounted form, as well as at higher rates chosen by some analysts.</p><p>When comparing alternatives, for example ITNs and IRS, the same or equivalent components for their delivery need to be costed. When using cost estimates reported in other studies, these need to have been presented transparently so that they can be adapted (e.g. by applying local prices to standard quantities). Otherwise, costs for a specific study need to be identified, measured and valued and to be assessed for consistency with other authors' estimates. For programmes that recover costs through user contributions, double counting needs to be avoided. For the societal perspective these costs should be included, whereas if the provider perspective is taken they should be excluded [e.g. [<xref ref-type="bibr" rid="B5">5</xref>]].</p></sec><sec><title>Data analysis</title><p>To reflect the uncertainty in measurements, a sensitivity analysis needs to be carried out on (at a minimum): discount rate, frequency of net impregnation, procurement cost of insecticide & net, number of people protected per net, life span of net. Where it is necessary to estimate a share of resources contributed from other programmes or interventions, the assumptions used should be subjected to sensitivity analysis.</p></sec><sec><title>Reporting of results</title><p>Cost estimates should be provided in US$ and local currency, and the year to which costs were adjusted needs to be specified. Projects producing nets locally for sale need to report on the cost of making/treating nets, not just the price at which these were sold to users, which may not fully measure the opportunity cost.</p><p>The cost per ITN delivered should be reported, as cost per person sleeping under a net varies depending on cultural practices and as cost per case of malaria prevented depends on the incidence of malaria. Both are useful for policy decisions in a given context, but the cost per net delivered is best suited for comparing delivery strategies. Cost-effectiveness studies should not just provide net costs (i.e. subtracting resources saved from total implementation costs), as these cannot be compared to results from costing studies.</p></sec><sec><title>Discussion of results</title><p>To put cost-estimates for a particular method, or combination of methods, into perspective, it is recommended that they are compared to the per capita expenditure on government health services in the country/region. This does not improve on the comparability of costing studies between countries, but will help readers to assess the affordability of the intervention [e.g. [<xref ref-type="bibr" rid="B3">3</xref>]]. When reporting results, one should beware of the assumptions required to extrapolate these to calculate costs of scaling-up an intervention. To date, studies that have investigated the cost of scaling-up health interventions are limited. Available data indicate that scaling-up costs are highly specific to both the type of intervention and its particular setting [<xref ref-type="bibr" rid="B30">30</xref>]. Without further study of potential additional cost and without considering factors such as human resources, geography and infrastructure, extrapolations are not valid because they assume constant marginal costs.</p></sec><sec><title>The recommendations are summarized in Table <xref ref-type="table" rid="T2">2</xref></title><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Recommendations for calculating and presenting cost results</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Viewpoint</td><td align="left">Use societal perspective.<break/>Sub-analyses can focus on specific perspectives such as provider, patient or Ministry of Health</td></tr></thead><tbody><tr><td align="left">Output</td><td align="left">Clarify whether intervention delivers nets or treatment or both<break/>Calculate net and treatment costs separately</td></tr><tr><td align="left">Resource identification</td><td align="left">Include all costs of behaviour change activities (including advertising)<break/>Include any costs of treating LLINs in campaigns<break/>Exclude research costs<break/>Include relevant overheads of collaborating organizations (e.g. NGO contributions to procurement, management, etc)<break/>Clarify what management capacity is assumed to exist and whether calculating average cost or incremental cost of adding intervention to an existing programme</td></tr><tr><td align="left">Resource measurement</td><td align="left">Attempt to measure and report leakage of resources (e.g. nets used by individuals outside target groups)<break/>Establish average lifespan of net under local conditions</td></tr><tr><td align="left">Resource valuation</td><td align="left">Treat nets as a capital item.<break/>Base case should assume life expectancy of 3 years and use discount rate of 3% Avoid double counting of user contributions where cost recovery applied, but ensure these are counted as user contributions when disaggregating costs by source</td></tr><tr><td align="left">Sensitivity analysis</td><td align="left">Conduct sensitivity analysis on: discount rate (0%, 5%, 10% at a minimum); frequency of net impregnation, procurement cost of net and insecticide; number of people protected per net; lifespan of net Consider impact of research where this is conducted alongside a programme on programme effectiveness (see ref 5)<break/>Vary proportion of shared resources where distribution carried out as part of another programme/campaign</td></tr><tr><td align="left">Reporting of results</td><td align="left">Provide costs in US$ and local currency<break/>Specify year in which costs calculated/adjusted<break/>Report cost per ITN delivered and cost per person sleeping under a net</td></tr><tr><td align="left">Discussion of results</td><td align="left">Compare costs with per capita government health expenditure to aid assessment of affordability<break/>Be cautious in using cost estimates to scale up, and make explicit assumptions about whether marginal cost constant, increasing or decreasing</td></tr></tbody></table></table-wrap></sec><sec><title>The reference case</title><p>To provide comparability between ITN costing studies it is suggested that analysts include a reference case scenario. If future costing studies adhere to this practice, it will finally be feasible to compare the costs of different options for ITN delivery. At the same time, authors maintain the freedom of analysing their data from other viewpoints that may be more relevant to their specific context. Based on the above recommendation the use of the following reference case scenario is proposed (Table <xref ref-type="table" rid="T3">3</xref>).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Reference case scenario</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Parameter</bold></td><td align="left"><bold>Suggested Reference Scenario</bold></td><td align="left"><bold>Explanation</bold></td></tr></thead><tbody><tr><td align="left">Perspective</td><td align="left">Societal</td><td align="left">To include all costs, not just those of the provider or patient.</td></tr><tr><td align="left">Currency</td><td align="left">US$</td><td align="left">A cost estimate in US$, indicating the year of conversion should be provided in addition to local currency</td></tr><tr><td align="left">Life-span of mosquito net</td><td align="left">3 years</td><td align="left">Assumptions have varied from 3 to 7 years, but field observations increasingly indicate a relatively short life span of polyester nets. Olyset<sup>® </sup>nets (made of polyethylene) are more durable and should be considered separately.</td></tr><tr><td align="left">Re-treatment</td><td align="left">Annually</td><td align="left">Treatment of mosquito nets with modern pyrethroids is generally assumed to last 6 – 12 months. Evidence for LLINs (Olyset<sup>® </sup>and PermaNet<sup>®</sup>) indicates that treatment lasts for the life span of the net [7, 8]. While LLINs are phased in, they may be retreated alongside conventional nets. It should be investigated if this is the case.</td></tr><tr><td align="left">Cost data</td><td align="left">Include:<break/>All intervention costs (e.g. mosquito nets, insecticide, wages, transport, advertising, etc.)<break/>All time cost, including care giving (formal/informal) and volunteers<break/>Transportation and other non-medical services<break/>Administrative costs for sick leave and for other transfers<break/>Donated items</td><td align="left">Costs for these and possible other ingredients (depending on programme specifics) should be collected. Costs can only be excluded once it has been established that they are insignificant in the context of the analysis [see reference 17 and 18 for further guidance). Quantities and prices need to be presented separately</td></tr><tr><td align="left">Revenue</td><td align="left">Value and include</td><td align="left">From the societal perspective, funds from cost-recovery need to be included. This needs careful attention to avoid double counting. Clearly indicate cost-recovery when presenting results.</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="left" colspan="3">Adjustment of financial costs to calculated economic costs</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="left">Annualisation</td><td align="left">Life expectancy of capital items as specified above</td><td align="left">To obtain an equivalent annual cost for each capital outlay, an annuitization procedure needs to be followed. This requires an estimate of the life expectancy of each capital item and a decision on the discount rate to be used (see below).</td></tr><tr><td align="left">Discount rate</td><td align="left">3%</td><td align="left">Base-case calculations should use 3%, to be consistent with World Bank recommendations [32]. This should be varied in the sensitivity analysis, e.g. from 0 – 10%.</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="left" colspan="3">Reporting of results</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="left">Cost estimate</td><td align="left">Cost per net ITN delivered</td><td align="left">For programmes delivering untreated nets and insecticide treatment, the cost of both components should be quoted separately. Indicate whether the cost per ITN is the composite of the two costs or is achieved at lower/higher cost</td></tr></tbody></table></table-wrap></sec></sec><sec><title>Conclusion</title><p>The cost of different ITN delivery strategies is important when deciding which ones to scale-up, yet it is one of the knowledge gaps remaining to be filled [<xref ref-type="bibr" rid="B10">10</xref>]. Many strategies have not been costed at all and only some of the existing cost-estimates have been derived using appropriate methods. Well-conducted studies have often used outputs that are difficult to compare. A template for costing of ITN interventions and reporting on results does not exist.</p><p>The limited role of health economics in generating essential evidence for scaling-up of ITNs cannot be explained by the absence of general guidelines or other supporting literature. More likely it partly results from the lack of ITN specific recommendations, which has led to the observed variation in methods and outputs, and partly from some authors being reluctant to use any economic methods, because these are misconceived as not applicable to health care [<xref ref-type="bibr" rid="B33">33</xref>]. Until these obstacles are overcome, decision makers will lack important data to guide their efforts of scaling-up coverage.</p><p>To advocate for the use of improved and standardised methods, this review has drawn attention to the variations in approaches taken to costing ITN programmes, and consequently of study outputs, and the limitations of these. This will hopefully allow ITN programme staff and their donors to recognise the potential impact that costing of their work could have, if results were readily comparable to other studies and could be interpreted in other contexts. The suggested Reference Case scenario is meant to further assist standardisation of new costing initiatives.</p></sec><sec><title>Authors' contributions</title><p>Kara Hanson suggested the topic for the review, provided advice on structure and content and was involved in re-drafting the paper. Jan Kolaczinski reviewed and summarised the literature, and wrote the first draft of the paper.</p></sec>
|
Intra-cluster correlation coefficients in adults with diabetes in primary care practices: the Vermont Diabetes Information System field survey
|
<sec><title>Background</title><p>Proper estimation of sample size requirements for cluster-based studies requires estimates of the intra-cluster correlation coefficient (ICC) for the variables of interest.</p></sec><sec sec-type="methods"><title>Methods</title><p>We calculated the ICC for 112 variables measured as part of the Vermont Diabetes Information System, a cluster-randomized study of adults with diabetes from 73 primary care practices (the clusters) in Vermont and surrounding areas.</p></sec><sec><title>Results</title><p>ICCs varied widely around a median value of 0.0185 (Inter-quartile range: 0.006, 0.037). Some characteristics (such as the proportion having a recent creatinine measurement) were highly associated with the practice (ICC = 0.288), while others (prevalence of some comorbidities and complications and certain aspects of quality of life) varied much more across patients with only small correlation within practices (ICC<0.001).</p></sec><sec><title>Conclusion</title><p>The ICC values reported here may be useful in designing future studies that use clustered sampling from primary care practices.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Littenberg</surname><given-names>Benjamin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>MacLean</surname><given-names>Charles D</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Medical Research Methodology
|
<sec><title>Background</title><p>Multi-level or clustered sampling designs are increasingly deployed in medical and health care surveys. In these designs, clusters are identified (<italic>e.g</italic>. medical practices) and then subjects (<italic>e.g</italic>. patients) are sampled from each cluster. The analysis and sample size estimation for such designs must take the clustering into account or the resultant significance tests (<italic>P </italic>values) and confidence intervals will be in error [<xref ref-type="bibr" rid="B1">1</xref>]. Generally, failure to account for clustering leads to nominal confidence intervals that are too narrow and to <italic>P </italic>values that are too small. To the extent that patient characteristics are independent of cluster, the effective sample size will be close to the number of individual subjects studied. If the subject characteristics are highly associated within clusters, the effective sample size approaches the number of clusters. In the extreme case, if all the subjects within a cluster are identical, there is no advantage to measuring more than one subject per cluster.</p><p>To estimate statistical power or required sample size in a study based on simple random sampling or allocation, one requires an estimate of the minimal important effect and (for continuous measures) the standard deviation of the outcome in the population studied. For clustered designs, however, one must also inflate the sample size to account for the clustering effect. The design effect, sometimes referred to as the variance inflation factor, is a function of the extent of correlation within clusters, the intraclass (or intra-cluster) correlation coefficient (ICC). Unfortunately, pre-study estimates of ICC are difficult to come by and obtaining them constitutes "the main difficulty in calculating sample size for cluster randomized studies" [<xref ref-type="bibr" rid="B2">2</xref>].</p><p>Several groups have published estimates of ICCs for various patient characteristics observed in large surveys of patients clustered within primary care or general practices from around the world [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B6">6</xref>]. Here we expand their estimates to include those derived from a survey of adults with diabetes clustered within primary care practices in the northeast United States.</p></sec><sec sec-type="methods"><title>Methods</title><p>This study was part of a larger project, the Vermont Diabetes Information System (VDIS), a cluster-randomized trial of a laboratory-based diabetes decision support system in a region-wide sample of 8808 adults with diabetes from 73 Primary Care practices in Vermont and nearby parts of the United States [<xref ref-type="bibr" rid="B7">7</xref>]. Primary care in these predominantly rural practices is provided by General Internists, Family Physicians, Physician Assistants, and Nurse Practitioners who provide the bulk of long-term care for these and other patients. There are few diabetes specialists in the region and most diabetes care is provided in the practices. All 119 eligible primary care practices near the thirteen participating hospitals were invited to participate [<xref ref-type="bibr" rid="B7">7</xref>]. The participating practices range in size from one provider (in 41 practices) to two practices with six providers each.</p><p>A field survey targeted at a sub-sample of subjects was designed to provide a better understanding of the non-laboratory features of the patients before intervention. Field survey subjects were selected at random from the patients participating in the VDIS and invited by telephone to participate in an in-home interview. Patient names were randomly sorted and patients contacted until a sample of approximately 15% of the patients from each practice agreed to an interview. We attempted to contact 4,209 patients and reached 1,576 (37%). Of these, 1,006 (64%) agreed to be interviewed.</p><p>Subjects who agreed were mailed a questionnaire and were scheduled for an interview by a trained field interviewer. During the visit, the interviewer reviewed any missing or ambiguous questionnaire items. If necessary, the interviewer read the questions aloud for subjects and recorded their responses for them. Then the interviewer measured the subject as described below and administered a few more instruments that were not included in the questionnaire. The interviews took place during the baseline phase of the study before any interventions were in place. All subjects provided written informed consent. The protocol was approved by the institutional review board of the University of Vermont.</p><sec><title>Demographic, social and economic characteristics</title><p>Income was recorded in seven ordered self-reported categories from less than US$15,000 per year to US$100,000 per year or more. Education was also recorded as the highest level completed in seven categories from "Less than 9th Grade" to "Graduate or Professional Degree." We collapsed self-reported race and ethnicity into two categories: Non-Hispanic white and all others. Marital status was collapsed into two categories: Married or living as married <italic>vs</italic>. all others (single, widowed, divorced or separated). We recorded the presence or absence of four types of health insurance: private (commercial indemnity or health maintenance organization benefits often supplied by an employer), Medicare (government health coverage for the elderly and disabled), Medicaid (government health coverage for low income patients), and military (including active duty or veteran's benefits). Subjects may have more than one insurance type.</p><p>The shortest driving distance from the patients' homes to their site of care was calculated in kilometers using ArcView 3.3 by Environmental Systems Research Institute, Inc., and a geographic data set purchased from TeleAtlas, Inc. Driving distance was defined as the shortest distance along roads and highways [<xref ref-type="bibr" rid="B8">8</xref>].</p></sec><sec><title>Physical characteristics</title><p>Height was measured using a portable stadiometer (SECA, Inc.), weight with a portable scale (LB Dial Scale HAP200KD-41, Healthometer, Inc.), and blood pressure with an automated sphygmomanometer (Omron Model HEM-711). Blood pressure was obtained in the seated position in the left arm (unless contraindicated), using the cuff size recommended by the manufacturer. Three readings were obtained at 5-minute intervals and were averaged for the final result. Body mass index was calculated as weight in kilograms divided by height in meters squared.</p></sec><sec><title>Laboratory results</title><p>Glycosolated hemoglobin A1C was measured at 13 clinical laboratories in the patients' home communities. All laboratories used the same high-pressure liquid chromatography method with identical reference ranges. Serum creatinine, urine microalbumin-to-creatinine ratio, total cholesterol, high density lipoprotein cholesterol, and triglycerides were likewise measured by the laboratories. Low density lipoprotein cholesterol (LDL) was calculated using the Friedwald formula (LDL = Total cholesterol - high density lipoprotein cholesterol - triglycerides/5) [<xref ref-type="bibr" rid="B9">9</xref>] from fasting specimens. Each patient was classified as being above or below certain laboratory value thresholds recommended by the American Diabetes Association (A1C >8%; A1C <7%; microalbumin-to-creatinine ratio <30 mg/mmol) [<xref ref-type="bibr" rid="B10">10</xref>]. If the LDL was 100 mg/dl or greater, or if it could not be calculated because the triglycerides were above 400 mg/dl, we categorized lipids as above goal. Tests were ordered by the primary care provider when clinically indicated. We report the most recent laboratory assays done before the home visit.</p></sec><sec><title>Quality of care</title><p>Where possible, we classified each subject as meeting or not meeting recommendations for care made by the American Diabetes Association [<xref ref-type="bibr" rid="B10">10</xref>] and the Vermont Program for Quality in Health Care [<xref ref-type="bibr" rid="B11">11</xref>]. Creatinine and urine microalbumin tests were due every year. A1C was on time if the latest test was within 3 months (6 months if the latest result was <7.0%). Lipid testing was on time if the latest test was within 6 months (12 months if the latest result showed LDL-cholesterol under 100 mg/dl). Additional measures indicate if the subject was both on time and had results on target for A1C and LDL. Pneumonia vaccine was recommended once ever. Influenza vaccine was considered up to date if the patient reported it was given in the current or previous calendar year.</p></sec><sec><title>Health habits</title><p>Alcohol consumption was measured by asking: "How many drinks of the following alcoholic beverages do you have in a typical week (including weekends)?</p><p>Bottles or cans of beer: _________</p><p>Glasses of wine or wine coolers: _________</p><p>Mixed drinks or shots of liquor: _________"</p><p>Subjects who indicated that they do not currently drink alcohol were assigned zero to each of the three beverage categories. A summary variable representing total consumption was constructed as the sum of the three beverage-specific responses. Subjects were also asked the four CAGE screening questions [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Tobacco use was assessed by asking: "Have you smoked a cigarette – even one puff – during the past seven days?" Those responding "yes" were asked "How many cigarettes do you smoke on an average day?"</p></sec><sec><title>Self care</title><p>We assessed self-care behavior with the Summary of Diabetes Self Care Activities Measure [<xref ref-type="bibr" rid="B13">13</xref>]. This instrument asks the subject to record how many days in the last week they performed recommended self-care activities such as following a healthful eating plan, or participating in at least 30 minutes of physical activity. Eleven items are used to generate 5 summary scores representing the fraction of days the subject performs recommended activities related to general diet, diabetes-specific diet, exercise, blood glucose self-monitoring, and foot care. Each score ranges from 0 to 100.</p></sec><sec><title>Literacy</title><p>The Short Test of Functional Health Literacy in Adults (STOFHLA) is a 7-minute timed instrument that measures the ability to read health-related material [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. The score ranges from 0 to 36 items answered correctly. Responses can be categorized at "inadequate" (STOFHLA 0–16), "marginal" (STOFHLA 17–22), and "adequate" (STOFHLA 23–36).</p></sec><sec><title>Comorbidity</title><p>The Self-Administered Comorbidity Questionnaire is a modification of the widely used Charlson Index. It uses patient interview or questionnaire responses rather than chart abstraction for assessment of comorbidity and has excellent agreement with the chart-based Charlson Index [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. We calculated the rate of endorsement of each of 18 specific conditions as well the number of conditions endorsed. We also calculated a score with one point if the condition is endorsed and additional points if the subject reports currently receiving treatment for it, or if it limits activities. Each condition may, therefore, contribute 0 to 3 points for a possible maximum of 54 points. One of the conditions, "eye, nerve, or kidney damage due to diabetes" may be considered a complication of diabetes rather than strictly a comorbidity.</p></sec><sec><title>Functional status and depression</title><p>The Medical Outcomes Trust SF-12 Health Survey is a widely used, validated instrument for assessment of general (rather than disease-specific) functional status [<xref ref-type="bibr" rid="B18">18</xref>]. Two summary scales are calculated: the Physical Component Summary and the Mental Component Summary. The Patient Health Questionnaire-9 is a brief self-report instrument that quantifies the presence and degree of mental depression [<xref ref-type="bibr" rid="B19">19</xref>].</p></sec><sec><title>Complications</title><p>We assessed the presence of diabetes complications by asking six questions. The responses were "Yes," "No," and "Don't know."</p><p>1. Have you ever had an ulcer or sore on your leg or foot that took more than 4 weeks to heal?</p><p>Has your doctor or health care provider ever told you that you have these problems:</p><p>2. Problems with vision or retinopathy <underline>related to your diabetes</underline>?</p><p>3. Pain, burning, or numbness in the feet or legs <underline>related to your diabetes</underline>?</p><p>4. Problems with stomach emptying <underline>related to your diabetes</underline>?</p><p>5. Problems with sexual function?"</p><p>6. Problems with your kidneys <underline>related to your diabetes</underline>?</p></sec><sec><title>Medications</title><p>The subjects were asked to produce "all medications you have used in the past month including prescriptions, over-the-counter products, vitamins, and herbs." The field assistant recorded the name, strength, dose, route, and frequency of each preparation.</p></sec><sec><title>Quality of life</title><p>The Audit of Diabetes-Dependant Quality of Life is an 18-item questionnaire regarding the impact of diabetes on specific aspects of a person's life with patient weighting of the impact of each domain [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>]. We employed 17 of the 18 domains of this instrument. The scores for each domain can range from -9 (maximum negative impact of diabetes on that domain) to +9 (maximum positive impact).</p></sec><sec><title>Resource utilization</title><p>The survey included items asking the subjects to record whether they had used various services in the last year: Endocrinologist, Dietician, Podiatrist, Diabetes Educator, Ophthalmologist, and Diabetes Class. Those answering "Don't Know" were assigned a value of "No." It also prompted subjects to report the number of Emergency Room visits they had in the last year and "In the past month, how many times have you been to a doctor or health care professional?"</p></sec><sec><title>Statistical analyses</title><p>In the random effects model, the ICC is the proportion of the total variance that is between clusters (practices).</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2288-6-20-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mi>C</mml:mi><mml:mo>=</mml:mo><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>b</mml:mi><mml:mn>2</mml:mn></mml:msubsup><mml:mo>/</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>b</mml:mi><mml:mn>2</mml:mn></mml:msubsup><mml:mo>+</mml:mo><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>w</mml:mi><mml:mn>2</mml:mn></mml:msubsup><mml:mo stretchy="false">)</mml:mo></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGjbqscqWGdbWqcqWGdbWqcqGH9aqpiiGacqWFdpWCdaqhaaWcbaGaemOyaigabaGaeGOmaidaaOGaei4la8IaeiikaGIae83Wdm3aa0baaSqaaiabdkgaIbqaaiabikdaYaaakiabgUcaRiab=n8aZnaaDaaaleaacqWG3bWDaeaacqaIYaGmaaGccqGGPaqkaaa@4137@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>where <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M2" name="1471-2288-6-20-i2" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>b</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaaiiGacqWFdpWCdaqhaaWcbaGaemOyaigabaGaeGOmaidaaaaa@30E2@</mml:annotation></mml:semantics></mml:math></inline-formula> is the between-cluster component of variance while <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M3" name="1471-2288-6-20-i3" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msubsup><mml:mi>σ</mml:mi><mml:mi>w</mml:mi><mml:mn>2</mml:mn></mml:msubsup></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaaiiGacqWFdpWCdaqhaaWcbaGaem4DaChabaGaeGOmaidaaaaa@310C@</mml:annotation></mml:semantics></mml:math></inline-formula> is the within-cluster component. If a measurement varies across patients without regard to which practice they are in, the ICC will be close to zero. If the value of the variable is largely a function of which practice they are in, the ICC will be close to 1.0 [<xref ref-type="bibr" rid="B2">2</xref>]. We used the analysis of variance estimator [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>] provided by the "loneway" command in STATA 8.2 (Stata Corp., College Station, Texas). This estimator uses the F statistic to calculate the ICC for N total subjects in k groups of size N<sub>o</sub>:</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M4" name="1471-2288-6-20-i4" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mi>C</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mi>o</mml:mi><mml:mi>b</mml:mi><mml:mi>s</mml:mi></mml:mrow></mml:msub><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mrow><mml:msub><mml:mi>F</mml:mi><mml:mrow><mml:mi>o</mml:mi><mml:mi>b</mml:mi><mml:mi>s</mml:mi></mml:mrow></mml:msub><mml:mo>−</mml:mo><mml:mn>1</mml:mn><mml:mo>+</mml:mo><mml:mi>g</mml:mi></mml:mrow></mml:mfrac></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGjbqscqWGdbWqcqWGdbWqcqGH9aqpdaWcaaqaaiabdAeagnaaBaaaleaaieGacqWFVbWBcqWFIbGycqWFZbWCaeqaaOGaeyOeI0IaeGymaedabaGaemOray0aaSbaaSqaaiab=9gaVjab=jgaIjab=nhaZbqabaGccqGHsislcqaIXaqmcqGHRaWkcqWGNbWzaaaaaa@41BD@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>where</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M5" name="1471-2288-6-20-i5" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mi>g</mml:mi><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>N</mml:mi><mml:mo>−</mml:mo><mml:mstyle displaystyle="true"><mml:msub><mml:mo>∑</mml:mo><mml:mi>i</mml:mi></mml:msub><mml:mrow><mml:msubsup><mml:mi>N</mml:mi><mml:mi>o</mml:mi><mml:mn>2</mml:mn></mml:msubsup><mml:mo>/</mml:mo><mml:mi>N</mml:mi></mml:mrow></mml:mstyle></mml:mrow><mml:mrow><mml:mi>k</mml:mi><mml:mo>−</mml:mo><mml:mn>1</mml:mn></mml:mrow></mml:mfrac><mml:mo>.</mml:mo></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGNbWzcqGH9aqpdaWcaaqaaiabd6eaojabgkHiTmaaqababaGaemOta40aa0baaSqaaiabd+gaVbqaaiabikdaYaaakiabc+caViabd6eaobWcbaGaemyAaKgabeqdcqGHris5aaGcbaGaem4AaSMaeyOeI0IaeGymaedaaiabc6caUaaa@3E53@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>Further,</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M6" name="1471-2288-6-20-i6" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msqrt><mml:mrow><mml:mo stretchy="false">(</mml:mo><mml:mi>V</mml:mi><mml:mo stretchy="false">)</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mi>C</mml:mi><mml:mo stretchy="false">)</mml:mo></mml:mrow></mml:msqrt></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaadaGcaaqaaiabcIcaOiabdAfawjabcMcaPiabcIcaOiabdMeajjabdoeadjabdoeadjabcMcaPaWcbeaaaaa@3499@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>is the asymptomatic standard error of the ICC, and the 100(1-α)% confidence interval is:</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M7" name="1471-2288-6-20-i7" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mi>C</mml:mi><mml:mo>±</mml:mo><mml:msub><mml:mi>z</mml:mi><mml:mrow><mml:mi>α</mml:mi><mml:mo>/</mml:mo><mml:mn>2</mml:mn></mml:mrow></mml:msub><mml:msqrt><mml:mrow><mml:mo stretchy="false">(</mml:mo><mml:mi>V</mml:mi><mml:mo stretchy="false">)</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mi>I</mml:mi><mml:mi>C</mml:mi><mml:mi>C</mml:mi><mml:mo stretchy="false">)</mml:mo></mml:mrow></mml:msqrt><mml:mo>.</mml:mo></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGjbqscqWGdbWqcqWGdbWqcqGHXcqScqWG6bGEdaWgaaWcbaacciGae8xSdeMaei4la8IaeGOmaidabeaakmaakaaabaGaeiikaGIaemOvayLaeiykaKIaeiikaGIaemysaKKaem4qamKaem4qamKaeiykaKcaleqaaOGaeiOla4caaa@3FDF@</mml:annotation></mml:semantics></mml:math></inline-formula></p><p>For each characteristic, we recorded the sample size (N), the sample size per cluster (N<sub>o</sub>) the mean (or proportion for dichotomous variables), the standard deviation (SD) for continuous variables, the standard error of the mean (or percentage) adjusted for clustering within practices (SE), the ICC, and the 95% confidence interval of the ICC. We assessed the association between the value (reported proportion) of binary variables and the ICC [<xref ref-type="bibr" rid="B25">25</xref>] with Spearman's non-parametric correlation coefficient. For proportions greater than 0.5, we used the complement of the proportion so that all proportions for this analysis were less than 0.5. To compare groups of ICCs, we used the two-sample Wilcoxon rank-sum (Mann-Whitney) test.</p></sec></sec><sec><title>Results</title><p>The results appear in Table <xref ref-type="table" rid="T1">1</xref>. The 112 ICCs ranged from 0 for 15 variables with negative values truncated at zero to 0.288 for the proportion with a creatinine measurement on time. The median value was 0.0185 with an inter-quartile range of (0.006, 0.037). Results were similar for 62 binary variables (median 0.022; IQR 0.006, 0.040) and 50 continuous variables (median 0.017; IQR 0.006, 0.032). A Wilcoxon rank-sum test gave a <italic>P </italic>value of 0.54 for the comparison between ICCs of continuous and binary variables.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Descriptive statistics and intra-practice correlation coefficients</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Variable</td><td align="center">N</td><td align="center">N<sub>o</sub></td><td align="center">Mean or percent</td><td align="center">SD</td><td align="center">SE</td><td align="center">ICC</td><td align="center">95% confidence interval</td></tr></thead><tbody><tr><td align="left" colspan="8"><bold>Demographic, social and economic characteristics</bold></td></tr><tr><td align="left">Sex (% male)</td><td align="center">8808</td><td align="center">119.7</td><td align="center">48.4</td><td></td><td align="center">1.3</td><td align="center">0.038</td><td align="center">0.021, 0.056</td></tr><tr><td align="left">Age (years)</td><td align="center">8286</td><td align="center">117.3</td><td align="center">63.1</td><td align="center">13.9</td><td align="center">0.5</td><td align="center">0.077</td><td align="center">0.045, 0.109</td></tr><tr><td align="left">Married (%)</td><td align="center">1004</td><td align="center">14.5</td><td align="center">62.5</td><td></td><td align="center">1.6</td><td align="center">0.009</td><td align="center">0, 0.036</td></tr><tr><td align="left">High School Graduate (%)</td><td align="center">999</td><td align="center">14.4</td><td align="center">75.4</td><td></td><td align="center">1.6</td><td align="center">0.011</td><td align="center">0, 0.038</td></tr><tr><td align="left">Income <$30,000 per year (%)</td><td align="center">931</td><td align="center">13.4</td><td align="center">59.0</td><td></td><td align="center">2.1</td><td align="center">0.042</td><td align="center">0.003, 0.081</td></tr><tr><td align="left">Private insurance (%)</td><td align="center">1001</td><td align="center">14.4</td><td align="center">58.4</td><td></td><td align="center">2.0</td><td align="center">0.050</td><td align="center">0.011, 0.090</td></tr><tr><td align="left">Medicare (%)</td><td align="center">997</td><td align="center">14.4</td><td align="center">59.9</td><td></td><td align="center">1.7</td><td align="center">0.018</td><td align="center">0, 0.047</td></tr><tr><td align="left">Medicaid (%)</td><td align="center">995</td><td align="center">14.3</td><td align="center">21.4</td><td></td><td align="center">1.8</td><td align="center">0.036</td><td align="center">0, 0.071</td></tr><tr><td align="left">Military or veterans' insurance (%)</td><td align="center">994</td><td align="center">14.3</td><td align="center">5.2</td><td></td><td align="center">0.8</td><td align="center">0.029</td><td align="center">0, 0.062</td></tr><tr><td align="left">No insurance (%)</td><td align="center">993</td><td align="center">14.3</td><td align="center">2.4</td><td></td><td align="center">0.6</td><td align="center">0.035</td><td align="center">0, 0.070</td></tr><tr><td align="left">Travel distance from home to Primary Care provider (km)</td><td align="center">2955</td><td align="center">76.5</td><td align="center">13.7</td><td align="center">15.3</td><td align="center">1.8</td><td align="center">0.196</td><td align="center">0.102, 0.289</td></tr><tr><td align="left">Non-Hispanic white (%)</td><td align="center">1004</td><td align="center">14.5</td><td align="center">97.3</td><td></td><td align="center">0.5</td><td align="center">0.003</td><td align="center">0, 0.028</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Physical characteristics</bold></td></tr><tr><td align="left">Heart Rate (beats per min)</td><td align="center">998</td><td align="center">14.4</td><td align="center">75.0</td><td align="center">13.0</td><td align="center">0.5</td><td align="center">0.015</td><td align="center">0, 0.043</td></tr><tr><td align="left">Systolic blood pressure (mmHg)</td><td align="center">999</td><td align="center">14.4</td><td align="center">140.3</td><td align="center">19.6</td><td align="center">0.8</td><td align="center">0.042</td><td align="center">0.005, 0.079</td></tr><tr><td align="left">Diastolic blood pressure (mmHg)</td><td align="center">999</td><td align="center">14.4</td><td align="center">78.3</td><td align="center">10.5</td><td align="center">0.4</td><td align="center">0.017</td><td align="center">0, 0.047</td></tr><tr><td align="left">Blood pressure below 130/80 mmHg (%)</td><td align="center">999</td><td align="center">14.4</td><td align="center">25.0</td><td></td><td align="center">1.6</td><td align="center">0.028</td><td align="center">0, 0.061</td></tr><tr><td align="left">Blood pressure over 140/90 mmHg (%)</td><td align="center">999</td><td align="center">14.4</td><td align="center">49.2</td><td></td><td align="center">2.3</td><td align="center">0.069</td><td align="center">0.024, 0.114</td></tr><tr><td align="left">Height (cm)</td><td align="center">998</td><td align="center">14.4</td><td align="center">165.2</td><td align="center">10.4</td><td align="center">0.4</td><td align="center">0.019</td><td align="center">0, 0.048</td></tr><tr><td align="left">Weight (kg)</td><td align="center">997</td><td align="center">14.4</td><td align="center">203.0</td><td align="center">47.7</td><td align="center">1.6</td><td align="center">0.011</td><td align="center">0, 0.038</td></tr><tr><td align="left">Body Mass Index (kg/m<sup>2</sup>)</td><td align="center">994</td><td align="center">14.3</td><td align="center">33.8</td><td align="center">7.4</td><td align="center">0.2</td><td align="center">0.011</td><td align="center">0, 0.039</td></tr><tr><td align="left">Body Mass Index >30 kg/m<sup>2 </sup>(%)</td><td align="center">994</td><td align="center">14.3</td><td align="center">67.2</td><td></td><td align="center">1.6</td><td align="center">0.010</td><td align="center">0, 0.038</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Laboratory results</bold></td></tr><tr><td align="left">Mean glycosolated hemoglobin (A1C) (%)</td><td align="center">8711</td><td align="center">118.4</td><td align="center">7.01</td><td align="center">1.45</td><td align="center">0.05</td><td align="center">0.055</td><td align="center">0.032, 0.079</td></tr><tr><td align="left">A1C >8.0% (%)</td><td align="center">8711</td><td align="center">118.4</td><td align="center">18.1</td><td></td><td align="center">1.0</td><td align="center">0.025</td><td align="center">0.013, 0.037</td></tr><tr><td align="left">A1C <7.0 (%)</td><td align="center">8711</td><td align="center">118.4</td><td align="center">61.7</td><td></td><td align="center">1.6</td><td align="center">0.046</td><td align="center">0.026, 0.066</td></tr><tr><td align="left">Total cholesterol (mg/dl)</td><td align="center">8167</td><td align="center">112.5</td><td align="center">184.4</td><td align="center">41.8</td><td align="center">1.2</td><td align="center">0.037</td><td align="center">0.020, 0.054</td></tr><tr><td align="left">LDL-cholesterol (mg/dl)</td><td align="center">7834</td><td align="center">111.0</td><td align="center">105.8</td><td align="center">34.2</td><td align="center">1.0</td><td align="center">0.045</td><td align="center">0.025, 0.065</td></tr><tr><td align="left">LDL <100 mg/dl (%) *</td><td align="center">7873</td><td align="center">111.6</td><td align="center">43.8</td><td></td><td align="center">1.4</td><td align="center">0.029</td><td align="center">0.020, 0.055</td></tr><tr><td align="left">Triglycerides (mg/dl)</td><td align="center">7969</td><td align="center">109.8</td><td align="center">190.5</td><td align="center">161.3</td><td align="center">3.3</td><td align="center">0.014</td><td align="center">0.006, 0.022</td></tr><tr><td align="left">Serum creatinine (mg/dl)</td><td align="center">8474</td><td align="center">115.1</td><td align="center">1.12</td><td align="center">0.70</td><td align="center">0.02</td><td align="center">0.080</td><td align="center">0.048, 0.112</td></tr><tr><td align="left">Urine microalbumin to creatinine ratio (mg/mmol)</td><td align="center">3039</td><td align="center">46.7</td><td align="center">47.0</td><td align="center">260.7</td><td align="center">10.0</td><td align="center">0.101</td><td align="center">0.051, 0.152</td></tr><tr><td align="left">Urine microalbumin to creatinine ratio <30 mg/mmol (%)</td><td align="center">3338</td><td align="center">49.0</td><td align="center">67.3</td><td></td><td align="center">1.5</td><td align="center">0.031</td><td align="center">0.011, 0.051</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Quality of care (Process Measures)</bold></td></tr><tr><td align="left">Creatinine on time (%)</td><td align="center">8808</td><td align="center">119.7</td><td align="center">79.4</td><td></td><td align="center">3.1</td><td align="center">0.288</td><td align="center">0.203, 0.373</td></tr><tr><td align="left">Urine microalbumin to creatinine ratio on time (%)</td><td align="center">8808</td><td align="center">119.7</td><td align="center">24.6</td><td></td><td align="center">2.3</td><td align="center">0.162</td><td align="center">0.105, 0.219</td></tr><tr><td align="left">A1C on time (%)</td><td align="center">8808</td><td align="center">119.7</td><td align="center">50.7</td><td></td><td align="center">2.4</td><td align="center">0.118</td><td align="center">0.074, 0.162</td></tr><tr><td align="left">Lipids on time (%)</td><td align="center">8808</td><td align="center">119.7</td><td align="center">70.3</td><td></td><td align="center">3.0</td><td align="center">0.199</td><td align="center">0.132, 0.265</td></tr><tr><td align="left">A1C on time & <7% (%)</td><td align="center">7308</td><td align="center">113.2</td><td align="center">35.5</td><td></td><td align="center">1.5</td><td align="center">0.040</td><td align="center">0.021, 0.060</td></tr><tr><td align="left">LDL on time & <100 mg/dl (%)</td><td align="center">7308</td><td align="center">113.2</td><td align="center">31.3</td><td></td><td align="center">1.4</td><td align="center">0.040</td><td align="center">0.021, 0.059</td></tr><tr><td align="left">Influenza vaccine given (%)</td><td align="center">995</td><td align="center">14.3</td><td align="center">80.5</td><td></td><td align="center">1.4</td><td align="center">0.058</td><td align="center">0.016, 0.100</td></tr><tr><td align="left">Pneumonia vaccine given (%)</td><td align="center">913</td><td align="center">13.2</td><td align="center">70.6</td><td></td><td align="center">1.8</td><td align="center">0.058</td><td align="center">0.014, 0.102</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Health habits</bold></td></tr><tr><td align="left">Alcoholic drinks per week</td><td align="center">988</td><td align="center">14.2</td><td align="center">1.6</td><td align="center">4.5</td><td align="center">0.2</td><td align="center">0.027</td><td align="center">0, 0.060</td></tr><tr><td align="left">CAGE score (0–4)</td><td align="center">946</td><td align="center">13.6</td><td align="center">0.2</td><td align="center">0.7</td><td align="center">0.2</td><td align="center">0.009</td><td align="center">0, 0.038</td></tr><tr><td align="left">Tobacco smoker (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">16.9</td><td></td><td align="center">1.2</td><td align="center">0.005</td><td align="center">0, 0.030</td></tr><tr><td align="left">Cigarettes per day</td><td align="center">159</td><td align="center">2.6</td><td align="center">17.2</td><td align="center">11.2</td><td align="center">1.0</td><td align="center">0.086</td><td align="center">0, 0.270</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Summary of Diabetes Self Care Activities Measure</bold></td></tr><tr><td align="left">General diet score</td><td align="center">953</td><td align="center">13.7</td><td align="center">58.4</td><td align="center">33.3</td><td align="center">1.2</td><td align="center">0.017</td><td align="center">0, 0.048</td></tr><tr><td align="left">Specific diet score</td><td align="center">980</td><td align="center">14.1</td><td align="center">52.2</td><td align="center">23.9</td><td align="center">0.8</td><td align="center">0.011</td><td align="center">0, 0.038</td></tr><tr><td align="left">Exercise score</td><td align="center">992</td><td align="center">14.3</td><td align="center">34.6</td><td align="center">32.7</td><td align="center">1.2</td><td align="center">0.029</td><td align="center">0, 0.062</td></tr><tr><td align="left">Blood glucose self-monitoring score</td><td align="center">933</td><td align="center">13.5</td><td align="center">57.9</td><td align="center">38.9</td><td align="center">1.5</td><td align="center">0.051</td><td align="center">0.010, 0.093</td></tr><tr><td align="left">Foot care score</td><td align="center">990</td><td align="center">14.3</td><td align="center">44.2</td><td align="center">35.5</td><td align="center">1.2</td><td align="center">0.002</td><td align="center">0, 0.026</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Literacy: Short Test of Functional Health Literacy in Adults</bold></td></tr><tr><td align="left">Test score (0–36 points)</td><td align="center">1002</td><td align="center">14.4</td><td align="center">29.7</td><td align="center">9.6</td><td align="center">0.3</td><td align="center">0.017</td><td align="center">0, 0.079</td></tr><tr><td align="left">Inadequate (0–16) (%)</td><td align="center">1002</td><td align="center">14.4</td><td align="center">10.5</td><td></td><td align="center">1.0</td><td align="center">0.011</td><td align="center">0, 0.039</td></tr><tr><td align="left">Adequate (23–36) (%)</td><td align="center">1002</td><td align="center">14.4</td><td align="center">82.9</td><td></td><td align="center">1.5</td><td align="center">0.037</td><td align="center">0.002, 0.073</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Comorbidity</bold></td></tr><tr><td align="left">Congestive heart failure (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">17.1</td><td></td><td align="center">1.4</td><td align="center">0.025</td><td align="center">0, 0.057</td></tr><tr><td align="left">Coronary artery disease (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">19.3</td><td></td><td align="center">1.4</td><td align="center">0.024</td><td align="center">0, 0.056</td></tr><tr><td align="left">Peripheral vascular disease (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">8.7</td><td></td><td align="center">1.0</td><td align="center">0.006</td><td align="center">0, 0.032</td></tr><tr><td align="left">Stroke (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">11.7</td><td></td><td align="center">1.1</td><td align="center">0.006</td><td align="center">0, 0.032</td></tr><tr><td align="left">Dementia (%) †</td><td align="center">1006</td><td align="center">14.5</td><td align="center">0.9</td><td></td><td align="center">0.3</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Asthma (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">20.2</td><td></td><td align="center">1.4</td><td align="center">0.014</td><td align="center">0, 0.043</td></tr><tr><td align="left">Arthritis (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">14.1</td><td></td><td align="center">1.1</td><td align="center"><0.001</td><td align="center">0, 0.024</td></tr><tr><td align="left">Peptic ulcer disease (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">14.3</td><td></td><td align="center">1.2</td><td align="center">0.012</td><td align="center">0, 0.040</td></tr><tr><td align="left">Cirrhosis (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">1.9</td><td></td><td align="center">0.3</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Paralysis (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">3.0</td><td></td><td align="center">0.6</td><td align="center">0.006</td><td align="center">0, 0.032</td></tr><tr><td align="left">Renal disease (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">5.0</td><td></td><td align="center">0.6</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Microvascular complications of diabetes (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">16.9</td><td></td><td align="center">1.2</td><td align="center">0.012</td><td align="center">0, 0.040</td></tr><tr><td align="left">Cancer (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">11.9</td><td></td><td align="center">1.0</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Leukemia (%) †</td><td align="center">1006</td><td align="center">14.5</td><td align="center">0.6</td><td></td><td align="center">0.2</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Lymphoma (%) †</td><td align="center">1006</td><td align="center">14.5</td><td align="center">0.5</td><td></td><td align="center">0.3</td><td align="center">0.018</td><td align="center">0, 0.048</td></tr><tr><td align="left">Metastatic cancer (%) †</td><td align="center">1006</td><td align="center">14.5</td><td align="center">0.7</td><td></td><td align="center">0.3</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">HIV disease (%) †</td><td align="center">1006</td><td align="center">14.5</td><td align="center">0.1</td><td></td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.024</td></tr><tr><td align="left">Depression (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">35.0</td><td></td><td align="center">1.7</td><td align="center">0.007</td><td align="center">0, 0.033</td></tr><tr><td align="left">Comorbid condition count</td><td align="center">1006</td><td align="center">14.5</td><td align="center">1.8</td><td align="center">1.7</td><td align="center">0.1</td><td align="center">0.031</td><td align="center">0, 0.065</td></tr><tr><td align="left">Self-reported comorbidity questionnaire score (0–54)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">3.6</td><td align="center">3.9</td><td align="center">0.2</td><td align="center">0.032</td><td align="center">0, 0.066</td></tr><tr><td align="left">Cardiovascular disease (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">30.9</td><td></td><td align="center">1.6</td><td align="center">0.018</td><td align="center">0, 0.048</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Functional status</bold></td></tr><tr><td align="left">SF-12 Physical Component Summary score (0–100)</td><td align="center">986</td><td align="center">14.2</td><td align="center">41.2</td><td align="center">12.4</td><td align="center">0.5</td><td align="center">0.028</td><td align="center">0, 0.062</td></tr><tr><td align="left">SF-12 Mental Component Summary score (0–100)</td><td align="center">986</td><td align="center">14.2</td><td align="center">50.0</td><td align="center">10.7</td><td align="center">0.4</td><td align="center">0.032</td><td align="center">0, 0.061</td></tr><tr><td align="left">Patient Health Questionnaire-9 depression score (0–27)</td><td align="center">591</td><td align="center">11.7</td><td align="center">4.1</td><td align="center">5.0</td><td align="center">0.2</td><td align="center">0.006</td><td align="center">0, 0.044</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Complications of diabetes</bold></td></tr><tr><td align="left">Foot ulcer (%)</td><td align="center">992</td><td align="center">14.3</td><td align="center">11.3</td><td></td><td align="center">1.0</td><td align="center">0.009</td><td align="center">0, 0.036</td></tr><tr><td align="left">Vision problems (%)</td><td align="center">950</td><td align="center">13.7</td><td align="center">20.1</td><td></td><td align="center">1.3</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Foot or leg pain (%)</td><td align="center">943</td><td align="center">13.6</td><td align="center">30.9</td><td></td><td align="center">1.8</td><td align="center">0.023</td><td align="center">0, 0.055</td></tr><tr><td align="left">Stomach emptying problems (%)</td><td align="center">901</td><td align="center">13.0</td><td align="center">6.2</td><td></td><td align="center">0.8</td><td align="center">0.003</td><td align="center">0, 0.030</td></tr><tr><td align="left">Sexual problems (%)</td><td align="center">923</td><td align="center">13.3</td><td align="center">26.4</td><td></td><td align="center">1.4</td><td align="center">0 §</td><td align="center">0, 0.026</td></tr><tr><td align="left">Kidney problems (%)</td><td align="center">490</td><td align="center">10.3</td><td align="center">9.2</td><td></td><td align="center">1.4</td><td align="center">0.040</td><td align="center">0, 0.096</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Medications</bold></td></tr><tr><td align="left">Insulin (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">18.6</td><td></td><td align="center">1.4</td><td align="center">0.014</td><td align="center">0, 0.042</td></tr><tr><td align="left">Oral hypoglycemic agent (%)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">66.6</td><td></td><td align="center">1.7</td><td align="center">0.022</td><td align="center">0, 0.053</td></tr><tr><td align="left">Total medication count (0–29)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">8.8</td><td align="center">4.5</td><td align="center">0.2</td><td align="center">0.028</td><td align="center">0, 0.061</td></tr><tr><td align="left">Prescription medication count (0–24)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">6.7</td><td align="center">3.8</td><td align="center">0.2</td><td align="center">0.055</td><td align="center">0.014, 0.095</td></tr><tr><td align="left">Non-prescription medication count (0–15)</td><td align="center">1006</td><td align="center">14.5</td><td align="center">2.0</td><td align="center">2.4</td><td align="center">0.1</td><td align="center">0.059</td><td align="center">0.017, 0.101</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Audit of Diabetes-Dependant Quality of Life</bold></td></tr><tr><td align="left">Do physically</td><td align="center">985</td><td align="center">14.2</td><td align="center">-1.9</td><td align="center">2.6</td><td align="center">0.1</td><td align="center">0.001</td><td align="center">0, 0.025</td></tr><tr><td align="left">Confidence in ability</td><td align="center">982</td><td align="center">14.2</td><td align="center">-2.0</td><td align="center">2.7</td><td align="center">0.1</td><td align="center">0.004</td><td align="center">0, 0.030</td></tr><tr><td align="left">Motivation</td><td align="center">981</td><td align="center">14.2</td><td align="center">-1.9</td><td align="center">2.7</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Freedom to eat</td><td align="center">985</td><td align="center">14.2</td><td align="center">-2.8</td><td align="center">2.9</td><td align="center">0.1</td><td align="center"><0.001</td><td align="center">0, 0.025</td></tr><tr><td align="left">Enjoyment of food</td><td align="center">986</td><td align="center">14.2</td><td align="center">-2.5</td><td align="center">2.9</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Future</td><td align="center">984</td><td align="center">14.2</td><td align="center">-0.6</td><td align="center">3.8</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Freedom to drink</td><td align="center">983</td><td align="center">14.2</td><td align="center">-1.5</td><td align="center">2.4</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Dependence</td><td align="center">982</td><td align="center">14.2</td><td align="center">-0.6</td><td align="center">2.8</td><td align="center">0.1</td><td align="center">0.004</td><td align="center">0, 0.029</td></tr><tr><td align="left">Working life</td><td align="center">981</td><td align="center">14.2</td><td align="center">-1.5</td><td align="center">2.3</td><td align="center">0.1</td><td align="center">0.014</td><td align="center">0, 0.042</td></tr><tr><td align="left">Family life</td><td align="center">985</td><td align="center">14.2</td><td align="center">-0.6</td><td align="center">2.7</td><td align="center">0.1</td><td align="center">0.035</td><td align="center">0, 0.070</td></tr><tr><td align="left">Social life</td><td align="center">983</td><td align="center">14.2</td><td align="center">-0.9</td><td align="center">2.0</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Sex life</td><td align="center">978</td><td align="center">14.1</td><td align="center">-1.0</td><td align="center">2.4</td><td align="center">0.1</td><td align="center">0.020</td><td align="center">0, 0.051</td></tr><tr><td align="left">Physical appearance</td><td align="center">980</td><td align="center">14.1</td><td align="center">-1.1</td><td align="center">2.3</td><td align="center">0.1</td><td align="center">0.012</td><td align="center">0, 0.041</td></tr><tr><td align="left">Holidays/Leisure</td><td align="center">983</td><td align="center">14.2</td><td align="center">-1.5</td><td align="center">2.5</td><td align="center">0.1</td><td align="center">0 §</td><td align="center">0, 0.025</td></tr><tr><td align="left">Travel</td><td align="center">981</td><td align="center">14.1</td><td align="center">-1.6</td><td align="center">2.6</td><td align="center">0.1</td><td align="center">0.022</td><td align="center">0, 0.054</td></tr><tr><td align="left">Society reaction</td><td align="center">977</td><td align="center">14.1</td><td align="center">-0.5</td><td align="center">1.6</td><td align="center">0.1</td><td align="center">0.026</td><td align="center">0, 0.058</td></tr><tr><td align="left">Finances</td><td align="center">978</td><td align="center">14.1</td><td align="center">-1.7</td><td align="center">2.7</td><td align="center">0.1</td><td align="center">0.009</td><td align="center">0, 0.037</td></tr><tr><td align="left">Average weighted impact</td><td align="center">990</td><td align="center">14.3</td><td align="center">-1.5</td><td align="center">2.0</td><td align="center">0.1</td><td align="center">0.004</td><td align="center">0, 0.029</td></tr><tr><td colspan="8"><hr></hr></td></tr><tr><td align="left" colspan="8"><bold>Resource utilization</bold></td></tr><tr><td align="left">Endocrinologist in last year (%)</td><td align="center">971</td><td align="center">14.0</td><td align="center">15.9</td><td></td><td align="center">2.0</td><td align="center">0.089</td><td align="center">0.037, 0.140</td></tr><tr><td align="left">Dietician in last year (%)</td><td align="center">984</td><td align="center">14.2</td><td align="center">34.3</td><td></td><td align="center">1.8</td><td align="center">0.018</td><td align="center">0, 0.048</td></tr><tr><td align="left">Podiatrist in last year (%)</td><td align="center">984</td><td align="center">14.2</td><td align="center">23.8</td><td></td><td align="center">1.6</td><td align="center">0.022</td><td align="center">0, 0.054</td></tr><tr><td align="left">Diabetes Educator in last year (%)</td><td align="center">970</td><td align="center">14.0</td><td align="center">23.8</td><td></td><td align="center">1.8</td><td align="center">0.041</td><td align="center">0.004, 0.078</td></tr><tr><td align="left">Ophthalmologist in last year (%)</td><td align="center">944</td><td align="center">13.6</td><td align="center">60.4</td><td></td><td align="center">2.1</td><td align="center">0.041</td><td align="center">0.003, 0.079</td></tr><tr><td align="left">Diabetes class in last year (%)</td><td align="center">999</td><td align="center">14.4</td><td align="center">35.0</td><td></td><td align="center">1.9</td><td align="center">0.035</td><td align="center">0.002, 0.073</td></tr><tr><td align="left">Emergency Room visits in last year (mean)</td><td align="center">994</td><td align="center">14.3</td><td align="center">0.64</td><td align="center">1.46</td><td align="center">0.05</td><td align="center">0.015</td><td align="center">0, 0.044</td></tr><tr><td align="left">Health professional visits in last month (mean)</td><td align="center">991</td><td align="center">14.3</td><td align="center">1.63</td><td align="center">2.39</td><td align="center">0.08</td><td align="center">0.035</td><td align="center">0, 0.070</td></tr></tbody></table><table-wrap-foot><p>N = Subjects analyzed; N<sub>o </sub>= Number per cluster; SD = standard deviation of mean; SE = standard error adjusted for clustering; ICC = Intra-cluster (intra-practice) correlation coefficient.</p><p>*In 39 cases in which LDL could not be estimated because triglycerides were over 400 mg/dl, LDL was assumed to be over 100 mg/dl.</p><p>† Because only a few subjects have this condition, the ICC may not be estimated consistently.</p><p>§The method used truncates the estimate of ICC (and its CI) at 0 for some values that may be negative.</p></table-wrap-foot></table-wrap><p>The ICCs for the 62 binary variables were significantly associated with their proportions (Spearman's correlation coefficient = 0.53; <italic>P </italic>< 0.0001) to a degree sometimes classified as "Large" [<xref ref-type="bibr" rid="B26">26</xref>].</p></sec><sec><title>Discussion</title><p>These data provide estimates of intra-cluster correlations for 112 patient characteristics relevant to the analysis of adults with diabetes receiving primary care in Vermont and nearby regions of the United States. They may usefully be applied to the design and sample size estimation of future surveys that are clustered on primary care practices.</p><p>In the design of clustered-based studies, the ICC may be used to calculate the design effect which is the degree to which the sample size must be inflated above that of a simple random sample to account for the loss of information inherent in the clustered design. If the average number of subjects sampled per cluster is <italic>m</italic>, the design effect is given by:</p><p>Design effect = 1 + (<italic>m</italic>-1)·ICC</p><p>If <italic>m </italic>or ICC is large, the total number of individual subjects needed may be substantially greater than suggested by a sample size calculation that is not adjusted for clustering. Alternatively, if both <italic>m </italic>and ICC are small, the design effect may be very close to 1.0 indicating that the clustered design does not inflate the sample size.</p><p>In the VDIS, the cost of enrolling subjects for laboratory data collection within a cluster, <italic>once the practice was enrolled</italic>, was relatively low. Therefore, large values of <italic>m </italic>(120.7 subjects per practice on average) were not problematic. However, the cost per patient of obtaining interview data was relatively high. Therefore, we reduced the mean sample size per cluster to 14.5 by random sampling within practice. The design effects experienced in VDIS are not representative of those faced in other designs unless they happen to have the same mean cluster sample size as VDIS (which is extremely unlikely). Therefore, unlike previous publications, we have elected not to report design effects. Study designers should use the ICCs and their own estimates of <italic>m </italic>to understand their own design effects.</p><p>Campbell et al [<xref ref-type="bibr" rid="B27">27</xref>] suggest that ICCs are higher for process measures than for outcomes. We see evidence for this in that the eight quality of care variables (process measures, see Table <xref ref-type="table" rid="T1">1</xref>) have a median ICC of 0.088 (IQR 0.049, 0.181) while the ten laboratory outcomes measures have a median ICC of 0.038 (IQR 0.029, 0.055). This difference is significant by Wilcoxon rank-sum test with <italic>P </italic>= 0.013. We note that the nine physical characteristics of the subjects, presumably under less control of the provider than either laboratory results or even process measures, have a median ICC of 0.017 (IQR 0.011, 0.028) and are significantly different than the laboratory measures (<italic>P </italic>= 0.034).</p><p>Within practice correlation was most prominent for process measures associated with quality of care. The likelihood of receiving a creatinine measurement on time had the highest ICC (0.288) with other quality of care measures also among the most highly correlated. This may represent that process measures are heavily influenced by the practice style of the practitioners and any office-based procedures (reminders, registries, flow sheets, etc.) that only some practices employ. In a similar vein, physiologic control of some aspects of diabetes (especially achievement of tight control of A1C and LDL) appears to vary importantly across practices with ICCs of 0.046 for A1C below 7% and 0.029 for LDL below 100 mg/dl.</p><p>Some demographic aspects of the population were correlated within practices: age, sex, and especially travel burden. As patients tend to stay with their primary provider as they age, some practices accumulate older patients (ICC = 0.077). Some patients express a preference for same-sex providers, with more women visiting practices that have female providers. This may account for the relatively high ICC for sex (0.038). Travel distance may be related to the geographic location of the practice office. Practices in more densely populated areas may tend to have lower typical travel burdens.</p><p>With the possible exception of blood pressure (which may be under the control of the providers to some degree), the physical characteristics and health habits of patients vary little across practices.</p><p>Although apparently under the control of the providers, the utilization of health care services had generally small ICCs. The exception was consultation by an Endocrinologist with an ICC of 0.089. This may reflect the geographic proximity of an endocrinologist to some of the practices.</p><p>The aspects of diabetes that are most directly experienced by patients (complications, quality of life, functional status, comorbidity, and self-care) vary little across practices. It does not appear that some primary care practices tend to accumulate more complicated or difficult patients than others. Likewise, low health literacy is a substantial, and perhaps unrecognized, problem for all practices, with little clustering within practices.</p><p>These data demonstrate a large correlation between the proportion of the 62 binary variables and their ICCs. This finding has been noted by others [<xref ref-type="bibr" rid="B25">25</xref>] and may be useful in estimating an ICC for sample size calculations.</p><p>For many of these variables, the impact of within-cluster correlation on sample size requirements appears to be relatively small. Thirty-three variables had an ICC <0.010 with a design effect less than 1.14 indicating that the VDIS clustered design required an increase in sample size of 14% compared to a non-clustered study. However, depending on the number of clusters and the number of subjects per cluster, even small ICCs can result in the need for costly increases in overall sample size.</p><p>These estimates of ICC are not useful for studies that cluster on factors other than practice (such as community, hospital, individual provider, family, classroom, <italic>etc</italic>.). The VDIS study population was drawn from Vermont and nearby New York and New Hampshire and is, therefore, predominantly white and rural. All the subjects were under care for diabetes. Although some have Type 1 diabetes, this older, overweight population is largely comprised of patients with Type 2 diabetes. The practices in the VDIS are small with a median of 2 providers per practice. For these reasons, generalization to other populations and settings may be problematic.</p><p>Several recent reports provide some comparisons from other settings (Table <xref ref-type="table" rid="T2">2</xref>). A study of British patients aged 75 and older reported intra-practice correlations from 106 general practices [<xref ref-type="bibr" rid="B4">4</xref>]. A study from Rhode Island and nearby Massachusetts enrolled 15 primary care practices [<xref ref-type="bibr" rid="B5">5</xref>]. Several surveys of general practices from Australia and New Zealand provided a few ICCs comparable to VDIS [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. The ICCs for most of these variables vary substantially across the studies. For instance, the ICCs for weight and body mass index varied between 0.011 and 0.081. Differences in the practice structures, referral patterns, social and geographic factors and practice patterns may explain these differences. Only recently have determinants of ICC begun to be studied [<xref ref-type="bibr" rid="B26">26</xref>-<xref ref-type="bibr" rid="B28">28</xref>]. We suggest that more catalogues of ICCs, drawn from a variety of settings, will be useful both to investigators designing new clustered studies and to researchers investigating the role of setting on patient characteristics.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Intra-practice correlation coefficients from four recent studies</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Characteristic</td><td align="center">VDIS</td><td align="center">BEACH</td><td align="center">CEART</td><td align="center">MRC</td><td align="center">HPOP</td><td align="center">JOGS</td></tr></thead><tbody><tr><td align="left">Age</td><td align="center">0.077</td><td align="center">0.153–0.159</td><td></td><td></td><td></td><td align="center">0.050</td></tr><tr><td align="left">Sex</td><td align="center">0.038</td><td align="center">0.055–0.066</td><td></td><td></td><td></td><td></td></tr><tr><td align="left">Height</td><td align="center">0.019</td><td></td><td align="center">0.053</td><td align="center">0.048</td><td></td><td></td></tr><tr><td align="left">Weight</td><td align="center">0.011</td><td></td><td align="center">0.020</td><td align="center">0.012</td><td></td><td align="center">0.043</td></tr><tr><td align="left">Body mass index</td><td align="center">0.011</td><td></td><td align="center">0.031</td><td align="center">0.022</td><td></td><td align="center">0.081</td></tr><tr><td align="left">Systolic blood pressure</td><td align="center">0.042</td><td></td><td align="center">0.048</td><td></td><td></td><td align="center">0.018</td></tr><tr><td align="left">Diastolic blood pressure</td><td align="center">0.017</td><td></td><td align="center">0.129</td><td></td><td></td><td align="center">0.046</td></tr><tr><td align="left">Heart rate</td><td align="center">0.015</td><td></td><td></td><td align="center">0.032</td><td></td><td></td></tr><tr><td align="left">Total cholesterol</td><td align="center">0.037</td><td></td><td align="center"><0.001</td><td></td><td></td><td align="center">0.004</td></tr><tr><td align="left">LDL-cholesterol</td><td align="center">0.045</td><td></td><td align="center">0.006</td><td></td><td></td><td></td></tr><tr><td align="left">LDL-cholesterol at goal</td><td align="center">0.038</td><td></td><td align="center">0.027</td><td></td><td></td><td></td></tr><tr><td align="left">Triglycerides</td><td align="center">0.014</td><td></td><td align="center">0.024</td><td></td><td></td><td></td></tr><tr><td align="left">Serum creatinine</td><td align="center">0.080</td><td></td><td align="center">0.00007</td><td></td><td></td><td></td></tr><tr><td align="left">Tobacco use</td><td align="center">0.005</td><td></td><td align="center">0.036</td><td align="center">0.006</td><td></td><td align="center">0.055</td></tr><tr><td align="left">Number of medications</td><td align="center">0.028</td><td></td><td></td><td></td><td align="center">0</td><td></td></tr><tr><td align="left">Influenza vaccination</td><td align="center">0.058</td><td></td><td></td><td></td><td align="center">0.062</td><td></td></tr><tr><td align="left">Depressed</td><td align="center">0.007</td><td></td><td></td><td></td><td align="center">0</td><td></td></tr></tbody></table><table-wrap-foot><p>VDIS = Vermont Diabetes Information System (the present study) [7].</p><p>BEACH = Bettering the Evaluation and Care of Health [3].</p><p>CEART = Cholesterol Education and Research Trial [5].</p><p>MRC = Medical Research Council Trial of the Assessment and Management of Older People in the Community [4].</p><p>HPOP = Health Promotion for Older People [6].</p><p>JOGS = Justification of Green Scripts trial [6]</p></table-wrap-foot></table-wrap></sec><sec><title>Conclusion</title><p>Intra-practice correlation coefficients in this survey of adults receiving care for diabetes varied widely. Some characteristics (such as the likelihood of having a recent creatinine measurement) were highly associated with the practice (ICC = 0.288), while others (prevalence of some comorbidities and complications and certain aspects of quality of life) varied much more across patients with virtually no correlation within practices (ICC<0.001). The values reported here may be useful in designing future clustered studies.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>BL conceptualized the research and performed the analyses. CDM oversaw the data collection and management. Both authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2288/6/20/prepub"/></p></sec>
|
IL-10, IL-6 and CD14 polymorphisms and sepsis outcome in ventilated very low birth weight infants
|
<sec><title>Background</title><p>Genetic variation in the innate immune system of the host may play a role in determining the risk of developing infection, as well as outcome from infection.</p></sec><sec sec-type="methods"><title>Methods</title><p>Infectious complications were retrospectively determined in 293 (233 African-American (AA), 57 Caucasian and 3 Hispanic) mechanically ventilated very low birth weight (VLBW) infants (<1500 grams at birth) who were genotyped for the IL-6 -174 G/C, IL-10 -1082 G/A and CD14 -260 C/T single nucleotide polymorphisms (SNPs).</p></sec><sec><title>Results</title><p>The IL-6 -174C allele was associated with an increased incidence of late blood stream infection (BSI) in AA but not Caucasian infants. In AA infants with the C allele the incidence of late BSI was 20/29 (69%) compared to 94/204 (46%) in homozygous GG infants (RR 2.6, 95% CI: 1.1–6.0, p = 0.021). The IL-10 -1082A allele was associated with an increased incidence of late BSI. One or more episodes of late BSI developed in 14 (35%) of 40 infants with the GG genotype, 71 (49%) of 145 infants with the GA genotype and 63 (58%) of 108 infants with the AA genotype (p = 0.036). Infants with the A allele (AA or GA genotypes) had an incidence of late BSI that was 134/253 (53%) compared to 14/40 (35%) in homozygous GG infants (RR 2.1, 95% CI: 1.04–4.19, p = 0.035). The CD14 -260 C/T SNP did not alter the overall risk for BSI in ventilated VLBW infants. Multiple BSI episodes were more common in the TT genotype group (CC: 17%, CT: 11%, TT: 30%, p = 0.022). This effect was due to the strong effect of the TT genotype on the incidence of multiple BSI in AA infants (CC: 15%, CT: 11%, TT: 39%, p = 0.003).</p></sec><sec><title>Conclusion</title><p>The IL-6 -174 G/C, IL-10 -1082 G/A and CD14 -260 C/T SNPs may alter risk for BSI in ventilated VLBW infants.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Baier</surname><given-names>R John</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Loggins</surname><given-names>John</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Yanamandra</surname><given-names>Krishna</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Medicine
|
<sec><title>Background</title><p>Sepsis is a persistent and vexing problem in very low birth weight (VLBW) infants. The development of sepsis in this population has several adverse implications including; prolongation of hospitalization, development of chronic lung disease, adverse neurodevelopmental outcome and excess mortality [<xref ref-type="bibr" rid="B1">1</xref>]. The incidence of one or more episodes of late onset blood stream infection (BSI) in this population ranges from 25–30% with higher rates for infants with birth weights less than 1000 grams [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B5">5</xref>].</p><p>Genetic variation in the innate immune system of the host may play a role in determining the risk of developing and outcome from infection. Variation in the ability to recognize pathogens may influence the risk of infection. One of the primary molecules that function in recognition of pathogens is CD14. CD14 is expressed on phagocytic cells, and along with LPS-binding protein, it acts to transfer lipopolysaccharide (LPS) and other bacterial ligands to the Toll-like receptor 4 (TLR4)/MD-2 signaling complex [<xref ref-type="bibr" rid="B6">6</xref>]. The engagement of CD14 and LPS-binding protein during recognition of Gram-negative bacteria results in activation of a complex of innate host defense mechanisms. CD14 expression is influenced by a single nucleotide polymorphism (SNP) found in the proximal CD14 promoter. A C to T substitution 260 base pairs (bp) prior to the start codon diminishes binding of the inhibitory factor sp3 resulting in increased expression of the CD14 gene [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. Homozygous carriers of the T allele have significant increases in both soluble and membrane-bound CD14 [<xref ref-type="bibr" rid="B9">9</xref>]. The CD14 -260 T allele may increase modify the risk for septic shock [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>Variation in the magnitude of inflammatory response may also influence the risk of sepsis. Outcome from sepsis depends to a considerable degree on the host response. While an absent or diminished host response leads to overwhelming infection, an excessive response can lead to systemic inflammation and multiple organ failure [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B15">15</xref>].</p><p>Interleukin-6 (IL-6) is a proinflammatory cytokine that plays an important role in the host response to infection [<xref ref-type="bibr" rid="B16">16</xref>]. The C allele of IL-6 -174 G/C promoter polymorphism is associated with increased IL-6 production in newborn infants [<xref ref-type="bibr" rid="B17">17</xref>]. The IL-6 -174 G/C SNP has been inconsistently associated with altered risk for and outcome from sepsis in several studies [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>].</p><p>Interleukin-10 (IL-10) is an anti-inflammatory cytokine produced by macrophages and T-helper-type II (TH2) lymphocytes that downregulates inflammatory mediator production by stimulated immune and epithelial cells [<xref ref-type="bibr" rid="B20">20</xref>-<xref ref-type="bibr" rid="B22">22</xref>]. Thus, IL-10 can potentially counterbalance the detrimental effects of excessive cytokine production in sepsis. The SNP at -1082 (G to A) modifies IL-10 secretion and may influence outcome in several disease states [<xref ref-type="bibr" rid="B23">23</xref>-<xref ref-type="bibr" rid="B25">25</xref>]. The IL-10 -1082 G/A SNP lies within a putative Ets transcription site and is associated (A allele) with lower IL-10 production in vitro [<xref ref-type="bibr" rid="B26">26</xref>]. The IL-10 -1082 SNP may modify the response to sepsis from a variety of organisms [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B27">27</xref>].</p><p>Variation in the IL-10, IL-6 and CD14 genes may be genetic factors influencing the development and outcome of sepsis in the premature newborn. The purpose of this study was to determine if there is a relationship between the IL-10 -1082 G/A, IL-6 -174 G/C and CD14 -260 C/T SNPs and risk for or outcome from sepsis in mechanically ventilated very low birth weight (VLBW) infants.</p></sec><sec sec-type="methods"><title>Methods</title><p>The genomic DNA used for this case controlled study was extracted from archival tracheal aspirate (TA) pellets (259) or blood (34 patients) collected prospectively as part of an ongoing study of genetic factors in the development of complications of prematurity. The TAs that were used as a source of genomic DNA were originally collected as part of long term longitudinal studies examining cytokine concentrations and the development of CLD [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B29">29</xref>]. Infants were included in the study if they fulfilled the following inclusion criteria: birth weight less than 1500 grams, mechanical ventilation (MV) during the first week of life, complete clinical data on infectious outcome and genomic DNA sample that could be used for genotyping. Infants were excluded if complete data on outcome were not available or suitable DNA was not available. In order to compare the frequency of the various polymorphisms with normal term infants, control DNA was extracted from cord blood spots from a random sample of 168 African-American and 96 Caucasian term infants (performed independently as part of another study examining genotype and asthma). Consent was obtained from the parents of study infants to use the TAs and blood samples. The study was approved by the Institutional Review Board for Human Research at Louisiana State University Health Sciences Center in Shreveport.</p><p>Cultures for genital mycoplasmas were performed on TA samples collected in the first few days of life. Clinical and outcome data were abstracted from the clinical record and included information on respiratory outcome, survival and development of complications of prematurity. The results of all cultures of blood, tracheal aspirates and cerebrospinal fluid were recorded from the patients' charts. Tracheal aspirates were obtained twice a week in all intubated patients according to unit policy. Infants were evaluated for sepsis at the discretion of the clinical staff when signs and symptoms compatible with sepsis developed. Generally, 2 blood cultures from separate sites were obtained when assessing infants for blood stream infection. For the purposes of this study any positive blood culture was considered a blood stream infection (bacteremia or fungemia). Blood stream infections (BSI) were divided into early if the culture was obtained during the first 3 days of life, and late (nosocomial) if it was obtained thereafter. Nosocomial pneumonia was diagnosed when there was radiological evidence of a new pulmonary infiltrate and the blood and endotracheal aspirate culture grew the same organism. Isolation of an organism from a TA culture in an infant greater than 3 days of age without a positive blood culture or a change in chest radiograph was considered colonization. Sepsis mortality was defined as mortality during an acute episode of sepsis.</p><sec><title>Laboratory methods</title><p>Isolation of total DNA from blood or TA pellets was performed using the QIAmp DNA Mini kits™ (Qiagen Incorporated, Chatsworth, CA). Briefly, TA pellets were suspended in 200 μl of sterile phosphate buffered saline by vigorous vortexing, then digested with proteinase K and applied to silica gel spin columns. Columns were washed with the manufacturer's supplied buffers and the total DNA was eluted in 200 μl elution buffer. Blood (200 μl) was extracted similarly to the TA pellets.</p><p>The IL-10 -1082 G/A and IL-6 -174 G/C SNPs were genotyped using published allele specific PCR methods [<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B30">30</xref>]. The CD14 -26/0 CT SNP was genotyped using a published restriction fragment length polymorphism method [<xref ref-type="bibr" rid="B9">9</xref>].</p></sec><sec><title>Data analysis</title><p>Data analysis consisted of comparing the incidence of infections and their complications between the various genotypes. All statistical analysis was performed using SPSS for Windows version 6.0 (SPSS Inc., Chicago, IL). Differences in frequencies of complications were assessed by Chi square. ANOVA or the Student t-test was used to assess normally distributed variables where appropriate. The Wilcox Rank Sum test was used for analysis of factors that were not normally distributed. A probability value of less than 0.05 was considered statistically significant. The data are presented as mean ± standard error of the mean (SEM).</p></sec></sec><sec><title>Results</title><p>Two hundred and ninety-three (293) patients had complete culture and clinical information. Two hundred and thirty-three (79%) were African-American, 57 (20%) were Caucasian and 3 (1%) were Hispanic. Mean gestational age and mean birth weight of the study population were 26.7 ± 0.1 weeks and 906 ± 13 grams, respectively. Male: female ratio was 176:117. All patients required MV at birth and 264 (90%) infants were treated with exogenous surfactant therapy (Survanta<sup>®</sup>, Ross Products Division, Abbott Laboratories, Columbus, OH). One hundred and fifty eight (54%) infants were oxygen dependent at 28 days and 64 (22%) were oxygen dependent at 36 weeks PCA. There were 40 (14%) patients who died during their initial hospitalization (25 before 28 days of age and 15 after 28 days). Late onset BSI was the major cause of late mortality in ventilated VLBW infants with 14/15 deaths occurring after 28 days of age directly attributable to BSI (RR 15.3, 95% CI 2.0–117.9; p < 0.001).</p><p>One patient had positive blood cultures (Group B streptococci) during the first 3 days of life (early BSI) and 147 (50%) had one or more episodes of late (nosocomial) BSI. Coagulase negative staphylococci (CONS) were the organisms most commonly isolated from blood cultures. Seventy-one (24%) infants had one or more episodes of bacteremia/fungemia other than from CONS and 46 (15.7%) had multiple episodes of bacteremia or fungemia. Organisms causing nosocomial BSI (bacteremia/fungemia) for the study population are shown in Table <xref ref-type="table" rid="T1">1</xref>. There were 15 (5%) deaths directly attributable to sepsis. Fourteen of 15 sepsis related deaths were associated with a non-CONS BSI (p < 0.001, RR = 59; 95% CI 7–421).</p><p>Endotracheal (ET) cultures obtained during the first 3 days of life grew <italic>Ureaplasma urealyticum </italic>in 86 (29%), <italic>Mycoplasma hominis </italic>in 35 (12%) and other bacteria in 12 (4%) infants. Subsequent bacterial colonization of the endotracheal tube (ETT) was detected in 145 (49%) patients. After birth, ETTs were most commonly colonized with CONS, less frequently with other organisms (Table <xref ref-type="table" rid="T1">1</xref>). Nosocomial pneumonia was diagnosed in 22 (7%) infants (Table <xref ref-type="table" rid="T1">1</xref>).</p><sec><title>IL-6 -174 G/C polymorphism and infectious complications</title><p>The frequency of the IL-6 -174 C allele in the study population was 0.128. The frequency of the C allele was significantly less in African-American (0.07) than in Caucasian infants (0.385) (p < 0.001). Two hundred and four (87.6%) African-American infants were homozygous GG, 27 (11.6%) were heterozygous GC and 2 (0.8%) were homozygous CC. In Caucasian infants the distribution of genotypes was: 19 (33.3%) GG, 32 (56.1%) GC and 6 (10.5%) CC. The 3 Hispanic infants were all GG. There were no differences in the frequency of the IL-6 -174C allele between study infants and ethnically matched control (term) infants [see <xref ref-type="supplementary-material" rid="S1">Additional File 1</xref>].</p><p>Birth weight, gestational age, gender, TA isolation of Uu or Mh and need for surfactant replacement did not differ among genotype groups in either Caucasian or African American infants. Because the frequency of the IL-6 -174C allele was significantly different between ethnic groups, the effects of this SNP on infectious complications were analyzed separately for Caucasian and African-American infants.</p><p>The IL-6 -174C allele was associated with an increased incidence of late BSI in African-American infants. One or more episodes of late BSI developed in 94 (46%) of 204 infants with the GG genotype, 18 (67%) of 27 infants with the GC genotype and 2 (100%) of 2 infants with the CC genotype (p = 0.046). In infants who had the C allele (CC or GC genotypes) the incidence of late BSI was 20/29 (69%) compared to 94/204 (46%) in infant who were homozygous GG (RR 2.6, 95% CI: 1.1–6.0, p = 0.021). There were no significant differences in organism-specific BSI rates among genotype groups (Table <xref ref-type="table" rid="T2">2</xref>). Neither the rates of non-CONS BSI (GG: 24%, GC: 44%, CC: 0%; p = 0.116) or multiple BSI episodes (GG: 15%, GC: 19%, CC: 0%, p = 0.752) were different among genotype groups.</p><p>The IL-6 -174 G/C polymorphism had no effect on sepsis-related mortality in African-American infants. Overall sepsis mortality was (GG: 5%, GC: 4%, CC: 0%; p = 0.916). The rate of sepsis mortality for African-American infants with late BSI was (GG: 11%, GC: 6%, CC: 0%; p = 0.717).</p><p>The rate of colonization of ETT tubes and incidence of nosocomial pneumonia were not affected by the IL-6 -174 G/C SNP. Nosocomial pneumonia rates were: (GG: 8%, GC: 7%, CC: 0%; p = 0.909). ETT colonization rates were 25/54 (46%) for infants with the GG genotype, 68/128 (53%) GC genotype, and 52/113 (46%) CC genotype (p = 0.489). CONS were the most common bacteria colonizing the ET tube in both groups. Organism-specific colonization rates were not different among genotype groups.</p><p>In contrast to that observed in African-American infants, the IL-6 -174 G/C SNP had no effect on the incidence of late BSI in Caucasian infants. One or more episodes of late BSI developed in 12 (63%) of 19 infants with the GG genotype, 16 (50%) of 32 infants with the GC genotype and 3 (50%) of 6 infants with the CC genotype (p = 0.643). The incidence of fungal BSI was increased in Caucasian infants with CC genotype (2/6 (33%)) compared to infants with the GG (5%) and GC (3%) (p = 0.027) (p = 0.008 comparing CC vs. GC and GG). There were no other significant differences in organism specific BSI rates among genotype groups (Table <xref ref-type="table" rid="T2">2</xref>). Neither the rates of non-CONS BSI (GG: 21%, GC: 28%, CC: 33%; p = 0.827) nor multiple BSI episodes (GG: 21%, GC: 13%, CC: 33%, p = 0.415) were different among genotype groups.</p><p>The IL-6 -174 G/C SNP had no effect on sepsis related-mortality in Caucasian infants. Overall sepsis mortality was (GG: 5%, GC: 6%, CC: 17%; p = 0.614). The rate of sepsis mortality for Caucasian infants with late BSI was (GG: 8%, GC: 7%, CC: 33%; p = 0.383).</p><p>The rate of colonization of ETT tubes and incidence of nosocomial pneumonia in Caucasian infants were not affected by the IL-6 -174 G/C polymorphism. Nosocomial pneumonia rates were: (GG: 5%, GC: 7%, CC: 0%; p = 0.813). ETT colonization rates were: 11/19 (58%) for infants with the GG genotype, 14/32 (44%) GC genotype and 3/6 (50%) CC genotype (p = 0.620). Organism-specific colonization rates were not different among genotype groups.</p><p>The incidences of necrotizing enterocolitis (NEC), CSF infection and infections in other sites were not different among genotype groups in either African-American or Caucasian infants.</p></sec><sec><title>IL-10 -1082 G/A polymorphism and infectious complications</title><p>The frequency of the IL-10 -1082A allele in the study population was 0.62. The frequency of the A allele was similar between African-American (0.63) and Caucasian infants (0.55) (p = 0.191). Forty (13.7%) infants were homozygous GG, 145 (47.3%) were heterozygous GA and 108 (36.9%) were homozygous AA. Distributions of genotypes were not significantly different between African-American and Caucasian infants (p = 0.088). There were no differences in the frequency of the IL-10 -1082 A allele between study infants and ethnically matched control (term) infants [see <xref ref-type="supplementary-material" rid="S1">Additional File 1</xref>]. However, the frequency of the A allele was significantly different between Caucasian and African-American controls [see <xref ref-type="supplementary-material" rid="S1">Additional File 1</xref>].</p><p>Birth weight, gestational age, gender, TA isolation of Uu or Mh and need for surfactant replacement were not different among genotype groups in either Caucasian or African American infants.</p><p>The IL-10 -1082A allele was associated with an increased incidence of late BSI in our study infants. One or more episodes of late BSI developed in 14 (35%) of 40 infants with the GG genotype, 71 (49%) of 145 infants with the GA genotype and 63 (58%) of 108 infants with the AA genotype (p = 0.036). In infants who had the A allele (AA or GA genotypes) the incidence of late BSI was 134/253 (53%) compared to 14/40 (35%) in infant who were homozygous GG (RR 2.1, 95% CI: 1.04–4.19, p = 0.035). Although they were not statistically significant, similar trends were observed when African-American and Caucasian infants were analyzed separately [see <xref ref-type="supplementary-material" rid="S2">Additional File 2</xref>, <xref ref-type="supplementary-material" rid="S3">Additional File 3</xref>, <xref ref-type="supplementary-material" rid="S4">Additional File 4</xref>]. There were no significant differences in organism-specific BSI rates among genotype groups (Table <xref ref-type="table" rid="T3">3</xref>). The rates of non-CONS BSI (GG: 15%, GA: 40%, AA: 26%; p = 0.676) or multiple BSI episodes (GG: 10%, GA: 16%, AA: 18%, p = 0.528) were not different among genotype groups. There was a trend towards increased incidence of non-CONS BSI in infants with the IL-10 -1082A allele (5/40 infants with GG compared to 66/253 infants GA/AA; p = 0.062).</p><p>The IL-10 -1082 G/A SNP had no effect on sepsis-related mortality. Overall sepsis mortality was: GG: 3%, GA: 6%, AA: 6% (p = 0.721). Sepsis mortality in infants with late BSI was: GG: 7%, GA: 11%, AA: 10% (p = 0.877).</p><p>The IL-10 -10814 G/A SNP did not affect ETT colonization or incidence of nosocomial pneumonia. Nosocomial pneumonia rates were: (GG: 11%, GA: 6%, A: 9%; p = 0.620). ETT colonization rates were: 18/40 (45%) for infants with the GG genotype, 74/145 (51%) GA genotype, and 53/108 (49%) AA genotype (p = 0.791). Organism-specific colonization rates were not different among genotype groups.</p><p>The incidences of necrotizing enterocolitis (NEC), CSF infection and infections in other sites were also not different among genotype groups in either African-American or Caucasian infants.</p></sec><sec><title>CD14 -260 C/T polymorphism and infectious complications</title><p>The frequency of the CD14-260T allele in the study population was 0.33. The frequency of the T allele was similar between African-American (0.32) and Caucasian infants (0.39) (p = 0.151). One hundred and thirty-two (45%) infants were homozygous CC, 128 (44.2%) were heterozygous CT and 33 (11.3%) were homozygous TT. The frequency of the CD24 -260T allele was significantly greater in Caucasian term infants than in Caucasian VLBW and African-American term infants [see <xref ref-type="supplementary-material" rid="S1">Additional File 1</xref>]. Birth weight, gestational age, gender, TA isolation of Uu or Mh and need for surfactant replacement were not different among genotype groups in either Caucasian or African American infants.</p><p>The CD14 -260CT SNP had no effect on the overall incidence of late BSI. One or more episodes of late BSI developed in 63 (48%) of 132 infants with the CC genotype, 66 (52%) of 128 infants with the CT genotype and 19 (58%) of 33 infants with the CC genotype (p = 0.570). However, there were significant differences in organism-specific rates of BSI between African-American and Caucasian infants (Table <xref ref-type="table" rid="T4">4</xref>). In particular, in African-American infants, Gram-negative BSI was associated with the CC genotype (Incidence of infection CC: 12%, CT: 2% and TT: 9%, p = 0.020). There was a trend for non-CONS BSI to be more frequent in infants with the TT genotype (CC: 24%, CT: 20%, TT: 39%; p = 0.074). Multiple BSI episodes were more common in the TT genotype group (CC: 17%, CT: 11%, TT: 30%, p = 0.022). This effect was due to the strong effect of the TT genotype on the incidence of multiple BSI in African-American infants (CC: 15%, CT: 11%, TT: 39%, p = 0.003).</p><p>The CD14 -260 CT SNP had no effect on sepsis-related mortality. Overall sepsis mortality was (CC: 6%, CT: 4%, TT: 6%; p = 0.709). Sepsis mortality in infants with late BSI was (CC: 10%, CT: 8%, TT: 20% (p = 0.741).</p><p>The rate of colonization of ETT tubes and incidence of nosocomial pneumonia were not affected by the CD14 -260 CT SNP. Nosocomial pneumonia rates were: (CC: 11%, CT: 5%, CC: 7%; p = 0.190). ETT colonization rates were: 66/132 (50%) for infants with the CC genotype, 58/128 (45%) CT genotype, and 21/33 (64%) TT genotype (p = 0.170). Organism-specific colonization rates were not different among genotype groups.</p><p>The incidences of necrotizing enterocolitis (NEC), CSF infection and infections in other sites were not different among genotype groups in either African-American or Caucasian infants.</p></sec><sec><title>Interactions between IL-10 -1082 G/A and IL-6 -174 G/C and incidence of sepsis</title><p>Since the IL-10 -1082A and the IL-6 -174C alleles were associated with an increased incidence of late BSI, the interaction of these two polymorphisms were examined on this and other outcomes. The effect of carriage of neither, either or both of the IL-10 -1082A and the IL-6 -174C alleles and IL-10 -1082:IL-6 -174 haplotypes was examined. Eight of the 9 possible haplotypes were present in our patient population. The IL-10 -1082 GG: IL-6 -174CC haplotype was not observed. Carriage of both the IL-10 -1082A and the IL-6 -174C alleles was associated with the greatest risk of late BSI in the overall population (p = 0.073) (Table <xref ref-type="table" rid="T5">5</xref>). The incidences of CONS BSI, non-CONS BSI, multiple BSI and sepsis-related mortality paralleled that of the overall late BSI rate but were not significant (Table <xref ref-type="table" rid="T5">5</xref>). This increase in risk was seen primarily in African-American infants (p = 0.031) [see <xref ref-type="supplementary-material" rid="S2">Additional File 2</xref> and <xref ref-type="supplementary-material" rid="S3">Additional File 3</xref>]. No significant trend between haplotype and risk for sepsis was observed, although the highest incidence of late BSI was in infants with the IL-10 -1082AA: IL-6 -174 CC haplotype [see <xref ref-type="supplementary-material" rid="S5">Additional File 5</xref>]. In multivariate analysis (logistic regression), carriage of the IL-10 -1082 A allele (p = 0.035) but not the IL-6 -174 C allele (p = 0.152) was a significant predictor of late BSI.</p></sec></sec><sec><title>Discussion</title><p>Genetically determined variation in the magnitude of inflammatory response may play a role in determining outcome from serious infections. IL-6 is a pro-inflammatory cytokine associated with the development of shock and mortality from sepsis [<xref ref-type="bibr" rid="B31">31</xref>]. Therefore genetic variants that influence production of this cytokine may have important implications in the development of and outcome from sepsis in the preterm neonate. The frequency of the IL-6 -174 C allele in both Caucasian and African-American VLBW infants was similar to control term infants and as described in other populations [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B32">32</xref>,<xref ref-type="bibr" rid="B33">33</xref>]. This suggests that this polymorphism has no effect on the incidence of prematurity or on the need for ventilation at birth. We found that the IL-6 C allele was associated with an increased risk for late BSI (all organisms) in African-American but not Caucasian infants. However, the incidence of fungal BSI was greatly increased in Caucasian infants with CC genotype compared to infants with GG or GC. This suggests that the IL-6 -174CC genotype may be a risk factor for fungal sepsis in specific ethnic groups. However, because there were relatively few Caucasian patients with the IL-6 -174 CC genotype (6), caution must be exercised in interpreting this finding. Further studies are needed to confirm this association.</p><p>Earlier studies had suggested that the IL-6 -174 G/C polymorphism was either not associated with increased risk of sepsis, or that VLBW infants who were homozygous GG were at increased risk of Gram positive sepsis [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B34">34</xref>]. However, the infants in these studies were all or predominately Caucasian. Ethnic differences and small numbers of Caucasian infants in our study may account for some of the discrepancies. In addition, our infants were significantly smaller, less mature and required mechanical ventilation, factors that significantly increase risk of infection and may overwhelm any effect of this polymorphism. There was no effect of the IL-6 -174 G/C polymorphism on mortality-related sepsis. The Ahrens study did not address the effect of polymorphisms on sepsis mortality. In adults, the GG genotype was associated with increased survival in sepsis, but had no effect on the incidence of sepsis (intensive care setting) [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>The role of the IL-6 -174 G/C polymorphism on IL-6 production is unclear. The IL-6 -174 C allele has been associated with decreased transcriptional activity in response to LPS and IL-1α [<xref ref-type="bibr" rid="B33">33</xref>]. The effects of this polymorphism, however, are more complex and may be stimulus dependent, cell line dependent and different <italic>in vivo </italic>from <italic>in vitro</italic>. <italic>In vitro </italic>IL-6 production in LPS-stimulated mononuclear cells is higher in the CC genotype in newborn infants [<xref ref-type="bibr" rid="B17">17</xref>]. Following coronary artery bypass surgery the C allele is associated with increased plasma IL-6, whereas following vaccination the G allele is associated with increased plasma IL-6 [<xref ref-type="bibr" rid="B35">35</xref>-<xref ref-type="bibr" rid="B37">37</xref>]. This complexity is further compounded by additional functional polymorphisms that are in linkage disequilibrium with the -174 SNP [<xref ref-type="bibr" rid="B35">35</xref>,<xref ref-type="bibr" rid="B36">36</xref>].</p><p>IL-10 downregulates inflammatory mediator production by stimulated immune and epithelial cells. Production of IL-10 can have a potentially beneficial effect by dampening excess inflammatory mediator production in sepsis. Excessive IL-10 production can, however, lead to the phenomenon of immunoparalysis by inhibiting the response of macrophages to pathogenic bacteria. The frequency of the IL-10 -1082 A allele in our ventilated VLBW infants is similar to that in control term infants and to that described in the literature [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B38">38</xref>-<xref ref-type="bibr" rid="B40">40</xref>]. This suggests little effect of this allele on either the incidence of preterm birth or the need for mechanical ventilation at birth. Our results are consistent with those of Kalish et al. [<xref ref-type="bibr" rid="B41">41</xref>], in which the IL-10 -1082 SNP was not associated with risk of preterm birth.</p><p>The IL-10 -1082A allele (lower IL-10 production) was associated with a two-fold increase in the incidence of late BSI in ventilated VLBW infants. Although not statistically significant, there was a trend towards increased non-CONS BSI (associated with higher mortality and morbidity) in infants with the A allele. This suggests that a more robust inflammatory response may protect the host from invasive disease. There was, however, no effect on sepsis-related mortality. The effect of the IL-10 -1082 G/A polymorphism on the incidence and outcome from infectious disease has been contradictory and may be organism-specific or vary according to ethnicity. There was no effect of the IL-10 -1082 G/A polymorphism on the incidence of invasive meningococcal disease, but disease severity and mortality were associated with the AA genotype [<xref ref-type="bibr" rid="B27">27</xref>]. In contrast, the G allele was associated with the development of septic shock in pneumococcal sepsis and severity of the systemic inflammatory response in community acquired pneumonia, whereas there was no association between genotype and risk of infection [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B42">42</xref>]. In the VLBW infant, an earlier smaller study showed no effect of the IL-10 -1082 G/A polymorphism on development of sepsis [<xref ref-type="bibr" rid="B34">34</xref>].</p><p>The role of polymorphic variation in the IL-10 gene and IL-10 production is unclear. Variation in IL-10 production in relation to the IL-10 -1082 SNP is affected by the nature of inducing stimulus and association of other polymorphisms [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B39">39</xref>,<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>-<xref ref-type="bibr" rid="B46">46</xref>]. The IL-10 -1082 SNP is in linkage disequilibrium with two other SNPs (-819 C/T and -592 C/A) and appears in three haplotypes [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B47">47</xref>,<xref ref-type="bibr" rid="B48">48</xref>]. The GCC haplotype (G at position -1082, C at position -819, C at -592) is associated with high IL-10 secretion, while the ACC and ATA haplotypes are associated with intermediate and low IL-10 secretion respectively [<xref ref-type="bibr" rid="B49">49</xref>]. The -2849 A/G SNP also affects IL-10 production functionally and should be studied [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B44">44</xref>].</p><p>Genetic variation in the ability to recognize and respond to invading organisms may also significantly influence the development of and outcome from infection. CD14 is an important component of innate immunity and plays a role in recognizing both Gram negative and positive organisms. We found that the CD14 -260 C/T SNP did not alter the overall risk for BSI in ventilated VLBW infants. However, the CC genotype was associated with an increased incidence of Gram negative BSI in African-American infants, suggesting a potential role for this polymorphism in determining risk for certain types of infection or in specific ethnic groups. In addition, multiple BSI episodes were more common in infants with the TT genotype owing to the strong effect of the TT genotype on the incidence of multiple BSI in African-American infants. Ahrens et al. demonstrated no association between this SNP and sepsis or sepsis mortality in VLBW Caucasian infants [<xref ref-type="bibr" rid="B18">18</xref>] The current study extends those findings by examining a higher risk group (ventilated) of infants with different ethnic backgrounds. In other studies involving primarily Caucasian adults, the CD14 -260 C/T polymorphism does not seem to have a major influence on the risk for or outcome from sepsis [<xref ref-type="bibr" rid="B50">50</xref>,<xref ref-type="bibr" rid="B51">51</xref>]. Only a single study suggested that the TT genotype (increased CD14) was associated with the development of septic shock and mortality [<xref ref-type="bibr" rid="B10">10</xref>]. Increased soluble CD14 concentrations were associated with mortality from Gram-negative septic shock in earlier studies [<xref ref-type="bibr" rid="B52">52</xref>].</p><p>In Caucasians the CD14 -260 T allele was significantly less frequent in VLBW infants than in term infants, suggesting a potential role for this polymorphism in premature birth. The role of CD14 polymorphisms in prematurity has not been studied but a larger prior study reported by Hartel et al. does not support a role for the CD14 -260CT SNP [<xref ref-type="bibr" rid="B53">53</xref>]. The frequency of the CD14 -260 T allele in our control population is similar to that reported in other control Caucasian populations (both adult and term infants), which varied between 0.352 and 0.548 [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B54">54</xref>]. The CD14 -260T allele was significantly more frequent in Caucasian control infants than in African-American controls, consistent with other reports [<xref ref-type="bibr" rid="B55">55</xref>]. The CD14 -260T allele frequency in African-American infants (both VLBW and term) was similar to that reported in the literature [<xref ref-type="bibr" rid="B56">56</xref>,<xref ref-type="bibr" rid="B57">57</xref>]. In contrast to that seen in Caucasians, no effect of the CD14 -260CT polymorphism on prematurity was seen in among African-American infants.</p><p>An individual's risk for developing sepsis (and its outcome) probably depends on interactions of several genetic factors. Polymorphisms of both pro-inflammatory (IL-6) and anti-inflammatory cytokines (IL-10) may interact either to increase or to decrease risk. In our study, co-carriage of both IL-6 -174 C and IL-10-1082 alleles was associated with a slightly increased risk of late BSI. Multivariate analysis, however, suggests that the IL-10 -1082 A allele is the dominant risk factor.</p><p>Other genetic differences may influence risk for and outcome from sepsis in VLBW infants. We recently reported that the TNF-α-308A allele did not affect the risk for sepsis but increased mortality in septic infants [<xref ref-type="bibr" rid="B58">58</xref>]. The 3020insC mutation of the NOD2 gene is also associated with increased risk of sepsis in VLBW infants [<xref ref-type="bibr" rid="B18">18</xref>]. Polymorphisms in other cytokines, their receptors and other bacterial pattern recognition molecules have been suggested to alter the course of sepsis and are logical candidates for further study in this population [<xref ref-type="bibr" rid="B59">59</xref>-<xref ref-type="bibr" rid="B62">62</xref>].</p><p>The observations of this retrospective case-controlled study are limited by selection bias. Our population of mechanically ventilated infants, most of whom were less than 1000 grams at birth, is at high risk for infection. As a result, the incidence of bacteremia/fungemia in our studied population (47%) is higher than generally reported for a VLBW population and also higher than for the population of VLBW infants in our NICU (approximately 30%) [<xref ref-type="bibr" rid="B3">3</xref>]. Mechanical ventilation and lower birth weight are known risk factors for late onset sepsis [<xref ref-type="bibr" rid="B63">63</xref>]. Because only infants who were mechanically ventilated were included, the true impact of the polymorphisms studied on the incidence of infectious complications of prematurity may be underestimated.</p></sec><sec><title>Abbreviations</title><p>IL-6 Interleukin-6</p><p>IL-10 Interleukin-10</p><p>VLBW Very Low Birth Weight</p><p>CONS Coagulase Negative Staphylococci</p><p>BSI Blood Stream Infection</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>RJB conceived and organized the study, prepared the manuscript and performed the statistical analyses. KY performed the genotyping, and assisted with manuscript preparation. JL oversaw collection tracheal aspirates, assisted with recruitment of subjects into the original studies, and assisted with editing of the manuscript. All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1741-7015/4/10/prepub"/></p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Genotype and allele frequencies of Caucasian and African-American term and preterm infants.</p></caption><media xlink:href="1741-7015-4-10-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p>Effects of the IL-10 -1082 GA polymorphisms and infectious complication in Caucasian and African-American infants.</p></caption><media xlink:href="1741-7015-4-10-S2.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional File 3</title><p>Effects of the IL-10 -1082 GA and IL-6 -174C allele carriage on nosocomial blood stream infections in Caucasian infants.</p></caption><media xlink:href="1741-7015-4-10-S3.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S4"><caption><title>Additional File 4</title><p>Effects of the IL-10 -1082 GA and IL-6 -174C allele carriage on nosocomial blood stream infections in African-American infants.</p></caption><media xlink:href="1741-7015-4-10-S4.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S5"><caption><title>Additional File 5</title><p>Effects of the IL-10 -1082 GA/IL-6 -174C haplotypes on nosocomial blood stream infections.</p></caption><media xlink:href="1741-7015-4-10-S5.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Inverse association of antioxidant and phytoestrogen nutrient intake with adult glioma in the San Francisco Bay Area: a case-control study
|
<sec><title>Background</title><p>Increasing evidence from epidemiologic studies suggest that oxidative stress may play a role in adult glioma. In addition to dietary antioxidants, antioxidant and weak estrogenic properties of dietary phytoestrogens may attenuate oxidative stress. Our hypothesis is that long-term consumption of dietary antioxidants and phytoestrogens such as genistein, daidzein, biochanin A, formononetin, matairesinol, secoisolariciresinol and coumestrol, may reduce the risk of adult glioma.</p></sec><sec sec-type="methods"><title>Methods</title><p>Using unconditional logistic regression models, we compared quartiles of consumption for several specific antioxidants and phytoestrogens among 802 adult glioma cases and 846 controls from two study series from the San Francisco Bay Area Adult Glioma Study, 1991 – 2000, controlling for vitamin supplement usage, age, socioeconomic status, gender, ethnicity and total daily calories. For cases, dietary information was either self-reported or reported by a proxy. For controls, dietary information was self-reported. Gender- and series- specific quartiles of average daily nutrient intake, estimated from food-frequency questionnaires, were computed from controls.</p></sec><sec><title>Results</title><p>Significant p-values (trend test) were evaluated using significance levels of either 0.05 or 0.003 (the Bonferroni corrected significance level equivalent to 0.05 adjusting for 16 comparisons). For all cases compared to controls, statistically significant inverse associations were observed for antioxidant index (p < 0.003), carotenoids (alpha- and beta-carotene combined, p < 0.05), daidzein (p = 0.003), matairesinol (p < 0.05), secoisolariciresinol (p < 0.003), and coumestrol (p < 0.003). For self-reported cases compared to controls, statistically significant inverse associations were observed for antioxidant index (p < 0.05) and daidzein (p < 0.05).</p></sec><sec><title>Conclusion</title><p>Our results support inverse associations of glioma with higher dietary antioxidant index and with higher intake of certain phytoestrogens, especially daidzein.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Tedeschi-Blok</surname><given-names>Nicole</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Lee</surname><given-names>Marion</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Sison</surname><given-names>Jennette D</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Miike</surname><given-names>Rei</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected] </email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Wrensch</surname><given-names>Margaret</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Cancer
|
<sec><title>Background</title><p>Epidemiologic evidence from studies of dietary antioxidant intake and adult glioma support a role for oxidative stress in gliomagenesis. Several studies have shown that antioxidant consumption from fruits and vegetables may be protective against adult glioma [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B4">4</xref>]. Most recently, a large case-control study in Nebraska showed that increased intake of carotenoids reduces the risk of glioma by 50 percent and suggests that antioxidant as well as phytochemical nutrient intake is likely to play a protective role in adult glioma [<xref ref-type="bibr" rid="B4">4</xref>]. Pro-inflammatory and pro-angiogenic mediators may also be implicated in glioma development and progression. Recent studies suggest that estrogen receptors may play a role in decreasing both pro-inflammatory and pro-angiogenic mediators [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Furthermore, estrogen receptor expression is negatively correlated with astrocytic tumor grade [<xref ref-type="bibr" rid="B7">7</xref>]. Phytoestrogens, plant-derived chemicals with both antioxidative and estrogenic properties, have structural similarities to 17-beta-estradiol and can be categorized to isoflavones (eg., genistein, daidzein, biochanin A, and formononetin), lignans (eg., matairesinol and secoisolariciresinol), and coumestans (eg., coumestrol). Some foods rich in isoflavones include soy, lentils, and beans. Foods rich in lignans include oilseeds, whole-grain cereals, legumes, and berries; while alfalfa and clover are foods rich in coumestans. In addition to their antioxidative and estrogenic properties, phytoestrogens are noted for their clinical potential in antiviral, antibacterial, antiproliferative, angiogenic activities [<xref ref-type="bibr" rid="B8">8</xref>].</p><p>If oxidative stress is involved in brain tumor development, then dietary intake of nutrients of fruits and vegetables with antioxidative and estrogenic properties may reduce the risk of adult glioma. We addressed this hypothesis by comparing nutrient intake between incident glioma cases and controls of the San Francisco Bay Area Adult Glioma Study, 1991–2000.</p></sec><sec sec-type="methods"><title>Methods</title><p>Human subject methods were approved by the University of California Committee on Human Research, approval number H6539-04956. Informed consent was documented through subject signature on the approved consent document.</p><sec><title>Case ascertainment</title><p>The San Francisco Bay Area Adult Glioma Study, 1991–2000, is described in detail elsewhere [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B12">12</xref>]. Briefly, cases were identified within a median of seven weeks of diagnosis using Northern California Cancer Center's rapid case ascertainment system. Eligible cases included those age 20 years and older, with histologically confirmed newly diagnosed glioma, International Classification of Disease for Oncology codes 9380 to 9481, who resided in 6-counties of the San Francisco Bay Area, which included Alameda, Contra Costa, Marin, San Mateo, San Francisco, or Santa Clara. Ascertainment periods included August 1991 to April 1994 (series 1) and from May 1997 to August 2000 (series II). Pathologic specimens were obtained and reviewed by a single neuropathologist in each series [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>], thereby minimizing misclassification of diagnosis.</p></sec><sec><title>Control ascertainment</title><p>In both series, controls were identified by random digit-dialing and frequency matched to cases on ten-year age groups, gender, and ethnicity (White, Black, Hispanic, Asian, or other). Initial sampling units included area code, 3-digit prefix, and 2 digits of cases' phone numbers. Eligible matches were found for each sampling unit through 2 digit suffixes generated from random number tables [<xref ref-type="bibr" rid="B2">2</xref>].</p></sec><sec><title>Dietary questionnaires and interviews</title><sec><title>Series I</title><p>Eligible cases or next-of-kin and controls were sent a letter and telephoned to schedule an in-person interview at subjects' home or other location of their choice [<xref ref-type="bibr" rid="B2">2</xref>]. Interviews were conducted with consenting subjects or their next-of-kin (proxies) when cases were not available due to death or disability. Subjects were sent a self-administered dietary questionnaire in advance, which was collected or completed at the in-person interview. All questions were asked for the year prior to glioma diagnosis for cases and for the previous year for controls.</p><p>The series I questionnaire consisted of 79-item food-frequency questions modified from the US National Cancer Institutes' (Block's) Health Habits and History Questionnaire [<xref ref-type="bibr" rid="B15">15</xref>] and the Los Angeles Glioma and Meningioma Study Questionnaire [<xref ref-type="bibr" rid="B16">16</xref>]. The modifications were originally made to emphasize antioxidant, nitrite, and nitrate – containing foods [<xref ref-type="bibr" rid="B2">2</xref>]. Food items were group into the following categories: fruits and juices; vegetables; breakfast foods; lunch foods; meat, fish, poultry and mixed dishes; breads, salty snacks and spreads; sweets; dairy products and beverages. Participants were asked to choose usual frequencies of consumption of each food item. The frequency choices included: 'never', 'less than 1 per month', '1 per month', '2–3 per month', '1 per week', '2 per week', '3–4 per week', '5–6 per week', '1 per day', and '2 or more per day'. Portion sizes were not asked, but were estimated using gender- and age- specific values from the Block and colleagues' National Cancer Institute's Health Habits and History Questionnaire [<xref ref-type="bibr" rid="B15">15</xref>]. Additional questions were asked about how often lemon juice was added to fish, fruit juices or tea was consumed with cured meat or bacon, and beer was consumed with meals. Frequency of vitamin supplement use (never or number of times per day, week, month, or year as well as dose) was also asked.</p></sec><sec><title>Series II</title><p>Similarly, eligible cases or next-of-kin and controls were sent a letter and telephoned to schedule an in-person interview at subjects' home or other location of their choice. Interviews were conducted with consenting subjects or their proxies when cases were not available due to death or disability. Rather than a self-administered diet questionnaire as in series I, interviewers asked the diet questions in series II. As in series I, all questions were asked for the year prior to glioma diagnosis for cases and for the previous year for controls in series II.</p><p>The series II diet questionnaire consisted of 96 food-frequency questions modified from the US National Cancer Institutes' (Block's) Health Habits and History Questionnaire [<xref ref-type="bibr" rid="B15">15</xref>] and included questions on portion sizes. Modifications included additional fruits, vegetables, and soy products (eg., avocado, banana, fresh peaches, mangoes or papayas, beets, cauliflower, celery, turnips, radishes, rhubarb, salsa, soy burgers, soy milk, and, tofu). The frequency choices included: 'never or less than 1 per month', '1 per month', '2–3 per month, '1 per week', '2 per week', '3–4 per week', '5–6 per week', '1 per day'. Portion sizes were asked according to the number of pieces, or either small, medium, large, or extra-large, for food frequencies of 1 per week or greater; using visual aids and 3-dimensional models. Questions on type (multi-vitamin or antioxidant), frequency and duration of vitamin supplements usage were also included.</p></sec><sec><title>Series I & II</title><p>For the purposes of assessing major differences of dietary consumption comparing eligible participants to eligible non-participants, an abbreviated questionnaire was administered to some unwilling to participate in the full interview. In a 5-minute phone interview, the abbreviated questionnaire asked basic demographic and dietary (including vitamin supplement use) questions.</p></sec></sec><sec><title>Dietary analyses</title><p>For both series, only one value was assigned for each food frequency choice by converting the unit of year, month or day to the number of times per day. For ranges of frequencies, the mid-value was assigned before conversion, e.g. 2.5 times per month was assigned if the '2–3 per month' category was given. Intake of each food item was converted to grams consumed per day by applying an appropriate algorithm for each series as described in the following: For series I, the frequency and an assigned gender- and age-specific medium portion size were multiplied against a nutrient database to produce grams consumed per day for each food item. The assigned gender- and age-specific medium portions were based on the Block and colleagues' National Cancer Institute's Health Habits and History Questionnaire Personal Computer System Packet [<xref ref-type="bibr" rid="B17">17</xref>]. For series II, the frequency and portion reported were multiplied against the same nutrient database to produce grams consumed per day for each food item. For both series, this nutrient database, which included a total of 27 nutrients per 100 grams for each food item, was based on the California Teacher's Study [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>], USDA nutrient database [<xref ref-type="bibr" rid="B20">20</xref>], and the Block and colleagues' National Cancer Institute's Health Habits and History Questionnaire Personal Computer System Packet [<xref ref-type="bibr" rid="B17">17</xref>]. The summation of nutrient amounts over all food items provided a subject's total daily intake for 27 nutrients. Of the 27 nutrients, there was a total of 15 antioxidants and phytoestrogens for our case-control comparisons.</p><p>Additionally for both series, a total daily antioxidant index intake was calculated by summing the product of grams consumed over all food items and units of antioxidant index per gram from an antioxidant index database for fruit, vegetable, juice and tea food items. We constructed the antioxidant index database based on data from Wang and colleagues [<xref ref-type="bibr" rid="B21">21</xref>] (for fruits and fruit juices), Cao and colleagues [<xref ref-type="bibr" rid="B22">22</xref>] (for fruits and vegetables) and Rice-Evans and Miller [<xref ref-type="bibr" rid="B23">23</xref>] (for fruits, vegetables, and teas). A total of 35 items of fruits, vegetables, juices and teas from either series were included in the antioxidant index database. All 35 food items were available for series II participants, while only 22 of the 35 food items were available for series I participants (see above for additional fruits and vegetables included in series II only). Therefore, because the antioxidant index was the sum of antioxidant values across food items reportedly consumed by each subject, series II participants had higher values for the total antioxidant index than series I participants. Antioxidant index values were units of micromoles Trolox equivalents per gram of food (fruits, vegetables, juices or teas). Trolox equivalents per gram of food were measured by the oxygen radical absorbance capacity (ORAC) assay [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. The ORAC assay measures the degree to which a sample inhibits oxidizing agent action using Trolox, a water-soluble analogue of vitamin E, as a control standard.</p><p>SAS version 8.02 (SAS Institute Inc, Cary, NC) was used for all statistical analyses. Demographic distributions of cases were compared to those of controls using t-tests or analysis of variance (ANOVA) for continuous variables and chi-square tests for categorical variables. Means and standard errors as well as geometric means with upper and lower 95% confidence limits for daily consumption of total antioxidant index as well as various individual antioxidants and phytoestrogens adjusted for total calories were generated with general linear models for all cases, self-reporting cases and controls using SAS procedure GLM. Logistic regression models estimated odds ratios for quartiles of consumption of antioxidant index or each nutrient adjusting for age (continuous), gender, ethnicity (White versus other), socioeconomic status (combined income and education categories), and total calories (continuous). Thus, to adjust for total energy intake, we employed the standard multivariate method [<xref ref-type="bibr" rid="B24">24</xref>]. In addition, vitamin supplement use and/or meat consumption were explored as covariates in the logistic regression models.</p><p>Case-control comparisons were analyzed separately by series since data on nutrient and antioxidant index intake was collected by different methods between study series. Because comparison of series-specific odds ratios and corresponding 95% confidence showed that odds ratios were largely in the same direction and their corresponding 95% confidence intervals overlapped between series, we also conducted combined series case-control comparisons with a study series indicator variable as a necessary covariate in logistic regression models. Gender- and series- specific quartile cutpoints were determined for the antioxidant index or nutrient intake levels based on food consumption among the controls. Test for trend were computed for case-control odds ratios of antioxidant index or nutrient quartiles. Results are described for p-values generated based on a 5% significance level and two-sided hypothesis tests. Finally, a Bonferroni correction for 16 simultaneous comparisons (an antioxidant index plus 15 nutrients) was applied to emphasize results based on a 0.003 (= 0.05/16) significance level.</p></sec></sec><sec><title>Results</title><p>The total number of subjects from both series that were identified as eligible to participate in the San Francisco Bay Area Adult Glioma Study, 1991–2000, includes 1,110 cases and 1,284 controls. Participation was greater among eligible cases (79%–80% for series I and II) than among eligible controls (67% for combined series data, 74% for series I, and 60% for series II) [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Dietary data for 98 percent (846/864) of controls, 94 (497/524) percent of self-reported cases and 87 (305/349) percent of proxy-reported cases were included in the dietary analyses, based on the number of items with missing food frequency and daily intake of total calories. Table <xref ref-type="table" rid="T1">1</xref> summarizes frequency distributions for vitamin supplement intake, demographics, and clinical factors for 802 cases (including responses from 305 proxies and 497 self-reporting cases) and 846 controls included in the dietary analyses. Because of better survival of younger cases, the mean age for self-reported cases was 50 years, significantly lower than that of the controls (p < 0.001, table <xref ref-type="table" rid="T1">1</xref>). For combined income and education, self-reported cases had greater socioeconomic status than controls (p < 0.001, table <xref ref-type="table" rid="T1">1</xref>); likely due to younger ages of self-reporting cases. Geometric means of total daily nutrients consumed adjusted for total calories among cases and controls are shown in table <xref ref-type="table" rid="T2">2</xref>. The observed average daily nutrients consumed were reasonable values for the U.S.</p><p>Information on vitamin type (multi-vitamin or antioxidant), asked for series II, was not considered during the analyses because the reported data did not appear to reflect a reasonable distribution of supplement types. In addition, differential absorption rates between supplements and foods prevent the combination of nutrients from supplement types with that from foods consumed. Therefore, all vitamin supplement analyses only considered intake (yes or no). Results of an abbreviated diet questionnaire for comparisons of participants and non-participants, showed that vitamin use among controls adjusted for age, gender, and ethnicity was similar for participants and non-participants (OR = 1.3, 95% CI = 0.91 to 1.82; data not shown). Interestingly, the percent of cases (among all and self-reported cases) that regularly consumed vitamin supplements was significantly lower than that of controls (p < 0.001, table <xref ref-type="table" rid="T1">1</xref>). Regardless of study series, a significantly lower percentage of cases (either all or self-reported) than controls reported use of vitamin supplements (p = 0.002 and p = 0.005 for series I and II, respectively; data not shown). For this reason, we also compared case-control nutrient consumption stratified by vitamin supplement use and observed a high degree of overlap among 95% confidence intervals of odds ratios across strata (data not shown).</p><p>Age- gender-, ethnicity-, socioeconomic status-, and supplement intake-adjusted odds ratios for quartiles of total calories consumed are shown in table <xref ref-type="table" rid="T3">3</xref>. Case-control odds ratios for quartiles of nutrient intakes adjusted for age, gender, ethnicity, socioeconomic status, total calories, and supplement intake are shown in tables <xref ref-type="table" rid="T4">4</xref> and <xref ref-type="table" rid="T5">5</xref>. Case-control distributions (= N) for quartiles of nutrient consumption are provided in supplemental table <xref ref-type="table" rid="T1">1</xref> [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. Examining these results for those that were statistically significant and consistent for all cases versus controls and self-reported cases versus controls, we found that odds ratios for intakes of antioxidant index, carotenoids (alpha- and beta-carotene combined), and the phytoestrogens, daidzein, and matairesinol were reduced (tables <xref ref-type="table" rid="T4">4</xref> and <xref ref-type="table" rid="T5">5</xref>). Although not consistent with self-reported cases compared to controls, other notable findings regarding nutrient intake include statistically significant differences among all cases compared to controls, which include consumption of vitamin C, and the phytoestrogens, genistein, formononetin, secoisolariciresinol and coumestrol (table <xref ref-type="table" rid="T4">4</xref>). After applying a Bonferroni correction, statistically significant p-values for trend tests were observed for all cases compared to controls among combined series data for antioxidant index (p = 0.002), daidzein (p = 0.003), secoisolariciresinol (p = 0.001) and coumestrol (p = 0.001). Additionally, after adjustment for meat consumption, the only noteworthy changes were for vitamin C consumption. Consumption of vitamin C for cases was no longer significantly different than that of controls after adjusting for meat consumption (OR = 0.77, 95% CI = 0.58 to 1.02, for second quartile compared to lowest; OR = 0.78, 95% CI = 0.59 to 1.05, for third quartile compared to lowest; OR = 0.76, 95% CI = 0.56 to 1.04, for highest quartile compared to lowest; combined series I and II for all cases compared to controls). The effect of adjustment for meat consumption was negligible for all other dietary variables.</p></sec><sec><title>Discussion</title><p>We previously discussed in detail major potential sources of bias in these case-control dietary data [<xref ref-type="bibr" rid="B2">2</xref>]. Briefly, bias could result from differential data quality between proxy and self-reported dietary histories with more accurate data from self-reports. Differential participation of controls who are more health conscious and therefore might eat foods that are considered to be healthier such as fruits and vegetables could also bias results. In addition, the dietary data collected may not accurately represent actual dietary data during the etiologically relevant period of brain tumor development; however, this etiologically relevant time period is unknown. For this reason, the period of dietary recall was limited to recent dietary habits to improve recall accuracy. Furthermore, non-differential dietary misclassification may affect series I data as a result of using computed measures of dietary nutrients because these computations assume consumption of average portion sizes for age group and gender specific subjects, which may decrease detectable differences in nutrient consumption. This issue is likely to be overcome in series II data, which ascertained portion sizes for foods consumed once a week or more. Also, since actual nutrient values of foods depend on soil quality and other factors, the nutrient database values used in these analyses may either over or underestimate the actual nutrients of the foods consumed. Since, to our knowledge, this is the first study to investigate an association between antioxidant index of foods consumed with a disease, we used estimates for antioxidant values based solely on literature review. Our findings of an inverse association between antioxidant index and glioma may encourage other researchers to further investigate these antioxidant indices and their physiologic relevance. In addition, since the series I food-frequency questionnaire lacked some foods with high quantities of phytoestrogens (such as tofu and soy, which are both high in genistein and daidzein), the estimated daily intake of such phytoestrogens are likely lower than actual values. While the series II food-frequency questionnaire included these additional food items that are likely to more accurately assess total intakes of phytoestrogens, evidence of an association is strong in both series data for nutrients driven largely by soy products (especially daidzein). Therefore, we are encouraged that we likely have an overall improved assessment of phytoestrogen intake with consideration of series II data. It should also be noted that individual variability is likely to contribute to differences in metabolism and bioavailability of dietary phytoestrogens [<xref ref-type="bibr" rid="B25">25</xref>]. Regarding the validity of the estimates of nutrients consumed for the food-frequency information assessed in this study, both antioxidant [<xref ref-type="bibr" rid="B26">26</xref>] and phytoestrogen (Horn-Ross, P. et al., <italic>Am J Epidemiol</italic>, in review, 2005) nutrients estimates closely approximate reference values in previous studies. Finally, although we report findings that were statistically significant at the 0.05 level, because of multiple comparisons, the actual p-values are likely to be higher than those calculated. For this reason, we also emphasize statistically significant results at the 0.003 level after applying a Bonferroni correction for 16 simultaneous comparisons.</p><p>Due to methodological differences in data collection between series, we presented the data separately by series. Because the direction and 95% confidence limits of odds ratios were similar across series, we also performed combined series analyses since increasing the sample size increases the power to detect statistically significant findings. It should be noted that our combined series analysis is similar to a meta-analysis, a common epidemiologic practice, with the advantage of having the raw data to directly estimate associations from the combined data. Combined series odds ratios were generated with addition of a series indicator covariate in the logistic regression models. Additionally, because of the many limitations with retrospective dietary assessment, we focus on results that were reasonably consistent in magnitude and statistical significance between all cases versus controls and self-reporting cases versus controls. For those nutrients with such consistent findings, the odds ratios for highest to lowest quartiles of intake tended to be further from the null among subjects without reported supplement use than those with such use. These findings could indicate that distributions of dietary choices of people taking supplements are more similar regardless of case-status.</p><p>Despite the many caveats inherent in retrospective dietary assessment, our findings, in light of previous epidemiologic and mechanistic studies, support the growing evidence of a role for oxidative stress in adult glioma. Vitamin C is a water soluble, well known scavenger of hydroxyl radicals that inhibits oxidative DNA lesions such as 8-hydroxydeoxyguanosine [<xref ref-type="bibr" rid="B27">27</xref>]. We observed an inverse association with adult glioma for vitamin C consumption, which is consistent with our previous results [<xref ref-type="bibr" rid="B2">2</xref>]. Regarding consumption of vitamin E, a lipid soluble free radical scavenger, we did not observe any statistically significantly differences between cases and controls, although total dietary antioxidant index was statistically significantly inversely associated with glioma in this study. However, since results from other studies regarding the roles of vitamin's C and E with brain tumor risk are inconsistent [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B30">30</xref>], the roles of antioxidant vitamins C & E may remain unclear. If the findings of inverse associations of total antioxidant intake from all fruits, vegetables, juices and teas with glioma are replicated it may suggest a protective role of these compounds for adult glioma.</p><p>For carotenoid consumption, more consistent results have been observed among studies of brain tumor risk. We observed an inverse association of carotenoid consumption with glioma risk, which is consistent with other studies [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B28">28</xref>] including a recent study in Nebraska, which showed a two-fold reduction in risk of glioma with increased carotenoid intake [<xref ref-type="bibr" rid="B4">4</xref>]. Carotenoids are lipophilic molecules with antioxidant properties implicated in scavenging peroxynitrite, modulating DNA repair [<xref ref-type="bibr" rid="B31">31</xref>], and possibly anti-inflammatory mechanisms [<xref ref-type="bibr" rid="B32">32</xref>-<xref ref-type="bibr" rid="B35">35</xref>].</p><p>We observed statistically significant inverse associations for consumption of several phytoestrogens suggesting protective effects. The most consistently significant inverse association was observed for daidzein. Our results also suggest protective effects against gliomas for formononetin, matairesinol, secoisolariciresinol, and coumestrol. Vaya and colleagues [<xref ref-type="bibr" rid="B36">36</xref>] showed that flavonoids, which include isoflavones and coumestans, inhibit low density lipoprotein oxidation. More specifically, mechanistic studies suggest that flavonoids may down-regulate both the expression of iNOS and cyclooxygenase-2 (COX-2), both of which have pro-inflammatory roles [<xref ref-type="bibr" rid="B37">37</xref>-<xref ref-type="bibr" rid="B39">39</xref>]. Epidemiologic findings also support a role for inflammatory and immune mediators in brain tumors. An inverse association of brain tumors with asthma and autoimmune disease has been observed [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B40">40</xref>], which might partly be related to an inverse association of non-steroidal anti-inflammatory drug (NSAID) use or other anti-inflammatory drugs [<xref ref-type="bibr" rid="B41">41</xref>].</p><p>Furthermore, the influence of phytoestrogens on pro-inflammatory mediators may be a result of estrogen receptor-dependent activities. Recent evidence from several studies supports that estrogen receptors, and especially estrogen receptor-beta, down-regulates iNOS and COX-2 gene expression [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Interestingly, expression of estrogen receptor-beta has been characterized in astrocytic tumors, showing that as tumor grade increases, estrogen receptor-beta expression decreases [<xref ref-type="bibr" rid="B7">7</xref>]. Furthermore, Hara and Okayasu recently showed a strong correlation of vascular endothelial growth factor (VEGF) and COX-2 expression, likely due to increased iNOS expression, with degree of angiogenesis in astrocytomas [<xref ref-type="bibr" rid="B42">42</xref>]. Could phytoestrogen intake play a role in preventing the progression of astrocytic tumors from low to high grade through estrogen receptor-dependent down-regulation of pro-inflammatory and pro-angiogenic mediators? Further studies of phytoestrogen intake and astrocytic tumors are necessary to clarify the suggestive association.</p></sec><sec><title>Conclusion</title><p>The observed inverse associations suggest a protective role against gliomagenesis for consumption of foods rich in antioxidants and certain phytoestrogens, especially daidzein. In light of the growing evidence supporting a role for oxidative stress in gliomagenesis, future studies in brain tumor research should focus on reactive nitrogen and oxygen species to further clarify their roles as well as identifying targets of treatment and prevention of glioma.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>We, the authors, are noted for the following significant contributions toward this study: ML for the overall idea; MW and ML for the study design; MW and RM for the collection of data; JDS and RM for analysis of data; NTB for writing the manuscript; and both ML and MW for advice and consultation.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2407/6/148/prepub"/></p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Supplmental_Table1, "Case-control distributions for quartiles of nutrient consumption; San Francisco Bay Area Adult Glioma Study, 1991–2000", 133 KB</p></caption><media xlink:href="1471-2407-6-148-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
The relation between deoxycytidine kinase activity and the radiosensitising effect of gemcitabine in eight different human tumour cell lines
|
<sec><title>Background</title><p>Gemcitabine (dFdC) is an active antitumour agent with radiosensitising properties, shown both in preclinical and clinical studies. In the present study, the relation between deoxycytidine kinase (dCK) activity and the radiosensitising effect of gemcitabine was investigated in eight different human tumour cell lines.</p></sec><sec sec-type="methods"><title>Methods</title><p>Tumour cells were treated with dFdC (0–100 nM) for 24 h prior to radiotherapy (RT) (γ-Co<sup>60</sup>, 0–6 Gy, room temperature). Cell survival was determined 7, 8, or 9 days after RT by the sulforhodamine B test. dCK activity of the cells was determined by an enzyme activity assay.</p></sec><sec><title>Results</title><p>A clear concentration-dependent radiosensitising effect of dFdC was observed in all cell lines. The degree of radiosensitisation was also cell line dependent and seemed to correlate with the sensitivity of the cell line to the cytotoxic effect of dFdC. The dCK activity of our cell lines varied considerably and differed up to three fold from 5 to 15 pmol/h/mg protein between the tested cell lines. In this range dCK activity was only weakly related to radiosensitisation (correlation coefficient 0.62, p = 0.11).</p></sec><sec><title>Conclusion</title><p>Gemcitabine needs to be metabolised to the active nucleotide in order to radiosensitise the cells. Since dFdCTP accumulation and incorporation into DNA are concentration dependent, the degree of radiosensitisation seems to be related to the extent of dFdCTP incorporated into DNA required to inhibit DNA repair. The activity of dCK does not seem to be the most important factor, but is clearly a major factor. Other partners of the intracellular metabolism of gemcitabine in relation to the cell cycle effects and DNA repair could be more responsible for the radiosensitising effect than dCK activity.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Pauwels</surname><given-names>Bea</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Korst</surname><given-names>Annelies EC</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Pattyn</surname><given-names>Greet GO</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Lambrechts</surname><given-names>Hilde AJ</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Kamphuis</surname><given-names>Juliette AE</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>De Pooter</surname><given-names>Christel MJ</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Peters</surname><given-names>Godefridus J</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Lardon</surname><given-names>Filip</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Vermorken</surname><given-names>Jan B</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Cancer
|
<sec><title>Background</title><p>Gemcitabine (2',2'-difluorodeoxycytidine, dFdC) is a synthetic pyrimidine nucleoside analogue that has a structure very similar to that of deoxycytidine and cytosine arabinoside (Ara-C) [<xref ref-type="bibr" rid="B1">1</xref>]. In clinical use, gemcitabine is active against a variety of solid tumours such as cancers of the pancreas, lung, head and neck, bladder, breast, and ovary. It is activated intracellularly by deoxycytidine kinase (dCK), which adds a phosphate group to the 5' position of the deoxyribose group. The diphosphate (dFdCDP) and triphosphate (dFdCTP) forms of the drug play an important role in the cytotoxic effect: dFdCDP is an inhibitor of ribonucleotide reductase, while dFdCTP is incorporated into DNA, both leading to the inhibition of DNA synthesis. At the same time, gemcitabine has several self-potentiation mechanisms that serve to increase intracellular levels of the active metabolite and increase cytotoxicity [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]</p><p>In addition to its cytotoxic effect, gemcitabine is a potent radiosensitiser when used in rodent and human tumour cells, including pancreatic tumours, non-small cell lung cancer, head and neck cancer, colorectal, breast, ovarian and bladder cancer [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B14">14</xref>]. Gemcitabine can lead to a radiosensitising effect in both monolayer and spheroid glioblastoma cultures [<xref ref-type="bibr" rid="B15">15</xref>]. Recent <italic>in vivo </italic>studies have confirmed these observations and have shown significant tumour growth delay with the combination of gemcitabine and ionising radiation in animal models [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. These results have prompted a variety of clinical trials using gemcitabine as a radiosensitiser [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B31">31</xref>]. The exact mechanism of radiosensitisation and most important factors are still not known yet. Although the effects of gemcitabine on cell cycle redistribution and deoxynucleotide triphosphate (dNTP) pools may contribute to, or even be necessary for, gemcitabine-mediated radiosensitisation [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B13">13</xref>], they do not ultimately determine whether or not gemcitabine treatment is going to result in enhanced radiosensitivity. For example, the role of dATP depletion could not be confirmed in a study using cells with different repair defects [<xref ref-type="bibr" rid="B32">32</xref>]. Wachters et al [<xref ref-type="bibr" rid="B33">33</xref>] have shown that gemcitabine can sensitise cells to radiation by specific interference with the homologous repair (HR) pathway. Others have suggested that gemcitabine in combination with radiotherapy is compromised by mismatch repair (MMR), because recovery from gemcitabine treatment is facilitated by, although not dependent on MMR proficiency [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B35">35</xref>]</p><p>As a prodrug, gemcitabine requires intracellular phosphorylation to its active triphosphate form by dCK and to a lesser extent by thymidine kinase (TK2) [<xref ref-type="bibr" rid="B36">36</xref>] to exhibit biological activity. Acquired resistance to gemcitabine has been associated with deficiency of dCK [<xref ref-type="bibr" rid="B37">37</xref>-<xref ref-type="bibr" rid="B40">40</xref>] It has been shown that the expression of dCK at mRNA, protein and activity level in cancer cell lines of different origin were closely related and a correlation between the sensitivity to gemcitabine and the activity of dCK was observed [<xref ref-type="bibr" rid="B41">41</xref>]. It has also been shown that sensitivity to nucleoside analogues could be restored by transfection of a wild type dCK cDNA [<xref ref-type="bibr" rid="B42">42</xref>-<xref ref-type="bibr" rid="B44">44</xref>]. The critical role of dCK in gemcitabine's radiosensitisation is not known. Previously a significant relationship between dCK activity and sensitivity of various xenografts to gemcitabine was described [<xref ref-type="bibr" rid="B41">41</xref>] while accumulation of the triphosphate, dFdCTP, was related to sensitivity to gemcitabine also [<xref ref-type="bibr" rid="B45">45</xref>]. Earlier data with a gemcitabine resistant cell line, i.e. lacking dCK, showed that these cells required a high concentration (50–100 μM) of gemcitabine in order to get radiosensitisation [<xref ref-type="bibr" rid="B46">46</xref>]. However, because radiosensitisation <italic>in vitro </italic>increases with the drug concentration [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B47">47</xref>])., it could be that the radiosensitising effect of gemcitabine also depends on the rate of drug phosphorylation which in turn depends on dCK activity [<xref ref-type="bibr" rid="B8">8</xref>] and drug concentration [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B37">37</xref>,<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B48">48</xref>]). Therefore, it has be hypothesized that dCK activity could be an important factor for the radiosensitising effect of gemcitabine.</p><p>The purpose of the present study was to further substantiate the role of dCK in gemcitabine's radiosensitisation. Most studies investigating the radiosensitising effect of gemcitabine are limited in cell types and only a few concentrations of gemcitabine are used. In this study, the radiosensitising effect of gemcitabine was investigated in eight human tumour cell lines, originating from different tissues, using a range of gemcitabine concentrations. Within that context, the role of dCK was further explored.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Chemicals and reagents</title><p>Dulbecco's Modified Eagle's Medium (DMEM), RPMI, Medium 199, fetal calf serum and the medium supplements L-glutamine and sodium pyruvate were all purchased from Invitrogen (Merelbeke, Belgium). Sulforhodamine B was obtained from ICN (Asse, Belgium). Gemcitabine was purchased from Eli Lilly (Indianapolis, USA).</p></sec><sec><title>Cell lines</title><p>The cell lines used in this study were human tumour cells differing in p53 status originating from different tissues: ECV304 (mt-p53) a human epidermoid bladder cancer cell line, H292 (wt-p53) a human mucoepidermoid lung cancer cell line, A549 (wt-p53) a human squamous lung cancer cell line, MCF-7 (wt-p53) a mammary carcinoma cell line, HT-29 (mt-p53) a colon adenocarcinoma cell line, Panc-1 (mt-p53) a human pancreatic epitheloid cell line, CAL-27 (mt-p53) a human squamous cell carcinoma cell line of the tongue and FaDu (mt-p53) a human squamous cell carcinoma cell line of the pharynx. H292 and A549 were cultured in RPMI-1640 medium, supplemented with glutamine, sodium pyruvate and 10% fetal calf serum. ECV304 was cultured in Medium-199 supplemented with 10% fetal calf serum. MCF-7, HT-29, CAL-27 and FaDu were cultured in DMEM medium, supplemented with glutamine and 10% fetal calf serum. Cultures were maintained in exponential growth in a humidified atmosphere at 37°C under 5% CO<sub>2</sub>/95% air.</p></sec><sec><title>Cell survival after treatment with gemcitabine and radiation</title><p>The sulforhodamine B (SRB) test is a suitable test system for in vitro radiosensitivity testing, which in the presently used cell lines has shown to be comparable in outcome with the clonogenic assay, when cells are allowed to undergo at least 6 doubling times after radiation treatment [<xref ref-type="bibr" rid="B49">49</xref>]. Therefore, in our experiments, ECV304, H292, A549, MCF-7, HT-29 and FaDu cells were incubated for 7 days, CAL-27 cells for 8 days and Panc-1 cells for 9 days after radiation treatment, before determination of the survival by the SRB assay. Optimal seeding densities were determined for each cell line to assure exponential growth during the assay.</p><p>Cells were harvested from exponential phase cultures by trypsinisation, counted and plated in 48-well plates. Following plating and a 24 h recovery period, cells were treated with gemcitabine (0–100 nM) dissolved in phosphate buffered saline (PBS) during 24 h immediately followed by radiation. PBS was added to control cells. Each concentration was tested six times within the same experiment. After irradiation at room temperature over a dose range of 0–6 Gy, using a <sup>60</sup>Co source (Alcyon, St Augustinus hospital, Antwerp), cells were washed with drug free medium. After 7, 8 or 9 days, the survival was determined by the SRB assay. For determination of cell survival after treatment with gemcitabine alone, the SRB assay was performed 4 days after the start of treatment.</p><p>The SRB assay was performed according to the method of Skehan and colleagues and Papazisis and colleagues, with minor modifications [<xref ref-type="bibr" rid="B50">50</xref>,<xref ref-type="bibr" rid="B51">51</xref>] Culture medium was aspirated prior to fixation of the cells by addition of 200 μl 10% cold trichloroacetic acid. After 1 h incubation at 4°C, cells were washed 5 times with deionised water. Then the cells were stained with 200 μl 0.1% SRB dissolved in 1% acetic acid for at least 15 minutes and subsequently washed 4 times with 1% acetic acid to remove unbound stain. The plates were left to dry at room temperature and bound protein stain was solubilised with 200 μl 10 mM unbuffered TRIS base (tris(hydroxymethyl)aminomethane) and transferred to 96 wells plates for reading the optical density at 540 nm (Biorad 550 microplate reader, Nazareth, Belgium).</p></sec><sec><title>dCK enzyme activity assays</title><p>To determine a possible correlation between dCK activity and the radiosensitising effect of gemcitabine, cells were harvested as described previously [<xref ref-type="bibr" rid="B41">41</xref>], and pellets were stored at -80°C until analysis. 25.10<sup>6 </sup>cells were used in order to be able to measure enzyme activity in a linear range for time and protein. dCK was determined essentially as described previously [<xref ref-type="bibr" rid="B41">41</xref>]. To measure dCK selectively and bypass TK2 mediated phosphorylation of deoxycytidine, we used radiolabeled chlorodeoxyadenosine ([<sup>3</sup>H]CdA) as the substrate [<xref ref-type="bibr" rid="B52">52</xref>], which is not activated by TK2. Enzyme activities were expressed as nmol product per h per mg protein (nmol/h/mg protein).</p></sec><sec><title>Statistical methods</title><p>The survivals were calculated by: mean optical density (OD) of treated cells/mean OD of control cells × 100%. The radiation survival curves were fitted according to the linear-quadratic model: surviving fraction = exp(- αD – βD<sup>2</sup>), using Winnonlin (Pharsight, USA).</p><p>The following parameters were calculated: ID50, the radiation dose causing 50% growth inhibition and IC50, the concentration gemcitabine causing 50% growth inhibition. The radiosensitising effect was represented by the dose enhancement factor (DEF): ID50(-dFdC)/ID50(+dFdC).</p><p>Unless otherwise indicated, all data are presented as the mean ± standard deviation. All experiments were performed at least three times. A two-sample t-test and one-way ANOVA analysis was used to determine significance between ID50 values and DEFs.</p><p>Radiosensitisation can be defined as a synergistic interaction between gemcitabine and radiation. For the determination of synergism, the combination index (CI) was calculated by the Chou-Talalay equation [<xref ref-type="bibr" rid="B47">47</xref>,<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B54">54</xref>], using CalcuSyn (Biosoft, USA, UK). The general equation for the classic isobologram is given by:</p><p><inline-graphic xlink:href="1471-2407-6-142-i1.gif"/></p><p>where (D<sub>x</sub>)<sub>1 </sub>and (D<sub>x</sub>)<sub>2 </sub>in the denominators are the doses (or concentrations) for D<sub>1 </sub>(gemcitabine) and D<sub>2 </sub>(radiation) alone that give x% inhibition, whereas (D)<sub>1 </sub>and (D)<sub>2 </sub>in the nominators are the doses of gemcitabine and radiation in combination that also inhibit x% (i.e., isoeffect).</p><p>The (D<sub>x</sub>)<sub>1 </sub>or (D<sub>x</sub>)<sub>2 </sub>(for gemcitabine and radiation) can be readily calculated from the median-effect equation of Chou:</p><p><inline-graphic xlink:href="1471-2407-6-142-i2.gif"/></p><p>Where f<sub>a </sub>is the fraction affected and D<sub>m </sub>is the median-effect dose (ID50 or IC50) that is obtained from the anti-log of the X-intercept of the median effect plot, X-log (D) versus Y = log [f<sub>a</sub>/(1 - f<sub>a</sub>] or D<sub>m </sub>= 10<sup>-(Y-intercept)/m</sup>, and m is the slope of the median effect plot.</p><p>For conservative mutually nonexclusive isobolograms of two agents, a third term,</p><p><inline-graphic xlink:href="1471-2407-6-142-i3.gif"/></p><p>is added. The third term is usually omitted, when the mutually exclusive (α = 0) assumption or classic isobologram is used [<xref ref-type="bibr" rid="B53">53</xref>,<xref ref-type="bibr" rid="B55">55</xref>] Data in table <xref ref-type="table" rid="T2">2</xref> are based on the assumption that α = 0.</p><p>A CI value between 0.9 and 1.1 indicates only additivity. Moderate synergism is depicted by CI values between 0.7 and 0.9, synergism by CI values below 0.7.</p></sec></sec><sec><title>Results</title><sec><title>Radiosensitisation by gemcitabine</title><p>A clear concentration-dependent radiosensitising effect of gemcitabine was observed in ECV304, FaDu, H292, A549, CAL-27, Panc-1, MCF-7 and HT-29 cells (Figure <xref ref-type="fig" rid="F1">1</xref> and <xref ref-type="fig" rid="F2">2</xref>). ID50 values and DEFs for the different gemcitabine concentrations are summarised in Table <xref ref-type="table" rid="T1">1</xref>.</p><p>The degree of radiosensitisation seemed to be cell line dependent. For ECV304 and MCF-7 cells, the DEF after treatment with 1 nM gemcitabine was 1.37, and 1.24, respectively, while this concentration had no radiosensitising effect in FaDu, H292, A549, CAL-27 and HT-29 cells with DEF values around 1. In Panc-1 cells, much higher concentrations (15–100 nM) of gemcitabine were required to obtain a radiosensitising effect than in the other cell lines (2–8 nM). Panc-1 cells were less sensitive for the cytotoxic effect of gemcitabine alone as depicted by the IC50 values in Table <xref ref-type="table" rid="T1">1</xref>. The radiosensitising effect seemed to correlate with the sensitivity of the cell line to the cytotoxicity effects of gemcitabine (correlation coefficient for mean IC50 and mean DEF: -0.82, p = 0.013).</p><p>The CI analysis showed that after treatment during 24 h with gemcitabine immediately before radiation treatment, there is synergism in ECV304 with gemcitabine concentrations of 2 nM or higher (CI ≤ 0.65), with concentrations of 4 nM or higher in A549 (CI ≤ 0.70), MCF-7 (CI ≤ 0.62) and FaDu (CI ≤ 0.50) and with concentrations of 6 nM and higher in HT-29 cells (CI ≤ 0.39). Only moderate synergism was observed with 6 nM or higher in H292 (CI ≤ 0.88) and CAL-27 (CI ≤ 0.72). In Panc-1 cells, concentrations of 7 nM and higher resulted in a moderate to synergistic interaction (CI ≤ 0.79) (Figure <xref ref-type="fig" rid="F3">3</xref>).</p></sec><sec><title>dCK</title><p>Since our cell lines had a normal sensitivity range (nM range) and had a DEF comparable to high and low values in literature, we measured dCK activity in order to determine whether there would be relation between DEF and dCK activity. Table <xref ref-type="table" rid="T2">2</xref> represents the dCK activity per cell line. The dCK activity varied considerably from cell line to cell line and differed up to three fold from 5 to 15 pmol/h/mg protein between the tested cell lines. In this range dCK activity was weakly related to DEF (correlation coefficient = 0.62, p = 0.11) (Figure <xref ref-type="fig" rid="F4">4</xref>).</p></sec></sec><sec><title>Discussion</title><p>This is the first study showing a clear concentration-dependent radiosensitising effect of gemcitabine in a large number of tumour cell lines, originating from different tissues. In addition, we found that that there was a weak positive correlation between dCK activity of these cells and the DEF. Probably, more factors, including other partners of the intracellular metabolism, cell cycle effects and DNA repair play a role. This offers an explanation for the variable results in the clinic.</p><p>In particular, we investigated the radiosensitising effect in 8 different cell lines, using various concentrations of gemcitabine. Our data demonstrate that gemcitabine increases the radiosensitivity of ECV304, H292, A549, MCF-7, HT-29, CAL-27, Panc-1 and FaDu cells in vitro when the cells are treated for 24 h immediately before radiation. The enhancement is concentration dependent, with an increasing DEF with higher concentrations of gemcitabine.</p><p>Our DEFs are comparable to literature data from other cell lines [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B46">46</xref>]). In most of these studies, the concentration of gemcitabine needed to achieve such a radiosensitising effect, was higher than in our study. This seems to be dependent on the sensitivity to the single agent gemcitabine of the cell lines studied since in gemcitabine-resistant cell lines even μM concentrations of gemcitabine were required [<xref ref-type="bibr" rid="B46">46</xref>]. In addition, this could also be due to differences in DNA repair between the cells. In our experiments, the radiosensitising effect was cell line dependent; for example, higher gemcitabine concentrations were needed to induce radiosensitisation in Panc-1 cells (Table <xref ref-type="table" rid="T1">1</xref>).</p><p>As mentioned gemcitabine is a prodrug that requires successive intracellular phosphorylations into its active trisphosphate form [<xref ref-type="bibr" rid="B56">56</xref>]. The enzyme dCK is required for the first phosphorylation step into dFdCMP, while non-specific kinases are responsible for the further phosphorylation steps. As it has been reported that the level of enzymatic activity of dCK could have a profound influence on cellular resistance to gemcitabine cytotoxicity, the present investigations were also designed to address the relationship between gemcitabine's radiosensitisation and the activity of dCK in various human tumour cell lines. The range of dCK activity found in our cell lines, is in agreement with the range found in other solid tumour cell lines, but lower than in leukemic cell lines [<xref ref-type="bibr" rid="B57">57</xref>]. For those cell lines investigated the extent of overall gemcitabine phosphorylation to dFdCTP is related to the drug concentration and the activity of dCK [<xref ref-type="bibr" rid="B37">37</xref>,<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B57">57</xref>]. However other enzymes also contribute to the overall accumulation and particularly the retention of dFdCTP, such as deoxycytidine deaminase, pyrimidine 5'nucleotidase (5'NT), aspecific nucleotidases and phosphatases, and incorporation into DNA and RNA.</p><p>Our results in 8 different human solid tumour cell lines with a varying radiosensitising effect of gemcitabine did not further validate the correlation between dCK activity of the cells and the radiosensitising effect of gemcitabine. Gregoire et al [<xref ref-type="bibr" rid="B58">58</xref>] reported a correlation between dCK activity and gemcitabine's radiosensitisation. This correlation held both for dCK mRNA expression and the protein versus radiosensitisation [<xref ref-type="bibr" rid="B58">58</xref>]. Contrary to this, we could only show a weak relation between the DEF and dCK activity of eight different tumour cells. Possibly the range of dCK activity in our panel was not large enough. In the xenografts investigated for a relation between dCK activity and gemcitabine's antitumor activity was larger [<xref ref-type="bibr" rid="B41">41</xref>].</p><p>Several hypotheses have been proposed to explain the radiosensitisation potential of gemcitabine [<xref ref-type="bibr" rid="B59">59</xref>]. Of these, most likely, cell cycle synchronization [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B60">60</xref>,<xref ref-type="bibr" rid="B61">61</xref>] and DNA repair [<xref ref-type="bibr" rid="B32">32</xref>] play an important role in the radiosensitisation. In addition to the intracellular metabolism of gemcitabine, for which dCK is a key factor, intrinsic variation in the cellular response to the analogue after ionising radiation is likely to also play a role in gemcitabine's radiosensitisation. The relative importance of dCK activity among these various parameters is, however, not clear. Possibly the overall rate of initial gemcitabine phosphorylation is also determined by the equilibrium between the activities of dCK and the degrading enzyme, 5'NT, since the ratio dCK/5'NT showed a better correlation with gemcitabine sensitivity than dCK activity alone [<xref ref-type="bibr" rid="B62">62</xref>]. Moreover, the final mode of action of gemcitabine of producing a "masked chain termination" after its incorporation into DNA, favours that DNA repair plays an important role with respect to sensitising the tumour cells to radiation.</p></sec><sec><title>Conclusion</title><p>Our study suggest that other partners of the intracellular metabolism of gemcitabine in relation to cell cycle effects and DNA repair could be more responsible for the radiosensitising effect of gemcitabine than dCK activity of the cell.</p></sec><sec><title>Competing interests</title><p>We have received financial support for doing studies with gemcitabine. Eli Lilly Company is not financing this manuscript.</p></sec><sec><title>Authors' contributions</title><p>BP participated in the design of the study, performed cell survival experiments, statistic analysis and drafted the manuscript; AEK, participated in the study design and coordination; GGP and HJL participated in the cell survival experiments and performed cell culture; JAK carried out the dCK activity measurement; CDP was involved in the irradiation experiments; GJP has made substantial contribution to the analysis and interpretation of the data and has been revising the manuscript critically; FL participated in the study design and coordination, helped to draft the manuscript and has been revising the manuscript critically; JBV participated in the coordination of the study, has been involved in drafting the manuscript and revising it critically.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2407/6/142/prepub"/></p></sec>
|
Genetic structure of Indian populations based on fifteen autosomal microsatellite loci
|
<sec><title>Background</title><p>Indian populations endowed with unparalleled genetic complexity have received a great deal of attention from scientists world over. However, the fundamental question over their ancestry, whether they are all genetically similar or do exhibit differences attributable to ethnicity, language, geography or socio-cultural affiliation is still unresolved. In order to decipher their underlying genetic structure, we undertook a study on 3522 individuals belonging to 54 endogamous Indian populations representing all major ethnic, linguistic and geographic groups and assessed the genetic variation using autosomal microsatellite markers.</p></sec><sec><title>Results</title><p>The distribution of the most frequent allele was uniform across populations, revealing an underlying genetic similarity. Patterns of allele distribution suggestive of ethnic or geographic propinquity were discernible only in a few of the populations and was not applicable to the entire dataset while a number of the populations exhibited distinct identities evident from the occurrence of unique alleles in them. Genetic substructuring was detected among populations originating from northeastern and southern India reflective of their migrational histories and genetic isolation respectively.</p></sec><sec><title>Conclusion</title><p>Our analyses based on autosomal microsatellite markers detected no evidence of general clustering of population groups based on ethnic, linguistic, geographic or socio-cultural affiliations. The existence of substructuring in populations from northeastern and southern India has notable implications for population genetic studies and forensic databases where broad grouping of populations based on such affiliations are frequently employed.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Kashyap</surname><given-names>VK</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A2" equal-contrib="yes" contrib-type="author"><name><surname>Guha</surname><given-names>Saurav</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" equal-contrib="yes" contrib-type="author"><name><surname>Sitalaximi</surname><given-names>T</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Bindu</surname><given-names>G Hima</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Hasnain</surname><given-names>Seyed E</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Trivedi</surname><given-names>R</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Genetics
|
<sec><title>Background</title><p>Human diversity in India is defined by 4693 different, documented population groups that include 2205 major communities, 589 segments and 1900 territorial units spread across the country [<xref ref-type="bibr" rid="B1">1</xref>]. Anthropologically, the populations are grouped into four major ethnic categories, which include the Australoid, Indo-Caucasoid, Indo-Mongoloid and Negrito populations and linguistically broadly classified as Indo-European, Dravidian, Austro-Asiatic and Sino-Tibetan speakers. The complex structure of the Indian population is attributed to incessant, historical waves of migrations into India, the earliest, by the Austric speakers around 70,000 years ago, followed by the Dravidian speakers from middle-east Asia and the Sino-Tibetan speakers from China and southeast Asia around 8000 to 10,000 years ago. The last major migration is believed to have occurred around 4000 years ago by several waves of Indo-European speakers [<xref ref-type="bibr" rid="B2">2</xref>]. Earlier genetic studies to understand the prevailing diversity among extant Indian populations analyzing populations that were predefined either based on ethnicity, language, culture or geography have interpreted existence of different levels of genetic relationships among population groups [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B6">6</xref>] that broadly attest the theories of migration and assimilation of different populations. However, recent molecular analyses have also asserted genetic similarity across populations spread over diverse geographic regions of the country, revealing a gradation of genetic lineages underscoring the genetic correlation amongst populations [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>].</p><p>The striking social attribute of the Indian populations is their strict practice of endogamy across all social ranks that has resulted in emergence of diverse population-specific social traditions and formation of distinct linguistic dialects due to subsequent isolation of populations. Although uniparental, biallelic markers have deciphered the common major Paleolithic contributions [<xref ref-type="bibr" rid="B9">9</xref>], resolution of many sub-lineages is still awaited in order to decipher finer genetic signatures defining populations that have resisted admixture for centuries. Patterns of variation across recently diverged populations can be successfully characterized with fast-evolving microsatellite markers [<xref ref-type="bibr" rid="B10">10</xref>][<xref ref-type="bibr" rid="B11">11</xref>][<xref ref-type="bibr" rid="B12">12</xref>]. Genetic drift among isolated, small populations manifests as characteristic allele frequency patterns that have been recently effectively characterized to identify genetic clusters that corresponded well with predefined geographically or linguistically similar populations [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>With these rationales, we have analyzed 15 highly polymorphic autosomal microsatellite markers including 13 core forensic loci, which have been extensively used to reveal the ethnological and anthropological affinity of diverse populations ([<xref ref-type="bibr" rid="B10">10</xref>][<xref ref-type="bibr" rid="B11">11</xref>][<xref ref-type="bibr" rid="B12">12</xref>], [<xref ref-type="bibr" rid="B14">14</xref>][<xref ref-type="bibr" rid="B15">15</xref>][<xref ref-type="bibr" rid="B16">16</xref>][<xref ref-type="bibr" rid="B17">17</xref>][<xref ref-type="bibr" rid="B18">18</xref>][<xref ref-type="bibr" rid="B19">19</xref>]). In order to decipher if geographic proximity, linguistic, ethnic and socio-cultural affiliations have played a role in genetic differentiation of extant Indian populations these markers were analyzed in over 3522 individuals drawn from 54 endogamous populations representing major ethnic and linguistic groups spread across diverse geographic regions of the country (Table <xref ref-type="table" rid="T1">1</xref>). Distribution of alleles across populations was evaluated to ascertain presence of group-specific patterns if any. Extent of molecular variance evident among pre-defined groups based on ethnicity, language, geography and socio-cultural hierarchy was evaluated to determine if such classifications were supported genetically. In addition, a model-based clustering algorithm was applied to infer population groups differentiated by their characteristic allele frequencies and to detect presence of cryptic population subdivisions.</p></sec><sec><title>Results</title><p>A number of alleles of the different microsatellite loci analyzed were found to be present unique to specific populations with discernable distribution along geographic and ethnic affiliations evident only among few of the populations. Populations like the Gond (a tribal population) from Chattisgarh; Irular, Chakkiliyar, Gounder and Pallar (Australoid populations) from the southern state of Tamil Nadu; showed genetic isolation, evident from the presence of alleles confined within these populations (Figure <xref ref-type="fig" rid="F1">1</xref>). On the contrary, allele 15.2 of the D3S1358 locus was found to be prevalent among the Gowda and Muslims in the state of Karnataka and allele 18.2 of the FGA locus was present among the Thakur and Kurmi of Uttar Pradesh exhibiting a regional distribution. Sharing of allele 24.2 of the FGA locus was also observed between Lepcha and the Nepali of Sikkim, who share similar ethnic and geographic origins.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Alleles with significant distribution among the different groups of India for the studied microsatellite markers. ○ represents alleles occurring at a high frequency and □ denotes unique alleles present in a population.</p></caption><graphic xlink:href="1471-2156-7-28-1"/></fig><p>Significantly, most frequent alleles were shared among some ethnically and linguistically related populations. The populations of Sikkim, Lai and Lusei of Mizoram that shared Mongol ancestry had a high frequency of allele 12 of the D7S820 locus. Analogous results were obtained for allele 13 of the D5S818 locus, which was in high incidence amongst the Bhutia of Sikkim and Mara of Mizoram. The Indo-Caucasoids, Lingayat of Karnataka; Yadav and Baniya of Bihar, and the geographically proximate Australoid, Kurmi had allele 7 of the Penta E locus in high frequency. Allele 18 of the same locus was present in high frequencies among the Dravidian speaking Australoids, Gowda of Karnataka; Irular of Tamil Nadu as well as among the Indo-European speaking Indo-Caucasoids, Khandayat and Gope of Orissa.</p><p>Analysis of molecular variance (Table <xref ref-type="table" rid="T2">2</xref>) failed to support the geographic, ethnic, linguistic or socio-cultural grouping of Indian populations suggesting little variation between the different groups. We then employed a cluster-based algorithm to ascertain the extent to which the observed discrete patterns of allele distributions would delineate populations. In order to maintain uniformity of estimated probabilities across runs for a given value of K with large datasets [<xref ref-type="bibr" rid="B20">20</xref>], we initially used small K to analyze the 54 populations in this study and then subdivided the dataset into smaller groups to dissect the regional diversity.</p><p>In the countrywide dataset, at K = 5, associated with maximum posterior probability (Table <xref ref-type="table" rid="T3">3</xref>), individuals displayed partial membership to multiple clusters with some populations exhibiting distinctive identities that did not correspond to geographic, linguistic or ethnic affiliation (Figure <xref ref-type="fig" rid="F2">2</xref>). Populations such as Thakur and Khatri from Uttar Pradesh and Baniya from Bihar showed similarity with southern populations such as Naikpod Gond and Chenchu from Andhra Pradesh and with a few individuals from Maharashtra and Lepcha of Sikkim. Populations from the northeastern state of Mizoram exhibited a distinct clustering, different from populations of similar ethnicity from Sikkim, while some individuals from Saora and Gope from the eastern state of Orissa shared a similar degree of membership as the Mizoram populations. Of the southern populations, those from Karnataka and Andhra Pradesh were differentiated into two groups with populations from Tamil Nadu exhibiting split membership to both groups.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p>Bar plot of estimation of the membership coefficient (Q) for each individual of the Indian population grouped on geographic distribution. Each individual is represented by a thin vertical line, which is partitioned into <italic>K </italic>colored segments that represent the individual's estimated membership fractions in <italic>K </italic>clusters. Black lines separate individuals of different population groups based on geography. Population groups are labeled below the figure, with their geographical affiliations above it. The figure shown for <italic>K </italic>= 5 is based on the highest probability run at that <italic>K</italic>.</p></caption><graphic xlink:href="1471-2156-7-28-2"/></fig><p>At the regional level (Figure <xref ref-type="fig" rid="F3">3</xref>), amongst northern Indian populations, at K = 5, where the highest posterior probability was associated, Thakur were identified to be distinct from Jat and Uttar Pradesh Kurmi. The Khatri were found substructured with few individuals exhibiting membership similar to the Thakur.</p><fig position="float" id="F3"><label>Figure 3</label><caption><p>Estimated population structure in different geographic regions. Bar plot estimation figures for North, East, Northeast, South, West, and Central were based on the highest probability run at that K.</p></caption><graphic xlink:href="1471-2156-7-28-3"/></fig><p>In the East, Bihar Brahmin, Bhumihar, Kayasth, Rajput, Yadav, Bihar Kurmi, Orissa Brahmin, Khandayat, Karan, Juang and Paroja shared similar membership to multiple clusters revealing a common genetic structure. Baniya of Bihar were found similar to two of the northern Indian populations while Gope and few individuals from Saora of Orissa shared similar identities as the populations from Mizoram of northeastern India.</p><p>The northeastern populations from Mizoram were identified to be distinct from those of Sikkim. Three clusters were evident with Hmar, Mara, Lai and Lusei of Mizoram all representing one group while Lepcha of Sikkim were distinct representing the second group and the third group comprised Nepali and Bhutia of Sikkim.</p><p>In the south, Lingayat, Gowda, Brahmin and Muslim of Karnataka along with Vanniyar, Gounder and Pallar of Tamil Nadu separated from rest of the populations. Irular of Tamil Nadu and Yerukula of Andhra Pradesh presented distinct identities while Chenchu and Naikpod Gond of Andhra Pradesh exhibited similar affinities. Rest of the populations from Tamil Nadu; Chakkiliyar, Paraiyar, Tanjore Kallar and from Andhra Pradesh; Brahmin, Raju, Komati, Kamma Chaudhury, Kapu Naidu, Reddy and Lambadi displayed mixed membership to multiple clusters.</p><p>Populations from western and central India showed absence of any distinct grouping with individuals having symmetrical membership across inferred clusters. The above results reveal genetic similarity across populations with a few presenting distinct identities that did not follow traditional groupings of geography, language or ethnicity. Populations from southern India and northeastern India largely exhibited structuring while most Indian populations shared similar membership in multiple clusters.</p></sec><sec><title>Discussion</title><p>Contemporary molecular studies on Indian populations were focused to uncover the genetic relationship among geographically, linguistically or ethnically related populations [<xref ref-type="bibr" rid="B21">21</xref>-<xref ref-type="bibr" rid="B25">25</xref>]. Recently, few studies involving a larger number of populations have correlated the genetic relatedness of the populations with linguistic [<xref ref-type="bibr" rid="B6">6</xref>] or socio-cultural affinities, [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B5">5</xref>] though genetic uniformity across populations has also been largely observed [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. The current study employs microsatellite markers to decipher allele frequency changes that would effectively detect recently isolated populations whose times of divergences were shorter than those detectable by uniparental markers. Distribution of alleles across the microsatellite loci studied among the populations predominantly demonstrates the occurrence of alleles unique only to a few populations (Figure <xref ref-type="fig" rid="F1">1</xref>). This pattern is probably due to the result of genetic isolation and drift experienced by the populations that follow strict endogamous practices. The distribution of the most frequent allele was in general, uniform across populations suggesting their common origin. Earlier reports have also suggested geographic contiguity favoring gene flow among populations [<xref ref-type="bibr" rid="B26">26</xref>]. Although ethnic and geographic propinquity were discernible from the allele distribution patterns across few populations in the current study, no consistent pattern across all populations of any particular group was observed. This was also evident from the analysis of molecular variance that failed to support any grouping; ethnic, linguistic, geographic or socio-cultural in contributing to the extant genetic structure of Indian populations.</p><p>The immense diversity within the ethnic and linguistic affiliations of the populations inhabiting India had always been a debatable issue, whether some of them had originated indigenously or were the results of earlier migrations [<xref ref-type="bibr" rid="B27">27</xref>-<xref ref-type="bibr" rid="B29">29</xref>]. The distinct grouping of the populations of Mizoram (Figure <xref ref-type="fig" rid="F3">3</xref>) does concord with earlier reports [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B30">30</xref>] that northeastern India was peopled by migration of Tibeto-Burman speakers from East Asia. However, Tibeto-Burman speaking populations of Sikkim grouped separately and exhibited considerable gene-flow with non-Tibeto-Burman speakers. It is probable that these two regions were peopled by different waves of migration from Southeast and East Asia. Interestingly, eastern Indian populations; Saora and Gope also exhibit similarities to the populations of Mizoram indicating shared genetic ancestry. Though the Lepcha were distinct at the highest-likelihood run for K = 4 (Figure <xref ref-type="fig" rid="F3">3</xref>), in other runs with lower K, they grouped with the rest of the populations from Nepal (data not shown).</p><p>Majority of the Indian populations in general exhibited extensive admixture with each population displaying membership to multiple clusters. Populations such as Khatri, Baniya, Chenchu, Yerukula and Naikpod Gond, however, were substructured. Interestingly, populations comprising the southern Indian region exhibited substructuring with a number of populations clustering into a separate group while the rest were found similar to the general Indian population structure. This group comprising Iyenger Brahmin, Lingayat, Gowda and Muslim from Karnataka and Gounder, Vanniyar and Pallar from Tamil Nadu probably represents those populations that have resisted recent geneflow, and accumulated characteristic allele frequencies because of genetic drift leading to their differentiation from the rest of the populations. In addition, Irular of Tamil Nadu and Yerukula of Andhra Pradesh were found distinctive while Chenchu and Naikpod Gond of Andhra Pradesh grouped together. However, these populations at lower K grouped into clusters similar to those of Tanjore Kallar, Paraiyar and Chakkiliyar of Tamil Nadu and Brahmin, Raju, Komati, Kamma Chaudhury, Kappu Naidu, Kapu Reddy and Lambadi of Andhra Pradesh.</p></sec><sec><title>Conclusion</title><p>Our analyses failed to reveal any genetic groups that correlate to language, geography, ethnicity or socio-cultural affiliation of populations. Of course, the absence of evidence of structuring of the Indian populations based on ethnic, linguistic, geographic or socio-cultural affiliations may be related to the ascertainment bias of selection of these highly polymorphic forensic microsatellite markers. Future studies employing a large number of microsatellites/SNPs might yield higher resolution to decipher stronger associations between populations. The occurrence of few populations distinct from the general populace suggests genetic drift due to isolation of such populations have resulted in their characteristic allele frequencies. This cryptic population structure would have significant implications in forensic investigations where computations of statistical significance of a DNA match rely on ethnic identities often defined by the country of origin. The existence of substructuring in populations from northeastern and southern India also cautions against broad grouping of populations based on geographic, ethnic or linguistic affiliation that are frequently employed in population genetic studies.</p></sec><sec sec-type="methods"><title>Methods</title><p>A total of 3522 consenting individuals from fifty four populations belonging to three major ethnic groups and affiliated to four major language families from across the country were included in this study (Table <xref ref-type="table" rid="T1">1</xref>) after approval of the ethical committee of the Central Forensic Science Laboratory. To ensure representations from all groups, information on geographic origin, ethnicity and linguistic affiliation were recorded for every individual sampled.</p><p>DNA was extracted either from blood or buccal swabs by standard methods [<xref ref-type="bibr" rid="B31">31</xref>]. Amplification was carried out using the Power Plex<sup>®</sup>16 system (Promega Corporation, Madison, USA) or AmpF<italic>l</italic>STR <sup>®</sup>Identifiler™ PCR Amplification Kit (Applied Biosystems, Foster City, California, USA) that coamplify fifteen microsatellite loci according to manufacturers' specifications. The amplified products were separated on a denaturing 5% polyacrylamide gel using the ABI Prism™ 377 DNA Sequencer (PE Applied Biosystems, Foster City, CA, USA). The genotypes were analyzed with GeneScan<sup>® </sup>Analysis 3.1, Genotyper<sup>® </sup>2.5 (PE Applied Biosystems, Foster City, CA, USA) and PowerTyper™ 16 Macro v2 (Promega Corporation, Madison, USA) softwares.</p><p>Analysis of molecular variance, AMOVA [<xref ref-type="bibr" rid="B32">32</xref>], was performed by Arlequin 2.0 software using all 15 loci to ascertain which of the attributes; ethnicity, social hierarchy, geographic or linguistic affiliation of the Indian populations contribute maximum to the extant genetic structure. Significance of the AMOVA values was estimated by use of 10,000 permutations.</p><p>We used a model-based clustering method for inferring population groups using genotype data consisting of unlinked markers as implemented in <italic>Structure </italic>2.1 program [<xref ref-type="bibr" rid="B33">33</xref>]. The model assumes there are <italic>K </italic>populations (where <italic>K </italic>may be unknown), each of which is characterized by a set of allele frequencies at each locus. Individuals in the sample are assigned probabilistically to populations, or jointly to two or more populations if their genotypes indicate they are admixed. Each run used 100,000 estimation iterations for K = 2 to 8 after a 20,000 burn-in length. Each run was carried out several times to ensure consistency of the results. Posterior probabilities for each K were computed for each set of runs.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>VKK designed the course of the study and contributed significantly in manuscript preparation. SG carried out statistical analysis and participated in manuscript preparation. TS analyzed the data and drafted the manuscript. GHB performed experiments on Andhra Pradesh samples and participated in manuscript preparation. SEH provided analytical inputs for manuscript preparation. RT provided critical information for data processing and manuscript preparation.</p><p>All authors read and approved the final manuscript.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Ethnic, linguistic and geographical affiliations of Indian populations included in the study</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="left">Population</td><td align="left">Sample size</td><td align="left">Ethnicity</td><td align="left">Language family</td><td align="left">Area of sampling</td><td align="left">Microsatellite data source</td></tr></thead><tbody><tr><td align="left">1</td><td align="left">Jat</td><td align="left">48</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Uttar Pradesh</td><td align="left">[34]</td></tr><tr><td align="left">2</td><td align="left">Khatri</td><td align="left">47</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Uttar Pradesh</td><td align="left">[34]</td></tr><tr><td align="left">3</td><td align="left">Kurmi</td><td align="left">52</td><td align="left">Australoid</td><td align="left">Austro-Asiatic</td><td align="left">Uttar Pradesh</td><td align="left">[34]</td></tr><tr><td align="left">4</td><td align="left">Thakur</td><td align="left">52</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Uttar Pradesh</td><td align="left">[34]</td></tr><tr><td align="left">5</td><td align="left">Desasth Brahmin</td><td align="left">107</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Maharashtra</td><td align="left">[35]</td></tr><tr><td align="left">6</td><td align="left">Dhangar</td><td align="left">112</td><td align="left">Australoid</td><td align="left">Indo-European</td><td align="left">Maharashtra</td><td align="left">[35]</td></tr><tr><td align="left">7</td><td align="left">Chitpavan Brahmin</td><td align="left">76</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Maharashtra</td><td align="left">[35]</td></tr><tr><td align="left">8</td><td align="left">Maratha</td><td align="left">102</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Maharashtra</td><td align="left">[35]</td></tr><tr><td align="left">9</td><td align="left">Agharia</td><td align="left">53</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Chattisgarh</td><td align="left">[36]</td></tr><tr><td align="left">10</td><td align="left">Gond</td><td align="left">28</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Chattisgarh</td><td align="left">[36]</td></tr><tr><td align="left">11</td><td align="left">Satnami</td><td align="left">44</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Chattisgarh</td><td align="left">[36]</td></tr><tr><td align="left">12</td><td align="left">Teli</td><td align="left">47</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Chattisgarh</td><td align="left">[36]</td></tr><tr><td align="left">13</td><td align="left">Baniya</td><td align="left">45</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[37]</td></tr><tr><td align="left">14</td><td align="left">Bhumihar</td><td align="left">65</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[38]</td></tr><tr><td align="left">15</td><td align="left">Bihar Brahmin</td><td align="left">58</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[38]</td></tr><tr><td align="left">16</td><td align="left">Bihar Kayasth</td><td align="left">45</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[38]</td></tr><tr><td align="left">17</td><td align="left">Bihar Kurmi</td><td align="left">49</td><td align="left">Australoid</td><td align="left">Austro-Asiatic</td><td align="left">Bihar</td><td align="left">[37]</td></tr><tr><td align="left">18</td><td align="left">Rajput</td><td align="left">58</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[38]</td></tr><tr><td align="left">19</td><td align="left">Yadav</td><td align="left">40</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Bihar</td><td align="left">[37]</td></tr><tr><td align="left">20</td><td align="left">Bhutia</td><td align="left">32</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Sikkim</td><td align="left">[39]</td></tr><tr><td align="left">21</td><td align="left">Lepcha</td><td align="left">44</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Sikkim</td><td align="left">[39]</td></tr><tr><td align="left">22</td><td align="left">Nepali</td><td align="left">63</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Sikkim</td><td align="left">[39]</td></tr><tr><td align="left">23</td><td align="left">Hmar</td><td align="left">40</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Mizoram</td><td align="left">[40]</td></tr><tr><td align="left">24</td><td align="left">Lai</td><td align="left">46</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Mizoram</td><td align="left">[40]</td></tr><tr><td align="left">25</td><td align="left">Lusei</td><td align="left">46</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Mizoram</td><td align="left">[40]</td></tr><tr><td align="left">26</td><td align="left">Mara</td><td align="left">44</td><td align="left">Mongoloid</td><td align="left">Tibeto-Burman</td><td align="left">Mizoram</td><td align="left">[40]</td></tr><tr><td align="left">27</td><td align="left">Gope</td><td align="left">60</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Orissa</td><td align="left">[41]</td></tr><tr><td align="left">28</td><td align="left">Juang</td><td align="left">50</td><td align="left">Australoid</td><td align="left">Austro-Asiatic</td><td align="left">Orissa</td><td align="left">[42]</td></tr><tr><td align="left">29</td><td align="left">Karan</td><td align="left">62</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Orissa</td><td align="left">[41]</td></tr><tr><td align="left">30</td><td align="left">Khandayat</td><td align="left">62</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Orissa</td><td align="left">[41]</td></tr><tr><td align="left">31</td><td align="left">Orissa Brahmin</td><td align="left">57</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Orissa</td><td align="left">[41]</td></tr><tr><td align="left">32</td><td align="left">Paroja</td><td align="left">77</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Orissa</td><td align="left">[42]</td></tr><tr><td align="left">33</td><td align="left">Saora</td><td align="left">35</td><td align="left">Australoid</td><td align="left">Austro-Asiatic</td><td align="left">Orissa</td><td align="left">[42]</td></tr><tr><td align="left">34</td><td align="left">Gowda</td><td align="left">59</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Karnataka</td><td align="left">[43]</td></tr><tr><td align="left">35</td><td align="left">Iyengar Brahmin</td><td align="left">65</td><td align="left">Caucasoid</td><td align="left">Dravidian</td><td align="left">Karnataka</td><td align="left">[43]</td></tr><tr><td align="left">36</td><td align="left">Karnataka Muslim</td><td align="left">45</td><td align="left">Caucasoid</td><td align="left">Dravidian</td><td align="left">Karnataka</td><td align="left">[43]</td></tr><tr><td align="left">37</td><td align="left">Lingayat</td><td align="left">98</td><td align="left">Caucasoid</td><td align="left">Dravidian</td><td align="left">Karnataka</td><td align="left">[43]</td></tr><tr><td align="left">38</td><td align="left">Chakkiliyar</td><td align="left">49</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[44]</td></tr><tr><td align="left">39</td><td align="left">Gounder</td><td align="left">56</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[44]</td></tr><tr><td align="left">40</td><td align="left">Irular</td><td align="left">54</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[44]</td></tr><tr><td align="left">41</td><td align="left">Pallar</td><td align="left">33</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[45]</td></tr><tr><td align="left">42</td><td align="left">Tanjore Kallar</td><td align="left">101</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[45]</td></tr><tr><td align="left">43</td><td align="left">Vanniyar</td><td align="left">86</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[45]</td></tr><tr><td align="left">44</td><td align="left">Paraiyar</td><td align="left">21</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Tamil Nadu</td><td align="left">[45]</td></tr><tr><td align="left">45</td><td align="left">Andhra Brahmin</td><td align="left">106</td><td align="left">Caucasoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[46]</td></tr><tr><td align="left">46</td><td align="left">Raju</td><td align="left">66</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[46]</td></tr><tr><td align="left">47</td><td align="left">Komati</td><td align="left">104</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[46]</td></tr><tr><td align="left">48</td><td align="left">Kamma Chaudhury</td><td align="left">106</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[47]</td></tr><tr><td align="left">49</td><td align="left">Kappu Naidu</td><td align="left">107</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[47]</td></tr><tr><td align="left">50</td><td align="left">Kapu Reddy</td><td align="left">107</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[47]</td></tr><tr><td align="left">51</td><td align="left">Chenchu</td><td align="left">100</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[48]</td></tr><tr><td align="left">52</td><td align="left">Yerukula</td><td align="left">101</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[48]</td></tr><tr><td align="left">53</td><td align="left">Naikpod Gond</td><td align="left">104</td><td align="left">Australoid</td><td align="left">Dravidian</td><td align="left">Andhra Pradesh</td><td align="left">[48]</td></tr><tr><td align="left">54</td><td align="left">Lambadi</td><td align="left">108</td><td align="left">Caucasoid</td><td align="left">Indo-European</td><td align="left">Andhra Pradesh</td><td align="left">[48]</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Analysis of molecular variance across different groups of Indian populations</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td></td><td></td><td align="center" colspan="3">Percentage of Variation</td><td></td></tr><tr><td></td><td></td><td></td><td colspan="3"><hr></hr></td><td></td></tr><tr><td align="center">Sample</td><td align="center">Number of groups</td><td align="center">Number of populations</td><td align="center">Within population</td><td align="center">Among populations within groups</td><td align="center">Among groups</td><td align="center">F<sub>st</sub></td></tr></thead><tbody><tr><td align="left">India</td><td align="center">1</td><td align="center">54</td><td align="center">98.16</td><td align="center">1.84</td><td align="center">-</td><td align="center">0.01840</td></tr><tr><td align="left">Geography</td><td align="center">6</td><td align="center">54</td><td align="center">97.95</td><td align="center">1.26</td><td align="center">0.79</td><td align="center">0.02051</td></tr><tr><td align="left">Language</td><td align="center">3</td><td align="center">54</td><td align="center">97.97</td><td align="center">1.51</td><td align="center">0.51</td><td align="center">0.02025</td></tr><tr><td align="left">Ethnicity</td><td align="center">4</td><td align="center">54</td><td align="center">97.99</td><td align="center">1.59</td><td align="center">0.43</td><td align="center">0.02012</td></tr><tr><td align="left">Caste & Tribe</td><td align="center">2</td><td align="center">54</td><td align="center">98.05</td><td align="center">1.76</td><td align="center">0.19</td><td align="center">0.01953</td></tr><tr><td align="left">Caste – Geography</td><td align="center">6</td><td align="center">39</td><td align="center">98.48</td><td align="center">1.00</td><td align="center">0.53</td><td align="center">0.01523</td></tr><tr><td align="left">Tribe – Geography</td><td align="center">5</td><td align="center">15</td><td align="center">96.74</td><td align="center">1.93</td><td align="center">1.32</td><td align="center">0.03258</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Estimates of log probability of data under admixture model for geographic groups of Indian populations</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Run</td><td align="center">K</td><td align="center">ln pr(x/k)</td><td align="center">p(k/x)</td></tr></thead><tbody><tr><td align="center">1</td><td align="center">2</td><td align="center">-190467.3</td><td align="center">0</td></tr><tr><td align="center">2</td><td align="center">3</td><td align="center">-190000.7</td><td align="center">0</td></tr><tr><td align="center">3</td><td align="center">4</td><td align="center">-189591</td><td align="center">6.95 × 10<sup>-15</sup></td></tr><tr><td align="center">4</td><td align="center">5</td><td align="center">-189558.4</td><td align="center">0.99</td></tr><tr><td align="center">5</td><td align="center">6</td><td align="center">-189697.2</td><td align="center">5.24 × 10<sup>-61</sup></td></tr><tr><td align="center">6</td><td align="center">7</td><td align="center">-189576.2</td><td align="center">1.86 × 10<sup>-8</sup></td></tr><tr><td align="center">7</td><td align="center">8</td><td align="center">-189822.7</td><td align="center">0</td></tr></tbody></table></table-wrap></sec>
|
Histological staining methods preparatory to laser capture microdissection significantly affect the integrity of the cellular RNA
|
<sec><title>Background</title><p>Gene expression profiling by microarray analysis of cells enriched by laser capture microdissection (LCM) faces several technical challenges. Frozen sections yield higher quality RNA than paraffin-imbedded sections, but even with frozen sections, the staining methods used for histological identification of cells of interest could still damage the mRNA in the cells. To study the contribution of staining methods to degradation of results from gene expression profiling of LCM samples, we subjected pellets of the mouse plasma cell tumor cell line TEPC 1165 to direct RNA extraction and to parallel frozen sectioning for LCM and subsequent RNA extraction. We used microarray hybridization analysis to compare gene expression profiles of RNA from cell pellets with gene expression profiles of RNA from frozen sections that had been stained with hematoxylin and eosin (H&E), Nissl Stain (NS), and for immunofluorescence (IF) as well as with the plasma cell-revealing methyl green pyronin (MGP) stain. All RNAs were amplified with two rounds of T7-based in vitro transcription and analyzed by two-color expression analysis on 10-K cDNA microarrays.</p></sec><sec><title>Results</title><p>The MGP-stained samples showed the least introduction of mRNA loss, followed by H&E and immunofluorescence. Nissl staining was significantly more detrimental to gene expression profiles, presumably owing to an aqueous step in which RNA may have been damaged by endogenous or exogenous RNAases.</p></sec><sec><title>Conclusion</title><p>RNA damage can occur during the staining steps preparatory to laser capture microdissection, with the consequence of loss of representation of certain genes in microarray hybridization analysis. Inclusion of RNAase inhibitor in aqueous staining solutions appears to be important in protecting RNA from loss of gene transcripts.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Wang</surname><given-names>Hongyang</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Owens</surname><given-names>James D</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Shih</surname><given-names>Joanna H</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Li</surname><given-names>Ming-Chung</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Bonner</surname><given-names>Robert F</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A6" corresp="yes" contrib-type="author"><name><surname>Mushinski</surname><given-names>J Frederic</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib>
|
BMC Genomics
|
<sec><title>Background</title><p>Microarray hybridization has been used to study the global gene expression from many different kinds of tissues and cell lines [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B4">4</xref>]. When it is desired to apply this technique only to certain cells that exist in a heterogeneous tissue, surrounded by cells of other types such as connective tissue cells, it is essential to minimize the contribution of mRNA from undesirable cells by enriching the percentage of desirable cell types [<xref ref-type="bibr" rid="B5">5</xref>]. Laser Capture Microdissection (LCM) is a valuable tool that makes this possible via the visual (microscopic) identification of cells of interest in intact tissues, followed by their excision and subsequent RNA extraction and analysis by microarray hybridization analysis [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]. Frozen sections are highly recommended to maximize quantity and quality of RNA recovery [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. However, in frozen sections it is often difficult to recognize histological details after routine staining, such as hematoxylin & eosin (H&E), in part because LCM requires desiccated sections with no cover slip. Specialized staining methods may be helpful for distinguishing cells of interest from surrounding stroma, e.g., Nissl stain (NS), Immunofluorescence (IF), and Immunohistochemistry (IHC) [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>], but these reagents potentially could result in RNA damage. Methyl Green Pyronin (MGP) is a special stain that has been useful in identifying plasma cells, a major interest of this laboratory, owing to its staining of the nucleus dark blue and the cytoplasm bright pink in visible light or fluorescing red under UV illumination of paraffin-imbedded sections [<xref ref-type="bibr" rid="B13">13</xref>]. Although MGP and all histological stains yield significantly less detail on frozen sections than on paraffin-embedded tissue, MGP did enable identification of plasma cells in the midst of other cell types in oil granulomas (Figure <xref ref-type="fig" rid="F1">1B</xref>, panel c). Since the pyronin reagent reacts with RNA, we worried that MGP might damage the RNA and compromise the subsequent gene expression profiling. Therefore, we investigated whether MGP or other commonly used staining methods themselves would affect gene expression profiling and to what extent.</p><p>Usually, a minimum of 5–50 μg total RNA is required for indirect or direct labeling of cDNA, respectively, with fluorochromes such as Cy3 and Cy5 to perform hybridization on one microarray chip to analyze gene expression [<xref ref-type="bibr" rid="B14">14</xref>]. RNA yields of this magnitude are usually impossible from the small number of cells that typically comprise LCM-derived samples, requiring amplification to make cDNA microarray analysis possible. RNA amplification by RNA polymerase-based <italic>in vitro </italic>transcription is thought to introduce the least bias to gene expression profiling [<xref ref-type="bibr" rid="B15">15</xref>-<xref ref-type="bibr" rid="B17">17</xref>], although it is possible that amplifications methods have the potential of amplifying the effect of damage to mRNAs introduced during LCM processing.</p><p>It has not been determined to what extent each of the steps in LCM sample processing damages the RNA, potentially leading to loss of representation of certain genes when amplified, reverse transcribed, labeled and hybridized to a microarray containing a wide spectrum of cDNA targets on a glass slide. The purpose of the present study was to assess how much the choice of histological staining before LCM would contribute to gene loss, recognized as dropout of hybridized spots, when compared with unstained samples. To render our analysis statistically significant we performed four biological replicates for each staining protocol, starting with 16 individual LCM-derived tissue samples. Each of these 16 was compared with a pool of RNA from an identical pellet of cells. We found that the MGP staining method, which included RNAase inhibitor, was the least detrimental to the RNA, while the Nissl method, which had no RNAase inhibitor, inflicted the most damage to the RNA.</p></sec><sec><title>Results</title><p>RNA was extracted from pellets of cultured TEPC 1165 murine plasma cell tumor cells, which had been flash frozen immediately after harvest. LCM-derived RNA was isolated from samples of microdissected frozen sections of an identical cell pellet, which had been treated with different staining methods. To obtain a sufficient amount of total RNA from LCM-captured cells, we procured cells with 2000–2500 hits of laser (approximately 4000–5000 cells in total). All LCM-captured samples yielded sufficient total RNA (10 – 20 μg) that could be quantitated by the low-range Ribogreen RNA Quantitation Kit (Molecular Probes). Each RNA sample was evaluated for suitability for use by electrophoresis using Agilent RNA Pico chips. RNA that was not degraded should have two clear bands representing the 28S and 18S rRNA, with ratios of 28S/18S absorbance exceeding 1.5. In addition, the RNA sample should be low in genomic DNA contamination, which would appear as bands larger than 28S rRNA. RNA from the unprocessed cell pellet had an acceptable 28S/18S rRNA ratio of about 1.5. The four LCM samples also showed clear 28 rRNA and 18 rRNA bands, but the 28S/18S ratios were different: 1.75 for the unprocessed cell pellet (A), 1.84 for MGP, 4.76 for H&E, 0.71 for NISSL, 2.05 for IF, indicating that a small amount of degradation appeared in the Nissl RNAs (Figure <xref ref-type="fig" rid="F2">2</xref>) the only sample with a ratio less than 1.7. The unusually high ratio on the H&E lane appears to stem from the unusual and not yet understood high background, since substantial amounts of 28S are clearly visible to the naked eye. Five ng of total RNA from all samples were subjected to 2 rounds of amplification. All samples yielded sufficient amplified antisense RNA for microarray hybridization. After two rounds of amplification the total yield of antisense RNA from all samples ranged from 40 μg to 60 μg. The size of the antisense RNA after 2 rounds of amplification was in the expected and satisfactory range between 200 NT to 700 NT for all the samples (data not shown), indicating that these antisense RNAs could be used directly for cDNA synthesis and probe labeling for microarray hybridization. We did not perceive any consistent difference in yield or size of amplified RNAs that could be attributed to the method of staining.</p><sec><title>Comparisons between LCM subgroups</title><p>We found significant differences between the molecular profiles of each "LCM + staining" group and the unprocessed cell pellet (A). One-sample t-tests identified 74, 76, 108, and 77 genes differently expressed at the 0.001 significance level between the unprocessed cell pellet and MGP, H&E, NS and IF, respectively. Testing 8,534 probes at this significance level, we expected that the average number of spuriously significant (false positive) results would be nine or less. Thus, the expression profiles in the staining groups are significantly different from those of the unprocessed cell pellet. Among these genes, 21, 44, 70, and 44 had at least 2-fold mean expression difference for MGP, H&E, NS and IF, respectively (Table <xref ref-type="table" rid="T1">1</xref>) of which 13, 29, 40, and 31 had lower expression. The heat map for the union of the genes differentially expressed (p < 0.001 and ≥ 2-fold mean expression difference) between each LCM + staining sample and the pool of cell pellet preparations (Figure <xref ref-type="fig" rid="F3">3</xref>) confirmed graphically that the MGP staining method introduced the fewest differences. The MGP vs. A comparison in the 4 left-most columns showed the largest number of black elements that indicate neither loss nor gain of signal following staining.</p><p>In addition, we tested the difference in expression profiles among the H&E, NS, IF samples and the MGP samples (Table <xref ref-type="table" rid="T2">2</xref>). The number of genes significant at the 0.001 level and with a mean expression difference greater than 2-fold is equal to 61, 147, and 112 for H&E, NS and IF, respectively. Most of these genes (52, 115, and 89 for H&E, NS and IF, respectively) had weaker signals in these three groups of samples, compared with the MGP-stained group, suggesting loss of signal at additional spots, owing to RNA damage acquired during staining. The global permutation tests also showed that the overall signals in the H&E, NS and IF samples were significantly weaker than in the MGP samples (p-values <0.02).</p><p>Loss of mRNA suitable for hybridizing to some of the cDNAs was the result expected from damage during processing, and this loss was expected to show no gene specificity. The accuracy of this prediction can be visually appreciated in the heat maps presented in Figure <xref ref-type="fig" rid="F3">3</xref>. Thus, no additional analysis was done to analyze which specific genes were involved in expression changes among the individual chips.</p><p>Superimposed upon this experimentally induced spot drop out are the differences inherent in the microarray hybridization method. The number of spots that show increased intensities (always lower than the number of spots with decreased intensities) may be a rough measure of this effect in each subgroup. Thus, we subtracted this value form the number of spots with lower intensities in the stained samples to yield a value that can be used for inter-group comparison.</p></sec></sec><sec><title>Discussion</title><p>The cDNA microarray hybridization method is now widely used to analyze simultaneously the expression of thousands of genes in cancer tissues [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B21">21</xref>], but the contamination of tumors by epithelial, stromal or immune cells presents problems for obtaining accurate expression profiles. Microarray analysis coupled with LCM is a good way to solve this problem. However, cDNA microarray analysis requires a large amount of total RNA (5 μg to 100 μg). It has been shown that one or two rounds of amplification can be used reliably, without compromising RNA quality or gravely skewing patterns of gene expression, when only small initial amounts of total RNA are available [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. In this study, we used two rounds of amplification to obtain amounts of antisense RNA sufficient for microarray hybridization yielding strong signal intensities, permitting statistical data processing.</p><p>To positively identify the specific cells desired for the microdissection, a staining protocol must be used, which permits identification of plasma cells in desiccated cryosections. Such sections, without cover slips, are required for microdissection and may pose a critical problem, since desiccate cryosections are not optically ideal for histological examination (cf. Figure <xref ref-type="fig" rid="F1">1</xref>). However, staining solutions may react with cellular components, such as RNA, potentially adversely affecting RNA integrity. In addition, the aqueous components of the staining reagents may contain or activate intracellular RNAases and result in RNA degradation. For this reason, it was important to test different conditions for tissue staining and section dehydration to assure high quality of RNA from LCM samples. To assure ourselves of adequate numbers of uniform populations of plasma cells, which are scarce in most tissues, we utilized pellets of cultured plasma cell tumor cells that were sectioned and stained as if they had been collected by LCM. We used this approach, because obtaining large enough sample of plasma cell tumor cells from tissues by LCM would have been impracticable.</p><p>Our modification (addition of RNAase inhibitor) to the standard MGP staining method has made it suitable for use with LCM. Sections treated as described here can be used to prepare total RNA of good quality. Though pyronin is a fluorescent molecule that can bind to RNA, we detected only 21 changes when compared with unstained LCM samples. This amounted to less than 0.3% of the analyzed microarray elements, and less than one half of the number changed due to the other staining methods tested.</p><p>It appears that inclusion of RNAase inhibitor in staining solutions can prevent RNA from damage by endogenous and exogenous RNAase. The Nissl-stained group was the only staining protocol that did not include RNAase inhibitor. The number of genes in this group with 2-fold changes, 70 spots, was the highest among all four groups, indicating that this staining procedure damaged the most mRNAs.</p><p>In summary, we performed a statistical analysis of 16 cDNA microarray chips comparing fluorescent-labeled cDNA generated from amplified RNA from 16 independently isolated and processed, LCM-procured, stained frozen sections with amplified RNA from an identical pellet of plasma cell tumor cells. This showed that a variable amount of spot fall-out was seen after staining of the frozen sections, depending on the staining method, indicating that the choice of staining method can be important in minimizing loss of mRNAs from microarray analysis of LCM-derived RNA. Happily, the degree of mRNA damage was small with the commonly used H&E method, so long as RNAase inhibitor was included in the aqueous hematoxylin solution. Nissl staining solution, which may not be compatible with RNAase inhibitor addition, introduced the greatest degree of damage. Fortunately, for our ongoing study of plasma cell tumor development, the MGP stain, chosen for its unique ability to flag plasma cells in tissue sections, was the least destructive of all. Our analysis suggests that maximizing the use of non-aqueous staining solutions is recommended for other specialty stains that might be advantageous for identification of other cell types in studies that require use of LCM. The inclusion of RNAase inhibitor in aqueous solutions appears to be important in rendering histological staining procedures suitable to extraction of intact RNA.</p></sec><sec><title>Conclusion</title><p>RNA damage can occur during the staining steps preparatory to laser capture microdissection, resulting in loss of representation of certain genes in microarray hybridization analysis. Inclusion of RNAase inhibitor in aqueous staining solutions appears to be important in protecting RNA from loss of gene transcripts.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Cell culture</title><p>The murine plasma cell tumor cell line TEPC 1165 [<xref ref-type="bibr" rid="B18">18</xref>] was used both as a source of frozen sections for LCM as well as a source of reference RNA, i.e., RNA extracted without LCM processing. Cells were grown in RPMI 1640 media supplemented with 10% fetal bovine serum, 10 ng/ml IL-6, 4 mM L-glutamine and 100 units/ml penicillin and 100 units/ml streptomycin. Duplicate cell pellets were collected by low-speed centrifugation in 50-ml conical centrifuge tubes and immediately snap frozen for RNA extraction and LCM preparation.</p></sec><sec><title>Preparation of LCM samples</title><p>The cell pellets were frozen in a dry ice-ethanol bath, embedded with OCT frozen tissue embedding medium, and stored at -80°C. Six-micrometer serial frozen sections were cut on a cryostat at -20°C and mounted on chilled, pre-cleaned, uncoated microscope glass slides that had been heated at 200°C overnight to eliminate RNAase. Slides were kept on dry ice or at -80°C until LCM was completed. Sections were stained individually with four different methods before LCM. RNAase-free (treated with diethylpyrocarbonate) water was used for all solutions. All procedures were carried out at room temperature, unless otherwise stated, and usually on the same day as the frozen sectioning.</p><sec><title>Methyl Green Pyronin staining</title><p>After fixation in 70% ethanol (10 sec), each slide was rinsed in DEPC-treated water (5 sec). Then, 50 μl of MGP staining solution (Sigma/Aldrich) to which RNAase inhibitor (SUPERase*In, Ambion) had been added to a final concentration of 500 U/ml, was applied to the section and incubated 2 min, followed by a quick dip in diethylpyrocarbonate-treated water, dehydration in 100% ethanol in 2 steps (quick dip, then for 10–15 sec), and xylenes in 2 steps (2 min then 3 min). It is important to note that MGP, like all histological stains, yields significantly less detail on frozen sections than on paraffin-embedded tissue, in part, owing to lack of a cover slip. MGP did enable identification of plasma cells in oil granulomas under these conditions. To be used in this way, however, one must include RNAase inhibitor if RNA extraction is desired after LCM.</p></sec><sec><title>Hematoxylin and eosin staining</title><p>A modification was incorporated into the manufacturer's (Sigma/Aldrich) protocol; namely, RNAase inhibitor (SUPERase*In, Ambion) was added to the aqueous hematoxylin solution to a final concentration of 500 U/ml. Briefly, slides were stained as follows: 70% EtOH for 10 sec, DEPC-treated water for 5 sec, hematoxylin with RNAase inhibitor for 20 sec, 70% EtOH for 30 sec, eosin Y in 100% EtOH for 20 sec, followed by dehydration with a series of alcohol for 30 sec each, and xylenes for 2 min.</p></sec><sec><title>Nissl staining</title><p>The frozen sections were fixed with 70% ethanol (30 sec), rinsed in DEPC-treated water (30 sec), immersed in NS staining solution (Arcturus Corp.) for 30 sec, dehydrated with graded alcohols (30 sec each) and xylenes (2 min and 3 min).</p></sec><sec><title>Immunofluorescence</title><p>Sections were fixed in ice-cold acetone for 2 min, rinsed in DEPC-treated PBS for 10 sec, incubated with a 1:250 dilution of goat anti-mouse Igκ conjugated with TXRD (Southern Biotechnology Associates, Inc.) to which RNAase inhibitor (SUPERase*In, Ambion) had been added to a final concentration of 500 U/ml, for 2 min in the dark, dehydrated with 100% ethanol and xylenes (11).</p></sec></sec><sec><title>Laser capture microscopy</title><p>Cells were lifted from the slide onto CapSure LCM plastic caps (Arcturus Corp.) using a 30-μm spot size of laser at the power of 50 μv using PixCell II LCM system (Arcturus Corp.).</p></sec><sec><title>RNA extraction and purification</title><p>Total RNA from the unprocessed cell pellet was extracted using TriZol reagent (Sigma), treated with RNAase-free DNAase I (Qiagen, Inc.), followed by RNA purification with Qiagen RNeasy kit according to the manufacturer's protocol. RNA concentration was measured spectrophotometrically.</p><p>RNA extraction of 16 LCM-derived samples (4 from each staining group) was performed with the PicoPure RNA extraction kit (Arcturus Corp.), simultaneously treated with RNAase-free DNAase I (Qiagen, Inc) according to the manufacturer's protocol. Briefly, 50 μl of extraction buffer was pipetted onto the tissue fragments that had been collected in the cap of a sterile plastic 1.5-micro-centrifuge tube (CapSure LCM caps, Arcturus Corp.), and they were incubated together in a hot-air oven at 42°C for 30 min. Cells and buffer were collected by centrifugation at 800 × g for 2 min. RNA was DNAase-treated and purified on a preconditioned RNA-purification spin column according to the manufacturer's recommendations. LCM-derived total RNA was quantitated with Ribogreen (Molecular Probes, Inc.) using Victor 2 1420 multilabel counter (Perkin-Elmer/Wallac). To assess the RNA quality and confirm the absence of genomic DNA contamination, the Agilent 2100 Bioanalyzer (Pico RNA chip) was used to perform electrophoresis on picogram amounts of total RNA. Only samples with intact 18S and 28S rRNA peaks were used for gene expression analysis.</p></sec><sec><title>RNA amplification</title><p>We performed antisense RNA amplification of each RNA sample with RiboAmp™ RNA Amplification kit (Arcturus, Inc.) based on T7 RNA polymerase <italic>in vitro </italic>transcription of cDNA synthesized by reverse transcription primed by oligo-dT attached to the T7 RNA polymerase promoter [<xref ref-type="bibr" rid="B15">15</xref>]. To avoid bias caused by different starting amounts of total RNA in the amplification process, 5 ng of total RNA were used in all experimental and reference samples. To obtain sufficient antisense RNA for microarray experiments, two rounds of amplification were performed on all samples. The total yield of amplification products was determined using Ribogreen (Molecular Probes, Inc.), and the sizes of the amplified RNAs were determined with the Agilent Bioanalyzer (nano RNA chip).</p></sec><sec><title>Probe labeling and cDNA array hybridization</title><p>Probes were synthesized and labeled from 4 μg of amplified RNA (aRNA) by the method of Xiang et al [<xref ref-type="bibr" rid="B14">14</xref>]. In brief, 4 μg of amplified RNA were combined with 4 μg amine-modified random primer and 5 units of RNAase inhibitor (SUPERase*In, Ambion). The mixture was incubated at 70°C for 10 min, then chilled on ice for 10 min, and left at room temperature for 10 min. Primer-RNA solution was added to the reverse transcriptase mix (including 0.5 mM dATP, dGTP, dCTP, 0.3 mM dTTP, and 0.2 mM aminoallyl-dUTP) and incubated at 42°C for 2 h. The reaction was terminated by adding 10 μl 0.5 M EDTA, and RNA was hydrolyzed with 10 μl 1 M NaOH at 65°C for 30 min. The solution was neutralized with 10 μl 1 M HCl, and then MinElute PCR purification kits (Qiagen, Inc) were used to purify the cDNAs. The cDNA was eluted three times with 10 μl of H<sub>2</sub>O. In each staining group 3 of the 4 reference RNA samples were labeled with Cy5, while 1 of the 4 was labeled with Cy3. Three of each group of 4 experimental samples were labeled with Cy3, and 1 of 4 was labeled with Cy5 to test for bias from labeling. Probes were mixed and hybridized overnight at 42°C in hybridization solution (New England Nuclear) to microarray chips with a total of 9998 cDNA elements (InCyte) printed by the NCI microarray facility.</p></sec><sec><title>Evaluation of gene-specific dye bias</title><p>Four arrays were hybridized for each of the 4 groups (MGP, HE, IF, NS). In each group 3 arrays had cDNA derived from the reference, unsectioned pellet of cells labeled with Cy5 and the LCM-derived test sample labeled with Cy3, and one array had the labeling reversed. We were interested in evaluating the possibility of gene-specific dye bias that was not removed by the normalization process. To evaluate this possibility, we computed the average difference in log2 ratio between normalized forward- and reverse-label experiments for each array element as an estimate of the residual dye bias for the associated gene. Of the 8534 array elements, none had an average difference of log2 ratio in absolute value greater than 1 (i.e., 2-fold difference). Thus, we concluded that after normalization there was no systematic bias favoring either dye.</p></sec><sec><title>Array scanning and data processing</title><p>Arrays were scanned at 10-μm resolution using a GenePix 4000 (Axon Inc.) array scanner. PMT voltages were varied to gain optimum intensity of the spots. Four duplicate hybridizations/slides were used in each experiment. LCM-derived RNAs were regarded as "experimental" samples, and 3 of the 4 cDNA probes derived from them were labeled with Cy3, whereas the "reference" samples, derived from RNA from unprocessed cell pellets were usually labeled with Cy5. To rule out color bias produced by the labeling process itself, the fluorochromes were reversed once in each group of four repeats. Quantification files were obtained by using the GenePixPro 4.0 software (Axon Inc.). The fluorescence signal intensity was determined as the volume in a fixed-size circle, and background was estimated as the median pixel value in a diamond-shaped region between each adjacent four-spot region.</p></sec><sec><title>Data filtering and normalization</title><p>Log-ratios of local median background-subtracted intensity levels were analyzed. For each array element, if the intensity was less than 75 in one channel but greater than 75 in the other channel, the intensity <75 was set at 75. Array elements that were flagged as poor quality during image analysis or that had intensities less than 75 in both channels were not reliable and treated as missing [<xref ref-type="bibr" rid="B19">19</xref>]. Six percent of the array elements were found to be missing in this way. Those samples with missing values for a particular array element were excluded from all analyses involving that array element. In order to have enough samples in each group to compare expression profiles between the groups, array elements were further filtered out and removed from analysis if their number of non-missing values in each group was less than three. Eleven percent of the array elements were filtered out in this way, resulting in 8534 array elements for analysis. Log-ratios for each microarray were normalized by the locally robust lowess smoother (20) to adjust for any dye bias and variation in PMT voltage settings.</p></sec><sec><title>Clustering</title><p>Average linkage hierarchical analysis of differentially expressed genes was performed by using 1-Pearson correlation as the distance metric to group genes based on their patterns of variation across the staining groups. The clusters and associated heat maps were implemented in R.</p></sec><sec><title>Enumeration of differentially expressed genes</title><p>Identification of array elements that were differentially expressed between each subgroup and the reference sample and among subgroups was done using one-sample t-tests and two-sample t-tests, respectively. Differentially expressed genes were identified as those genes that were significant at the 0.001 level (p < 0.001) and which were at least 2-fold different in the geometric mean of the expression measurements. A treated subgroup was considered to have weaker expression than another treated subgroup if, among the genes that were differentially expressed between these two subgroups, it had more lower-expressed genes than higher-expressed genes. A global permutation test was used as a global test to assess if one subgroup had statistically significant weaker expression than the other subgroup. Specifically, two-sample t-tests were performed by randomly permuting the group labels. In each permutation, the difference between the number of lower-expressed genes and the number of higher-expressed genes was recorded. The p-value of the permutation test was the proportion of permutations with the same or larger difference as those observed in the original data. Two subgroups would be defined as having globally significant different expression profiles if the p-value from the permutation test is less than 0.05.</p></sec></sec><sec><title>Abbreviations</title><p>LCM (Laser Capture Microdissection), MGP (Methyl Green Pyronin), H&E (Hematoxylin and Eosin), NS (Nissl Stain), IF (Immunofluorescence).</p></sec><sec><title>Authors' contributions</title><p>HW, RFB and JFM conceived the idea for the project. HW and RFB performed the laser capture microdissection. JDO prepared and quality controlled some of the RNAs. HW performed the microarray hybridizations and scans. JHS and M-CL performed the statistical analysis and interpretation.</p></sec>
|
Alu elements contain many binding sites for transcription factors and may play a role in regulation of developmental processes
|
<sec><title>Background</title><p>The human genome contains over one million Alu repeat elements whose distribution is not uniform. While metabolism-related genes were shown to be enriched with Alu, in structural genes Alu elements are under-represented. Such observations led researchers to suggest that Alu elements were involved in gene regulation and were selected to be present in some genes and absent from others. This hypothesis is gaining strength due to findings that indicate involvement of Alu elements in a variety of functions; for example, Alu sequences were found to contain several functional transcription factor (TF) binding sites (BSs). We performed a search for new putative BSs on Alu elements, using a database of Position Specific Score Matrices (PSSMs). We searched consensus Alu sequences as well as specific Alu elements that appear on the 5 Kbp regions upstream to the transcription start site (TSS) of about 14000 genes.</p></sec><sec><title>Results</title><p>We found that the upstream regions of the TSS are enriched with Alu elements, and the Alu consensus sequences contain dozens of putative BSs for TFs. Hence several TFs have Alu-associated BSs upstream of the TSS of many genes. For several TFs most of the putative BSs reside on Alu; a few of these were previously found and their association with Alu was also reported. In four cases the fact that the identified BSs resided on Alu went unnoticed, and we report this association for the first time. We found dozens of new putative BSs. Interestingly, many of the corresponding TFs are associated with early markers of development, even though the upstream regions of development-related genes are Alu-poor, compared with translational and protein biosynthesis related genes, which are Alu-rich. Finally, we found a correlation between the mouse B1 and human Alu densities within the corresponding upstream regions of orthologous genes.</p></sec><sec><title>Conclusion</title><p>We propose that evolution used transposable elements to insert TF binding motifs into promoter regions. We observed enrichment of biosynthesis genes with Alu-associated BSs of developmental TFs. Since development and cell proliferation (of which biosynthesis is an essential component) were proposed to be opposing processes, these TFs possibly play inhibitory roles, suppressing proliferation during differentiation.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Polak</surname><given-names>Paz</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" corresp="yes" contrib-type="author"><name><surname>Domany</surname><given-names>Eytan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Genomics
|
<sec><title>Background</title><p>Over 90% of the human and other vertebrate genomes don't have any known functional role [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>], and as such were considered until recently as "Junk DNA" or genomic "dark matter". A large fraction of the "non-functional" DNA originated from mobile elements [<xref ref-type="bibr" rid="B1">1</xref>] such as Alu, which comprise about 10% of the nucleotides of the human genome [<xref ref-type="bibr" rid="B1">1</xref>] with over one million inserted copies. This abundance is somewhat of a surprise, since Alu is a non-autonomous retroelement, i.e. it doesn't encode proteins that assist its mobilization, and therefore it needs to rely on the cell's machinery for its duplication in the genome [<xref ref-type="bibr" rid="B3">3</xref>]. Currently Alu is believed to be a parasite of the mobilization machinery of long interspersed elements (LINE) [<xref ref-type="bibr" rid="B4">4</xref>].</p><p>A typical Alu is a dimer, comprised of a central A-rich region which is flanked by two similar sequence elements of about 130 bp (left and right arms). During evolution the Alu elements have spread in the human genome in several bursts at different times, which allow their classification into several families and subfamilies, on the basis of their insertions, deletions and mutations. The major families are: old AluJ, whose members are AluJb and AluJo (both spread in the genome 80–100 mya); the middle, AluS family, which includes AluSx, AluSg, AluSp AluSc and AluSq (35 –50 mya), and the youngest Alu family, AluY (25 mya) [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><p>One of the biggest mysteries associated with Alu concerns the non-random manner in which these elements are distributed throughout the human genome [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. The distribution of Alu was found to be highly correlated with local GC content and with the density of genes and with intron density [<xref ref-type="bibr" rid="B8">8</xref>]. Furthermore, the density of Alu is higher in intragenic (12.5% of the nucleotides) than in intergenic regions (9.6% of the nucleotides) [<xref ref-type="bibr" rid="B8">8</xref>], which is surprising if indeed Alu elements are non-functional bits of DNA. In addition, Alu elements are negatively selected in imprinted regions of primate genomes [<xref ref-type="bibr" rid="B9">9</xref>]. Transposable elements are underrepresented in the region surrounding the TSS of genes [<xref ref-type="bibr" rid="B10">10</xref>]. Another analysis revealed that almost 20% of the genes contain transposable elements in the 3' UTR [<xref ref-type="bibr" rid="B11">11</xref>]. Significant differences were observed also when Alu density was compared among full length gene sequences of different biological process classes (this study [<xref ref-type="bibr" rid="B12">12</xref>] was limited to chromosomes 21 and 22). Genes that were related to metabolism, transport and signalling were mostly Alu rich, whereas genes that had poor Alu content belonged mostly to structural proteins and to information processing and storage pathways [<xref ref-type="bibr" rid="B12">12</xref>]. It was also shown recently that housekeeping genes contain more Alu than tissue specific genes [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>This non-random of distribution of Alu, in particular the high concentration near and within genes, might be the result of positive feedback; selective pressure drives the organism to use these elements for some new functions, and the functions acquired by Alu generate advantage for phenotypes that have high Alu concentration near genes. Indeed, the possible functional roles of Alu in genomic organization and gene expression has received increasing attention during the last two decades (see [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>] for reviews). One of the more fascinating possible consequences of integration of mobile elements near genes is acquisition of a transcriptional regulatory role by the Alu element, as a carrier of different TF binding sites. The Alu element harbours BSs of several transcription factors that were shown to bind to Alu; most of these TFs were nuclear factors, hormones, calcium nuclear factors, as well as other TFs [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B22">22</xref>]. The idea of Alu acting as a carrier of cis regulatory elements was suggested by Britten [<xref ref-type="bibr" rid="B23">23</xref>] and by Oei at el. [<xref ref-type="bibr" rid="B20">20</xref>] who found YY1 and SP1 BSs on young Alu family members. These experimental results imply that Alu elements have influence on the regulation of expression levels of many genes that have Alu in their promoter regions; however, such a wide ranged regulatory impact has not yet been tested experimentally.</p><p>The main aim of our study was to find new putative BSs that reside on Alu. Very recently, several groups have performed similar analysis for all transposable element classes; however, the set of PSSMs studied [<xref ref-type="bibr" rid="B24">24</xref>,<xref ref-type="bibr" rid="B25">25</xref>] contained about half of what is presented here. In addition, our analysis is complemented by an extensive literature search of past experimental work on the Alu-associated binding sites that we found, as well as the biological functions of the corresponding TFs and target genes. We started by searching the Alu <italic>consensus sequences </italic>and found new, previously unnoticed BSs, many of which turned out to be related to developmental processes.</p><p>The regulatory role played by Alu for genes that are related to differentiation or development was already established for at least six genes [<xref ref-type="bibr" rid="B26">26</xref>]. In these cases Alu was shown to act as enhancer or silencer involved in regulating the expression of six developmental genes. For one of these, the T-cell marker gene <italic>CD</italic>8α, the silencing mechanism was shown to act via binding of <italic>LYF-1</italic>, <italic>bHLH </italic>and <italic>GATA3 </italic>to the Alu situated in the last intron of <italic>CD</italic>8α [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. These experimental results demonstrated the regulatory impact of Alu on developmental genes and during developmental stages. To explore further and enhance the possible regulatory roles of Alu we focused on the genomic region most likely to have a regulatory function – the upstream regions of genes. We scanned these regions and searched for BSs on the <italic>particular Alu sequences </italic>that were identified. Our findings show for the first time that TFs that are known to regulate the developmental processes may bind to Alu elements that are incorporated in the <italic>promoter regions </italic>of genes that need to be activated or suppressed during development (such binding has already been demonstrated for Gfi-1, PITX2 and Nkx2.5 – see below).</p></sec><sec><title>Results and discussion</title><sec><title>The Alu consensus sequences are enriched with Binding Motifs</title><p>We focused on the eight major Alu subfamilies AluJo, Jb, Sx, Sz, Sc, Sq, Sp and Y [<xref ref-type="bibr" rid="B3">3</xref>], whose sequences were downloaded from RepBase [<xref ref-type="bibr" rid="B29">29</xref>]. We refer to the ensemble of sequences that bind a TF as it's Binding Motif (BM). In order to find BMs of a TF, we used its Position Specific Score Matrix (PSSM); 410 PSSMs were downloaded from two databases [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. For each PSSM we set two threshold scores, T5 and T1, selected so that the probability to find a target BM with score higher than T5 in a suitably chosen random sequence (see Methods) is less than 0.05 (0.01 for T1). Furthermore, for each BM we identified Tmax, the maximal score observed on all the above listed Alu subfamilies' consensus sequences. For 66 BMs Tmax was higher than T5 and for 25 TFs – higher than T1 (see Table <xref ref-type="table" rid="T1"> 1</xref>). A good measure for the statistical significance of enrichment of Alu with BMs is the number of BMs whose score exceeds T5; for example, we found on AluSx 35 BMs with scores higher than their respective T5. The probability to find this many BMs in a random model (i.e. p-value) is 0.023 (see Methods).</p><p>In order to control the problem of multiple testing we performed a false discovery rate (FDR) analysis at 20% FDR[<xref ref-type="bibr" rid="B32">32</xref>]. The 14 BMs that passed the FDR test with Q = 0.20 in AluSx are listed in Table <xref ref-type="table" rid="T1">1</xref>, where we also compare the IUPAC consensus sequences of these BMs [<xref ref-type="bibr" rid="B33">33</xref>] with the target sequence that had the highest score in all the major Alu subfamilies. Most sequences with scores higher than T1 differ at not more than two nucleotides from the IUPAC consensus (see Table <xref ref-type="table" rid="T1">1</xref> and <xref ref-type="supplementary-material" rid="S1">Additional data file 1</xref>), and there are several PSSMs, such as PITX2_Q2, LYF-1.01, PAX4.01, SIX3.01, DELTAEF.01 and NKX25.01 (Table <xref ref-type="table" rid="T1">1</xref>, and <xref ref-type="supplementary-material" rid="S1">Additional data file 1</xref>), whose IUPAC consensus sequences agree with the highest-scoring target sites on at least one of the Alu consensus sequences.</p></sec><sec><title>BMs tend to cluster together on specific parts of Alu consensus sequences, that also tend to be more conserved during evolution</title><p>We checked the BM's locations along the consensus sequences on both strands of the eight Alu subfamilies. In general, most of the putative BSs were found on the sense strand. It is interesting to note that Alu elements show a 55% to 45% bias towards antisense orientation [<xref ref-type="bibr" rid="B34">34</xref>] relative to nearby genes. It would be of considerable interest to check experimentally whether this orientational difference is reflected in functional roles of the TFs. The target sites that passed T5 cover, on average, 50% of the consensus sequence of each subfamily. On the other hand, the total length of all the binding motifs found on one Alu consensus sequence exceeds, on the average, 420 nucleotides (150% of the Alu's length). This redundancy in putative target sites is caused by the fact that several PSSMs have very similar structure, giving top scores to the same DNA sequence.</p><p>The short sequences GGTAATCCC on the 5' to 3' strand and its complementary sequence GGGATTACC serve as a putative BS of five TFs, four of which are homeodomain proteins (Table <xref ref-type="table" rid="T1">1</xref> and <xref ref-type="supplementary-material" rid="S1">Additional data file 1</xref>, TFs and locations marked by *). This sequence is part of the consensus of all eight Alu subfamilies, and in the five middle AluS subfamilies the BS appears on both the left and right monomers. Moreover, this is precisely the sequence identified by Britten [<xref ref-type="bibr" rid="B23">23</xref>] as one of the most conserved subsequences on Alu, which was seen as a surprise since it did not have any role in retrotransportation or for any other function. Later on, it was found that in the germ line the Specific Alu Binding Protein (SABP) can bind to this sequence and prevent methylation [<xref ref-type="bibr" rid="B35">35</xref>], but since the importance of this function of is still not understood, and hence the high level of conservation is still a mystery. Another region is a putative harbour for several BMs is the middle poly A- stretch (that bridges the two Alu monomers). We found this sequence to be a putative target site for many binding motifs such as all MEF2 family members, HNF1.03, OC.2 and PAX4 (marked by ** in Table <xref ref-type="table" rid="T1">1</xref> and <xref ref-type="supplementary-material" rid="S1">Additional data file 1</xref>). These putative target sites were also found to accumulate less mutations than the expected rate [<xref ref-type="bibr" rid="B23">23</xref>]. More recently both the central poly A stretch and the subsequence mentioned above [<xref ref-type="bibr" rid="B36">36</xref>-<xref ref-type="bibr" rid="B38">38</xref>] were found to be associated with A-to-I editing.</p></sec><sec><title>Existing evidence for TF binding to Alu (on DNA in solution)</title><p>Having identified BSs of various TFs on Alu, we searched for existing experimental evidence for binding and regulation associated with these motifs. Indeed, several TFs that are known to bind to motifs that passed our thresholds have already been shown to bind to DNA, and the fact that the binding sequences reside on Alu has been noticed (denoted by * on Table <xref ref-type="table" rid="T2">2</xref>). We show in Table <xref ref-type="table" rid="T2">2</xref> four additional BMs that we found and report here for the first time; their corresponding TFs were shown to bind to the target sites identified by us, but the fact that they reside on Alu has not been previously noted. These TFs (bold in Table <xref ref-type="table" rid="T2">2</xref>) are PITX2, Nkx2.5 Gfi-1 and LUN1. The most striking example was LUN-1; Chu et al. [<xref ref-type="bibr" rid="B39">39</xref>] observed (by EMSA) that it binds to 55 DNA sequences, 27 of which are palindromic; we discovered that all of these 27 reside on Alu. The situation is different for Gfi-1; there the PSSM [<xref ref-type="bibr" rid="B30">30</xref>] is based on 21 target sites, only five of which are part of Alu (Table <xref ref-type="table" rid="T3">3</xref>). As to PITX2, it's PSSM [<xref ref-type="bibr" rid="B31">31</xref>] was based on target sites on the promoter of <italic>PLOD1</italic>, all of which reside on Alu subsequences [<xref ref-type="bibr" rid="B40">40</xref>]. Finally, the heart development marker Nkx2.<italic>5 </italic>is the only one whose PSSM was calculated using only off-Alu target sites – nevertheless, the target site found by us, that resides on Alu, is the one with the highest possible score.</p></sec><sec><title>Evidence (in live cells) for regulation by TFs bound to Alu</title><p>The regulatory role of single – gene BSs on Alu was demonstrated for several TFs by a transfection assay for hormone ligands [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B22">22</xref>], for YY1 [<xref ref-type="bibr" rid="B20">20</xref>], and for Estrogen Receptors (ER) [<xref ref-type="bibr" rid="B18">18</xref>]. We found experimental evidence indicating that Gfi-1, PITX2 and Nkx2.5 regulate a reporter gene's expression [<xref ref-type="bibr" rid="B40">40</xref>-<xref ref-type="bibr" rid="B42">42</xref>] via BSs that reside on Alu; where the PITX2 function was validated in two different experiments [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B42">42</xref>]. In addition, Alu was essential for regulation of <italic>PLOD1 </italic>by PITX2, shown by using two constructs of <italic>PLOD1 </italic>promoter: one with several Alu elements and one without any Alu. These researchers [<xref ref-type="bibr" rid="B40">40</xref>] found a 3-fold change in the induction of a reporter gene with constructs that contained several Alu elements with respect to the expression level of the construct from which Alu elements were excluded. The same procedure of comparing promoters with and without Alu was done again on <italic>PLOD1 </italic>[<xref ref-type="bibr" rid="B42">42</xref>], but this time they checked PITX2 together with Nkx2.5 in order to prove the synergic effect on <italic>PLOD1 </italic>expression, and again a big difference between the expression levels of the reporter gene was found.</p><p><bold>ChIP analysis </bold>was done on 21 genes' promoters that contain Gfi-1 putative target sites; five of these promoters contained suspected BSs on Alu (table <xref ref-type="table" rid="T3">3</xref>). However, only one of these five genes (<italic>JAK3</italic>) had its Gfi-1 BS exclusively on Alu [<xref ref-type="bibr" rid="B41">41</xref>]. Hence the positive result of the ChIP experiment on this gene shows that Gfi-1 actually binds <italic>in vivo </italic>to Alu.</p></sec><sec><title>Developmental and stress response TFs constitute a majority of the BSs on Alu</title><p>Most of the stress response TFs that bind to Alu are nuclear factors and hormone ligands that were validated experimentally, such as ER, RAR, LXRE and SP1 (Tables <xref ref-type="table" rid="T2">2</xref>, <xref ref-type="table" rid="T4">4</xref>). As was mentioned above, a subset of them are common to all Alu subfamilies while some appear only on the young Alu sequences, or on Alu elements that were mutated [<xref ref-type="bibr" rid="B20">20</xref>]. Another interesting group of TFs is developmental (usually referred to either as early markers or regulators of development); more than 30 members of this group passed the T5 threshold on at least one Alu consensuses sequence. The largest set of developmental TF binding sites that appear on Alu belong to hemateopoietic transcription factors [<xref ref-type="bibr" rid="B43">43</xref>] such as Gfi-1, LYF-1, GATA3, GATA2 and ONECUT2 (Table <xref ref-type="table" rid="T4">4</xref>). Another set of BMs belong to TFs that are related to muscle and heart development [<xref ref-type="bibr" rid="B44">44</xref>], such as PITX2, MEF2, NKX2.5 and LUN1 (Tables <xref ref-type="table" rid="T2">2</xref>, <xref ref-type="table" rid="T4">4</xref>), and some to brain development: PAX4, NRL, OTX2 (Tables <xref ref-type="table" rid="T2">2</xref>, <xref ref-type="table" rid="T4">4</xref>). Note that even though the PAX6.01 PSSM did pass our thresholds (and hence appears in Table <xref ref-type="table" rid="T4">4</xref>), it has been demonstrated [<xref ref-type="bibr" rid="B45">45</xref>] that PAX6 actually binds to very specific Alu sequences, that constitute only 0.2% of the genome-wide Alu elements. The PSSM that was used in our analysis (taken from MatInspector) was constructed from BMs that were identified in different experiments and its dominant binding sequences do not coincide with those that [<xref ref-type="bibr" rid="B45">45</xref>] identified as the Alu mediated PAX6 BMs.</p></sec><sec><title><bold>The distribution of transposable elements in the 5' upstream regions of human genes</bold></title><p>The distribution of transposable elements in the 5' upstream regions of human genesis not uniform. We computed the density of different transposable elements on 5000 bp long upstream sequences of about 14000 genes. In the region between 2000 bp and 5000 bp from the transcription start site (TSS), 45% of the nucleotides belong to transposable elements. This fraction is the same as the overall density in the entire genome – see Fig <xref ref-type="fig" rid="F1">1</xref>. In contrast, in the region closer to the TSS, between 0 and 2000 bp – the density of transposable elements is lower. The Alu density exhibits a more interesting behaviour; for short distances – between 0 and 600 bp – it is lower than the genomic average of about 10%. This is not surprising since the immediate proximal region to the TSS is rich with important binding sites for initiating factors and insertion of Alu and other transposable elements in this region is most probably selected against. The density increases steadily till about 2000 bp, where it reaches a level of 18% of the base pairs [Fig. <xref ref-type="fig" rid="F1">1</xref>], after which it maintains a steady level (and has to decrease to the genomic average as we move further away into the intergenic region). The Alu content in the deep intergenic spacer regions is less than 10%; hence the density of on-Alu BSs in these regions is also lower than in the upstream 5000 bp regions of genes. The presences of less BSs in these regions might imply less functional relevance. Finally, 85% of the genes in our database contained at least one Alu insertion in the 5 Kbp upstream region.</p></sec><sec><title>Non random distribution of Alu elements on promoter regions of different gene groups</title><p>We analyzed the distribution of Alu elements in the 5000 bp upstream regions of genes, treating separately groups of genes that belong to different biological processes. Protein biosynthesis genes are enriched with Alu compared to the genome-wide average value of 3.63 copies in the 5000 bp long upstream regions (Table <xref ref-type="table" rid="T5">5</xref>). In contrast, genes that are related to various kinds of developmental processes, tissue formation (such as cell communication, central nervous system development genes and muscle development genes) contain a smaller number of Alu elements in their 5000 bp upstream region than the average value (see Table <xref ref-type="table" rid="T5">5</xref>). Many of these differences between the Alu content of genes of a particular biological process and the Alu content of the entire genome (in the 5 Kbp upstream regions) were found (by t-test) to be highly significant (Table <xref ref-type="table" rid="T5">5</xref>). As mentioned above, similar differences in Alu content were reported [<xref ref-type="bibr" rid="B12">12</xref>] when the full length gene sequences on chromosomes 21 and 22 were analyzed.</p></sec><sec><title>Studying BS content in specific (non-consensus) Alu elements in the promoter regions</title><p>Hundreds of thousands of putative BSs reside on Alu elements in the upstream regions of genes. For several TFs one observes that nearly all the putative BSs (more than 90%) are located on Alu. The search for BMs presented in the previous sections was done on the consensus sequences of various Alu subfamilies. Since, however, Alu elements have mutated after insertion, there is considerable deviation from the consensus in the specific Alu sequences that are found across the genome. This variation may change the BS content of specific individual Alu elements versus the consensus. Hence we repeated our search of BSs in the actual 5000 bp upstream regions of about 14,000 genes, and if a BS passed a significance filter, we checked whether it resides on a specific Alu. This search was limited to those 66 BMs that passed our filter based on the consensus sequences, i.e. their largest score, Tmax, exceeded T5 for the consensus of at least one of the eight Alu subfamilies studied. For each of these 66 BMs we searched the 14,000 upstream regions for putative BMs whose score exceeds (the consensus-based) Tmax.</p><p>We found that for 29 BMs out of the 66 the majority of putative BSs that pass Tmax reside on Alu sequences; the relative on and off Alu BM content is presented, for 19 out of these 66, in Fig <xref ref-type="fig" rid="F2">2</xref>. For example, for 5 out the 6 nuclear factors (Table <xref ref-type="table" rid="T4">4</xref>) whose BMs appear among the 66, between 66% and 99% of the BMs (found in the 5000 bp upstream regions) are on Alu (see Fig. <xref ref-type="fig" rid="F2">2</xref>). The highest over-representations in Alu were observed for the BS of PITX2; over 99% of the 34000 BSs that pass the threshold were on Alu, and for LXRE's BSs over 99% of it's 9000 BSs that pass the threshold were on Alu, (see Fig. <xref ref-type="fig" rid="F2">2</xref>). In general, the distributions of putative BSs of TFs that bind to TGTAATCCC (Table <xref ref-type="table" rid="T1">1</xref>) exhibit remarkable differences between the number of putative BS on and off Alu; over 90% of the total number of putative BSs of these TFs are located within Alu sequences. More moderate differences were found for developmental TFs such as SIX3.01, PAX4.01, NRL.01, MEF2.04 and NKX25.01 (Fig. <xref ref-type="fig" rid="F2">2</xref>). In general, for every BM that passed FDR of Q = 0.2 (see Table <xref ref-type="table" rid="T1">1</xref>) over 80% of the BSs on the 5 kbp upstream region are located on Alu.</p><p>On the other hand, for the 37 (out of 66) remaining PSSMs the majority of BSs were not located on Alu, like IRF1.01 (48% out of the total number of 4000 BS were located on Alu) and HNF1.03 (27%). Still, even for these TFs, with mainly off-Alu BMs, the Alu repeats provide between hundreds to thousands of putative BSs in the 5 kbp upstream regions (see Table <xref ref-type="table" rid="T6">6</xref>).</p><p>As mentioned above, in the 500 bp upstream region, which constitutes the proximal promoter [<xref ref-type="bibr" rid="B46">46</xref>] and hence is believed to be most important for regulation, the Alu density is below the genome average (Fig. <xref ref-type="fig" rid="F1">1</xref>). To study this region in more detail, we repeated the analysis that was done for the 5000 bp region, and compared the distributions of BSs on and off Alu in the first 500 bp upstream to the TSS (see Fig <xref ref-type="fig" rid="F3">3</xref> for 19 representative TFs). Only 9 out of the 66 BMs had more on Alu than off Alu BMs. These BMs include almost all homeodomain proteins, LXRE.01, LYF-1 and PAX4.01 (but not SIX3, see Fig. <xref ref-type="fig" rid="F3">3</xref>). In spite of the relatively low Alu density, on-Alu BMs do constitute a sizeable fraction of the putative BMs that are found in the 500 bp long upstream region (Fig. <xref ref-type="fig" rid="F3">3</xref> and Table <xref ref-type="table" rid="T6">6</xref>).</p></sec><sec><title>Genes of different biological processes contain different amounts of BSs as a result of differences in their Alu content</title><p>We used the results of the search for on and off Alu BMs to look for differences between genes that belong to different biological processes. As expected, for those biological processes whose genes were previously found to be enriched with Alu (Table <xref ref-type="table" rid="T6">6</xref>), we also found that a high percentage of their regulatory signals were on-Alu. The Alu rich groups in this category include metabolic and bio-synthesis associated genes. Since for several TFs a large majority of BMs reside on Alu, we find that Alu-rich functional families have a large number of these Alu-associated BMs, whereas Alu-poor biological processes (such as developmental genes) will have less of these BMs (see Fig. <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="supplementary-material" rid="S2">Additional data file 2</xref>). For example, the average number of putative BSs of PITX2 is almost 4 times greater in protein biosynthetic genes than in CNS developmental genes (Fig. <xref ref-type="fig" rid="F4">4</xref>). Interestingly, as a result, developmental TFs such as LYF-1, PAX4, MEF2D and PITX2, that have on-Alu BMs, appear less in the (Alu-poor) upstream region of developmental genes than in the Alu-rich non-developmental group of protein biosynthesis related genes. One possible explanation of this may be that the developmental TFs main role is to suppress pathways that are in competition with or exclusive of the developmental process (versus acting directly on development – associated genes).</p><p>One should emphasize that the statements made above are based mainly on in-silico evidence, which must be supplemented by firm experimental observations of binding and functional roles of the bound TFs.</p></sec><sec><title>Inter-species comparisons: B1, the murine 'Alu like' elements, contain several BSs that are similar to the primate Alu BSs</title><p>We scanned the B1 consensus sequences that were downloaded from RepBase [<xref ref-type="bibr" rid="B29">29</xref>]. These sequences were less enriched with BSs that pass the T5 threshold; this might be due to the fact that the length of the B1 sequences is only half that of Alu. Twenty three binding motifs passed T5 on B1 (see Table <xref ref-type="table" rid="T7">7</xref>) and we found that 11 of these passed T5 also in Alu, as expected from the similarity between B1 and Alu [<xref ref-type="bibr" rid="B47">47</xref>] (bold in Table <xref ref-type="table" rid="T7">7</xref>). One of these shared BSs was T(G/T)TAATCCC where G appears in all Alu families and T appears in the consensus of B1s. This BS is the target of OTX2.01, a homeodomain protein in both human and mouse. An additional homeodomain protein, CRX01.01, is also bound to this BS in mouse; it did not pass T5 in Alu. The members of the MEF2 family of BMs appear in mouse (see Table <xref ref-type="table" rid="T7">7</xref>) with a weaker score than in Alu, (unpublished data). The PSSM with the most significant score on B1 was PAX6.02, which is not surprising since this PSSM is built from those BSs of PAX6 that were known to reside on B1 [<xref ref-type="bibr" rid="B48">48</xref>]. In addition, a group of BMs that are related to nuclear factors and response proteins passed T5 on B1 (TTF1.01 and ERR_01) but not on Alu. This is surprising since there are experiments that validated that the TFs that correspond to these BMs bind to primate Alu [<xref ref-type="bibr" rid="B16">16</xref>].</p></sec><sec><title>The distributions of B1 in mouse promoters and Alu in human promoters are positively correlated</title><p>When the mouse genome was first sequenced, regions orthologous to human DNA were identified. The densities of B1 and B2 in mouse and Alu in human were measured in windows of 500,000 bp and the Alu density in humans was found to be correlated with B2 and B1 density in mouse [<xref ref-type="bibr" rid="B49">49</xref>]. A high correlation was quite unexpected, since Alu and the mouse SINE elements B1 and B2 spread in the rodent and human genomes <italic>after </italic>the divergence of rodents and primates. We checked whether this trend is present also in the regulatory regions of genes, of size 5 kbp, and we found a positive correlation between B1 and ALU densities even in this relatively short region. In more than 5400 pairs of orthologous genes, that had available upstream sequences in both human and mouse, we found positive correlation (see methods) of 0.495 (pval<1e-05). In addition we ordered the genes by increasing Alu content in the 5000 upstream bp region, and divided the genes on this basis into 27 sets, with each set containing 200 genes. We then computed the average Alu and B1 associated nucleotide densities for each of the 27 sets, and plotted the two average Alu/B1 contents (Fig <xref ref-type="fig" rid="F5">5</xref>). Each point in Fig <xref ref-type="fig" rid="F5">5</xref> represents a group of 200 genes, with the x (y) coordinates representing the Alu (B1) densities in human (mouse). The monotonic behaviour evident in Fig <xref ref-type="fig" rid="F5">5</xref> indicates that the Alu and B1 contents of orthologous genes are largely similar in human and mouse. This similarity is far from being obvious, since spreading of Alu and B1 across the genome is believed to have happened <italic>after </italic>the divergence of primates and rodents.</p></sec></sec><sec><title>Conclusion</title><sec><title>Summary</title><p>This research suggests that evolution used transposable elements to insert modules of transcription factor binding motifs into promoters and, by means of their presence, assemble higher level regulatory networks. In order to explore this question we focused on Alu elements, which are good potential candidates to be part of the building blocks of regulatory networks for two reasons. First, Alu elements are abundant in the upstream region of the TSS of genes, and second, Alu elements contain dozens of putative BSs for TFs. Some of these BSs were found before and their association with Alu was also reported, whereas in some cases although the BSs were found, the fact that they reside on Alu went unnoticed. Finally, we list here also BSs on Alu that were not identified previously. Our findings imply that the biological pathway on which Alu-mediated regulation appears to have the most significant impact is the development process. Many of the TFs that have binding motifs on Alu are associated with development; moreover, some of these BSs were previously demonstrated to be functional <italic>in vivo </italic>and essential to regulation of some target genes.</p><p>We have also shown that a few subsequences on Alu, that were known to be highly conserved, provide many of the binding sites for the transcription factors. We studied the Alu distribution in the upstream region of genes and found that while Alu are "repelled" from the 500 bp long region immediately preceding the transcription starts site, the 4500 bp long sequence further upstream is enriched (compared to the entire genome). The Alu content in the promoters varies between functional families of genes.</p><p>Comparison of Alu in human and B1 in mice shows a significant correlation between their distributions (with respect to orthologs), and high similarity between the binding motifs that reside on the two mobile elements.</p><p>Several particular cases (TFs and genes) were discussed in detail.</p></sec><sec><title>Future directions</title><p>Most importantly, more experiments are needed to verify that the BMs that were found by us (see Table <xref ref-type="table" rid="T1">1</xref> and <xref ref-type="supplementary-material" rid="S1">Additional data file 1</xref>) indeed bind their corresponding TFs and, furthermore, to verify their regulatory roles. Of particular interest is the role of Alu in developmental processes. We found that promoters of genes involved in biosynthesis and metabolism are enriched (via Alu) in BMs of several developmental TFs (Table <xref ref-type="table" rid="T4">4</xref>). It is our belief that since development and cell proliferation (of which biosynthesis is an essential component) are opposing processes [<xref ref-type="bibr" rid="B50">50</xref>], these BMs play an inhibitory role, suppressing proliferation during differentiation. This hypothesis is supported by the known suppressory role of GFI1, but the presently known activities on PITX2 and NKX2.5 do not fit this picture. Further experiments that address this issue by measuring the regulatory effects of other Alu-associated developmental TFs are needed. Clearly, many of the genes and their functions date earlier than the rodent/primate divergence, whereas the Alu insertions happened later. This poses an open and debated [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B51">51</xref>] question about the mechanism by which pre-existing regulatory circuits must have been replaced, modified or supplemented by Alu insertions.</p></sec></sec><sec sec-type="methods"><title>Methods</title><sec><title>Downloads</title><p>The Alu consensus sequences were downloaded from Repbase [<xref ref-type="bibr" rid="B29">29</xref>].</p><p>To scan for BMs we downloaded 410 vertebrate PSSMs from the MathInspector [<xref ref-type="bibr" rid="B30">30</xref>] database, with two additional PSSMs ($PITX2_Q2 and &LUN1_01) taken from the TRANSFAC database [<xref ref-type="bibr" rid="B31">31</xref>].</p><p>The coordinates of the upstream regions of 21000 human genes were taken from [<xref ref-type="bibr" rid="B52">52</xref>]. The annotations of the U133 Array (Affymetrix, Santa Clara, CA) were used in order to remove multiple upstream regions for different splice variants of the same gene, (a single upstream region was associated with each gene symbol). We also removed from our database those Refseq genes that have duplicates on chromosomes X and Y, and those with upstream regions that were not fully sequenced. All this left 13686 promoters of different Refseq genes for analysis.</p><p>The coordinates of Alu elements and other transposable elements in the human genome were extracted from the file chromOut.zip, which was downloaded from [<xref ref-type="bibr" rid="B52">52</xref>]. This is the output file of RepeatMasker [<xref ref-type="bibr" rid="B53">53</xref>]. The sequences of the Alu that reside in promoters were then extracted from the human genome, by intersecting these Alu coordinates with the upstream sequences of our 13686 genes.</p><p>Similarly, for mouse we downloaded from [<xref ref-type="bibr" rid="B54">54</xref>] the files upstream5000.zip which contained the coordinates and sequences of the regions upstream to the TSS of mouse genes, and the RepeatMasker output file of the mouse genome, chromOut.zip. We used annotations of Mouse 430_2 Array (Affymetrix, Santa Clara, CA) in order to remove multiple promoters.</p><p>In order to annotate genes to biological processes and molecular functions we used the DAVID database (based on the GO annotation) [<xref ref-type="bibr" rid="B55">55</xref>].</p></sec><sec><title>Calculating scores and probabilities</title><p>For each PSSM we calculated the maximal score, on both sense and antisense orientations.</p><p>In order to find the probability of getting some maximal score for a BM on an Alu sequence we produced 100,000 random 281 bp long sequences, of di-nucleotide distribution similar to that of the AluSx consensus sequence. The next step was to calculate the maximal score for a given PSSM for each random sequence. The p-value for each given score was the percent of random sequences that contained a target site with higher score. Although, such procedure might seem to be biased towards AluSx, it can be seen in Tables <xref ref-type="table" rid="T8">8</xref> and <xref ref-type="table" rid="T9">9</xref> that the subfamilies on which we focus in this research have similar nucleotide and di- nucleotide contents. Note that since we did not use a twice repeated random sequence of 140 nucleotides, our p-values are an upper bound to the true ones.</p></sec><sec><title>Orthologous genes in mouse and human</title><p>We compared gene symbols of human and mouse and got about 5400 orthologous gene pairs and their respective promoters in human and mouse. The Pearson's correlation coffeeicient between the Alu and B1 densities on human and mouse genes was calculated over these 5400 pairs of promoters.</p></sec></sec><sec><title>Authors' contributions</title><p>PP participated in the conception of the study, carried out the mapping of the Alu elements and BSs in the upstream of genes, analyzed the data, wrote the initial draft of the manuscript. ED conceived of the study, and participated in its design coordination and supervision, helped to draft and edited the manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Sixty six binding motifs whose scores pass T5 on at least one major Alu subfamily sequence. The consensus sites of the PSSMs (from the IUPAC convention [<xref ref-type="bibr" rid="B33">33</xref>]) is in the second column and the the target sequences with the highest scores among all the major Alu consensus sequences – in the third, with nucleotides that agree with the consensus denoted by capital letters. The fourth column contains the locations of the putative target sites on the Alu sequences of the subfamily with the highest score (in case the same sequence appears on several Alu elements, we choose the one with the highest number of copies in the 5 kb upstream regions). Two Alu subsequences serve as putative target sites of several TFs (designated by *, and **). The fifth column contains the p-values (see text and methods) and the number of subfamilies on which the BSs of the third column resides is listed in column 6.</p></caption><media xlink:href="1471-2164-7-133-S1.xls" mimetype="application" mime-subtype="vnd.ms-excel"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p>We present for 66 PSSMs the number of putative BSs in the 5 Kbp region upstream, averaged over the genes that belong to various biological processes. The PSSMs that are shown are those for which Tmax>T5. The number of genes in each biological process is given in column 2. The Alu density and number of Alu repeats per gene are given in columns 3 and 4 respectively.</p></caption><media xlink:href="1471-2164-7-133-S2.xls" mimetype="application" mime-subtype="vnd.ms-excel"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Operon information improves gene expression estimation for cDNA microarrays
|
<sec><title>Background</title><p>In prokaryotic genomes, genes are organized in operons, and the genes within an operon tend to have similar levels of expression. Because of co-transcription of genes within an operon, borrowing information from other genes within the same operon can improve the estimation of relative transcript levels; the estimation of relative levels of transcript abundances is one of the most challenging tasks in experimental genomics due to the high noise level in microarray data. Therefore, techniques that can improve such estimations, and moreover are based on sound biological premises, are expected to benefit the field of microarray data analysis</p></sec><sec><title>Results</title><p>In this paper, we propose a hierarchical Bayesian model, which relies on borrowing information from other genes within the same operon, to improve the estimation of gene expression levels and, hence, the detection of differentially expressed genes. The simulation studies and the analysis of experiential data demonstrated that the proposed method outperformed other techniques that are routinely used to estimate transcript levels and detect differentially expressed genes, including the sample mean and SAM t statistics. The improvement became more significant as the noise level in microarray data increases.</p></sec><sec><title>Conclusion</title><p>By borrowing information about transcriptional activity of genes within classified operons, we improved the estimation of gene expression levels and the detection of differentially expressed genes.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Xiao</surname><given-names>Guanghua</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Martinez-Vaz</surname><given-names>Betsy</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Pan</surname><given-names>Wei</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Khodursky</surname><given-names>Arkady B</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Genomics
|
<sec><title>Background</title><p>Genome-wide monitoring of transcription by means of DNA microarrays is used to infer transcriptional and regulatory networks in living organisms. In most of microarray experiments, transcript levels of thousands of genes are measured with a relatively small number of replications, so the estimates of true expression levels from microarray data may be poor, mostly due to a small sample size. To address this problem, several statistical methods have been proposed to borrow information from other genes to improve detection of the differentially expressed ones [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. The main idea is to borrow information from other genes to estimate either the distributions of genes's expression levels or the distribution of error terms. The underlying assumption is that it should be possible to improve the estimates of expression levels of genes by borrowing information about transcriptional activity across the sets of genes that are biologically, or physically, related. In some cases, the expression levels may significantly vary across the genes, then borrowing information from unrelated genes may not improve, or even worsen, the estimates of gene expression levels [<xref ref-type="bibr" rid="B10">10</xref>]. However, if, based on biological knowledge, we can expect that some genes are more likely to express at similar levels (i.e. co-express), then we can improve the inference by using information about the activity of those genes.</p><p>An operon [<xref ref-type="bibr" rid="B11">11</xref>] is a set of linearly juxtaposed genes transcribed as a single mRNA; operons are commonly found in prokaryotic genomes such as <italic>Escherichia coli</italic>. Transcription of operons of <italic>E. coli </italic>has been examined, and operons have been predicted in many studies [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B19">19</xref>], which provides background information about the <italic>E. coli </italic>regulatory network. The genes within the same operon usually have similar expression levels, hence show some local structure in expression profiles [<xref ref-type="bibr" rid="B20">20</xref>], and this fact has been successfully used in operon prediction [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>].</p><p>Based on the existing information about the structure of operons, we propose a hierarchical Bayesian model which improves gene expression estimation by borrowing information from genes within the same operon. Most existing methods for detecting differently expressed genes [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B9">9</xref>] borrow information from other genes in the whole genome, while our proposed method only borrows information from other genes within the same operon, which has sound biological basis. Wren <italic>et al </italic>[<xref ref-type="bibr" rid="B23">23</xref>] have proposed a simulated annealing approach to adjust gene expression data by using existing microarray measurements obtained on the same organism, which effectively reduced the noise and made it possible to compare different microarray experiments. But their method relies on reference microarray experiments that cover the dynamic range of transcript abundances for most of the genes, which may be difficult to select or unavailable. Instead of using existing microarray measurements, we use existing information about the operons' structure to reduce the noise in microarray data.</p><p>A more accurate estimation of transcript abundances of individual genes will improve our ability to evaluate transcriptional activity on a genome-wide scale, and hence facilitate the exploration of gene regulatory networks. Our proposed method provides a better way to estimate relative transcript levels, which is critical for distinguishing differentially expressed (DE) genes from equally expressed (EE) genes. Herein, we refer to the logarithm of a ratio of the fluorescent intensities of the test and control samples as the observed gene expression level. The genes with estimated expression levels significantly different from zero are identified as DE genes, otherwise as EE genes.</p><p>Using more than 200 microarray experiments, we obtained the evidence of co-transcription of genes within <italic>E. coli </italic>operons on a genome-wide scale. We applied the proposed method to three simulated and one experimentally obtained data sets. The simulation studies and the real data application demonstrated that the proposed method performed better than the sample mean and the SAM t statistics in estimating the gene expression levels as well as in detecting differentially expressed genes. The improvement became more significant as the noise level in microarray data increased.</p></sec><sec><title>Results</title><sec><title>Simulation study</title><p>We carried out three simulations, with similar settings and the noise level gradually increasing from simulation 1 to 3. In simulations, we assumed that the genes within an operon are co-transcribed. Since the true expression levels in a simulated data set were known, we could calculate the mean squared errors of the estimated expression levels (summarized in Table <xref ref-type="table" rid="T1">1</xref>). Without incorporating the information from operons, the sample mean of the observed expression levels would be a natural estimator of a gene's expression level. The mean squared errors of the two estimates, the posterior mean from the proposed model and the sample mean, are shown in Table <xref ref-type="table" rid="T1">1</xref> for comparison. It can be easily seen that the incorporation of operon information leads to a better estimate, with a smaller mean squared error, of expression levels. When the noise level increased, the improvement from incorporating operon information also increased.</p><p>To evaluate the performance of the proposed method in detecting DE genes, the estimated expression level of each gene was used to rank the genes, and the highly ranked genes were identified as DE genes. Since the identities of DE genes in the simulation studies were known, we compared the performances of the proposed method, sample mean, and SAM t statistics in detecting DE genes using receiver operating-characteristic (ROC) curves (Figure <xref ref-type="fig" rid="F1">1</xref>). In a ROC curve, the <italic>sensitivity </italic>is plotted against 1 – <italic>specificity</italic>. The sensitivity is denned as a fraction of true DE genes being correctly detected and the specificity is a fraction of the true EE genes being correctly identified. The ROC curves in Figure <xref ref-type="fig" rid="F1">1</xref> demonstrate that the performance of the sample mean and of the SAM t statistic were very close, and our hierarchical model, incorporating operon information, outperformed both of them. The difference in the performance became greater as the noise level increased. For example, as the specificity equals to 0.8, the sensitivities of the methods using the sample mean or SAM t were about 0.91, 0.80, and 0.68 for simulations 1,2, and 3, respectively, while the sensitivities of the proposed method were 0.95, 0.89 and 0.83, respectively.</p></sec><sec><title>Application to E. coli data</title><p>To verify the assumption that the genes organized in operons are co-expressed, we pooled together data from 217 microarray experiments, obtained in 53 conditions [<xref ref-type="bibr" rid="B24">24</xref>]. The distribution of pairwise correlations between expression profiles of genes in operons was greatly skewed towards positive values, with the mean correlation of 0.62 (Figure <xref ref-type="fig" rid="F2">2A</xref>). Unlike the profiles of genes organized in operons, expression profiles of randomly picked pairs of genes were not correlated; the corresponding distribution of correlation coefficients was almost symmetric around 0, with the mean correlation of 0.012 (Figure <xref ref-type="fig" rid="F2">2B</xref>). This result demonstrated the similarity of transcriptional activity of genes within operons and served as a motivation for borrowing information from other genes within the same operon.</p><p>The proposed method [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>] was used to analyze differential transcriptional activity in an <italic>E. coli </italic>mutant lacking the <italic>flhDC </italic>gene, a master regulator of transcription of genes whose products mediate bacterial motility and chemotaxis [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>]. The genes were ranked by their estimated expression levels, i.e. their posterior means of <italic>μ<sub><italic>i </italic></sub></italic>obtained from the proposed model. For the sake of comparison, the sample mean and SAM t statistics were also used to rank the genes [see <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>]. Using the functional annotation from Macnab [<xref ref-type="bibr" rid="B25">25</xref>] as a standard, we obtained the number of false positives at different cut-off levels (total positives). The comparison revealed that ranking genes by the proposed method produced fewer false positives than the ranking based on the SAM t or sample mean statistics (Figure <xref ref-type="fig" rid="F3">3</xref>).</p><p>To find a reasonable cutoff value for differently expressed genes, we calculated the false discovery rate (FDR) [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B27">27</xref>]based on the posterior probability [<xref ref-type="bibr" rid="B7">7</xref>]. In this experiment, we also estimated the FDR by using the functional annotation from Macnab [<xref ref-type="bibr" rid="B25">25</xref>] as a reference. Comparison of the estimated False Discovery Rates revealed that the estimated FDR from the posterior probability was a little lower than that derived from the annotation, which could be due to the partial incompleteness of the reference (Fig. <xref ref-type="fig" rid="F4">4</xref>). Overall, the estimated FDR from the posterior probability is close to the FDR using the reference list of genes, indicating that our method for estimating the FDR is adequate. We set the cutoff for the FDR to be 0.01, which identified the top 44 genes as DE genes. At such a cutoff, the estimated number of false negatives is 14 and the estimated false negative rate is about 0.003 (see the "Methods" section for details). The top 44 genes are listed in Table <xref ref-type="table" rid="T2">2</xref>. Note that, in Table <xref ref-type="table" rid="T2">2</xref>, the gene expression level is on the log scale and the FDR corresponds to a specific number of DE genes and not to each individual gene itself. According to Macnab's classification [<xref ref-type="bibr" rid="B25">25</xref>], 41 genes out of the 44 were expected to be differentially expressed in the <italic>flhDC- </italic>dependent manner, whereas the lists of 44 genes identified by using SAM t and sample mean contained only 36 and 38 expected genes, respectively.</p><p>We examined some genes and operons in more detail, to demonstrate the advantages of borrowing information from within an operon. For example, an operon <italic>argT-hisJQMP </italic>contains 5 genes (argT, hisJ, hisM, hisP, hisQ) and is not expected to be differentially expressed under the examined experimental condition, according to our biological knowledge [<xref ref-type="bibr" rid="B25">25</xref>]. But from the microarray data, the mean expression level (an average log ratio) of the gene <italic>argT was </italic>-2.73, which ranked 15th among all the expression levels in the <italic>E. coli </italic>genome. This "high expression" of the <italic>argT </italic>could possibly be caused by random noise in microarray measurements and/or in biological samples, since the mean expression levels of other genes within the same operon were close to 0 (those for genes hisP, hisM, hisQ, hisJ were 0.00, -0.05, 0.03, and 0.05, respectively). However, accounting for expression levels of other genes within the same operon lowered the estimate of the expression level (posterior mean of <italic>μ<sub><italic>i </italic></sub></italic>of the <italic>argT to </italic>-0.02, rank of 1456, indicating that this gene was not differentially expressed. Analysis of another operon, <italic>fliDST</italic>, illustrates a complimentary case. Transcription of the <italic>fliDST operon </italic>(containing 3 genes, <italic>fliD, fliS, fliT</italic>) is known to be controlled by the FlhDC [<xref ref-type="bibr" rid="B25">25</xref>], and thus, under the experimental condition, differential expression of genes in that operon would be expected. While the expression level of the <italic>fliT</italic>, estimated by the sample mean at -0.90, ranked only 65th, the estimated expression level of the gene after borrowing information from two other genes in the operon was -3.25, which ranked 19th.</p><p>In general, through borrowing information, our Bayesian method worked in a way giving more consistent estimates of the expression levels for the genes of the same operon. For example, compared with using the sample mean to estimate expression levels, the Bayesian method tended to yield smaller standard deviations of the expression estimates for within-operon genes (see Figure <xref ref-type="fig" rid="F5">5</xref>).</p></sec></sec><sec><title>Discussion</title><p>In this paper, we proposed and applied a hierarchical Bayesian model, to estimate relative gene expression levels and detect differentially expressed genes by borrowing expression information within operons. The performance of the proposed method was compared with that of the sample mean and SAM t statistics. Through the simulation studies, we showed that the proposed method outperformed the sample mean and the SAM t statistics in estimating gene expression levels and detecting DE genes. The proposed method was used to analyze differential expression in an <italic>E. coli </italic>mutant with a defect in transcription of motility/chemotaxis genes, giving results more consistent with the existing biological knowledge than those obtained by using the other statistics.</p><p>A major advantage of the proposed approach is in borrowing expression information from other genes within the same operon. The approach is developed within a statistically sound Bayesian model and it offers necessary flexibility with respect to the amount of information that needs to be borrowed from other genes. By borrowing information we can obtain stabilized estimates of expression levels from rather noisy microarray data. As a result, the estimates of transcript levels within the same operon become more similar to each other, more so than without borrowing information; this is consistent with a biological fact that genes within the same operon are transcribed as a single mRNA molecule. With the proposed method, the estimated expression levels of genes in "differentially expressed" operons are consistently high, and more importantly, transcript abundances of genes in "equally expressed" operons are stabilized towards zero. In the experimentally obtained microarray data, the within operon variation was smaller than the variation among the replicate data points for the same genes, indicating that the expression levels of the genes within an operon were very similar.</p><p>In our model, the ratio of parameters <italic>τ</italic><sup>2 </sup>and <inline-graphic xlink:href="1471-2164-7-87-i1.gif"/> determines how much information comes from the observed expression of a gene and how much comes from the average expression of an operon, when estimating the expression level of an individual gene within an operon. A smaller <italic>τ</italic>, as compared to <inline-graphic xlink:href="1471-2164-7-87-i1.gif"/>, puts more weight on the average expression of an operon. Here we assumed the same <italic>τ </italic>value for all operons, implying that all operons had similar within-the-operon variability. However, this assumption might not be realistic. In some operons and physiological conditions, the genes might express very similarly, but in others, especially under the control of internal promoters, transcription of individual genes may be more heterogeneous. In the future, we will investigate the effect of an operon-specific <italic>τ</italic>. Operonal organization of genes is common in prokaryotes and also present in some eukaryotic organisms [<xref ref-type="bibr" rid="B28">28</xref>-<xref ref-type="bibr" rid="B30">30</xref>], and the proposed method can be extended to biological systems where the operonal structure is unknown. Many biological studies have demonstrated that co-expressed genes tend to cluster on the chromosome. Although the nature of this phenomenon is not quite understood, a positional clustering of co-expressed genes can be found in many eukaryotes including yeast [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>], worm [<xref ref-type="bibr" rid="B33">33</xref>], fly [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B35">35</xref>], mouse [<xref ref-type="bibr" rid="B36">36</xref>], and human [<xref ref-type="bibr" rid="B37">37</xref>-<xref ref-type="bibr" rid="B39">39</xref>]. These findings indicate that the genes are likely to co-transcribe with their chromosomal neighbors. In those cases, instead of borrowing information from genes in the same operon, we can borrow information from gene neighbors on the chromosome. Another extension of our method would involve incorporation of gene annotation information into the analysis of expression data. The approach would be very similar to the one described in this paper: based on biological knowledge, the genes belonging to the same functional group are more likely to be co-expressed, so we can use a hierarchical model to borrow information from and for the genes within the same functional group to improve the estimates of gene expression levels.</p></sec><sec><title>Conclusion</title><p>The information about operon structure leads to a better estimation of gene expression levels. Using simulated and experimental data sets, we have demonstrated that the proposed method performs better than the sample mean and the SAM t statistics in estimating the relative levels of transcript abundances and detecting differentially expressed genes.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>RegulonDB database and E. coli. microarray data</title><p>RegulonDB [<xref ref-type="bibr" rid="B40">40</xref>] is a database containing information about known operons in <italic>E. coli</italic>. According to the RegulonDB annotations, 1486 genes (about one third of all genes predicted in the <italic>E. coli </italic>genome) are organized in 600 operons.</p><p>The <italic>E. coli </italic>data set contains results of 217 microarrays collected in 53 different experimental conditions. The fluorescent intensities of the test and control samples were measured, and the average log ratio of the intensities for each gene under the same condition was used here to represent an observed gene expression level under that condition [<xref ref-type="bibr" rid="B24">24</xref>].</p><p>An <italic>E. coli </italic>motility expression data set ([<xref ref-type="bibr" rid="B41">41</xref>] series accession number: GPL2101) was obtained in a direct pair-wise comparison between a knock-out mutant of the <italic>flhDC</italic>, a master regulator of the motility/chemotaxis regulon [<xref ref-type="bibr" rid="B25">25</xref>], and its isogenic wild type strain. Total RNA samples of a mutant <italic>E. coli </italic>(test samples) and an isogenic wild type <italic>E. coli </italic>(control samples) were labeled with red (Cy5) and green (Cy3) fluorophors. The intensities from the red and the green channels were normalized by the lowess method [<xref ref-type="bibr" rid="B42">42</xref>]. There were 4281 genes (G = 4281) with four replicates for each gene (n = 4). Let <italic>Y</italic><sub><italic>i</italic>,<italic>j </italic></sub>be defined as the log ratio of the intensities between the test and control samples for gene <italic>i </italic>on array <italic>j; </italic>that is,</p><p><inline-graphic xlink:href="1471-2164-7-87-i2.gif"/></p></sec><sec><title>Hierarchical models</title><p>We propose a hierarchical Bayesian model,</p><p><inline-graphic xlink:href="1471-2164-7-87-i3.gif"/></p><p>where, <italic>Y</italic><sub><italic>i</italic>,<italic>j </italic></sub>is the log ratio of gene <italic>i </italic>in replicate <italic>j, μ<sub><italic>i </italic></sub></italic>and <italic>σ</italic><sub><italic>i </italic></sub>are the true expression level and the standard deviation, respectively. In our method, the posterior mean of <italic>μ<sub><italic>i </italic></sub></italic>is used as the estimated expression level of gene <italic>i</italic>, while the sample mean, <inline-graphic xlink:href="1471-2164-7-87-i4.gif"/>., is referred to as the observed expression level.</p><p>As prior knowledge, we assume that if several genes belong to the same operon, in accordance with the RegulonDB annotation, then their expression levels are from a normal distribution, with the mean <italic>λ</italic><sub><italic>p </italic></sub>and the variance <italic>τ</italic><sup>2</sup>. Specifically,</p><p><inline-graphic xlink:href="1471-2164-7-87-i5.gif"/></p><p>where <italic>O</italic><sub><italic>p </italic></sub>denotes operon <italic>p</italic>. <italic>λ</italic><sub><italic>p </italic></sub>represents the expression level of the operon <italic>p</italic>, which is the average of the mean expression levels of all genes within the operon <italic>p</italic>. <italic>τ</italic><sup>2 </sup>is the within operon variation, and is assumed to be the same across all operons. A non-informative prior is assigned to <italic>λ</italic><sub><italic>p</italic></sub>, that is <italic>Pr</italic>(<italic>λ</italic><sub><italic>p</italic></sub>) α 1, to reflect the lack of prior information, <inline-graphic xlink:href="1471-2164-7-87-i1.gif"/> and <italic>τ</italic><sup>2 </sup>have vague priors, which are inverse Gamma distributions with the shape and rate parameters equal to 0.01 and 0.01 respectively [<xref ref-type="bibr" rid="B43">43</xref>]. If gene <italic>i </italic>is not in any operon, then</p><p><inline-graphic xlink:href="1471-2164-7-87-i6.gif"/></p><p>so the posterior mean of <italic>μ<sub><italic>i </italic></sub></italic>is just the the sample mean <inline-graphic xlink:href="1471-2164-7-87-i4.gif"/>; if gene <italic>i </italic>is in operon <italic>p</italic>, then the conditional distribution of <italic>μ<sub><italic>i </italic></sub></italic>can be derived as:</p><p><inline-graphic xlink:href="1471-2164-7-87-i7.gif"/></p><p>where</p><p><inline-graphic xlink:href="1471-2164-7-87-i8.gif"/></p><p>Equation (3) shows that, when borrowing information from the other genes within the same operon, the estimated expression level of the gene <italic>i </italic>becomes the weighted average of the observed expression level of gene <italic>i </italic>and the expression level of operon <italic>p</italic>, given that gene <italic>i </italic>belongs to operon <italic>p</italic>. The weights are inversely proportional to the variances. In this model, a key concept is to shrink the observed expression level <inline-graphic xlink:href="1471-2164-7-87-i4.gif"/>, towards <italic>λ</italic><sub><italic>p</italic></sub>, the expression level of an operon, based on the knowledge of the operon structure. The degree of shrinkage is determined by the variability of <inline-graphic xlink:href="1471-2164-7-87-i4.gif"/>. and <italic>λ</italic><sub><italic>p</italic></sub>. Without incorporating operon information, the estimated expression level would be close to the observed expression level, <inline-graphic xlink:href="1471-2164-7-87-i4.gif"/> In the hierarchical model, <italic>λ</italic><sub><italic>p </italic></sub>represents the expression level of operon <italic>p</italic>, and</p><p><inline-graphic xlink:href="1471-2164-7-87-i9.gif"/></p><p>where, <italic>m</italic><sub><italic>p </italic></sub>is the number of genes in operon <italic>p</italic>, and <italic>i </italic>∈ <italic>O </italic><sub><italic>p </italic></sub>denotes that gene <italic>t </italic>is in operon <italic>p</italic>. Although the posterior distribution was not available in a closed form, we could derive a closed form of the full conditional distribution, and used Markov chain Monte Carlo (MCMC) to simulate the parameters from the posterior distribution. With this closed form expression, the model could be easily coded in R [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>] for MCMC simulation using Gibbs sampling [<xref ref-type="bibr" rid="B44">44</xref>]. The expression level of gene <italic>i </italic>is estimated by the posterior mean of <italic>μ<sub><italic>i</italic></sub></italic>, and the genes are ranked by the absolute values of the posterior means of the <italic>μ<sub><italic>i</italic></sub>'</italic>s. Genes with high rankings were designated as differentially expressed (DE) genes.</p></sec><sec><title>SAM t statistic</title><p>To evaluate the performance of the proposed method in estimating gene expression levels and identifying DE genes, we compared the proposed method to the sample mean and SAM t statistics [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Because of its good performance, the SAM t statistic [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>] is widely used to rank genes and detect DE genes [<xref ref-type="bibr" rid="B45">45</xref>]. We denote the SAM t statistic for the gene <italic>i </italic>as <italic>Z</italic><sub><italic>i</italic></sub>, then</p><p><inline-graphic xlink:href="1471-2164-7-87-i10.gif"/></p><p>where <inline-graphic xlink:href="1471-2164-7-87-i11.gif"/> and <italic>S</italic><sub><italic>i </italic></sub>are the sample mean and sample standard deviation for the gene <italic>i</italic>, and <italic>So </italic>is the 90 <italic>th </italic>percentile of <italic>S</italic><sub><italic>i</italic></sub>'s.</p></sec><sec><title>Simulation settings</title><p>We conducted three simulation studies to assess the usefulness of our method. The operon structure of the <italic>E. coli </italic>genome from the RegulonDB database [<xref ref-type="bibr" rid="B40">40</xref>] was used in simulation studies. We randomly chose 100 operons (involving about 340 genes) and assumed that genes from those operons were differentially expressed DE genes. Then we randomly picked a subset of non-operon genes to be DE genes and adjusted the total number of DE genes to 400. Let <italic>μ<sub><italic>i </italic></sub></italic>be the expression level of gene <italic>i</italic>, for <italic>i = </italic>1, 2,..., 4821. For DE genes, <italic>μ<sub><italic>i</italic></sub>'</italic>s were simulated from an equal mixture of <italic>N</italic>(1, 0.25<sup>2</sup>) and <italic>N(–</italic>1,0.25<sup>2</sup>) distributions, and genes within the same operon were from the same component of the mixture distribution. For EE genes, <italic>μ</italic><sub><italic>i </italic></sub>~ <italic>N</italic>(0, 0.25<sup>2</sup>). We simulated 4 replicates from a normal distribution for each gene, <italic>Y</italic><sub><italic>ij </italic></sub>~ <italic>N</italic>(μ<sub><italic>i</italic></sub>,<inline-graphic xlink:href="1471-2164-7-87-i1.gif"/>), where <italic>Y</italic><sub><italic>i</italic>,<italic>j </italic></sub>was the log ratio of transcript abundances for the gene <italic>i </italic>on the array <italic>j</italic>. To provide increasing noise levels for simulations 1, 2 and 3, the σ<sub><italic>i</italic></sub>'s were simulated from the <italic>uniform</italic>(0.25, 0.75), <italic>uniform(</italic>0.5,1.0), and <italic>uniform(</italic>0.75,1.25), respectively.</p></sec><sec><title>Estimation of false positives and false negatives</title><p>Using the posterior distributions, we can evaluate the FDR for specific number of DE genes. Using <italic>Pr</italic>(|<italic>μ</italic><sub><italic>i</italic></sub>| > <italic>δ'</italic>|<italic>Y</italic><sub><italic>i,j</italic></sub>) to estimate the probability of gene <italic>i </italic>to be a DE gene, we can estimate the number of false positives for a cut off value <italic>k </italic>[<xref ref-type="bibr" rid="B7">7</xref>]:</p><p><inline-graphic xlink:href="1471-2164-7-87-i12.gif"/></p><p>Here, the genes are ranked based on the estimated mean expression level <italic>μ<sub><italic>i</italic></sub></italic>. In this study, we set δ = 1,</p><p>which corresponding to the commonly used 2-fold cutoff.</p><p>The false discovery rate (FDR) for the cut off <italic>k </italic>can be derived as:</p><p><inline-graphic xlink:href="1471-2164-7-87-i13.gif"/></p><p>Similarly, the number of false negatives for the cut off <italic>k </italic>can be calculated as:</p><p><inline-graphic xlink:href="1471-2164-7-87-i14.gif"/></p></sec><sec><title>Algorithm for Gibbs sampler</title><p>The algorithm is implemented below:</p><p>Set initial values:</p><p><inline-graphic xlink:href="1471-2164-7-87-i15.gif"/></p><p>FOR t FROM 1 TO T, draws random samples:</p><p><inline-graphic xlink:href="1471-2164-7-87-i16.gif"/></p><sec><title>END FOR</title><p>where <italic>V</italic><sub><italic>i </italic></sub>is the sample variance of gene <italic>i</italic>, and <italic>V</italic><sub>0 </sub>is the median of <italic>V</italic><sub><italic>i</italic></sub>'s. <italic>n</italic><sub><italic>p </italic></sub>is the number of genes in operon <italic>p</italic>. <italic>T </italic>is the total number of iteration. To diminish the effect of the initial values, we discard the results from the early iterations <italic>(t ≤ T</italic><sub><italic>B</italic></sub>, where <italic>T</italic><sub><italic>B </italic></sub>is the burn in time). The posterior mean of <italic>μ</italic><sub><italic>i</italic></sub>, of gene <italic>i </italic>is calculated by:</p><p><inline-graphic xlink:href="1471-2164-7-87-i17.gif"/></p><p>In our proposed method, the expression level of gene <italic>i </italic>is estimated by <inline-graphic xlink:href="1471-2164-7-87-i18.gif"/>. In the real data example, <italic>T</italic><sub><italic>B </italic></sub>and <italic>T </italic>are 500 and 2000, respectively.</p></sec></sec></sec><sec><title>Authors' contributions</title><p>GX initiated the study, implemented the methods and conducted data analysis. WP participated in development of the methods and co-wrote the paper. BMV generated the <italic>E. coli </italic>motility data. ABK generated the <italic>E. coli </italic>data set containing 217 microarrays, supervised the project and co-wrote the paper. All authors contributed to the writing, read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Operon.r – The R code used in the study.</p></caption><media xlink:href="1471-2164-7-87-S1.R" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p>Motility.txt – E. coli motility data set.</p></caption><media xlink:href="1471-2164-7-87-S2.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional File 3</title><p>Result.txt – The result for E. coli motility data, including posterior mean of proposed method, sample mean and SAM t statistics</p></caption><media xlink:href="1471-2164-7-87-S3.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Construction of a nurse shark (<italic>Ginglymostoma cirratum</italic>) bacterial artificial chromosome (BAC) library and a preliminary genome survey
|
<sec><title>Background</title><p>Sharks are members of the taxonomic class Chondrichthyes, the oldest living jawed vertebrates. Genomic studies of this group, in comparison to representative species in other vertebrate taxa, will allow us to theorize about the fundamental genetic, developmental, and functional characteristics in the common ancestor of all jawed vertebrates.</p></sec><sec><title>Aims</title><p>In order to obtain mapping and sequencing data for comparative genomics, we constructed a bacterial artificial chromosome (BAC) library for the nurse shark, <italic>Ginglymostoma cirratum</italic>.</p></sec><sec><title>Results</title><p>The BAC library consists of 313,344 clones with an average insert size of 144 kb, covering ~4.5 × 10<sup>10 </sup>bp and thus providing an 11-fold coverage of the haploid genome. BAC end sequence analyses revealed, in addition to LINEs and SINEs commonly found in other animal and plant genomes, two new groups of nurse shark-specific repetitive elements, NSRE1 and NSRE2 that seem to be major components of the nurse shark genome. Screening the library with single-copy or multi-copy gene probes showed 6–28 primary positive clones per probe of which 50–90% were true positives, demonstrating that the BAC library is representative of the different regions of the nurse shark genome. Furthermore, some BAC clones contained multiple genes, making physical mapping feasible.</p></sec><sec><title>Conclusion</title><p>We have constructed a deep-coverage, high-quality, large insert, and publicly available BAC library for a cartilaginous fish. It will be very useful to the scientific community interested in shark genomic structure, comparative genomics, and functional studies. We found two new groups of repetitive elements specific to the nurse shark genome, which may contribute to the architecture and evolution of the nurse shark genome.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Luo</surname><given-names>Meizhong</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Kim</surname><given-names>HyeRan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Kudrna</surname><given-names>Dave</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Sisneros</surname><given-names>Nicholas B</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Lee</surname><given-names>So-Jeong</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Mueller</surname><given-names>Christopher</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Collura</surname><given-names>Kristi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Zuccolo</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Buckingham</surname><given-names>E Bryan</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A10" contrib-type="author"><name><surname>Grim</surname><given-names>Suzanne M</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A11" contrib-type="author"><name><surname>Yanagiya</surname><given-names>Kazuyo</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A12" contrib-type="author"><name><surname>Inoko</surname><given-names>Hidetoshi</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A13" contrib-type="author"><name><surname>Shiina</surname><given-names>Takashi</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A14" contrib-type="author"><name><surname>Flajnik</surname><given-names>Martin F</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A15" contrib-type="author"><name><surname>Wing</surname><given-names>Rod A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A16" corresp="yes" contrib-type="author"><name><surname>Ohta</surname><given-names>Yuko</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
BMC Genomics
|
<sec><title>Background</title><p>Bacterial artificial chromosome (BAC) libraries are indispensable for many applications in genomic studies [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]. BAC-end sequences have been used to develop sequence-tagged connector (STC) frameworks [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>], to survey genome structures [<xref ref-type="bibr" rid="B6">6</xref>], and for comparative analysis of gene structure and synteny. Although the whole-genome shotgun method has been used to produce genome draft sequences [<xref ref-type="bibr" rid="B7">7</xref>], mapped BACs usually have been required to provide a framework for sequence assembly and templates to complete the sequences of complex genomes [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B8">8</xref>]. BACs and BAC-based maps have also been used in sequencing of targeted genome regions [<xref ref-type="bibr" rid="B9">9</xref>], chromosomal landing or positional cloning [<xref ref-type="bibr" rid="B10">10</xref>], genome function investigation [<xref ref-type="bibr" rid="B11">11</xref>], and evolutionary and comparative studies [<xref ref-type="bibr" rid="B12">12</xref>]. Through BAC sequencing, a full-length avian androgen receptor gene was identified, which had not been detected with conventional methods [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>Sharks belong to the phylogenetic taxon comprising the oldest jawed vertebrates, the cartilaginous fish, which diverged from the common ancestor of all other jawed vertebrates 460–520 million years ago [<xref ref-type="bibr" rid="B14">14</xref>]. Genomic and genetic studies of this group, in comparison to representative species in other vertebrate taxa, will allow us to theorize about the fundamental genetic, developmental, and functional characteristics in the common ancestor of all jawed vertebrates. This ancient taxon is of particular interest to us since it is the oldest group of living animals having an adaptive immune system with underlying molecules and mechanisms similar to those of mammals [<xref ref-type="bibr" rid="B15">15</xref>]. While most sharks and other cartilaginous fish have large genome sizes (~80% of species studied have larger genome sizes than human, with some up to 5 times as large), the nurse shark genome size is relatively small at 4 × 10<sup>9 </sup>bp/haploid genome, only slightly larger than that of humans (3.4 × 10<sup>9</sup>). Thus, the nurse shark is a candidate to become a model species in the biomedical and genomics fields. However, most of the shark genes and intergenic regions are much larger than those of mammals, and thus large-insert genomic libraries are essential to obtain sufficient genomic information. Previously, BAC libraries of other cartilaginous fish (clearnose skate and horn shark)were described by Miyake and Amemiya [<xref ref-type="bibr" rid="B1">1</xref>]. In this paper we report the construction and characterization of a publicly-available nurse shark BAC library and carry out a preliminary genome survey.</p></sec><sec><title>Results and discussion</title><sec><title>BAC library construction</title><p>We constructed a nurse shark BAC library, consisting of a total of 313,344 clones that were deposited in 816 barcode-ordered 384-well microtiter plates. To complete this large library, approximately 20 ligations were performed. Since colonies with a diameter of <1.5 mm did not grow well in the freezing media, only those colonies with a diameter of >1.5 mm on selection-agar plates were picked.</p></sec><sec><title>Insert size distribution</title><p>To evaluate the quality of the BAC library, we first analyzed the insert sizes of 408 sampling BAC clones with <italic>Not </italic>I (cutting GCGGCCGC sequence), which liberates the inserts from the BAC vector (Figure <xref ref-type="fig" rid="F1">1</xref>). The 408 sampling BAC clones were selected by picking one clone from A01 position of every other 384-well plate of the library and arranged in the same order as the library plates. Of the 408 BAC clones, 6 did not yield detectable DNA. Of the remaining 402 clones, only one had no insert, indicating a negligible empty-vector rate (1/402). While most clones produced only one <italic>Not </italic>I insert band, ~16% of the clones produced more than one insert band due to the presence of internal <italic>Not </italic>I sites. Eight such clones produced a single or very few small insert bands with high DNA densities disproportional to the DNA density of the vector band, suggesting co-migration of multiple repetitive bands. In fact, when these clones were digested with <italic>Swa </italic>I (cutting ATTTAAAT sequence) that also liberates the inserts, they generated single large-insert bands (e.g. 194 kb). These clones may be of interest because they might contain large repetitive and structurally distinctive regions with extremely high GC content. The average insert size of the 402 clones is 144 kb, and thus the entire BAC library covers ~4.5 × 10<sup>10 </sup>bp (313,344 × 144,000 bp). The nurse shark genome content was reported to be ~8 pg DNA per cell [<xref ref-type="bibr" rid="B16">16</xref>], ~4 × 10<sup>9 </sup>bp per haploid genome. Thus, our BAC library covers ~11 haploid genome equivalents. Approximately 94% of clones contain inserts greater than 110 kb. Although multiple ligations were used, no significant difference was observed between different library segments corresponding to the different ligations except that one segment of 14,976 clones, corresponding to the library plates 628 to 666 (number 314 to 333 in Figure <xref ref-type="fig" rid="F1">1</xref>), has very large inserts. This segment resulted from two parallel ligations and has an average insert size of 209 kb.</p></sec><sec><title>BAC end sequence analysis</title><p>We sequenced the first 96 of the above 408 sampling clones at both ends. One hundred seventy-seven BAC end sequences (BES) were obtained with an average high-quality base-pair (bp) number of 548 bp ([GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549372">CZ549372</ext-link>~<ext-link ext-link-type="gen" xlink:href="CZ579549">CZ579549</ext-link>], excluding [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549507">CZ549507</ext-link>] which is the vector sequence), covering a total of ~100,000 bp. The average GC content of the BES is 41%, ranging from 26–66%. Sequences from other parts of the nurse shark genome (~300 kb) have an average of 42–45% GC content (YO, MFF; personal observation), which is consistent with this finding. It is worth noting that the GC content was calculated in windows of the BES lengths, and some smaller regions may contain higher GC content. Indeed, we previously detected simple repeats with a high GC content (70–80%) in a region of the Major Histocompatibility Complex (MHC) class I gene [<xref ref-type="bibr" rid="B17">17</xref>]. Analysis of the BES also revealed 4 sequences containing significantly long microsatellite repeats (simple sequence repeats, SSRs): [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549546">CZ549546</ext-link>] (AG)<sub>9</sub>, [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549467">CZ549467</ext-link>] (AG)<sub>19</sub>, [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549532">CZ549532</ext-link>] (AT)<sub>9 </sub>and [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549547">CZ549547</ext-link>] (AC)<sub>14</sub>. However, no significant tri-nucleotide or longer motifs of SSRs were found in our BES. The CpG dinucleotide frequency was under-represented (the observed value is only 1/4 of the expected one) while the AA and TT dinucleotide frequencies were over-represented in this set of BES.</p><p>BlastN and BlastX searches of these BES against the 'nr' database in GenBank and tBlastX and tBlastN searches against an in-house collection of conserved domains of non-LTR retroelements were carried out with an E-value of below 1e-10. The Blast analysis results are classified in Table <xref ref-type="table" rid="T1">1</xref> (BlastN results can be found in our web site [<xref ref-type="bibr" rid="B18">18</xref>]). Nine BES share a significant similarity with various shark genomic sequences; one [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549423">CZ549423</ext-link>] is similar to 18S ribosomal RNA; two showed hits to zebrafish and <italic>Tetraodon </italic>genomic sequences that remain 'unclassified'; and one hundred and two showed 'No Hits'. Twenty-nine and five BES contained LINEs and SINEs respectively, indicating that these repetitive elements are major components of the nurse shark genome. Most of them hit regions of published nurse shark genomic sequences, especially to an intron of the <italic>LMP7-like </italic>pseudogene, and non-coding sequences from genes in other shark species (<italic>RAG </italic>[<xref ref-type="bibr" rid="B19">19</xref>] and <italic>HOX </italic>[<xref ref-type="bibr" rid="B20">20</xref>]), and the 3'-untranslated region (3'UTR) of a non-classical class I gene, <italic>UAA-NC1 </italic>[<xref ref-type="bibr" rid="B17">17</xref>] that contains a retrotransposon. Most of the LINEs contain partial open reading frames (ORFs) encoding reverse transcriptases related to the CR1 family, first found in chicken [<xref ref-type="bibr" rid="B21">21</xref>]. The high representation of CR1-like SINEs/LINEs is thus likely to be a common feature of shark genomes. In the sandbar shark (<italic>Carcharhinus plumbeus</italic>), a total of 7 CR1-like SINE/LINE elements were found in the 9.4 kb intergenic region of <italic>RAG1 </italic>and <italic>RAG2 </italic>[<xref ref-type="bibr" rid="B22">22</xref>].</p></sec><sec><title>NSRE1 and NSRE2 analysis
</title><p>The remaining 29 BES and one other [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549472">CZ549472</ext-link>] that also contains a LINE element are of interest. They hit only nurse shark BAC draft sequences when searched against 'htgs' (high-throughput genomic sequence) database in GenBank. When BlastN searches were performed using the whole BES set as the database and the BES as queries, they were categorized into two groups designated as NSRE1 (<underline>n</underline>urse <underline>s</underline>hark <underline>r</underline>epetitive <underline>e</underline>lement 1) and NSRE2 (24 and 6 for NSRE1 and NSRE2 respectively, Figures <xref ref-type="fig" rid="F2">2</xref> and <xref ref-type="fig" rid="F3">3</xref>). Searching against the 'nr' database in GenBank revealed that 21 of 24 NSRE1 sequences hit the nurse shark <italic>UAA-NC1 </italic>cDNA ([GenBank:<ext-link ext-link-type="gen" xlink:href="AF357922">AF357922</ext-link>]) in the 5'UTR (nt positions 2–103), and 22 of 24 NSRE1 sequences hit the nurse shark LMP7 pseudogene, exon 4 ([GenBank:<ext-link ext-link-type="gen" xlink:href="AF357928">AF357928</ext-link>]) at nt positions 3274–3350 (see Additional file <xref ref-type="supplementary-material" rid="S1">1</xref>). The 698 bp-sequence of GC_Ba0153A01.f [Genbank:<ext-link ext-link-type="gen" xlink:href="CZ549510">CZ549510</ext-link>] hit the subject regions in two query regions (nt positions 1–60 and 538–635), showing two NSRE1 in one BES. Two BAC clones (GC__Ba0143A01 and GC__Ba0109A01) hit the subject regions at both ends. When the subject sequence [GenBank: <ext-link ext-link-type="gen" xlink:href="AF357922">AF357922</ext-link>] nt positions 2–103 region was used as a query to search the 'htgs' database in GenBank, it only hit all the 9 nurse shark BAC draft sequences then present in the GenBank in multiple regions with an average of 7.2 hits/BAC, from a minimum of 3 to a maximum of 11 (data of Oct. 2, 2005). Searching the NSRE2 sequence [GenBank:<ext-link ext-link-type="gen" xlink:href="CZ549534">CZ549534</ext-link>] against the 'htgs' database in GenBank found an average of 3.4 hits/BAC on 8 of the 9 nurse shark BAC draft sequences ranging from a minimum of 1 to a maximum of 5. These results indicate that NSRE1 and NSRE2 are repetitive elements, and NSRE1 is more frequent in the nurse shark genome than NSRE2. BlastX searches of both NSRE1 and NSRE2 did not detect any protein matches. To further sample the representation and organization of NSRE1 and NSRE2 elements in the nurse shark genome including coding and non-coding regions, we sequenced the BAC clone GC_Ba0754I06. The draft sequence of this clone revealed seventeen NSRE1 and six NSRE2 in a ~170 kb region dispersed among SINEs/LINEs repetitive elements. All identified repetitive elements are found most abundantly in regions outside of the gene (e.g. fatty acid synthase FASN, Figure <xref ref-type="fig" rid="F4">4</xref>).</p><p>To confirm that these repetitive sequences were present in a high copy number in the nurse shark genome and to test for their existence in related animals, we did Southern blotting and library screening using 'overgo' probes of NSRE1 and NSRE2. Both the NSRE1 and NSRE2 probes hybridized to the nurse shark genomic DNA, but not to DNA from other elasmobranchs (sand tiger shark, little skate, and lemon shark), <italic>Xenopus laevis</italic>, rat, human and zebra finch (Figure <xref ref-type="fig" rid="F5">5</xref> for NSRE1, data not shown for NSRE2). Both the NSRE1 and NSRE2 hybridization signals are smear, indicating that these elements are highly repetitive and dispersed in the nurse shark genome, consistent with our sequence analysis results above. From a library screening of 36,864 clones, many colonies hybridized with different intensities (data not shown), perhaps correlating with the copy numbers within the BAC clones. Our data suggest that the expansion of NSRE1 and NSRE2 repetitive elements occurred after the divergence of nurse shark from other shark lineages. However, we examined only distantly related shark species in this study. Closely related shark species, belonging to the same family (Orectolobiformes) as nurse shark (e.g. wobbegong shark, bamboo shark), must be examined for the presence/absence of these repetitive elements. It is possible that different, unique repetitive elements will be found in other shark species.</p></sec><sec><title>BAC library screening with gene-specific probes</title><p>To further assess the quality and demonstrate the utility of the BAC library, we screened the entire BAC library with gene-specific probes. These probes are listed in Table <xref ref-type="table" rid="T2">2</xref>. With all single-copy gene probes except <italic>TAP1 </italic>(transporter associated with antigen processing) and <italic>Ring3</italic>, we obtained 6–28 positive clones (Table <xref ref-type="table" rid="T2">2</xref>, library screening). A similar number of positive clones was observed when we used the <italic>Factor B </italic>probe, present in two tandem copies in the nurse shark genome. We further confirmed the positive clones by colony hybridization (Table <xref ref-type="table" rid="T2">2</xref>, colony hybridization). Most of the gene probes except <italic>CD83 </italic>[<xref ref-type="bibr" rid="B23">23</xref>] resulted in 8–19 true positives, consistent with the 11x coverage calculated from the average insert size and total number of clones. The low percentage (50%) of true positives from MHC class I and <italic>CD83 </italic>may be due to weakly hybridizing false-positives, which appeared only after long film exposures. In some cases, we observed multiple weaker signals in the vicinity of the stronger signals in the same double-spotting pattern, presumably due to carryover during filter production. In fact, the low percentage (62%) of true positives for the <italic>TAP1 </italic>probe was most probably due to such carryover of positive clones into neighboring wells. Once we carefully selected putative positive clones for <italic>TCR </italic>probes (i.e. only selected the strongest signals among the neighboring signals), the percentage of true positives increased (see Table <xref ref-type="table" rid="T2">2</xref>, 89–91% <italic>TCR</italic>s). <italic>TAP1 </italic>and <italic>Ring3 </italic>are members of large gene families containing conserved domains, which might cross-hybridize to other family members and thus result in higher number of positive clones than expected. With the <italic>LMP7 </italic>probe, 19 positive clones seem to be low for a multi-copy gene, however we have found at least 2 pseudogenes containing only small fragments of the gene [<xref ref-type="bibr" rid="B17">17</xref>]. These results showed that most genetic regions tested in this study (except <italic>CD83</italic>) were well represented in the BAC library.</p><p>Unlike most other species, the nurse shark MHC genes are much (3–5 times) larger than those of mammals and have even larger distances in intergenic regions [<xref ref-type="bibr" rid="B17">17</xref>]. Previously, we constructed a genomic cosmid library with an average insert size of ~40 kb. However, most cosmid clones contain at most a single gene (data not shown). In this study, we used several linked genes <italic>(TAP1</italic>, MHC class I, class II, <italic>Factor B, LMP7, LMP2 </italic>[<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B24">24</xref>], and <italic>Ring3 </italic>(unpublished data)) to quickly glean the number of genes in a single BAC clone; we found up to four genes in a single BAC clone, making physical mapping possible. Thus far, our analysis has convincingly shown that the BAC library is a useful tool (and perhaps the only way) to obtain genetic information for this species.</p></sec></sec><sec><title>Conclusion</title><p>We report in this paper a large insert, deep-coverage and high-quality BAC library for a cartilaginous fish that will be very useful to the scientific community for gene isolation, genetic analysis, and comparative genomics. We found two new groups of repetitive elements, designated as NSRE1 and NSRE2, which are specific to the nurse shark genome. These repetitive elements may contribute to the architecture and evolution of the nurse shark genome. The BAC library, HDR filters and individual clones are available to the public from the Arizona Genomics Institute's BAC/EST Resource Center [<xref ref-type="bibr" rid="B18">18</xref>].</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Isolation of high molecular weight DNA</title><p>Blood was obtained from the nurse shark individual "Yellow" using a Heparinized 18G 1 1/2" needle from the caudal vein. To obtain ~30 micrograms of DNA per 80-μl agarose plug, approximately 4 × 10<sup>6 </sup>erythrocytes were embedded in 1% InCert agarose (FMC, Rockland, ME) prepared in 1/2x PBS and molded in ~200 Plug Molds (BIO-RAD, Hercules, CA). Twenty plugs were then submerged in 50 ml of cell lysis solution (1% lithium dodecyl sulfate, 10 mM Tris, pH8.0, 100 mM EDTA, pH8.0) and incubated overnight at 37°C with occasional swirling. The cell lysis solution was replaced with 50 ml 20% NDS (0.2% N-lauroylsarcosine, 2 mM Tris, pH9.0, 0.14M EDTA, pH9.0), and DNA plugs were shaken gently at room temperature for two hours, and then kept at 4°C [<xref ref-type="bibr" rid="B25">25</xref>].</p></sec><sec><title>BAC vector preparation</title><p>We used a modified version of the BAC vector pBeloBAC11, pIndigoBAC536Swa. A first modified version of pBeloBAC11 [GenBank:<ext-link ext-link-type="gen" xlink:href="U51113">U51113</ext-link>], pIndigoBAC536, was a gift from Dr. H. Shizuya of Caltech. pIndigoBAC536 has the internal <italic>Eco </italic>R1 site of pBeloBAC11 destroyed so that the unique <italic>Eco </italic>R1 site in the multiple cloning sites can be used for cloning, and also contains a random point mutation in the <italic>lac</italic>Z gene that provides colonies with a darker blue-color on X-gal/IPTG selection. We further inserted two <italic>Swa </italic>I sites (ATTTAAAT) near and internal to the two <italic>Not </italic>I sites of pIndigoBAC536 (this new version is named pIndigoBAC536Swa) to facilitate insert-size estimation of clones from GC-rich organisms (Luo et al, unpublished data). We then cloned this single-copy BAC vector pIndigoBAC536Swa into the high-copy vector pGEM-4Z (this composed high-copy plasmid is named pAGIBAC1) to facilitate the preparation of the single-copy BAC vector as we did for pIndigoBAC536 (the composed high-copy pIndigoBAC536-pGEM-4Z plasmid is named pCUGIBAC1) [<xref ref-type="bibr" rid="B26">26</xref>]. pCUGIBAC1 is available through Clemson University Genomics Institute [<xref ref-type="bibr" rid="B27">27</xref>] and pAGIBAC1 is available through Arizona Genomics Institute [<xref ref-type="bibr" rid="B18">18</xref>]. The linearized and dephosphorylated single-copy BAC vector pIndigoBAC536Swa can be prepared from the high-copy pAGIBAC1 according to our previously published method [<xref ref-type="bibr" rid="B28">28</xref>].</p></sec><sec><title>Generation and size selection of large DNA fragments for BAC cloning</title><p>Large genomic DNA fragments for BAC cloning were prepared according to our previously published method [<xref ref-type="bibr" rid="B28">28</xref>]. The DNA-agarose plugs were washed thoroughly with TE buffer (10 mM Tris/1 mM EDTA, pH8.0) and stored in 70% ethanol at -20°C. A desired number of DNA plugs were transferred to TE buffer the day before use and kept at 4°C overnight. The DNA plugs were test-digested with various amount of <italic>Hin </italic>dIII (1–50U) for 20 minutes at 37°C to optimize partial digestion conditions and the fragmented DNAs were separated on 1% agarose gels by Pulsed Field Gel Electrophoresis (PFGE) (CHEF Mapper, BIO-RAD) at 1–50 sec linear ramp, 6 volts/cm, 14°C in 0.5X TBE buffer for 18–20 hours. Bulk digestions were then carried out using the conditions that produced the most DNA fragments in the range of 100–400 kb. Fragmented DNAs were separated on a 1% CHEF gel in the same conditions described above. DNA fractions ranging from 150–250 kb and 250–350 kb were excised from the gel and subjected to a second size selection on a 1% CHEF gel at 4 sec constant time, 6 volts/cm, 14°C in 0.5X TBE buffer for 18–20 hours. DNA fragments were electroeluted with dialysis tubing as described by Strong et al [<xref ref-type="bibr" rid="B29">29</xref>] or with an Electro-eluter Model 422 (BIO-RAD) following the manufacture's instructions. DNA concentrations were determined on subsequent agarose gels.</p></sec><sec><title>Ligation and transformation</title><p>One hundred to two hundred nanograms of size-selected DNA fragments were ligated with 20 nanograms of dephosphorylated BAC vector in a 100 μl of volume at 16°C overnight. The ligation reactions were terminated at 65°C for 15 min, and the ligation products were desalted in 0.1 M glucose : 1% agarose cones for 1.5 hours on ice as described by Atrazhev and Elliott [<xref ref-type="bibr" rid="B30">30</xref>] and electroporated into the <italic>E. coli </italic>strain DH10B T1 phage resistant (F-<italic>mcr</italic>A Δ(<italic>mrr</italic>-<italic>hsd</italic>RMS-<italic>mcr</italic>BC) φ80d<italic>lac</italic>Z ΔM15 Δ<italic>lac</italic>X74 <italic>deo</italic>R <italic>rec</italic>A1 <italic>end</italic>A1 <italic>ara</italic>D139 Δ(<italic>ara, leu</italic>) 7697 <italic>gal</italic>U <italic>gal</italic>K λ-<italic>rps</italic>L <italic>nup</italic>G) electrocompetent cells (Invitrogen, Carlsbad, CA). Transformants were grown on LB plates supplemented with 12.5 mg/L chloramphenicol, 80 mg/L X-gal (5-bromo-4-chloro-3-indolyl-beta-D-galactoside or 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside) and 100 mg/L IPTG (Isopropyl-beta-D-thiogalactoside or Isopropyl-beta-D-thiogalactopyranoside) at 37°C overnight.</p></sec><sec><title>Library arraying and high density replica (HDR) filters</title><p>A total of 313,344 individual recombinant clones (white color on X-gal plates) were picked robotically (Genetix, New Milton, UK) and arrayed into 816 barcode-ordered 384-well microtiter plates containing freezing media (10 g/L Bacto tryptone, 5 g/L Bacto yeast extract, 10 g/L NaCl, 36 mM K<sub>2</sub>HPO<sub>4</sub>, 13.2 mM KH<sub>2</sub>PO<sub>4</sub>, 1.7 mM Na-citrate, 6.8 mM (NH<sub>4</sub>)<sub>2</sub>SO<sub>4</sub>, 4.4% glycerol, autoclaved and added filter-sterilized MgSO<sub>4 </sub>solution to the final concentration of 0.4 mM) supplemented with 12.5 mg/L of chloramphenicol. After an overnight incubation at 37°C, empty wells were back-filled manually and duplicate copies were replicated. The master library and the two copies were then stored in -80°C freezers at different locations. The whole BAC library was gridded onto 17 22.5 cm × 22.5 cm Hybond N+ membrane filters (Amersham, Piscataway, NJ) in high density, double spots, and 4 × 4 patterns with Genetix Q-bots (Genetix). Each 22.5 cm × 22.5 cm filter supports 18,432 clones in duplicate in 6 fields. The filters were placed on LB media supplemented with 12.5 mg/L of chloramphenicol and incubated overnight at 37°C. The filters were then soaked in 0.5N NaOH/1.5M NaCl for 7 min, in 1.5M NaCl/0.5M Tris-HCl (pH8) for 7 min, air dried for 1–2 hours, soaked in 0.4N NaOH for 20 min, in 20x SSPE for 7 min, and air-dried overnight.</p></sec><sec><title>DNA analysis of BAC clones</title><p>BAC DNAs were extracted with Tomtec Quadra 96 model 320 (Tomtec, Hamden, CT) in a 96-well format at AGI. The 408 sampling BAC clones were selected by picking one clone from A01 position of every other 384-well plate of the library and arranged in the same order of the library plates. Inserts were liberated by digesting with <italic>Not </italic>I or <italic>Swa </italic>I and their sizes were determined on CHEF gels.</p></sec><sec><title>BAC end sequencing</title><p>BAC DNAs were sequenced at both ends using BigDye Terminator v.3 (Applied Biosystems, ABI, Foster City, CA) according to manufacturer's instruction. The T7 primer (5' TAA TAC GAC TCA CTA TAG GG 3') was used as the "forward" primer and the BES_HR primer (5' CAC TCA TTA GGC ACC CCA 3') was used as the "reverse" primer. Cycle sequencing was performed using PTC-200 thermal cyclers (MJ Research, Waltham, MA) in a 384-well format with the following regime: 150 cycles of 10 sec at 95°C, 5 sec at 55°C, and 2.5 min at 60°C. After the cycle-sequencing step, the DNA was purified by magnetic beads, CleanSeq (Agencourt, Beverly, MA) according to manufacturer's instruction. Samples were eluted into 20 μl of water and separated on ABI 3730xl DNA capillary sequencers with default conditions. Sequence data was collected by data collection software (Applied Biosystems), extracted using sequence analysis software (Applied Biosystems) and transferred to a UNIX workstation. Sequences were base-called using the program Phred [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>]; vector and low-quality (Phred value <16) sequences were removed by CROSS_MATCH [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>].</p></sec><sec><title>Bioinformatics analyses of sequences</title><p>Similarity searches against public GenBank and in-house database were carried out using the Blast algorithm. Composition analyses as well as searches for inverted repeats were done using the programs "composition" and "palindrome" respectively, both of which are included in the package EMBOSS [<xref ref-type="bibr" rid="B33">33</xref>]. SSR were searched using the software "Sputnik" [<xref ref-type="bibr" rid="B34">34</xref>].</p></sec><sec><title>NSRE alignment</title><p>The nucleotide sequences of both NSRE1 and NSRE2 were aligned using ClustalW. NSRE1 motifs were extracted from BES and full-length BES sequences were aligned for NSRE2.</p></sec><sec><title>Southern blotting for NSREs</title><p>Five μg genomic DNAs were digested with 80 units of restriction enzyme <italic>Hin </italic>dIII for 6 hours at 37°C. DNA fragments were separated in a 0.8% agarose gel by electrophoresis and blotted onto a nylon membrane. Overlapping oligonucleotide (overgo) hybridization was performed according to Ross et al [<xref ref-type="bibr" rid="B35">35</xref>] with modifications. The NSRE1 and NSRE2 overgo probes were designed from the sequence [GenBank:<ext-link ext-link-type="gen" xlink:href="AF357922">AF357922</ext-link>] nt positions 2–103 region and sequence [GenBank:<ext-link ext-link-type="gen" xlink:href="AF357928">AF357928</ext-link>] nt position 4406–4529 region respectively. Primers used for NSRE1 are: 5' TCT CGG CCC GAA ACG TCA GCT TTC 3' and 5' AGC ATC AGA GGA GCA CGA AAG CTG 3'. Primers used for NSRE2 are: 5' TGC TGT TCC TGC AAC CTT CGG GTA 3' and 5' AAT GCC ACA ACG ACG CTA CCC GAA 3'. Each set of primers overlaps by 8 base pairs. Probes were labeled with both <sup>32</sup>P-dCTP and <sup>32</sup>P-dATP using Klenow enzyme. Hybridization was carried out overnight in a solution containing 1% Bovine Serum Albumin (BSA), 1 mM EDTA pH8.0, 7% SDS, 0.5M sodium phosphate at 60°C. Membranes were washed in 4x SSC, 0.1% SDS at room temperature, followed by 1.5x SSC, 0.1% SDS at 60°C. Membranes were exposed to screens and scanned using the phosphor imager.</p></sec><sec><title>BAC sequencing and assembly</title><p>The BAC clone GC_Ba0754I06 that covered 170 kb was bidirectionally shotgun sequenced with an average redundancy of about 6, which was sufficient for assembly and analysis of the entire sequence using previously established procedures [<xref ref-type="bibr" rid="B36">36</xref>]. The draft sequence was searched using Blast2 for NSRE1 and NSRE2, BlastN and BlastX against the 'nr' database in GenBank for SINEs/LINEs and CR1-like SINEs/LINEs, respectively. During the BlastX search, at least 27 exons were identified with significant similarity to other species' fatty acid synthase (FASN) exons (E-values of 3e-56 and 3e-46 to chicken [GenBank:<ext-link ext-link-type="gen" xlink:href="AAB46389">AAB46389</ext-link>] and rat [GenBank:<ext-link ext-link-type="gen" xlink:href="AAA41145">AAA41145</ext-link>], respectively).</p></sec><sec><title>BAC library screening</title><p>The seventeen HDR filters of the BAC library were pre-hybridized in high-stringency hybridization solution (50% Formamide, 6x SSC, 0.5% SDS, 5x Denhardt's solution) [<xref ref-type="bibr" rid="B37">37</xref>] supplemented with 100μg/ml of denatured salmon sperm DNA for ~4 hours at 42°C. Probes (50 ng) were radiolabeled using the random-priming method (Roche, Indianapolis, IN) or by incorporating α<sup>32</sup>P-dCTP in the polymerase chain reaction (PCR) [<xref ref-type="bibr" rid="B38">38</xref>]. Hybridization was performed overnight at 42°C and membranes were washed at room temperature in pre-warmed (42°C) 2x SSC/1% SDS for 20 minutes, followed by washing in 0.2x SSC/0.1% SDS for 20 minutes at 65°C. Membranes were exposed to X-ray film for various lengths of time to obtain positive signals and the desired background. General protocols for high-density BAC library filter screening and address determination of positive signals are publicly available from our website [<xref ref-type="bibr" rid="B18">18</xref>]. Putative positive clones were re-spotted on nylon membranes for colony hybridization to confirm true positives. For hybridization with NSRE overgo probes, we performed in the same hybridization solution as described above for Southern Blotting. Filters were washed in 4x SSC, 0.1% SDS at room temperature, followed by 0.75x SSC, 0.1% SDS at 60°C, exposed to screens, and scanned using the phosphor imager.</p></sec></sec><sec><title>Authors' contributions</title><p>RAW, MFF, ML and YO participated in the design of this study. ML constructed the BAC library. ML, YO, HK, DK, NBS, S-JL, CM, KC, AZ, EBB and SMG participated in library characterization and data analysis. KY, HI, and TS participated in sequencing and contig assembly of the BAC clone GC_Ba0754I06. ML, MFF, and YO prepared this manuscript. All authors have read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Analysis of NSRE1 and NSRE2 sequences. GenBank accession numbers containing NSRE1 and NSRE2 identified in this study are listed in this table. Some of them also matched the two nurse shark sequences [GenBank:<ext-link ext-link-type="gen" xlink:href="AF357922">AF357922</ext-link>] and [GenBank:<ext-link ext-link-type="gen" xlink:href="AF357928">AF357928</ext-link>]; the length and the % identities at the nucleotide level are also shown.</p></caption><media xlink:href="1471-2164-7-106-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
C-reactive protein in degenerative aortic valve stenosis
|
<p>Degenerative aortic valve stenosis includes a range of disorder severity from mild leaflet thickening without valve obstruction, "aortic sclerosis", to severe calcified aortic stenosis. It is a slowly progressive active process of valve modification similar to atherosclerosis for cardiovascular risk factors, lipoprotein deposition, chronic inflammation, and calcification. Systemic signs of inflammation, as wall and serum C-reactive protein, similar to those found in atherosclerosis, are present in patients with degenerative aortic valve stenosis and may be expression of a common disease, useful in monitoring of stenosis progression.</p>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Sanchez</surname><given-names>Pedro L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Mazzone</surname><given-names>AnnaMaria</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
Cardiovascular Ultrasound
|
<sec><title>Introduction</title><p>Degenerative aortic stenosis (AS) is at present time the most frequent valvulopaty in developed countries, and as life expectancy increases, the incidence and prevalence of AS will also rise fundamentally at the expense of the degenerative form.</p><p>The scientific interest of this valvar disorder was parked for many years, as the image tools necessary to quantify it are at our disposal and we have a certain agreement of the clinical moment for indicating the valve replacement. Nevertheless, in the past years different studies have demonstrated a common pathogenic mechanism between degenerative AS and atherosclerosis [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]. This is consistent with histopathological evidence that the lesions in AS involve active cellular processes that have classical "response to injury features", namely inflammatory infiltrates containing macrophages, T cells, and smooth muscle cells [<xref ref-type="bibr" rid="B4">4</xref>]. Therefore, investigators have needed only a half turn of the head to transfer the early past experience of blood markers of inflammation in the atherosclerosis setting to the AS scenario.</p><p>C-reactive protein (CRP) is the marker of inflammation most widely studied in patients with coronary artery disease and hence has become the marker of reference for any other inflammatory-based disease [<xref ref-type="bibr" rid="B5">5</xref>]. On this basis, CRP has emerged as a leading candidate for a better understanding of AS pathogenesis, for predicting AS progression, and for driving therapies in AS.</p></sec><sec><title>C-reactive protein is increased in patients with degenerative aortic stenosis</title><p>The first study showing increased CRP levels in patients with degenerative AS was published by Galante et al, in 2001 [<xref ref-type="bibr" rid="B6">6</xref>]. They compared serum CRP levels of 68 consecutive patients with severe degenerative trileaflet AS and absence of coronary atherosclerotic lesions admitted for elective cardiac surgery with 92 healthy controls. CRP levels were higher in patients with AS than in controls (0.85 ± 1.42 vs. 0.39 ± 0.50; p = 0.0001). They also showed an independent association of CRP with AS; the odds ratio for the disease according to CRP levels was 2.62 (95% confidence interval: 1.06 to 6.49). No association between CRP and aortic jet velocity, aortic valve area, or degree of calcification was found, notwithstanding that all patients had severe aortic stenosis and were waiting for surgery.</p><p>Further studies have confirmed [<xref ref-type="bibr" rid="B1">1</xref>] and expanded these observations. Serum CRP is also elevated in patients with asymptomatic aortic stenosis [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>], does not rise in accordance with increasing severity of valvar disease [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>], and decreased from before to 6 months after aortic valve replacement [<xref ref-type="bibr" rid="B9">9</xref>]. However, other studies have shown a week association of CRP and AS in the initial stages of the disease (aortic sclerosis) [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>One of the principal characteristics of AS is the important degree of valvar calcification that takes place along the different stages of the disease. Thus, investigating whether CRP could be related to the mechanism of calcification appears very atractive. The role of serum CRP in the tissue calcification process has been investigated in a recently elegant study by Warrier et al,. When aortic wall was exposed to an excess amount of CRP in an in vitro simulating model, the calcification rate of aortic wall increased as the concentration of CRP. The results of this work could reveal the role of CRP present in physiological fluid in aortic valvar calcification [<xref ref-type="bibr" rid="B11">11</xref>]. Data providing contribution of serum CRP to valve calcification in the clinical setting is available in patients with renal failure. Valvar calcification in patients with renal failure is associated with enhanced inflammation [<xref ref-type="bibr" rid="B12">12</xref>]. Furthermore, in chronic hemodyalisis patients in steady clinical conditions with no clinical evidence of either infectious or inflammatory diseases, a high CaxPO4 is associated with high CRP concentrations and hence associated with valvar calcification [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. However, preliminary transversal evidence evaluating the association of CRP and calcification in patients with AS and no renal failure is controversial [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]; thus, the long-term predictive value of the serum CRP level for the development of aortic calcification should be addressed in future well-designed prospective trials.</p><p>Recently, Skowasch et al, have observed localization of CRP in valve tissue of degenerative AS and degenerative aortic valve bioprostheses [<xref ref-type="bibr" rid="B15">15</xref>]. Furthermore, serum CRP showed a significant correlation with the valvar CRP expression (r = 0.54; P < 0.001). Consequently, we must integrate our previous understanding of the physiological role of CRP in inflamed tissues, thereby promoting local anti-inflammatory and proinflammatory effects [<xref ref-type="bibr" rid="B16">16</xref>]. Therefore, elevated CRP levels, whether preceding the development of degenerative valvular aortic stenosis or established at some time during the course of the disease, may have a detrimental influence on the natural history of the disease by inducing local activation of complement and subsequent amplification of local inflammation and cellular damage [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B18">18</xref>].</p></sec><sec><title>C-reactive protein and progression of aortic stenosis</title><p>We have assessed whether serum CRP levels could predict rapid AS progression. We measured serum high sensitivity CRP in 43 asymptomatic subjects with AS at baseline and six months later. Plasma CRP concentration was significantly higher in patients with rapid AS progression (5.1 [2.3 to 11.3] mg/L) compared to patients with slow AS progression (2.1 [1.0 to 3.1] mg/L, p = 0.007). The rate of progression was higher in patients who had a cutpoint level above the median CRP concentration (3 mg/L) than those who had levels ≤3 mg/L (66.7% vs 33.3%, p = 0.012 for aortic jet velocity and 62.5% vs 37.5%, p = 0.063 for aortic valve area. Little is known of the mechanisms responsible for progression of AS albeit mechanical, clinical, and metabolic variables have been suggested to contribute to rapid progression of AS [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Our data suggest that elevated CRP levels may be a marker of AS progression and could have important clinical implications as interventions that reduce CRP levels may be beneficial in the prevention of AS and perhaps also in reducing AS progression.</p><p>Finally, baseline CRP concentration was similar in patients who developed symptoms compared to those asymptomatic during follow-up. Data in patients with aortic sclerosis suggest a positive association between the risk of adverse cardiovascular events and the presence of coronary artery disease (hazard ratio [HR] 3.23, p = 0.003) and enhanced inflammation (HR 2.2, p = 0.001), and not as a result of the effects of valvar heart disease per se [<xref ref-type="bibr" rid="B21">21</xref>].</p></sec><sec><title>Targeting C-reactive protein for the therapy of aortic stenosis</title><p>The presence of higher serum CRP levels and the tissue location of CRP in patients with AS have raised the important question of whether medical therapies with agents such as statins and ACE inhibitors, which have already been shown to delay the progression of atherosclerosis may also affect the progression of AS.</p><p>Preclinical studies have shown that experimental hypercholesterolemia provide evidence of a proliferative atherosclerosis-like process in the aortic valve that is inhibited by statins [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>]; ie, atorvastatin inhibited calcification in the aortic valve by increasing eNOS protein and serum nitrite concentrations [<xref ref-type="bibr" rid="B23">23</xref>], and decreasing Lrp5 (low-density receptor-related protein) receptors involve in cellular proliferation and osteoblastogenesis via the beta-catenin signaling pathway [<xref ref-type="bibr" rid="B24">24</xref>].</p><p>Although recent retrospective clinical studies suggest that statins also may slow the hemodynamic progression of AS [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B29">29</xref>], the results of the SALTIRE study have been discouraging for all of us who believe that AS is an active disease process akin to atherosclerosis with lipoprotein deposition, chronic inflammation, and active leaflet calcification [<xref ref-type="bibr" rid="B30">30</xref>]. Therefore, it is important to analyze the neutral effect of high doses of atorvastatin in such an attractive hypothesis. First, the SALTIRE study differs not only because of its prospective design but also because the indications for therapy were different. In the retrospective trials, statin therapy was indicated for the treatment of hyperlipidemia, whereas in the prospective trial, patients in whom statins were indicated for the treatment of hyperlipidemia were excluded. Second, statin doses in the retrospective studies were lower. Finally, although the observation periods in the various studies were similar, patients in the retrospective studies were already receiving therapy at the time of inclusion in the study [<xref ref-type="bibr" rid="B31">31</xref>]. Aside from the already commented differences with retrospective studies, high proportion of patients were on drugs with anti-inflammatory effects (aspirin, betablockers or angiotensin converting enzyme inhibitors) that could have mitigated the pleiotropic effect of statins. For example, in the retrospective trials that found a lower rate of progression among patients treated with statins, the percentage of patients on aspirin was always important and significantly higher in the group with positive results [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B29">29</xref>]. Importantly, in the SALTIRE study, half of the patients in each group were on aspirin, a prevalence that would have a bearing on the results of the study. In addition, markers of inflammation would have been useful as a means to predict or monitor the individual's response to atorvastatin. If the authors were not just looking for a reduction cholesterol effects it is surprising that they have not yet provided any information about inflammatory effects. As the main finding, Skowasch et al, [<xref ref-type="bibr" rid="B15">15</xref>] in their recent study showed that both valvar CRP expression and serum CRP levels were found to be lower in patients on statins.</p></sec><sec><title>Limitations</title><p>The main limitations of all the studies that have evaluated the association between CRP and AS are similar to the previous experience in the atherosclerosis setting: variability and laboratory methodology in CRP determination [<xref ref-type="bibr" rid="B32">32</xref>,<xref ref-type="bibr" rid="B33">33</xref>]; cardiovascular risk factors and other variables known to affect CRP concentrations [<xref ref-type="bibr" rid="B34">34</xref>]; CRP was determined once, so longitudinal data are not available; other non-inflammatory mechanisms are likely to have an important role in the pathogenesis of degenerative aortic valve disease [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]; small study populations; short follow-up intervals; etcetera.</p></sec><sec><title>Conclusion</title><p>Moving forward, we must learn more about the pathogenic mechanisms of AS. We must integrate the atherosclerotic background of inflammatory biomarkers in our future research, and finally we must focus on the development of prospective, randomised trials using CRP to monitor the individual's response to treatments.</p></sec><sec><title>Abbreviations</title><p>Angiotensin Converting ezyme (ACE)</p><p>Aortic stenosis (AS)</p><p>C-reactive protein (CRP)</p><p>Hazard Ration (HR)</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>Both authors contributed equally to the review.</p></sec>
|
The complete chloroplast genome sequence of the chlorophycean green alga <italic>Scenedesmus obliquus </italic>reveals a compact gene organization and a biased distribution of genes on the two DNA strands
|
<sec><title>Background</title><p>The phylum Chlorophyta contains the majority of the green algae and is divided into four classes. While the basal position of the Prasinophyceae is well established, the divergence order of the Ulvophyceae, Trebouxiophyceae and Chlorophyceae (UTC) remains uncertain. The five complete chloroplast DNA (cpDNA) sequences currently available for representatives of these classes display considerable variability in overall structure, gene content, gene density, intron content and gene order. Among these genomes, that of the chlorophycean green alga <italic>Chlamydomonas reinhardtii </italic>has retained the least ancestral features. The two single-copy regions, which are separated from one another by the large inverted repeat (IR), have similar sizes, rather than unequal sizes, and differ radically in both gene contents and gene organizations relative to the single-copy regions of prasinophyte and ulvophyte cpDNAs. To gain insights into the various changes that underwent the chloroplast genome during the evolution of chlorophycean green algae, we have sequenced the cpDNA of <italic>Scenedesmus obliquus</italic>, a member of a distinct chlorophycean lineage.</p></sec><sec><title>Results</title><p>The 161,452 bp IR-containing genome of <italic>Scenedesmus </italic>features single-copy regions of similar sizes, encodes 96 genes, <italic>i.e</italic>. only two additional genes (<italic>infA </italic>and <italic>rpl12</italic>) relative to its <italic>Chlamydomonas </italic>homologue and contains seven group I and two group II introns. It is clearly more compact than the four UTC algal cpDNAs that have been examined so far, displays the lowest proportion of short repeats among these algae and shows a stronger bias in clustering of genes on the same DNA strand compared to <italic>Chlamydomonas </italic>cpDNA. Like the latter genome, <italic>Scenedesmus </italic>cpDNA displays only a few ancestral gene clusters. The two chlorophycean genomes share 11 gene clusters that are not found in previously sequenced trebouxiophyte and ulvophyte cpDNAs as well as a few genes that have an unusual structure; however, their single-copy regions differ considerably in gene content.</p></sec><sec><title>Conclusion</title><p>Our results underscore the remarkable plasticity of the chlorophycean chloroplast genome. Owing to this plasticity, only a sketchy portrait could be drawn for the chloroplast genome of the last common ancestor of <italic>Scenedesmus </italic>and <italic>Chlamydomonas</italic>.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>de Cambiaire</surname><given-names>Jean-Charles</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Otis</surname><given-names>Christian</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Lemieux</surname><given-names>Claude</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" corresp="yes" contrib-type="author"><name><surname>Turmel</surname><given-names>Monique</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Evolutionary Biology
|
<sec><title>Background</title><p>The complete chloroplast DNA (cpDNA) sequences currently available for green plants (green algae and land plants) point to radically divergent evolutionary trends of the chloroplast genome in the phyla Streptophyta and Chlorophyta. The Streptophyta [<xref ref-type="bibr" rid="B1">1</xref>] comprises all land plants and their closest green algal relatives, the members of the class Charophyceae sensu Mattox and Stewart [<xref ref-type="bibr" rid="B2">2</xref>]. In this phylum are currently available the chloroplast genome sequences of about 35 land plants and six or seven charophycean green algae (six algae if the controversial phylogenetic position of <italic>Mesostigma viride </italic>at the base of the Streptophyta and Chlorophyta [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>] proves to be correct and seven algae if its association with the Streptophyta is confirmed [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]). The Chlorophyta [<xref ref-type="bibr" rid="B9">9</xref>] comprises the Prasinophyceae, Ulvophyceae, Trebouxiophyceae and Chlorophyceae. The Prasinophyceae represent the most basal divergence of the Chlorophyta [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>] and, although the branching order of the Ulvophyceae, Trebouxiophyceae and Chlorophyceae (UTC) remains uncertain [<xref ref-type="bibr" rid="B12">12</xref>], chloroplast and mitochondrial genome data suggest that the Trebouxiophyceae emerged before the Ulvophyceae and Chlorophyceae [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>]. Complete chloroplast genome sequences have been reported for five chlorophytes: the prasinophyte <italic>Nephroselmis olivacea </italic>[<xref ref-type="bibr" rid="B16">16</xref>], the trebouxiophyte <italic>Chlorella vulgaris </italic>[<xref ref-type="bibr" rid="B17">17</xref>], the ulvophytes <italic>Oltmannsiellopsis viridis </italic>[<xref ref-type="bibr" rid="B13">13</xref>] and <italic>Pseudendoclonium akinetum </italic>[<xref ref-type="bibr" rid="B15">15</xref>] and the chlorophycean green alga <italic>Chlamydomonas reinhardtii </italic>[<xref ref-type="bibr" rid="B18">18</xref>].</p><p>In nearly all photosynthetic lineages investigated thus far in the Streptophyta, the chloroplast genome harbours the same quadripartite structure and the same gene partitioning pattern, genes are densely packed and most of the genes are organized into conserved clusters [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>], the origin of which dates back to the common ancestor of all chloroplasts [<xref ref-type="bibr" rid="B16">16</xref>]. The typical quadripartite structure is characterized by the presence of two copies of a large inverted repeat sequence (IR) separating a small single-copy (SSC) and a large single-copy (LSC) region. The rRNA operon always resides in the IR and is transcribed toward the SSC region. Although the IR readily expands or contracts by gaining or losing genes from the neighbouring single-copy regions [<xref ref-type="bibr" rid="B21">21</xref>], each of the three genomic partitions (IR, SSC and LSC) shows a distinctive and highly conserved gene content. Including 106 to 137 genes, the gene repertoire appears to have progressively shrunk from charophycean green algae to land plants [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. Slight changes in intron composition of the chloroplast genome also occurred during streptophyte evolution [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. The vast majority of introns were likely acquired early during the evolution of charophycean green algae.</p><p>In the Chlorophyta, the chloroplast genome shows extraordinary variability at the levels of its quadripartite structure, global gene organization and intron composition. The cpDNA of the prasinophyte <italic>Nephroselmis </italic>features the largest gene repertoire (128 genes) and the most ancestral features, including the quadripartite structure and gene partitioning pattern observed in streptophytes [<xref ref-type="bibr" rid="B16">16</xref>]. In contrast, all four completely sequenced UTC algal cpDNAs encode fewer genes (94–112) and are substantially rearranged [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Moreover, genes in these cpDNAs are more loosely packed than in <italic>Nephroselmis </italic>and most streptophyte cpDNAs, intergenic spacers usually contain short dispersed repeats (SDRs) and the coding regions of some protein-coding genes are expanded [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. Of the four UTC cpDNAs, that of the trebouxiophyte <italic>Chlorella </italic>has retained the highest degree of ancestral characters; it lacks an IR but has retained many ancestral gene clusters. Both ulvophyte cpDNAs feature an atypical quadripartite structure that deviates from the ancestral type displayed by <italic>Nephroselmis </italic>and streptophyte cpDNAs. In each genome, one of the single-copy regions features many genes characteristic of both the ancestral SSC and LSC regions, whereas the opposite single-copy region features only genes characteristic of the ancestral LSC region. Moreover, the rRNA genes in the IR are transcribed toward the latter single-copy region. From their observations, Pombert <italic>et al</italic>. [<xref ref-type="bibr" rid="B13">13</xref>] concluded that a dozen genes were transferred from the LSC to the SSC region before or soon after emergence of the Ulvophyceae and that the transcription direction of the rRNA genes changed. In the chlorophycean green alga <italic>Chlamydomonas</italic>, the single-copy regions are similar in size and both their gene contents and gene organizations display tremendous differences relative to the same cpDNA regions in ulvophytes, implying that numerous genes were exchanged between opposite single-copy regions during the evolutionary period separating the Ulvophyceae and the chlorophycean clade represented by <italic>Chlamydomonas </italic>(a clade known as the Chlamydomonadales or CW clade [<xref ref-type="bibr" rid="B11">11</xref>]). Gene reshuffling was so extensive that no reliable scenario of gene rearrangements can be predicted to explain the observed differences.</p><p>To gain insights into the various changes that underwent the chloroplast genome in the Chlorophyceae, we have undertaken the complete sequencing of the chloroplast genome from distinct lineages of this class. We report here the 161,452 bp chloroplast genome sequence of <italic>Scenedesmus obliquus</italic>, a member of the lineage that appears to share a sister relationship with the Chlamydomonadales (Sphaeropleales or DO clade) [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref>]. All swimming cells in this lineage are biflagellates with a directly opposed (DO) arrangement of basal bodies, instead of the clockwise (CW) arrangement seen in the Chlamydomonadales. <italic>Scenedesmus </italic>cpDNA was found to be a compact genome that carries as many derived features as its <italic>Chlamydomonas </italic>homologue. It shares with <italic>Chlamydomonas </italic>cpDNA single-copy regions of similar sizes, an almost identical gene repertoire and several derived gene clusters; however, the sets of genes in the single-copy regions of these chlorophycean genomes are very different. These extensive differences in global gene arrangement underscore the remarkable plasticity of the chloroplast genome in the Chlorophyceae.</p></sec><sec><title>Results</title><sec><title>General features</title><p>The <italic>Scenedesmus </italic>cpDNA sequence assembles as a circular molecule of 161,452 bp encoding a total of 96 genes (not counting intron ORFs, free-standing ORFs and duplicated genes) (Fig. <xref ref-type="fig" rid="F1">1</xref> and Table <xref ref-type="table" rid="T1">1</xref>). With an overall A+T content of 73.1%, this chloroplast genome is the most A+T rich among completely sequenced chlorophyte cpDNAs. Two identical copies of an IR sequence of 12,022 bp are separated from one another by single-copy regions differing by only 7.5 kbp in size (SC1 and SC2). For both isomeric forms of the genome, a remarkably strong bias is observed in the distribution of genes between the two DNA strands. In the isomer shown in Fig. <xref ref-type="fig" rid="F1">1</xref>, 82 genes occupy one strand whereas, only 20, including the six present in the IR, reside on the opposite strand. Furthermore, 64 consecutive genes, encompassing more than half of the genome, feature the same polarity. The genes in <italic>Scenedesmus </italic>cpDNA are more tightly packed than in the four other completely sequenced UTC algal cpDNAs, their density (67.2%) being comparable to that found in <italic>Nephroselmis </italic>cpDNA (68.7%). Intergenic spacers have an average size of 465 bp and feature short dispersed repeats (SDRs). A total of nine introns, seven group I and two group II introns, were identified in <italic>Scenedesmus </italic>cpDNA; five of these introns display ORFs.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Gene map of <italic>Scenedesmus </italic>cpDNA and compared patterns of gene partitioning in <italic>Chlamydomonas </italic>and <italic>Scenedesmus </italic>cpDNAs</bold>. The two copies of the rRNA operon-containing IR (IR<sub>A </sub>and IR<sub>B</sub>) are represented by thick lines; the transcription direction of the rRNA genes is indicated by arrows. Genes (filled boxes) on the outside of the map are transcribed in a clockwise direction; those on the inside of the map are transcribed counterclockwise. The colour-code denotes the genomic regions containing the homologous genes in <italic>Chlamydomonas </italic>cpDNA: cyan, SC1; magenta, SC2; yellow, IR. Genes and ORFs absent from <italic>Chlamydomonas </italic>cpDNA are shown in grey. Labelled brackets denote the gene clusters shared specifically by <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs (see Table 4 for the gene content of these clusters). tRNA genes are indicated by the one-letter amino acid code followed by the anticodon in parentheses (Me, elongator methionine: Mf, initiator methionine). Identical copies of the <italic>trnE</italic>(uuc) genes are denoted by asterisks. Introns are represented by open boxes and intron ORFs are denoted by narrow, filled boxes. The intron sequences bordering the <italic>psaA </italic>exons (<italic>psaA </italic>exon 1 and <italic>psaA </italic>exon 2) are spliced in <italic>trans </italic>at the RNA level. Note that only one of the two isomeric forms of the genome is shown here; these isomers differ with respect to the relative orientation of the single-copy regions.</p></caption><graphic xlink:href="1471-2148-6-37-1"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>General features of <italic>Scenedesmus</italic> and other UTC algal cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>Feature</bold></td><td align="center"><bold><italic>Chlorella</italic></bold></td><td align="center"><bold><italic>Oltmannsiellopsis</italic></bold></td><td align="center"><bold><italic>Pseudendoclonium</italic></bold></td><td align="center"><bold><italic>Scenedesmus</italic></bold></td><td align="center"><bold><italic>Chlamydomonas</italic></bold></td></tr></thead><tbody><tr><td align="center">Size (bp)</td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="left"> Total</td><td align="center">150,613</td><td align="center">151,933</td><td align="center">195,867</td><td align="center">161,452</td><td align="center">203,827</td></tr><tr><td align="left"> IR</td><td align="center">– <sup>a</sup></td><td align="center">18,510</td><td align="center">6,039</td><td align="center">12,022</td><td align="center">22,211</td></tr><tr><td align="left"> LSC</td><td align="center">– <sup>a</sup></td><td align="center">33,610</td><td align="center">140,914</td><td align="center">72,440 <sup>b</sup></td><td align="center">81,307 <sup>b</sup></td></tr><tr><td align="left"> SSC</td><td align="center">– <sup>a</sup></td><td align="center">81,303</td><td align="center">42,875</td><td align="center">64,968 <sup>c</sup></td><td align="center">78,088 <sup>c</sup></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center">A+T (%)</td><td align="center">68.4</td><td align="center">59.5</td><td align="center">68.5</td><td align="center">73.1</td><td align="center">65.5</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center">Coding sequences (%) <sup>d</sup></td><td align="center">60.9</td><td align="center">59.2</td><td align="center">62.3</td><td align="center">67.2</td><td align="center">50.1</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center">Genes (no.) <sup>e</sup></td><td align="center">112</td><td align="center">105</td><td align="center">105</td><td align="center">96</td><td align="center">94</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="center">Introns (no.)</td><td></td><td></td><td></td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="left"> Group I</td><td align="center">3</td><td align="center">5</td><td align="center">27</td><td align="center">7</td><td align="center">5</td></tr><tr><td align="left"> Group II</td><td align="center">0</td><td align="center">0</td><td align="center">0</td><td align="center">2</td><td align="center">2</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Because <italic>Chlorella </italic>cpDNA lacks an IR, only the total size of this genome is given.</p><p><sup>b</sup>In this study, this region is designated SC1 rather than LSC because it displays major differences in gene content relative to the LSC region in <italic>Mesostigma </italic>and <italic>Nephroselmis </italic>cpDNAs [13].</p><p><sup>c </sup>This region is designated SC2 rather than SSC because it displays major differences in gene content relative to the SSC region in <italic>Mesostigma </italic>and <italic>Nephroselmis </italic>cpDNAs [13].</p><p><sup>d</sup>Conserved genes, unique ORFs and introns were considered as coding sequences.</p><p><sup>e</sup>Genes present in the IR were counted only once. Unique ORFs and intron ORFs were not taken into account.</p></table-wrap-foot></table-wrap></sec><sec><title>Gene content and gene structure</title><p>The gene repertoire of the <italic>Scenedesmus </italic>genome differs from that of its <italic>Chlamydomonas </italic>homologue only by the presence of two additional genes, <italic>infA </italic>and <italic>rpl12 </italic>(Table <xref ref-type="table" rid="T1">1</xref>). As in <italic>Chlamydomonas </italic>cpDNA, two identical copies of the <italic>trnE</italic>(uuc) gene are found on opposite strands outside the IR. Five ORFs with more than 65 codons were identified in intergenic regions (Fig. <xref ref-type="fig" rid="F1">1</xref>). The largest one, ORF932, resides in SC1 and is part of the long segment carrying genes with identical polarity. The protein encoded by this ORF shows limited sequence similarity with bacterial reverse transcriptases, the observed similarity being restricted to domain X. The four remaining ORFs display no homology with any known DNA sequences. All five ORFs differ from the conserved protein-coding genes at the levels of codon usage and nucleotide composition.</p><p>As is the case for <italic>Chlamydomonas </italic>cpDNA, the <italic>rpoB </italic>and <italic>rps2 </italic>genes in <italic>Scenedesmus </italic>cpDNA each occur as two contiguous ORFs (Fig. <xref ref-type="fig" rid="F1">1</xref>) and two separate genes, <italic>clpP </italic>and <italic>rps3</italic>, have extensions in their coding sequences that are absent from other chlorophyte and streptophyte cpDNAs (Table <xref ref-type="table" rid="T2">2</xref>). The extra coding sequences in each of these genes share the same insertion site in the two chlorophycean algae. Note that the intein gene previously identified within the <italic>Chlamydomonas eugametos clpP </italic>gene [<xref ref-type="bibr" rid="B25">25</xref>] lies at a position different from the insertion sequence reported here for <italic>Scenedesmus clpP</italic>. Six additional protein-coding genes in both <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs resemble their <italic>Chlorella </italic>and/or ulvophyte homologues in exhibiting expanded coding regions (Table <xref ref-type="table" rid="T2">2</xref>). In sharp contrast, the rRNA and tRNA genes of UTC algae show less than 1% deviation in size relative to their homologues in <italic>Nephroselmis </italic>and <italic>Mesostigma</italic>.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Expanded genes in <italic>Scenedesmus</italic> and other UTC algal cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Gene</bold></td><td align="center" colspan="2"><bold><italic>Chlorella</italic></bold></td><td align="center" colspan="2"><bold><italic>Oltmannsiellopsis</italic></bold></td><td align="center" colspan="2"><bold><italic>Pseudendoclonium</italic></bold></td><td align="center" colspan="2"><bold><italic>Scenedesmus</italic></bold></td><td align="center" colspan="2"><bold><italic>Chlamydomonas</italic></bold></td></tr><tr><td></td><td colspan="10"><hr></hr></td></tr><tr><td></td><td align="center"><bold>Size (bp)</bold></td><td align="center"><bold>Factor </bold><sup>a</sup></td><td align="center"><bold>Size (bp)</bold></td><td align="center"><bold>Factor </bold><sup>a</sup></td><td align="center"><bold>Size (bp)</bold></td><td align="center"><bold>Factor </bold><sup>a</sup></td><td align="center"><bold>Size (bp)</bold></td><td align="center"><bold>Factor </bold><sup>a</sup></td><td align="center"><bold>Size (bp)</bold></td><td align="center"><bold>Factor </bold><sup>a</sup></td></tr></thead><tbody><tr><td align="left"><italic>cemA </italic><sup>b</sup></td><td align="center">801</td><td align="center">1.6</td><td align="center">1059</td><td align="center">2.2</td><td align="center">909</td><td align="center">1.8</td><td align="center">1278</td><td align="center">2.6</td><td align="center">1503</td><td align="center">3.1</td></tr><tr><td align="left"><italic>clpP</italic></td><td align="center">606</td><td align="center">0.9</td><td align="center">588</td><td align="center">0.9</td><td align="center">597</td><td align="center">0.9</td><td align="center">1614</td><td align="center">2.3</td><td align="center">1575</td><td align="center">2.3</td></tr><tr><td align="left"><italic>ftsH </italic><sup>b</sup></td><td align="center">5163</td><td align="center">1.9</td><td align="center">6879</td><td align="center">2.6</td><td align="center">7791</td><td align="center">2.9</td><td align="center">10998</td><td align="center">4.1</td><td align="center">8916</td><td align="center">3.3</td></tr><tr><td align="left"><italic>infA </italic><sup>b</sup></td><td align="center">240</td><td align="center">1.2</td><td align="center">222</td><td align="center">1.1</td><td align="center">330</td><td align="center">1.6</td><td align="center">306</td><td align="center">1.5</td><td align="center">– <sup>c</sup></td><td align="center">– <sup>c</sup></td></tr><tr><td align="left"><italic>rpoA</italic></td><td align="center">837</td><td align="center">0.9</td><td align="center">1527</td><td align="center">1.6</td><td align="center">1734</td><td align="center">1.8</td><td align="center">1437</td><td align="center">1.5</td><td align="center">2213 <sup>d</sup></td><td align="center">2.3</td></tr><tr><td align="left"><italic>rpoB</italic></td><td align="center">3906</td><td align="center">1.2</td><td align="center">4251</td><td align="center">1.3</td><td align="center">6537</td><td align="center">2.0</td><td align="center">4896 <sup>e</sup></td><td align="center">1.5</td><td align="center">4967 <sup>e</sup></td><td align="center">1.5</td></tr><tr><td align="left"><italic>rpoC1</italic></td><td align="center">2511</td><td align="center">1.3</td><td align="center">3066</td><td align="center">1.5</td><td align="center">4737</td><td align="center">2.4</td><td align="center">4590</td><td align="center">2.3</td><td align="center">5739 <sup>e</sup></td><td align="center">2.9</td></tr><tr><td align="left"><italic>rpoC2</italic></td><td align="center">4689</td><td align="center">1.3</td><td align="center">5580</td><td align="center">1.5</td><td align="center">10389</td><td align="center">2.8</td><td align="center">7659</td><td align="center">2.1</td><td align="center">9363</td><td align="center">2.5</td></tr><tr><td align="left"><italic>rps2</italic></td><td align="center">804</td><td align="center">1.2</td><td align="center">717</td><td align="center">1.0</td><td align="center">777</td><td align="center">1.1</td><td align="center">2747 <sup>e</sup></td><td align="center">4.0</td><td align="center">2731 <sup>e</sup></td><td align="center">4.0</td></tr><tr><td align="left"><italic>rps3</italic></td><td align="center">696</td><td align="center">1.1</td><td align="center">708</td><td align="center">1.1</td><td align="center">690</td><td align="center">1.1</td><td align="center">2103</td><td align="center">3.3</td><td align="center">2139</td><td align="center">3.3</td></tr><tr><td align="left"><italic>rps18 </italic><sup>b</sup></td><td align="center">312</td><td align="center">1.4</td><td align="center">237</td><td align="center">1.1</td><td align="center">258</td><td align="center">1.2</td><td align="center">567</td><td align="center">2.6</td><td align="center">414</td><td align="center">1.9</td></tr><tr><td align="left"><italic>ycf1 </italic><sup>b</sup></td><td align="center">2460</td><td align="center">2.0</td><td align="center">2427</td><td align="center">2.0</td><td align="center">2505</td><td align="center">2.0</td><td align="center">7008</td><td align="center">5.7</td><td align="center">5988</td><td align="center">4.9</td></tr></tbody></table><table-wrap-foot><p><sup>a </sup>Expansion factor relative to the corresponding <italic>Mesostigma </italic>gene. Each value was obtained by dividing the size of the UTC algal gene by the size of the corresponding <italic>Mesostigma </italic>gene.</p><p><sup>b </sup>Genes also expanded in streptophyte cpDNAs. Note that the <italic>ftsH </italic>gene is designated as <italic>ycf2 </italic>in <italic>Chlorella, Chlamydomonas </italic>and land plants cpDNAs.</p><p><sup>c </sup><italic>infA </italic>is missing in <italic>Chlamydomonas </italic>cpDNA</p><p><sup>d </sup>Size derived from our unpublished sequence data. A portion of the <italic>rpoA </italic>region is not annotated in GenBank:<ext-link ext-link-type="gen" xlink:href="NC_005353">NC_005353</ext-link> as a result of a sequencing error introducing a frameshift mutation.</p><p><sup>e </sup>Gene occurring as two independent ORFs. The indicated size includes the spacer separating the ORFs.</p></table-wrap-foot></table-wrap></sec><sec><title>Gene partitioning and gene clustering</title><p>Our comparison of the gene complements found in the three genomic regions of <italic>Scenedesmus </italic>cpDNA with those observed in the <italic>Chlamydomonas </italic>genome reveals dramatic differences in the gene composition of the single-copy regions (Fig. <xref ref-type="fig" rid="F1">1</xref>). In each single-copy region of <italic>Scenedesmus </italic>cpDNA, we find numerous genes whose homologues map to the opposite single-copy region in the <italic>Chlamydomonas </italic>genome. Of the 43 genes displayed by <italic>Scenedesmus </italic>SC1 (largest single-copy region), 24 are located in the SC1 region (largest single-copy region) of <italic>Chlamydomonas </italic>(genes shown in cyan in Fig. <xref ref-type="fig" rid="F1">1</xref>), whereas all the others map to the alternate SC2 region (genes shown in magenta). Similarly, 19 of the 47 genes present in <italic>Scenedesmus </italic>SC2 reside in the SC1 region of <italic>Chlamydomonas</italic>, whereas all the others lie in the opposite SC2 region. Note here that the single-copy regions of both <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>were arbitrarily designated (see Table <xref ref-type="table" rid="T1">1</xref>) and that the genes shared by the SC1 or SC2 regions of these algae were not necessarily confined to the same single-copy region in the chloroplast genome of the last common ancestor of the two algae. Assuming that the SC1 or SC2 regions in <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs are equivalent, it would be necessary to propose that the transcription direction of the rRNA operon was altered during the evolution of chlorophycean green algae concurrently with the extensive exchanges of genes that took place between the single-copy regions.</p><p>To identify the ancestral clusters carried by <italic>Scenedesmus </italic>cpDNA as well as the derived clusters that are shared with other UTC algal cpDNAs, we carried out a detailed comparative analysis of gene order. As a first step, we investigated the 24 gene clusters present in both <italic>Mesostigma </italic>and <italic>Nephroselmis </italic>cpDNAs and found that <italic>Scenedesmus </italic>cpDNA is identical to its <italic>Chlamydomonas </italic>counterpart in terms of composition of ancestral clusters (Fig. <xref ref-type="fig" rid="F2">2</xref>). Both chlorophycean green algal cpDNAs harbour a single, intact ancestral cluster (<italic>psbB-T-/N-/H</italic>, with the slash indicating a change in gene polarity) and the remains of four other ancestral clusters: altogether, these conserved clusters encode 20 genes. These observations confirm the notion that the lowest degree of ancestral clusters among UTC algal cpDNAs is found in the chlorophycean lineage [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B26">26</xref>]. As previously reported by Pombert <italic>et al</italic>. [<xref ref-type="bibr" rid="B13">13</xref>], the chloroplast genome of the trebouxiophyte <italic>Chlorella </italic>exhibits the highest conservation of ancestral clusters among completely sequenced UTC algal cpDNAs; it has retained 11 intact clusters and four partially conserved ones that include 62 genes. The chloroplast genomes of the ulvophytes <italic>Oltmannsiellopsis </italic>and <italic>Pseudendoclonium </italic>have retained only five of the 24 ancestral clusters in an intact state as well as 13 and 10 partially conserved clusters, respectively.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Conservation of ancestral gene clusters in <italic>Scenedesmus </italic>and other UTC algal cpDNAs</bold>. Black boxes represent the 89 genes found in the 24 clusters shared by <italic>Mesostigma </italic>and <italic>Nephroselmis </italic>cpDNAs as well as the genes in UTC algal cpDNAs that have retained the same order as those in these ancestral clusters. For each genome, the set of genes making up each of the identified clusters (either an intact or fragmented ancestral cluster) is shown as black boxes connected by a horizontal line. Black boxes that are contiguous but not linked together indicate that the corresponding genes are not adjacent on the genome. Gray boxes denote genes in UTC algal cpDNAs that have been relocated elsewhere on the chloroplast genome; open boxes denote genes that have disappeared from the chloroplast genome. Although the <italic>rpl22 </italic>gene is missing from <italic>Nephroselmis </italic>cpDNA, it is shown as belonging to the large ribosomal protein cluster equivalent to the contiguous <italic>S10</italic>, <italic>spc </italic>and α operons of <italic>Escherichia coli </italic>because it is present in this cluster in the cpDNAs of <italic>Mesostigma</italic>, streptophytes and algae from other lineages. Note also that the <italic>psbB </italic>cluster of <italic>Oltmannsiellopsis </italic>and <italic>Pseudendoclonium </italic>cpDNAs differs from the ancestral cluster found in other genomes by the presence of <italic>psbN </italic>on the alternate DNA strand.</p></caption><graphic xlink:href="1471-2148-6-37-2"/></fig><p>The overall gene order in <italic>Scenedesmus </italic>cpDNA was also compared to the global gene arrangements of other completely sequenced UTC algal cpDNAs. As expected, we found that <italic>Scenedesmus </italic>cpDNA most closely resembles its <italic>Chlamydomonas </italic>counterpart at the level of derived gene clusters. The two chlorophycean genomes share 11 clusters that are not conserved in <italic>Chlorella</italic>, <italic>Oltmannsiellopsis </italic>and <italic>Pseudendoclonium </italic>cpDNAs (Fig. <xref ref-type="fig" rid="F1">1</xref> and Table <xref ref-type="table" rid="T4">4</xref>). Three of these clusters (clusters 9, 10 and 11) are adjacent on the <italic>Scenedesmus </italic>genome, whereas no shared derived clusters show contiguity on <italic>Chlamydomonas </italic>cpDNA. Only the <italic>rpoB</italic>-/<italic>psbF</italic>-<italic>psbL </italic>cluster originated by fusion of a fragment of ancestral cluster (<italic>psbF</italic>-<italic>psbL</italic>) with other genes. Two of the 11 derived clusters, <italic>petA</italic>-<italic>petD </italic>and <italic>trnL</italic>(uag)<italic>-clpP </italic>clusters, as well segments of the <italic>atpA-</italic><underline><italic>psbI-cemA</italic></underline> and <italic>psbE-rps9-</italic><underline><italic>ycf4-ycf3</italic></underline> clusters (relevant segments are underlined) have been previously identified not only in <italic>C. reinhardtii </italic>and its close relative <italic>C. gelatinosa </italic>[<xref ref-type="bibr" rid="B27">27</xref>] but also in the distantly pair of interfertile algae <italic>C. eugametos </italic>and <italic>C. moewusii </italic>[<xref ref-type="bibr" rid="B26">26</xref>] and their close relative <italic>C. pitschmannii </italic>[<xref ref-type="bibr" rid="B28">28</xref>], indicating their conservation in the Chlamydomonodales. The <italic>Scenedesmus </italic>chloroplast genome shares no specific gene clusters with <italic>Oltmannsiellopsis </italic>and <italic>Pseudendoclonium </italic>cpDNA and only one pair of genes (<italic>psaJ-rps12</italic>, a subset of cluster 3 in Table <xref ref-type="table" rid="T4">4</xref>) with <italic>Chlorella </italic>cpDNA.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Introns in <italic>Scenedesmus</italic> cpDNA and homologous introns at identical gene locations in other green algal cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center" colspan="4"><bold><italic>Scenedesmus </italic>introns</bold></td><td align="left" colspan="2"><bold>Homologous introns</bold></td></tr></thead><tbody><tr><td align="left"><bold>Designation</bold></td><td align="center"><bold>Subgroup </bold><sup>a</sup></td><td align="center"><bold>ORF location </bold><sup>b</sup></td><td align="center"><bold>ORF type </bold><sup>c</sup></td><td align="left"><bold>Green alga </bold><sup>d</sup><bold>/Intron number </bold><sup>e</sup></td><td align="left"><bold>Accession no.</bold></td></tr><tr><td colspan="6"><hr></hr></td></tr><tr><td align="left">So.<italic>L(uaa)</italic>.1</td><td align="center">IC3</td><td align="center">–</td><td align="center">–</td><td align="left"><italic>Bryopsis plumosa </italic>(U)</td><td align="left">[GenBank:<ext-link ext-link-type="gen" xlink:href="M61159">M61159</ext-link>]</td></tr><tr><td></td><td></td><td></td><td></td><td align="left"><italic>Chlorella vulgaris </italic>(T)</td><td align="left">[GenBank:<ext-link ext-link-type="gen" xlink:href="NC_001865">NC_001865</ext-link>]</td></tr><tr><td align="left">So.<italic>psaB</italic>.1</td><td align="center">IA1</td><td align="center">–</td><td align="center">–</td><td align="left"><italic>Chlamydomonas moewusii </italic>(C)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="M90641">M90641</ext-link>]</td></tr><tr><td align="left">So.<italic>psbA</italic>.1</td><td align="center">IA1</td><td align="center">L5</td><td align="center">H-N-H</td><td align="left">–</td><td align="left">-</td></tr><tr><td align="left">So.<italic>psbA</italic>.2</td><td align="center">IA1</td><td align="center">–</td><td align="center">–</td><td align="left">–</td><td align="left">-</td></tr><tr><td align="left">So.<italic>psbA</italic>.3</td><td align="center">IA1</td><td align="center">L5</td><td align="center">H-N-H</td><td align="left"><italic>Pseudendoclonium akinetum </italic>i6 (U)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="AY835431">AY835431</ext-link>]</td></tr><tr><td align="left">So.<italic>rrl</italic>.1</td><td align="center">IB4</td><td align="center">L6</td><td align="center">LAGLIDADG</td><td align="left"><italic>Chlamydomonas eugametos </italic>i5 (C)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="Z17234">Z17234</ext-link>]</td></tr><tr><td align="left">So.<italic>rrl</italic>.2</td><td align="center">IA3</td><td align="center">L6</td><td align="center">LAGLIDADG</td><td align="left"><italic>Chlamydomonas reinhardtii </italic>(C)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="NC_005353">NC_005353</ext-link>]</td></tr><tr><td></td><td></td><td></td><td></td><td align="left"><italic>Chlorella vulgaris </italic>(T)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="NC_001865">NC_001865</ext-link>]</td></tr><tr><td></td><td></td><td></td><td></td><td align="left"><italic>Oltmannsiellopsis viridis </italic>i3 (U)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="DQ291132">DQ291132</ext-link>]</td></tr><tr><td></td><td></td><td></td><td></td><td align="left"><italic>Pseudendoclonium akinetum </italic>(U)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="AY835431">AY835431</ext-link>]</td></tr><tr><td align="left">So.<italic>petD</italic>.1</td><td align="center">IIA</td><td align="center">DIV</td><td align="center">RT-X-Zn</td><td align="left">–</td><td align="left">–</td></tr><tr><td align="left">So.<italic>psaA</italic>.1</td><td align="center">IIB</td><td align="center">–</td><td align="center">–</td><td align="left"><italic>Chlamydomonas reinhardtii </italic>i2 (C)</td><td align="left">[GenBank: <ext-link ext-link-type="gen" xlink:href="NC_005353">NC_005353</ext-link>]</td></tr></tbody></table><table-wrap-foot><p><sup>a </sup>Group I introns were classified according to Michel and Westhof [66], whereas classification of group II introns was according to Michel <italic>et al</italic>. [65].</p><p><sup>b </sup>L followed by a number refers to the loop extending the base-paired region identified by the number; D refers to a domain of group II intron secondary structure.</p><p><sup>c </sup>For group I intron ORFs, the conserved motif in the predicted homing endonuclease is given; for the <italic>petD </italic>intron ORF, RT, X and Zn refer to the reverse transcriptase, maturase and nuclease domains of reverse transcriptases, respectively.</p><p><sup>d </sup>For the two <italic>rrl </italic>introns, only the homologues identified in completely sequenced genomes are given. The complete lists of introns homologous to So. <italic>rrl</italic>.1 and So.<italic>rrl</italic>.2 can be obtained in Turmel <italic>et al</italic>. [35] and Pombert <italic>et al</italic>. [13], respectively. The letter in parentheses indicates the chlorophyte lineage comprising the green alga indicated. U, Ulvophyceae; T, Trebouxiophyceae; C, Chlorophyceae.</p><p><sup>e </sup>The intron number is given when more than one intron is present.</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Derived gene clusters shared by <italic>Scenedesmus</italic> and <italic>Chlamydomonas</italic> cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Cluster no. </bold><sup>a</sup></td><td align="left"><bold>Gene composition</bold></td></tr></thead><tbody><tr><td align="left">1</td><td align="left"><italic>rps3-rpoC2</italic></td></tr><tr><td align="left">2</td><td align="left"><italic>rpoBb-rpoBa-/psbF-psbL</italic><sup>b</sup></td></tr><tr><td align="left">3</td><td align="left"><italic>psbD-psaAa-psbJ-atpI-psaJ-rps12</italic></td></tr><tr><td align="left">4</td><td align="left"><italic>rps7-atpE</italic></td></tr><tr><td align="left">5</td><td align="left"><italic>psaC-petL-trnN</italic>(guu)</td></tr><tr><td align="left">6</td><td align="left"><italic>atpA-psbI-cemA</italic></td></tr><tr><td align="left">7</td><td align="left"><italic>trnL</italic>(uag)<italic>-clpP</italic></td></tr><tr><td align="left">8</td><td align="left"><italic>ccsA-psbZ-psbM</italic></td></tr><tr><td align="left">9</td><td align="left"><italic>petA-petD</italic></td></tr><tr><td align="left">10</td><td align="left"><italic>chlL-rpl36</italic></td></tr><tr><td align="left">11</td><td align="left"><italic>psbE-rps9-ycf4-ycf3</italic></td></tr></tbody></table><table-wrap-foot><p><sup>a </sup>Clusters are labelled as in Fig.1.</p><p><sup>b </sup>The slash indicates a change in gene polarity.</p></table-wrap-foot></table-wrap></sec><sec><title>Introns</title><p>The seven group I introns of <italic>Scenedesmus </italic>interrupt four genes: three introns occur in <italic>psbA</italic>, two in <italic>rrl </italic>and the two others in <italic>psaB </italic>and <italic>trnL</italic>(uaa). These introns fall within four different subgroups (IA1, IA3, IB4 and IC3), with the IA1 subgroup including the four introns present in <italic>psaB </italic>and <italic>psbA </italic>(Table <xref ref-type="table" rid="T3">3</xref>). At 255 bp, the IC3 intron in <italic>trnL</italic>(uaa) is the smallest of the <italic>Scenedesmus </italic>introns. As homologous introns are inserted at the same position not only in the chloroplast <italic>trnL</italic>(uaa) genes of the chlorophytes <italic>Bryopsis </italic>and <italic>Chlorella </italic>(Table <xref ref-type="table" rid="T3">3</xref>) but also in the <italic>trnL</italic>(uaa) genes of streptophytes and algae from other lineages, this intron is thought to have been inherited by vertical inheritance from the common ancestor of all chloroplasts [<xref ref-type="bibr" rid="B29">29</xref>]. The IA3 and IB4 introns in <italic>Scenedesmus rrl </italic>are also positionally and structurally homologous to previously reported introns in green plant cpDNAs (Table <xref ref-type="table" rid="T3">3</xref>). Although the four IA1 introns revealed relatively poor sequence similarity with one another, two of these introns, So.<italic>psaB</italic>.1 and So.<italic>psbA</italic>.3, were found to be clearly homologous to introns inserted at identical gene locations in <italic>Chlamydomonas moewusii </italic>and <italic>Pseudendoclonium </italic>cpDNAs, respectively (Table <xref ref-type="table" rid="T3">3</xref>). So. <italic>psbA</italic>.3 and its <italic>Pseudendoclonium </italic>homologue display not only similar primary sequences and secondary structures, but also similar ORFs encoding potential homing endonucleases carrying the H-N-H motif (44% identity at the protein sequence level). The two other IA1 introns of <italic>Scenedesmus</italic>, So.<italic>psbA</italic>.1 and So.<italic>psbA</italic>.2, represent unique insertion positions in the <italic>psbA </italic>gene; they are located only 5 bp and 6 bp away from the second and fourth introns in <italic>Pseudendoclonium psbA</italic>, respectively. For these two pairs of closely linked introns, similarity was found to be limited to the So. <italic>psbA</italic>.1 intron-encoded H-N-H homing endonuclease, which shares 33% sequence identity with the protein encoded by second intron in <italic>Pseudendoclonium psbA</italic>.</p><p>One of the two group II introns of <italic>Scenedesmus </italic>is spliced in <italic>trans </italic>at the RNA level. This intron occurs in <italic>psaA </italic>and is inserted at exactly the same site as the second of the two <italic>trans</italic>-spliced introns in <italic>Chlamydomonas psaA </italic>[<xref ref-type="bibr" rid="B30">30</xref>] (Table <xref ref-type="table" rid="T3">3</xref>). Like its <italic>Chlamydomonas </italic>homologue, it has no ORF. These positionally identical <italic>Scenedesmus </italic>and <italic>Chlamydomonas psaA </italic>introns both belong to the IIB subgroup, share poor sequence similarity and are both fragmented within domain IV (Fig. <xref ref-type="fig" rid="F3">3</xref>). The second group II intron in <italic>Scenedesmus </italic>cpDNA lies within <italic>petD </italic>and has no known homologue (Table <xref ref-type="table" rid="T3">3</xref>). This intron harbours in its domain IV an ORF encoding a reverse transcriptase [<xref ref-type="bibr" rid="B31">31</xref>] with the typical reverse transcriptase, maturase and nuclease domains [<xref ref-type="bibr" rid="B32">32</xref>].</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Secondary structure model of the <italic>Scenedesmus psaA </italic>intron</bold>. Intron modelling was according to the nomenclature proposed by Michel <italic>et al</italic>. [65]. Exon sequences are shown in lowercase letters. Roman numerals specify the six major structural domains of group II introns. Tertiary interactions are represented by blocked arrows. EBS and IBS refer to exon-binding and intron-binding sites, respectively. Numbers inside the loops denote the sizes of these regions. The 5' and 3' strand polarities of the <italic>psaA</italic>a and <italic>psaA</italic>b transcripts are indicated by arrows.</p></caption><graphic xlink:href="1471-2148-6-37-3"/></fig></sec><sec><title>Short dispersed repeats</title><p>SDR elements were identified in many intergenic spacers and some coding regions of <italic>Scenedesmus </italic>cpDNA; however, they are less numerous than those found in other UTC algal cpDNAs (Table <xref ref-type="table" rid="T5">5</xref>). The repeats ≥ 30 bp in the <italic>Scenedesmus </italic>genome represent 8.7% of the total size of the intergenic spacers, whereas the fraction of the intergenic regions represented by such repeats in <italic>Chlamydomonas </italic>cpDNA reaches 31.9%. Analysis of the most abundant repeat elements in the <italic>Scenedesmus </italic>genome using RepeatFinder revealed three distinct groups of repeat units featuring sequences of 15 or 16 bp (Table <xref ref-type="table" rid="T6">6</xref>). With a total of 41 copies, repeat unit B represents the most abundant group of repeats. Repeat units A and C are restricted to intergenic regions, whereas some copies of unit B are also found within the coding regions of five genes that are expanded relative to their <italic>Mesostigma </italic>counterparts, <italic>i.e. cemA</italic>, <italic>ftsH</italic>, <italic>infA</italic>, <italic>rpoBa </italic>and <italic>rpoC2 </italic>(Fig. <xref ref-type="fig" rid="F4">4</xref> and see Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Abundance of SDRs in <italic>Scenedesmus</italic> and other UTC algal cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>cpDNA</bold></td><td></td><td align="center" colspan="2"><bold>Number of repeats </bold><sup>a</sup></td><td align="center" colspan="3"><bold>Non-overlapping repeats ≥ 30 bp</bold><sup>b</sup></td></tr><tr><td></td><td></td><td colspan="5"><hr></hr></td></tr><tr><td></td><td align="center"><bold>Maximal size of repeats (bp)</bold></td><td align="center"><bold>≥ 30 bp</bold></td><td align="center"><bold>≥ 45 bp</bold></td><td align="center"><bold>Total size (bp) </bold></td><td align="center"><bold>Fraction of genome (%)</bold></td><td align="center"><bold>Fraction of intergenic regions (%)</bold></td></tr></thead><tbody><tr><td align="left"><italic>Chlorella</italic></td><td align="center">84</td><td align="center">269</td><td align="center">44</td><td align="center">11,743</td><td align="center">7.8</td><td align="center">20.8</td></tr><tr><td align="left"><italic>Oltmannsiellopsis</italic></td><td align="center">172</td><td align="center">1,205</td><td align="center">161</td><td align="center">18,033</td><td align="center">11.9</td><td align="center">30.1</td></tr><tr><td align="left"><italic>Pseudendoclonium</italic></td><td align="center">171</td><td align="center">1,047</td><td align="center">203</td><td align="center">10,073</td><td align="center">5.1</td><td align="center">13.6</td></tr><tr><td align="left"><italic>Scenedesmus</italic></td><td align="center">112</td><td align="center">86</td><td align="center">21</td><td align="center">4,817</td><td align="center">3.0</td><td align="center">8.7</td></tr><tr><td align="left"><italic>Chlamydomonas</italic></td><td align="center">221</td><td align="center">3,247</td><td align="center">551</td><td align="center">32,244</td><td align="center">15.8</td><td align="center">31.9</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Repeats with identical sequences on the same DNA strand or different DNA strands were identified using REPuter.</p><p><sup>b</sup>The repeats identified with REPuter were annotated and masked on the genome sequence using RepeatMasker.</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T6"><label>Table 6</label><caption><p>SDR repeat units in <italic>Scenedesmus</italic> cpDNA</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Designation</bold></td><td align="center"><bold>Size (bp)</bold></td><td align="left"><bold>Sequence</bold></td><td align="center" colspan="2"><bold>Copy number </bold><sup>a</sup></td></tr><tr><td></td><td></td><td></td><td colspan="2"><hr></hr></td></tr><tr><td></td><td></td><td></td><td align="center"><bold>Forward strand</bold></td><td align="center"><bold>Reverse strand</bold></td></tr></thead><tbody><tr><td align="left">A</td><td align="center">15</td><td align="left">TTTACGCTTTTTTTC</td><td align="center">17</td><td align="center">10</td></tr><tr><td align="left">B</td><td align="center">15</td><td align="left">TTCTTCTTCATTTTT</td><td align="center">22</td><td align="center">19</td></tr><tr><td align="left">C</td><td align="center">16</td><td align="left">TGCTTTGCTGCTTTTT</td><td align="center">16</td><td align="center">22</td></tr></tbody></table><table-wrap-foot><p><sup>a</sup>Copies of each SDR unit were identified using FINDPATTERNS (one mismatch was allowed).</p></table-wrap-foot></table-wrap><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Positions of SDR elements in <italic>Scenedesmus </italic>cpDNA</bold>. Lines connect cpDNA loci displaying repeats ≥30 bp with identical sequences either on the same strand or different strands. For this analysis carried out with REPuter, one copy of the IR sequence (IR<sub>A</sub>) was deleted; the location of this deleted copy is indicated by the long, vertical arrow. The loci containing repeat units A, B and C are represented by symbols of different shapes outside the gene map: triangles, repeat unit A; squares, repeat unit B; circles, repeat unit C. Filled symbols denote the repeats occupying the + strand; open symbols denote the repeats found on the alternate strand. A symbol accompanied by an asterisk indicates the presence of two or more copies. Small arrows point to gene coding regions containing copies of repeat unit B.</p></caption><graphic xlink:href="1471-2148-6-37-4"/></fig><p>Although repeat units A, B and C occur on both strands of <italic>Scenedesmus </italic>cpDNA, they are not evenly distributed throughout the genome (Fig. <xref ref-type="fig" rid="F4">4</xref>). Many intergenic spacers entirely lack copies of these repeat units and tend to be clustered in distinct cpDNA regions (<italic>e.g</italic>., the regions in the vicinity of <italic>petD</italic>, <italic>psbA </italic>and <italic>rpoC1</italic>). On the other hand, numerous intergenic spacers are populated by two or more copies of the same repeat unit and/or by several copies representing different units (Fig. <xref ref-type="fig" rid="F4">4</xref>). The repeats in the latter spacers often form longer repeated sequences that are not randomly distributed on the <italic>Scenedesmus </italic>genome. In Fig. <xref ref-type="fig" rid="F4">4</xref>, it can be seen that the great majority of the repeats exceeding 30 bp in size are confined to one half of the genome. While most of the intergenic spacers harbouring SDRs reside in regions that differ in gene order relative the <italic>Chlamydomonas </italic>genome, some occur in shared, derived gene clusters (clusters 1, 2, 3, 5, 6 and 7).</p><p>In the intergenic spacers displaying copies of the same repeat unit, these copies are often arranged in direct orientation (Fig. <xref ref-type="fig" rid="F4">4</xref>) and separated by 23–25 bp; identical repeats also occur on different strands but, in this configuration, their distances are highly variable (8–116 bp). In intergenic regions of <italic>Chlamydomonas </italic>cpDNA, repeated elements appear to show arrangements similar to those reported here for <italic>Scenedesmus </italic>cpDNA. In contrast, in <italic>Oltmannsiellopsis </italic>and <italic>Pseudendoclonium </italic>cpDNAs, SDRs occur predominantly as stem-loop structures [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. None of the repeat units of <italic>Scenedesmus </italic>cpDNA was identified as being part of SDRs in other UTC algal cpDNAs.</p></sec></sec><sec><title>Discussion</title><p>Our comparative analyses of the <italic>Scenedesmus </italic>chloroplast genome with previously sequenced chlorophyte cpDNAs highlight the remarkable plasticity displayed by the chloroplast genome in the Chlorophyceae. As expected, we found that <italic>Scenedesmus </italic>cpDNA shows the most similarities with <italic>Chlamydomonas </italic>cpDNA. The almost identical gene repertoires displayed by these chlorophycean green algal cpDNAs contrasts with the tremendous differences they exhibit at the level of gene order and pattern of gene partitioning between the single-copy regions. This highly variable gene organization is not the only surprising result that emerged from our study. Three other features of the <italic>Scenedesmus </italic>genome were found to be peculiar: (1) its high gene density, which mirrors that found for <italic>Nephroselmis </italic>cpDNA and diverges from the tendency of previously studied UTC algal cpDNAs to grow in size by gaining sequences in intergenic regions and selected gene coding regions [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B18">18</xref>], (2) the low abundance of its repeated sequences, which represents the lowest level identified thus far in a UTC algal cpDNA and (3) the strongly biased distribution of its genes between the two DNA strands.</p><p>The features shared by <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs provide information on the cpDNA of the last common ancestor of DO and CW green algae; however, the portrait that could be drawn for this ancestral genome is rather sketchy owing to the major differences observed at the levels of gene order and intron content. We infer that the chloroplast genome of the last common ancestor of DO and CW green algae harboured a total of 96 genes, including a duplicated <italic>trnE</italic>(uuc) gene, that one third of these genes were organized in the same order as those found in the 11 gene clusters specifically shared by <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs, that both <italic>rps2 </italic>and <italic>rpoB </italic>were fragmented in two pieces and that <italic>clpP </italic>and <italic>rps3 </italic>each displayed an insertion sequence. During the evolution of chlamydomonads, <italic>infA </italic>and <italic>rpl12 </italic>genes disappeared from the chloroplast genome and <italic>rpoC1 </italic>was broken into two separate reading frames. We also predict with confidence that the ancestral genome contained introns in <italic>rrl </italic>and <italic>psaA </italic>at the same positions as those shared by <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs as well as introns in <italic>trnL</italic>(uaa), <italic>psaB</italic>, and <italic>rrl </italic>at the same positions as those shared by <italic>Scenedesmus </italic>and other chlorophyte cpDNAs. Homologues of all these introns, with the exception of the <italic>trnL</italic>(uaa) intron, have been identified in chlamydomonads distantly related to <italic>C. reinhardtii </italic>(<italic>rrl</italic>, [<xref ref-type="bibr" rid="B33">33</xref>-<xref ref-type="bibr" rid="B35">35</xref>]; <italic>psaA</italic>, [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B36">36</xref>]; <italic>psaB</italic>, [<xref ref-type="bibr" rid="B37">37</xref>]). In contrast, the <italic>trnL</italic>(uaa) intron shows a broader distribution among green algae and is thought to have been inherited from the common ancestor of all chloroplasts [<xref ref-type="bibr" rid="B29">29</xref>]. Undoubtedly, as reported for <italic>C. reinhardtii </italic>cpDNA [<xref ref-type="bibr" rid="B38">38</xref>], the <italic>psaA </italic>intron of the last common ancestor of DO and CW green algae featured a break in domain IV, the two <italic>psaA </italic>exons were unlinked and transcribed independently along with an intron fragment, and the intron was spliced <italic>in trans</italic>. In the CW lineage, a second <italic>trans</italic>-spliced group II intron (a tripartite intron comprising the RNA species encoded by the chloroplast <italic>tscA </italic>gene) took residence within <italic>psaA </italic>[<xref ref-type="bibr" rid="B38">38</xref>], group I introns inserted at several new sites within <italic>rrl </italic>[<xref ref-type="bibr" rid="B33">33</xref>,<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B39">39</xref>] and members of the group I family also invaded multiple sites of the <italic>rrs </italic>[<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B41">41</xref>], <italic>psbA </italic>[<xref ref-type="bibr" rid="B42">42</xref>] and <italic>psbC </italic>[<xref ref-type="bibr" rid="B37">37</xref>] genes.</p><p>The unusual structures displayed by the expanded <italic>clpP </italic>and <italic>rps3 </italic>genes and the fragmented <italic>rps2 </italic>and <italic>rpoB </italic>genes are inventions that arose in the Chlorophyceae. For both <italic>clpP </italic>and <italic>rps3</italic>, it has been shown that the insertion sequence is not removed at the RNA level [<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B44">44</xref>]. Characterization of the chloroplast ClpP/R protease complex of <italic>C. reinhardtii </italic>revealed that the approximately 30 kDa insertion sequence in <italic>clpP</italic>, designated as IS1, could be a new type of intein [<xref ref-type="bibr" rid="B45">45</xref>]. Two distinct proteins derived from the chloroplast <italic>clpP </italic>gene, a long version containing IS1 and a shorter version lacking this sequence element, were found to be stable components of this complex. IS1 has been hypothesized to prevent interaction with the HSP100 chaperone and to be localized in only one of the two heptamers forming the complex, thus prohibiting access of protein substrates to the proteolytic chamber of the ClpP/R complex via one of its axial pores. In contrast, a proteomic analysis of ribosomal proteins in the small subunit of the chloroplast ribosome from <italic>C. reinhardtii </italic>revealed that the insertion sequence in <italic>rps3 </italic>is an integral part of the mature product of this gene [<xref ref-type="bibr" rid="B46">46</xref>]. In this same analysis, Rps2 was also found to be an unusually large ribosomal protein; this protein of 570 amino acid residues encoded by <italic>rps2b </italic>(the largest of the two ORFs showing sequence similarity to the bacterial rps2 genes), contains an N-terminal extension and a C-terminal half with homology to characterized Rps2 proteins from other organisms. No peptides were found to be derived from <italic>rps2a</italic>, indicating that the latter sequence may be a pseudogene. The biological significance of the additional domains found in Rps3 and Rps2 remains uncertain. These domains, which are exposed to the solvent side and are located near each other and around the neck of the 30S subunit, may be related to unique features of translational regulation, or they may be orthologues of nonribosomal proteins [<xref ref-type="bibr" rid="B46">46</xref>]. Finally, it is not yet clear how the fragmented <italic>rpoB </italic>gene is expressed at the protein level. This gene is undoubtedly essential for cell survival in view of the fact that, unlike their homologues in land plants, the <italic>C. reinhardtii </italic>nuclear genome does not appear to encode a chloroplast-targeted RNA polymerase [<xref ref-type="bibr" rid="B47">47</xref>,<xref ref-type="bibr" rid="B48">48</xref>].</p><p>The considerable differences in gene density and abundance of SDRs observed in <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs raise questions about the status of the chloroplast genome of the common ancestor of DO and CW green algae with regards to these features. From the data derived from previously sequenced chlorophyte cpDNAs, Pombert <italic>et al</italic>. [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>] proposed that proliferation of repeated sequences in intergenic regions and selected genes occurred progressively during the evolution of UTC algae, thereby accounting for the observation that the <italic>Chlamydomonas </italic>genome is the most rich in SDR elements and the least tightly packed with genes. The results reported here are compatible with this idea and support the presence of SDRs in the common ancestor of <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs provided that specific loss of numerous SDRs occurred concurrently with streamlining of the genome in the <italic>Scenedesmus </italic>lineage. On the other hand, considering that no common SDRs have been identified in different UTC algal cpDNAs, the idea that these genetic elements were independently acquired in UTC lineages cannot be ruled out.</p><p>The single-copy regions of <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs are almost equal in size but differ radically in gene content, indicating that many genes were exchanged between opposite single-copy regions during the evolution of the DO and CW algae. This observation contrasts with the situation reported for the cpDNAs of <italic>C. reinhardtii </italic>and <italic>C. moewusii </italic>[<xref ref-type="bibr" rid="B26">26</xref>]. These representatives of deeply branched chlamydomonad lineages also display extensive gene rearrangements in their cpDNAs; however, these rearrangements are mainly confined to individual single-copy regions. Only two (<italic>atpA </italic>and <italic>psbI</italic>) of the 77 genes mapped on the <italic>C. reinhardtii </italic>and <italic>C. moewusii </italic>genomes [<xref ref-type="bibr" rid="B26">26</xref>-<xref ref-type="bibr" rid="B28">28</xref>] moved from one single-copy region to the other.</p><p>To compare the level of gene rearrangements displayed by <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs with those exhibited by other pairs of chlorophyte cpDNAs, we examined the orders of the 90 genes common to seven green algal cpDNAs (Table <xref ref-type="table" rid="T7">7</xref>). In these analyses, changes in gene order were assumed to occur only by inversions, an hypothesis that is strongly supported by previous mapping analyses of chloroplast genes from closely related chlamydomonads [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. We estimated that a minimum of 58 inversions would be required to convert the gene order of <italic>Scenedesmus </italic>cpDNA to that of <italic>Chlamydomonas </italic>cpDNA. Although extensive, this level of gene rearrangements is less important than those observed in the comparisons involving each chlorophycean genome and a genome from another green algal class (Table <xref ref-type="table" rid="T7">7</xref>). The similar levels of gene rearrangements observed in these interclass comparisons suggest that the gene organizations of both <italic>Chlamydomonas </italic>and <italic>Scenedesmus </italic>diverged considerably from that of their last common ancestor.</p><table-wrap position="float" id="T7"><label>Table 7</label><caption><p>Minimal numbers of inversions accounting for gene rearrangements between green algal cpDNAs</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Compared cpDNA</bold></td><td align="center" colspan="6"><bold>Number of inversions </bold><sup>a</sup></td></tr><tr><td></td><td colspan="6"><hr></hr></td></tr><tr><td></td><td align="center"><bold><italic>Nephroselmis</italic></bold></td><td align="center"><bold><italic>Chlorella</italic></bold></td><td align="center"><bold><italic>Oltmannsiellopsis</italic></bold></td><td align="center"><bold><italic>Pseudendoclonium</italic></bold></td><td align="center"><bold><italic>Scenedesmus</italic></bold></td><td align="center"><bold><italic>Chlamydomonas</italic></bold></td></tr></thead><tbody><tr><td align="left"><italic>Mesostigma</italic></td><td align="center">43</td><td align="center">46</td><td align="center">54</td><td align="center">54</td><td align="center">75</td><td align="center">75</td></tr><tr><td align="left"><italic>Nephroselmis</italic></td><td></td><td align="center">47</td><td align="center">55</td><td align="center">55</td><td align="center">73</td><td align="center">74</td></tr><tr><td align="left"><italic>Chlorella</italic></td><td></td><td></td><td align="center">50</td><td align="center">52</td><td align="center">74</td><td align="center">75</td></tr><tr><td align="left"><italic>Oltmannsiellopsis</italic></td><td></td><td></td><td></td><td align="center">55</td><td align="center">78</td><td align="center">75</td></tr><tr><td align="left"><italic>Pseudendoclonium</italic></td><td></td><td></td><td></td><td></td><td align="center">76</td><td align="center">77</td></tr><tr><td align="left"><italic>Scenedesmus</italic></td><td></td><td></td><td></td><td></td><td></td><td align="center">58</td></tr></tbody></table><table-wrap-foot><p><sup>a </sup>Numbers of gene permutations by inversions were computed using GRIMM.</p></table-wrap-foot></table-wrap><p>The high gene density and strongly biased distribution of genes between the two DNA strands in the <italic>Scenedesmus </italic>genome most probably reflect the influence of natural selection on genome organization. A bias in clustering of adjacent genes on the same DNA strand has also been reported for the <italic>Chlamydomonas </italic>chloroplast genome [<xref ref-type="bibr" rid="B49">49</xref>]; however, this bias is less conspicuous than that observed for <italic>Scenedesmus </italic>cpDNA. For <italic>Chlamydomonas </italic>cpDNA, a parametric bootstrap approach was used to test if gene order evolves under selection [<xref ref-type="bibr" rid="B49">49</xref>]. In this analysis, the putative gene order in the common ancestor of <italic>Chlamydomonas </italic>and <italic>Chlorella </italic>was inferred and subjected to random rearrangements. It was found that the multiple gene rearrangements in the <italic>Chlamydomonas </italic>lineage resulted in an increased sidedness, <italic>i.e</italic>. an increased propensity of adjacent genes to be located on the same strand. Sidedness indices of 0.6966 and 0.8710 were scored for the common ancestor and <italic>Chlamydomonas</italic>, respectively, and simulated genomes showed a significant decrease in sidedness relative to the ancestral genome. At 0.8842, the sidedness index we calculated for <italic>Scenedesmus </italic>cpDNA is slightly higher than that reported for its <italic>Chlamydomonas </italic>counterpart.</p><p>Coding strand biases have also been reported for the plastid genomes of the parasitic green alga <italic>Helicosporidium </italic>sp. [<xref ref-type="bibr" rid="B50">50</xref>], the euglenozoan alga <italic>Euglena gracilis </italic>[<xref ref-type="bibr" rid="B51">51</xref>], and apicomplexan parasites [<xref ref-type="bibr" rid="B52">52</xref>-<xref ref-type="bibr" rid="B55">55</xref>]. This feature is prominent in the highly reduced <italic>Helicosporidium </italic>genome where a symmetry in strand bias of coding regions has been observed, with nearly all genes on each half of the genome being encoded on one strand. The two strands of the <italic>Helicosporidium </italic>and <italic>Euglena </italic>genomes are also biased with regards to nucleotide composition and this compositional bias switches at the putative origin of DNA replication [<xref ref-type="bibr" rid="B50">50</xref>,<xref ref-type="bibr" rid="B51">51</xref>]. It has been proposed that the coding strand bias observed in these genomes is generated by selection to code highly expressed genes on the leading strand to limit collisions between RNA and DNA polymerases, thereby increasing the rates of both replication and transcription. Unlike their <italic>Helicosporidium </italic>and <italic>Euglena </italic>homologues, <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs show no strand bias in nucleotide composition (our unpublished results and [<xref ref-type="bibr" rid="B49">49</xref>]), thus providing no support for the notion that gene orders in chlorophycean genomes are selected to maximize the rate of replication. The high degree of sidedness observed for <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs could result mainly from selection of polycistronic transcription to coordinate gene expression [<xref ref-type="bibr" rid="B49">49</xref>].</p></sec><sec><title>Conclusion</title><p>Our study revealed that, although <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs display nearly identical gene repertoires and a high level of sidedness in the distribution of their genes on the two DNA strands, their gene orders are highly scrambled. In future studies, it will be interesting to investigate whether remodelling of the chloroplast genome is subjected to different constraints in the DO and CW lineages and whether the derived state of <italic>Scenedesmus </italic>and <italic>Chlamydomonas </italic>cpDNAs arose early during the evolution of chlorophycean green algae. To test this hypothesis, it will be necessary to examine other representatives of the DO and CW clades as well as members of more basal lineages of the Chlorophyceae.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Strain and culture conditions</title><p><italic>Scenedesmus obliquus </italic>(Turp.) Kürtz was obtained from the Culture Collection of Algae at the University of Texas at Austin (UTEX 393) and grown in modified Volvox medium [<xref ref-type="bibr" rid="B56">56</xref>] under 12 h light/dark cycles.</p></sec><sec><title>Isolation and sequencing of cpDNA</title><p>An A+T rich fraction containing cpDNA was isolated and sequenced as described in Pombert <italic>et al</italic>. [<xref ref-type="bibr" rid="B14">14</xref>]. Sequences were edited and assembled with AUTOASSEMBLER 2.1.1 (Applied Biosystems). The fully annotated chloroplast genome sequence has been deposited in [GenBank:<ext-link ext-link-type="gen" xlink:href="DQ396875">DQ396875</ext-link>].</p></sec><sec><title>Analyses of genome sequence</title><p>Gene content was determined by Blast homology searches [<xref ref-type="bibr" rid="B57">57</xref>] against the nonredundant database of the National Center for Biotechnology and Information (NCBI) server. Protein-coding genes and open reading frames (ORFs) were localized precisely using ORFFINDER at NCBI, various programs of the GCG version 10.2 package (Accelrys, Burlington, Mass.) and other applications from the EMBOSS version 2.6.0 package [<xref ref-type="bibr" rid="B58">58</xref>]. Genes coding for tRNAs were localized using tRNAscan-SE 1.23 [<xref ref-type="bibr" rid="B59">59</xref>]. Repeated sequences were identified using REPuter 2.74 [<xref ref-type="bibr" rid="B60">60</xref>] and classified using REPEATFINDER [<xref ref-type="bibr" rid="B61">61</xref>]. Numbers of SDR units were determined with FINDPATTERNS of the GCG Wisconsin Package version 10.2. The total length of genome sequences containing repeated elements was estimated with RepeatMasker [<xref ref-type="bibr" rid="B62">62</xref>] running under the WU-BLAST 2.0 search engine [<xref ref-type="bibr" rid="B63">63</xref>].</p><p>The GRIMM web server [<xref ref-type="bibr" rid="B64">64</xref>] was used to infer the minimal number of gene permutations by inversions in pairwise comparisons of chloroplast genomes. Because GRIMM cannot deal with duplicated genes and requires that the compared genomes have the same gene content, genes within one of the two copies of the IR were excluded and only the genes common to all the compared genomes were analysed. The data set used in the comparative analyses reported in Table <xref ref-type="table" rid="T7">7</xref> contained 90 genes; the three exons of the <italic>trans</italic>-spliced <italic>psaA </italic>gene were coded as distinct fragments (for a total of 92 gene loci).</p></sec></sec><sec><title>Abbreviations</title><p>cpDNA, chloroplast DNA; CW, clockwise; DO, directly opposed; IR, inverted repeat; LSC, large single copy; SDR, short dispersed repeat; SSC, small single copy; UTC, Ulvophyceae/Trebouxiophyceae/Chlorophyceae.</p></sec><sec><title>Authors' contributions</title><p>JCC participated in the conception of this study, carried out part of the genome sequencing, performed all sequence analyses, annotated the genome, generated the tables and figures, and drafted the manuscript. CO participated in the sequencing and contributed to the assembly of the genome sequence. CL and MT conceived and supervised the study, contributed to the interpretation of the data and prepared the manuscript. All authors read and approved the final manuscript.</p></sec>
|
A simplified explanation for the frameshift mutation that created a novel C-terminal motif in the <italic>APETALA3 </italic>gene lineage
|
<sec><title>Background</title><p>The evolution of type II MADS box genes has been extensively studied in angiosperms. One of the best-understood subfamilies is that of the <italic>Arabidopsis </italic>gene <italic>APETALA3 </italic>(<italic>AP3</italic>). Previous work has demonstrated that the ancestral paleo<italic>AP3 </italic>lineage was duplicated at some point within the basal eudicots to give rise to the paralogous <italic>TM6 </italic>and eu<italic>AP3 </italic>lineages. This event was followed in eu<italic>AP3 </italic>orthologs by the replacement of the C-terminal paleoAP3 motif with the derived euAP3 motif. It has been suggested that the new motif was created by an eight-nucleotide insertion that produced a translational frameshift.</p></sec><sec><title>Results</title><p>The addition of 25 eudicot <italic>AP3 </italic>homologs to the existing dataset has allowed us to clarify the process by which the euAP3 motif evolved. Phylogenetic analysis indicates that the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication maps very close to the base of the core eudicots, associated with the families Trochodendraceae and Buxaceae. We demonstrate that although the transformation of paleoAP3 into euAP3 was due to a frameshift mutation, this was the result of a single nucleotide deletion. The use of ancestral character state reconstructions has allowed us to demonstrate that the frameshift was accompanied by few other nucleotide changes. We further confirm that the sequence is evolving as coding region.</p></sec><sec><title>Conclusion</title><p>This study demonstrates that the simplest of genetic changes can result in the remodeling of protein sequence to produce a kind of molecular 'hopeful monster.' Moreover, such a novel protein motif can become conserved almost immediately on the basis of what appears to be a rapidly generated new function. Given that the existing data on the function of such C-terminal motifs are somewhat disparate and contradictory, we have sought to synthesize previous findings within the context of the current analysis and thereby highlight specific hypotheses that require further investigation before the significance of the euAP3 frameshift event can be fully understood.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Kramer</surname><given-names>Elena M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Su</surname><given-names>Huei-Jiun</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Wu</surname><given-names>Cheng-Chiang</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Hu</surname><given-names>Jer-Ming</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Evolutionary Biology
|
<sec><title>Background</title><p>An increasing body of research has demonstrated that changes in gene regulation play a major role in the evolution of morphological form (reviewed [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]). That is not to say, however, that the evolution of coding sequence does not also contribute. Multiple examples from both plants and animals demonstrate that even minor changes in coding sequence can impact both biochemical and developmental functions (e.g., [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B7">7</xref>]). Interestingly, a common theme among many of these examples is gene duplication, which serves to release resultant paralogs from the selective pressures experienced by the single ancestral locus. In order to begin to understand the process by which non-synonymous mutation leads to changes in gene function, we need to be able to isolate such changes and characterize the pattern of sequence evolution in detail. This is facilitated by a thorough understanding of taxonomic and gene lineage evolution as well as a relatively recent evolutionary timescale. All of these criteria are met by the <italic>APETALA3 </italic>(<italic>AP3</italic>) lineage of type II MADS box genes.</p><p>Members of the type II MADS box family control many important aspects of plant development (reviewed [<xref ref-type="bibr" rid="B8">8</xref>]). Extensive phylogenetic analyses have identified multiple subfamilies, which are particularly well understood in the seed plants (reviewed [<xref ref-type="bibr" rid="B9">9</xref>]). This interest was largely triggered by the central role that type II MADS box genes play in the genetic program controlling floral organ identity. The so-called ABC model [<xref ref-type="bibr" rid="B10">10</xref>] describes how floral organ identity is determined by an overlapping set of three gene activities that produce distinct combinatorial codes: A class genes code for first whorl sepals; A+B, for second whorl petals; B+C, for third whorl stamens; and C alone, for fourth whorl carpels. Subsequent studies have identified additional critical gene classes, including the "E" class that acts in all floral whorls to facilitate the function of A, B and C class genes [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. All but one of the ABCE class loci are type II MADS box genes [<xref ref-type="bibr" rid="B13">13</xref>], which are also known as MIKC MADS box genes due to the canonical structure displayed by the members. Starting at the N-terminal end of the gene, the 'M' or MADS domain is highly conserved across eukaryotes, and mediates DNA binding and protein dimerization [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. The next two regions, referred to as I and K, are primarily involved with protein dimerization [<xref ref-type="bibr" rid="B14">14</xref>], while the last, the C domain, has been associated with a number of different functions. These include mediating higher-order interactions among MADS protein dimers [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>], transcriptional activation [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>], and post-translational modification [<xref ref-type="bibr" rid="B20">20</xref>]. A notable feature of the C-terminal domain is that although it shows a lower degree of overall sequence conservation than the other regions, each of the major MIKC subfamilies possesses short, highly conserved diagnostic motifs at their C-terminal end (reviewed [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>]). In the majority of cases, the specific function of these motifs remains unknown.</p><p>As our understanding of the evolution of MIKC MADS box genes has grown, it has become increasingly clear that their evolutionary history is one of frequent gene duplication across all phylogenetic levels (reviewed [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B23">23</xref>]). One subfamily that demonstrates this phenomenon especially well is defined by the <italic>APETALA3 </italic>(<italic>AP3</italic>) and <italic>PISTILLATA </italic>(<italic>PI</italic>) gene lineages, which include the <italic>Arabidopsis </italic>petal and stamen identity genes of the same names. These two lineages are sister groups within the larger MIKC MADS gene family [<xref ref-type="bibr" rid="B24">24</xref>] and are the product of a gene duplication event that predated the diversification of the angiosperms [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B27">27</xref>]. Early studies recognized that there were, in fact, two paralogous lineages of <italic>AP3</italic>-like genes in the core eudicots: one termed eu<italic>AP3 </italic>that contains <italic>AP3 </italic>itself and the other named <italic>TM6</italic>, which lacks a representative in <italic>Arabidopsis </italic>but has been identified in many other core eudicot taxa [<xref ref-type="bibr" rid="B28">28</xref>,<xref ref-type="bibr" rid="B29">29</xref>]. Although clearly related, the eu<italic>AP3 </italic>and <italic>TM6 </italic>lineages have a number of distinct features, the most striking of which is their C-terminal motifs. In the <italic>TM6 </italic>and ancestral paleo<italic>AP3 </italic>lineages, the C-terminal motif has the consensus YGxHDLRLA (x indicating a variable site) [<xref ref-type="bibr" rid="B28">28</xref>]. This sequence, the paleoAP3 motif, is conserved throughout angiosperms and is recognizable in gymnosperm <italic>AP3</italic>/<italic>PI </italic>ancestors as well as the even more distantly related B<sub>sister </sub>lineage [<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. In the eu<italic>AP3 </italic>lineage, however, the paleoAP3 motif is completely absent and in its place is the so-called euAP3 motif with the consensus SDLTTFALLE [<xref ref-type="bibr" rid="B28">28</xref>]. The differences in this region and other sites reveal eu<italic>AP3 </italic>to be a divergent paralogous lineage relative to both its ancestral and sister lineages.</p><p>The patterns of sequence evolution associated with the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication raise questions regarding the functional significance of the C-terminal motifs in general and the eu<italic>AP3 </italic>divergence in particular. From the biochemical standpoint, we can say with certainty that the euAP3 motif is important for proper <italic>AP3 </italic>function <italic>in vivo</italic>, and that the paleoAP3 and euAP3 motifs are not functionally equivalent [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. In terms of the genes' developmental roles, the suggestion has been made that following the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication, the eu<italic>AP3 </italic>lineage acquired a new role in petal development [<xref ref-type="bibr" rid="B6">6</xref>]. The evidence to support this conclusion is diverse, and includes: 1) the fact that the expression patterns of paleo<italic>AP3 </italic>orthologs in the petals of non-core eudicots are much more variable than those observed for eu<italic>AP3 </italic>representatives within the core eudicots [<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B33">33</xref>]; 2) that a chimeric <italic>AP3 </italic>bearing a paleoAP3 motif is especially poor at promoting petal identity in <italic>Arabidopsis </italic>[<xref ref-type="bibr" rid="B6">6</xref>]; and 3) that the sole <italic>TM6 </italic>ortholog to be functionally characterized, <italic>PhTM6 </italic>from <italic>Petunia</italic>, only contributes to stamen identity ([<xref ref-type="bibr" rid="B34">34</xref>], Vandenbussche and Gerats, pers. comm). On the other hand, paleo<italic>AP3 </italic>orthologs are almost always expressed in petaloid organs (e.g., [<xref ref-type="bibr" rid="B35">35</xref>-<xref ref-type="bibr" rid="B37">37</xref>]) and appear to function in the identity of petal-derived organs in the grasses [<xref ref-type="bibr" rid="B38">38</xref>,<xref ref-type="bibr" rid="B39">39</xref>]. One explanation that could encompass all of the current evidence is to posit that although paleo<italic>AP3 </italic>members play variable roles in petal identity, this function was canalized at the base of the core eudicots in conjunction with changes in biochemical aspects of eu<italic>AP3 </italic>function and subsequent subfunctionalization in the <italic>TM6 </italic>lineage [<xref ref-type="bibr" rid="B40">40</xref>,<xref ref-type="bibr" rid="B41">41</xref>].</p><p>In regards to the evolution of the euAP3 motif itself, it was recently recognized that a frameshift event in the coding sequence of the paleoAP3 motif could generate components of the euAP3 motif [<xref ref-type="bibr" rid="B22">22</xref>]. The model of Vandenbussche et al. proposes that an eight nucleotide insertion contributed to the evolution of the euAP3 motif both by the addition of novel sequence and by causing a frameshift mutation. In the current study, we have sought to better establish the timing of the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication event and the nature of the evolution of the euAP3 motif. The addition of 25 new <italic>AP3 </italic>homologs has particularly provided insight into the latter issue by demonstrating that the derivation of the euAP3 motif was even simpler than previously suggested. We conclude that a single nucleotide deletion transformed the ancestral paleoAP3 motif into the euAP3 motif with relatively few associated nucleotide changes. Furthermore, we provide evidence that the region is being conserved at the amino acid level, suggesting that the almost immediate conservation of the euAP3 motif was due to new function of the novel protein sequence.</p></sec><sec><title>Results and discussion</title><sec><title>Characterization and phylogenetic analysis of <italic>AP3 </italic>homologs</title><p>In an effort to better understand the evolution of the <italic>AP3 </italic>lineage in the eudicots, we used RT-PCR to isolate <italic>AP3 </italic>homologs from five taxa representing every lineage of the basal eudicots as well as eight taxa drawn from core eudicot lineages that had been poorly sampled (Fig. <xref ref-type="fig" rid="F1">1</xref>). This process yielded 25 <italic>AP3 </italic>homologs, 5 of which have been published in the context of previous studies [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B37">37</xref>] (see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref> for GenBank accession numbers). All of the basal eudicot loci exhibit well-conserved C-terminal paleoAP3 motifs (<xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). <italic>Pachysandra</italic>, <italic>Meliosma </italic>and <italic>Platanus </italic>were found to express multiple paralogs with high degrees of sequence similarity, most likely indicating recent gene duplication events. As expected, two types of loci were identified in the core eudicots, some with paleoAP3 motifs and others with euAP3 motifs (<xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). Both types were obtained from <italic>Saxifraga</italic>, <italic>Corylopsis </italic>and <italic>Ilex</italic>, but in the other five taxa we were only able to detect one of the two classes. Only paleoAP3-containing loci were found in <italic>Phytolacca</italic>, <italic>Paeonia</italic>, <italic>Vitis </italic>and <italic>Loranthus</italic>, while only euAP3-containing genes were identified in <italic>Kalanchoe</italic>. Multiple closely related paralogs were identified in <italic>Kalanchoe</italic>, <italic>Phytolacca </italic>and <italic>Corylopsis</italic>. The detection of only one <italic>AP3 </italic>class may have several different causes including actual paralog loss; low levels of paralog expression, which could hamper RT-PCR-based identification; and sequence divergence that prevented the success of current primer combinations.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Simplified eudicot phylogeny with newly sampled taxa. </bold>Simplified eudicot phylogeny based on [42] and [43] with newly sampled taxa noted. The inferred timing of the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication based on phylogenetic analyses of the current dataset (Fig. 2) is indicated by the blue box.</p></caption><graphic xlink:href="1471-2148-6-30-1"/></fig><p>We performed phylogenetic analysis using maximum likelihood (ML) on a nucleotide dataset (<xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>) containing all of the new loci in addition to previously identified basal and core eudicot sequences, with magnoliid dicot, monocot and ANITA grade <italic>AP3 </italic>homologs serving as outgroups to the eudicot sequences (Fig. <xref ref-type="fig" rid="F2">2</xref>). The recovered phylogeny is consistent with previous analyses [<xref ref-type="bibr" rid="B26">26</xref>,<xref ref-type="bibr" rid="B28">28</xref>] in showing two major core eudicot lineages (eu<italic>AP3 </italic>and <italic>TM6</italic>) that were derived from an ancestral lineage (paleo<italic>AP3</italic>), which is represented in the basal eudicot and outgroup taxa. There is strong ML bootstrap support for the core eudicot eu<italic>AP3 </italic>and <italic>TM6 </italic>clades but little support for the other backbone nodes. Marginal support is seen for the clade containing <italic>Trochodendron AP3</italic>, the <italic>Pachysandra AP3 </italic>homologs and the other core eudicot sequences. The ML tree places <italic>Trochodendron </italic>and <italic>Pachysandra </italic>close to the gene duplication event that produced <italic>euAP3 </italic>and <italic>TM6</italic>. Based on a strict interpretation of the current phylogeny, this duplication would be inferred to have occurred after the divergence of Trochodendraceae but before the split of Buxaceae (star in Fig. <xref ref-type="fig" rid="F2">2</xref>). However, the lack of support for the backbone nodes allows alternative hypotheses. Most notably, the multiple loci from <italic>Aquilegia </italic>are not monophyletic (Fig. <xref ref-type="fig" rid="F2">2</xref>), suggesting additional duplications that may not be independent from eu<italic>AP3 </italic>and <italic>TM6</italic>. It has been demonstrated that there are at least three paralogous <italic>AP3 </italic>lineages in the Ranunculales [<xref ref-type="bibr" rid="B37">37</xref>] but this study did not test whether these events are related to that which gave rise to eu<italic>AP3 </italic>and <italic>TM6</italic>. Analysis of a dataset focused on complete sampling of the Ranunculales (including an additional 45 sequences) recovers all of the Ranunculid representatives as a single clade with moderate support (data not shown, Kramer, in prep). This indicates that the Ranunculid gene duplication events are, in fact, independent from that of eu<italic>AP3</italic>/<italic>TM6</italic>. While this increased sampling improves the resolution of the Ranunculid representatives, it is otherwise identical to the analysis shown in Fig. <xref ref-type="fig" rid="F2">2</xref>, both in terms of the positions of the Trochodendraceae and Buxaceae homologs, and in the lack of support for their positions.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Maximum likelihood phylogeny derived from analysis of the <italic>AP3 </italic>nucleotide dataset. </bold>Bootstrap percentages (above 50) are placed at the nodes. The name of each taxon is in parentheses following the locus name. The node corresponding to the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication is indicated with a star while the branch associated with the subsequent eu<italic>AP3</italic>-specific frameshift event is indicated with an arrow. Colored vertical bars on the right are used to indicate the paralog lineage membership of the adjacent loci: the purple bars represent the eu<italic>AP3 </italic>lineage; the green bar, the <italic>TM6 </italic>lineage; and the blue bars, the ancestral paleo<italic>AP3 </italic>lineage. The phylogenetic positions of the associated taxa are denoted as Core Eudicot, Basal Eudicot, Magnoliid, Monocot or ANITA (See <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>, [94] and [95]). Colored branches are used to indicate the "frameshift potential" of each locus: black branches mean that a single nucleotide frameshift in the paleoAP3 or euAP3 motif would recover 0–3 amino acids of the other motif; orange branches, 4–6 amino acids; and red branches, seven or more amino acids. For instance, as shown in Fig. 3A, the first reading frame of <italic>PloAP3-1 </italic>encodes a perfect paleoAP3 motif but the second reading frame would produce a motif with seven out of the ten euAP3 motif residues, and, therefore, the <italic>PloAP3-1 </italic>branch is red. In contrast, the second reading frame of the <italic>AreAP3 </italic>paleoAP3 motif would have only two amino acids similar to the euAP3 motif, which is indicated by a black branch for <italic>AreAP3</italic>. Additionally, some paleo<italic>AP3 </italic>cDNAs would encode a positionally correct stop codon for a euAP3 motif in their second frame. These loci are denoted with red asterisks.</p></caption><graphic xlink:href="1471-2148-6-30-2"/></fig><p>The major departure of the current phylogeny from previous studies is the position of the <italic>Pachysandra AP3 </italic>homologs, representing sampling from two species, which are placed as sister to the eu<italic>AP3 </italic>lineage <italic>s.s</italic>. after the duplication event. This position is somewhat surprising given that none of the <italic>Pachysandra </italic>loci contain euAP3 motifs, which have previously been considered diagnostic for the eu<italic>AP3 </italic>lineage. However, in the I and K regions of the protein sequence (<xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>), the <italic>Pachysandra AP3 </italic>homologs share other character states that have been identified as eu<italic>AP3 </italic>lineage synapomorphies [<xref ref-type="bibr" rid="B28">28</xref>]. It should be noted that in maximum parsimony (MP) analyses, the <italic>Pachysandra </italic>loci sometimes are placed as an earlier branch, just before the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication event (data not shown), underscoring the poorly supported position of these loci.</p><p>This analysis does allow us to make some conclusions regarding the timing of the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication event. The duplication clearly occurred before the last common ancestor of all core eudicots, including the family Gunneraceae, which has been identified as sister to the traditionally defined core eudicot clade [<xref ref-type="bibr" rid="B42">42</xref>]. It seems likely that the duplication occurred after the early lineages of the basal eudicots, including the Ranunculales, Proteales and Sabiaceae. Based on the current analysis, we cannot determine with certainty how the timing of the duplication event related to the origin of the Trochodendraceae and Buxaceae lineages. Similarly, recent phylogenetic studies of the eudicots place these two families as sister to the core eudicots including Gunneraceae without strong support for their exact branching order (Fig. <xref ref-type="fig" rid="F1">1</xref>) [<xref ref-type="bibr" rid="B42">42</xref>,<xref ref-type="bibr" rid="B43">43</xref>]. Most likely, these difficulties reflect the very rapid diversification that occurred during this period of angiosperm evolution, which dates to ~95–115 mya [<xref ref-type="bibr" rid="B44">44</xref>].</p></sec><sec><title>Evidence for a single nucleotide frameshift event at the base of the eu<italic>AP3 </italic>clade</title><p>What is interesting about the current dataset is that all of the paleo<italic>AP3 </italic>lineage members and the <italic>Pachysandra AP3 </italic>homologs possess fairly normal paleoAP3 motifs with no clear sign of intermediates with the highly diverged euAP3 motif (<xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). The explanation for this lack of 'missing links' has recently become apparent. In the course of characterizing the <italic>AP3 </italic>representatives from <italic>Platanus </italic>[<xref ref-type="bibr" rid="B37">37</xref>], we noticed that while the first reading frame encoded a perfect paleoAP3 motif, the second frame in the same region had the potential to encode an amino acid sequence with strong similarity to the euAP3 motif (Fig. <xref ref-type="fig" rid="F3">3A</xref>). The 3' UTR of <italic>PloAP3-1 </italic>even contains a stop codon in the correct frame and position. It has similarly been suggested by other researchers that a frameshift event transformed the paleoAP3 motif into the euAP3 motif, but this model posited an eight nucleotide insertion [<xref ref-type="bibr" rid="B22">22</xref>]. Examination of our basal eudicot sequences suggests a much simpler model whereby a single nucleotide deletion gave rise to the novel motif without the necessity for the insertion of new nucleotides. In fact, it is possible to construct a theoretical nucleotide sequence that encodes a chemically conserved paleoAP3 motif in the first reading frame and a perfect euAP3 motif in the second (Fig. <xref ref-type="fig" rid="F3">3F</xref>). We will subsequently refer to this phenomenon, the capacity of a given nucleotide sequence to simultaneously encode a paleoAP3 motif in the first reading frame and a recognizable euAP3 motif in the second, as 'frameshift potential.' Naturally occurring frameshift potential is particularly noticeable in other basal eudicot loci (Fig. <xref ref-type="fig" rid="F2">2</xref>, <xref ref-type="fig" rid="F3">3B</xref>). <italic>AP3 </italic>homologs from the magnoliid dicots, monocots or ANITA grade show little frameshift potential by comparison (Fig. <xref ref-type="fig" rid="F2">2</xref>, <xref ref-type="fig" rid="F3">3C</xref>). Similarly, core eudicot eu<italic>AP3 </italic>and <italic>TM6 </italic>lineage members exhibit relatively little frameshift potential (in the case of eu<italic>AP3</italic>, this would be a kind of 'reverse' frameshift potential to regenerate paleoAP3 sequence from euAP3; Fig. <xref ref-type="fig" rid="F2">2</xref>, 3D-E).</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Frameshift potentials of <italic>Platanus PloAP3-1 </italic>(A), <italic>Pachysandra PtAP3-1 </italic>(B), <italic>Aristolochia AreAP3 </italic>(C), <italic>S. lycopersicon TM6 </italic>(D), <italic>Antirrhinum DefA </italic>(E) and a theoretical paleoAP3-encoding sequence (F). </bold>A-D, Nucleotide sequences of the paleoAP3-encoding regions of <italic>Platanus PloAP3-1 </italic>(A), <italic>Pachysandra PtAP3-2 </italic>(B), <italic>Aristolochia AreAP3 </italic>(C) and <italic>S. lycopersicon TM6 </italic>(D) with first and second predicted translation frames. E, Nucleotide sequence of the euAP3-encoding region of <italic>Antirrhinum DefA </italic>with first and third predicted translation frames. F, Nucleotide sequence of a theoretical DNA sequence that encodes a chemically conserved paleoAP3 motif in the first translational reading frame and a perfect euAP3 motif in the second translational reading frame. Chemical similarity with the paleoAP3 motif consensus YGxHDLRLA is indicated by purple letters while chemical similarity with the euAP3 motif consensus SDLTTFALLE is indicated by blue letters.</p></caption><graphic xlink:href="1471-2148-6-30-3"/></fig><p>The phylogenetically-structured nature of euAP3/paleoAP3 frameshift potential suggests that it is dependent on patterns of codon usage and, therefore, that this region is behaving as normal coding region. This conclusion is significant since one possible explanation for the observed phenomenon is that the region is conserved at the nucleotide level rather than at the amino acid level, such as would be the case for something like a microRNA binding site, for example. The prediction of this scenario, however, is that the sequence should not evolve in a pattern typical of coding region, where the first and second codon positions exhibit lower nucleotide diversity than the third positions. An alternative model is that the region is subject to programmed translational frameshift, a phenomenon previously observed in fungal, prokaryotic, plastid and viral genomes (reviewed [<xref ref-type="bibr" rid="B45">45</xref>]). This process is associated with perturbations in the expected pattern of sequence evolution such that substitutions are concentrated in the third positions of the <italic>original </italic>reading frame rather than in the third positions of the new frame. In addition, the encoded amino acid sequence of the original frame is conserved (e.g., [<xref ref-type="bibr" rid="B46">46</xref>,<xref ref-type="bibr" rid="B47">47</xref>]). Thus, under the first hypothesis, the paleoAP3 sequence would be conserved at the nucleotide level and would not bear the hallmarks of coding sequence evolution, while under the second hypothesis, the sequence should evolve like coding sequence but in the original reading frame.</p><p>Our general observations, as well as those of others [<xref ref-type="bibr" rid="B22">22</xref>], are not consistent with these models but we wanted to test this further by directly analyzing patterns of nucleotide diversity in the region. Figure <xref ref-type="fig" rid="F4">4</xref> shows a comparison of position-by-position nucleotide diversity values for the region spanning the inferred frameshift event (see also <xref ref-type="supplementary-material" rid="S6">Additional file 6</xref>). In the codons before the frameshift, first and second positions generally show lower nucleotide diversity than third positions. This pattern is maintained in both the paleoAP3-encoding and frameshifted euAP3-encoding sequences. Comparison of the appropriate paleoAP3 and euAP3 positions reveals that when the first position nucleotides of the paleoAP3 motif become third positions in euAP3-encoding sequences, the nucleotide diversity generally increases. Similarly, third positions in the paleoAP3 motif show high diversity but these values tend to decrease when the nucleotide becomes shifted to the second position in the euAP3 motif. Overall, position-by-position nucleotide diversity differs between the paleoAP3 and euAP3 regions, which suggests that the patterns of conservation do change following the frameshift event. Taken together, these findings confirm that both regions show all of the evolutionary hallmarks of sequence that is being conserved at the amino acid level in the first reading frame, allowing us to reject both the nucleotide-level conservation and programmed translational frameshift hypotheses. We conclude, therefore, that the ancestral paleoAP3 motif, which was conserved over more than 200 million years [<xref ref-type="bibr" rid="B31">31</xref>], was completely replaced by a new amino acid motif via a single nucleotide deletion following gene duplication. The euAP3 motif appears to have been conserved due to its protein function rather than any underlying nucleotide-level function. This clarification of the model for euAP3 evolution has been facilitated by the greatly improved sampling of basal eudicot lineages, which, in turn, allowed the refinement of the <italic>AP3 </italic>alignment to include fewer indels than that used by Vandenbussche et al. [<xref ref-type="bibr" rid="B22">22</xref>].</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Comparison of position-by-position nucleotide diversity values for paleoAP3 and euAP3 motif encoding loci (see also Add. Files 4-6). </bold>The yellow bars indicate the values for a dataset including all <italic>TM6 </italic>lineage members and basal eudicot paleo<italic>AP3 </italic>loci. The codon positions of each nucleotide and the corresponding amino acids are shown immediately below the chart. Third positions are highlighted in yellow. The blue bars indicate the values for a dataset including all eu<italic>AP3 </italic>lineage members. The codon positions of each nucleotide and the corresponding amino acid are shown at the bottom, with third positions highlighted in blue. The position of the euAP3 frameshift is represented by a dash mark. Note that some of the euAP3 positions have zero nucleotide diversity. n/a = not applicable.</p></caption><graphic xlink:href="1471-2148-6-30-4"/></fig><p>As shown in Fig. <xref ref-type="fig" rid="F3">3F</xref>, it is possible that the single nucleotide deletion was accompanied by few additional nucleotide changes. In an effort to investigate the potential range of nucleotide changes, we used MP and ML methods to reconstruct the ancestral nucleotide character states for critical nodes in the current <italic>AP3 </italic>phylogeny (Fig. <xref ref-type="fig" rid="F5">5</xref>). We also conducted the same analyses on alternative topologies to control for the fact that there is little or no support for the backbone of our phylogeny, (see Fig. <xref ref-type="fig" rid="F5">5</xref> and Methods). Due to the high level of conservation in this region, the ancestral character state reconstructions were very similar for the MP and ML approaches, regardless of the models of substitution or the details of the topology. Based on these results, it appears that 4–6 nucleotide changes occurred coincidently with the frameshift event, which in the current phylogeny would be inferred to have occurred along the branch at the base of the eu<italic>AP3 </italic>clade after the separation of the Buxaceae (represented by <italic>Pachysandra</italic>; Fig. <xref ref-type="fig" rid="F2">2</xref>). We cannot predict the order of the nucleotide changes relative to the frameshift event, however; and due to the nature of the frameshift, some changes that are synonymous before the deletion event are non-synonymous after (and vice versa). Figs. <xref ref-type="fig" rid="F5">5C</xref> and <xref ref-type="fig" rid="F5">5D</xref> reconstruct two alternative scenarios using the ancestral character states shown in Fig. <xref ref-type="fig" rid="F5">5B</xref> (which infers six nucleotide changes). In Figs. <xref ref-type="fig" rid="F5">5C</xref> and <xref ref-type="fig" rid="F5">5D</xref>, each line represents a stepwise set of changes that could have occurred during the transition from the states reconstructed for node B1 to those recovered for node B2. The first scenario is a 'minimal' model in which only one of the six changes is nonsynonymous and this one change is chemically conservative (Fig. <xref ref-type="fig" rid="F5">5C</xref>). The second is a 'maximal' model where all six changes are nonsynonymous. In this case, three out of the nine paleoAP3 amino acids are changed before the frameshift and three of the ten euAP3 amino acids are changed afterward (a total of four of these changes are chemically non-conservative). Even under the 'maximal' model, the frameshift event was clearly more significant in terms of sequence remodeling, resulting in the replacement of all but one of the paleoAP3 amino acids. Overall, these findings demonstrate that it is possible for the euAP3 motif to have been generated by single nucleotide deletion without significant additional nonsynonymous changes.</p><fig position="float" id="F5"><label>Figure 5</label><caption><p><bold>Nucleotide ancestral character state reconstructions and evolutionary scenarios. </bold>A, The MP ancestral character state reconstructions for the pre- and post-frameshift nodes (A1 and A2, respectively, as indicated to the right) of the recovered phylogeny (Fig. <xref ref-type="fig" rid="F2">2</xref>). These nucleotide sequences were recovered with the accelerated transitions (ACCTRAN) setting. Under the delayed transitions (DELTRAN) setting, the inferred sequences were identical to those shown in B. In the schematic phylogeny to the right, the star indicates the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication node and the red branch denotes the timing of the frameshift event. B, The MP ancestral character state reconstructions for an alternative topology where the <italic>Pachysandra </italic>loci predate the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication (again indicated by a star on the schematic phylogeny to the right). In this case, the frameshift occurred along the red branch immediately following the duplication event. Node B1 represents the duplication while B2 represents the ancestor of the eu<italic>AP3 </italic>clade. The sequences recovered with the ACCTRAN and DELTRAN settings were identical. C. Evolutionary scenario that minimizes the number of non-synonymous changes associated with the frameshift event. Each line represents stepwise changes that would have occurred during the transition from the sequence indicated as 'node B1' to that denoted 'node B2.' Under this model, four nucleotide changes occurred before the frameshift (line 'pre-FS') and two after (line 'node B2'), but only one of these changes was non-synonymous (indicated by red asterisk). D. Evolutionary scenario that maximizes the number of non-synonymous changes associated with the frameshift event. Again, each line represents stepwise changes that would have occurred during the transition from the sequence indicated as 'node B1' to that denoted 'node B2.' Under this model, three nucleotide changes occurred before the frameshift (line 'pre-FS') and three after (line 'node B2'), but all of these changes were non-synonymous (indicated by red asterisks). Scenarios in C and D are both based on the reconstructions shown in B. All nucleotide changes are indicated with red letters. FS = frameshift.</p></caption><graphic xlink:href="1471-2148-6-30-5"/></fig></sec><sec><title>Evidence for independent frameshift events in the <italic>AP3 </italic>lineage</title><p>The eu<italic>AP3 </italic>frameshift event seems so extraordinary that it naturally begs the question of how often this sort of thing happens. Similar events have been described in other MADS box genes lineages [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B48">48</xref>] as well as vertebrate gene families [<xref ref-type="bibr" rid="B49">49</xref>]. We examined the larger <italic>AP3 </italic>dataset for additional examples and found three (Fig. <xref ref-type="fig" rid="F6">6</xref>). The first we will consider is a single nucleotide insertion very close to the 3' end of the coding region in eu<italic>AP3 </italic>orthologs of the Solanaceae (Fig. <xref ref-type="fig" rid="F6">6A</xref>). Other eu<italic>AP3 </italic>loci from the Asterids, including the basal Solanaceous genus <italic>Petunia </italic>[<xref ref-type="bibr" rid="B50">50</xref>], show the complete euAP3 motif with a terminal glutamic acid. In comparison to these sequences, the eu<italic>AP3 </italic>homologs of more derived members of the Solanaceae have a single A insertion in the eighth codon of the euAP3 motif, which results in a single amino acid truncation of the motif. Such a minor change seems unlikely to have major biochemical significance, potentially explaining why the frameshifted form could be maintained. In contrast to this example, the other two instances are from taxa that have multiple recent <italic>AP3 </italic>paralogs. In <italic>Paeonia</italic>, there are two <italic>TM6 </italic>lineage members that share 91% identity at the nucleotide level. Their C-terminal regions are completely divergent, however, with <italic>PesTM6-1 </italic>having a recognizable paleoAP3 motif while <italic>PesTM6-2 </italic>has only the first tyrosine of the consensus (Fig. <xref ref-type="fig" rid="F6">6B</xref>). Examination of the nucleotide alignment reveals two indels in the 3' end of the coding region, the more significant of which is a 7-nucleotide deletion in <italic>PesTM6-2 </italic>that falls within the first codon of the paleoAP3 motif. This results in the complete replacement of the paleoAP3 sequence with a novel coding region derived from the 3' UTR and a second indel region. Similar to this case, a frameshift is observed in one of the four paleo<italic>AP3 </italic>paralogs of the magnoliid dicot <italic>Drimys</italic>, which is a recently polyploid genus [<xref ref-type="bibr" rid="B51">51</xref>]. The nucleotide identity among these paralogs ranges from 84–93% and three of the four paralogs have canonical paleoAP3 motifs. The fourth, <italic>DrwAP3-1</italic>, diverges in sequence in the second half of the motif, corresponding with an eight nucleotide deletion of this region. It has been argued that compensating mechanisms such as the presence of closely related paralogs or splicing variants can enable frameshift mutations to persist and eventually lead to functional divergence [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B49">49</xref>]. This model is consistent with the current observations for <italic>Paeonia </italic>and <italic>Drimys</italic>, as well as for the ancient eu<italic>AP3</italic>/<italic>TM6 </italic>duplication. The frameshifts detected in <italic>Solanum</italic>, <italic>Paeonia </italic>and <italic>Drimys </italic>may also indicate that this type of event occurs with relative frequency. Although sequence remodeling events such as those in <italic>PesTM6-2 </italic>and <italic>DrwAP3-1 </italic>may very well be lost over a short evolutionary timescale, it only takes one successful event to found a divergent paralogous lineage such as eu<italic>AP3</italic>.</p><fig position="float" id="F6"><label>Figure 6</label><caption><p><bold>Additional identified frameshift events in the <italic>APETALA3 </italic>lineage. </bold>A, Amino acid (left) and corresponding nucleotide (right) alignments of the C-terminal regions of select Asterid eu<italic>AP3 </italic>cDNAs. Loci from <italic>Antirrhinum </italic>(<italic>DefA</italic>), <italic>Syringa </italic>(<italic>SvAP3</italic>) and <italic>Petunia </italic>(<italic>pMADS1</italic>) show the typical euAP3 motif but a single nucleotide insertion in the Solanaceous taxa <italic>Nicotiana tobaccum </italic>(<italic>NTDEF</italic>), <italic>Solanum lycopersicon </italic>(<italic>LeAP3</italic>) and <italic>Solanum tuberosum </italic>(<italic>StDef</italic>) has produced a one amino acid truncation. B, Amino acid (left) and corresponding nucleotide (right) alignments of the C-terminal regions of the <italic>Paeonia suffructosa TM6 </italic>lineage members <italic>PesTM6-1 </italic>and <italic>PesTM6-2</italic>. A seven nucleotide deletion in <italic>PesTM6-2 </italic>has given rise to a novel C-terminal motif that replaces the paleoAP3 motif (which is moderately conserved in <italic>PesTM6-1</italic>). There is an additional indel between the two loci in the region of the <italic>PesTM6-2 </italic>stop codon. C, Amino acid (left) and corresponding nucleotide (right) alignments of the C-terminal regions of the <italic>Drimys winterii </italic>paleo<italic>AP3 </italic>lineage members <italic>DrwAP3-1</italic>, <italic>-2</italic>, -<italic>3 </italic>and -<italic>4</italic>. An eight nucleotide deletion in <italic>DrwAP3-1 </italic>results in remodeling of the last four amino acids in the paleoAP3 motif. For A-C: Asterisks indicate translational stops. The stop codons used in the separate reading frames are boxed. Numbers at right indicate the position in the amino acid or nucleotide sequence of each locus. See also Additional Files <xref ref-type="supplementary-material" rid="S2">2</xref> and <xref ref-type="supplementary-material" rid="S3">3</xref>.</p></caption><graphic xlink:href="1471-2148-6-30-6"/></fig></sec><sec><title>Molecular 'hopeful monsters'</title><p>The term 'hopeful monster' was coined by Goldschmidt [<xref ref-type="bibr" rid="B52">52</xref>] to describe new species that arise abruptly by macromutation. Very rarely, he argued, such profound mutations could be beneficial and allow the organism to rapidly adapt to a new mode of life. On the molecular level, the impact of a frameshift mutation on protein sequence is similarly drastic – replacing most, if not all, of the ancestral amino acids with new residues. It seems very likely that the vast majority of such mutations will not be retained, but the eu<italic>AP3</italic>/<italic>TM6 </italic>example, as well as others [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B49">49</xref>], demonstrates that there are isolated cases in which frameshifts have become conserved. Although this phenomenon would seem to be so unlikely as to be vanishingly rare, the role of gene duplication in this process means that it is essentially a matter of numbers, particularly in plants. It has been suggested that plants are especially subject to frequent gene duplications [<xref ref-type="bibr" rid="B53">53</xref>], due to everything from genome-scale events to single locus tandem duplications. In particular, loci involved in transcriptional regulation and signal transduction appear to be preferentially retained [<xref ref-type="bibr" rid="B54">54</xref>,<xref ref-type="bibr" rid="B55">55</xref>]. Phylogenetic analyses of multiple gene families bear out this impression, displaying evidence of duplications at every phylogenetic level (e.g., [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B56">56</xref>-<xref ref-type="bibr" rid="B58">58</xref>]). The lower eudicots appear to be a particularly active period for MADS box gene duplication (reviewed [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B59">59</xref>]), leading to the suggestion that at least one genome duplication occurred during this period [<xref ref-type="bibr" rid="B60">60</xref>]. Given what may be a relatively high rate of paralog generation, even very rare events such as the appearance of an adaptive frameshift mutation will occur at low frequency. Once such a frameshifted allele appears, it will be subject to the usual microevolutionary forces and may be fixed due to selection or neutral processes. Along these lines, it has been suggested that periods of paralog maintenance due to neutral forces or subfunctionalization may eventually facilitate neofunctionalization [<xref ref-type="bibr" rid="B61">61</xref>,<xref ref-type="bibr" rid="B62">62</xref>].</p><p>Of course, it is only the evolutionarily successful events, or the fairly recent ones, that can be easily detected. Many such molecular 'monsters' may have come and gone over the course of plant evolution. This is not to say that frameshift-based evolution is restricted to plants, since it has also been identified in vertebrates [<xref ref-type="bibr" rid="B49">49</xref>]. In these cases, the presence of differentially spliced transcripts is associated with frameshift sequence remodeling. It remains to be seen whether duplication-related frameshift will also be uncovered in animals or if the variable transcript phenomenon will predominate. Other instances of clustered non-synonymous nucleotide changes have been identified [<xref ref-type="bibr" rid="B63">63</xref>], which demonstrate that such events can be maintained by selection. These examples may also provide candidates to be re-examined for evidence of frameshift mutation since the failure to recognize a frameshift mutation would result in a nucleotide alignment with the signature of successive non-synonymous substitutions. It is important to note, however, that the 'hopeful monster' analogy only applies to the evolutionary pattern of the protein sequence. At the nucleotide level, the sequence changes are, in fact, quite gradual.</p></sec><sec><title>Implications for the evolution of the <italic>AP3 </italic>lineage and the ABC program</title><p>The rapid generation and fixation of the euAP3 motif raises obvious questions regarding its biochemical function and its evolutionary significance. In order to consider these issues, we must first outline our basic knowledge of B gene function in model species. In <italic>Arabidopsis</italic>, AP3 and PI function as obligate heterodimers to promote petal and stamen identity [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B64">64</xref>]. All aspects of their function appear to be interconnected since their heterodimerization through the I and K domains is a requirement for protein stability [<xref ref-type="bibr" rid="B65">65</xref>,<xref ref-type="bibr" rid="B66">66</xref>], nuclear localization [<xref ref-type="bibr" rid="B67">67</xref>], DNA binding [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B68">68</xref>] and the maintenance of gene expression [<xref ref-type="bibr" rid="B69">69</xref>,<xref ref-type="bibr" rid="B70">70</xref>]. The contribution of the C-terminal motifs to these functions is not well understood. As mentioned previously, it has been demonstrated that the euAP3 motif is required for proper AP3 function and that the paleoAP3 motif is not biochemically equivalent to the euAP3 in <italic>Arabidopsis </italic>[<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. The study of Lamb and Irish further determined that the euAP3 motif is capable of conferring AP3-specific function to PI. This result is particularly intriguing since it suggests that dimers between the endogenous PI and chimeric PI<sub>cAP3 </sub>proteins were stabilized when one of the PI proteins possessed a euAP3 motif. Although indirect, this is the best evidence we have to support a role for the euAP3 motif in mediating protein-protein interactions. As to the paleo<italic>AP3 </italic>motif, a study in <italic>Lilium </italic>has argued that this region contributes to the novel homodimerization capacity of the paleo<italic>AP3 </italic>homolog and, further, that the Lilium paleo<italic>AP3 </italic>motif is sufficient to confer homodimerization capability on <italic>AP3 </italic>itself [<xref ref-type="bibr" rid="B71">71</xref>]. These findings are highly surprising given that all previous studies have shown that the C domain as a whole plays no role in AP3/PI dimerization [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B72">72</xref>]. Additionally, other analyses of both <italic>TM6 </italic>and paleo<italic>AP3 </italic>orthologs have not recovered any evidence of homodimerization [<xref ref-type="bibr" rid="B34">34</xref>,<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B73">73</xref>,<xref ref-type="bibr" rid="B74">74</xref>]. Despite the conflicting nature of this set of results, it remains true that all specific investigations of AP3 motif function have indicated that it plays a role in mediating protein-protein interactions.</p><p>Following from this statement, it is natural to now consider the known interaction partners of AP3. The current model of ABCE gene function holds that AP3/PI dimers form higher order complexes with other type II MADS box proteins from the A, C and E classes. In <italic>Arabidopsis</italic>, these genes are represented by <italic>APETALA1 </italic>(<italic>AP1</italic>) in the A class, <italic>AGAMOUS </italic>(<italic>AG</italic>) in the C class and the <italic>SEPALLATA1</italic>-<italic>4 </italic>loci in the E class (reviewed [<xref ref-type="bibr" rid="B75">75</xref>]). Therefore, in petals AP3/PI would interact with AP1/SEP dimers and in the stamens, with AG/SEP dimers [<xref ref-type="bibr" rid="B76">76</xref>]. This model is assumed to essentially hold for all other core eudicots, with supporting evidence in <italic>Antirrhinum </italic>and <italic>Petunia </italic>[<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B77">77</xref>-<xref ref-type="bibr" rid="B80">80</xref>]. Unfortunately, the broader findings concerning the functions of C-terminal motifs within the context of these higher order complexes tend to be somewhat contradictory. On the one hand, complete deletion of the motifs does not generally affect complex formation in yeast three- or four-hybrid analyses [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B19">19</xref>] but, on the other hand, a separate yeast three-hybrid study recovered mutations in the C-terminal PI motif that did affect interactions with SEP proteins [<xref ref-type="bibr" rid="B17">17</xref>]. Similarly, the ability of <italic>PI</italic><sub>cAP3 </sub>to rescue <italic>AP3 </italic>function may suggest a role for the euAP3 motif in higher order interactions [<xref ref-type="bibr" rid="B6">6</xref>]. Since the C-terminus is not required for AP3/PI dimerization [<xref ref-type="bibr" rid="B14">14</xref>], the apparent stabilization of the PI/PI<sub>cAP3 </sub>dimer is unlikely to be due to a direct interaction between the euAP3 motif and PI. It is more probable that the presence of the euAP3 motif allows the weakly associated dimer to interact with other proteins, thereby stabilizing the whole complex. One explanation for this diverse set of results is that there are other proteins participating in complex formation <italic>in planta </italic>that are not represented in the yeast experiments and it is these co-factors that are the targets of C-terminal motif interactions. Alternatively, it may simply be that the yeast system is not always sensitive enough to detect alterations in interaction strength that are significant <italic>in vivo</italic>.</p><p>Given that our current understanding of C-terminal motif functions is confusing at best, it is also useful to consider the evolutionary histories of the loci thought to interact with AP3. In the case of <italic>PI</italic>, there is currently no clear evidence for a coincident gene duplication. Moreover, although there are sequence synapomorphies for core eudicot <italic>PI </italic>homologs, none of these map to the C-terminus and the MIK-associated residues do not represent obvious candidates for co-evolutionary changes (Kramer and Hu, unpublished data; [<xref ref-type="bibr" rid="B28">28</xref>]). Interestingly, the <italic>AG </italic>and <italic>SEP1</italic>/<italic>4 </italic>lineages both duplicated close to the base of the core eudicots [<xref ref-type="bibr" rid="B81">81</xref>,<xref ref-type="bibr" rid="B82">82</xref>]. However, AG has been shown to be unable to interact with AP3/PI on its own [<xref ref-type="bibr" rid="B19">19</xref>] and neither <italic>AG </italic>nor <italic>SEP1 </italic>underwent any major sequence remodeling in association with their basal eudicot duplications [<xref ref-type="bibr" rid="B81">81</xref>,<xref ref-type="bibr" rid="B82">82</xref>]. In contrast, the gene lineage containing <italic>AP1 </italic>is of particular interest given that it exhibits an evolutionary pattern which closely parallels that of <italic>AP3 </italic>[<xref ref-type="bibr" rid="B48">48</xref>]. Specifically, this lineage duplicated close to the base of the core eudicots to produce the paralogous eu<italic>AP1 </italic>and eu<italic>FUL </italic>lineages. Similar to eu<italic>AP3</italic>, the eu<italic>AP1 </italic>genes are divergent in sequence relative to both eu<italic>FUL </italic>and the ancestral <italic>FUL</italic>-like lineage. Perhaps most surprising is that the remodeling of the eu<italic>AP1 </italic>C-terminus also involved a frameshift mutation, although the exact extent of this phenomenon remains unclear [<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B48">48</xref>]. In the case of eu<italic>AP1</italic>, the single ancestral FUL-like motif was lost and two new conserved motifs evolved: one being involved in transcriptional activation (termed the euAP1 motif) and the other a site of post-translational farnesylation [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. No clear data exist, however, regarding the function of the ancestral FUL-like motif or to suggest that the eu<italic>AP1 </italic>motifs play a role in higher order complex formation.</p><p>Although it has been proposed that the appearance of the euAP3 and euAP1 motifs may have been a co-evolutionary phenomenon [<xref ref-type="bibr" rid="B22">22</xref>], there are at least two variations on this theme that could fit the data. These two hypotheses yield sets of opposing and, most importantly, testable predictions. One possibility is that the new motifs promote interaction with each other in a manner that their ancestors did not. This theory is consistent with the idea that eu<italic>AP1 </italic>and eu<italic>AP3 </italic>acquired their common role in petal identity at the base of the core eudicots [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B22">22</xref>]. Supporting evidence includes the fact that AP1 orthologs can interact with AP3/PI heterodimers on their own, although this does not appear to be dependent on their C-terminal motifs [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B19">19</xref>]. Also, as opposed to the equivocal situation with eu<italic>AP3 </italic>homologs [<xref ref-type="bibr" rid="B41">41</xref>], significant data exist to suggest that the role of eu<italic>AP1 </italic>in petal identity is specific to the core eudicots [<xref ref-type="bibr" rid="B35">35</xref>,<xref ref-type="bibr" rid="B48">48</xref>]. A second scenario is that it was the ancestral FUL-like and paleoAP3 motifs that directly interacted and that, following the gene duplications, the loss of one of these motifs released the other from selection and allowed it to diverge to new function. This theory is more consistent with the lack of data indicating a protein interaction function for the eu<italic>AP1 </italic>motifs. It is interesting to note that the FUL-like motif is strongly similar to the C-terminal motif of the <italic>SEP </italic>lineages [<xref ref-type="bibr" rid="B48">48</xref>,<xref ref-type="bibr" rid="B81">81</xref>], which are found within the same subfamily as <italic>AP1</italic>/<italic>FUL </italic>[<xref ref-type="bibr" rid="B8">8</xref>]. It may be that the loss of the FUL-like motif in eu<italic>AP1 </italic>could be compensated by its conservation in the SEP proteins, which are thought to participate in the same complex. In terms of testable hypotheses, analyses of protein interactions among pre-duplication taxa could help to distinguish between the two models. On the whole, we are left with an intense sense of coincidence – that the <italic>AP3 </italic>and <italic>AP1/FUL </italic>lineages both duplicated and experienced C-terminal frameshift mutation in the same approximate phylogenetic vicinity. Understanding the full significance of this coincidence awaits the definitive establishment of the functions of the C-terminal motifs.</p></sec></sec><sec><title>Conclusion</title><p>Phylogenetic analysis of an expanded set of <italic>AP3 </italic>homolog sequences indicates that the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication event occurred very close to the base of the core eudicots in association with the Trochodendraceae and Buxaceae lineages. The current dataset also reveals that the transition from the ancestral paleoAP3 motif to the derived euAP3 motif was primarily mediated by a single nucleotide deletion. The new motif appears to have become conserved with relatively little additional change, a somewhat extraordinary finding highlighting the potential for 'punctuated equilibrium' [<xref ref-type="bibr" rid="B83">83</xref>] to act at the molecular level as well as the morphological. It seems likely that the existence of a conserved second paralog facilitated the maintenance of the frameshift mutation. This finding fits with original models of gene duplication as a major source for genetic and biochemical diversification [<xref ref-type="bibr" rid="B84">84</xref>]. Current evidence regarding the biochemical functions of these C-terminal motifs is largely indirect and often contradictory, underscoring the importance of targeting these regions for further analysis.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Characterization of <italic>APETALA3 </italic>homologs</title><p>Homologs of <italic>AP3 </italic>were cloned from select taxa (see Fig. <xref ref-type="fig" rid="F1">1</xref>) using reverse transcriptase polymerase chain reaction (RT-PCR) on floral RNA following the protocol described by Stellari et al. [<xref ref-type="bibr" rid="B27">27</xref>] and Kramer et al. [<xref ref-type="bibr" rid="B28">28</xref>]. 5' rapid amplification of cDNA ends (RACE) was performed on <italic>TroAP3 </italic>using 5' RACE system (Invitrogen<sup>™ </sup>Life Technologies, Carlsbad CA). Reverse primers are as follows: for the first round of PCR, TroAP3-KR1 5' CTTTTTCCTGTCCGTCTCAGTCTG, and for the second round, TroAP3-KR2 5' TCCACCCGTCCTTCGCCCAATTTC. Sequences have been deposited in GenBank under accession numbers DQ453773-DQ453775 and DQ479353-DQ479368 (see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>).</p></sec><sec><title>Phylogenetic analyses</title><p>In addition to the 20 new loci obtained in the current study, 61 other core eudicot, basal eudicot, magnoliid, monocot and ANITA grade <italic>AP3 </italic>homologs were identified based on previously published analyses and BLAST searches [<xref ref-type="bibr" rid="B85">85</xref>] (see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref> for references and accession numbers). In cases where GenBank contained nearly identical sequences from the same taxon, only one representative sequence was included. Full-length nucleotide alignments of the loci were initially compiled using ClustalW. ClustalW multiple alignment parameters were gap penalty 8 and gap extension penalty 2, transitions weighted for the nucleotide alignment. The alignments were then refined by hand using MacClade 4.06 [<xref ref-type="bibr" rid="B86">86</xref>]. The hypothesized single nucleotide deletion in the C-terminus of eu<italic>AP3 </italic>lineage members was incorporated into the alignment (see <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref> for complete alignment in NEXUS format).</p><p>Maximum likelihood (ML) phylogenetic analyses were performed using PAUP* [<xref ref-type="bibr" rid="B87">87</xref>]. We used Modeltest [<xref ref-type="bibr" rid="B88">88</xref>] with the standard Akaike Information Criterion (AIC) to determine the simplest and most appropriate evolutionary model for our dataset. The models selected were a general time-reversible model (GTR) with a proportion of invariable sites (I) and a gamma approximation to the rate of variation among sites (Γ). The ML analysis used a single heuristic search with 100 random addition replicates, TBR branch swapping, MULPARS, and the steepest descent options. Branch support was estimated by performing 100 replicates of nonparametric bootstrapping using the same parameters as the original analysis. We also performed maximum parsimony (MP) analysis on the dataset using a heuristic tree search with 1000 random addition sequence replicates and TBR branch swapping. Support was estimated by performing 1000 bootstrap support replicates each with 10 random sequence addition replicates. The MP phylogeny is not shown (see text).</p></sec><sec><title>Analysis of nucleotide diversity and ancestral character state reconstructions</title><p>The program DnaSP [<xref ref-type="bibr" rid="B89">89</xref>] was used to determine the position-by-position nucleotide diversity of two small alignments derived from the full-length nucleotide dataset. The first alignment contains the C-terminal paleoAP3 motif-encoding region of loci from the <italic>TM6 </italic>lineage and the paleo<italic>AP3 </italic>lineage of basal eudicots. All indels were removed from the DnaSP alignment (see <xref ref-type="supplementary-material" rid="S4">Additional file 4</xref>). The second alignment contains the C-terminal euAP3 motif-encoding region of loci from the eu<italic>AP3 </italic>lineage (all core eudicots). All indels were removed from the DnaSP alignment except for the single nucleotide deletion that produced the euAP3 motif (see <xref ref-type="supplementary-material" rid="S5">Additional file 5</xref>). The DNA Polymorphism function was used to determine the nucleotide diversity (π, [<xref ref-type="bibr" rid="B90">90</xref>]) for each position in the two alignments.</p><p>Ancestral nucleotide character state reconstructions were performed using both MP and ML methods. For these analyses, we used the complete nucleotide alignment and the ML phylogeny. MP reconstructions were performed using the accelerated transitions (ACCTRAN) and delayed transitions (DELTRAN) options as they are implemented in MacClade 4.0 [<xref ref-type="bibr" rid="B86">86</xref>]. ML reconstructions were performed using the approach of Yang et al. [<xref ref-type="bibr" rid="B91">91</xref>] that is implemented in PAML [<xref ref-type="bibr" rid="B92">92</xref>]. As has been found in other cases where changes are relatively rare ([<xref ref-type="bibr" rid="B93">93</xref>] and references therein), the MP and ML reconstructions were identical. Given the fact that the relevant nodes have poor support, we also performed ancestral character state reconstructions with alternative topologies. Specifically, we tested a phylogeny where the <italic>Pachysandra </italic>loci are placed before the eu<italic>AP3</italic>/<italic>TM6 </italic>duplication (see Fig. <xref ref-type="fig" rid="F5">5B</xref>). In addition, we rearranged the eu<italic>AP3 </italic>and <italic>TM6 </italic>clade members such that their relationships were consistent with published core eudicot relationships. For this set of analyses, we tried two alternative topologies, one consistent with Soltis et al. 2003 [<xref ref-type="bibr" rid="B42">42</xref>] and the other, with Kim et al. [<xref ref-type="bibr" rid="B43">43</xref>].</p></sec></sec><sec><title>Authors' contributions</title><p>EK characterized <italic>AP3 </italic>homologs from <italic>Ilex</italic>, <italic>Kalanchoe</italic>, <italic>Saxifraga</italic>, <italic>Corylopsis</italic>, <italic>Pachysandra</italic>, <italic>Phytolacca</italic>, <italic>Paeonia </italic>and <italic>Vitis</italic>; conducted the phylogenetic analyses and ancestral state reconstructions; and drafted the manuscript. HJS and JMH characterized <italic>AP3 </italic>homologs from <italic>Loranthus </italic>and <italic>Trochodendron</italic>; and helped draft the manuscript. CCW characterized the <italic>AP3 </italic>homolog from <italic>Nelumbo </italic>in the laboratory of JMH. All authors read and approved of the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional file 1</title><p><bold>Table with Locus information </bold>Taxa of origin, GenBank accession numbers and reference information for all loci included in the alignment (sorted alphabetically by taxon).</p></caption><media xlink:href="1471-2148-6-30-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional file 2</title><p><bold>Alignment of C-terminal regions of predicted proteins of paleo<italic>AP3</italic>, <italic>TM6 </italic>and eu<italic>AP3 </italic>representatives. </bold>Phylogenetic affinities of the taxa are indicated by the bars on the left (BE = Basal Eudicot; based on [42, 95]. The PI Motif-derived region is boxed in green; paleoAP3 motifs, in blue; and euAP3 motifs, in purple. Residues showing chemical conservation with the consensus for each of these regions [28] are shaded in grey. Red arrows at the right indicate the loci that appear to have experienced independent frameshift mutations.</p></caption><media xlink:href="1471-2148-6-30-S2.eps" mimetype="application" mime-subtype="postscript"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional file 3</title><p><bold><italic>APETALA3 </italic>nucleotide alignment </bold>NEXUS format file of complete <italic>APETALA3 </italic>nucleotide alignment used in current phylogenetic analyses.</p></caption><media xlink:href="1471-2148-6-30-S3.nex" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S6"><caption><title>Additional file 6</title><p><bold>Comparison of position-by-position nucleotide diversity values for paleoAP3 and euAP3 motif containing loci. </bold>Complete dataset of nucleotide diversity values of paleoAP3 and euAP3 containing loci. Region spans the entire C-terminal motif. The yellow bars indicate the values for a dataset including all <italic>TM6 </italic>lineage members and basal eudicot paleo<italic>AP3 </italic>loci. The codon positions of each nucleotide are indicated by vertical hash marks and the corresponding amino acids are shown immediately below the chart. Note that the last four nucleotides in the paleoAP3 alignment are 3' UTR. The blue bars indicate the values for a dataset including all eu<italic>AP3 </italic>lineage members. The codon positions of each nucleotide are indicated by vertical hash marks and the corresponding amino acid are shown at the bottom. The position of the euAP3 frameshift is represented by a dash mark. n/a = not applicable.</p></caption><media xlink:href="1471-2148-6-30-S6.eps" mimetype="application" mime-subtype="postscript"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S4"><caption><title>Additional file 4</title><p><bold>PaleoAP3 alignment for nucleotide diversity calculation </bold>Alignment of paleoAP3 encoding regions of <italic>Pachysandra </italic>loci, <italic>TM6 </italic>orthologs and basal eudicot paleo<italic>AP3 </italic>representatives. Indels were removed from the alignment.</p></caption><media xlink:href="1471-2148-6-30-S4.nex" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S5"><caption><title>Additional file 5</title><p><bold>EuAP3 alignment for nucleotide diversity calculation </bold>Alignment of euAP3 motif encoding regions of eu<italic>AP3 </italic>lineage members. All indels were removed except for the single nucleotide deletion corresponding to the euAP3 motif frameshift.</p></caption><media xlink:href="1471-2148-6-30-S5.nex" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Hybridization interactions between probesets in short oligo microarrays lead to spurious correlations
|
<sec><title>Background</title><p>Microarrays measure the binding of nucleotide sequences to a set of sequence specific probes. This information is combined with annotation specifying the relationship between probes and targets and used to make inferences about transcript- and, ultimately, gene expression. In some situations, a probe is capable of hybridizing to more than one transcript, in others, multiple probes can target a single sequence. These 'multiply targeted' probes can result in non-independence between measured expression levels.</p></sec><sec><title>Results</title><p>An analysis of these relationships for Affymetrix arrays considered both the extent and influence of exact matches between probe and transcript sequences. For the popular HGU133A array, approximately half of the probesets were found to interact in this way. Both real and simulated expression datasets were used to examine how these effects influenced the expression signal. It was found not only to lead to increased signal strength for the affected probesets, but the major effect is to significantly increase their correlation, even in situations when only a single probe from a probeset was involved. By building a network of probe-probeset-transcript relationships, it is possible to identify families of interacting probesets. More than 10% of the families contain members annotated to different genes or even different Unigene clusters. Within a family, a mixture of genuine biological and artefactual correlations can occur.</p></sec><sec><title>Conclusion</title><p>Multiple targeting is not only prevalent, but also significant. The ability of probesets to hybridize to more than one gene product can lead to false positives when analysing gene expression. Comprehensive annotation describing multiple targeting is required when interpreting array data.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Okoniewski</surname><given-names>Michał J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Miller</surname><given-names>Crispin J</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Bioinformatics
|
<sec><title>Background</title><p>Sources of noise in microarray experiments may be numerous [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>], thus most researchers try to minimize its influence or estimate it through various quality control, normalization and outlier filtering procedures [<xref ref-type="bibr" rid="B3">3</xref>]. One source of variation is cross-hybridization (CH), which occurs when unintended sequences hybridize to a probe alongside the intended target. In the case of Affymetrix arrays, which use a set of short (typically 25-mer) oligonucleotide probes to target a transcript, hybridization conditions are carefully controlled with the aim of minimizing the effect of CH due to non-specific binding [<xref ref-type="bibr" rid="B4">4</xref>]. In addition, each Perfect Match (PM) probe is accompanied by a Mismatch probe (MM), in which the middle residue has been changed. The intention is that this can be used to provide a measure of the level of CH associated with each PM probe. A more detailed discussion of CH in short oligo arrays may be found in [<xref ref-type="bibr" rid="B5">5</xref>]. From October 2004, Affymetrix also started to display brief summaries of cross-hybridization within their own NetAffx service [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>In some circumstances, probes may match exactly to more than one transcript. This is important because these probes can no longer be identified with a unique transcript, but are instead dependent on more than one gene product. The situation is rendered somewhat more complex by the fact that Affymetrix arrays use more than one probe (typically, 11 PM/MM pairs – together referred to as a "probeset") to target each transcript. Recently, several databases have been built to provide a mapping of Affymetrix probesets to known transcripts [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>], to sequences from cDNA microarrays [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>], or for applying algorithmic approaches to cross-platform or cross-species comparisons [<xref ref-type="bibr" rid="B7">7</xref>]. A recent paper [<xref ref-type="bibr" rid="B13">13</xref>] presents a global overview of the interpretation of GeneChip arrays, and the need to update annotation to match the continued evolution of genomic databases. The solution includes the redefinition of CDF files, similar to what was proposed initially in [<xref ref-type="bibr" rid="B10">10</xref>], which may be sufficient in many cases.</p><p>The issue of 'multiply targeted' probes is important because they have the potential to result in cross-talk between the probesets they are part of. If their effects are significant, and expression summarizing algorithms are unable to control for them, then one outcome of this will be that otherwise unrelated probesets will appear correlated, since they are being driven by a shared signal.</p><p>The ADAPT database [<xref ref-type="bibr" rid="B4">4</xref>] was used to investigate the extent and significance of multiply-targeted probesets in Affymetrix expression data (see methods). Use is made of the fact that the platform's combination of short oligos and strict hybridization conditions, which are designed to maximize binding to the PM probes whilst minimizing binding to the MM ones. This makes it viable to use <italic>in silico </italic>methods to identify which probes are likely to bind with 100% identity to which transcripts. We refer to cases of exact matches between probe and transcript as Multiple Targeting (MT), to distinguish from the more general case of cross-hybridization, in which matches with less than 100% identity may occur.</p><p>Particular attention is directed at the influence MT can have on the apparent correlation between probesets' expression measurements. Since Pearson correlation is scale independent, it is not influenced by the overall magnitude of either signal being compared, but rather on the similarity in their shapes. Although it may seem counter-intuitive, when two signals are superimposed, the amount of correlation found between each of the original signals and the combined one is driven by the relative variance of those two signals, not by their mean intensity (an example and further discussion of this can be found in the supplemental material). Many microarray data analysis techniques rely on correlation analysis, with the majority of methodologies aiming to draw a distinction between genes that are, in some way, co-occurring, co-expressed or correlated and those that do not follow a significant common pattern. Methodologies such as hierarchical clustering [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>] and relevance networks [<xref ref-type="bibr" rid="B16">16</xref>-<xref ref-type="bibr" rid="B18">18</xref>] make direct use of the Pearson correlation coefficient of expression values between probesets, whilst others (such as ANOVA and more general linear models), are ultimately based on correlation-like principles.</p></sec><sec><title>Results</title><p>Affymetrix arrays use a series of probes to target a transcript. These probes are grouped together to form a probeset; expression processing algorithms such as MAS5 [<xref ref-type="bibr" rid="B19">19</xref>], RMA [<xref ref-type="bibr" rid="B20">20</xref>] and GCRMA [<xref ref-type="bibr" rid="B21">21</xref>] combine the signals from each probe in a probeset to provide a single summary value representing an estimate of the concentration of that transcript in solution. The issue of MT arises because certain probes are capable of hybridizing to more than one transcript, leading to non-independence, while in other situations probes from more than one probeset are capable of hybridizing to a single transcript (Figure <xref ref-type="fig" rid="F1">1</xref>). In general, these interactions combine to form a complex lattice (Figure <xref ref-type="fig" rid="F2">2</xref>, see also <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>).</p><p>In this paper, we consider the extent and structure of these relationships, followed by an investigation of how much effect they have both on signal strength and on the correlation between probesets.</p><sec><title>The prevalence of multiple targeting in oligo arrays</title><p>An analysis of the HG_U133A array reveals that many transcripts (Ensembl: 7,257; RefSeq: 6,702) are matched with multiple probesets (i.e. case a in Fig. <xref ref-type="fig" rid="F1">1</xref>) while almost half (10,223) of the total 22,215 probesets (excluding control probesets) show exact matches (with 1 or more PM probes) to more than one Ensembl (9,460) or RefSeq (9,666) transcript (i.e. case b in Fig. <xref ref-type="fig" rid="F1">1</xref>). For comparison, 18,722 probesets were found to match to at least one well-known transcript.</p><p>The effect of MM probes is minimal: the number of MM probes that can hybridize exactly to known transcripts is about 1,000 times smaller (Ensembl: 1,899 MM matches vs. over 1,956,000 PM matches, RefSeq: 1,962 MM vs. 1,922,000 PM) – most of them singleton matches to unrelated sequences. Thus we exclude MM probes from subsequent analyses. Since MM probes were not considered, and RMA makes no use of these probes in its computations, RMA processed data is used for all calculations presented here, although similar effects were also observed with MAS5 processing.</p><p>Affymetrix probeset names are supposed to identify probesets that are associated with multiple targeting. In particular, those marked "_x_at" are identified as being non-specific. Similarly, "_s_at" probesets are identified as potentially targeting different gene family members or splice variants. The analysis shows that many of the probesets associated with MT are not identified in this way and are simply annotated " _at" (2,189 according to Ensembl matches; 1,496 for RefSeq). These numbers are likely to be underestimates because ADAPT was built using only well characterized sequences. Thus, a significant number of the standard "_at" probesets are involved in MT.</p></sec><sec><title>Structures of multiple targeting in oligo arrays</title><sec><title>Basic motifs</title><p>The two basic building blocks of MT interaction networks are Probeset-Transcript-Probeset (PTP) motifs (Figure <xref ref-type="fig" rid="F1">1a</xref>), and Transcript-Probeset-Transcript (TPT) motifs (Figure <xref ref-type="fig" rid="F1">1b</xref>). Depending on the robustness of the analysis algorithms used to process array data, the presence of either motif can be expected to lead to non-independence between the expression profiles of the participating probesets.</p><p>A search for both types of motifs confirms the prevalence of MT in oligo arrays. Table <xref ref-type="table" rid="T1">1</xref> summarizes the rates of occurrence of both motifs for a variety of Affymetrix arrays. The PTP motif is especially common – it involves almost half the probesets on the HG_U133A array and over a third of those on the HG_U133Plus2. Generally, the more recent arrays have a larger proportion of probesets involved in MT.</p></sec><sec><title>Families of related probesets</title><p>Probesets may be involved in multiple PTP and TPT motifs, resulting in an MT-network. This can be expressed as a graph in which nodes represent transcripts and probesets, while edges represent matches between transcripts and probesets, labelled with the number of matching probes involved in the interaction. Such graphs are informative because so many probesets have the potential to be involved in MT (almost half for HGU133A arrays). Since Affymetrix arrays measure the binding of cRNA sequences to sequence-specific probes, the searches used to define MT help catalogue which binding events are possible. Knowledge of MT interactions is important because it begins to describe what is actually being measured in a microarray experiment.</p><p>Figure <xref ref-type="fig" rid="F2">2</xref> shows one such graph, laid out using LGL [<xref ref-type="bibr" rid="B22">22</xref>]. Edges attached to RefSeq transcripts are painted red, Ensembl ones, green. Blue is used to mark the strength of MT, with intensity corresponding to the number of matching probes. The LGL graph, when magnified, shows a set of disconnected families -detached sub-graphs of various complexity. Thus, almost all of the MT relationships are local ones.</p><p>To build families, the database was queried to identify all PTP motifs. Then, a simple search algorithm used to identify the maximal graph that can be reached from a starting probeset using the identified motifs. Probesets that are not involved in any PTP motifs result in trivial families that consist of just a single probeset. An additional step is used to eliminate "hub probesets", as described below.</p><p>For HG_U133A arrays, this process results in the identification of 3,859 families containing at least 2 probesets (for examples – see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). The mean number of probesets in the family is not high -about 2.56. Interestingly, 429 families (involving a total of 1,529 probesets), were found in which family members were annotated to different genes. Importantly, these families were not simply comprised of "_x_at" probesets: 456 were annotated "_at" and 497 – "_s_at".</p><p>A full list of MT families is included in the supplementary data (see <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>), along with an applet that allows the exploration of these families, attached to exemplary expression data (see <xref ref-type="supplementary-material" rid="S4">Additional file 4</xref>).</p></sec><sec><title>Hub probesets</title><p>There is a group of probesets (not always annotated by Affymetrix as " _x_at") that match a large number of transcripts, usually with a small number of probes. They may be called "hub" probesets, because their expression combines signals from many available transcripts. In the network of probeset-transcript relationships, hub probesets often join together smaller families of probesets, often many at a time. A typical example of a hub probeset is "221992_at" which matches to 44 RefSeq or Ensembl transcripts, with an average 3.18 probes per match, or "210524_x_at" (127 matches, 1.5 probe on average).</p><p>Hubs were selected for the family search algorithm described above if the average number of matching probes was less than 3 and the total number of transcripts greater than 30, or if the total number of transcript matches was greater than 70. This resulted in 277 hub probesets being selected, allowing the granularity of families to be kept to a reasonable level (also see Table <xref ref-type="table" rid="T2">2</xref> for hub selection criteria).</p></sec></sec><sec><title>Quantitation of the effect of multiple targeting</title><p>Probes found by the database searches to target multiple transcripts, generally have a higher measured signal than those that target unique transcripts. For example, the average measured expression level in the Gene Atlas data is 16% higher for multiply targeted PM probes and over 80% higher when the PM – MM difference for individual PM:MM probe pairs is considered.</p><p>These numbers refer to differences in the raw probe intensities, which are subsequently grouped into probesets and processed by an expression summary tool such as MAS5 or RMA. The following sections investigate whether these changes at the probe level are carried through to the MAS5 or RMA processed expression summaries, and the influence they have on Pearson correlation.</p><sec><title>Real data, same transcript</title><p>Figure <xref ref-type="fig" rid="F1">1</xref> draws a distinction between transcripts that share a probeset, and probesets that share a transcript. The first case (PTP, la) is relatively trivial: we should expect to see correlation between these probesets. The extent of the excessive correlation is confirmed by Figure <xref ref-type="fig" rid="F3">3</xref>, which shows the distribution of the Pearson correlation coefficient calculated between every probeset pair on the array. The resultant distribution is almost normal, with a slight displacement (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2105-7-276-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mover accent="true"><mml:mi>r</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuWGYbGCgaqeaaaa@2E31@</mml:annotation></mml:semantics></mml:math></inline-formula> = 0.02, for Gene Atlas data processed by RMA, for other datasets the mean is comparably small). By contrast, when only multiply targeted probesets are considered (as in Figure <xref ref-type="fig" rid="F1">1a</xref>), the distribution is strongly distorted towards positive values (<inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M2" name="1471-2105-7-276-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mover accent="true"><mml:mi>r</mml:mi><mml:mo>¯</mml:mo></mml:mover><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuWGYbGCgaqeaaaa@2E31@</mml:annotation></mml:semantics></mml:math></inline-formula> = 0.55). Thus, as expected, probesets targeting the same transcript show much higher correlation than those that are not linked in this way. Similar results were also seen with MAS5 and GCRMA processed data (not shown). Importantly, this effect is not confined to probesets in which 11/11 probes match. Figure <xref ref-type="fig" rid="F4">4</xref> shows the distribution of Pearson correlation for probesets in which only a subset of probes are involved in MT. It can be seen that even a single matching probe can result in increased correlation. This is surprising given that oligo array data processing methods such as MAS5 and RMA are designed to be robust against outliers – a single probe behaving differently from its peers may not be expected to have a large influence on the data. This is investigated in more detail below.</p></sec><sec><title>Simulation data</title><sec><title>Intensity</title><p>Figure <xref ref-type="fig" rid="F1">1b</xref> shows a situation where the expression level of a probeset might be expected to be driven by two different transcripts. Since there is no independent estimate available for the expression levels of the individual transcripts involved in TPT motifs, simulation experiments were performed to mimic the effect by artificially spiking raw expression data.</p><p>Figure <xref ref-type="fig" rid="F5">5</xref> shows the results of one such simulation, designed to consider the effects of the presence of an additional transcript in equal abundance to the intended target. It can be seen that as the number of spiked probes increases, the signal becomes more pronounced. As previously observed with real data, a single matching probe can have a significant influence on the computed expression level. Even when the expression level is relatively high the signal from only 2 probes can be sufficient to lead to apparent differential expression. Even so, the largest fold changes are generally restricted to the lower intensity probesets, indicating that both MAS5 and RMA do a good job at reducing the effects of outliers.</p></sec><sec><title>Correlation</title><p>In a second simulation experiment, spiking was achieved by adding the signals from a second set of probes to the first set. In this way, the case shown in Figure <xref ref-type="fig" rid="F1">1b</xref> was simulated – i.e., a probeset hybridizing to two different transcripts (one with all the probes matching, the other with a varying number of matches). The second group of probesets was produced by randomly selecting up to 500 probesets. Variance filtering was performed to ensure that at least one of the transcripts had an expression profile that varied. Since Pearson correlation is not dependent on the mean intensity of the signals, but rather the similarity of their shapes, filtering was performed on variance, not intensity. Pearson correlation, <italic>r </italic>was calculated between each member of the first list and its corresponding partner in the second. Before spiking, the two sets should be uncorrelated; spiking is expected to increase correlation. As in the real data, the signal from the spiked probes contributes significantly to the correlation, even when only a small number of probes are involved. It can be seen from Figure <xref ref-type="fig" rid="F6">6</xref> that even when high variance probesets are the recipients of additional spiked signals changes in <italic>r </italic>are possible. Thus, the effects are not restricted to probesets with low signal (see also <xref ref-type="supplementary-material" rid="S5">Additional files 5</xref>, <xref ref-type="supplementary-material" rid="S6">6</xref>, <xref ref-type="supplementary-material" rid="S7">7</xref>, <xref ref-type="supplementary-material" rid="S8">8</xref>).</p></sec></sec><sec><title>Intensity vs. correlation</title><p>Both real and artificial datasets demonstrate that MT can have a significant effect on correlation, even when only a small proportion of probesets are involved in the interaction. Algorithms such as RMA and MAS5 successfully employ robust averaging techniques (such as median polishing or a Tukey's biweight) to reduce the effect of outliers. Thus, when only a small number of probes in a probeset are involved in MT, changes in measured expression level are expected to be generally small. This is confirmed in both the real and simulation datasets.</p><p>However, even when overall changes in intensity are minimal, increase in Pearson correlation can still be high. This is because Pearson correlation is driven by similarity in profile shape, not intensity; small amounts of stray signal can lead to large increases in <italic>r</italic>, even if the overall mean between probesets are very different. Since Pearson correlation centers each variable about its mean, and scales it by its standard deviation, correlation is entirely dependent on the relative shape and variance of the two signals, not their overall intensity. When two signals, <italic>a </italic>and <italic>b </italic>are compared to their sum, <italic>s</italic>, the signal that is most correlated with <italic>s </italic>depends not on their relative sizes, but on their relative variance. This is counter-intuitive but important to recognize when considering the effects of interacting signals on correlation (see <xref ref-type="supplementary-material" rid="S9">Additional file 9</xref>).</p><p>This effect can be demonstrated by varying the amount of contribution made by the spiking probes (<italic>f f </italic>-see Methods) to the resulting value. Figure <xref ref-type="fig" rid="F7">7</xref> shows that even when only 5% of the spike signal was present, the influence on Pearson correlation can still be large, even though the resultant fold change is generally small.</p><p>Together Figures <xref ref-type="fig" rid="F6">6</xref> and <xref ref-type="fig" rid="F7">7</xref> show that increases in correlation are not simply confined to those cases in which a large and varying signal is being added to a low-variance probesets.</p><p>In situations where probesets are already strongly correlated, the addition of extra signal due to cross hybridization with another transcript might be expected to reduce correlation. Spiking experiments found this to be the case (data not shown). Interestingly, however, even though there are occasions where <italic>r </italic>is reduced by multiple targeting, the general tendency is towards significantly increased correlation (as shown in Figure <xref ref-type="fig" rid="F3">3</xref>; similar figure for simulation experiments – see <xref ref-type="supplementary-material" rid="S10">Additional file 10</xref>).</p><p>False positive rates will also be raised because otherwise absent probesets with signals resulting only from background levels of non-specific hybridization can experience additional, structured, signal due to exact matches to transcripts other than the intended target.</p></sec></sec><sec><title>Functional homogeneity and spurious correlations in families of probesets</title><p>Analysis of MT-families shows that out of the 3,859 shown in Figure <xref ref-type="fig" rid="F2">2</xref>, 395 contained probesets annotated (using the BioConductor annaffy package [<xref ref-type="bibr" rid="B3">3</xref>]) to 2 or more UniGene clusters. When gene symbols are considered, annotation becomes even more ambiguous: 429 families contained transcripts annotated to different genes. Thus, even though the majority of families are homogenous with respect to UniGene and gene symbols – some 10–15% (depending on size of the family and source of annotation) may be annotated to different genes. This translates to about 1000 probesets.</p><p>As we have shown using both real and artificial data, MT leads to increased correlation. A consequence of this is that probesets associated by MT should be drawn closer to one another in dendrograms, such as those used to cluster probesets for visualisation using heatmaps. For example, the heatmap in Figure <xref ref-type="fig" rid="F8">8</xref> was created using three groups of probesets. The first (annotated to the genes RPS29,HFL-B5,EIF4A1,RPL36A and RPL18), was identified in [<xref ref-type="bibr" rid="B23">23</xref>] as discriminating between standard-risk and high-risk TEL-AML1 cytogenetic abnormalities. Non of these probesets are associated by MT and can thus be considered to form a "well behaving" biological family. The second set (annotated to TUBB6, TUBB2 and TUBB3) constitute another biological family, but they are also associated by MT to each other, as well as to other tubulin genes. This family thus represents a mixed biological – MT family. The third group of probesets are associated with RPS10, but also to numerous pseudogene transcripts. This group represents an "MT family" where the relationship is expected to be artefactual. These three sets of probesets were added to a further set of randomly selected probesets, to act as "background", and then clustered. The MT-family, the tubulins and the biological family are found as separate clusters (the MT-family with even closer links than others), demonstrating that the hierarchical clustering is unable to make a distinction between probable real (i.e. biological) and probably artefactual (i.e. MT driven) clusters.</p></sec></sec><sec><title>Discussion</title><p>It is clear that multiple targeting is an important artefact within microarray data: nearly half of all probesets on the HG_U133A array are associated with MT. When real expression data are considered, it can be seen that these probesets are significantly more correlated than would be expected by chance. These results are also supported by simulation experiments, using datasets derived from real experimental data, that allow MT to be considered in a more controlled framework. MT can lead to increased correlation between associated probesets, even when only a small proportion of their probes are involved. Although expression summary algorithms are successful at reducing the effects of outlier probes, the do not remove them completely, and small amounts of stray signal can still have a significant influence on correlation. The reason for this apparent paradox is the scale-invariance of Pearson correlation; absolute signal is not important. What is important are the variance and (effectively) the relative similarity in shape of the expression profiles. For this reason, particular care must be taken when analysing expression data using correlation-based approaches. The situation is also further complicated by the fact that MT occurs at a probe level – adding additional signal to individual probes within a probeset – but correlation is calculated after normalization and expression summarization using an algorithm such as RMA or MAS5. This additional complexity makes it difficult to reliably predict what will happen when signals are combined. However, empirical data (Figure <xref ref-type="fig" rid="F6">6</xref>) show that influence on correlation is dependent on the relative variance of the two probesets being combined. As expected, high variance spiked probes generally have more of an effect than low variance spikes, but interestingly, adding low variance spikes to low variance data (the magenta line in Figure <xref ref-type="fig" rid="F6">6</xref>) has more of an effect than adding high variance spikes to low variance data (the cyan line). This is likely to be a consequence of the expression summarization and normalization that is imposed on the data.</p><p>One consequence of MT is that because it serves to add structure to otherwise random probesets with no genuine signal, it can lead to the detection of false positives unless the presence of cross-matching probesets is known. Analysis of the intensity distributions for MT and non-MT probesets shows a considerable degree of overlap (see <xref ref-type="supplementary-material" rid="S11">Additional file 11</xref>). This means that MT probesets cannot be removed simply by filtering on intensity. In fact, because MT generally increases signal strength, such filtering might actually serve to enrich for MT probes.</p><p>MT is ultimately a sequence-based event; it occurs when two sequences show 100% identity across the 25bp targeted by a probe. At the level of a probeset, this is most likely to occur when transcripts show a high degree of sequence similarity. The relationships is troublesome, because a major use of expression data is to identify probesets (and, via annotation, genes) with correlated expression profiles, and to use these relationships to infer functional similarities. Since sequence similarity is itself often the basis by which common function is inferred [<xref ref-type="bibr" rid="B24">24</xref>], sequence similarity combined with MT has the potential to become a self-fulfilling prophecy.</p><p>A search of the database found that about 5% of family members contained probesets annotated to different genes. Thus, the chances of finding a spurious functional relationship due to MT between a pair of randomly selected genes is small. However, this is optimistic, because microarray analysis generally involves filtering to produce a set of significant probesets (either by magnitude of change, or by statistical confidence). The result of such filtering is to enrich the final 'hit list' not only for real biological effects, but also for anything else that is consistent, including biochemical or sequence based artefacts such as MT. This is illustrated by the heatmap in the Figure <xref ref-type="fig" rid="F8">8</xref>; MT families fall into separate clusters against a background of randomly selected probesets.</p><p>One possible solution to MT is to redefine probesets so that probes targeting the same transcript are placed into larger probesets representing the entire sequence, as proposed by [<xref ref-type="bibr" rid="B10">10</xref>]. This is the approach taken also by [<xref ref-type="bibr" rid="B13">13</xref>], but authors conclude that "under many circumstances it is not possible to generate transcript-specific probe sets for genes with multiple transcripts based on probes available on the current generation of GeneChips". Thus they may be used to make distinction at the level of genes, but not at the level of transcripts or splice variants – MT with all its consequences, still exists. There is a compromise to be made between generalisation and maintaining the ability to resolve subtle differences between transcripts and, for example, splice variants.</p><p>The issue becomes more significant with the new generation of microarrays such as the Affymetrix exon array [<xref ref-type="bibr" rid="B25">25</xref>] that deliberately use multiple probesets to distinguish between individual transcripts from within a set of splice variants expressed by a particular gene. The result is a many-many relationship between gene, transcript and probeset.</p><p>Annotation schemes that attempt to compress these many-many relationships into a one-to-one mapping lose the complexities inherent in the system. Grouping together a probeset that targets more than one transcript with probesets that target one or more individual transcripts, results in MT occurring between the new probeset and all the other transcripts it shares probes with. From one perspective, many these issues are simply down to annotation. The apparent aretefacts in the data only exist because the probeset annotations do not accurately reflect the transcripts they bind to.</p><p>With all solutions, including those that attempt to solve the problem by aggregating probes into larger probesets, annotation is crucial, since inaccuracies will arise unless all the many-many relationships that occur within the data are represented explicitly.</p></sec><sec><title>Conclusion</title><p>Cross hybridization between probesets is a significant effect that has real consequences for the interpretation of microarray data. It may cause a variety of problems during analysis including false positives and negatives, and generally increased correlation between multiply-targeted probesets. Although the results presented here are for Affymetrix arrays, it is reasonable to expect similar effects to occur with other expression-based technologies. The use of short oligos and strict hybridization conditions makes it possible to perform the <italic>in silico </italic>searches required to identify MT within Affymetrix data. However, CH is not exclusive to any one platform, and similar behaviour is likely to be seen elsewhere. Expression summary algorithms must correct not only for variation across arrays, but also for variation between individual probes within a probeset. This is generally performed using some kind of robust averaging procedure, but even small amounts of stray signal can lead to high correlations between probesets. Although algorithms such as RMA and MAS5 do a very good job of significantly reducing the influence of outlier probes, they do not always remove it completely – and this is manifested by significantly increased correlation between probesets, even when only a small subset of probes are involved.</p><p>Many of the issues described above can be avoided with more detailed annotation. Often the terms 'gene', 'transcript' and 'probeset' are used interchangeably. This is dangerous, because the relationship is not one-one-one, and the existence of MT networks can lead to apparent biological relationships that are, in fact, artefactual. Expression data that is presented simply as a gene list is difficult to interpret properly, since the complexities of the interaction networks implicit within the data are lost. The community should ensure that the actual probeset IDs are always available alongside gene names or transcript accessions. This allows the graph structures associated with gene-transcript-probeset mappings to be explored where necessary and used to fully interpret the complexities of gene expression data.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Graph rendering</title><p>MT networks and interaction graphs were produced by extracting data from ADAPT and redirecting the output for visualization to LGL [<xref ref-type="bibr" rid="B22">22</xref>], and our own visualization software. As global layouts of graphs such as LGL are static, thus not interactive, because the number of vertices is too big for efficient real-time rendering, an applet was developed for fast and flexible analyses of individual families. These small, local graphs within the applet were realized with the JUNG API [<xref ref-type="bibr" rid="B26">26</xref>].</p></sec><sec><title>Databases, experimental data sources and data processing</title><p>ADAPT [<xref ref-type="bibr" rid="B4">4</xref>] is a database of mappings between Affymetrix probesets, transcripts and genes. It is populated by searching all probe sequences for exact matches to transcript data taken from RefSeq (Release 11 at the time of writing) [<xref ref-type="bibr" rid="B27">27</xref>] and Ensembl (V30 at the time of writing) [<xref ref-type="bibr" rid="B28">28</xref>]. For RefSeq, both "known" and "model" sequences are used; for Ensembl, ADAPT uses those assigned "known", "novel" or "pseudo" status. Both databases are used because they employ different methods to predict transcript/gene sequences.</p><p>The ADAPT database was queried (using SQL and RdbiPqSQL database link to R) to extract a set of tables describing all possible MT links between probesets and transcripts, excluding anti-sense strand matches. The probesets may match transcripts with anything from 1 to 16 matching probes. The tables implicitly define an unconnected graph (see Figure <xref ref-type="fig" rid="F2">2</xref> and supplemental file 1), and form the basis for all subsequent explorations of MT. In order to consider the strength of the MT effect and its consequences on expression studies, data from ADAPT were combined with expression data from experiments generated using the HG-U133A array. Expression levels were produced using MAS5 and RMA, as implemented in Bioconductor (packages <italic>affy</italic> and <italic>simpleaffy </italic>[<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B29">29</xref>,<xref ref-type="bibr" rid="B30">30</xref>]).</p><p>Results were analogous when experiments were repeated with MAS5. All plots presented were generated using the Novartis Gene Atlas [<xref ref-type="bibr" rid="B31">31</xref>] dataset. Similar results were seen with both leukaemia [<xref ref-type="bibr" rid="B23">23</xref>] and sarcoma [<xref ref-type="bibr" rid="B32">32</xref>] datasets – publicly available from ArrayExpress.</p><p>Pearson correlation was calculated for all the pairs of probesets found to be targeting the same transcript. The distribution of correlation coefficients was calculated for all probeset pairs and for all pairs where one of the probesets matches to a transcript with less than a specified number of probes.</p></sec><sec><title>Simulation data</title><p>A subset of 50 HG_U133A arrays from Gene Atlas V2 was used as the basis for simulation experiments designed to explore how the number of MT probes influences expression measurements from RMA processed data.</p><p>Spiking was conducted as follows: prior to expression summary generation using RMA, 500 probesets were selected at random to be spiked and 500 (at random) to act as a source of spiking data. No filtering was applied to these probesets. Probesets were randomly paired, and between 1 and 10 probe-pairs selected for each probeset (again at random). The signals from the spike-sources were added to the original signals for the spike-targets. In this way TPT motifs were simulated. The resulting simulated data were batch normalized using RMA and compared to the original un-spiked data (again batch normalized using RMA, separately from the first set). In all simulation experiments spiking for the selected probesets was carried out across the entire set of arrays.</p><p>In the second experiment, a set of 500 probesets was selected, as before. A second set with the same number of probesets was then chosen at random. These probesets were selected from a subset of the probesets available, generated by filtering the expression data on variance. In this way, both sets could be sampled from probesets with specifically high, average or low variance of expression. High and low variance are defined as the top or bottom 2000 probesets, sorted by of variance, excluding the 100 most extreme ones. Each probeset in the second list was used to supply data for the probeset in the first list; between 1 and 11 probes were chosen and the probe intensities from the second list added to the corresponding probes in the first list. Various levels of influence were applied, adding a specific proportion of one probeset signal to another: <italic>PM</italic>1<sub>1<italic>after </italic></sub>= <italic>PM</italic><sub>1<italic>before </italic></sub>+ <italic>f f </italic>* <italic>PM</italic><sub>2</sub>, where for <italic>f f </italic>ranging from 0.05 to 1. In this way, cross-hybridization between probesets in the first list, and the transcripts represented by the probesets in the second list, was simulated.</p></sec></sec><sec><title>Abbreviations</title><p>ADAPT – "A Database of Affymetrix Probesets and Transcripts"</p><p>CH – cross-hybridization</p><p>LGL – Large Graph Layout</p><p>MAS5 – MicroArray Suite – Affymetrix algorithm (MAS 5.0)</p><p>MM – mismatch probe</p><p>MT – multiple targeting</p><p>PM – perfect match probe</p><p>PTP – probeset-transcript-probeset network motif</p><p>RMA – Robust Multichip Average algorithm</p><p>TPT – transcript-probeset-transcript network motif</p></sec><sec><title>Authors' contributions</title><p>MO developed the concept of interactions in families of probesets, carried out database and statistical analyses and drafted the manuscript, CM conceived the study on probes alignments to transcript and its implications, supervised and participated in its design and helped to draft the manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p>Graph of MT families for HGU133A array. Animated GIF, 3D visualization of MT-families in the array.</p></caption><media xlink:href="1471-2105-7-276-S1.gif" mimetype="image" mime-subtype="gif"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p>Examples of 3 families of probesets and transcripts. Screenshots from the applet. Big nodes signify probesets (green – positive detection call), small magenta ones – transcripts. The width of edges is proportional to the quantity of MT probes. Probes are marked with a name, annotation in Affymetrix or BioConductor and expression values. Presented are families associated mainly with PAX8, RUNX1/RPL22 and tubulins.</p></caption><media xlink:href="1471-2105-7-276-S2.tiff" mimetype="image" mime-subtype="tiff"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional File 3</title><p>List of MT families for HGU133A array. The CSV file lists all the discovered MT probeset families along with their gene-level annotations according to Affymetrix and BioConductor.</p></caption><media xlink:href="1471-2105-7-276-S3.csv" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S4"><caption><title>Additional File 4</title><p>Applet for families exploration. <ext-link ext-link-type="uri" xlink:href="http://bioinformatics.picr.man.ac.uk/adaptnet"/>. An applet for browsing graphs of MT-families in the HGU133A array. Big nodes represent HGU133A probesets: green ones have "Present" detection call, pink ones "Absent". They are labelled with Affymetfix and BioConductor annotations, detection call and expression value in the experiment. Small magenta nodes represent transcripts. There is a possibility to add Exon 1.0ST probesets (blue) to the graph. The width of edges is proportional to the number of matching probes. The applet is intended for online use – it is connected to an application server and ADAPT database.</p></caption><media xlink:href="1471-2105-7-276-S4.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S5"><caption><title>Additional File 5</title><p>Spiking experiment, signal filtering 1. Scatter plot and correlation distribution, generated as in Figure <xref ref-type="fig" rid="F7">7</xref>, but filtered by average signal intensity. Low intensity: the 10% probesets with lowest mean signal. High intensity: the 10% probesets with highest mean signal. Low intensity spikes added to high targets. The plots in <xref ref-type="supplementary-material" rid="S5">Additional files 5</xref>, <xref ref-type="supplementary-material" rid="S6">6</xref>, <xref ref-type="supplementary-material" rid="S7">7</xref>, <xref ref-type="supplementary-material" rid="S8">8</xref> prove that with any sort of signal intensity filtering, the shift in correlation coefficient occurs.</p></caption><media xlink:href="1471-2105-7-276-S5.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S6"><caption><title>Additional File 6</title><p>Spiking experiment, signal filtering 2. As <xref ref-type="supplementary-material" rid="S5">Additional file 5</xref>, but high intensity spikes added to high targets.</p></caption><media xlink:href="1471-2105-7-276-S6.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S7"><caption><title>Additional File 7</title><p>Spiking experiment, signal filtering 3. As <xref ref-type="supplementary-material" rid="S5">Additional file 5</xref>, but high intensity spikes added to low targets.</p></caption><media xlink:href="1471-2105-7-276-S7.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S8"><caption><title>Additional File 8</title><p>Spiking experiment, signal filtering 4. As <xref ref-type="supplementary-material" rid="S5">Additional file 5</xref>, but low intensity spikes added to low targets.</p></caption><media xlink:href="1471-2105-7-276-S8.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S9"><caption><title>Additional File 9</title><p>R code for correlation experiment. A simple experiment to consider the relationship between correlation cooefficient and variance using 10,000 randomly generated cases.</p></caption><media xlink:href="1471-2105-7-276-S9.R" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S10"><caption><title>Additional File 10</title><p>Influence of specific number of spiked probes on correlation. Changes in Pearson correlation following spiking to simulate MT between probesets. Red plot – correlation before spiking, orange – 1 spiked probe per probeset, magenta – up to 3 probes, blue – up to 7 probes, green – all probes spiked. As in the case of real data – even a single probe may influence the distribution of correlation, however in that case there are no effects of biological similarity – that's why the effect exists, but is smallest for single probes.</p></caption><media xlink:href="1471-2105-7-276-S10.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S11"><caption><title>Additional File 11</title><p>Distribution of the expression signal for MT and non-MT probesets after processing with RMA and MAS5. The plots (normalized distributions of summarized expression values) indicate a slight increase in the high signal values for MT probesets (blue) against non-MT probesets (green).</p></caption><media xlink:href="1471-2105-7-276-S11.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Applying dynamic Bayesian networks to perturbed gene expression data
|
<sec><title>Background</title><p>A central goal of molecular biology is to understand the regulatory mechanisms of gene transcription and protein synthesis. Because of their solid basis in statistics, allowing to deal with the stochastic aspects of gene expressions and noisy measurements in a natural way, Bayesian networks appear attractive in the field of inferring gene interactions structure from microarray experiments data. However, the basic formalism has some disadvantages, e.g. it is sometimes hard to distinguish between the origin and the target of an interaction. Two kinds of microarray experiments yield data particularly rich in information regarding the direction of interactions: time series and perturbation experiments. In order to correctly handle them, the basic formalism must be modified. For example, dynamic Bayesian networks (DBN) apply to time series microarray data. To our knowledge the DBN technique has not been applied in the context of perturbation experiments.</p></sec><sec><title>Results</title><p>We extend the framework of dynamic Bayesian networks in order to incorporate perturbations. Moreover, an exact algorithm for inferring an optimal network is proposed and a discretization method specialized for time series data from perturbation experiments is introduced. We apply our procedure to realistic simulations data. The results are compared with those obtained by standard DBN learning techniques. Moreover, the advantages of using exact learning algorithm instead of heuristic methods are analyzed.</p></sec><sec><title>Conclusion</title><p>We show that the quality of inferred networks dramatically improves when using data from perturbation experiments. We also conclude that the exact algorithm should be used when it is possible, i.e. when considered set of genes is small enough.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Dojer</surname><given-names>Norbert</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Gambin</surname><given-names>Anna</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Mizera</surname><given-names>Andrzej</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Wilczyński</surname><given-names>Bartek</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Tiuryn</surname><given-names>Jerzy</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Bioinformatics
|
<sec><title>Background</title><p>Since most genetic regulatory systems involve many components connected through complex networks of interactions, formal methods and computer tools for modeling and simulating are needed. Therefore, various formalisms were proposed to describe genetic regulatory systems, including Boolean networks and their generalizations, ordinary and partial differential equations, stochastic equations and Bayesian networks (see [<xref ref-type="bibr" rid="B1">1</xref>] for a review).</p><p>While differential and stochastic equations describe the biophysical processes at a very refined level of detail and prove useful in simulations of well studied systems, Bayesian networks appear attractive in the field of inferring the regulatory network structure from gene expression data [<xref ref-type="bibr" rid="B2">2</xref>]. The reason is that their learning techniques have a solid basis in statistics, allowing them to deal with the stochastic aspects of gene expressions and noisy measurements in a natural way. Other formalisms applied to this task include Boolean networks [<xref ref-type="bibr" rid="B3">3</xref>], weighted graphs [<xref ref-type="bibr" rid="B4">4</xref>], ordinary differential equations [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>] and information-theoretic approaches [<xref ref-type="bibr" rid="B8">8</xref>].</p><p>A <italic>Bayesian network </italic>(BN) is a representation of a joint probability distribution over a set of random variables. It consists of two components:</p><p>• a directed acyclic graph whose vertices correspond to random variables and edges indicate conditional dependence relations</p><p>• a family of conditional distributions for each variable, given its parents in the graph.</p><p>Together, these two components determine a unique joint distribution.</p><p>When applying Bayesian networks to genetic regulatory systems, vertices are identified with genes and their expression levels, edges indicate interactions between genes and conditional distributions describe these interactions. Given a set of gene expression data, the learning techniques for Bayesian networks allow one to infer networks that match this set well. However, as it was shown in [<xref ref-type="bibr" rid="B9">9</xref>], the problem of finding an optimal network is NP-hard. Consequently, one has to choose between restricting to small gene networks (a relatively quick exponential algorithm was given in [<xref ref-type="bibr" rid="B10">10</xref>]) and inferring suboptimal networks by heuristic search methods (see [<xref ref-type="bibr" rid="B11">11</xref>]).</p><p>It should be also pointed out that the basic BN formalism has some major limitations. First, several networks with the same undirected graph structure but different directions of some edges may represent the same distribution. Hence, relying on expression levels only, the origin and the target of an interaction become indistinguishable. Second, the acyclicity constraint rules out feedback loops, essential in genetic networks. Third, the dynamics of a gene regulatory system is not taken into account.</p><p>The above limitations may be overcome by <italic>Dynamic Bayesian networks </italic>(DBNs), which model the stochastic evolution of a set of random variables over time. In comparison with BNs, discrete time is introduced and conditional distributions are related to the values of parent variables in the previous time point. Moreover, in DBNs the acyclicity constraint is relaxed.</p><p>Given a set of time series of expression data, the learning techniques adapted from BNs allow one to infer dynamic networks that match well the temporal evolution contained in the series. The papers [<xref ref-type="bibr" rid="B12">12</xref>] and [<xref ref-type="bibr" rid="B13">13</xref>] initiated a series of biological applications of DBNs [<xref ref-type="bibr" rid="B14">14</xref>-<xref ref-type="bibr" rid="B19">19</xref>].</p><p>A special treatment is required for experiments in which expression of some genes was perturbed (e.g. knockout experiments). Since perturbations change the structure of interactions (regulation of affected genes is excluded), the learning techniques have to use data selectively.</p><p>It should be also pointed out that not every perturbation experiment may be realized in practice. The reason is that some perturbations of a regulatory mechanism may be lethal to the organism. On the other hand data from perturbation experiments are particularly rich in information regarding causal relationships.</p><p>Inferring networks from perturbed expression profiles by means of BNs was investigated in [<xref ref-type="bibr" rid="B14">14</xref>] and [<xref ref-type="bibr" rid="B20">20</xref>]. To our knowledge the DBN technique has not been applied in the context of perturbation experiments. In the present paper we extend the framework of DBNs to deal with microarray data from perturbation experiments. We propose an exact algorithm for inferring an optimal network under BDe scoring function. Moreover, we propose a method of discretization of expression levels, suitable for the data coming from time series perturbation experiments. The above methodology is applied to realistic simulations data. We perform statistical analysis, via a suitably defined p-value, which assesses the statistical significance of the inferred networks. As a way of assessing the quality of the scoring function we estimate the percentage of networks with scores better than the score of the original network. We show that the quality of inferred networks dramatically improves when using data from perturbations. We also show some advantages of our exact algorithm over heuristics like Markov chain Monte Carlo (MCMC) method.</p></sec><sec><title>Data and preprocessing</title><p>When analysing learning procedure's efficiency, the procedure should be applied to the data generated by a known network, which then might be compared with the inferred one. To this aim, most studies apply procedures to gene expression data and compare inferred interactions with those found in biological literature. The disadvantage of this approach is that our knowledge of the structures of real biological networks is far from being complete even in the most deeply investigated organisms. Although many interactions between genes are known, there are very few results stating the absence of some interactions. Thus no conclusion can be drawn from the fact that the procedure inferred unknown interaction. The above disadvantage is no longer present when data are generated from a mathematical model simulating real networks. However, a danger of this approach is that the employed model simplifies the real process, losing important biological features. In that case, analysis of a learning procedure is aimed at its ability to infer an artificial model rather than real biological networks.</p><p>Husmeier in [<xref ref-type="bibr" rid="B17">17</xref>] suggests that a satisfactory compromise between the above two extremes is to apply the learning procedure to data generated by a system of ordinary differential equations.</p><sec><title>Basic model</title><p>In the present study we generate data using the model proposed in [<xref ref-type="bibr" rid="B21">21</xref>]. The model consists of 54 species of molecules, representing 10 genes with their transcription factors, promoters, mRNAs, proteins and protein dimers, connected through 97 elementary reactions, including transcription factor binding, transcription, translation, dimerization, mRNA degradation and protein degradation. The dynamics of the model is described by the system of ordinary differential equations of the following form:</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2105-7-249-i1" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>t</mml:mi>
</mml:mrow>
</mml:mfrac>
<mml:mo>=</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>i</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:msup>
<mml:mrow>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>G</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mn>2</mml:mn>
</mml:msup>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>u</mml:mi>
<mml:mi>d</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>b</mml:mi>
<mml:mi>i</mml:mi>
<mml:mi>H</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>+</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>u</mml:mi>
<mml:mi>b</mml:mi>
<mml:mi>H</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo>⋅</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>e</mml:mi>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>t</mml:mi>
</mml:mrow>
</mml:mfrac>
<mml:mo>=</mml:mo>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>b</mml:mi>
<mml:mi>i</mml:mi>
<mml:mi>H</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>+</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>u</mml:mi>
<mml:mi>b</mml:mi>
<mml:mi>H</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo>⋅</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>m</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>t</mml:mi>
</mml:mrow>
</mml:mfrac>
<mml:mo>=</mml:mo>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>m</mml:mi>
<mml:mi>H</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>m</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>+</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mi>r</mml:mi>
<mml:mi>H</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>+</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mi>r</mml:mi>
<mml:mi>H</mml:mi>
<mml:mi>G</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:msub>
<mml:mi>G</mml:mi>
<mml:mn>2</mml:mn>
</mml:msub>
<mml:mo>⋅</mml:mo>
<mml:mi>p</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>t</mml:mi>
</mml:mrow>
</mml:mfrac>
<mml:mo>=</mml:mo>
<mml:mo>−</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>e</mml:mi>
<mml:mi>H</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo>+</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mi>l</mml:mi>
<mml:mi>H</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>m</mml:mi>
<mml:mi>H</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@F62C@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p><p>where <italic>t </italic>represents time, <italic>k</italic><sub><italic>x </italic></sub>are kinetic constants of related reactions, [<italic>X</italic>] means concentration, <italic>pX</italic>, <italic>mX</italic>, <italic>X </italic>and <italic>X</italic><sub>2 </sub>are promoter, mRNA, protein and dimer <italic>X</italic>, respectively, finally <italic>X</italic>·<italic>Y </italic>stands for a transcription factor bound to a promoter.</p><p>The system is composed of structures reported in the biological literature [<xref ref-type="bibr" rid="B22">22</xref>-<xref ref-type="bibr" rid="B24">24</xref>], i.e. a hysteretic oscillator, a genetic switch, cascades and a ligand binding mechanism that influences transcription (during the simulation, the ligand is injected for a short time). The whole network is shown in Fig. <xref ref-type="fig" rid="F1">1(a)</xref>.</p><p>The total time of each simulation is set to 5000 minutes. At time 1000 minutes the ligand is injected for 10 minutes, changing the expression levels of all genes. The influence of the injection to expression decays throughout the interval 1500–3200 minutes (depending on the gene), but at time 2400 minutes system dynamics becomes similar to the initial one.</p><p>All the equations and parameters of the model, as well as the MATLAB code to integrate it, are available in the supplementary materials to [<xref ref-type="bibr" rid="B21">21</xref>].</p><p>This model is chosen for two reasons. First, differential equations formalism and biological origin of the structure guarantee realistic simulations. Second, small size of the system (note that, according to microarray experiments data, the learning procedure will be provided with mRNA concentrations only) allows to avoid a noise arising from heuristic search methods, which are necessary when dealing with large networks. Such noise might influence comparison of methods.</p></sec><sec><title>Modified models</title><p>Since genes <italic>G </italic>and <italic>K </italic>from the model are regulated by the same gene <italic>C </italic>and have the same kinetic constants, trajectories of their concentrations are identical. Consequently, their contributions to the regulatory interactions are indistinguishable, given the expression data. For this reason, Husmeier in [<xref ref-type="bibr" rid="B17">17</xref>] tests efficiency of DBN based learning techniques using the model slightly modified by identifying both genes.</p><p>In the present study we introduce perturbations to the model. It is done by replacing the differential equation regarding the mRNA of a perturbed gene by the following equation:</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M2" name="1471-2105-7-249-i2" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>m</mml:mi>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mi>d</mml:mi>
<mml:mi>t</mml:mi>
</mml:mrow>
</mml:mfrac>
<mml:mo>=</mml:mo>
<mml:msub>
<mml:mi>k</mml:mi>
<mml:mrow>
<mml:mi>p</mml:mi>
<mml:mi>e</mml:mi>
<mml:mi>X</mml:mi>
</mml:mrow>
</mml:msub>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>c</mml:mi>
<mml:mo>−</mml:mo>
<mml:mo stretchy="false">[</mml:mo>
<mml:mi>m</mml:mi>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">]</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaadaWcaaqaaiabdsgaKjabcUfaBjabd2gaTjabdIfayjabc2faDbqaaiabdsgaKjabdsha0baacqGH9aqpcqWGRbWAdaWgaaWcbaGaemiCaaNaemyzauMaemiwaGfabeaakiabcIcaOiabdogaJjabgkHiTiabcUfaBjabd2gaTjabdIfayjabc2faDjabcMcaPaaa@4585@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p><p>which makes the concentration exponentially converging to <italic>c</italic>. Taking <italic>c </italic>= 0 yields system with gene knocked out, while setting <italic>c </italic>to maximal (with respect to the basic system) expression level of a gene makes it overexpressed. 21 simulations altogether are executed: one simulation with the basic system and 20 simulations with one gene knocked out or overexpressed.</p><p>Octave scripts for generating expression time series are available in the supplementary materials [<xref ref-type="bibr" rid="B25">25</xref>].</p></sec><sec><title>Sampling and discretization</title><p>Husmeier in [<xref ref-type="bibr" rid="B17">17</xref>] chooses for his test 12 time points in equal length intervals between 1100 and 1600 minutes. He argues that more information is contained in the data derived from the system which is driven out of equilibrium by the ligand injection. In our tests Husmeier's choice is repeated and other intervals are tried, as reported below.</p><p>Before applying our learning procedure (as well as Husmeier's), the expression levels need to be discretize. One of the simplest methods of discretizing is partition of the interval of real numbers covering mRNAs concentrations of each gene into subintervals of equal length, relevant to particular discretized values. Another strategy is to base the discretization procedure on the meanings of introduced discrete expression levels (e.g. 'on'-'off' or 'under-expressed'-'baseline'-'over-expressed').</p><p>Husmeier in [<xref ref-type="bibr" rid="B17">17</xref>] applies the former approach with 3 discretized expression levels, and we follow him in the case of unperturbed data.</p><p>The specificity of perturbed data suggests the latter approach. The simulation of an unperturbed system specifies the reference point for expression levels of perturbed data. Thus discretization consists in comparing each concentration value from a perturbed system simulation with the concentration value of the same gene at the same time point of the unperturbed system simulation. When the values are close to each other, i.e.</p><p><inline-graphic xlink:href="1471-2105-7-249-i3.gif"/></p><p>the expression level is set to 'baseline', otherwise it is set to 'over-' or 'under-expressed'. The log-ratios of concentration values in knockout simulations are shown in Fig. <xref ref-type="fig" rid="F2">2</xref>. The threshold 0.5 seems to minimize the loss of information, inevitable in the discretization process. However, this choice, as well as the choice of the number of thresholds, is arbitrary.</p><p>Discretized expression data files are available in the supplementary materials [<xref ref-type="bibr" rid="B25">25</xref>].</p></sec></sec><sec sec-type="methods"><title>Methods</title><sec><title>Dynamic Bayesian networks</title><p>A <italic>dynamic Bayesian network </italic><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M3" name="1471-2105-7-249-i4" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mi mathvariant="script">N</mml:mi><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBamrtHrhAL1wy0L2yHvtyaeHbnfgDOvwBHrxAJfwnaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaWaaeGaeaaakeaaimaacqWFneVtaaa@383B@</mml:annotation></mml:semantics></mml:math></inline-formula> is a representation of stochastic evolution of a set of random variables <bold>X </bold>= {<italic>X</italic><sub>1</sub>,..., <italic>X</italic><sub><italic>n</italic></sub>} over discretized time. Represented temporal process is assumed to be <italic>Markovian</italic>, i.e.</p><p><italic>P</italic>(<bold>X</bold>(<italic>t</italic>)|<bold>X</bold>(0), <bold>X</bold>(1),..., <bold>X</bold>(<italic>t </italic>- 1)) = <italic>P</italic>(<bold>X</bold>(<italic>t</italic>)|<bold>X</bold>(<italic>t </italic>- 1))</p><p>and <italic>time homogenous</italic>, i.e.<italic>P</italic>(<bold>X</bold>(<italic>t</italic>)|<bold>X</bold>(<italic>t </italic>- 1)) are independent of <italic>t</italic>. The representation consists of two components:</p><p>• a directed graph <italic>G </italic>= (<bold>X</bold>, <bold>E</bold>) encoding con ditional (in-)dependencies</p><p>• a family of conditional distributions <italic>P</italic>(<italic>X</italic><sub><italic>i</italic></sub>(<italic>t</italic>)|<bold>Pa</bold><sub><italic>i</italic></sub>(<italic>t </italic>- 1)), where <bold>Pa</bold><sub><italic>i </italic></sub>= {<italic>X</italic><sub><italic>j </italic></sub>∈ <bold>X</bold>|(<italic>X</italic><sub><italic>j</italic></sub>, <italic>X</italic><sub><italic>i</italic></sub>) ∈ <bold>E</bold>}</p><p>By assumption, the joint distribution over all the possible trajectories of the process decomposes into the following product form</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M4" name="1471-2105-7-249-i5" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mn>0</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>,</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>,</mml:mo>
<mml:mo>…</mml:mo>
<mml:mo>,</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>T</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>=</mml:mo>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mn>0</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>T</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>|</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo>−</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGqbaucqGGOaakieqacqWFybawcqGGOaakcqaIWaamcqGGPaqkcqGGSaalcqWFybawcqGGOaakcqaIXaqmcqGGPaqkcqGGSaalcqWIMaYscqGGSaalcqWFybawcqGGOaakcqWGubavcqGGPaqkcqGGPaqkcqGH9aqpcqWGqbaucqGGOaakcqWFybawcqGGOaakcqaIWaamcqGGPaqkcqGGPaqkdaqeWbqaaiabdcfaqjabcIcaOiab=HfayjabcIcaOiabdsha0jabcMcaPiabcYha8jab=HfayjabcIcaOiabdsha0jabgkHiTiabigdaXiabcMcaPiabcMcaPaWcbaGaemiDaqNaeyypa0JaeGymaedabaGaemivaqfaniabg+Givdaaaa@5C7E@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p><p>Consequently, the evolution of the random variables is given by</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M5" name="1471-2105-7-249-i6" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:mtable columnalign="right">
<mml:mtr columnalign="right">
<mml:mtd columnalign="right">
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>,</mml:mo>
<mml:mo>…</mml:mo>
<mml:mo>,</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>T</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>|</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mn>0</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>=</mml:mo>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>T</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>|</mml:mo>
<mml:mi>X</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo>−</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>=</mml:mo>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>T</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>i</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>n</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:msub>
<mml:mi>X</mml:mi>
<mml:mi>i</mml:mi>
</mml:msub>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>|</mml:mo>
<mml:mi>P</mml:mi>
<mml:msub>
<mml:mi>a</mml:mi>
<mml:mi>i</mml:mi>
</mml:msub>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo>−</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>=</mml:mo>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr columnalign="right">
<mml:mtd columnalign="right">
<mml:mrow>
<mml:mo>=</mml:mo>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>i</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>n</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∏</mml:mo>
<mml:mrow>
<mml:mi>t</mml:mi>
<mml:mo>=</mml:mo>
<mml:mn>1</mml:mn>
</mml:mrow>
<mml:mi>T</mml:mi>
</mml:munderover>
<mml:mrow>
<mml:mi>P</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:msub>
<mml:mi>X</mml:mi>
<mml:mi>i</mml:mi>
</mml:msub>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo>|</mml:mo>
<mml:mi>P</mml:mi>
<mml:msub>
<mml:mi>a</mml:mi>
<mml:mi>i</mml:mi>
</mml:msub>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>t</mml:mi>
<mml:mo>−</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo stretchy="false">)</mml:mo>
<mml:mo stretchy="false">)</mml:mo>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
</mml:mtd>
</mml:mtr>
</mml:mtable>
<mml:mtext>     </mml:mtext>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mn>1</mml:mn>
<mml:mo>)</mml:mo>
</mml:mrow>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@9F49@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p></sec><sec><title>Inferring networks</title><p>The problem of learning a DBN is understood as follows: find a network graph that best matches a given dataset of time series of <bold>X</bold>-instances. The notion of a good match is formalized by means of a <italic>scoring function</italic>, usually Bayesian Dirichlet equivalence (BDe) score [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B26">26</xref>], derived from the posterior probability of the network, given the data (the prior distributions over the network parameters have to be assumed). Owing to the product decomposition (1), BDe score decomposes into the sum over the set of random variables. This property is extremely useful in learning procedures, since the parents of each variable may be computed independently. When the set of variables is small enough (the boundary is approximately 20), one may score each subset as a possible parent set for each variable and choose the best match. Otherwise, heuristic search methods have to be applied and the decomposition property is helpful in reducing the computation cost when scoring locally changed networks.</p><p>Since our training datasets consist of mRNA concentrations of 10 genes only, we can apply an exact algorithm. This choice allows us to avoid the noise caused by using heuristic search methods.</p><p>Edges in the inferred network graph witness conditional dependence between variables in neighboring time points, which is interpreted as interaction between corresponding genes. However, a special care is required when inferring self-regulation. In this case it is clear that <italic>X</italic><sub><italic>i</italic></sub>(<italic>t </italic>+ 1) depends on <italic>X</italic><sub><italic>i</italic></sub>(<italic>t</italic>) because of natural inertia of mRNA production and degradation. Such dependence cannot be distinguished from actual auto-regulation by the scoring function currently used to select the best DBN model for the data. In the particular case of our experiments we have observed that with different choice of the number of time points we obtained all or none of the genes with self loops. This issue was addressed in other studies [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B15">15</xref>] by explicitly forbidding or forcing the presence of self-loops in all considered models. We take the same approach in the present paper. However it remains an open question whether the DBN scoring functions could be extended to distinguish between inertia and self-dependence.</p></sec><sec><title>Perturbations</title><p>When expression of a gene is perturbed in an experiment (e.g. by knocking it out), its natural regulation is blocked and replaced by the perturbation scheme. Consequently, regarding that gene's regulation mechanisms, the experiment contributes noise to the model instead of information. On the other hand, the remaining interactions might be significantly reflected in data, in particular those in which the gene acts as a regulator. Therefore our learning procedure has to make use of data from perturbation experiments selectively.</p><p>Recall that the parent sets of each gene may be inferred independently. Thus, when inferring parents of a particular gene, we apply the standard learning procedure to the dataset restricted to those experiments, in which this gene's expression was not perturbed. When parents of all genes are computed, the network graph is composed. A related method in the framework of static BNs was successfully used in [<xref ref-type="bibr" rid="B14">14</xref>] and [<xref ref-type="bibr" rid="B20">20</xref>]. Summarizing, our exact algorithm can be expressed as follows:</p><p>for each gene <bold>G</bold></p><p>   choose all experiments with unperturbed expression of <bold>G</bold></p><p>   for each potential parent set Pa of <bold>G</bold></p><p>      compute the local score in <bold>G</bold> for <bold>Pa</bold> and chosen experiments</p><p>   choose the parent set of <bold>G</bold> yielding optimal score compose the network from the chosen parent sets</p><p>Software for finding optimal DBNs is available in the supplementary materials [<xref ref-type="bibr" rid="B25">25</xref>].</p></sec></sec><sec><title>Results and discussion</title><p>In the present section our exact algorithm is applied to the datasets from the model modified by introducing perturbations. The results are compared with those obtained from the basic model, as well as with those obtained by Bayesian learning with Markov chain Monte Carlo (MCMC) method [<xref ref-type="bibr" rid="B17">17</xref>].</p><sec><title>Experiments</title><p>In the first experiment we followed the procedure of Husmeier [<xref ref-type="bibr" rid="B17">17</xref>] (the system restricted to 9 genes, 12 time points chosen in equal length intervals between 1100 and 1600 minutes, simple discretization into 3 levels). The sensitivity of our exact algorithm was similar to Husmeier's heuristics – see Fig. <xref ref-type="fig" rid="F3">3</xref>.</p><p>Next we turned to the knockout data. Recall that the entire system with all 10 genes was considered and the discretization was made according to the comparison of expression levels from perturbed and unperturbed profiles.</p><p>The first set of time points was chosen as in the above experiment, resulting in 7 edges corresponding to direct transcriptional regulation, 1 edge due to an interaction triggered by the ligand and 6 spurious edges (see Fig. <xref ref-type="fig" rid="F4">4(a)</xref>).</p><p>The dataset used in the experiment was quite large: 10 series, 12 time points each gives 120 slices. On the other hand, the variability of discretized expression levels is rather low – as is shown in Fig. <xref ref-type="fig" rid="F2">2</xref>, the thresholds are usually crossed at most one time per series. Therefore the steadiness of expression is represented in enormous proportion. Moreover, the sampling rate fails to match the delay of regulation processes. Since edges in DBNs represent conditional dependencies in neighboring time points, the learning process is affected. Consequently, a large number of false positives appears in the inferred network. The variability of expression is reduced by the discretization process. The choice of our discretization threshold 0.5 is aimed at minimizing this reduction (see the section <italic>Sampling and discretization</italic>). The variability may be increased by allowing more discretization levels, but it can disturb inferring. The reason is that the BN formalism disregards the structure of sets of possible values of random variables. For example, the information that the discretized expression level '0' is closer to the level '1' than to '2' is ignored. Consequently, the learning procedure treats equally the situation in which some configuration of regulators causes a regulon to assume the value '0' or '1' with the one in which it is caused to assume the value '0' or '2'. Our experiments with gene expression discretized into more than 3 levels do not improve results (data not shown).</p><p>The disproportion between a large dataset and a low variability may be avoided by decreasing a number of samples. Hence we decided to choose for the next experiment 3 time points in equal length intervals between 1100 and 1600 minutes. The accuracy significantly improved – the inferred network contains 7 edges corresponding to direct transcriptional regulation, 1 reflecting posttranscriptional regulation and 2 spurious edges (see Fig. <xref ref-type="fig" rid="F4">4(c)</xref>).</p><p>Another time intervals were tried, resulting in networks less accurate than the two above (data not shown), which confirms Husmeier's assertion of low information content of signals from a system being in equilibrium.</p><p>Corresponding experiments were also executed for the overexpression data, as well as for both kinds of perturbed data together. Accuracy of overexpression experiments does not match that of knockout ones. However, it is worth pointing out that, unlike the knockout data case, the edges <italic>A</italic>→<italic>C</italic>, <italic>B</italic>→<italic>C </italic>and <italic>E</italic>→<italic>F </italic>were inferred correctly.</p><p>The best results were obtained when both kinds of perturbations were used together. As it is shown on Fig. <xref ref-type="fig" rid="F4">4(e)</xref>, the inferred network contains 8 edges corresponding to direct transcriptional regulation, 1 edge due to an interaction triggered by the ligand, 1 reflecting posttranscriptional regulation and 3 spurious edges. The last experiment aimed at comparing our exact algorithm with heuristic methods of searching networks with optimal scores. We adapted the MCMC algorithm of Husmeier [<xref ref-type="bibr" rid="B17">17</xref>] to work with perturbations and applied it to our data. The accuracy of obtained networks was lower then the one of networks resulting from our exact algorithm – see Fig. <xref ref-type="fig" rid="F4">4</xref>. Moreover, the experiments showed two disadvantages of this method. First, when the number of genes is small (10 genes in the considered network), the MCMC algorithm is significantly slower than the exact one. Second, due to the non-deterministic character of the algorithm, the networks inferred in succeeding simulations were highly variable. For example, Husmeier's experiment on unperturbed data, repeated 100 times, resulted in the set of networks with the number of correctly inferred edges varying from 1 to 5. Moreover, the network obtained by Husmeier (see Fig. <xref ref-type="fig" rid="F3">3(b)</xref>) did not appear among them.</p></sec><sec><title>Statistical analysis</title><p>We define the <italic>p-value </italic>of a network with <italic>k </italic>true out of <italic>m </italic>inferred edges to be the probability of finding at least <italic>k </italic>true when choosing <italic>m </italic>edges at random. According to the hypergeometric distribution, the probability of <italic>n </italic>successful selections out of <italic>m </italic>from a set of <italic>N </italic>true and <italic>M </italic>- <italic>N </italic>false edges is given by</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M6" name="1471-2105-7-249-i7" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:msub>
<mml:mi>p</mml:mi>
<mml:mi>n</mml:mi>
</mml:msub>
<mml:mo>=</mml:mo>
<mml:mfrac>
<mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mi>N</mml:mi>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mi>n</mml:mi>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mi>M</mml:mi>
<mml:mo>−</mml:mo>
<mml:mi>N</mml:mi>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mi>m</mml:mi>
<mml:mo>−</mml:mo>
<mml:mi>n</mml:mi>
</mml:mrow>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
</mml:mrow>
<mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mi>M</mml:mi>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mi>m</mml:mi>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
</mml:mrow>
</mml:mfrac>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGWbaCdaWgaaWcbaGaemOBa4gabeaakiabg2da9maalaaabaWaaeWaaeaafaqabeGabaaabaGaemOta4eabaGaemOBa4gaaaGaayjkaiaawMcaamaabmaabaqbaeqabiqaaaqaaiabd2eanjabgkHiTiabd6eaobqaaiabd2gaTjabgkHiTiabd6gaUbaaaiaawIcacaGLPaaaaeaadaqadaqaauaabeqaceaaaeaacqWGnbqtaeaacqWGTbqBaaaacaGLOaGaayzkaaaaaaaa@4182@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p><p>consequently,the p-value of a network is defined by</p><p><inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M7" name="1471-2105-7-249-i8" overflow="scroll">
<mml:semantics definitionURL="" encoding="">
<mml:mrow>
<mml:mi>p</mml:mi>
<mml:mi>v</mml:mi>
<mml:mi>a</mml:mi>
<mml:mi>l</mml:mi>
<mml:mo>=</mml:mo>
<mml:mstyle displaystyle="true">
<mml:munderover>
<mml:mo>∑</mml:mo>
<mml:mrow>
<mml:mi>n</mml:mi>
<mml:mo>=</mml:mo>
<mml:mi>k</mml:mi>
</mml:mrow>
<mml:mrow>
<mml:mi>m</mml:mi>
<mml:mi>i</mml:mi>
<mml:mi>n</mml:mi>
<mml:mo stretchy="false">(</mml:mo>
<mml:mi>m</mml:mi>
<mml:mo>,</mml:mo>
<mml:mi>N</mml:mi>
<mml:mo stretchy="false">)</mml:mo>
</mml:mrow>
</mml:munderover>
<mml:mrow>
<mml:mfrac>
<mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mi>N</mml:mi>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mi>n</mml:mi>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mi>M</mml:mi>
<mml:mo>−</mml:mo>
<mml:mi>N</mml:mi>
</mml:mrow>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mrow>
<mml:mi>m</mml:mi>
<mml:mo>−</mml:mo>
<mml:mi>n</mml:mi>
</mml:mrow>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
</mml:mrow>
<mml:mrow>
<mml:mrow>
<mml:mo>(</mml:mo>
<mml:mrow>
<mml:mtable>
<mml:mtr>
<mml:mtd>
<mml:mi>M</mml:mi>
</mml:mtd>
</mml:mtr>
<mml:mtr>
<mml:mtd>
<mml:mi>m</mml:mi>
</mml:mtd>
</mml:mtr>
</mml:mtable>
</mml:mrow>
<mml:mo>)</mml:mo>
</mml:mrow>
</mml:mrow>
</mml:mfrac>
</mml:mrow>
</mml:mstyle>
</mml:mrow>
<mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGWbaCcqWG2bGDcqWGHbqycqWGSbaBcqGH9aqpdaaeWbqaamaalaaabaWaaeWaaeaafaqabeGabaaabaGaemOta4eabaGaemOBa4gaaaGaayjkaiaawMcaamaabmaabaqbaeqabiqaaaqaaiabd2eanjabgkHiTiabd6eaobqaaiabd2gaTjabgkHiTiabd6gaUbaaaiaawIcacaGLPaaaaeaadaqadaqaauaabeqaceaaaeaacqWGnbqtaeaacqWGTbqBaaaacaGLOaGaayzkaaaaaaWcbaGaemOBa4Maeyypa0Jaem4AaSgabaacbiGae8xBa0Mae8xAaKMae8NBa4MaeiikaGIaemyBa0MaeiilaWIaemOta4KaeiykaKcaniabggHiLdaaaa@5350@</mml:annotation>
</mml:semantics>
</mml:math></inline-formula></p><p>where <italic>M </italic>equals 10<sup>2 </sup>for the full considered network or 9<sup>2 </sup>for the network restricted to 9 genes (used for unperturbed data). The value of <italic>N </italic>depends on it if we allow direct transcriptional regulation only. P-values of the networks from Fig. <xref ref-type="fig" rid="F3">3</xref> and <xref ref-type="fig" rid="F4">4</xref> are grouped in the Table <xref ref-type="table" rid="T1">1</xref>.</p><p>The above considerations refer to inferring <italic>local </italic>interactions between genes, represented by particular edges. In order to analyse the ability to infer a <italic>global </italic>interaction system, one has to compare the score of the original network with the scores of other networks. Since it is impossible to compute the scores of all graphs (there are <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M8" name="1471-2105-7-249-i9" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msup><mml:mn>2</mml:mn><mml:mrow><mml:msup><mml:mi>n</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:msup></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqaIYaGmdaahaaWcbeqaaiabd6gaUnaaCaaameqabaGaeGOmaidaaaaaaaa@3050@</mml:annotation></mml:semantics></mml:math></inline-formula> directed graph structures with <italic>n </italic>nodes), we sampled randomly 100 000 graphs. For each graph, edges were generated independently, each with the same probability. The uniform distribution on the space of all graphs could be obtained by setting this probability to 1/2, but it would cause scores of most of graphs to be dominated by high penalties due to excessive structures. In order to get networks with scores close to the original one, there was chosen the probability resulting in the expected number of 12 edges in the graph (11 edges between different nodes in the cases of forbidden and forced self-connecting edges).</p><p>Original and randomly generated graphs are available in the supplementary materials [<xref ref-type="bibr" rid="B25">25</xref>].</p><p>Table <xref ref-type="table" rid="T2">2</xref> shows how many generated networks received (in various experiments) score better then the original one. We compare the results obtained with forbidden, permitted or forced self-loops, i.e. edges leading from a vertex to itself.</p><p>The tables show that using perturbed data significantly improves the possibility of inferring the original network. The results obtained in the experiments with 3 time points are usually better than those in the experiment with 12 time points, but the differences between them are relatively small.</p><p>Comparison of the values for particular versions of the algorithm shows that the best results are obtained when self-loops are forbidden, slightly worse when self-loops are permitted and significantly worse when they are forced. The analysis of the best scored networks with permitted self-loops leads to the statement that self-regulation of genes cannot be handled within our framework correctly and requires more refined methods. Therefore, with respect to our algorithm's variants, the best choice is to forbid self-loops.</p></sec></sec><sec><title>Conclusion</title><p>In the present paper the framework of dynamic Bayesian networks is extended in order to handle gene expression perturbations. A new discretization method specialized for datasets from time series gene perturbation experiments is also introduced. Networks inferred from realistic simulations data by our method are compared with those obtained by DBNs learning techniques.</p><p>The comparison shows that application of our method substantially improves quality of inference. Moreover, our results lead to the suggestion that the exact algorithm should be applied when it is possible, i.e. when the set of genes is small enough. The reason is high variability of the networks resulting from heuristics and their lower accuracy.</p><p>Since self-regulating interactions appeared to be intractable by DBN learning techniques, we also suggest to forbid self-connecting edges in inferred networks. Our experiments show that this choice makes the learning procedure more sensitive to other interactions than it would be with self-loops permitted or forced. An important problem for designing time series expression experiments is to determine sampling rates properly. Our experiments show that assuming too short rate results in noisy expression profiles, just as when the samples are chosen from the system being in equilibrium. Consequently, networks inferred from over-sampled datasets have low accuracy.</p><p>The reason of this surprising finding is the Markovian assumption of DBNs, which states that the value of an expression profile from a particular time point depends on the value of the profile from the preceding time point only. It means that the sampling rate has to match the delay of regulation processes. Most learning procedures working with time series gene expression data make similar assumptions. This is unlike those working with independent expression profiles (e.g. BNs), which assume that considered profiles represent steady states.</p></sec><sec><title>Authors' contributions</title><p>ND designed the extension of DBN framework incorporating perturbations, performed the statistical analysis and participated in executing the experiments. AG participated in the design of the study. AM implemented and tested the exact algorithm. BW implemented modifications concerning perturbations: to the system of differential equations of regulatory network [<xref ref-type="bibr" rid="B21">21</xref>] and to the MCMC learning procedure [<xref ref-type="bibr" rid="B17">17</xref>] and participated in executing the experiments. JT coordinated the study. ND, BW and JT edited the final manuscript. All authors read and approved the final manuscript.</p></sec>
|
The statistics of identifying differentially expressed genes in Expresso and TM4: a comparison
|
<sec><title>Background</title><p>Analysis of DNA microarray data takes as input spot intensity measurements from scanner software and returns differential expression of genes between two conditions, together with a statistical significance assessment. This process typically consists of two steps: data normalization and identification of differentially expressed genes through statistical analysis. The Expresso microarray experiment management system implements these steps with a two-stage, log-linear ANOVA mixed model technique, tailored to individual experimental designs. The complement of tools in TM4, on the other hand, is based on a number of preset design choices that limit its flexibility. In the TM4 microarray analysis suite, normalization, filter, and analysis methods form an analysis pipeline. TM4 computes integrated intensity values (IIV) from the average intensities and spot pixel counts returned by the scanner software as input to its normalization steps. By contrast, Expresso can use either IIV data or median intensity values (MIV). Here, we compare Expresso and TM4 analysis of two experiments and assess the results against qRT-PCR data.</p></sec><sec><title>Results</title><p>The Expresso analysis using MIV data consistently identifies more genes as differentially expressed, when compared to Expresso analysis with IIV data. The typical TM4 normalization and filtering pipeline corrects systematic intensity-specific bias on a per microarray basis. Subsequent statistical analysis with Expresso or a TM4 <italic>t</italic>-test can effectively identify differentially expressed genes. The best agreement with qRT-PCR data is obtained through the use of Expresso analysis and MIV data.</p></sec><sec><title>Conclusion</title><p>The results of this research are of practical value to biologists who analyze microarray data sets. The TM4 normalization and filtering pipeline corrects microarray-specific systematic bias and complements the normalization stage in Expresso analysis. The results of Expresso using MIV data have the best agreement with qRT-PCR results. In one experiment, MIV is a better choice than IIV as input to data normalization and statistical analysis methods, as it yields as greater number of statistically significant differentially expressed genes; TM4 does not support the choice of MIV input data. Overall, the more flexible and extensive statistical models of Expresso achieve more accurate analytical results, when judged by the yardstick of qRT-PCR data, in the context of an experimental design of modest complexity.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Sioson</surname><given-names>Allan A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Mane</surname><given-names>Shrinivasrao P</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Li</surname><given-names>Pinghua</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Sha</surname><given-names>Wei</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Heath</surname><given-names>Lenwood S</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Bohnert</surname><given-names>Hans J</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Grene</surname><given-names>Ruth</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Bioinformatics
|
<sec><title>Background</title><p>DNA microarrays are a powerful means of monitoring the expression of thousands of genes simultaneously. A variety of computational and statistical methods have been proposed to extract information from the large quantity of data generated from microarray experiments. Many methods assume, as we do here, the use of cDNA labeled with one of two fluorescent dyes to differentiate two treatments on a single microarray, implying data from two images to be analyzed. These methods include a number of data normalization techniques to reduce the effects of systematic errors and various kinds of statistical tests to identify differentially expressed genes in comparisons among different experimental conditions. There is as yet no single method that can be recommended under all circumstances for either normalization or identification of differential gene expression.</p><p>In recent years, ANOVA methods have gained popularity for identification of differential gene expression. The power of ANOVA methods derives from their flexibility in fitting and comparing different models to a given set of data [<xref ref-type="bibr" rid="B1">1</xref>]. One such method is the two-stage, log-linear ANOVA mixed models technique of Wolfinger, <italic>et al</italic>., [<xref ref-type="bibr" rid="B2">2</xref>]. Its first stage uses a normalization model designed to remove global effects across all microarrays. Its second stage uses a gene-specific model to estimate gene-treatment interactions as ratios of gene expression under control and treated conditions, along with a statistical significance. Kerr [<xref ref-type="bibr" rid="B3">3</xref>] notes that the global normalization model employed in this technique is conducive to combining data across genes for realistic and robust models of error, especially when random effects are included. Pan [<xref ref-type="bibr" rid="B4">4</xref>] compares different microarray statistical analysis methods and demonstrates that the log-linear ANOVA mixed model approach performs better than the <italic>t</italic>-test and regression approaches. The regression approach, although flexible and robust, assumes that the data is drawn from a normal distribution, while the <italic>t</italic>-test is limited due to very few degrees of freedom. Chu, <italic>et al</italic>., [<xref ref-type="bibr" rid="B5">5</xref>] compare two log-linear ANOVA mixed models for probe-level, oligonucleotide array data and found that both types of models capture key measurable sources of variability of oligonucleotide arrays for real and simulated data. Cui and Churchill [<xref ref-type="bibr" rid="B6">6</xref>] review the use of a mixed ANOVA model for analyzing a cDNA microarray experiment and conclude that such models provide a powerful way to obtain information from experiments with multiple factors or sources of variation. Rosa, <italic>et al</italic>., [<xref ref-type="bibr" rid="B7">7</xref>] review issues of analyzing cDNA microarrays with mixed linear models and puts such analysis in the larger context of Bayesian analysis procedures and adjustments for multiple testing.</p><p>Data normalization is the first step in analyzing microarray data; numerous data normalization methods have been proposed and investigated. While refinements of existing methods continue to appear (e.g., Futschik and Crompton [<xref ref-type="bibr" rid="B8">8</xref>]), naive methods, such as total intensity normalization, are still in use (e.g., Held, <italic>et al</italic>., [<xref ref-type="bibr" rid="B9">9</xref>]). Xie, <italic>et al</italic>., [<xref ref-type="bibr" rid="B10">10</xref>] did a comparative study of normalization methods and test statistics to analyze the results of a DNA-protein binding microarray experiment. Using performance and bias correction criteria, Bolstad, <italic>et al</italic>., [<xref ref-type="bibr" rid="B11">11</xref>] evaluate the cyclic lowess method, the contrast method, the quantile method, and baseline array scaling methods, both linear and non-linear; they demonstrate that normalization methods incorporating data from all microarrays perform better than methods employing a baseline array.</p><p>Several software tools that combine data normalization and statistical analysis are currently available. Dudoit, <italic>et al</italic>., [<xref ref-type="bibr" rid="B12">12</xref>] review these software tools with an emphasis on the TM4 microarray software suite, Bioconductor in R, and the BioArray Software Environment (BASE) system. Saeed, et al., [<xref ref-type="bibr" rid="B13">13</xref>] describe the features and capabilities of TM4, while Quackenbush [<xref ref-type="bibr" rid="B14">14</xref>] describes the normalization and transformation methods implemented in it. Williams, et al., [<xref ref-type="bibr" rid="B15">15</xref>], Zhu, et al., [<xref ref-type="bibr" rid="B16">16</xref>], and Khaitovich, et al., [<xref ref-type="bibr" rid="B17">17</xref>] have used TM4 in microarray data analysis. Another system is Expresso, an experiment management system that serves as a unifying framework to study data driven applications such as microarray experiments [<xref ref-type="bibr" rid="B18">18</xref>-<xref ref-type="bibr" rid="B20">20</xref>]. Expresso has adapted the two-stage ANOVA mixed models technique of Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>] to the particular needs of individual microarray data sets. Our experience with numerous such data sets has demonstrated that modeling the underlying experiment carefully and completely is essential to obtaining meaningful and defensible results. Use of tools that require experiments to conform to their analysis methods are less than satisfactory.</p><p>In this paper, we compare the Expresso analysis methodology to the approach provided in the TM4 microarray analysis software suite [<xref ref-type="bibr" rid="B13">13</xref>]. Each is invoked to identify differentially expressed genes in two experimental data sets, each of which uses an <italic>Arabidopsis thaliana </italic>oligonucleotide array. Along the way, we demonstrate differences between the use of integrated intensity values (IIV) and median intensity values (MIV) as inputs. We report interactions between normalization and gene identification methods. We use quantitative reverse-transcriptase PCR (qRT-PCR) results to assess the consistency of genes reported by TM4 and Expresso as having significant differential expression.</p></sec><sec><title>Results and discussion</title><p>Here, we report a portion of the results obtained in our comparison of Expresso analysis and the TM4 pipeline (see Materials and Methods). Figure <xref ref-type="fig" rid="F1">1</xref> illustrates the overall flow of the statistical analyses of microarray data that were done in this study. We began with microarray data in GPR format from Experiment 1 and Experiment 2. Median intensity values (MIV) from the GPR files can be analyzed by the Expresso GP and GOT models directly. ExpressConverter provides integrated intensity values (IIV) for further Expresso and TM4 analysis. The MIDAS normalization and filtering pipeline executes these steps in order: total intensity normalization (subscript T), lowess normalization (subscript L), standard deviation regularization (subscript S), and low intensity filter (subscript F). MIDAS allows tapping the output of any step in the pipeline; for example, IIV<sub>TL </sub>signifies an MEV file after total intensity normalization followed by lowess normalization. The identification of genes with significant differential expression was performed on all GPR and MEV files, using the Expresso GP and GOT models and the <italic>t</italic>-test in MEV.</p><sec><title>Normalization and low intensity filtering in TM4</title><p>Quackenbush [<xref ref-type="bibr" rid="B14">14</xref>] describes the use of ratio-intensity plots (RI-plots) to detect and normalize for any systematic intensity-dependent dye bias using lowess normalization (see Materials and Methods). We evaluated the effect of lowess normalization within the context of the flow in Figure <xref ref-type="fig" rid="F1">1</xref> by creating RI-plots after each step for the second replicate microarray in Experiment 1, WT plant. Supplementary Figure 1 - see <xref ref-type="supplementary-material" rid="S1">Additional file: 1</xref> contains these RI-plots. The IIV<sub>TL </sub>is indeed effective, for this data set, in correcting systematic dye bias, suggesting that preprocessing by these two normalization steps in MIDAS may be a good practice in many situations.</p><p>The normalization and filtering pipeline affects the number of genes identified as differentially expressed in both the GP and GOT models. See Table <xref ref-type="table" rid="T1">1</xref>. For example, in Experiment 1, the GP model using IIV input data identifies 567 up-expressed genes in the WT microarrays, while it identifies only 460 WT genes as up-expressed if IIV<sub>TLSF </sub>(processed by the complete MIDAS pipeline) input data is used.</p><p>Small changes in the number of genes identified as up- or down-expressed after successive MIDAS steps may mask larger changes in the composition of sets of up- and down-expressed genes. To obtain a more precise view of the effects of MIDAS changes, we computed retention counts (RC) and retention percentages (RP) between the gene successive sets whose numbers are in Table <xref ref-type="table" rid="T1">1</xref>. RC is the number of genes in the set before the MIDAS step that remain in the set after the step. RP is the percentage of remaining genes with respect to the number of genes in the set after the MIDAS step. Table <xref ref-type="table" rid="T2">2</xref> contains the RC and RP values corresponding to the counts in Table <xref ref-type="table" rid="T1">1</xref>. For Experiment 1, there is a tremendous drop in retention during the lowess normalization that follows the total intensity normalization. There is not a drop of corresponding magnitude for Experiment 2. For both experiments, normalization has a significant effect on the sets of genes identified as differentially expressed.</p><p>In Experiment 1 results, the number of genes commonly assessed by Expresso as significantly expressed when using IIV and IIV<sub>T </sub>is high. For example, there is 95.45% retention of WT genes (545 total) assessed as up-expressed when using IIV<sub>T </sub>data in Expresso compared to that when using IIV data. Retention percentage of these genes assessed as expressed however went down after doing lowess normalization. There is only 15.20% retention of WT genes (74 total) assessed as up-expressed in the results when using IIV<sub>TS </sub>data in Expresso compared to when using IIV<sub>T</sub>. While we observe increase in the retention percentage in IIV<sub>TLS </sub>(from IIV<sub>TL</sub>) and IIV<sub>TLSF </sub>(from IIV<sub>TLS</sub>), there's low retention percentage in the results using IIV<sub>TLSF </sub>from IIV data. This can be traced in the low retention percentage of IIV<sub>TL </sub>from IIV<sub>T</sub>. Hence, the normalization method that affects the results in Experiment 1 the most is lowess normalization.</p><p>The results of Expresso on Experiment 2 show that low retention percentages happen after appli-cation of total intensity normalization (lowest is 59.30%) and after application of low intensity filtering (lowest is 61.19%). The low retention percentages shown in the IIV ∩ IIV<sub>TLSF </sub>column implies that the normalization pipeline also significantly affects the results in Expresso analysis of Experiment 2.</p></sec><sec><title>Choice of intensity signal data</title><p>The input to statistical analysis of microarray experiments is a set of real numbers that represent the measured intensity signal for each spot in a microarray. Much statistical analysis of microarray data has traditionally used median intensity values (MIV). The alternative used in TM4 is the integrated intensity value (IIV). (See Materials and Methods.) Since IIV is intended to integrate the measured intensity across the biological sample printed at a spot, one might expect IIV to be a more accurate assessment of the biological measurement than MIV data. For example, a spot having 100 pixels and a median intensity of 5,000 has the same IIV as a spot having 50 pixels and a median intensity of 10,000.</p><p>This study provides the opportunity to observe the difference that choosing MIV or IIV makes on the sets of genes ultimately identified as differentially expressed. We used the GP model to analyze unnormalized MIV and IIV data from Experiment 1, and we used the GOT model to analyze unnormalized MIV and IIV data from Experiment 2. Table <xref ref-type="table" rid="T3">3</xref> reports a summary of the results. In Experiment 1, 725 WT genes are assessed as up-expressed and 774 WT genes are down-expressed when MIV data are used in Expresso. These numbers decreased to 567 up-expressed genes and 552 down-expressed genes when IIV data are used instead. A similar trend is observed in Experiment 2 results when using MIV and IIV data. These results suggest that employing IIV input data with Expresso analysis leads to more conservative results than employing MIV input data.</p></sec><sec><title>Comparison of statistical methods</title><p>We compared the performance of the GP model, the GOT model, and the <italic>t</italic>-test of MEV in identifying differentially expressed genes in Experiment 2. We used the IIV<sub>TLSF </sub>data of Experiment 2 as input to these methods. We also contrast these results with MIV data analyzed in GP. Table <xref ref-type="table" rid="T4">4</xref> reports counts for these analyses.</p><p>The plot in Figure <xref ref-type="fig" rid="F2">2a</xref> demonstrates that the estimates of Iog<sub>2</sub>(fold change) are the same in GP and GOT. As Figure <xref ref-type="fig" rid="F2">2b</xref> shows, the <italic>p </italic>values by GOT are smaller compared to the <italic>p </italic>values calculated by GP. Use of the MEV <italic>t</italic>-test resulted in fewer genes assessed as significantly expressed when compared to the numbers for the GP model. The results obtained when MIV data was used as input to GP, is closest to the results when using IIV<sub>TLSF</sub>.</p><p>To compare the effectiveness of Expresso and TM4 in identifying gene differential expression, we compared the identified direction of differential expression of a select set of genes per genotype in Experiment 2 with results obtained by qRT-PCR. See Table <xref ref-type="table" rid="T5">5</xref>. The lowest overall percentage (71.9%) of agreement is between the qRT-PCR results and the MEV <italic>t</italic>-test results using IIV<sub>TLSF</sub>. The log(fold change) estimates of the GP model has 77.1% percentage agreement with the qRT-PCR results, which is slightly higher than the percentage for the MEV <italic>t</italic>-test. The results of the GP model using MIV data demonstrated the greatest agreement, 90.1%, with the qRT-PCR results.</p><p>Figures <xref ref-type="fig" rid="F3">3</xref>, <xref ref-type="fig" rid="F4">4</xref>, and <xref ref-type="fig" rid="F5">5</xref> present the actual assessed Iog<sub>2</sub>(fold change) values for 50 selected Col-0 genes in Experiment 2, along with their qRT-PCR values. These are the 50 Col-0 genes, among the 55 with qRT-PCR values, for which we have expression values for all methods. For each gene, a histogram of the Iog<sub>2</sub>(fold change) estimates of qRT-PCR, the GP model using MIV, the GP model using IIV<sub>TLSF</sub>, and the MEV <italic>t</italic>-test is given. The 50 histograms are spread over three figures to enhance readability and are in increasing order by qRT-PCR estimated change. In general, the log<sub>2</sub>(fold change) estimates of the GP model and of the MEV <italic>t</italic>-test, all IIV<sub>TLSF </sub>input data, are approximately the same, while being slightly different from estimates of the GP model using MIV input data. As might be expected, disagreement between qRT-PCR and microarray results are more prevalent for small estimated log<sub>2</sub>(fold change) values. The histograms for genes AT4G09020 (Figure <xref ref-type="fig" rid="F3">3</xref>), AT1G35580 (Figure <xref ref-type="fig" rid="F4">4</xref>), and AT3G29360 (Figure <xref ref-type="fig" rid="F4">4</xref>) show that the direction of log(fold change) estimate of qRT-PCR matches the direction of the GP model using MIV input data, while differing from the direction of the estimates of the GP model and the MEV <italic>t</italic>-test using IIV<sub>TLSF </sub>input data.</p><p>Table <xref ref-type="table" rid="T6">6</xref> summarizes genotype-specific correlation results, which demonstrate that the GP model us-ing MIV input data has the highest correlation with qRT-PCR compared to the GP model and the MEV <italic>t</italic>-test using IIV<sub>TLSF </sub>input data. The highest correlation of 0.85 is for the Col-0 results of qRT-PCR versus the GP model using MIV input. The corresponding correlations for Cvi-0, WS, and Th are 0.73, 0.66, and 0.84, respectively, which are all best among the analysis methods.</p></sec></sec><sec><title>Conclusion</title><p>Our integration and comparison of Expresso analysis and the capabilities of TM4 has highlighted successes in microarray analysis, some similarities, and some differences. The success of microarray analysis is demonstrated by considerable agreement between qRT-PCR results and the results of all the examined microarray analysis methods. The greatest agreement was found when median intensity value (MIV) inputs were analyzed with the Expresso GP analysis model. We also found that the use of integrated intensity value (IIV) inputs for Expresso analysis consistently resulted in fewer genes identified as differentially expressed when compared to results from MIV inputs. This suggests that the use of IIV inputs is more conservative than the use of MIV inputs, while MIV inputs may give greater agreement to qRT-PCR results than IIV inputs.</p><p>Our results demonstrate that the MIDAS normalization and filtering pipeline corrects systematic intensity-dependent dye bias on a per microarray basis. The normalization stage in Expresso analysis removes global effects across all microarrays and complements the per microarray normalization methods of MIDAS. The generally better agreement of Expresso analysis with qRT-PCR results when compared to the MEV <italic>t</italic>-test suggests that it would be desirable for MEV to have an ANOVA test that has the greater flexibility of the Expresso gene model.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Median and integrated intensity values</title><p>This research considers two ways of measuring spot intensity, one or both of which are reported by typical microarray image processing software. The median intensity value (MIV) of a spot is the median value of all the pixels identified as part of the spot. The integrated intensity value (IIV) of a spot is the total value of all the pixels identified as part of the spot. In this research, both are background-corrected values. If the IIV data is unavailable, but the radius of the bounding circle of the spot and its average intensity value are available, then the IIV data can be estimated as the product of the average intensity and the number of pixels in the circle. This is the estimate used by ExpressConverter (below) when it converts GPR format data to MEV format data.</p></sec><sec><title>Microarray data sets</title><p>We used data sets from two experiments that used the Arabidopsis Oligonucleotide Microarrays [<xref ref-type="bibr" rid="B21">21</xref>], which include 25,712 elements, each a gene-specific 70-mer (Qiagen/Operon, Valencia, CA) for a known or putative open reading frames in <italic>Arabidopsis thaliana</italic>. There are 48 blocks per microarray, 25 rows by 24 columns (600 spots) per block, and 28,800 spots per microarray, including spots for the 25,712 gene-specific 70-mers and 302 control elements. The remaining 2,786 spots are blank.</p><sec><title>Experiment 1</title><p>Experiment 1 compares the responses of <italic>Arabidopsis thaliana </italic>wild type, ecotype Columbia (henceforth, WT), and of an antisense plant for phospholipase D α (antiPLD) in Columbia background to drought stress [<xref ref-type="bibr" rid="B22">22</xref>]. Plants were harvested at a single time point, and two biological replicate hybridizations were done for each of WT and antiPLD. Scan Array Express (PerkinElmer Life and Analytical Sciences, Inc., Boston, MA USA) was used to quantitate the four microarrays. By default, ScanArray Express performs a global lowess normalization of median intensities per microarray. ScanArray Express output its results to four files in GenePix (GPR) format, which constitute the Experiment 1 data set.</p></sec><sec><title>Experiment 2</title><p>Li, et al., [<xref ref-type="bibr" rid="B23">23</xref>] compare the responses to elevated CO<sub>2 </sub>of a wild <italic>Arabidopsis thaliana </italic>relative (<italic>Thellungiella halophila</italic>, ecotype Shandong; Th) and of three <italic>Arabidopsis thaliana </italic>ecotypes: Wassilewskija (WS), Columbia (Col-0), and Cape Verde Islands (Cvi-0). Three biological replicate hybridizations were done for each genotype. GenePix (Axon Instruments, Union City, CA USA) was used to quantitate the twelve microarrays. GenePix also performs by default a global lowess normalization of median intensities per microarray. The output of GenePix is twelve GPR files, which constitute the Experiment 2 data set.</p></sec></sec><sec><title>Real-time quantitative RT-PCR</title><p>For verification of microarray results in Experiment 2, Li, et al., [<xref ref-type="bibr" rid="B23">23</xref>] performed real-time quantitative reverse-transcriptase PCR (qRT-PCR) for selected genes — 55 in Col-0; 52 in Cvi-0; 59 in WS; 26 in Th. Supplementary Table 1 - see <xref ref-type="supplementary-material" rid="S2">Additional file: 2</xref> contains the annotation of the selected genes. In brief, primer pairs were selected to represent unique sequences in the <italic>Arabidopsis thaliana </italic>genome and in the <italic>Thellungiella </italic>sequences deposited in NCBI. <italic>Thellungiella </italic>actin (CX129618) cDNA primers and <italic>Arabidopsis thaliana</italic>. Ubiquitin-10 cDNA primers were used as internal controls in the qRT-PCR analyses. RT-PCR products were detected using the fluorescent dye SYBR-green (Applied Biosystems, Foster City, CA USA) and the ABI PRISM/Taqman 7900 Sequence Detection System (Applied Biosystems, Foster City, CA USA). Dissociation curves were generated for each reaction to ensure specific amplification. Three repeats were done for each gene. The averaged threshold cycle numbers were used to estimate original mRNA levels.</p><p>The microarray results of Experiment 2 suggested that exposure to elevated CO2 resulted in changes in expression of many genes associated with carbon metabolism, and those associated with pho-tosynthetic carbon metabolism in particular. This included genes encoding proteins that transport pho-tosynthate out of the chloroplast, where carbon fixation takes place, for export to other parts of the cell, and also genes encoding transport proteins that export carbon skeletons out of the cell to other tissues where growth is taking place. Because of this finding, it was important to validate the results obtained for gene expression associated with carbon metabolism with qRT-PCR. Hence, a number of the genes in Supplementary Table 1 - see <xref ref-type="supplementary-material" rid="S2">Additional file: 2</xref> are related to carbon metabolism.</p></sec><sec><title>Expresso analysis</title><p>Expresso analysis employs a general and flexible method to identify differentially expressed genes that is adapted from the two-stage analysis method of Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>]. In general, Expresso analysis consists of two log-linear ANOVA mixed models, called the normalization model and gene model. The first estimates and removes the experiment-wise systematic errors, while the second estimates and removes the gene specific errors. The residual that remains is the log-ratio estimate for each gene. In particular, the Tukey-Kramer multiple comparison of treatment effects on each gene is performed to estimate its expression level and the significance of (confidence in) that expression level. Expresso analysis is implemented for and executed on SAS (SAS/STAT version 8.2, SAS Institute Inc., Gary, NC USA).</p><p>The original model of Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>] includes the treatment and the array as the main effects. In previous Expresso analysis, we have extended that model to experiment-appropriate models that include additional fixed and random effects. Here, the design of the two-dye oligonucleotide microarray used in Experiment 1 and Experiment 2 includes various controls strategically positioned in different blocks of the microarray. This makes it possible to estimate the random block effect in each microarray. Furthermore, the dye effect is included in the normalization model to estimate and remove the global dye bias.</p><p>For this research, we developed two Expresso models, one whose gene model assesses the gene-(plant sample) effect (the GP model) and the other whose gene model assesses the gene-genotype-treatment effect (the GOT model). The GP model is much like previous Expresso models and is applicable to both Experiment 1 and Experiment 2. However, the GOT model is specific to analyzing Experiment 2. In both experiments, we used the GP model to estimate the differences in response of individual genotypes to treatment (drought stress versus control or ozone stress versus ambient ozone). However, the GOT model was used to estimate the effects of treatment (ozone stress), aside from the effect of individual genotypes.</p><sec><title>Expresso GP model</title><p>The normalization model is</p><p><italic>y</italic><sub><italic>spdab </italic></sub>= <italic>μ </italic>+ <italic>P</italic><sub><italic>p </italic></sub>+ <italic>D</italic><sub><italic>d </italic></sub>+ <italic>A</italic><sub><italic>a </italic></sub>+ (<italic>P </italic>× <italic>A</italic>)<sub><italic>pa </italic></sub>+ <italic>B</italic><sub><italic>ba </italic></sub>+ <italic>r</italic><sub><italic>spdab</italic></sub>.</p><p>Each <italic>y</italic><sub><italic>spdab </italic></sub>value is the log<sub>2</sub>-transformed intensity of spot <italic>s </italic>within the dye <italic>d </italic>image in block <italic>b </italic>of array <italic>a</italic>. (A spot may represent a gene, a control, or a blank.) The global mean of the <italic>y</italic><sub><italic>spdab </italic></sub>values, over all microarrays, is <italic>μ</italic>. The fixed effects in the model are the plant sample effect <italic>P</italic><sub><italic>p</italic></sub>, where <italic>p </italic>indexes the various distinct plant samples from which mRNA was obtained, and the dye effect <italic>D</italic><sub><italic>d</italic></sub>, where <italic>d </italic>has two values for the two dyes. The random effects in the model are the array effect <italic>A</italic><sub><italic>a</italic></sub>, where <italic>a </italic>indexes the microarrays, the interaction effect (<italic>P </italic>× <italic>A</italic>)<sub><italic>pa </italic></sub>of plant sample <italic>p </italic>with microarray <italic>a</italic>, and the block effect <italic>B</italic><sub><italic>ba</italic></sub>, where <italic>b </italic>identifies the block within microarray <italic>a</italic>. The model residual is <italic>r</italic><sub><italic>spdab</italic></sub>. This differs from the normalization model in Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>], in that it incorporates dye and block effects. It is a refinement of the Expresso normalization model in Watkinson, et al., [<xref ref-type="bibr" rid="B18">18</xref>], in that it has no printing pin effect, which is specific to the analysis in the earlier paper, and includes the block effect.</p><p>The second stage of the analysis uses the residual values <italic>r</italic><sub><italic>spdab </italic></sub>computed in the first stage to estimate the interaction between an individual gene <italic>g </italic>and each plant sample <italic>p </italic>at a significance level ≤ α = 0.05. Index <italic>g </italic>is added to the residual values <italic>r</italic><sub><italic>spdab </italic></sub>resulting to <italic>r</italic><sub><italic>gspdab</italic></sub>. The value of <italic>g </italic>is determined using the mapping of <italic>s </italic>index values to <italic>g </italic>index values. The gene model is</p><p><italic>r</italic><sub><italic>gspdab </italic></sub>= <italic>G</italic><sub><italic>g </italic></sub>+ (<italic>G </italic>× <italic>P</italic>)<sub><italic>gp </italic></sub>+ (<italic>G </italic>× <italic>D</italic>)<sub><italic>gd </italic></sub>+ (<italic>G </italic>× <italic>A</italic>)<sub><italic>ga </italic></sub>+ λ<sub><italic>gspdab</italic></sub>.</p><p>Here, <italic>g </italic>is a spot that represents a gene (not a blank or control) within the dye <italic>d </italic>image in block <italic>b </italic>of array <italic>a</italic>. The value <italic>G</italic><sub><italic>g </italic></sub>is the mean of residual values for all spots that represent gene <italic>g </italic>in all images. The interactions (<italic>G </italic>× <italic>P</italic>)<sub><italic>gp </italic></sub>of gene <italic>g </italic>with plant sample <italic>p </italic>and of (<italic>G </italic>× <italic>D</italic>)<sub><italic>gd </italic></sub>of gene <italic>g </italic>with dye <italic>d </italic>are the fixed effects. The interaction (<italic>G </italic>× <italic>A</italic>)<sub><italic>ga </italic></sub>of gene <italic>g </italic>with mi-croarray a is a random effect. The λ<sub><italic>gspdab </italic></sub>values are stochastic errors. This differs from the gene model in Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>], in that it incorporates interactions between gene and dye and between gene and array. It refines the Expresso gene model in Watkinson, et al., [<xref ref-type="bibr" rid="B18">18</xref>] to include the interaction between gene and array.</p><p>The estimate of the expression level of each gene in each treatment comparison is done by computing the pair-wise least square mean differences of gene-treatment effects. The Tukey-Kramer multiple comparison of gene-(plant sample) effects on each gene is made to estimate the <italic>p </italic>values associated with each calculated expression level. If there are ρ plant samples, then there are <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1" name="1471-2105-7-215-i1" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mrow><mml:mtable><mml:mtr><mml:mtd><mml:mi>ρ</mml:mi></mml:mtd></mml:mtr><mml:mtr><mml:mtd><mml:mn>2</mml:mn></mml:mtd></mml:mtr></mml:mtable></mml:mrow><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaadaqadaqaauaabeqaceaaaeaaiiGacqWFbpGCaeaacqaIYaGmaaaacaGLOaGaayzkaaaaaa@30FB@</mml:annotation></mml:semantics></mml:math></inline-formula> possible pairwise comparisons. If we index the plant samples from 1 to ρ, then the null hypothesis for gene <italic>g </italic>and comparison <italic>i</italic>, <italic>j</italic>, where 1 ≤ <italic>i </italic><<italic>j </italic>≤ ρ, is</p><p><italic>H</italic><sub>o</sub>: (<italic>G </italic>× <italic>P</italic>)<sub><italic>gi </italic></sub>= (<italic>G </italic>× <italic>P</italic>)<sub><italic>gi</italic></sub>.</p><p>The difference (<italic>G </italic>× <italic>P</italic>)<sub><italic>gi </italic></sub>- (<italic>G </italic>× <italic>P</italic>)<sub><italic>gj </italic></sub>is the estimate of the log<sub>2</sub>(fold change) of gene <italic>g </italic>in the experimental comparison <italic>P</italic><sub><italic>i </italic></sub>versus <italic>P</italic><sub><italic>j</italic></sub>. The analysis also yields a <italic>p</italic>-value for the statistical confidence in each difference.</p><p>The above GP model was used to analyze both Experiment 1 and Experiment 2. In both experiments, there are 48 blocks per array. In Experiment 1, there are four arrays and four plant samples, namely, WT-control, WT-stressed, antiPLD-control, and antiPLD-stressed. In Experiment 2, there are 12 arrays and eight plant samples, namely, Col-0-test, Col-0-control, Cvi-0-test, Cvi-0-control, WS-test, WS-control, Th-test, and Th-control.</p></sec><sec><title>Expresso GOT model</title><p>We wanted to estimate the gene-treatment effects separately from gene-genotype-treatment interaction effects using just one model. To do this, we designed the gene-genotype-treatment model, an alternative set of log-linear ANOVA mixed models, for the elevated CO<sub>2 </sub>experiment where we unfold the genotype information from the plant sample factor in the GP model. This resulted in a normalization model that includes the genotype effect (<italic>O</italic><sub><italic>o</italic></sub>) with 4 levels (Col-0, Cvi-0, WS, and Th) and basic treatment effect (<italic>T</italic><sub><italic>t</italic></sub>) with 2 levels (test and control). The random array (<italic>A</italic><sub><italic>a</italic></sub>) effect however needs to be removed from the model since it confounds the genotype effect.</p><p>The normalization model is</p><p><italic>y</italic><sub><italic>sotdab </italic></sub>= <italic>μ </italic>+ <italic>O</italic><sub><italic>o </italic></sub>+ <italic>T</italic><sub>T </sub>+ <italic>D</italic><sub><italic>d </italic></sub>+ (<italic>O </italic>× <italic>T</italic>)<sub><italic>ot </italic></sub>+ <italic>B</italic><sub><italic>ba </italic></sub>+ <italic>r</italic><sub><italic>sotdab</italic></sub>.</p><p>Each <italic>y</italic><sub><italic>sotdab </italic></sub>value is the log<sub>2</sub>-transformed intensity of spot <italic>s </italic>for genotype <italic>o </italic>and treatment <italic>t </italic>within the dye <italic>d </italic>image in block <italic>b </italic>of array <italic>a</italic>. We have that <italic>μ </italic>is as in the GP model. The fixed effects in the model are the genotype effect <italic>O</italic><sub><italic>o</italic></sub>, where <italic>o </italic>indexes the genotype (organism), the treatment effect <italic>T</italic><sub><italic>t</italic></sub>, where <italic>t </italic>is the treatment, and the dye effect <italic>D</italic><sub><italic>d</italic></sub>, where <italic>d </italic>has two values for the two dyes. The random effects in the model are the interaction effect (<italic>O </italic>× <italic>T</italic>)<sub><italic>ot </italic></sub>of genotype <italic>o </italic>with microarray <italic>a</italic>, and the block effect <italic>B</italic><sub><italic>ba</italic></sub>, where <italic>b </italic>identifies the block within microarray <italic>a</italic>. The model residual is <italic>r</italic><sub><italic>sotdab</italic></sub>. This differs from the normalization model in Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>], in that it incorporates genotype (organism), dye, and block effects. It is a refinement of the Expresso normalization model in Watkinson, et al., [<xref ref-type="bibr" rid="B18">18</xref>], in that it has no printing pin effect, and includes the genotype and block effects.</p><p>The second stage of the analysis uses the residual values <italic>r</italic><sub><italic>sotdab </italic></sub>computed in the first stage to estimate the interaction among an individual gene <italic>g</italic>, each genotype <italic>o</italic>, and each treatment <italic>t</italic>, at a significance level ≤ α = 0.05. Index <italic>g </italic>is added to the residual values <italic>r</italic><sub><italic>sotdab </italic></sub>resulting to <italic>r</italic><sub><italic>gsotdab</italic></sub>. The value of <italic>g </italic>is determined using the mapping of <italic>s </italic>index values to <italic>g </italic>index values. The gene model is</p><p><italic>r</italic><sub><italic>gsotdab </italic></sub>= <italic>G</italic><sub><italic>g </italic></sub>+ (<italic>G </italic>× <italic>O</italic>)<sub><italic>go </italic></sub>+ (<italic>G </italic>× <italic>T</italic>)<sub><italic>gt </italic></sub>+ (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>got </italic></sub>+ (<italic>G </italic>× <italic>D</italic>)<sub><italic>gd </italic></sub>+ λ<sub><italic>gsotdab</italic></sub></p><p>Here, <italic>G</italic><sub><italic>g </italic></sub>are as for the GP model. The interactions (<italic>G </italic>× <italic>O</italic>)<sub><italic>go </italic></sub>of gene <italic>g </italic>with genotype <italic>o</italic>, (<italic>G </italic>× <italic>T</italic>)<sub><italic>gt </italic></sub>of gene <italic>g </italic>with treatment <italic>t</italic>, (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>got </italic></sub>of gene <italic>g </italic>with genotype <italic>o </italic>and treatment <italic>t</italic>, and (<italic>G </italic>× <italic>D</italic>)<sub><italic>gd </italic></sub>of gene <italic>g </italic>with dye <italic>d </italic>are fixed effects. The λ<sub><italic>gsotdab </italic></sub>values are stochastic errors. This differs from the gene model in Wolfinger, et al., [<xref ref-type="bibr" rid="B2">2</xref>], in that it incorporates interactions between gene and genotype, between gene and genotype with treatment, and between gene and array. It refines the Expresso gene model in Watkinson, et al., [<xref ref-type="bibr" rid="B18">18</xref>] to include interactions between gene and genotype and between gene and genotype with treatment.</p><p>We do pairwise comparisons as in the GP model, but there are two plausible classes of null hypotheses to test within the GOT gene model. If <italic>τ </italic>is the number of treatments, then the class 1 null hypothesis for gene <italic>g </italic>and comparison <italic>i</italic>,<italic>j</italic>, where 1 ≤ <italic>i </italic><<italic>j </italic>≤ <italic>τ</italic>, is</p><p><italic>H</italic><sub>1</sub>: (<italic>G </italic>× <italic>T</italic>)<sub><italic>gi </italic></sub>= (<italic>G </italic>× <italic>T</italic>)<sub><italic>gj</italic></sub>.</p><p>The difference (<italic>G </italic>× <italic>T</italic>)<sub><italic>gi </italic></sub>- (<italic>G </italic>× <italic>T</italic>)<sub><italic>gj </italic></sub>is the estimate of the log<sub>2</sub>(fold change) of gene <italic>g </italic>in the <italic>T</italic><sub><italic>i </italic></sub>versus <italic>T</italic><sub><italic>j</italic></sub>comparison. This particular comparison looks for gene-treatment effects that are independent of genotype.</p><p>We can still estimate the expression level of each gene with respect to a specific genotype level by computing the pair-wise least square differences of gene-genotype-treatment interaction effects. The class 1 null hypothesis for gene <italic>g </italic>and comparison <italic>i</italic>, <italic>j</italic>, where 1 ≤ <italic>i </italic><<italic>j </italic>≤ <italic>τ</italic>, is</p><p><italic>H</italic><sub>2</sub>: (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>goi </italic></sub>= (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>goj</italic></sub>.</p><p>The difference (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>goi </italic></sub>- (<italic>G </italic>× <italic>O </italic>× <italic>T</italic>)<sub><italic>goj </italic></sub>is the estimate of the log<sub>2 </sub>(fold change) of gene <italic>g </italic>of a specific genotype <italic>o </italic>for the <italic>T</italic><sub><italic>i </italic></sub>versus <italic>T</italic><sub><italic>j </italic></sub>comparison. The analysis also yields a <italic>p</italic>-value for the statistical confidence in each difference.</p><p>The GOT model applies to Experiment 2 in a straightforward way. There are 12 arrays, two treatments, and four genotypes, namely, Col-0, Cvi-0, WS, and Th.</p></sec></sec><sec><title>TM4 microarray software suite</title><p>The TM4 microarray software suite consists of several components freely available from the TM4 web site [<xref ref-type="bibr" rid="B24">24</xref>]. In this research, we employed these components: ExpressConverter, Microarray Data Analysis Software (MIDAS), and Microarray Experiment Viewer (MEV). ExpressConverter converts microarray data from various data formats, such as the GenePix Results (GPR) format, to the MEV format, which is used by MIDAS and MEV. MEV format includes only integrated intensity values (IIV), which is the kind of intensity values expected of all TM4 components.</p></sec><sec><title>MIDAS data normalization methods and filters</title><p>Low intensity and saturated spots are marked by quantitation programs. These spots are filtered out from the data before doing any further normalization or statistical analysis. Data normalization methods proceed from the assumption that only a relatively small proportion of the genes change significantly in expression level between the two hybridized mRNA samples. This assumption is reasonable for these data sets since the hybridizations and subsequent analysis address nearly all <italic>Arabidopsis thaliana </italic>genes. The MIDAS component of TM4 provides a number of data normalization methods and filters and supports applying them in a pipelined fashion [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>].</p><sec><title>Total intensity normalization</title><p>While our assumption implies that the average measured intensities of the two channels of a cDNA or oligonucleotide microarray should be almost the same, these averages are often significantly different, due to differences in the inherent fluorescence of the two dyes. The total intensity normalization step in TM4 is a straightforward means to eliminate this global dye bias. For each spot <italic>i</italic>, where 1 ≤ <italic>i </italic>≤ <italic>n</italic>, let <italic>R</italic><sub><italic>i </italic></sub>and <italic>G</italic><sub><italic>i </italic></sub>be the measured intensities of the spot in the two channels. The normalized intensity data for spot <italic>i </italic>is <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M2" name="1471-2105-7-215-i2" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msub><mml:msup><mml:mi>G</mml:mi><mml:mo>′</mml:mo></mml:msup><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuWGhbWrgaqbamaaBaaaleaacqWGPbqAaeqaaaaa@2F56@</mml:annotation></mml:semantics></mml:math></inline-formula> = κ<italic>G</italic><sub><italic>i </italic></sub>and <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M3" name="1471-2105-7-215-i3" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:msub><mml:msup><mml:mi>R</mml:mi><mml:mo>′</mml:mo></mml:msup><mml:mi>i</mml:mi></mml:msub></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacuWGsbGugaqbamaaBaaaleaacqWGPbqAaeqaaaaa@2F6C@</mml:annotation></mml:semantics></mml:math></inline-formula> = <italic>R</italic><sub><italic>i</italic></sub>, where κ is the normalization factor <inline-formula><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M4" name="1471-2105-7-215-i4" overflow="scroll"><mml:semantics definitionURL="" encoding=""><mml:mrow><mml:mi>κ</mml:mi><mml:mo>=</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mstyle displaystyle="true"><mml:msubsup><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>n</mml:mi></mml:msubsup><mml:mrow><mml:msub><mml:mi>R</mml:mi><mml:mi>i</mml:mi></mml:msub></mml:mrow></mml:mstyle><mml:mo stretchy="false">)</mml:mo><mml:mo>/</mml:mo><mml:mo stretchy="false">(</mml:mo><mml:mstyle displaystyle="true"><mml:msubsup><mml:mo>∑</mml:mo><mml:mrow><mml:mi>i</mml:mi><mml:mo>=</mml:mo><mml:mn>1</mml:mn></mml:mrow><mml:mi>n</mml:mi></mml:msubsup><mml:mrow><mml:msub><mml:mi>G</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo stretchy="false">)</mml:mo></mml:mrow></mml:mstyle></mml:mrow><mml:annotation encoding="MathType-MTEF">
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaaiiGacqWF6oWAcqGH9aqpcqGGOaakdaaeWaqaaiabdkfasnaaBaaaleaacqWGPbqAaeqaaaqaaiabdMgaPjabg2da9iabigdaXaqaaiabd6gaUbqdcqGHris5aOGaeiykaKIaei4la8IaeiikaGYaaabmaeaacqWGhbWrdaWgaaWcbaGaemyAaKgabeaakiabcMcaPaWcbaGaemyAaKMaeyypa0JaeGymaedabaGaemOBa4ganiabggHiLdaaaa@4680@</mml:annotation></mml:semantics></mml:math></inline-formula> Quackenbush [<xref ref-type="bibr" rid="B14">14</xref>] discusses this normalization in the context of sev-eral variations that are possible to address differing channel intensities.</p></sec><sec><title>Lowess normalization</title><p>Beyond the global dye bias, there is dye bias that is dependent on the measured spot intensities [<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B26">26</xref>]. TM4 constructs a scatter plot, called an RI-plot, of the points (<italic>x</italic><sub><italic>i</italic></sub>, <italic>y</italic><sub><italic>i</italic></sub>), where 1 ≤ <italic>i </italic>≤ <italic>n</italic>, given by <italic>x</italic><sub><italic>i </italic></sub>= log<sub>10</sub>(<italic>R</italic><sub><italic>i</italic></sub><italic>G</italic><sub><italic>i</italic></sub>) and <italic>y</italic><sub><italic>i </italic></sub>= log<sub>2</sub>(<italic>R</italic><sub><italic>i</italic></sub>/<italic>G</italic><sub><italic>i</italic></sub>). Under our assumption, the RI-plot should be very nearly symmetric with respect to the line <italic>y </italic>= 0. In lowess normalization, TM4 applies the lowess method of Cleveland [<xref ref-type="bibr" rid="B27">27</xref>] to fit a locally weighted regression curve to the RI-plot; TM4 then adjusts spot intensities to eliminate any systematic intensity-dependent bias. Additional details on correcting intensity-dependent bias is found in [<xref ref-type="bibr" rid="B14">14</xref>].</p></sec><sec><title>Standard deviation regularization</title><p>After total intensity and lowess normalizations eliminate dye bias on a global (per microarray) scale, TM4 employs standard deviation regularization to ensure that the per-block variances of log(<italic>R</italic><sub><italic>i</italic></sub>/<italic>G</italic><sub><italic>i</italic></sub>) values are the same [<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. Quackenbush [<xref ref-type="bibr" rid="B14">14</xref>] provides the formulas for this normalization step.</p></sec><sec><title>Low intensity filtering</title><p>Since the relative error in the log(<italic>R</italic><sub><italic>i</italic></sub>/<italic>G</italic><sub><italic>i</italic></sub>) values increases if <italic>R</italic><sub><italic>i </italic></sub>or <italic>G</italic><sub><italic>i </italic></sub>is close to background levels, spots with low intensities are filtered out. Quackenbush [<xref ref-type="bibr" rid="B14">14</xref>] provides additional details, which essentially require that both <italic>R</italic><sub><italic>i </italic></sub>and <italic>G</italic><sub><italic>i </italic></sub>intensities be above two standard deviations of the respective backgrounds.</p></sec><sec><title>The MIDAS pipeline</title><p>We applied a MIDAS pipeline consisting of total intensity normalization, lowess normalization, standard deviation regularization, and low intensity filtering to both microarray data sets. MIDAS default parameters were used throughout; the default low intensity filter cut-off is <italic>R</italic><sub><italic>i</italic></sub><italic>G</italic><sub><italic>i </italic></sub>< 10,000.</p></sec></sec><sec><title>TM4 MEV analysis</title><p>The Multi Experiment Viewer (MEV) component of TM4 provides a number of statistical analyses and clustering algorithms to identify differentially expressed genes. We report results from the one-class <italic>t</italic>-test analysis applied to output of the MIDAS pipeline. This test assumes that the paired distribution of treated and control groups is normally distributed. Since the intensities measured from the same spot are correlated, we can apply the one-class <italic>t</italic>-test for the two-group comparison.</p></sec></sec><sec><title>Authors' contributions</title><p>AAS performed the data preprocessing, Expresso analysis of all data sets, developed the database of statistical results, and drafted the manuscript. SPM performed the drought stress experiment (experiment 1), while PL performed the elevated CO<sub>2 </sub>experiment (experiment 2). PL performed the qRT-PCR of 192 genes from experiment 2. WS and PL performed TM4 analysis of experiments 1 and 2 respectively. LSH supervised the data analysis. LSH, HJB, and RG conceived of the study and coordinated the work. All authors read and approved the manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p><bold>Intensity-dependent dye bias in Rl-plots</bold>. Supplementary Figure <xref ref-type="fig" rid="F1">1</xref> is a PDF file that contains six RI-plots that illustrate specific intensity-dependent dye bias as the microarray data is processed through the normalization steps.</p></caption><media xlink:href="1471-2105-7-215-S1.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p>Genes subjected to qRT-PCR. Supplementary Table <xref ref-type="table" rid="T1">1</xref> is a PDF file that contains the annotation of genes subjected to qRT-PCR and the functional categories represented. For verification of microarray results in Experiment 2, Li, <italic>et al</italic>., [<xref ref-type="bibr" rid="B23">23</xref>] performed real-time quantitative reverse-transcriptase PCR (qRT-PCR) for selected genes — 55 in Col-0; 52 in Cvi-0; 59 in WS; 26 in Th. The AT numbers of those genes, their annotation, and categorizations into biological functions are in the table.</p></caption><media xlink:href="1471-2105-7-215-S2.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Chloroquine/Sulphadoxine-Pyrimethamine for Gambian Children with Malaria: Transmission to Mosquitoes of Multidrug-Resistant <named-content content-type="genus-species">Plasmodium falciparum</named-content>
|
<sec id="st1"><title>Objectives:</title><p>In the Gambia, chloroquine (CQ) plus sulphadoxine-pyrimethamine (SP) is the first-line antimalarial treatment. <named-content content-type="genus-species">Plasmodium falciparum</named-content> parasites carrying mutations associated with resistance to each of these drugs were present in 2001 but did not cause a significant loss of therapeutic efficacy among children receiving the combination CQ/SP. We measured their effect on parasite transmission to <named-content content-type="genus-species">Anopheles gambiae</named-content> mosquitoes.</p></sec><sec id="st2"><title>Design:</title><p>We conducted a single-blind, randomised, controlled trial with follow-up over 28 d. Mosquito feeding experiments were carried out 7, 10, or 14 d after treatment.</p></sec><sec id="st3"><title>Setting:</title><p>The study took place in the town of Farafenni and surrounding villages in the Gambia.</p></sec><sec id="st4"><title>Participants:</title><p>Participants were 500 children aged 6 mo to 10 y with uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> malaria<bold>.</bold>
</p></sec><sec id="st5"><title>Interventions:</title><p>Children were randomised to receive CQ, SP, or CQ/SP.</p></sec><sec id="st6"><title>Outcome Measures:</title><p>Outcomes related to transmission were determined, including posttreatment gametocyte prevalence and density. Infectiousness was assessed by membrane-feeding <named-content content-type="genus-species">A. gambiae</named-content> mosquitoes with blood from 70 gametocyte-positive patients. Mutations at seven loci in four genes associated with drug resistance were measured pre- and posttreatment and in the midguts of infected mosquitoes.</p></sec><sec id="st7"><title>Results:</title><p>After SP treatment, the infectiousness of gametocytes was delayed, compared to the other two treatment groups, despite comparable gametocyte densities. Among bloodmeal gametocytes and the midguts of infected mosquitoes, the presence of the four-locus multidrug-resistant haplotype TYRG (consisting of mutations <italic>pfcrt</italic>-76T, <italic>pfmdr1</italic>-86Y, <italic>pfdhfr</italic>-59R, and <italic>pfdhps</italic>-437G) was associated with significantly higher oocyst burdens after treatment with the combination CQ/SP.</p></sec><sec id="st8"><title>Conclusions:</title><p>Parasites with a multidrug-resistant genotype had a substantial transmission advantage after CQ/SP treatment but did not have a significant impact on in vivo efficacy of this drug combination. Protocols that include measuring transmission endpoints as well as therapeutic outcomes may be a useful strategy when monitoring the evolution of drug resistance in malaria parasites in vivo.</p></sec>
|
<contrib contrib-type="author"><name><surname>Hallett</surname><given-names>Rachel L</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Dunyo</surname><given-names>Samuel</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Ord</surname><given-names>Rosalynn</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Jawara</surname><given-names>Musa</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Pinder</surname><given-names>Margaret</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Randall</surname><given-names>Anna</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Alloueche</surname><given-names>Ali</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Walraven</surname><given-names>Gijs</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Targett</surname><given-names>Geoffrey A. T</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Alexander</surname><given-names>Neal</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Sutherland</surname><given-names>Colin J</given-names></name><xref ref-type="aff" rid="aff1">1</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib>
|
PLoS Clinical Trials
|
<sec id="s1"><title>INTRODUCTION</title><p>The widespread occurrence of drug-resistant parasites in malaria endemic regions has led to rapid changes in antimalarial treatment policies in many countries. In some parts of Southeast Asia, multidrug-resistant parasites are prevalent and have greatly reduced the efficacy of the majority of monotherapy regimens. This has led to the implementation of combination therapies, including artemisinin-based treatment throughout the region [<xref rid="pctr-0010015-b001" ref-type="bibr">1</xref>]. In sub-Saharan Africa, monotherapies such as chloroquine (CQ) and the antifolate fixed combination sulphadoxine-pyrimethamine (SP) have continued to be widely used. Parasites resistant to CQ are common and known to contribute to excess severe malaria cases [<xref rid="pctr-0010015-b002" ref-type="bibr">2</xref>], and SP-resistant genotypes have spread widely across the continent from a few focal origins [<xref rid="pctr-0010015-b003" ref-type="bibr">3</xref>]. Thus, despite there being few studies of multidrug-resistant <named-content content-type="genus-species">Plasmodium falciparum</named-content> in Africa, it is likely that parasites harbouring mutations conferring resistance to both CQ and SP circulate in many areas. Policy changes towards the implementation of artemisinin-containing combination therapies (ACTs) are now occurring in many countries [<xref rid="pctr-0010015-b004" ref-type="bibr">4</xref>].</p><p>One of the potential benefits of ACT use is that the artemisinin component has the demonstrated ability to reduce gametocyte carriage and thus transmission to mosquitoes [<xref rid="pctr-0010015-b005" ref-type="bibr">5</xref>]. Other antimalarial drugs vary in their effect on gametocytes; CQ and other 4-aminoquinolines exert their effect on the parasite's haem-metabolising stages, which includes early-stage gametocytes. In contrast, antifolates such as SP are only effective during asexual development and will not affect parasites once these are committed to the sexual pathway [<xref rid="pctr-0010015-b006" ref-type="bibr">6</xref>].</p><p>The implementation of ACTs is costly and is currently beset with problems of supply due to the very rapid expansion in demand for artemisinin drugs in the last few years. Therefore, some African Ministries of Health have moved to combinations of established antimalarials as a first step away from monotherapies, in anticipation of a second switch to ACTs as soon as it is feasible. Senegal has implemented SP plus amodiaquine as first-line replacement for CQ monotherapy, with a move to amodiaquine plus artesunate planned for 2006. In the Gambia, where CQ monotherapy is no longer efficacious [<xref rid="pctr-0010015-b007" ref-type="bibr">7</xref>], it has been replaced by CQ combined with SP (CQ/SP), with the longer-term goal being implementation of ACT.</p><p>Earlier studies in the Gambia had shown that CQ plus SP was as efficacious as SP alone as treatment for uncomplicated malaria in children and was more effective at providing rapid symptomatic relief [<xref rid="pctr-0010015-b008" ref-type="bibr">8</xref>]. However, in studies of the infectiousness to mosquitoes of treated Gambian children with uncomplicated falciparum malaria, we have found that SP and CQ both permit substantial transmission during the first week following treatment [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>,<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>] as does the combination CQ/SP [<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>]. CQ-resistant parasites were already highly prevalent in the Farafenni area when these studies were carried out and displayed enhanced gametocyte carriage and higher intensity of mosquito infection than wild-type parasites in CQ-treated individuals [<xref rid="pctr-0010015-b012" ref-type="bibr">12</xref>,<xref rid="pctr-0010015-b013" ref-type="bibr">13</xref>]. Furthermore, we have shown that mutant forms of the <italic>pfdhfr</italic> and <italic>pfdhps</italic> loci associated with SP resistance and carriage of gametocytes after SP treatment [<xref rid="pctr-0010015-b014" ref-type="bibr">14</xref>] were also common [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. We were concerned therefore that the combination CQ/SP would selectively enhance the transmission and circulation of multidrug-resistant parasites. Such an enhancement may persist for some time after treatment, given the long half-life of SP, and the longevity of <named-content content-type="genus-species">P. falciparum</named-content> gametocytes.</p><p>In 2001, we conducted a single-blind, randomised, controlled trial of CQ versus SP versus CQ/SP in Gambian children aged 6 mo to 10 y presenting to Farafenni Hospital with uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> malaria. We report in the accompanying paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>] the therapeutic efficacy of the regimens used, and the identification of a four-locus parasite genotype that was associated with treatment failure and may represent a multidrug-resistant phenotype in the Gambia. Posttreatment gametocyte carriage and infectiousness of gametocyte carriers to mosquitoes are examined here. We also evaluated the contribution of parasites carrying a multidrug-resistant genotype to transmission success in each treatment group.</p></sec><sec id="s2"><title>METHODS</title><p>The study was a single-blind, randomised, controlled trial to compare efficacy of treatment of children with uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> malaria with CQ, SP, or CQ/SP. Full details of the participants and interventions are described in the accompanying paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p><sec id="s2a"><title>Objectives</title><p>The primary objectives were to measure posttreatment gametocyte prevalence and density in each of the three treatment groups and to assess how this related to infectiousness to mosquitoes. Within the SP group, the effect of time after treatment was also examined, because gametocyte samples were collected on three different follow-up days. Secondary measurements included the contribution of genetically defined drug-resistant parasites to mosquito oocyst burden and the effect of drug group on transmission success.</p></sec><sec id="s2b"><title>Outcomes</title><p>Outcomes measured were gametocyte carriage and density on defined follow-up days, and the proportion of mosquitoes infected. The number of oocysts was counted in each infected mosquito midgut. For all infections used for mosquito feeding, the genotype at drug resistance-associated loci was determined at three time points: pretreatment, and posttreatment in the gametocyte-positive blood feed sample and in the mosquito midguts arising from successful infections.</p></sec><sec id="s2c"><title>Sample Size and Randomisation</title><p>The sample size, randomisation, and blinding are described in the accompanying paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p></sec><sec id="s2d"><title>Posttreatment Follow-Up</title><p>Follow-up procedures relating to efficacy endpoints are described in the accompanying paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. In order to fulfill the transmission objectives, children were collected from their homes either on day 7, 10, or 14 after treatment and taken to the Medical Research Council Field Station in Farafenni, where they were clinically examined and finger prick blood samples were obtained for thick blood film preparation and packed cell volume estimation. The blood films were stained with Field's stain and examined immediately for malaria parasites. Children with gametocytes were requested to provide 3.0 ml of venous blood for mosquito infectivity experiments as previously described [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>,<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>]. For those who had received CQ or CQ/SP, this happened on day 7 only. Among the SP-treated group, children were assigned a gametocyte screening day according to the day of the week they were recruited to overcome the logistical problems that would be caused by a nested second randomisation. Thursday and Friday recruits receiving SP were screened for gametocytes on day 7, Tuesday and Saturday recruits were screened on day 10, and Monday and Wednesday recruits were screened on day 14. Recruiting did not occur on Sundays. SP-treated children who were screened on day 10 or 14 were not visited on day 7 and therefore were not subject to follow-up equivalent to that of SP-treated children screened on day 7 or of children in the other treatment groups, all of whom were screened on day 7.</p></sec><sec id="s2e"><title>Membrane-Feeding and Mosquito Dissection</title><p>Venous blood samples were obtained from children with a packed cell volume of >20%, who had been free of peripheral gametocytes at the time of recruitment (day 0) but who were gametocytaemic (limit of detection, 5 gametocytes/μl) when screened on day 7, 10, or 14. Specific consent for the procedure was obtained from the parents or guardians. If no children fulfilling these criteria were available as donors but cages of mosquitoes had been prepared, children who had been gametocyte positive on day 0 were selected. Membrane-feeding was performed as described elsewhere [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>,<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>]. In brief, venous blood in citrate-phosphate dextrose was centrifuged, and the plasma was removed. After being washed, the red blood cell pellet was split into two aliquots of 300–500 μl each. These were resuspended to a packed cell volume of 33% in, respectively, the original autologous plasma and in pooled AB serum from European donors with no history of malaria exposure (control serum). Each suspension was then fed to approximately 50 3- to 5-d-old female <named-content content-type="genus-species">Anopheles gambiae</named-content> mosquitoes (obtained as F<sub>1</sub> progeny of wild-caught gravid females) via an artificial membrane attached to a water-jacketed glass feeder maintained at 37 °C. Mosquito midguts were dissected out 7–8 d later, and the number of malaria oocysts on each one was recorded. Each oocyst-positive midgut was retained in a separate tube of 95% ethanol.</p></sec><sec id="s2f"><title>Parasite Genotyping</title><p>DNA was extracted from day 0 bloodspots on filter paper and from the remainder of the venous blood sample taken for membrane-feeding by boiling with chelex-100 [<xref rid="pctr-0010015-b016" ref-type="bibr">16</xref>]. Oocyst DNA was obtained from infected mosquito midguts as follows: each sample was rehydrated in 500 μl of a 1:1 mix of ethanol and TE buffer (10 mM Tris-HCl, 1 mM EDTA) before being incubated at 56 °C overnight in 200 μl of 6% chelex containing 20 μg of proteinase K. Samples were then boiled for 30 min and 5 μl of the supernatant was used in PCR amplifications [<xref rid="pctr-0010015-b017" ref-type="bibr">17</xref>].</p><p>A sequence-specific oligonucleotide probing assay was used to detect polymorphisms at loci associated with CQ and SP resistance in all PCR-positive DNA extracts [<xref rid="pctr-0010015-b012" ref-type="bibr">12</xref>,<xref rid="pctr-0010015-b018" ref-type="bibr">18</xref>]. These were <italic>pfcrt</italic>-K76T, <italic>pfmdr1</italic>-N86Y, <italic>pfmdr1</italic>-Y184F (linked to CQ resistance), and <italic>pfdhfr</italic>-N51I, <italic>pfdhfr</italic>-C59R, <italic>pfdhfr</italic>-S108N, <italic>pfdhps</italic>-A437G, and <italic>pfdhps</italic>-K540E (linked to SP resistance). For some analyses, where mixtures of drug-sensitive and drug-resistant parasites were detected, they were scored as resistant to reflect the expected phenotype of the infection.</p></sec><sec id="s2g"><title>Statistical Methods</title><p>All data were double entered and verified using Epi-Info, version 6 (Centers for Disease Control and Prevention, Atlanta, Georgia, United States). Clinical, parasitological, and entomological data entered in Epi-Info were transferred to Stata 8.1 (Stata Corporation, College Station, Texas, United States) for statistical analysis. Categorical variables were compared among groups using odds ratios and significance tested by the Fisher exact test. Gametocyte densities counted in thick films and oocyst density on mosquito midguts were compared between treatment groups in pairwise fashion. The ratio of the arithmetic means was calculated, and generalized linear models with a logarithmic link function and negative binomial distribution family were fitted to the data. This models the logarithm of the (arithmetic) mean gametocyte density, or oocyst density as previously described [<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>,<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>]. Ratios of mean densities were obtained by exponentiating the relevant regression coefficients. Oocyst data were corrected for clustering within mosquito cages fed on blood from the same individual.</p></sec></sec><sec id="s3"><title>RESULTS</title><sec id="s3a"><title>Participant Flow</title><p>A total of 500 eligible patients were enrolled into the trial, as described in detail elsewhere [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. After allocation to treatment groups, 193, 181, and 126 children were treated with CQ/SP, SP, and CQ, respectively. Among the SP-treated group, 79, 40, and 62 children were scheduled for gametocyte screening on days 7, 10, and 14, respectively. A trial profile is presented in <xref ref-type="fig" rid="pctr-0010015-g001">Figure 1</xref>, and clinical and parasitological outcomes are presented in the accompanying paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p><fig id="pctr-0010015-g001" position="float"><label>Figure 1</label><caption><title>CONSORT Flowchart</title><p>A total of 1,366 children was screened, and 500 were randomised. Children randomised to receive SP were scheduled for gametocyte screening on day 7, 10, or 14 (see text). Shaded boxes represent gametocyte screening days. Loss to follow-up is detailed in the main trial paper [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p></caption><graphic xlink:href="pctr.0010015.g001"/></fig></sec><sec id="s3b"><title>Gametocyte Prevalence and Density</title><p>The prevalence of posttreatment gametocyte carriage at each day of follow-up among children without gametocytes at presentation is shown in <xref ref-type="fig" rid="pctr-0010015-g002">Figure 2</xref>. Approximately 20% of children in each treatment group were gametocyte-positive at day 0. This is consistent with our previous findings [<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>]. These children, who had a 90% chance of carrying gametocytes at some point during follow-up after treatment (data not shown), are omitted from prevalence estimates at subsequent time points. In pairwise comparisons, there was no significant difference in gametocyte carriage between the CQ and CQ/SP treatment groups at day 3 (<italic>p</italic> = 0.71), day 7 (<italic>p</italic> = 0.066), day 14 (<italic>p</italic> = 0.82), or day 28 (<italic>p</italic> = 0.26). In contrast, SP-treated children were significantly more likely to become gametocyte carriers than were CQ/SP-treated or CQ-treated children at day 3 (<italic>p</italic> = 0.009, 0.054, respectively), day 7 (<italic>p</italic> = 0.002, < 0.001), and day 14 (<italic>p</italic> < 0.001 in both cases), but not at day 28 (<italic>p</italic> = 0.089, 0.66).</p><fig id="pctr-0010015-g002" position="float"><label>Figure 2</label><caption><title>Prevalence and Positive Arithmetic Mean Density of Gametocytes in Treated Children</title><p>Those carrying gametocytes at day 0 were excluded from the denominator at subsequent follow-up days (see data table). Bars represent prevalence of gametocyte carriage, with error bars showing 95% CIs. Lines represent arithmetic mean gametocyte density in carriers only.</p></caption><graphic xlink:href="pctr.0010015.g002"/></fig><p>Gametocyte carriage in the CQ treatment group was closely associated with treatment failure, as previously observed in this population [<xref rid="pctr-0010015-b010" ref-type="bibr">10</xref>,<xref rid="pctr-0010015-b012" ref-type="bibr">12</xref>]. Gametocyte carriage during follow-up was observed in 14 of 39 children (35.9%) presenting free of gametocytes and successfully treated with CQ, compared to 40 of 53 children (75.5%) who carried <named-content content-type="genus-species">P. falciparum</named-content> trophozoites at any point during the 28 d of follow-up (odds ratio, 5.17; 95% confidence interval [CI], 1.95–13.9; <italic>p</italic> < 0.0001). This was not seen in either of the other treatment groups, which were both highly efficacious [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p><p>A pairwise comparison of mean gametocyte density between treatment groups at each time point was carried out, including all evaluable children who were not gametocyte carriers at presentation. Gametocyte density in the SP-treated group was significantly higher than in the CQ/SP group at days 7, 14, and 28 (ratio of means, 3.06 [<italic>p</italic> = 0.031], 3.68 [<italic>p</italic> = 0.001], and 31.2 [<italic>p</italic> < 0.001], respectively). Gametocyte density in the CQ/SP-treated group was higher than in the CQ group at day 3 (ratio of means, 3.76 [<italic>p</italic> = 0.046]), but there was no significant difference between these two groups at day 7 (ratio of means, 1.73 [<italic>p</italic> = 0.30]) or day 14 (ratio of means, 1.69 [<italic>p</italic> = 0.25]). At day 28, gametocyte density in the CQ/SP-treated group was significantly lower than in the CQ group (ratio of means, 0.111 [<italic>p</italic> = 0.001]).</p></sec><sec id="s3c"><title>Infectiousness of Gametocyte Carriers</title><p>Venous blood was donated by 72 gametocyte carriers for membrane-feeding, resulting in 70 evaluable feeding experiments, and the dissection of 2033 blood-fed adult female mosquitoes. Results of the feeding experiments in both autologous plasma and control serum are presented in <xref ref-type="table" rid="pctr-0010015-t001">Table 1</xref>.</p><table-wrap id="pctr-0010015-t001" content-type="2col" position="float"><label>Table 1</label><caption><p>Results of Membrane-Feeding Experiments</p></caption><graphic xlink:href="pctr.0010015.t001"/></table-wrap><p>As a measure of infectiousness, the mean number of oocysts per infected midgut was compared among treatment groups and across different feed days in the SP-treated group (<xref ref-type="fig" rid="pctr-0010015-g003">Figure 3</xref> and <xref ref-type="table" rid="pctr-0010015-t002">Table 2</xref>). In regression analysis, the mean oocyst burden did not differ significantly between the CQ and CQ/SP groups (data not shown). Among SP-treated donors, those identified 10 or 14 d after treatment were significantly more infectious than were day 7 donors (<italic>p</italic> = 0.008, <italic>p</italic> = 0.020 respectively), despite high gametocyte densities in the latter group (<xref ref-type="fig" rid="pctr-0010015-g003">Figure 3</xref> and <xref ref-type="table" rid="pctr-0010015-t002">Table 2</xref>).</p><fig id="pctr-0010015-g003" position="float"><label>Figure 3</label><caption><title>Relationship between Gametocyte Density and Oocyst Burden in Feed Samples</title></caption><graphic xlink:href="pctr.0010015.g003"/></fig><table-wrap id="pctr-0010015-t002" content-type="2col" position="float"><label>Table 2</label><caption><p>Mean Oocyst Densities after Feeds from All Gametocyte Carriers at d 7 after Treatment, and in SP-Treated Carriers at Days 10 and 14</p></caption><graphic xlink:href="pctr.0010015.t002"/></table-wrap><p>When we compared SP-treated children who carried gametocytes on day 0 with those who did not, significantly greater infectiousness in day 10 and 14 gametocyte carriers was seen only among those with gametocytes at presentation (<xref ref-type="table" rid="pctr-0010015-t002">Table 2</xref>). There is therefore an important contribution of gametocyte maturity to infectivity.</p></sec><sec id="s3d"><title>Contribution of Resistant Genotypes to Circulating Gametocyte Populations</title><p>We report elsewhere on the prevalence of resistance-associated parasite genotypes in the study population, and the contribution of such parasites to treatment failure [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. It was found that multidrug-resistant parasites exemplified by the four-allele genotype TYRG (i.e., simultaneous carriage of <italic>pfcrt</italic>-76T, <italic>pfmdr1</italic>-86Y, <italic>pfdhfr</italic>-59R, <italic>pfdhps</italic>-437G) were contributing significantly to parasitological failure after treatment with SP. Parasites with the TYRG genotype were also present in some cases after treatment with CQ/SP and thus likely to be selected with widespread combination drug pressure. We therefore tested for associations between either posttreatment gametocyte emergence or transmission success and resistance-associated alleles at seven loci alone and in combination.</p><p>Carriage of genetically resistant parasites at presentation was associated with the presence of gametocytes during the follow-up period, as shown in <xref ref-type="fig" rid="pctr-0010015-g004">Figure 4</xref>. Gametocyte donors carried higher pretreatment prevalences of CQ and SP resistance-associated markers than a random sample of the trial population who had remained gametocyte negative. This increase was significant for the CQ resistance markers <italic>pfcrt</italic>-76T and <italic>pfmdr1</italic>-86Y, but not for <italic>pfmdr1</italic>-184F. Similarly, the difference was significant when analysing the pyrimethamine-resistant mutations in the <italic>pfdhfr</italic> gene (codons 51, 59, and 108) both individually and together as presence of the “triple mutant.” <italic>Pfdhps</italic>-437G was the only sulphadoxine-resistant mutation detected in this population, and was found at a higher prevalence in gametocyte donors at day 0, but this difference was not significant. When grouping mutations into the multidrug resistance-associated genotype TYRG as defined in Dunyo et al. [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>] we found that this combination of mutations showed an association with posttreatment gametocytaemia (odds ratio, 3.0; 95% CI, 0.78<bold>–</bold>17.0; two-sided Fisher exact, <italic>p</italic> = 0.11), although this was also not statistically significant.</p><fig id="pctr-0010015-g004" position="float"><label>Figure 4</label><caption><title>Contribution of Resistance-Associated Parasite Genotypes at Day 0 to Subsequent Gametocyte Emergence</title><p>Only subjects free of circulating gametocytes at enrollment are included. Pretreatment genotype data were obtained for 27 patients who remained gametocyte-free during follow-up (white bars) and 46 patients with emergent gametocytes (grey bars). Pretreatment genotype prevalences for 47 gametocyte feed donors are shown by the black bars. * indicates a significantly higher prevalence of the designated genotype than among the baseline nongametocyte carriage at presentation.</p></caption><graphic xlink:href="pctr.0010015.g004"/></fig></sec><sec id="s3e"><title>Selection of Drug Resistance Markers among Gametocyte Donors</title><p>We have previously presented evidence of within-host directional selection from <italic>pfcrt</italic>-76K to T after CQ treatment [<xref rid="pctr-0010015-b013" ref-type="bibr">13</xref>]. However, in the current dataset, too few genotype changes occurred between recruitment and mosquito-feeding to perform this analysis for any of the loci examined. Resistance markers were highly prevalent in presentation samples and, in general, genotypes detected in gametocyte-positive feed samples and in resulting oocysts reflected what was present at day 0 (<xref ref-type="supplementary-material" rid="pctr-0010015-st001">Table S1</xref>).</p><p>In two feeds (MF27 and MF40), mixed infections were detected in some midguts, but not others, at the <italic>pfdhfr</italic> and <italic>pfdhps</italic>-437 sites examined. In one further feed (MF59), six of seven infected midguts carried <italic>pfmdr1</italic>-184F and one had <italic>pfmdr1</italic>-184Y. In all other cases where multiple oocyst-positive midguts arose from a single feed, they contained identical parasite genotypes (<xref ref-type="supplementary-material" rid="pctr-0010015-st001">Table S1</xref>).</p><p>The most infectious gametocyte sample (MF68) originated from a CQ/SP-treated patient, producing 13 infected midguts with a mean of 42 oocysts on each (range, 2–100). Eleven of the 13 midguts were PCR-positive for all four resistance-associated genes. CQR alleles <italic>pfcrt</italic>-76T, <italic>pfmdr1</italic>-86Y, and <italic>pfmdr1</italic>-184F were detected pre- and posttreatment and in all 11 midguts. The <italic>pfdhfr</italic> genotype was mixed with both wild-type and resistance-associated alleles detected at positions 51, 59, and 108 both pre- and posttreatment. However, none of these wild-type alleles were detected in the resulting 11 midguts, all of which harboured pure <italic>pfdhfr</italic> triple mutants (51-I, 59-R, and 108-N).</p></sec><sec id="s3f"><title>Do Resistant Infections at Day 0 Contribute More to Later Transmission?</title><p>Very few changes in parasite genotype at resistance-associated loci were observed from pretreatment sampling to identification of markers in infected mosquito midguts. Therefore, we used day 0 data to look for associations between multidrug-resistant genotypes and mean oocyst count in mosquitoes membrane-fed on posttreatment emergent gametocytes (i.e., considering only those subjects without gametocytes at time of treatment). The presence of parasites with the four-locus multidrug-resistant genotype TYRG at day 0 was associated with higher oocyst numbers in SP and CQ/SP groups, but this was not statistically significant in either group (for SP, <italic>N</italic> = 644 dissected mosquitoes: mean ratio, 4.14; 95% CI, 0.722–23.8; <italic>p</italic> = 0.111; for CQ/SP, <italic>N</italic> = 415: mean ratio, 5.53; 95% CI, 0.742–41.2; <italic>p</italic> = 0.095). In contrast, in the CQ group, only one donor harboured TYRG parasites at day 0, and there was no transmission from this child's gametocytes. Parasites carrying CQ resistance markers at <italic>pfcrt</italic> and <italic>pfmdr1</italic> without accompanying antifolate resistance mutations in <italic>pfdhfr</italic> and <italic>pfdhps</italic> were highly transmissible in the CQ group (<italic>N</italic> = 186; mean oocyst ratio: 261; 95% CI, 24.8–2736; <italic>p</italic> < 0.001). This is consistent with previous findings [<xref rid="pctr-0010015-b013" ref-type="bibr">13</xref>].</p></sec><sec id="s3g"><title>Gametocyte and Oocyst Genotypes Impact Significantly on Transmission Success</title><p>An unambiguous four-locus genotype could be assigned to gametocyte feed samples from 56 children. Where mixed infections were detected, these were scored as resistant. Fourteen of the 56 were from children who had harboured gametocytes at the time of treatment; four, eight, and two children from the CQ, SP and CQ/SP treatment groups, respectively. We analysed the contribution of TYRG parasites to transmission in the 1680 mosquitoes that had fed on these 56 gametocyte samples, after stratifying by drug group. This was carried out by calculating the mean oocyst number on midguts resulting from TYRG-positive feed samples and comparing this to the mean oocyst number on midguts resulting from feeds containing any other genotypes, and producing a ratio of the means. In the same way, we then compared oocyst burden in the 48 midguts for which four-locus genotype data were available. Results of both analyses are presented in <xref ref-type="table" rid="pctr-0010015-t003">Table 3</xref>.</p><table-wrap id="pctr-0010015-t003" content-type="2col" position="float"><label>Table 3</label><caption><p>Contribution to Transmission Success of Multilocus Resistant Genotypes in Gametocytes and Oocysts</p></caption><graphic xlink:href="pctr.0010015.t003"/></table-wrap><p>The presence of TYRG in gametocyte samples and in oocyst DNA was associated with a significantly lower oocyst burden from CQ-treated individuals. In the SP-treated group, TYRG gametocytes also led to lower oocyst burdens, with borderline significance (<italic>p</italic> = 0.051). However, after CQ/SP treatment, parasites carrying TYRG resulted in significantly greater transmission success compared to other genotypes in this treatment group.</p></sec></sec><sec id="s4"><title>DISCUSSION</title><sec id="s4a"><title>Interpretation</title><p>When compared to the rapidly waning efficacy of CQ monotherapy in our study area [<xref rid="pctr-0010015-b007" ref-type="bibr">7</xref>], addition of SP to CQ provided therapeutic efficacy of over 90% after PCR correction in both this trial in 2001, and in a trial carried out 1 y later [<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. CQ/SP has been the recommended treatment for uncomplicated malaria in the Gambia since 2004. SP alone has had a measured parasitological efficacy of approximately 90% over 28 d in the Gambia [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>,<xref rid="pctr-0010015-b019" ref-type="bibr">19</xref>]. However, both SP and CQ/SP are characterized by substantial rates of posttreatment gametocyte emergence [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>,<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>], and because alleles of <italic>pfdhfr</italic> and <italic>pfdhps</italic> associated with SP failure are present in the Farafenni area, we expected that parasites resistant to both drugs would be frequently transmitted following combination treatment.</p><p>In this study, CQ/SP treatment was followed by little reduction in posttreatment prevalence and density of gametocytes, compared to CQ alone. Only at the day 28 follow-up point was gametocyte carriage in the CQ/SP-treated group significantly lower than in the CQ group. However, gametocytes occurred more frequently and at higher density in the SP group throughout posttreatment follow-up and particularly at day 28 (<xref ref-type="fig" rid="pctr-0010015-g003">Figure 3</xref>), suggesting that CQ/SP-treated children may be infectious for a shorter period of time following treatment than those treated with SP alone. This may be the result of more rapid asexual parasite clearance by CQ/SP [<xref rid="pctr-0010015-b008" ref-type="bibr">8</xref>], combined with the activity of CQ against early developing gametocytes [<xref rid="pctr-0010015-b006" ref-type="bibr">6</xref>].</p><p>Gametocyte carriers from all three treatment groups were infectious to <named-content content-type="genus-species">A. gambiae</named-content> in experimental feeding. Remarkably, the high gametocyte densities observed among the SP-treated group at day 7 in particular did not translate into substantial infectiousness (<xref ref-type="fig" rid="pctr-0010015-g003">Figure 3</xref>). Gametocyte carriers in the SP group identified at day 10 or 14 were significantly more infectious to mosquitoes, suggesting that gametocytes circulating at day 7 after SP treatment, but not CQ or CQ/SP treatment, have not attained sufficient maturity to infect mosquitoes efficiently. One explanation for this observation is that, because SP appears unable to kill even the earliest developing gametocyte stages sequestered in the host tissues at the time of treatment [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>], many of the emergent gametocytes seen on day 7 are not as yet fully mature, although morphologically they appear so. In contrast, CQ will kill genotypically sensitive gametocytes in the early stages of development, and so day 7 emergent gametocytes are those that were already some way to maturity at the time of treatment. However, this does not explain why the significant numbers of CQ-resistant parasites were not also less infectious at day 7 in the CQ treatment group, because their early development is expected to be unimpaired by drug. An alternative hypothesis is that SP treatment in some way perturbs the gametocyte sex ratio, so that the accumulation of one sex is delayed, thus preventing fertilization until sufficient of both sexes are present in the gametocyte population. It remains unclear why the maturity of gametocytes was a particularly important factor in the infectiousness of children who had also harboured gametocytes at day 0.</p><p>Resistance-associated alleles at five of the seven loci examined were more common among gametocyte carriers than among noncarriers in a randomly selected group of pretreatment isolates, as was the <italic>pfdhfr</italic> “triple mutant” haplotype of 51-I, 59-R, and 108-N (<xref ref-type="fig" rid="pctr-0010015-g004">Figure 4</xref>). The multilocus genotype TYRG (<italic>pfcrt-</italic>76T, <italic>pfmdr1</italic>-86Y, <italic>pfdhfr</italic>-59R, and <italic>pfdhps-</italic>437G) was also more common among gametocyte carriers, and among children donating gametocytes for feeding experiments, than among noncarriers, although this was not statistically significant. In the accompanying paper, the TYRG genotype was found to be weakly associated with parasitological failure after SP treatment, but not after CQ/SP treatment [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>].</p><p>The presence of the TYRG multilocus genotype among either gametocytes membrane-fed to mosquitoes, or among <named-content content-type="genus-species">P. falciparum</named-content> oocysts present in infected mosquito midguts was associated with significantly higher oocyst burdens in the CQ/SP combination group. This was independent of feed type or the presence of gametocytes prior to treatment, and was much stronger than the association between this genotype and treatment failure [<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. TYRG carrying parasites thus appear to enjoy a substantial survival advantage, which leads to favourable transmission under CQ/SP treatment. In contrast, TYRG genotypes were less successfully transmitted than other genotypes in both monotherapy arms; in the CQ treatment group, this difference was statistically significant. This supports the view that drug-resistant parasites are likely to be less fit, relative to other genotypes in the population, in the absence of the particular drug or combination of drugs that has led to their selection [<xref rid="pctr-0010015-b020" ref-type="bibr">20</xref>].</p></sec><sec id="s4b"><title>Generalizability</title><p>Our data did not provide clear evidence of within-host selection for the TYRG genotype, as we have previously observed in CQ-treated patients [<xref rid="pctr-0010015-b012" ref-type="bibr">12</xref>,<xref rid="pctr-0010015-b013" ref-type="bibr">13</xref>]. This is probably due to the fact that some of the alleles, particularly <italic>pfcrt-</italic>76T and <italic>pfdhfr-</italic>108N, were at a high starting prevalence and so wild-type alleles were rare.</p><p>Taken together, our analyses of therapeutic and transmission endpoints of this clinical trial demonstrate that multidrug-resistant parasites enjoyed a survival advantage in terms of their likelihood of transmission to mosquitoes, but that the particular genotypes encountered in the Gambia did not threaten the therapeutic efficacy of CQ/SP [<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010015-b015" ref-type="bibr">15</xref>]. Pretreatment genotyping of 90 children in a subsequent trial in 2002 [<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>] showed that <italic>pfdhfr</italic> and <italic>pfdhps</italic> mutation prevalences remained stable (<italic>pfdhfr</italic> triple mutant present in 50 of 89 or 56.2%; 95% CI, 45.9<bold>–</bold>66.5; <italic>pfdhps</italic> 437G present in 39 of 79 or 49.4%; 95% CI, 38.4<bold>–</bold>60.4) (R. Hallett and C. Sutherland, unpublished data). However, CQ monotherapy remained first line treatment at this time and CQ/SP was not being commonly used. The apparent anomaly between the widespread prevalence of the <italic>pfdhfr</italic> triple mutant in Africa and its weak association with risk of treatment failure [<xref rid="pctr-0010015-b021" ref-type="bibr">21</xref>] may simply be because its primary effect has been to enhance transmission of parasites under SP drug pressure, and this has been sufficient to ensure its persistence and spread.</p></sec><sec id="s4c"><title>Overall Evidence</title><p>The work presented in this paper extends our previous studies of CQ-treated, SP-treated, and CQ/SP-treated patients in this study area [<xref rid="pctr-0010015-b009" ref-type="bibr">9</xref>–<xref rid="pctr-0010015-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010015-b013" ref-type="bibr">13</xref>]. There are no previous published studies to our knowledge that measure transmission endpoints among African patients treated with CQ/SP. We have shown that multidrug-resistant <named-content content-type="genus-species">P. falciparum</named-content> are prevalent in West Africa and that transmission of these genotypes is favoured when CQ and SP are combined for treatment of uncomplicated malaria in children. Thus, evolution of resistance to this combination will continue despite any reduction in its therapeutic efficacy, so far. These findings demonstrate the importance of measuring transmission endpoints in studies of the evolution of resistance to antimalarial chemotherapy.</p></sec></sec><sec sec-type="supplementary-material" id="s5"><title>SUPPORTING INFORMATION</title><supplementary-material content-type="local-data" id="pctr-0010015-sd001"><label>CONSORT Checklist</label><media xlink:href="pctr.0010015.sd001.doc"><caption><p>Click here for additional data file.</p></caption></media><p>(50 KB DOC)</p></supplementary-material><supplementary-material content-type="local-data" id="pctr-0010015-sd002"><label>Trial Protocol</label><media xlink:href="pctr.0010015.sd002.doc"><caption><p>Click here for additional data file.</p></caption></media><p>(88 KB DOC)</p></supplementary-material><supplementary-material content-type="local-data" id="pctr-0010015-st001"><label>Table S1</label><caption><p>(343 KB DOC)</p></caption><media xlink:href="pctr.0010015.st001.doc"><caption><p>Click here for additional data file.</p></caption></media></supplementary-material></sec>
|
Randomised Trial of Chloroquine/Sulphadoxine-Pyrimethamine in Gambian Children with Malaria: Impact against Multidrug-Resistant <named-content content-type="genus-species">P. falciparum</named-content>
|
<sec id="st1"><title>Objectives:</title><p>In the Gambia, the combination of chloroquine (CQ) and sulphadoxine-pyrimethamine (SP) has replaced CQ monotherapy for treatment of malaria caused by <named-content content-type="genus-species">Plasmodium falciparum</named-content>. We measured the efficacy of the combination CQ/SP, and the prevalence of parasites carrying alleles associated with resistance to CQ or SP.</p></sec><sec id="st2"><title>Design:</title><p>We conducted a single-blind, randomised, controlled trial to compare the efficacy of CQ/SP to that of SP or CQ alone.</p></sec><sec id="st3"><title>Setting:</title><p>The study took place in the town of Farafenni and surrounding villages in the Gambia.</p></sec><sec id="st4"><title>Participants:</title><p>Participants were <bold>c</bold>hildren aged 12 mo to 10 y presenting as outpatients with uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> malaria.</p></sec><sec id="st5"><title>Interventions:</title><p>500 children were randomised to receive CQ, SP, or CQ/SP as supervised treatment and actively followed over 28 d.</p></sec><sec id="st6"><title>Outcome Measures:</title><p>Primary outcome was parasitaemia at any time during follow-up. Secondary outcomes were PCR-confirmed recrudescent infections among treatment failures, and clinical failure requiring rescue medication by day 28. Pretreatment parasite isolates from 161 patients were tested for the presence of resistance-associated genetic markers.</p></sec><sec id="st7"><title>Results:</title><p>The prevalence of parasitological failure by day 28 for the CQ group was 60.3%, compared to 17.6% for SP (odds ratio [OR], 0.106; 95% confidence interval [CI], 0.057–0.194; <italic>p</italic> < 0.001) and 13.9% for CQ/SP (OR versus CQ, 0.140; 95% CI, 0.078–0.250; <italic>p</italic> < 0.001). There was no difference between the SP and CQ/SP groups (OR, 1.324; 95% CI, 0.705–2.50). The projected prevalence of PCR-corrected treatment failure was 30.2, 6.06, and 3.94% in the CQ, SP, and CQ/SP groups, respectively. The <italic>pfdhfr</italic>-triple mutant and <italic>pfdhps</italic>-437G mutation were common, with prevalences of 67.4 and 51.2%, respectively. Pretreatment carriage of <italic>pfdhps-</italic>437G and of multidrug-resistant parasite genotypes was associated with treatment failure in the SP group, but not in the CQ or CQ/SP groups.</p></sec><sec id="st8"><title>Conclusions:</title><p>The combination of CQ/SP was an efficacious treatment for uncomplicated malaria in Gambian children in this study, but the frequent occurrence of multidrug-resistant parasites suggests that this observed efficacy is not sustainable.</p></sec>
|
<contrib contrib-type="author"><name><surname>Dunyo</surname><given-names>Samuel</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Ord</surname><given-names>Rosalynn</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Hallett</surname><given-names>Rachel</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Jawara</surname><given-names>Musa</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Walraven</surname><given-names>Gijs</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Mesa</surname><given-names>Eduardo</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Coleman</surname><given-names>Rosalind</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Sowe</surname><given-names>Maimuna</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Alexander</surname><given-names>Neal</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Targett</surname><given-names>Geoffrey A. T</given-names></name><xref ref-type="aff" rid="aff2">2</xref></contrib><contrib contrib-type="author"><name><surname>Pinder</surname><given-names>Margaret</given-names></name><xref ref-type="aff" rid="aff1">1</xref></contrib><contrib contrib-type="author"><name><surname>Sutherland</surname><given-names>Colin J</given-names></name><xref ref-type="aff" rid="aff2">2</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib>
|
PLoS Clinical Trials
|
<sec id="s1"><title>INTRODUCTION</title><p>Malaria control in Africa has until recently relied heavily on chemotherapy with chloroquine (CQ), a cheap and safe antimalarial drug [<xref rid="pctr-0010014-b001" ref-type="bibr">1</xref>,<xref rid="pctr-0010014-b002" ref-type="bibr">2</xref>]. CQ remains widely distributed and readily available even in the most remote villages in sub-Saharan Africa. In the Gambia, each village with a population of 400 or more has a trained village health worker who is provided with chloroquine and other basic drugs [<xref rid="pctr-0010014-b003" ref-type="bibr">3</xref>]. Chloroquine-resistant strains of <italic>Plasmodium falciparum,</italic> first observed in East Africa in 1987, have now been reported in all countries of tropical Africa [<xref rid="pctr-0010014-b004" ref-type="bibr">4</xref>].</p><p>Studies between 1994 and 2000 found that 65–73% of Gambian children treated with chloroquine were parasitaemic at some point over 28 days of follow-up [<xref rid="pctr-0010014-b005" ref-type="bibr">5</xref>,<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>], demonstrating that as in much of Africa, CQ had ceased to be a satisfactory first-line treatment for uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> malaria and an alternative was urgently required [<xref rid="pctr-0010014-b007" ref-type="bibr">7</xref>]. There have been a number of trials in sub-Saharan Africa to measure the efficacy, effectiveness, and impact on transmission of a variety of combination antimalarial regimens, including those incorporating a member of the artemisinin family [<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>,<xref rid="pctr-0010014-b008" ref-type="bibr">8</xref>–<xref rid="pctr-0010014-b014" ref-type="bibr">14</xref>]. Although the use of these newer combinations as first-line treatment for malaria is being adopted as policy in many African countries, interim solutions have been urgently sought by some prior to the full-scale introduction of artemisinin-based combination therapy.</p><p>In the Gambia, the combination of CQ and sulfadoxine-pyrimethamine (SP) has been shown to be a more effective symptomatic treatment than SP alone for malaria [<xref rid="pctr-0010014-b008" ref-type="bibr">8</xref>]. CQ/SP has therefore been used increasingly in the face of spreading resistance to CQ, and the combination was officially adopted as an affordable alternative frontline therapy in 2004. However, few reliable data are available on the efficacy of the combination CQ/SP in the Gambia, and although recent work has described the contribution of mutations in the CQ-resistance-associated loci <italic>pfcrt</italic> and <italic>pfmdr1</italic> to CQ treatment failure, to enhanced transmission to mosquitoes, and to an excess of severe malaria among rural Gambian children (<xref rid="pctr-0010014-b015" ref-type="bibr">15</xref>–<xref rid="pctr-0010014-b017" ref-type="bibr">17</xref>), there are no recent Gambian studies examining the prevalence of mutations associated with SP resistance, in the <italic>pfdhfr</italic> and <italic>pfdhps</italic> genes, nor of their effect on treatment outcome.</p><p>This paper reports the result of a randomised controlled trial conducted in 2001 to evaluate the efficacy of CQ/SP compared to SP or CQ alone. We also measured the pretreatment prevalence of parasites carrying resistance-associated alleles of four genes previously implicated in treatment failure of CQ and SP, and evaluate their contribution to therapeutic outcome.</p></sec><sec id="s2"><title>METHODS</title><sec id="s2a"><title>Participants</title><p>The study took place from September to December 2001 in Farafenni, a rural town on the north bank of the Gambia River. It is located on the Senegal border about 170 km from the Atlantic coast. The town has been the site for clinical trials to determine the efficacy of antimalarial drug combinations with gametocyte carriage and transmission as major endpoints since 1998 [<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>,<xref rid="pctr-0010014-b010" ref-type="bibr">10</xref>,<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010014-b013" ref-type="bibr">13</xref>,<xref rid="pctr-0010014-b014" ref-type="bibr">14</xref>]. The climate is characteristic of the sub-Sahel with a short rainy season from mid-June to mid-October. Malaria is thus seasonal with most clinical episodes occurring during a limited period of 8 to 10 wk at the end of the rains.</p><p>Recruitment took place at the Maternal and Child Health clinic according to an established protocol [<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>]. Briefly, children 0.5–10 y of age living within 15–20 km radius of Farafenni who presented to the Maternal and Child Health clinic or the General Hospital with history of fever and/or current fever (axillary temperature, ≥37.5 °C) and other symptoms suggestive of acute malaria infection, a carriage of <named-content content-type="genus-species">P. falciparum</named-content> parasites at a density ranging between 500 and 250,000 parasites/μl of blood, and a packed cell volume (PCV) ≥20% were enrolled in the study after obtaining the free and informed consent of their parents or guardians. Excluded from the study were children with anaemia (PCV, <20%), any signs or symptoms of severe malaria, inability to take drugs orally or any evidence of chronic disease, malnutrition or any other acute infection, including non-falciparum malaria. If there was evidence of treatment with any antimalarial drug within the past 2 wk, either from notes on the clinic card carried by children under 5 y of age, or after questioning of the caregiver, the child was excluded. The study protocol was reviewed and approved by the Medical Research Council/Gambian Government Joint Ethics Committee, and the Ethics Committee of the London School of Hygiene and Tropical Medicine.</p></sec><sec id="s2b"><title>Interventions</title><p>We performed a single-blind, randomised, controlled trial of oral treatment of uncomplicated falciparum malaria with CQ, SP, and the combination CQ/SP. A thorough clinical history (including demographic data) was taken, and a medical examination was performed by a study clinician on the day of screening, day 0. At screening and at each visit, body temperature was measured with a digital thermometer (Toshiba, Tokyo, Japan). The treatments used were CQ alone (25 mg of CQ base per kilogram of body weight over a 3-d period: 10 mg/kg on days 0 and 1; 5 mg/kg on day 2), SP alone given in a single dose (25 mg/kg sulphadoxine/1.25 mg/kg pyrimethamine), or the combination of CQ/SP. The group sizes were 130:180:200, respectively. This ratio was derived a priori as likely to ensure sufficient power for both efficacy and transmission endpoints of the trial, taking into account that children randomised to receive SP were further divided into three groups according to the day of the week on which they were recruited, as part of the accompanying study of post-treatment transmission [<xref rid="pctr-0010014-b018" ref-type="bibr">18</xref>]. These three SP-treated groups differed in their follow-up schedule, and only SP-treated patients recruited on Thursday or Friday received identical follow-up to children receiving CQ or CQ/SP. Therefore, only these patients are included in the analysis of clinical efficacy (<xref ref-type="fig" rid="pctr-0010014-g001">Figure 1</xref>). Treatment allocation and prescription of the required medication were performed by a study clinician, and medicines were administered by a study nurse. Treatment was performed in a separate room but was not deliberately obscured from the view of the study team. Older children chewed or swallowed the tablets whole with water. For younger children, the tablets were crushed together in a cup and mixed with water before administration. Subsequent doses, for those in chloroquine groups, were supervised by trained field assistants in the patients' homes. All treated children were observed for 30 min, and any child who vomited was administered a replacement dose. Children who vomited both the initial and the repeat medication were excluded from the study and either given parenteral CQ, which at the time of the study was standard care in the Gambia for children with uncomplicated malaria who could not retain oral medication, or admitted to the paediatric ward of Farafenni Hospital.</p><fig id="pctr-0010014-g001" position="float"><label>Figure 1</label><caption><title>CONSORT Flowchart</title><p>A total of 1,366 children was screened, and 500 were randomised. Children randomised to receive SP on Thursday or Friday (<italic>N</italic> = 79), were scheduled for standard follow-up and thus fulfilled the criteria for efficacy evaluation. Follow-up data are shown for these 79, but not for the other SP-treated children. Shaded boxes represent gametocyte screening days [<xref rid="pctr-0010014-b018" ref-type="bibr">18</xref>].</p></caption><graphic xlink:href="pctr.0010014.g001"/></fig><p>In addition to the study drugs, oral paracetamol (10 mg/kg, three times daily) was given together with the study medication in the clinic, and additional doses (to last for 3 d) were given to parents/guardians for administration to their children at home.</p></sec><sec id="s2c"><title>Objectives</title><p>The principal objective was to assess the efficacy of SP and CQ/SP for the treatment of uncomplicated falciparum malaria in children, compared to the efficacy of CQ monotherapy. A secondary objective was to estimate the prevalence of alleles of parasite genes associated with resistance to CQ and SP.</p></sec><sec id="s2d"><title>Outcomes</title><p>The primary outcome was parasitaemia at any time during follow-up. Secondary outcomes were PCR-corrected group estimates of treatment failure, and clinical failure requiring rescue medication by day 28. Pretreatment parasite isolates from 161 patients were tested for the presence of resistance-associated genetic markers.</p></sec><sec id="s2e"><title>Sample Size</title><p>The sample size required in each treatment group was estimated with transmission endpoints in mind. We expected to enroll approximately 500 patients, based on data from 1998–2000. In our previous studies, the observed prevalence of gametocyte carriage 7 d after treatment with SP was >70% (excluding those who had gametocytes on day 0), and 40% after treatment with CQ [<xref rid="pctr-0010014-b013" ref-type="bibr">13</xref>]. We therefore expected that the prevalence of gametocytes after SP/ CQ would be at least 28% (i.e., 40% × 70%). Based on these figures, and to ensure that at least 30 mosquito feeds were performed in each treatment group, we set out to enroll (a) 200 subjects in the CQ + SP group, (b) 50, 60, and 70 subjects in the three groups receiving SP and being screened for gametocytes on days 7, 10, and 14, respectively, and (c) 130 subjects in the CQ group. All CQ- and CQ/SP-treated children were to be screened for gametocytes on day 7 only. This sample size was expected to provide sufficient power at the 90% confidence level to detect a 10-fold difference in infectiousness at day 7 between gametocyte-positive children treated with SP alone, and those treated with SP + CQ [<xref rid="pctr-0010014-b018" ref-type="bibr">18</xref>]. This was not expected to provide sufficient power to detect any difference in parasitological efficacy between SP and CQ/SP, which was estimated a priori to be approximately 90% for both regimens.</p></sec><sec id="s2f"><title>Randomisation—Sequence Generation</title><p>Eligible patients were randomly assigned to treatment using a predetermined computer generated randomisation list. The list was generated by a statistician not otherwise engaged in the study.</p></sec><sec id="s2g"><title>Randomisation—Allocation Concealment</title><p>Allocation was not concealed, but taken from a list with 550 allocations on it, in the order of recruitment. Access to this list was restricted.</p></sec><sec id="s2h"><title>Randomisation—Implementation</title><p>Patients underwent initial clinical and parasitological screening to confirm a diagnosis of malaria before being asked for informed consent by a field worker able to speak Wolof, Mandinka, and Fulani. This person did not have access to the allocation list. If consent was given, a clinical examination, treatment allocation, and prescription of the required medication were performed in a separate room by a study clinician who had access to the randomisation list.</p></sec><sec id="s2i"><title>Randomisation—Blinding</title><p>Only the study clinician responsible for initial recruitment and treatment, and the field assistants visiting the children during follow-up, had knowledge of drug allocations. Field assistants did not have access to the full allocation list, but knew the treatment group of children in their care. All laboratory staff, microscopists, entomologists, and the principal investigator were blinded to treatment allocations.</p></sec><sec id="s2j"><title>Post-Treatment Follow-Up</title><p>Field assistants visited children at home on days 1 and 2 to supervise treatment (to those in CQ groups) and on days 3, 14, and 28 to enquire from the caretaker about the child's condition of health and to collect finger prick samples that provided thick blood films for microscopy and filter paper blood spots for parasite typing by PCR. If on any of these visits the caretaker was concerned about the child's health, he or she was asked to attend the clinic as soon as possible. Parents/guardians were also requested to bring their children to the Maternal and Child Health clinic at any time in the event of clinical aggravation. If a parent or guardian reported that a child was unwell, the child was examined by a study clinician. If the child had parasitaemia and fever (axillary temperature, ≥37.5 °C) or recent history of fever and did not have other conditions that could explain the symptoms, and fewer than 29 d had elapsed since recruitment, the child was considered a clinical treatment failure and given rescue treatment as follows: those in the CQ treatment group received SP, and those in the other two groups received a standard course of oral quinine. Children presenting with persistent symptoms of malaria within 3 d of enrollment with >90% reduction in parasitaemia were not considered to be clinical treatment failures, and did not receive rescue treatment. They were monitored until complete recovery.</p><p>On day 7 after treatment, the children were collected from their homes and taken to the Medical Research Council Field Station in Farafenni where they were clinically examined and finger prick blood samples were obtained for thick blood film preparation and PCV estimation. The blood films were stained with Field's stain and read immediately for malaria parasites. Some children identified as gametocyte carriers at this time contributed to transmission experiments described elsewhere [<xref rid="pctr-0010014-b018" ref-type="bibr">18</xref>].</p></sec><sec id="s2k"><title>Blood Sampling and Laboratory Measurements</title><p>At screening and at day 3, 7, 14, and 28 follow-up visits, blood samples were obtained by finger prick for thick blood smears for malaria microscopy. At day 7, or any time that the child had symptoms consistent with clinical malaria, two thick smears were made; the first was stained with Field's stain and read immediately. The second was dried overnight, stained with Giemsa, and read later by two experienced laboratory assistants to provide definitive parasite counts. One hundred high-power fields were read before a slide was declared negative. At screening and at day 7, blood samples were collected in heparinised capillary tubes and spun using a micro-haematocrit centrifuge (Hawksley, West Sussex, United Kingdom) for PCV determination. Staff performing all laboratory investigations were blinded to treatment regimens.</p></sec><sec id="s2l"><title>Molecular Genotyping</title><p>
<named-content content-type="genus-species">P. falciparum</named-content> genotypes circulating in children with detectable asexual parasitaemia during posttreatment follow-up were compared with those present in the same child prior to treatment. DNA was extracted from dried blood spots using a Chelex-based method [<xref rid="pctr-0010014-b019" ref-type="bibr">19</xref>]. Alleles of the polymorphic locus <italic>pfmsp2</italic> were compared between pretreatment and posttreatment parasite isolates by PCR [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010014-b020" ref-type="bibr">20</xref>]. The procedure of Cattamanchi et al. [<xref rid="pctr-0010014-b021" ref-type="bibr">21</xref>] was followed, in that indeterminate samples in which a majority of novel bands appeared in the posttreatment infection were scored as new infections.</p><p>Resistance-associated loci encoding amino acid positions 72–76 in <italic>pfcrt</italic> (sensitive allele, Cys-Val-Met-Asn-Lys [CVMNK]; resistant allele, Cys-Val-Ile-Glu-Thr [CVIET]), position 86 in <italic>pfmdr1</italic> (resistant allele, Asn [N]; sensitive allele, Tyr [Y]), positions 51, 59, and 108 in <italic>pfdhfr</italic> (resistant alleles, Ile, Arg, and Asn [I, R, N], respectively; sensitive alleles, Asn, Cys, and Ser [N, C, S], respectively), and positions 437 and 540 in <italic>pfdhps</italic> (sensitive alleles, Ala and Lys [A, K], respectively; resistant alleles, Gly and Glu [G, E], respectively) were identified as previously described [<xref rid="pctr-0010014-b015" ref-type="bibr">15</xref>,<xref rid="pctr-0010014-b022" ref-type="bibr">22</xref>,<xref rid="pctr-0010014-b023" ref-type="bibr">23</xref>]. Briefly, fluorescent-labeled oligonucleotide probes specific for each allele of interest were hybridised to PCR products spotted in 12 × 8 arrays on nylon membranes, and hybridisation signals were detected by chemiluminescence.</p></sec><sec id="s2m"><title>Statistical Methods</title><p>All data were double entered and verified using Epi-Info, version 6 (Centers for Disease Control and Prevention, Atlanta, Georgia, United States) and transferred to Stata 7.0 (Stata Corporation, College Station, Texas, United States) for statistical analysis. All children with follow-up data were included in the primary analysis. Any child presenting with danger signs on days 0–3 in the presence of malaria parasites or fever/history of fever plus parasitaemia any time from days 4 to 28 was given rescue medication and treated as a clinical therapeutic failure. Proportions were compared using the χ<sup>2</sup> statistic for parasitological data (denominators >50 in each case) or using Fisher exact test for clinical failure data and molecular genotyping data, because in both these analyses expected proportions in some cells of 2 × 2 tables were ≤5. Asexual parasite density was compared between groups using the ratio of arithmetic means. This was done by fitting generalized linear models with the negative binomial distribution family and logarithmic link function of the (arithmetic) mean parasite density as previously described [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>]. Confidence intervals around projected relative risk estimates were calculated from the variance of the log relative risk as described [<xref rid="pctr-0010014-b024" ref-type="bibr">24</xref>], after modification to allow for each risk being the product of two proportions.</p></sec></sec><sec id="s3"><title>RESULTS</title><sec id="s3a"><title>Recruitment and Participant Flow</title><p>A total of 1,368 children was screened. Two records were without parasitaemia data and were therefore excluded from further analysis. Of the 1,366 children with complete information on malaria infection, 844 (61.8%) were positive for malaria parasites according to microscopy at the recruiting clinic, and were therefore eligible for enrolment into the study. Of these, 500 were enrolled into the trial; 126, 181, and 193 in the CQ alone, SP alone, and CQ/SP groups, respectively. The most common reasons for ineligibility for enrollment were low density parasitaemia (<500/μl of blood) (<italic>n</italic> = 114), refusal to give parental consent (<italic>n</italic> = 40), anemia (PCV, <20%) (<italic>n</italic> = 34), residence outside the clinic catchment area (<italic>n</italic> = 24), high density parasitaemia (≥250,000 parasites/μl of blood), and/or other severe signs of malaria (<italic>n</italic> = 18) and presence of concomitant diseases and other reasons (<italic>n</italic> = 51). A trial profile is shown in <xref ref-type="fig" rid="pctr-0010014-g001">Figure 1</xref>. Loss to follow-up by day 28 was 25.4, 31.5, and 32.6% in the CQ, SP, and CQ/SP treatment groups, respectively.</p></sec><sec id="s3b"><title>Baseline Data</title><p>There were no differences in baseline demographic characteristics or malaria infection between the treatment groups (<xref ref-type="table" rid="pctr-0010014-t001">Table 1</xref>).</p><table-wrap id="pctr-0010014-t001" content-type="1col" position="float"><label>Table 1</label><caption><p>Baseline Demographic and Clinical Characteristics of Patients Randomised to CQ Alone, SP Alone, or CQ/SP Treatment</p></caption><graphic xlink:href="pctr.0010014.t001"/></table-wrap></sec><sec id="s3c"><title>Outcomes and Estimation</title><sec id="s3c1"><title>Clinical outcomes and adverse advents.</title><p>The frequency of posttreatment, parasitologically confirmed clinical malaria (clinical failure) among evaluable study patients in each treatment group is shown in <xref ref-type="table" rid="pctr-0010014-t002">Table 2</xref>. By day 28, as expected from previous studies, more children in the CQ treatment group had experienced an episode of clinical malaria posttreatment than in either of the other groups. There was weak evidence that clinical failure was more common among SP-treated children than among those receiving CQ/SP (OR, 2.68; 95% CI, 0.764–9.34). This was apparently due to persistence of symptoms during the first 3 d (<xref ref-type="table" rid="pctr-0010014-t002">Table 2</xref>), consistent with the findings of Bojang et al. [<xref rid="pctr-0010014-b008" ref-type="bibr">8</xref>]. However, clinical endpoints were not the main focus of the trial, and we did not collect sufficient data to make a valid comparison with earlier results [<xref rid="pctr-0010014-b008" ref-type="bibr">8</xref>]. There were no serious adverse events reported during follow-up, and no deaths among the 500 recruited participants during the study period.</p><table-wrap id="pctr-0010014-t002" content-type="1col" position="float"><label>Table 2</label><caption><p>Clinical Failure Rates in Children Treated with CQ, SP, or CQ/SP for Uncomplicated <named-content content-type="genus-species">P. falciparum</named-content> Malaria</p></caption><graphic xlink:href="pctr.0010014.t002"/></table-wrap></sec><sec id="s3c2"><title>Parasitological outcomes.</title><p>The point prevalence of parasitological treatment failure for each treatment group at each day of follow-up is shown in <xref ref-type="fig" rid="pctr-0010014-g002">Figure 2</xref>. This is a per protocol analysis. At each time point, parasitological failures were most common among CQ-treated children, with an uncorrected cumulative treatment failure rate over 28 d of 60.3% for CQ, compared to 17.6% for SP (OR, 0.106; 95% CI, 0.057–0.194; <italic>p</italic> < 0.001) and 13.9% for CQ/SP (OR compared to CQ, 0.140; 95% CI, 0.078–0.250; <italic>p</italic> < 0.001). To compare the parasitological outcomes of each group in more detail, the ratio between arithmetic mean parasite densities at each day of active follow-up was calculated, and the statistical significance of this difference tested by negative binomial regression, including those with zero parasites (<xref ref-type="table" rid="pctr-0010014-t003">Table 3</xref>). We have previously used this method to compare gametocyte densities among treatment groups [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>,<xref rid="pctr-0010014-b014" ref-type="bibr">14</xref>]. We found that the parasitological efficacy of SP was significantly higher than that of CQ at days 7, 14, and 28, and that of CQ/SP was significantly higher than that of CQ at days 3, 7, and 14 (<xref ref-type="table" rid="pctr-0010014-t003">Table 3</xref>). No significant difference in mean parasite density was found between the SP and CQ/SP groups at any point (data not shown).</p><fig id="pctr-0010014-g002" position="float"><label>Figure 2</label><caption><title>Point Prevalence of <named-content content-type="genus-species">Plasmodium falciparum</named-content> Asexual Parasitaemia 3, 7, 14, and 28 Days after Treatment of Children with CQ, SP, or CQ/SP</title><p>Error bars represent the upper 95% confidence limit of the proportion. Denominators for these data are (for CQ, SP, and CQ/SP groups, respectively): day 3, 111, 160, 168; day 7, 107, 52, 152; day 14, 98, 134, 143; and day 28, 94, 124, 130.</p></caption><graphic xlink:href="pctr.0010014.g002"/></fig><table-wrap id="pctr-0010014-t003" content-type="1col" position="float"><label>Table 3</label><caption><p>Parasitological Benefit of SP and CQ/SP Treatment Compared to CQ Treatment at Each Point of Follow-Up</p></caption><graphic xlink:href="pctr.0010014.t003"/></table-wrap><fig id="pctr-0010014-g003" position="float"><label>Figure 3</label><caption><title>Prevalence of Resistance-Associated Loci among 90 Pretreatment Parasite Isolates</title><p>Genotypes at seven loci in four <named-content content-type="genus-species">P. falciparum</named-content> genes were determined, as was the prevalence of the <italic>pfdhfr</italic> triple-mutant IRN, and of the putative multidrug-resistant genotype TYRG.</p></caption><graphic xlink:href="pctr.0010014.g003"/></fig></sec></sec><sec id="s3d"><title>Correction of Parasitological Failure Rates by Msp2 Genotyping</title><p>Alleles of <italic>msp2</italic> in pretreatment DNA samples were compared to those in posttreatment DNA samples to distinguish recrudescent parasites from parasites newly emergent from the liver during follow-up, as previously described [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>]. PCR was performed for paired isolates from 20 patients with posttreatment parasitaemia from each treatment group, and interpretable data obtained for 16, 15, and 17 pairs from the CQ, SP, and CQ/SP groups, respectively. Results are tabulated in <xref ref-type="table" rid="pctr-0010014-t004">Table 4</xref>, and used to estimate the true failure rate in each group by extrapolation. We found a corrected failure rate for CQ of 30.2%, for SP of 6.06%, and for CQ/SP of 3.94%. Using these estimates, we also projected a “virtual” relative risk of posttreatment recrudescence in each treatment group (<xref ref-type="table" rid="pctr-0010014-t004">Table 4</xref>). Both treatment groups containing SP were substantially better than CQ monotherapy in this analysis, whereas there was no significant benefit identified for CQ/SP over SP alone.</p><table-wrap id="pctr-0010014-t004" content-type="2col" position="float"><label>Table 4</label><caption><p>Cumulative 28-d Parasitological Failure Rates Corrected by <italic>msp2</italic> Genotyping</p></caption><graphic xlink:href="pctr.0010014.t004"/></table-wrap></sec><sec id="s3e"><title>Baseline Prevalence of Resistance-Associated Alleles</title><p>The prevalence of drug resistance-associated mutations at seven loci in four genes was measured among 90 day 0 samples randomly selected from among all trial participants (<xref ref-type="fig" rid="pctr-0010014-g003">Figure 3</xref>). The prevalence of the <italic>pfdhfr</italic> triple mutant 51-I, 59-R, 108-N (IRN) was also measured, as was prevalence of the combination genotype <italic>pfcrt-</italic>76T, <italic>pfmdr1</italic>-86Y, <italic>pfdhfr-</italic>59R, and <italic>pfdhps-</italic>437G (TYRG), a possible multidrug-resistant genotype. Combination genotypes IRN and TYRG were assigned to infections in which each allele was present. However, in some such individuals, wild-type alleles were also present at some of the loci in the combination, and so it is possible that IRN or TYRG did not occur as a single haplotype. Of the 60 patients designated as IRN, 49 (81.7%) unequivocally harboured true triple-mutant haplotypes, whereas the other 11 patients were of mixed genotype at two or three of the <italic>dhfr</italic> resistance-associated loci. Of the 17 patients who harboured parasites carrying the mutations <italic>pfcrt-</italic>76T, <italic>pfmdr1</italic>-86Y, <italic>pfdhfr-</italic>59R, and <italic>pfdhps-</italic>437G, the resistant allele alone was detected at three or four of these loci in 11 cases (64.7%). Thus, we are certain that parasites of the haplotype TYRG must occur in these infections. Mixed sensitive and resistant alleles were detected at two loci in four further isolates, and at three and all four loci in the remaining two isolates, respectively. Thus, in these latter six isolates, the haplotype TYRG may occur, but this is not certain. All were retained in subsequent analysis of associations with treatment failure.</p></sec><sec id="s3f"><title>Resistance-Associated Alleles and Treatment Failures</title><p>The risk of treatment failure associated with resistant genotypes at presentation was investigated by examining resistance-associated loci in pretreatment isolates from 100 patients who subsequently failed treatment. These comprised 44 children who received CQ, 31 who received SP, and 25 who received CQ/SP, of which four, three, and three children, respectively, required rescue treatment at the time of failure due to recurrent or persistent clinical signs of malaria. The prevalence of each marker in these pretreatment samples was compared to the prevalence among 61 isolates from children who were successfully treated. These children, 7, 22, and 32, respectively, from the CQ, SP, and CQ/SP groups, were among the 90 randomly selected isolates described in <xref ref-type="fig" rid="pctr-0010014-g003">Figure 3</xref>. Because 29 of these children subsequently failed treatment, only 61 were used for the purposes of estimating the odds of failure associated with each pretreatment genotype. A major purpose of these exploratory analyses was to identify a haplotype adequately representing “multidrug resistance” that could be defined at a single locus for each gene and so did not require using data at all seven loci. Estimates of the association between treatment failure and both single locus and selected multilocus genotypes is presented in <xref ref-type="table" rid="pctr-0010014-t005">Table 5</xref>. The only statistically significant association with treatment failure for any single locus was in the SP treatment group, where children presenting with the <italic>pfdhps-</italic>437G allele were more likely to fail treatment than those carrying the wild-type 437A allele (<xref ref-type="table" rid="pctr-0010014-t005">Table 5</xref>). Other weak associations could be discerned in the data, but given the multiple testing performed, these findings were not considered reliable. The combined dual locus genotype TY defined at <italic>pfcrt-</italic>76 and <italic>pfmdr1-</italic>86, previously associated with CQ treatment failure in this population [<xref rid="pctr-0010014-b015" ref-type="bibr">15</xref>], appeared to associate with treatment failure in the SP group only. The combined dual-locus genotype RG defined at <italic>pfdhfr-</italic>59 and <italic>pfdhps-</italic>437 was strongly associated with treatment failure, also in the SP group only, suggesting that the combination of these two alleles comprise the dominant SP-resistant haplotype in this study area. The four locus genotype combining both of these two allele genotypes, TYRG, was also strongly associated with treatment failure only in the SP treatment group (<xref ref-type="table" rid="pctr-0010014-t005">Table 5</xref>).</p><table-wrap id="pctr-0010014-t005" content-type="2col" position="float"><label>Table 5</label><caption><p>Odds of Subsequent Treatment Failure for Each Single-Locus Genotype, the <italic>pfdhfr</italic> Triple Mutant, and the Four-Locus Genotypes TFRG and TYRG at Day 0, in Each Treatment Group</p></caption><graphic xlink:href="pctr.0010014.t005"/></table-wrap></sec></sec><sec id="s4"><title>DISCUSSION</title><sec id="s4a"><title>Interpretation</title><p>In this study, we have examined the efficacy of the combination CQ/SP for treating uncomplicated falciparum malaria in Gambian children, compared to either CQ or SP alone. Pairwise comparisons of parasitological outcomes between regimens at each point of follow-up demonstrated a sustained benefit of treatment with SP or CQ/SP compared to CQ alone, but no difference in efficacy was found between the two SP-containing treatment groups at any time point. Cumulative failure rates, after correction by <italic>msp2</italic> genotyping to estimate rates of recrudescence, did not differ between the SP- and CQ/SP-treated groups of patients. The clinical failure rate of CQ monotherapy over 28 d was 22.6% in this study, and thus CQ is no longer useful as first-line treatment for malaria in children. The combination CQ/SP is an efficacious treatment for uncomplicated malaria in Gambian children, but the high prevalence of CQ-resistant parasites coupled with a small but measurable SP failure rate strongly suggest that this level of efficacy is unsustainable in our study area. These data were collected in late 2001, and continued use of CQ/SP as first-line treatment may have further eroded the efficacy of this combination since that time. However, subsequent studies show that crude parasitological efficacy for CQ/SP in Farafenni has remained good at 88% in 2002 [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>] and 86% in 2003 (P. Milligan and S. Dunyo, unpublished data). CQ/SP has been the first-line antimalarial regimen in the Gambia since 2004.</p><p>In 1995, Bojang et al. [<xref rid="pctr-0010014-b008" ref-type="bibr">8</xref>] measured the parasitological efficacy of SP and CQ/SP in children aged 1–10 y presenting to health facilities at Sibanor and Basse, the Gambia. These authors found the cumulative (uncorrected) parasitological failure rate at day 28 was 10% for SP (15 of 150) and 5% (7 of 141) for CQ/SP. These compare with uncorrected rates of 17.6% (29 of 165) and 13.9% (24 of 173) in the present study. Assuming the methodologies used are compatible, this suggests that children treated with SP in 2001 had a relative risk of parasitological failure of 1.31 (<italic>p</italic> = 0.053) compared to children treated with SP in 1995. Children treated with CQ/SP in 2001 had a relative risk of failure of 1.47 (<italic>p</italic> = 0.009). This exploratory comparison does suggest that the efficacy of both SP and CQ/SP has diminished in the 5-y period between these two studies. Bojang et al. did find that CQ/SP provided more rapid resolution of malaria symptoms than treatment with SP alone, but this was not tested in our study.</p><p>We used <italic>msp2</italic> genotyping in a subset of parasitological treatment failures to estimate the recrudescence rate in each treatment group. This method is prone to underestimate the true number of recrudescent infections in areas of stable transmission, because the PCR method will not pick up minority genotypes that may be present in the pretreatment sample at low abundance. A multiplicity of infection is seen in most infected patients in sub-Saharan Africa, and, among children enrolled in our study, there were on average four genotypes per infection prior to treatment (R. Ord and C. J. Sutherland, unpublished data). Minority genotypes that are resistant may flourish under drug selection and become dominant during follow-up. These will appear to be a “new” infection, but are actually recrudescent [<xref rid="pctr-0010014-b020" ref-type="bibr">20</xref>]. Thus PCR correction must be interpreted carefully, and for this reason we prefer to project a comparative estimate of the true failure rate among the treatment groups, rather than assign each posttreatment parasitaemia its own status as a recrudescent or a new infection.</p><p>In our study area, addition of SP to CQ provided a therapeutic efficacy of over 90% after PCR correction [6; this study], and is currently the recommended regimen in the Gambia. SP alone also provides parasitological efficacy of approximately 90% over 28 d [9; this study]. We were therefore surprised to find high prevalences of both the <italic>pfdhfr</italic> “IRN” triple mutant and the <italic>pfdhps-</italic>437G allele in our pretreatment population. The absence of the <italic>pfdhfr-</italic>164L and <italic>pfdhps-</italic>540E mutations in this study area (R. Hallett, R. Ord, and A. Randall, unpublished data) may explain why a reasonable level of SP efficacy is retained, although a lack of correlation between mutations and outcomes in individual patients is frequently observed in clinical trials of antimalarial treatment [<xref rid="pctr-0010014-b025" ref-type="bibr">25</xref>,<xref rid="pctr-0010014-b026" ref-type="bibr">26</xref>]. Nevertheless, the single marker <italic>pfdhps-</italic>437G was strongly associated with SP treatment failure in univariate analyses (<xref ref-type="table" rid="pctr-0010014-t005">Table 5</xref>).</p></sec><sec id="s4b"><title>Generalizability</title><p>Our study suffered from a high dropout rate due to logistic constraints exacerbated by national elections in the Gambia in October 2001, which meant that many staff and participants traveled away from the study area. Nevertheless, estimates of efficacy were consistent with other studies in the area in 1998 and 1999 (SP) [<xref rid="pctr-0010014-b013" ref-type="bibr">13</xref>], in 1998 and 2000 (CQ) [<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>,<xref rid="pctr-0010014-b013" ref-type="bibr">13</xref>], and in 2002 and 2003 (CQ/SP) [<xref rid="pctr-0010014-b011" ref-type="bibr">11</xref>]. Therefore, finding that SP and CQ/SP are efficacious treatments despite the common occurrence of resistance-associated genotypes in four genes of interest is likely to be generally true. Multidrug-resistant parasite genotypes have recently been described in a single case report from South Africa [<xref rid="pctr-0010014-b027" ref-type="bibr">27</xref>], but have not been widely investigated in African parasite populations, nor has the impact of such genotypes on treatment outcomes been measured in clinical trials. We found that <named-content content-type="genus-species">P. falciparum</named-content> parasites with the multidrug-resistant genotype TYRG, defined at the four loci <italic>pfcrt-</italic>76, <italic>pfmdr1-</italic>86, <italic>pfdhfr</italic>-59, and <italic>pfdhps-</italic>437 respectively, are relatively common among Gambian children presenting with uncomplicated malaria. The TYRG genotype was associated with therapeutic failure after treatment with SP, but not with the combination CQ/SP. Unexpectedly, there was also a weak association between SP treatment failure and the carriage of CQ resistance-associated alleles of <italic>pfcrt</italic> and <italic>pfmdr1</italic>. This may reflect the relatively low power of our study, and the exploratory nature of our analyses, which did not correct for multiple testing. Furthermore, the high prevalence of <italic>pfcrt-</italic>76T in the parasite population is likely to have masked associations between CQ-resistance loci and treatment outcomes. Nevertheless, these preliminary results do demonstrate the need for carefully designed studies to measure the contribution of multidrug-resistant parasites to inadequate treatment of uncomplicated malaria in African children as combination treatments become more widely deployed. Inadequate treatment is likely to increase risk of progression to severe disease, particularly severe malarial anaemia [<xref rid="pctr-0010014-b017" ref-type="bibr">17</xref>], and thus short-term gains in terms of improved treatment efficacy achieved with combinations such as CQ/SP and amodiaquine/SP may be quickly eroded if multidrug-resistant genotypes enjoy a selective advantage in the treated host. In the accompanying paper, we investigate the nature of that selective advantage in <named-content content-type="genus-species">Anopheles gambiae</named-content> mosquitoes experimentally fed on gametocytes from children who had received the CQ/SP combination [<xref rid="pctr-0010014-b018" ref-type="bibr">18</xref>]. The results suggest that children harbouring multidrug-resistant parasites are significantly more infectious to mosquitoes than other CQ/SP-treated children.</p></sec><sec id="s4c"><title>Overall Evidence</title><p>A recent systematic review found a poor evidence base for the therapeutic efficacy of CQ/SP [<xref rid="pctr-0010014-b028" ref-type="bibr">28</xref>], yet this drug combination has been first-line therapy for uncomplicated malaria in the Gambia since 2004. We have identified multidrug-resistant parasite genotypes of <named-content content-type="genus-species">P. falciparum</named-content> carrying alleles implicated in resistance to both CQ and SP, and found these to be common in our study area in 2001. Nevertheless, these parasites did not substantially challenge the therapeutic efficacy of SP or CQ/SP in this study, which were found to have 82.4 and 86.1% efficacy against recurrent parasitaemia, respectively. The prevalence of these genotypes suggests they are advantageous to the parasite. Therefore, continued use of CQ/SP may favour an increase in the prevalence of SP resistance-associated alleles, and should the absent <italic>pfdhps-</italic>540E mutation be introduced to the population the efficacy of the combination may then drastically decline [<xref rid="pctr-0010014-b025" ref-type="bibr">25</xref>]. The risk of this occurring is heightened by the prevalence of CQ-R parasites in the Gambia [15–17; this study], such that SP has been added to a drug that is already failing. It has been demonstrated that the addition of artesunate to CQ produces a poor therapeutic combination [<xref rid="pctr-0010014-b006" ref-type="bibr">6</xref>]. A possible interim solution in the Gambia for the period leading up to implementation of ACTs may be amodiaquine plus SP, a combination found to work very well in other settings where CQ resistance is high but amodiaquine remains efficacious [<xref rid="pctr-0010014-b029" ref-type="bibr">29</xref>].</p></sec></sec><sec sec-type="supplementary-material" id="s5"><title>SUPPORTING INFORMATION</title><supplementary-material content-type="local-data" id="pctr-0010014-sd001"><label>CONSORT Checklist</label><media xlink:href="pctr.0010014.sd001.doc"><caption><p>Click here for additional data file.</p></caption></media><p>(50 KB DOC)</p></supplementary-material><supplementary-material content-type="local-data" id="pctr-0010014-sd002"><label>Trial Protocol</label><media xlink:href="pctr.0010014.sd002.doc"><caption><p>Click here for additional data file.</p></caption></media><p>(88 KB DOC)</p></supplementary-material></sec>
|
Investigating the basis of substrate recognition in the pC221 relaxosome
|
<p>The nicking of the origin of transfer (<italic>oriT</italic>) is an essential initial step in the conjugative mobilization of plasmid DNA. In the case of staphylococcal plasmid pC221, nicking by the plasmid-specific MobA relaxase is facilitated by the DNA-binding accessory protein MobC; however, the role of MobC in this process is currently unknown. In this study, the site of MobC binding was determined by DNase I footprinting. MobC interacts with <italic>oriT</italic> DNA at two directly repeated 9 bp sequences, <italic>mcb1</italic> and <italic>mcb2</italic>, upstream of the <italic>oriT nic</italic> site, and additionally at a third, degenerate repeat within the <italic>mobC</italic> gene, <italic>mcb3</italic>. The binding activity of the conserved sequences was confirmed indirectly by competitive electrophoretic mobility shift assays and directly by Surface Plasmon Resonance studies. Mutation at <italic>mcb2</italic> abolished detectable nicking activity, suggesting that binding of this site by MobC is a prerequisite for nicking by MobA. Sequential site-directed mutagenesis of each binding site in pC221 has demonstrated that all three are required for mobilization. The MobA relaxase, while unable to bind to <italic>oriT</italic> DNA alone, was found to associate with a MobC–<italic>oriT</italic> complex and alter the MobC binding profile in a region between <italic>mcb2</italic> and the <italic>nic</italic> site. Mutagenesis of <italic>oriT</italic> in this region defines a 7 bp sequence, <italic>sra</italic>, which was essential for nicking by MobA. Exchange of four divergent bases between the <italic>sra</italic> of pC221 and the related plasmid pC223 was sufficient to swap their substrate identity in a MobA-specific nicking assay. Based on these observations we propose a model of layered specificity in the assembly of pC221-family relaxosomes, whereby a common MobC:<italic>mcb</italic> complex presents the <italic>oriT</italic> substrate, which is then nicked only by the cognate MobA.</p>
|
<contrib contrib-type="author"><name><surname>Caryl</surname><given-names>Jamie A</given-names></name></contrib><contrib contrib-type="author"><name><surname>Thomas</surname><given-names>Christopher D</given-names></name><xref ref-type="corresp" rid="cor1">*</xref></contrib><aff><institution>Astbury Centre for Structural Molecular Biology, Institute of Molecular and Cellular Biology, University of Leeds</institution><addr-line>Leeds LS2 9JT, UK</addr-line></aff>
|
Molecular Microbiology
|
<sec><title>Introduction</title><p>The initial events in mobilization of bacterial plasmids are characterized by two distinct cellular and molecular processes: DNA processing reactions, involved with preparing the plasmid for transfer; and utilization of the conjugative transfer apparatus involved in the mating-pair formation between donor and recipient cells.</p><p>The complex most frequently associated with conjugative DNA processing is the relaxosome, a nucleoprotein structure that includes a relaxase, which is a site- and strand-specific transesterase, typically associated with accessory proteins. These protein components interact with the origin of transfer (<italic>oriT</italic>) DNA and result in a phosphodiester bond cleavage from which transfer of a single-stranded DNA is initiated. The relaxosome thus provides a substrate for the membrane-spanning conjugative transfer apparatus.</p><p>Much of the current understanding of these processes has been derived from work on plasmids in Gram-negative bacteria (reviewed by <xref ref-type="bibr" rid="b42">Zechner <italic>et al</italic>., 2000</xref>). In contrast, the transfer regions of conjugative and mobilizable plasmids in Gram-positive bacteria have received comparatively little attention (<xref ref-type="bibr" rid="b13">Grohmann <italic>et al</italic>., 2003</xref>). Staphylococcal plasmid pC221 presents a simple system embodying the DNA processing reaction and is one of a number of closely related small (< 5 kb), mobilizable antibiotic-resistance plasmids found in the staphylococci, which also include pC223 (<xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>), pS194 (<xref ref-type="bibr" rid="b32">Projan <italic>et al</italic>., 1988</xref>), pRJ6 (<xref ref-type="bibr" rid="b25">Netz <italic>et al</italic>., 2001</xref>) and pRJ9 (<xref ref-type="bibr" rid="b26">Netz <italic>et al</italic>., 2002</xref>) of <italic>Staphylococcus aureus</italic>; and pIP1629, pIP1630 (<xref ref-type="bibr" rid="b3">Aubert <italic>et al</italic>., 1998</xref>) and pSK639 (<xref ref-type="bibr" rid="b2">Apisiridej <italic>et al</italic>., 1997</xref>) of <italic>Staphylococcus epidermidis</italic>.</p><p>Typically for a small plasmid, pC221 encodes only those functions required for DNA processing and contains four such loci: a <italic>cis</italic>-acting <italic>oriT</italic>; a DNA relaxase, MobA; and the accessory proteins MobB and MobC. It is thus mobilizable only in the presence of a compatible, co-resident conjugative plasmid such as pGO1, which encodes the required conjugative transfer apparatus (<xref ref-type="bibr" rid="b30">Projan and Archer, 1989</xref>). The minimal requirements for nicking have been defined using purified components <italic>in vitro</italic> (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). The MobA relaxase makes a reversible site- and strand-specific nick at a highly conserved sequence within <italic>oriT</italic> (5′-GCTTG′CCAAA-3′), whereupon it forms a phosphodiester bond with the 5′ end of the nicked strand. Nicking of supercoiled plasmid DNA substrate by MobA is dependent on the presence of the dimeric accessory protein MobC and divalent metal ions such as Mg<sup>2+</sup>; however, the process by which MobC mediates the cleavage by the relaxase is not yet understood.</p><p>While relaxase proteins themselves tend to share a core set of conserved functions, such as the presence of N-terminally encoded catalytic motifs and the requirement for a single-stranded DNA substrate and divalent metal cations, accessory proteins have more diverse lineages. This may reflect the origin of these proteins as components sequestered from the cellular background within which the respective plasmids have evolved (<xref ref-type="bibr" rid="b29">Parker <italic>et al</italic>., 2005</xref>). Almost all relaxases characterized to date are able to cleave single-stranded DNA substrates containing a nick site sequence in the absence of accessory proteins (<xref ref-type="bibr" rid="b42">Zechner <italic>et al</italic>., 2000</xref>); thus one role for the accessory proteins may be to provide a relaxase substrate with sufficient single-stranded character within a supercoiled DNA context. This may be achieved by enhancing strand denaturation at <italic>nic</italic>. In the case of plasmid R1162, the MobC protein extends MobA-induced DNA strand melting at <italic>oriT</italic> up to the nick site, potentially revealing a single-stranded substrate for cleavage (<xref ref-type="bibr" rid="b43">Zhang and Meyer, 1995</xref>; <xref ref-type="bibr" rid="b44">1997</xref>).</p><p>Alternatively, accessory proteins may act as recognition complexes to initiate relaxosome assembly, either by direct binding proximal to <italic>nic</italic> or indirectly by binding elsewhere and inducing significant bending at the cleavage site. In plasmid RP4, the TraJ accessory protein binds to the proximal arm of an inverted repeat upstream from the nick site (<xref ref-type="bibr" rid="b45">Ziegelin <italic>et al</italic>., 1989</xref>), and is believed to act as a recognition complex for the TraI relaxase. The same is true for the NikA accessory protein of plasmid R64: it has been proposed that binding of NikA bends the DNA at that point with the resulting conformational change playing a role in nicking (<xref ref-type="bibr" rid="b10">Furuya and Komano, 1997</xref>). The accessory proteins TraYp and IHF bind sequentially to <italic>oriT</italic> of plasmid F to facilitate binding by the TraIp relaxase. Such binding may present a suitable nicking substrate via possible protein–protein distortion of the DNA at the nick site (<xref ref-type="bibr" rid="b17">Howard <italic>et al</italic>., 1995</xref>; <xref ref-type="bibr" rid="b5">Byrd and Matson, 1997</xref>).</p><p>In addition, accessory proteins may also act in a regulatory manner: F plasmid TraYp binds a second site at the P<sub>Y</sub> promoter and is required for upregulation of <italic>tra</italic> gene expression (<xref ref-type="bibr" rid="b36">Silverman and Sholl, 1996</xref>). In R388, the TrwA accessory protein acts as both a transcriptional repressor of the <italic>trwABC</italic> operon and an enhancer of TrwC relaxase-mediated nicking (<xref ref-type="bibr" rid="b24">Moncalian <italic>et al</italic>., 1997</xref>). Both TraJ and TraK of RP4 repress their <italic>traJp</italic> and <italic>traKp</italic> promoters respectively (<xref ref-type="bibr" rid="b41">Zatyka <italic>et al</italic>., 1994</xref>).</p><p>The MobC proteins of pC221 and pC223 are interchangeable with regard to nicking, with the MobA proteins displaying specificity towards their respective substrates despite the highly conserved nature of the <italic>oriT</italic> region (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). The pC221 MobA relaxase represents one of three unrelated relaxases from Gram-positive bacterial plasmids that have been subjected to detailed characterization to date. The remaining two are TraA of the enterococcal plasmid pIP501 (<xref ref-type="bibr" rid="b39">Wang and Macrina, 1995</xref>; <xref ref-type="bibr" rid="b21">Kurenbach <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b18">Kopec <italic>et al</italic>., 2004</xref>), and MobM of the streptococcal rolling-circle replicon pMV158 (<xref ref-type="bibr" rid="b15">Guzman and Espinosa, 1997</xref>; <xref ref-type="bibr" rid="b12">Grohmann <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b1">de Antonio <italic>et al</italic>., 2004</xref>). These relaxases are grouped into the MOB<sub>Q</sub> and pMV158/Pre families respectively (<xref ref-type="bibr" rid="b8">Francia <italic>et al</italic>., 2004</xref>), which account for the majority of relaxases identified in Gram-positive bacterial plasmids (<xref ref-type="bibr" rid="b13">Grohmann <italic>et al</italic>., 2003</xref>). Indeed, the transfer functions of large conjugative staphylococcal plasmids pGO1 and pSK41 are also grouped within the MOB<sub>Q</sub> family of relaxases. In contrast, the pC221-family of relaxases are grouped within the MOB<sub>P</sub> family, most commonly associated with the plasmids of Gram-negative bacteria. This family also includes TraI of the conjugative plasmid RP4 and VirD2 of the agrobacterium pTi plasmid pTiC58 (<xref ref-type="bibr" rid="b8">Francia <italic>et al</italic>., 2004</xref>).</p><p>Of the mobilization functions of Gram-positive plasmids listed above, those of the pC221 family represent the only examples among the staphylococci for which detailed biochemical analysis has been reported. Elucidation of the mechanism of relaxosome formation and DNA processing in pC221 will provide a useful comparison to the well-known MOB<sub>P</sub> family member RP4, as well as providing the model for all pC221-related mobilization functions.</p><p>In this work we present data regarding the formation and specificity of the pC221 relaxosome, with the objective of differentiating the respective roles of the MobC and MobA proteins in substrate recognition. We report the use of DNase I footprinting to identify the sites of interaction of both MobC and MobA proteins with <italic>oriT</italic>, and confirm the sequence specificity of MobC through both electrophoretic mobility shift assay (EMSA) and Surface Plasmon Resonance (SPR) techniques. By mobilization studies <italic>in vivo</italic> and cleavage assays <italic>in vitro</italic> we have identified a minimal, functional <italic>oriT</italic> which itself contains a minimal nicking region. Site-directed mutagenesis of the latter has revealed elements essential for plasmid-specific cleavage by the cognate MobA relaxases of pC221 and pC223. Based on these observations we present a model of layered specificity that governs the assembly of the pC221 relaxosome.</p></sec><sec><title>Results</title><sec><title>MobC protein binds to a specific sequence at three positions within <italic>oriT</italic></title><p>The propensity of MobC to bind linear <italic>oriT</italic> DNA has been described previously (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>); a supercoiled DNA substrate is required for the nicking reaction. An alternative DNase I footprinting methodology that permitted the probing of supercoiled substrates was employed. The technique, described by <xref ref-type="bibr" rid="b11">Gralla (1985)</xref>, uses end-labelled oligonucleotide primers in a primer extension reaction to map strand discontinuities introduced within the target region by DNase I. As the DNase I cleavage products are not isotopically labelled, they can be can be re-probed a number of times with the same or different primers, or stored indefinitely. In addition, the ability to use large templates increases the site-specificity of protein–DNA interactions by providing abundant non-specific binding sites. Furthermore, under appropriate conditions the influence of secondary structure, promoted in the context of a supercoiled template, can be assessed. Finally, the method offers a high degree of sensitivity, even at relatively low protein concentrations, due to multiple rounds of primer extension on the template strand (<xref ref-type="bibr" rid="b11">Gralla, 1985</xref>).</p><p>Three control samples were used in all gels. First, a lane containing a sample of untreated supercoiled substrate DNA subjected to primer extension (<xref ref-type="fig" rid="fig01">Fig. 1</xref>, lane 1), thus revealing natural strand discontinuities resulting from the native DNA sequence/structure. Second, a lane containing a sample of supercoiled substrate DNA incubated only with the Mob protein prior to primer extension (<xref ref-type="fig" rid="fig01">Fig. 1</xref>, lane 2) was used to control for any strand discontinuities introduced by the proteins, or contaminants thereof, in the absence of DNase I cleavage. Finally, a lane that consists of a naked supercoiled substrate DNA treated with DNase I and subjected to primer extension, thus providing the control with which to assess potential footprints (<xref ref-type="fig" rid="fig01">Fig. 1</xref>, last lane). The observed strand discontinuities in all examples presented were independent of the observed patterns of DNase I protection and hypersensitivity observed in the protection experiments, thus the profiles observed can be attributed to the effect of Mob protein binding.</p><fig id="fig01" position="float"><label>Fig. 1</label><caption><p>Analysis of GSH–MobC binding at <italic>oriT</italic> and <italic>mobC</italic> by DNase I footprinting. DNase I footprinting was performed on both the upper (A) and lower (B) strands upstream of <italic>oriT</italic>, and (C) the upper strand of <italic>mobC</italic> located downstream of <italic>oriT</italic>. A supercoiled pC221<italic>cop903</italic> template was used and the resulting fragments probed by primer extension using primers NIC+98, NIC−178 and NIC+203 respectively. Lane identity is as follows: lane 1, DNA only (no DNase I); lane 2, GSH–MobC only (no DNase I); lanes 3–6, GSH–MobC (nM): 1.2, 12, 120, 372; lane 7, DNase I only. Sequencing ladder standards G, A, T, C are as indicated. Solid bars indicate protection, designated <italic>mcb1</italic>, <italic>mcb2</italic> and <italic>mcb3</italic>. Diminished and increased DNase I hypersensitivity are indicated by dotted bars and arrowheads respectively. The position of <italic>nic</italic> is indicated.</p></caption><graphic xlink:href="mmi060-1302-f1"/></fig><p>DNase I footprinting analysis was performed on pC221<italic>cop903</italic> DNA, incubated with varying concentrations of GSH–MobC (<xref ref-type="fig" rid="fig01">Fig. 1A and B</xref>). GSH–MobC is a thrombin-cleaved variant of the purified (His)<sub>6</sub>-tagged MobC, containing only the additional N-terminal residues GSH, thus reducing the possibility of any steric effects of the full tag on DNase I footprinting. The strands were probed by primer extension, initially using a primer that binds 98 nucleotides (nt) (NIC+98) upstream of <italic>nic</italic> in the upper strand (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>), and 178 nt (NIC−178) upstream from the <italic>nic</italic> complement in the lower strand (<xref ref-type="fig" rid="fig01">Fig. 1B</xref>). Footprinting on both strands revealed two corresponding regions of MobC–mediated DNase I protection. These regions, designated <italic>mcb1</italic> (<underline>M</underline>ob<underline>C b</underline>inding) and <italic>mcb2</italic> in the upper strand, were located 101 bp and 9 bp upstream from the <italic>oriT nic</italic> site and encompassed 16 bp and 17 bp of DNase I protected DNA respectively.</p><p>The regions of protection were comparable on both strands, not being significantly asymmetric, and were flanked by sites of DNase I hypersensitivity. Both <italic>mcb1</italic> and <italic>mcb2</italic> were protected with comparable affinity, apparent at 120 nM GSH–MobC, and clearer still at 372 nM. The ratio of MobC:DNA at 120 nM is also comparable to that used in plasmid nicking assays (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). At the higher concentration of MobC used, a pattern of hyper- and diminished-DNase I hypersensitivity with a periodicity of 10–12 bp was more evident between <italic>mcb1</italic> and <italic>mcb2</italic>.</p><p>A 9 bp sequence (5′-AGTGGCTAG-3′), conserved in both <italic>mcb1</italic> and <italic>mcb2</italic> was identified. Searches for this sequence elsewhere identified a match of 7 bp located 115 bp downstream from <italic>nic</italic>, designated <italic>mcb3</italic>, which is located within the <italic>mobC</italic> gene itself. The upper strand of this region was re-probed using an alternate primer (NIC+203) and although less distinct, demonstrated evidence of protection by MobC, flanked by DNase I hypersensitivity (<xref ref-type="fig" rid="fig01">Fig. 1C</xref>). Similar searches were performed in a selection of related plasmids, which identified similarly conserved sequences in comparable positions, with only a few instances of base substitutions (<xref ref-type="fig" rid="fig02">Fig. 2</xref>). In each of the seven plasmids aligned, <italic>mcb1</italic> was separated from the corresponding <italic>nic</italic> by 103–108 bp; however, in each case the <italic>mcb2</italic> was 15 bp upstream from <italic>nic</italic>.</p><fig id="fig02" position="float"><label>Fig. 2</label><caption><p>Sequence alignment of seven related <italic>oriT</italic> regions containing conserved motifs: pC221 [GenBank Accession No. X02166, as modified by <xref ref-type="bibr" rid="b37">Smith and Thomas (2004)</xref>: nucleotides 3019–3280], pC223 (GenBank Accession No. NC_005243: nucleotides 1737–1998), pS194 (GenBank Accession No. X06627: nucleotides 820–1080), pSE-12228-02 (GenBank Accession No. NC_005007: nucleotides 2973–3233), pSK639 (GenBank Accession No. U40259: nucleotides 246–417), pIP1629 (GenBank Accession No. AF045240: nucleotides 326–590), pRJ6 (GenBank Accession No. AF241888: nucleotides 744–476 on the complementary strand). Length (bp) of intervening sequences are indicated in brackets with percentage identity of those sequences below.</p></caption><graphic xlink:href="mmi060-1302-f2"/></fig><p>The juxtaposition of <italic>mcb1</italic> and <italic>mcb2</italic> with respect to nearby inverted repeats was not conserved (<xref ref-type="fig" rid="fig05">Fig. 5</xref>), with <italic>mcb1</italic> corresponding to the left arm of IR1 and <italic>mcb2</italic> to the loop region adjacent to the right arm of IR4 (<xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>). Thus we are unable to correlate the implied secondary structure with binding in the double-stranded context. An identical footprinting experiment using a linearized pC221<italic>cop903</italic> substrate displayed an identical protection and hypersensitivity profile to that observed with an ostensibly supercoiled substrate (data not presented). This, together with sequence conservation between the binding sites, suggests a sequence-based interaction; although involvement of these potential hairpins in other processes cannot be ruled out.</p><fig id="fig05" position="float"><label>Fig. 5</label><caption><p>Summary of footprinting data for the binding of MobC and MobA. Double-stranded DNA encompassing bases 3012–3285 (5′−3′) of pC221 (GenBank Accession No. X02166 as modified by <xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>). Sequences in black type correspond to the regions footprinted in <xref ref-type="fig" rid="fig01">Figs 1</xref> and <xref ref-type="fig" rid="fig04">4</xref>. Principle regions protected by MobC are indicated by solid boxes and labelled. The conserved sequence is highlighted in black. Those protected by MobA are indicated by dashed boxes. <italic>Nic</italic> is indicated by an arrow. Regions of diminished DNase I hypersensitivity upon binding of MobC are indicated by dashed bars. The predicted promoter for <italic>mobC</italic> is labelled (−35, −10) and the start codon indicated by a bent arrow. The positions of inverted repeats, based on <xref ref-type="bibr" rid="b37">Smith and Thomas (2004)</xref>, are indicated.</p></caption><graphic xlink:href="mmi060-1302-f5"/></fig><fig id="fig04" position="float"><label>Fig. 4</label><caption><p>Analysis of MobA binding at <italic>oriT</italic> by DNase I footprinting. DNase I footprinting was performed on the upper strand of <italic>oriT</italic> in the absence (A) and presence (B) of pre-bound GSH–MobC. A supercoiled pC221<italic>cop903</italic> template was used and fragments probed by primer extension; data obtained using primer NIC+98 is shown. Lane identity of (A): lane 1, DNA only/no DNase I; lane 2, MobAH<sub>6</sub> only/no DNase I; lanes 3–6, MobAH<sub>6</sub> (nM): 50, 100, 250, 450; lane 7, DNase I only. Lane identity of (B): lanes 1 and 2, as (A); lane 3, GSH–MobC<sup>221</sup> only: 120 nM; lanes 4–8, GSH–MobC (120 nM) + MobAH<sub>6</sub> (nM): 50, 100, 250, 450, 450; lane 9, DNase I only control. Sequencing ladder standards G, A, T, C are as indicated. Solid bars indicate protection, designated <italic>mcb1</italic>, <italic>mcb2</italic> and <italic>sra</italic>. MobA-induced DNase I hypersensitivity (<italic>HSA</italic>) is indicated by black arrowheads. The position of <italic>nic</italic> is indicated with an arrow.</p></caption><graphic xlink:href="mmi060-1302-f4"/></fig></sec><sec><title>The conserved 9 bp <italic>mcb</italic> sequence contains the MobC binding site</title><p>To differentiate the specific MobC binding site sequence from those flanking sequences protected due to steric hindrance, we performed competitive EMSA experiments. The conservation of the 9 bp sequence within the DNase I protected regions was suggestive of it being active in MobC binding, especially given the lack of conservation in the flanking regions of <italic>mcb1</italic> and <italic>mcb2</italic>. To confirm the extent to which the respective sequences were involved in binding by MobC, a 20 bp oligoduplex DNA (21P) containing the <italic>mcb2</italic> sequence and wild-type flanking DNA was synthesized and used as a competitor DNA by titrating it into a pre-formed MobC–<italic>oriT</italic> complex (<xref ref-type="fig" rid="fig03">Fig. 3A</xref>). The MobC–<italic>oriT</italic> complex represents isoform 3, the third and largest MobC–<italic>oriT</italic> complex observed before non-specific protein–DNA interactions are observed (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>); and which likely represents saturation of all the binding sites by MobC.</p><fig id="fig03" position="float"><label>Fig. 3</label><caption><p>A and B. Competition assay between pre-formed <italic>oriT</italic>–MobC complex and an increasing concentration of a synthetic oligoduplex DNA. Pre-formed <italic>oriT</italic>–MobC complex consisted of 10.75 nM BstXI/HindIII-digested pCER21T and 300 nM GSH–MobC. The complex was titrated with a 0–5-fold molar excess of 21P, an oligoduplex corresponding to the <italic>mcb2</italic> (sequence highlighted in light grey) and wild-type flanking sequence; or 21Pmut, an oligoduplex containing a mutated 9 bp <italic>mcb2</italic> (sequence highlighted in light grey), but maintaining wild-type flanking sequences. Short arrows indicate position of the free 20-mer oligoduplex DNA. Long arrows indicate free <italic>oriT</italic> DNA and the three <italic>oriT</italic>–MobC complex isoforms. Lane values refer to concentration of oligoduplex (nM).</p><p>C and D. SPR using an NTA-chip bound H<sub>6</sub>MobC<sup>221</sup> ligand and analytes 21P and 21Pmut at concentrations of 1000 nM (black), 100 nM (dark grey), buffer only control (light grey).</p></caption><graphic xlink:href="mmi060-1302-f3"/></fig><p>The 21P substrate demonstrated efficient competition with the 494 bp <italic>oriT</italic> fragment for binding by GSH–MobC, indicating that the sequence is active in MobC–<italic>oriT</italic> interaction. As a control, an identical assay was performed using a control competitor (21Pmut) a 20 bp oligoduplex in which the 9 bp consensus sequence was randomized, maintaining identical base composition; while retaining the <italic>mcb2</italic> flanking sequences (<xref ref-type="fig" rid="fig03">Fig. 3B</xref>). This 21Pmut substrate was unable to compete for binding by MobC, confirming that it is the conserved 9 bp sequence that contains crucial elements constituting the binding target.</p><p>In addition to these indirect studies, SPR was used to measure direct interaction between MobC and <italic>mcb</italic>. MobC was immobilized to the sensor chip surface via an NTA–His interaction and subjected to buffer containing competitor and control oligoduplexes as analytes. Binding was evident between 21P and MobC, with a K<sub>D</sub> in the order of 1.8 µM, but not with 21Pmut, thus confirming the observed effect of randomization at <italic>mcb2</italic> (<xref ref-type="fig" rid="fig03">Fig. 3C and D</xref>). The term <italic>mcb</italic> is therefore used to describe the 9 bp containing the MobC binding site.</p></sec><sec><title>MobA alters the footprint profile of MobC binding at <italic>mcb2</italic> and <italic>nic</italic></title><p>Unlike MobC, specific DNA binding by MobA was not apparent in earlier EMSA experiments (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). Given the requirement for MobC for nicking by MobA, we assessed the potential of MobA to protect <italic>oriT</italic> DNA from DNase I in the presence and absence of MobC. As with the earlier DNase I protection assays, control samples containing untreated supercoiled DNA in the presence and absence of Mob proteins confirmed that the observed pattern of strand discontinuities and hypersensitivity in experimental sample lanes were not due to innate DNA structure or contaminating endonucleases. A DNase I only treated supercoiled DNA was used to assess the extent to which MobA protected the substrate DNA. MobAH<sub>6</sub> was titrated against supercoiled pC221<italic>cop903</italic> and probed with primer NIC+98, which extends along the upper strand (<xref ref-type="fig" rid="fig04">Fig. 4A</xref>). No discernable protection or hypersensitivity was observed, relative to the DNase I only control.</p><p>To assess whether MobC facilitates binding or interaction of the MobA with substrate DNA, the titration was repeated using a supercoiled pC221<italic>cop903</italic> substrate pre-incubated with GSH–MobC (120 nM) (<xref ref-type="fig" rid="fig04">Fig. 4B</xref>). The purified template was probed as above and the resulting observations summarized (<xref ref-type="fig" rid="fig05">Fig. 5</xref>). A control lane containing MobC alone presented an identical profile to that previously observed (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>, also summarized in <xref ref-type="fig" rid="fig05">Fig. 5</xref>). With the addition of MobAH<sub>6</sub> the observed profile was modified, such that there was a loss of distinct MobC-induced DNase I hypersensitivity within a short (∼7 bp) region located between <italic>mcb2</italic> and the nick site (<italic>nic</italic>). This suggests that the addition of MobA may modulate the binding of the MobC proximal to <italic>nic</italic>, such that the MobC-induced DNase I hypersensitivity is lost. Alternatively, MobA may be associated with and positioned over this region, thus protecting these sites from DNase I cleavage. MobA also appeared to induce DNA bending or kinking downstream of the <italic>nic</italic> site, as indicated by the appearance of MobA-induced DNase I hypersensitivity 5 bp and 18 bp downstream from <italic>nic</italic>. The MobA-induced modifications were not apparent at the corresponding region downstream of <italic>mcb1</italic>, despite the presence of comparable MobC-induced hypersensitivity.</p></sec><sec><title>Derivation of a minimal target susceptible to nicking</title><p>One functional region defining <italic>oriT</italic> of pC221 corresponds to a cloned 514 bp fragment (roughly corresponding to the pC221 MboI–ClaI fragment) referred to as <italic>oriTA</italic>, which encompasses all three <italic>mcb</italic> sites, <italic>nic</italic> and the N-terminal coding region of <italic>mobC</italic>. When a comparable fragment is cloned in a pUC19-derivative (pCER21T) this construct can be nicked <italic>in vitro</italic> (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). When present within the shuttle vector, pCΔ<italic>oriTA</italic>, the fragment is seen to be nicked in whole cell lysate analysis (data not presented) and can be mobilized at a comparable frequency to the wild-type pC221 plasmid (<xref ref-type="fig" rid="fig06">Fig. 6A</xref>).</p><fig id="fig06" position="float"><label>Fig. 6</label><caption><p>A. Schematic diagram detailing the regions of <italic>oriT</italic> cloned and assessed for nicking and mobilization. The co-ordinates given are those of pC221 (GenBank Accession No. X02166 as modified by <xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>). The MobC binding sites (<italic>mcb</italic>) are indicated by numbered boxes. <italic>Nic</italic> is indicated by a small arrow and <italic>mobC</italic> by a bent arrow. <italic>oriTA</italic> (514 bp), <italic>oriTB</italic> (256 bp), <italic>oriTC</italic> (128 bp), <italic>oriTD</italic> (386 bp). Substrate nicking <italic>in vitro</italic> is indicated (+). The efficiency of mobilization of each construct is shown, and represents the number of target <italic>oriT</italic> molecules transferred for every transfer of pGO1 (calculated as described in the <italic>Experimental procedures</italic>). In the pC221Δ<italic>mcb</italic> mutants ‘X’ signifies mutation at that position.</p><p>B. Agarose gel electrophoresis of whole cell lysates showing <italic>in vivo</italic> nicking phenotype of pC221<italic>mcb</italic> mutants present in RN4220 donor cells also carrying pGO1. A control lane (–) indicates lysate of a pGO1-only donor.</p></caption><graphic xlink:href="mmi060-1302-f6"/></fig><p>To determine a minimal target susceptible to nicking, we cloned alternative regions of pC221 <italic>oriT</italic> DNA that encompassed either <italic>mcb2</italic> and <italic>mcb3</italic> (386 bp)<italic>, mcb1</italic> and <italic>mcb2</italic> (256 bp), or <italic>mcb2</italic> alone (139 bp and 77 bp). Constructs, prepared in the context of shuttle vector pCΔ (see <xref ref-type="table" rid="tbl1">Table 1</xref>), were initially tested using an <italic>in vitro</italic> nicking assay (<xref ref-type="fig" rid="fig06">Fig. 6A</xref>). In all cases the cloned <italic>oriT</italic> fragments were susceptible to nicking, thus the two smallest regions cloned demonstrated that a fragment containing only <italic>mcb2</italic> is sufficient for nicking.</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><p>Bacterial strains, plasmids and oligonucleotides</p></caption><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Strain, plasmid or oligonucleotide</th><th align="left" rowspan="1" colspan="1">Relevant characteristic(s)</th><th align="left" rowspan="1" colspan="1">Reference or source</th></tr></thead><tbody><tr><td colspan="3" rowspan="1">Strain</td></tr><tr><td align="left" rowspan="1" colspan="1"> <italic>E. coli</italic> DH5α</td><td align="left" rowspan="1" colspan="1">F<sup>–</sup>, φ80d<italic>lacZ</italic>ΔM15, <italic>endA</italic>1, <italic>recA</italic>1, <italic>HsdR</italic>17 (r<sub>k</sub><sup>–</sup>, m<sub>k</sub><sup>+</sup>), <italic>supE</italic>44, <italic>thi-1</italic>, <italic>GyrA</italic>96, <italic>relA</italic>1, Δ(<italic>lacZYA-argF</italic>)<italic>U169</italic>, λ<sup>−</sup></td><td align="left" rowspan="1" colspan="1">BRL</td></tr><tr><td align="left" rowspan="1" colspan="1"> <italic>S. aureus</italic> RN4220</td><td align="left" rowspan="1" colspan="1">Attenuated, <italic>mec</italic>A, TSSE<sup>–</sup>, <italic>rsbU</italic><sup>–</sup>, <italic>agr</italic><sup>–</sup>, restriction deficient, modification proficient derivative of strain + 8325–4.</td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b20">Kreiswirth <italic>et al</italic>. (1983)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> <italic>S. aureus</italic> RN2677</td><td align="left" rowspan="1" colspan="1">8325 derived φ11, φ12, φ13 lysogen, <italic>agr</italic><sup>+</sup>, <italic>sak</italic><sup>+</sup>, <italic>rsbU</italic><sup>–</sup>, <italic>hib</italic><sup>–</sup>, restriction deficient, spontaneous Nv<sup>r</sup> and Rf<sup>r</sup> mutant, recipient strain for conjugation.</td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b19">Kornblum <italic>et al</italic>. (1986)</xref> T.J. Foster <xref ref-type="table-fn" rid="tf1-1">a</xref></td></tr><tr><td colspan="3" rowspan="1">Plasmids</td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221</td><td align="left" rowspan="1" colspan="1">4555 bp: Cm<sup>r</sup>, <italic>inc4</italic>, <italic>repD</italic>, <italic>mobCAB</italic></td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b27">Novick and Bouanchaud (1971)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221<italic>cop903</italic></td><td align="left" rowspan="1" colspan="1">4168 bp: Cm<sup>r</sup>, <italic>inc4</italic>, <italic>repD</italic>, <italic>mobCAB</italic> (Δ806-1185: copy no. mutation)</td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b31">Projan <italic>et al</italic>. (1985)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER19</td><td align="left" rowspan="1" colspan="1">3063 bp: pUC19: Ap<sup>r</sup>, <italic>ori</italic> (pMB1), <italic>lacZ</italic>, <italic>cer</italic> (dimer resolution)</td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>. (2004)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER21T</td><td align="left" rowspan="1" colspan="1">3575 bp: pCER19 containing pC221<italic>cop903 oriT</italic></td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>. (2004)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER23T</td><td align="left" rowspan="1" colspan="1">3584 bp: pCER19 containing pC223 <italic>oriT</italic></td><td align="left" rowspan="1" colspan="1"><xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>. (2004)</xref></td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER194T</td><td align="left" rowspan="1" colspan="1">3560 bp: pCER19 containing pS194 <italic>oriT</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER21<italic>wt</italic></td><td align="left" rowspan="1" colspan="1">3127 bp: wild type 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER<italic>77mcb</italic></td><td align="left" rowspan="1" colspan="1">3127 bp: 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> mutated at <italic>mcb</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER<italic>77sra</italic></td><td align="left" rowspan="1" colspan="1">3127 bp: 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> mutated at <italic>sra</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER<italic>77nic</italic></td><td align="left" rowspan="1" colspan="1">3127 bp: 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> mutated at <italic>nic</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER21<italic>sra</italic>23</td><td align="left" rowspan="1" colspan="1">3127 bp: 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC223 <italic>sra</italic> swap</td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER23<italic>sra</italic>21</td><td align="left" rowspan="1" colspan="1">3127 bp: 77 bp pC223 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC221 <italic>sra</italic> swap</td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCER23<italic>wt</italic></td><td align="left" rowspan="1" colspan="1">3127 bp: wild type 77 bp pC223 <italic>oriT</italic><sub><italic>nic</italic></sub></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ</td><td align="left" rowspan="1" colspan="1">5625 bp: pE194 : pUC19 based shuttle vector.</td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ21wt</td><td align="left" rowspan="1" colspan="1">5689 bp: Shuttle vector containing wt 77 bp pC221 <italic>oriT</italic><sub><italic>nic</italic></sub></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ<italic>oriTA</italic></td><td align="left" rowspan="1" colspan="1">6128 bp: Shuttle vector containing 514 bp <italic>oriTA</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ<italic>oriTB</italic></td><td align="left" rowspan="1" colspan="1">5880 bp: Shuttle vector containing 256 bp <italic>oriTB</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ<italic>oriTC</italic></td><td align="left" rowspan="1" colspan="1">5753 bp: Shuttle vector containing 139 bp <italic>oriTC</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pCΔ<italic>oriTD</italic></td><td align="left" rowspan="1" colspan="1">6000 bp: Shuttle vector containing 386 bp <italic>oriTD</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221Δ<italic>mcb1</italic></td><td align="left" rowspan="1" colspan="1">4555 bp: pC221 with a 9 bp mutation of <italic>mcb1</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221Δ<italic>mcb2</italic></td><td align="left" rowspan="1" colspan="1">4555 bp: pC221 with a 9 bp mutation of <italic>mcb2</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221Δ<italic>mcb3</italic></td><td align="left" rowspan="1" colspan="1">4555 bp: pC221 with a 9 bp mutation of <italic>mcb3</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pC221Δ<italic>mcb1–3</italic></td><td align="left" rowspan="1" colspan="1">4555 bp: pC221 with a 9 bp mutations of <italic>mcb1</italic> and <italic>mcb3</italic></td><td align="left" rowspan="1" colspan="1">This work</td></tr><tr><td align="left" rowspan="1" colspan="1"> pGO1</td><td align="left" rowspan="1" colspan="1">52 314 bp: Gm<sup>r</sup>, Tp<sup>r</sup>, Qam<sup>r</sup>, <italic>tra</italic><sup>+</sup></td><td align="left" rowspan="1" colspan="1">T.J. Foster <xref ref-type="table-fn" rid="tf1-1">a</xref></td></tr></tbody></table><table frame="hsides" rules="groups"><thead><tr><th align="left" rowspan="1" colspan="1">Oligonucleotides</th><th align="left" rowspan="1" colspan="1">Sequence (5′−3′) <xref ref-type="table-fn" rid="tf1-2">b</xref></th><th align="center" rowspan="1" colspan="1">Description</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">PforORITA</td><td align="left" rowspan="1" colspan="1">AA<underline>CTGCAG</underline>AAATTTATGAACGTATAGC</td><td align="left" rowspan="1" colspan="1">PstI forward <italic>oriTA</italic></td></tr><tr><td align="left" rowspan="1" colspan="1">BrevORITA</td><td align="left" rowspan="1" colspan="1">CG<underline>GGATCC</underline>TTGGCAATCGATTGTCGC</td><td align="left" rowspan="1" colspan="1">BamHI reverse <italic>oriTA</italic></td></tr><tr><td align="left" rowspan="1" colspan="1">PforORITC</td><td align="left" rowspan="1" colspan="1">AA<underline>CTGCAG</underline>GATATAATCCCATCAAGCCG</td><td align="left" rowspan="1" colspan="1">PstI forward <italic>oriT</italic>C</td></tr><tr><td align="left" rowspan="1" colspan="1">PrevORITC</td><td align="left" rowspan="1" colspan="1">CG<underline>GGATCC</underline>ACGTTTCTTCGAAATATATTAC</td><td align="left" rowspan="1" colspan="1">BamHI reverse <italic>oriTC</italic></td></tr><tr><td align="left" rowspan="1" colspan="1">NIC+203</td><td align="left" rowspan="1" colspan="1">CATATTCGGATTCGCTC</td><td align="left" rowspan="1" colspan="1">Primer extension + strand</td></tr><tr><td align="left" rowspan="1" colspan="1">NIC+98</td><td align="left" rowspan="1" colspan="1">CACTCATTCAATCCCACC</td><td align="left" rowspan="1" colspan="1">Primer extension + strand</td></tr><tr><td align="left" rowspan="1" colspan="1">NIC−178</td><td align="left" rowspan="1" colspan="1">GAACGTATAGCAACCAC</td><td align="left" rowspan="1" colspan="1">Primer extension – strand</td></tr><tr><td align="left" rowspan="1" colspan="1">HPA23–</td><td align="left" rowspan="1" colspan="1">AAGTTTATAGCAACCAC</td><td align="left" rowspan="1" colspan="1">Primer extension + strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21P+</td><td align="left" rowspan="1" colspan="1">AACGAAGTGGCTAGAATATA</td><td align="left" rowspan="1" colspan="1"><italic>mcb</italic> oligoduplex + strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21P–</td><td align="left" rowspan="1" colspan="1">TATATTCTAGCCACTTCGTT</td><td align="left" rowspan="1" colspan="1"><italic>mcb</italic> oligoduplex – strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21Pmut+</td><td align="left" rowspan="1" colspan="1">AACGA<bold>GAGTAGGTC</bold>AATATA</td><td align="left" rowspan="1" colspan="1">mutagenic <italic>mcb</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21Pmut–</td><td align="left" rowspan="1" colspan="1">TATATT<bold>GACCTACTC</bold>TCGTT</td><td align="left" rowspan="1" colspan="1">mutagenic <italic>mcb</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–MCB+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGCCGGTATATTCAGAACGA<bold>GAGTAGGTC</bold>AATATACGACGCTTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>mcb</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–MCB–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAAGCGTCGTATATT<bold>GACCTACTC</bold>TCGTTCTGAATATACCGGCTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>mcb</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–SRA+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGCCGGTATATTCAGAACGAAGTGGCTAGAATAT<bold>CATTATA</bold>TTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>sra</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–SRA–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAA<bold>TATAAGT</bold>ATATTCTAGCCACTTCGTTCTGAATATACCGGCTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>sra</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–NIC+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGCCGGTATATTCAGAACGAAGTGGCTAGAATATACGACGCTT<bold>T</bold>CCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>nic</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">ORIT–NIC–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGG<bold>A</bold>AAGCGTCGTATATTCTAGCCACTTCGTTCTGAATATACCGGCTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with mutated <italic>nic</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21WT+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGCCGGTATATTCAGAACGAAGTGGCTAGAATATACGACGCTTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21WT–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAAGCGTCGTATATTCTAGCCACTTCGTTCTGAATATACCGGCTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21SRA23+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGCCGGTATATTCAGAACGAAGTGGCTAGAATAT<bold>CCAGT</bold>GCTTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC223 <italic>sra</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">21SRA23–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAAGC<bold>ACTGG</bold>ATATTCTAGCCACTTCGTTCTGAATATACCGGCTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC223 <italic>sra</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">23SRA21+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGTACGGCGTGCTAGAACGAAGTGGATAGCAAAG<bold>ACGAC</bold>GCTTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC223 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC221 <italic>sra</italic>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">23SRA21–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAAGC<bold>GTCGT</bold>CTTTGCTATCCACTTCGTTCTAGCACGCCGTACTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC223 <italic>oriT</italic><sub><italic>nic</italic></sub> with pC221 <italic>sra</italic>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">23WT+</td><td align="left" rowspan="1" colspan="1">GGGTATGGGATATAATCCCATCAAGTACGGCGTGCTAGAACGAAGTGGATAGCAAAGCCAGTGCTTGCCAAACCACAG</td><td align="left" rowspan="1" colspan="1">pC223 <italic>oriT</italic><sub><italic>nic</italic></sub>+ strand</td></tr><tr><td align="left" rowspan="1" colspan="1">23WT–</td><td align="left" rowspan="1" colspan="1"><underline>GATCC</underline>TGTGGTTTGGCAAGCACTGGCTTTGCTATCCACTTCGTTCTAGCACGCCGTACTTGATGGGATTATATCCCATACC<underline>CTGCA</underline></td><td align="left" rowspan="1" colspan="1">pC223 <italic>oriT</italic><sub><italic>nic</italic></sub>– strand</td></tr><tr><td align="left" rowspan="1" colspan="1">PST–BSTXI+</td><td align="left" rowspan="1" colspan="1">AGG<underline>CTGCAG</underline>TAGCAA<underline>CCACATTTTTGG</underline></td><td align="left" rowspan="1" colspan="1">PstI and BstXI forward</td></tr><tr><td align="left" rowspan="1" colspan="1">BAM–CLAI–</td><td align="left" rowspan="1" colspan="1">CG<underline>GGATCC</underline>A<underline>ATCGAT</underline>TGTCGCGTTTC</td><td align="left" rowspan="1" colspan="1">BamHI and ClaI reverse</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB1A+</td><td align="left" rowspan="1" colspan="1">AGCTTCATCATGCTGTATG</td><td align="left" rowspan="1" colspan="1">Forward <italic>mcb1</italic></td></tr><tr><td align="left" rowspan="1" colspan="1">MCB1A–</td><td align="left" rowspan="1" colspan="1">AAAGTG<bold>GACCTACTC</bold>CATAGTTCAAGACCTAATTC</td><td align="left" rowspan="1" colspan="1">Reverse <italic>mcb1</italic> mutation</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB1B+</td><td align="left" rowspan="1" colspan="1">ACTATG<bold>GAGTAGGTC</bold>CACTTTGCCACTCATTTTTTGCG</td><td align="left" rowspan="1" colspan="1">Forward <italic>mcb1</italic> mutation</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB2A–</td><td align="left" rowspan="1" colspan="1">TATATT<bold>GACCTACTC</bold>TCGTTCTGAATATACCGG</td><td align="left" rowspan="1" colspan="1">Reverse <italic>mcb2</italic> mutation</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB2B+</td><td align="left" rowspan="1" colspan="1">GAACGA<bold>GAGTAGGTC</bold>AATATACAACGCTTGCC</td><td align="left" rowspan="1" colspan="1">Forward <italic>mcb2</italic> mutation</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB3A–</td><td align="left" rowspan="1" colspan="1">CAAATC<bold>TGATGCAAG</bold>ATTATTATCATATTCACTC</td><td align="left" rowspan="1" colspan="1">Reverse <italic>mcb3</italic> mutation</td></tr><tr><td align="left" rowspan="1" colspan="1">MCB3B+</td><td align="left" rowspan="1" colspan="1">AATAAT<bold>CTTGCATCA</bold>GATTTGTCTGTTGGGGAAAAC</td><td align="left" rowspan="1" colspan="1">Forward <italic>mcb3</italic> mutation</td></tr></tbody></table><table-wrap-foot><fn id="tf1-1"><label>a</label><p>T.J. Foster, Department of Microbiology, Trinity College Dublin</p></fn><fn id="tf1-2"><label>b</label><p>Underlined nucleotides indicate restriction tags, whilst those emboldened denote nucleotides that are mutated with respect to the wild type</p></fn></table-wrap-foot></table-wrap><p>While an interaction between MobC and <italic>oriT</italic> at <italic>mcb2</italic> has been identified, this does not confirm that binding by the MobC at this site is a prerequisite for nicking by MobA. The minimal 77 bp target identified as being susceptible to nicking by the MobA and MobC proteins of pC221, although not mobilized <italic>in vivo</italic>, was referred to as <italic>oriT</italic><sub><italic>nic</italic></sub>. This <italic>oriT</italic><sub><italic>nic</italic></sub> was cloned into pCER19, as pCER21<italic>wt</italic> (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>), and employed as a control to examine the effect of mutation within the <italic>mcb2</italic> site. Using this substrate as a basis, we designed a variant, pCER<italic>77mcb</italic> (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>), with the 9 bp <italic>mcb2</italic> randomized to a sequence previously shown to demonstrate no binding (<xref ref-type="fig" rid="fig03">Fig. 3</xref>). This substrate was not susceptible to nicking <italic>in vitro</italic> by 21MobA and GSH–MobC (<xref ref-type="fig" rid="fig07">Fig. 7B</xref>). Barring any pleiotropic effects of mutating <italic>mcb2</italic>, this suggests that MobC binding at <italic>mcb2</italic> is indeed a requirement for nicking by the MobA relaxase.</p><fig id="fig07" position="float"><label>Fig. 7</label><caption><title>Effect on nicking activity of mutation within <italic>oriT</italic><sub><italic>nic</italic></sub>.</title><p>A. Alignment of pC221 (GenBank Accession No. X02166), pS194 (GenBank Accession No. X06627) and pC223 (GenBank Accession No. NC_005243) <italic>oriT</italic> regions. Bases that are not conserved between pC221/pS194 and pC223 are highlighted black. The positions of MobC binding (<italic>mcb2</italic>), MobA recognition (<italic>sra</italic>) and hypersensitivity in the presence of MobA (<italic>HSA</italic>) are labelled. <italic>Nic</italic> is indicated by an arrow. The susceptibility to nicking in the presence of either pC221 MobA (21MobA) or pC223 MobA (23MobA) is listed for each. The <italic>oriT</italic><sub><italic>nic</italic></sub> substrates that were synthesized with incorporated mutations made at <italic>mcb2</italic> (pCER<italic>77mcb</italic>), <italic>sra</italic> (pCER<italic>77sra</italic>) and <italic>nic</italic> (pCER<italic>77mcb</italic>) were based on the pC221 sequence of the alignment. The <italic>oriT</italic><sub><italic>nic</italic></sub> substrates pCER21<italic>wt</italic> and pCER23<italic>wt</italic> correspond to the wild-type pC221 and pC223 sequences, and were used as controls for the <italic>sra</italic>-swap experiment. The four bases swapped between these sequences to generate the <italic>sra</italic>-swap substrates pCER21<italic>sra</italic>23 and pCER23<italic>sra</italic>21 are indicated.</p><p>B and C. Nicking assays performed on supercoiled cloned mutant and <italic>sra</italic>-swap <italic>oriT</italic><sub><italic>nic</italic></sub> substrates. The nicking reactions were prepared as described (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). DNA only controls for each substrate are indicated (–). The position of nicked, open-circular DNA (OC) and supercoiled DNA (SC) is given. (B) Nicking of pC221 sequence based <italic>mcb</italic>, <italic>sra</italic> and <italic>nic</italic> mutants was performed using pC221 MobAH<sub>6</sub> and GSH–MobC, with pCER21<italic>wt</italic> as a positive control. (C) Nicking of <italic>sra</italic>-swap <italic>oriT</italic><sub><italic>nic</italic></sub> substrates was in the presence of pC221 MobAH<sub>6</sub> (21A) or pC223 MobAH<sub>6</sub> (23A), and pC221 GSH–MobC. Percentages indicate percentage of total plasmid DNA in the nicked open-circular DNA form.</p></caption><graphic xlink:href="mmi060-1302-f7"/></fig><p>The <italic>nic</italic> sites of the pC221-family <italic>oriT</italic> regions are identical, suggesting a strict sequence requirement at this region. This was investigated by mutation of the pCER21<italic>wt</italic> substrate at the 5′ nucleotide of the <italic>nic</italic> dinucleotide from 5′-GC-3′ to 5′-TC-3′ to create the substrate pCER<italic>77nic</italic> (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>). This substrate was also not susceptible to nicking in the <italic>in vitro</italic> assay (<xref ref-type="fig" rid="fig07">Fig. 7B</xref>).</p></sec><sec><title>All three binding sites are required for mobilization of pC221</title><p>The presence of multiple MobC binding sites within <italic>oriT</italic> suggests a potential for a higher-order structure formation between MobC and more than one <italic>mcb</italic> site. While the presence of <italic>mcb2</italic> alone is sufficient for nicking, <italic>mcb1</italic> and <italic>mcb3</italic> may be involved in the subsequent mobilization of <italic>oriT</italic>. The pCΔ-based shuttle plasmid constructs described above were assayed for mobilization in staphylococcal filter-mating experiments. This involved introducing them into a donor <italic>S. aureus</italic> RN4220 strain containing pGO1, encoding the conjugative transfer apparatus, and wild-type pC221 as a helper plasmid to provide <italic>mob</italic> functions <italic>in trans</italic>. Recipients were scored for their ability to grow on erythromycin, conferred by pCΔ, and both novobiocin and rifampicin, conferred by the recipient strain RN2677. Within the sensitivity of the experiment, these assays demonstrated that the gross deletion of regions containing <italic>mcb1</italic> or <italic>mcb3</italic> resulted in an 18-fold reduction in mobilization efficiency, suggesting all three <italic>mcb</italic> sites are required to constitute a fully functional <italic>oriT</italic>.</p><p>To more fully investigate the requirement for each MobC binding site in isolation, we used overlap extension PCR to generate mutations in either <italic>mcb1, mcb2, mcb3</italic> or <italic>mcb1</italic> + <italic>mcb3</italic>, which were then cloned in the context of pC221 (see <xref ref-type="table" rid="tbl1">Table 1</xref>). This yielded four mutant plasmids that differ from the wild type only at those <italic>mcb</italic> sites mutated. Binding sites <italic>mcb1</italic> and <italic>mcb2</italic> were mutated to the same randomized sequences described previously (<xref ref-type="fig" rid="fig03">Fig. 3</xref>). In order to mutate <italic>mcb3</italic> within the <italic>mobC</italic> gene, it was necessary to encode a silent mutation to maintain the amino acid sequence, although 9 bp were still altered. The resulting pC221 mutants were introduced into donor RN4220 cells containing pGO1, filter-mated with RN2677 and the transconjugant cells scored for resistance to chloramphenicol, as well as the novobiocin and rifampicin encoded by the recipient RN2677. It was observed that mutation of <italic>mcb1</italic>, <italic>mcb2</italic>, <italic>mcb3</italic> or <italic>mcb1</italic> + <italic>mcb3</italic>, in the context of pC221, resulted in a loss of mobilizable phenotype (<xref ref-type="fig" rid="fig06">Fig. 6A</xref>). Pleiotropic effects aside, this suggests that all three MobC binding sites are required for mobilization.</p><p>To assess whether the loss of mobilization was due to abolishment of nicking, whole cell lysate analysis was performed on the donor strains containing each of the pC221 mutants (<xref ref-type="fig" rid="fig06">Fig. 6B</xref>). No nicking was observed on the plasmid containing mutation at <italic>mcb2</italic>, which confirmed the earlier <italic>in vitro</italic> data from pCER<italic>77mcb</italic> (<xref ref-type="fig" rid="fig07">Fig. 7B</xref>). In all the other mutants a nicked open-circular complex, comparable to wild-type pC221, was observed, signifying successful relaxosome formation. This represents the first time <italic>in vivo</italic> nicking has been observed in the absence of mobilization in pC221 containing wild-type Mob proteins.</p></sec><sec><title>The <italic>sra</italic> sequence is the basis of substrate recognition</title><p>An alignment between the <italic>oriT</italic> regions of the three closely related plasmids pC221 (and <italic>cop903</italic>), pC223 and pS194 highlights the clustering of divergent sequences within <italic>oriT</italic><sub><italic>nic</italic></sub>, which could bear functional significance (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>). We have previously shown that the MobA proteins of pC221 and pC223 encode the plasmid-specific sequence specificity, such that these proteins recognize only the cognate <italic>nic</italic> (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). In a comparison of MobA activities against an additional cloned substrate we found that MobA of pC221 will also nick the pS194 <italic>oriT</italic> (pCER194T) at an identical position, whereas the pC223 MobA does not (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>).</p><p>The alteration of the MobC-induced DNase I hypersensitivity in the region between <italic>mcb</italic> and <italic>nic</italic> on addition of MobA suggests this sequence is a specific site for interaction with MobA. This 7 bp sequence was randomized (maintaining wild-type base composition) in the <italic>oriT</italic><sub><italic>nic</italic></sub> mutant pCER<italic>77sra</italic> (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>), and was found to not be susceptible to nicking by 21MobA in the <italic>in vitro</italic> nicking assay (<xref ref-type="fig" rid="fig07">Fig. 7B</xref>). It is therefore likely that this sequence, designated <italic>sra</italic> (<underline>s</underline>ite <underline>r</underline>ecognized by Mob<underline>A</underline>), plays a role in the interaction of the relaxosome at <italic>oriT</italic>.</p><p>The <italic>sra</italic> sequence of different plasmids may offer a means of substrate-specific interaction with the cognate MobA protein. Therefore, the four bases within <italic>sra</italic> that are not conserved between pC221 and pC223 were exchanged (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>), and the resultant substrates assessed for nicking by MobA of either pC221 or pC223, in the presence of pC221 MobC protein. Four <italic>oriT</italic><sub><italic>nic</italic></sub> substrates were compared: two controls consisting of wild-type pC221 (pCER21<italic>wt</italic>) and pC223 (pCER23<italic>wt</italic>) sequences; and two <italic>sra</italic>-swap substrates, pCER21<italic>sra</italic>23 and pCER23<italic>sra</italic>21. <italic>In vitro</italic> nicking assays demonstrated that exchange of these 4 bp within the <italic>sra</italic> resulted in swapping of substrate identities and consequently preferential cleavage by MobA (<xref ref-type="fig" rid="fig07">Fig. 7C</xref>); with only residual nicking of pCER21<italic>sra</italic>23 by 21MobA (14%), compared with 23MobA (39%); and no detectable nicking of pCER23<italic>sra</italic>21 by 23MobA (0%), compared with 21MobA (40%). The use of MobC derived from pC223, rather than pC221, yielded comparable results (data not presented). Thus the four exchanged base pairs represent the determinants within <italic>sra</italic> of pC221 and pC223 MobA specificity.</p></sec></sec><sec><title>Discussion</title><p>The MobA relaxases of pC221 and pC223 display specificity towards their cognate origins of transfer. This is in contrast to the MobC accessory proteins essential for the site- and strand-specific transesterification reaction mediated by the relaxase, which are interchangeable between the two related relaxosomes (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). In the present study we have distinguished the respective roles of the MobA and MobC proteins in substrate recognition and subsequent nicking specificity.</p><p>The pC221 MobC protein is a dimeric DNA-binding protein able to bind <italic>oriT</italic> DNA of either pC221 or pC223 (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). We have used DNase I footprinting, mutational analysis and SPR to investigate the interaction of MobC at <italic>oriT</italic> and identified three distinct binding sites within the <italic>oriT</italic> region. Two of the binding sites, designated <italic>mcb1</italic> and <italic>mcb2</italic>, which contain a conserved asymmetric 9 bp direct repeat, are located 106 bp (<italic>mcb1</italic>) and 15 bp (<italic>mcb2</italic>) upstream of <italic>nic</italic>. A third, designated <italic>mcb3</italic>, is encoded within the N-terminal coding region of <italic>mobC</italic>, and contains 7 bp in common with the direct repeat sequence. Thus the binding sites appear to conform to direct repeats despite the presence of local inverted repeats (<xref ref-type="fig" rid="fig05">Fig. 5</xref>). SPR analysis suggests a K<sub>D</sub> of 1.8 µM for the interaction between MobC and the single <italic>mcb2</italic> site in isolation; however, given the presence of three such sites within the <italic>oriT</italic> region, it is quite possible that cooperativity may enhance this affinity.</p><p>All three binding sites are highly conserved in terms of both sequence and location within the mobilon regions of a number of plasmids from <italic>Staphylococcus</italic> previously identified. The conservation of these sequences between pC221 and pC223 explains the interchangeable nature of the respective MobC proteins with regard to binding observed in earlier EMSA experiments (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>); a single base pair difference between the pC221 <italic>mcb2</italic> and the corresponding site on pC223 appears to have no effect on this activity. This is in contrast to plasmid RP4, where the accessory protein TraJ is able to distinguish between its cognate binding site, <italic>srj</italic>, and that of the related plasmid R751, which differ in 4 of 10 bp. Heterologous nicking and transfer of a cloned RP4 <italic>oriT</italic> by R751 proteins is only possible if RP4 <italic>traJ</italic> is encoded <italic>in cis</italic> (<xref ref-type="bibr" rid="b45">Ziegelin <italic>et al</italic>., 1989</xref>). Thus in this latter example, in addition to initiating assembly of the relaxosome at <italic>oriT</italic>, the accessory protein is able to fulfil a role in substrate specificity.</p><p>The presence of three binding sites within the pC221-family <italic>oriT</italic>s has the potential for DNA looping or other high-order structures with a core of MobC mediating interactions between these positions. This possibility is further supported by the pattern of diminished- and hyper-sensitivity to DNase I cleavage between <italic>mcb1</italic> and <italic>mcb2</italic>, evident at higher concentrations, indicative of increased DNase I access to the minor groove on the outside of bent DNA and restricted access to the minor groove within such a bend (<xref ref-type="bibr" rid="b16">Hochschild and Ptashne, 1986</xref>). However, given the excess of MobC over binding substrate, it remains possible that such a pattern results from non-specific DNA interactions.</p><p>A minimal <italic>oriT</italic>, functional in mobilization, <italic>in trans</italic>, is defined by a 524 bp fragment of pC221. Deletion of 97 bp (including <italic>mcb1</italic>) from the 5′ end and 253 bp (including <italic>mcb3</italic>) from the 3′ end of this fragment does not diminish nicking susceptibility, but compromises transfer efficiency. This bears similarity to the plasmid RP4, which has a functional <italic>oriT</italic> comprising ∼350 bp; however, a 194 bp <italic>oriT</italic> is also mobilizable but with 100–200-fold reduced mobilization efficiency. This was found to result from the loss of the TraK binding site, the binding of which is stimulatory to nicking and is proposed to alter the superhelical density at <italic>nic</italic> (<xref ref-type="bibr" rid="b46">Ziegelin <italic>et al</italic>., 1992</xref>).</p><p>Sequential site-directed mutagenesis of the three <italic>mcb</italic> sites, in the context of pC221, indeed confirms that an <italic>oriT</italic> containing all three <italic>mcb</italic> sites is required for wild-type levels of mobilization. The loss of <italic>mcb2</italic> abolishes <italic>in vivo</italic> nicking activity and therefore subsequent mobilization. This parallels observations where mutagenesis of accessory protein binding sites reduces nicking and transfer frequency in RP4 (via TraJ; <xref ref-type="bibr" rid="b40">Waters <italic>et al</italic>., 1991</xref>) and R64 (via NikA; <xref ref-type="bibr" rid="b10">Furuya and Komano, 1997</xref>). The loss of <italic>mcb1</italic> or <italic>mcb3</italic>, while having no impact on nicking, effectively abolishes mobilization. Thus MobC may function at several levels: primarily to initiate nicking by MobA, presumably via interaction at <italic>mcb2</italic>; but furthermore by mediating a potential high-order complex formation, additionally involving <italic>mcb1</italic> and <italic>mcb3</italic>, to yield a mobilizable substrate. Such a complex may even regulate the transcription of <italic>mobC</italic>.</p><p>In pC221, the binding of MobC induces DNase I hypersensitivity in the DNA flanking the protected sequence, indicative of a localized bending of DNA often observed on binding of a protein to DNA. Such bending could serve to stabilize subsequent relaxosome formation. On addition of MobA to a MobC–<italic>oriT</italic> complex, the observed pattern of DNase I protection and hypersensitivity is altered, revealing MobA protection over a 7 bp region (<italic>sra</italic>) between <italic>mcb2</italic> and <italic>nic</italic>. In addition, MobA contributes to increased DNase I hypersensitivity downstream from the <italic>nic</italic>, indicating that distortion of DNA structure may be important within the relaxosome.</p><p>Mutation of <italic>sra</italic> abolished susceptibility of a minimal 77 bp <italic>oriT</italic><sub><italic>nic</italic></sub> substrate to cleavage, which taken together with footprinting data identifies this as a site of interaction with MobA. This conclusion is supported by several other observations. First, the region coincides with a region of sequence divergence between the pC221 and pC223 <italic>oriT</italic><sub><italic>nic</italic></sub> (<xref ref-type="fig" rid="fig07">Fig. 7A</xref>), which, despite their similarity, can only be nicked by their cognate relaxases.</p><p>Second, comparison of the pC221 <italic>oriT</italic><sub><italic>nic</italic></sub> and a comparable region of the IncP plasmid RP4 demonstrates the similar juxtaposition of the RP4 TraJ accessory protein binding site upstream of <italic>nic</italic>. Mutations within a 10 bp sequence upstream of <italic>nic</italic>, comparable to the position of <italic>sra</italic>, in the identical plasmid RK2 result in complete loss of nicking activity (<xref ref-type="bibr" rid="b40">Waters <italic>et al</italic>., 1991</xref>). A 6 bp core TraI recognition sequence within this region, <italic>sri</italic>, was subsequently determined in the context of RP4 using oligonucleotide cleavage assays (<xref ref-type="bibr" rid="b28">Pansegrau <italic>et al</italic>., 1993</xref>).</p><p>Finally, conservation of the <italic>mcb2</italic> site and interchangeable nature of heterologous MobC proteins suggests that while necessary for nicking activity, this interaction is unable to account for the observed nicking specificity. The interaction of MobA at <italic>sra</italic> does explain such specificity.</p><p>Mutation of the region proposed as the <italic>sra</italic> results in a loss of nicking activity. By swapping those bases within <italic>sra</italic> that are divergent between pC221 and pC223 we were able to turn a substrate of the pC221 MobA protein into one preferentially cleaved by that of pC223 and <italic>vice versa</italic>. Thus <italic>sra</italic> represents not only a site of interaction between <italic>oriT</italic> and MobA, but also provides differentiation between pC221 and pC223.</p><p>The <italic>oriT</italic><sub><italic>nic</italic></sub> of the pC221 mobilon family thus consists of at least three sequence domains: <italic>mcb</italic>, <italic>sra</italic> and <italic>nic</italic>, each of which is required for correct nicking activity. The current model of pC221 relaxosome assembly could be considered one of layered specificity. The MobC proteins of pC221 and pC223 are interchangeable with respect to nicking; this is borne out by the conservation of the binding sequence between the respective substrates, with a single base exception in <italic>mcb2</italic> of pC223. A substrate encoding a mutated <italic>mcb2</italic> is not nicked, thus, barring any pleiotropy due to such mutation, MobC binding is a prerequisite for subsequent nicking by MobA, and can be considered the first layer of relaxosome specificity.</p><p>The role of MobC is not clear cut, but apparently facilitates the recruitment of MobA to the nick site. In this regard, MobC may perform a structural or positional role where it could mediate MobA recruitment by structurally altering the substrate, perhaps to present a recognition complex. Alternatively, it could act to position the MobA over its recognition sequence, which may be presented as single-stranded DNA. Evidence suggests that a high-order structure between MobC and <italic>mcb1</italic>, <italic>2</italic> and <italic>3</italic> is important for subsequent mobilization of a nicked plasmid; the nature of such a complex is yet to be elucidated. Each of these roles are not mutually exclusive and not dissimilar in principle to the binding of DnaA at <italic>oriC</italic> prior to the recruitment of the DnaBC helicase complex in the initiation of bacterial DNA replication (<xref ref-type="bibr" rid="b23">Messer, 2002</xref>), or the recruitment of ParF into the DNA partition complex via interaction with sequence-specific ParG protein (<xref ref-type="bibr" rid="b4">Barilla and Hayes, 2003</xref>).</p><p>In addition to recruitment, specific recognition of <italic>sra</italic> is required for subsequent cleavage by MobA. Whether recognition is mediated through extrusion of a single-stranded DNA as a potential result of MobC binding is yet to be established. Single-stranded DNA recognition and binding has, however, been demonstrated for the relaxase domain, TraI36, of the unrelated F-plasmid (<xref ref-type="bibr" rid="b38">Stern and Schildbach, 2001</xref>) and TrwC of R388 (<xref ref-type="bibr" rid="b14">Guasch <italic>et al</italic>., 2003</xref>).</p><p>We have not yet been able to demonstrate a direct protein–protein interaction between MobC and MobA, nor is there any unequivocal data for such interactions between TraJ–TraI of RP4 (<xref ref-type="bibr" rid="b45">Ziegelin <italic>et al</italic>., 1989</xref>) or NikA–NikB (<xref ref-type="bibr" rid="b9">Furuya and Komano, 1995</xref>). While not conclusive, a protein–protein interaction between the homologous <italic>Agrobacterium tumefaciens</italic> pTi plasmid VirD1–VirD2 proteins has been suggested on the basis of nuclear localization and co-immunoprecipitation in mammalian cells (<xref ref-type="bibr" rid="b33">Relic <italic>et al</italic>., 1998</xref>).</p><p>In summary, we have located three sequences within <italic>oriT</italic> required for interaction with the accessory protein MobC, and a further site for the subsequent interaction with the relaxase MobA. Nicking by the latter is dependent on the presence of the plasmid-specific sequence adjacent to <italic>nic</italic>, in concert with a single <italic>mcb</italic> site. However, all three <italic>mcb</italic> sites are required for formation of a productive complex capable of mobilization. Finally, it is worth noting the presence of the third MobC binding site within the <italic>mobC</italic> gene itself. Such a location presents the possibility of an autoregulatory role for MobC; a role that would link the establishment of the relaxation complex with the expression of these proteins required for its formation.</p></sec><sec sec-type="methods"><title>Experimental procedures</title><sec><title>Bacterial strains, growth conditions and manipulation of DNA</title><p>Bacterial strains, plasmids and primers are listed in <xref ref-type="table" rid="tbl1">Table 1</xref>. <italic>E. coli</italic> strain DH5α was used for the preparation and maintenance of plasmid constructs. <italic>S. aureus</italic> strain RN4220 was used as a donor in all filter mating experiments; RN2677 (kindly provided by T.J. Foster) was used as the recipient.</p><p><italic>Escherichia coli</italic> DH5α was cultured at 37°C using 2YT broth or Luria–Bertani (LB) agar (<xref ref-type="bibr" rid="b34">Sambrook and Russell, 2001</xref>). <italic>S. aureus</italic> was cultured at 30–37°C using Brain-Heart Infusion (BHI) broth or agar (Oxoid, Oxford, UK). For electroporation, recovery was in B2 broth (<xref ref-type="bibr" rid="b35">Schenk and Laddaga, 1992</xref>) but using 0.5% (w/v) NaCl rather than their stated 2.5% (A. O’Neill, pers. comm.). When appropriate, antibiotics were used at the following concentrations: Ampicillin (Ap) 100 µg ml<sup>−1</sup>; Chloramphenicol (Cm) 10 µg ml<sup>−1</sup>; Erythromycin (Em) 5 µg ml<sup>−1</sup>; Gentamicin (Gm) 5 µg ml<sup>−1</sup>; Novobiocin (Nv) 5 µg ml<sup>−1</sup>; Rifampicin (Rf) 5 µg ml<sup>−1</sup>.</p><p>Plasmid DNA preparations were performed as described (<xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>). Standard nucleotide sequencing was performed commercially (Lark technologies, Essex, UK). Oligonucleotides were prepared by commercial synthesis (MWG Biotech, Ebersberg, Germany).</p></sec><sec><title>Preparation of plasmid constructs</title><p>All plasmid constructs were prepared using <italic>E. coli</italic> DH5α. The shuttle vector pCΔ was constructed by digestion of pCER19, a pUC19-derivative containing the <italic>cer</italic> resolution sequence (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>), and the staphylococcal plasmid pE194 with restriction endonucleases EcoRI and BspEI respectively. Overhanging ends were filled in with DNA polymerase I Klenow fragment and the fragments ligated such that pCER19 <italic>bla</italic> was in series with pE194 <italic>repF</italic>. The pE194 <italic>pre</italic> (<underline>p</underline>lasmid <underline>re</underline>combinase) was subsequently excised via an Acc651–BsrGI digestion and self-ligation of the compatible ends. For constructs based on the pCΔ shuttle vector, <italic>oriT</italic> fragments were generated by PCR from a pC221 template using standard methods and <italic>Pfu</italic> turbo polymerase (Stratagene). All PCR products contained PstI and BamHI restriction tags, which were digested for directional cloning into similarly digested pCΔ. The primer sequences used for preparation of <italic>oriT</italic> shuttle constructs are listed in <xref ref-type="table" rid="tbl1">Table 1</xref>. <italic>oriTA</italic> was produced using primers PforORITA/BrevORITA. <italic>oriTB</italic> was prepared by digestion of <italic>oriTA</italic> with PstI/BstBI and cloning into pCΔ digested with PstI/AccI. <italic>oriTC</italic> was prepared using PforORITC and BrevORITC; and <italic>oriTD</italic> using PforORITC and BrevORITA. Ligated recombinants were transformed into <italic>E. coli</italic> DH5α (<xref ref-type="bibr" rid="b7">Dagert and Ehrlich, 1979</xref>) under Ap selection. Recombinant shuttle vectors prepared from <italic>E. coli</italic> were screened by restriction analysis and confirmed by DNA sequence analysis. They were then introduced into the mating donor strain (<italic>S. aureus</italic> RN4220, pGO1, pC221) after first passaging through plasmid-free <italic>S. aureus</italic> RN4220 (<xref ref-type="table" rid="tbl1">Table 1</xref>) by electroporation as described (<xref ref-type="bibr" rid="b35">Schenk and Laddaga, 1992</xref>) using 0.2 cm gap cuvette, 2.3 kV pulse voltage, 2.5 ms.</p><p>Site-directed mutations of <italic>mcb</italic> sites in the context of pC221 <italic>oriT</italic> were prepared by overlap extension PCR using pC221 as a template. Each mutant <italic>oriT</italic> was prepared in two separate reactions (A and B) with primers for the two halves overlapping at the site of the desired mutation. The primers used for mutation at <italic>mcb1</italic> were MCB1A+ and MCB1A– (reaction A), and MCB1B+ and BAM–CLAI– (reaction B). Primers for <italic>mcb2</italic> were PST–BSTXI+ and MCB2A– (reaction A), and MCB2B+ and BAM–CLAI– (reaction B). Primers for <italic>mcb3</italic> were PST–BSTXI+ and MCB3A– (reaction A), and MCB3B+ and BAM–CLAI– (reaction B). The products of reactions A and B were gel purified and subjected to a final PCR amplification using primers (PST–BSTXI+ and BAM–CLAI–) in order to amplify the final product from the overlap-extended A+B template, and incorporate appropriate restriction site tags. The mutagenized <italic>oriT</italic> DNA (511 bp) products were digested with BamHI and PstI and cloned into similarly digested pCER19 and transformed into <italic>E. coli</italic> DH5α. Recombinant DNA was selected by α-complementation using a complete LB/S-Gal/IPTG medium (Sigma) and screened by restriction analysis. Once isolated, recombinant DNA for each mutant was digested with BstXI and ClaI, releasing the mutagenized <italic>oriT</italic> DNA. This fragment was ligated with a similarly digested reciprocal part of the pC221 and transformed by electroporation as described above. Mutant plasmids were confirmed by sequencing.</p><p>For <italic>oriT</italic><sub><italic>nic</italic></sub> constructs based on pCER19, complementary strands corresponding to either wild-type or specifically mutated 77 bp <italic>oriT</italic><sub><italic>nic</italic></sub> were synthesized incorporating PstI and BamHI cohesive ends for directional cloning. Complementary strands were annealed and ligated with BamHI–PstI-digested pCER19. Selection of recombinant plasmids was made as above, and confirmed by sequencing.</p></sec><sec><title>Proteins</title><p>MobA and MobC from pC221 and pC223 were purified as described previously (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). Where cited in text, 21MobA and 23MobA represent the C-terminally hexahistidine-tagged MobA proteins derived from pC221 and pC223 respectively. The N-terminal (His)<sub>6</sub>-tag was removed from the MobC proteins by thrombin cleavage, yielding GSH–MobC. Cleavage was performed in batches of 1 mg protein using 40 U thrombin (Pharmacia) in a total volume of 5 ml containing 200 mM KCl, 25 mM Tris.HCl (pH 8.0), 2 mM CaCl<sub>2</sub> and incubated at 21°C for 2 h. Following cleavage, the mixture was adjusted to 0.5 M KCl and passed through a benzamidine column (Pharmacia; 1 ml) at 0.5 ml min<sup>−1</sup> to remove thrombin. Protein solutions were dialysed and concentrated as previously reported (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>).</p></sec><sec><title>DNase I footprinting</title><p>Footprinting experiments were performed using the methodology of <xref ref-type="bibr" rid="b22">Leblanc and Moss (2001)</xref> and <xref ref-type="bibr" rid="b11">Gralla (1985)</xref>. Purified MobC (and/or MobA) and 0.37 pmol supercoiled DNA were mixed in a volume of 50 µl containing 50 mM KCl and 20 mM Tris.HCl (pH 8.0). Reaction mixes were incubated at 30°C for 30 min whereupon an equal volume (50 µl) of cofactor solution (10 mM MgCl<sub>2</sub> and 5 mM CaCl<sub>2</sub>) was added and the temperature adjusted to 25°C. 0.5 Kunitz units of DNase I (D-4263, Sigma) were added to each sample and incubated for 3 min. Reactions were stopped by addition of 100 µl reaction stop buffer (SDS 1% v/v, 200 mM NaCl, 20 mM EDTA pH 8.0) containing 40 µg ml<sup>−1</sup> tRNA (Sigma) as a carrier. The digested fragments were purified by phenol extraction (<xref ref-type="bibr" rid="b34">Sambrook and Russell, 2001</xref>) and 160 µl of the aqueous phase was precipitated with ethanol and resuspended to a concentration of 20 fmol µl<sup>−1</sup>. Control reactions consisted of a sample lacking protein but treated with DNase I and a sample containing protein, but without DNase I treatment.</p></sec><sec><title>Primer extension analysis</title><p>Primers NIC+98 and NIC+203 were designed to anneal 98 nt and 203 nt respectively, from the pC221 <italic>oriT nic</italic> on the plus-strand; primer NIC−178 binds 178 nt upstream from the position of <italic>nic</italic> on the complementary strand (<xref ref-type="table" rid="tbl1">Table 1</xref>). The DNase I-digested templates were probed by primer extension using an AmpliCycle<sup>®</sup> Sequencing Kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. Briefly, oligonucleotide (10 pmol) was 5′ end-labelled in a forward reaction by incubation with T4 polynucleotide kinase (30 units) (Invitrogen) and 10 pmol γ-[<sup>33</sup>P]-ATP (3000 Ci mmol<sup>−1</sup>) (MP Biomedicals, Irvine, CA).</p><p>Primer extension reactions contained 0.07 pmol of DNase I-digested plasmid DNA, 1.2 pmol 5′-end-labelled oligonucleotide primer, 0.25 units of AmpliTaq DNA polymerase CS and 160 pmol of each deoxyribonucleoside triphosphate (dNTP) (Invitrogen) in 8 µl 50 mM Tris.HCl pH 8.9, 10 mM KCl, 2.5 mM MgCl<sub>2</sub>, 0.025% v/v Tween 20. Dideoxy nucleotide sequencing ladders were prepared according to the manufacturer’s instructions using 0.02 pmol supercoiled pC221<italic>cop903</italic> plasmid DNA as a template. An additional control reaction was prepared in a similar manner to the sequencing ladder, but using 160 pmol of each dNTP in place of ddNTP.</p><p>All reactions (8 µl) were overlaid with 20 µl mineral oil (USB, Cleveland, OH) and were subjected to the following thermocycling parameters: 1× (95°C, 2 min); 25× (95°C, 1 min; 45–50°C, 1 min; 72°C, 1 min); 1× (72°C, 5 min). The reactions were stopped by addition of a formamide dye solution (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol FF) and primer extension products were resolved on a standard 6% polyacrylamide sequencing gel (<xref ref-type="bibr" rid="b34">Sambrook and Russell, 2001</xref>). Radiolabelled gels were imaged using Fuji BAS plates and a FujiBas-1000 phosphorimager (Fuji, Japan).</p></sec><sec><title>Competitive EMSA</title><p>Pre-formed MobC–<italic>oriT</italic> complexes were prepared in a manner similar to that previously reported (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>). BstXI/HindIII digestion of pCER21T DNA (0.22 pmol), yielded fragments of 2789 bp, 494 bp (containing pC221 <italic>oriT</italic>) and 292 bp, which were mixed with purified GSH–MobC (6 pmol). To equal aliquots of the MobC–<italic>oriT</italic> complex, 20 bp synthetic competitor or control DNA was added at concentrations of 25, 50, 100, 200, 400, 800 and 1600 nM. The reactions were assembled in a total volume of 20 µl 100 nM KCl, 20 mM Tris.HCl pH 8.0 and incubated at 30°C for 30 min. Reaction mixtures were separated by electrophoresis using 0.8% agarose gels in TBE as running buffer, limiting at 5 V cm<sup>−1</sup> for 100 min, then stained with ethidium bromide (1 µg ml<sup>−1</sup>) for 30 min and destained for 20 min in 1× TBE. Gel images were captured digitally using a GDS-8000 gel documentation system (UVP, Cambridge).</p></sec><sec><title>Surface Plasmon Resonance</title><p>Surface Plasmon Resonance was performed using a BIACORE 2000 unit with NTA sensor chip (Biacore AB, Uppsala). Prior to loading with protein the sensor chip was pre-rinsed with a 1 min pulse of EDTA (0.35 M) at 40 µl min<sup>−1</sup> to remove loose metal ions. The NTA was charged with a 1 min pulse of NiSO<sub>4</sub> (500 µM) at 5 µl min<sup>−1</sup>. H<sub>6</sub>MobC<sup>221</sup> at a concentration of 100 nM in running buffer 50 mM KCl, 20 mM Tris.HCl (pH 8.0), 50 µM EDTA, 0.005% P20 was injected over the NTA chip at a flow rate of 5 µl min<sup>−1</sup> to an immobilization of 358 RU.</p><p>Flow cell 1 was treated with Ni<sup>2+</sup> only and served as a reference surface to correct for refractive index changes and other bulk effects. Flow cell 4 was left untreated as a control for non-specific association of protein or DNA with the carboxymethylated dextran to which the NTA is immobilized.</p><p>Binding experiments were performed at a flow rate of 5 µl min<sup>−1</sup> in running buffer at 25°C. Injections of two concentrations of each oligoduplex, 100 nM and 1000 nM, were performed over all four flow cells, separated by buffer only controls, for 3 min pulses at a flow rate of 40 µl min<sup>−1</sup>. After each injection the chip was regenerated with a 3 min pulse of high-salt buffer [300 mM KCl, 20 mM Tris.HCl (pH 8.0), 50 µM EDTA, 0.005% P20] to displace any remaining DNA, and rinsed prior to injection of the next analyte/concentration.</p></sec><sec><title>Nicking assays</title><p>Assembly of relaxosomes in nicking assays, and where appropriate the quantification of nicking, was performed as previously reported (<xref ref-type="bibr" rid="b6">Caryl <italic>et al</italic>., 2004</xref>).</p></sec><sec><title>Whole cell lysate analysis</title><p>Whole cell lysate analysis was performed as previously reported (<xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>).</p></sec><sec><title>Filter mating experiments</title><p>The mobilizable phenotype of pC221 and pCΔ shuttle constructs containing alternate <italic>oriT</italic> fragments was assessed using a modification of the filter mating methodology reported previously (<xref ref-type="bibr" rid="b37">Smith and Thomas, 2004</xref>). RN4220 containing pGO1 and pC221 was used as a donor strain. Overnight cultures of donor and recipient (RN2677) were grown in BHI broth and diluted into fresh media, in the presence of appropriate antibiotics, to an OD<sub>600</sub> equivalent of 0.1 and incubated at 30°C until mid-log phase.</p><p>Volumes of donor and recipient cultures equivalent to ∼3 × 10<sup>8</sup> cells (1 ml culture at OD<sub>600</sub> of 0.6) were harvested and resuspended separately in 1 ml fresh BHI broth. The washed donor and recipient cells were mixed and a further 1 ml fresh BHI broth added before filtering through a sterile 13 mm diameter, 0.2 µm pore nitrocellulose filter (Millipore). The filter was extracted and placed bacteria side up on a BHI agar plate, without selection, and incubated at 30°C for 18 h.</p><p>Cells were recovered from filters by vortexing in 1 ml sterile phosphate-buffered saline and serial dilutions were plated onto separate BHI agar plates containing appropriate selection for donors or transconjugants. The efficiency of transfer was calculated as a ratio of the mobilization frequency (erythromycin- or chloramphenicol-resistant transconjugants per donor) to the conjugative frequency (gentamicin-resistant transconjugants per donor cell). In those assays where no transconjugants were observed, the resolution of experimental sensitivity is given. Average values from three independent experiments, performed in duplicate, are presented.</p></sec></sec>
|
Tapping the nucleotide pool of the host: novel nucleotide carrier proteins of <italic>Protochlamydia amoebophila</italic>
|
<p><italic>Protochlamydia amoebophila</italic> UWE25 is related to the <italic>Chlamydiaceae</italic> comprising major pathogens of humans, but thrives as obligate intracellular symbiont in the protozoan host <italic>Acanthamoeba</italic> sp. The genome of <italic>P. amoebophila</italic> encodes five paralogous carrier proteins belonging to the nucleotide transporter (NTT) family. Here we report on three <italic>P. amoebophila</italic> NTT isoforms, <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, which possess several conserved amino acid residues known to be critical for nucleotide transport. We demonstrated that these carrier proteins are able to transport nucleotides, although substrate specificities and mode of transport differ in an unexpected manner and are unique among known NTTs. <italic>Pam</italic>NTT2 is a counter exchange transporter exhibiting submillimolar apparent affinities for all four RNA nucleotides, <italic>Pam</italic>NTT3 catalyses an unidirectional proton-coupled transport confined to UTP, whereas <italic>Pam</italic>NTT5 mediates a proton-energized GTP and ATP import. All NTT genes of <italic>P. amoebophila</italic> are transcribed during intracellular multiplication in acanthamoebae. The biochemical characterization of all five NTT proteins from <italic>P. amoebophila</italic> in this and previous studies uncovered that these metabolically impaired bacteria are intimately connected with their host cell’s metabolism in a surprisingly complex manner.</p>
|
<contrib contrib-type="author"><name><surname>Haferkamp</surname><given-names>Ilka</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Schmitz-Esser</surname><given-names>Stephan</given-names></name><xref ref-type="aff" rid="au2">2</xref></contrib><contrib contrib-type="author"><name><surname>Wagner</surname><given-names>Michael</given-names></name><xref ref-type="aff" rid="au2">2</xref></contrib><contrib contrib-type="author"><name><surname>Neigel</surname><given-names>Nadjeschka</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Horn</surname><given-names>Matthias</given-names></name><xref ref-type="aff" rid="au2">2</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib><contrib contrib-type="author"><name><surname>Neuhaus</surname><given-names>H Ekkehard</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><aff id="au1"><label>1</label><institution>Pflanzenphysiologie, Technische Universität Kaiserslautern</institution><addr-line>Erwin Schrödinger Str., D-67653 Kaiserslautern, Germany</addr-line></aff><aff id="au2"><label>2</label><institution>Department für Mikrobielle Ökologie, Universität Wien</institution><addr-line>A-1090 Vienna, Austria</addr-line></aff>
|
Molecular Microbiology
|
<sec><title>Introduction</title><p>Survival and reproduction of obligate intracellular bacteria take place within eukaryotic host cells. This cellular association depends upon metabolic interactions between both organisms, which may also be crucial for reproduction, survival and pathogenicity of the host (<xref ref-type="bibr" rid="b11">Görtz and Brigge, 1998</xref>; <xref ref-type="bibr" rid="b39">Taylor and Hoerauf, 2001</xref>; <xref ref-type="bibr" rid="b47">Zientz <italic>et al</italic>., 2004</xref>). In general the obligate intracellular lifestyle correlates with a drastically reduced genome of the bacteria, typically showing defects in essential biosynthetic pathways (<xref ref-type="bibr" rid="b27">Moran, 2002</xref>) and compensatory transport mechanisms for the import of key metabolites from the host cytosol (<xref ref-type="bibr" rid="b28">Moulder, 1991</xref>).</p><p>For example, members of the family <italic>Chlamydiaceae,</italic> comprising exclusively major obligate intracellular pathogens of humans like <italic>Chlamydia trachomatis</italic>, lost the ability to synthesize nucleotides <italic>de novo</italic>, and it has been shown that these essential building blocks are instead imported from the host cell’s cytosol. This metabolic interaction, commonly referred to as energy parasitism, is mediated by nucleotide transport proteins (NTT), which have so far only been identified in few obligate intracellular bacteria and plastids of plants (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b46">Winkler and Neuhaus, 1999</xref>; <xref ref-type="bibr" rid="b23">Linka <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>), and which have been grouped into the ATP:ADP antiporter (AAA) family (<xref ref-type="bibr" rid="b35">Saier, 2000</xref>; <xref ref-type="bibr" rid="b34">Ren <italic>et al</italic>., 2004</xref>). <italic>C. trachomatis</italic> possesses two NTT isoforms: <italic>Ct</italic>NTT1 catalyses the import of host-derived ATP in strict counter exchange with bacterial ADP, whereas <italic>Ct</italic>NTT2 catalyses an unidirectional, proton-driven uptake of all four RNA nucleoside triphosphates (<xref ref-type="bibr" rid="b37">Stephens <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>).</p><p>The recent identification of bacteria related to the <italic>Chlamydiaceae</italic> in the environment revealed a previously underestimated diversity and an unexpected distribution of chlamydiae in nature (<xref ref-type="bibr" rid="b3">Birtles <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b30">Ossewaarde and Meijer, 1999</xref>; <xref ref-type="bibr" rid="b8">Fritsche <italic>et al</italic>., 2000</xref>; <xref ref-type="bibr" rid="b19">Horn <italic>et al</italic>., 2000</xref>; <xref ref-type="bibr" rid="b18">Horn and Wagner, 2001</xref>). For example, the newly identified environmental chlamydia strain <italic>Protochlamydia amoebophila</italic> UWE25 resides as an endosymbiont in the ubiquitous protozoan host <italic>Acanthamoeba</italic> sp. (<xref ref-type="bibr" rid="b8">Fritsche <italic>et al</italic>., 2000</xref>; <xref ref-type="bibr" rid="b5">Collingro <italic>et al</italic>., 2005a</xref>). Like their pathogenic counterparts, environmental chlamydiae are obligate intracellular bacteria, and possess an unique developmental cycle alternating between an infectious, metabolically inert form (the elementary body) and a replicative, metabolically active form (the reticulate body). Genome sequencing of <italic>P. amoebophila</italic> revealed that it contains, in contrast to members of the <italic>Chlamydiaceae</italic>, in total five genes with sequence similarity to known NTT genes (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>). These analyses suggested that the presence of NTTs is an ancient chlamydial trait and it has been hypothesized that other bacterial human pathogens and plants received <italic>ntt</italic> genes from a chlamydial ancestor via horizontal gene transfer (<xref ref-type="bibr" rid="b12">Greub and Raoult, 2003</xref>; <xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>).</p><p>Three of the five NTT proteins from <italic>P. amoebophila</italic> have so far been analysed on the functional level, namely <italic>Pam</italic>NTT1, <italic>Pam</italic>NTT2 and <italic>Pam</italic>NTT4. <italic>Pam</italic>NTT1 represents an ATP/ADP counter exchanger (<xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>) similar to corresponding transporters in <italic>C. trachomatis, Rickettsia prowazekii, Caedibacter caryophilus</italic> or <italic>Holospora obtusa</italic> (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b23">Linka <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b7">Daugherty <italic>et al</italic>., 2004</xref>). <italic>Pam</italic>NTT2 transports all four RNA nucleotides, and most remarkably, <italic>Pam</italic>NTT4 has been identified as the first carrier protein able to transport nicotineamide adenine dinucleotide (NAD<sup>+</sup>) in an intact form across a biological membrane (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>).</p><p>In this work we expressed the so far not characterized NTT genes from <italic>P. amoebophila</italic> (<italic>ntt3</italic> and <italic>ntt5</italic>) heterologously in <italic>Escherichia coli</italic>, as no genetic system is available currently for chlamydiae. We identified the biochemical properties of the corresponding transporter proteins, and extended our previous analysis of the transport properties of <italic>Pam</italic>NTT2. The elucidation of the biochemical characteristics of all five NTT proteins from <italic>P. amoebophila</italic> allowed us to develop a metabolic interaction scenario, which embeds these carriers into biosynthetic abilities and restrains of these elusive intracellular bacteria. These analyses revealed a unique adaptation of <italic>P. amoebophila</italic> to its intracellular lifestyle.</p></sec><sec><title>Results</title><sec><title>Primary and secondary structure of NTT2, NTT3 and NTT5 from <italic>P. amoebophila</italic></title><p>A detailed analysis of primary and predicted secondary structures of NTT proteins can help to get a deeper insight into structure/function relationships of these transport proteins. <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 show sizes of 489–536 amino acids that are comparable to other known NTT type nucleotide transporters (<xref ref-type="supplementary-material" rid="SD1">Fig. S1</xref>). A number of highly conserved amino acid residues critical for the function of the ATP/ADP transporter NTT1 from <italic>Arabidopsis thaliana</italic> (<xref ref-type="bibr" rid="b44">Trentmann <italic>et al</italic>., 2000</xref>), namely the amino acid residues K65, E155, E303 and K446 (counted on basis of <italic>Pam</italic>NTT1), are present in all three <italic>P. amoebophila</italic> NTT proteins analysed in this study (<xref ref-type="supplementary-material" rid="SD1">Fig. S1</xref>). Among characterized NTT proteins, <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 exhibit comparable amino acid sequence identity to bacterial ATP/ADP antiporters and H<sup>+</sup>/nucleotide symporters (33–48% for <italic>Pam</italic>NTT2 and <italic>Pam</italic>NTT3, 28–32% for <italic>Pam</italic>NTT5). Consistently, in phylogenetic trees containing all known NTT proteins <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 formed deep branches between the ATP/ADP translocases from rickettsiae, chlamydiae, plant plastids and other intracellular bacteria, and the proton-driven nucleotide importers of chlamydiae (<xref ref-type="supplementary-material" rid="SD1">Fig. S2</xref>). They thus could not be clearly assigned to a specific NTT subgroup. Like all other known NTT proteins (except for <italic>Pam</italic>NTT4) <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 show (11–) 12 predicted transmembrane domains, a characteristic feature of members of the solute transporter family (<xref ref-type="supplementary-material" rid="SD1">Fig. S3</xref>; <xref ref-type="bibr" rid="b35">Saier, 2000</xref>).</p></sec><sec><title>Preferred substrates of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5</title><p>To reveal the transport properties of the remaining two transporters (<italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5) we cloned the corresponding genes, expressed these heterologously in <italic>E. coli</italic> and measured nucleotide transport mediated by the recombinant carrier proteins. It has been shown that this heterologous expression system allows to decipher biochemical properties of a wide range of NTT proteins from both, bacteria and plants, which are similar to those of the authentic carriers in their respective native membrane (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b26">Möhlmann <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b41">Tjaden <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b42">1999</xref>). In addition to <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, we also included a more detailed characterization of <italic>Pam</italic>NTT2, as this protein has so far only been characterized partially (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>).</p><p>To get a first impression on the spectrum of nucleotides transported by <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, we initially ensured (by sodium dodecyl-lauryl-sulphate-polyacrylamide gel analysis and autoradiography of [<sup>35</sup>S]-labelled NTT proteins) that a sufficient amount of these proteins was inserted into the <italic>E. coli</italic> membrane after heterologous synthesis (data not shown). Subsequently, we raised first evidence on preferred substrates of each carrier. This was done by incubating <italic>E. coli</italic> expressing <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 or <italic>Pam</italic>NTT5 in buffer containing radioactively labelled ATP, UTP, GTP or CTP (each at a concentration of 50 µM, <xref ref-type="table" rid="tbl1">Table 1</xref>).</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><p>Uptake of [α-<sup>32</sup>P]-labelled nucleotides into <italic>E. coli</italic> cells expressing <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 or <italic>Pam</italic>NTT5.</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th colspan="3" rowspan="1">Rate of Transport (nmol mg<sup>−1</sup> protein h<sup>−1</sup>)</th></tr><tr><th rowspan="1" colspan="1"/><th colspan="3" rowspan="1"><hr/></th></tr><tr><th align="left" rowspan="1" colspan="1">Nucleotide</th><th align="left" rowspan="1" colspan="1"><italic>Pam</italic>NTT2</th><th align="left" rowspan="1" colspan="1"><italic>Pam</italic>NTT3</th><th align="left" rowspan="1" colspan="1"><italic>Pam</italic>NTT5</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">ATP</td><td align="char" char="." rowspan="1" colspan="1"> 38.5 ± 1.8</td><td align="char" char="." rowspan="1" colspan="1">< 0.02</td><td align="char" char="." rowspan="1" colspan="1">7.0 ± 0.5</td></tr><tr><td align="left" rowspan="1" colspan="1">GTP</td><td align="char" char="." rowspan="1" colspan="1"> 95.6 ± 4.1</td><td align="char" char="." rowspan="1" colspan="1">< 0.02</td><td align="char" char="." rowspan="1" colspan="1">63.2 ± 5.2</td></tr><tr><td align="left" rowspan="1" colspan="1">CTP</td><td align="char" char="." rowspan="1" colspan="1">145.3 ± 14.6</td><td align="char" char="." rowspan="1" colspan="1">< 0.01</td><td align="char" char="." rowspan="1" colspan="1">< 1.0</td></tr><tr><td align="left" rowspan="1" colspan="1">UTP</td><td align="char" char="." rowspan="1" colspan="1"> 36.2 ± 5.1</td><td align="char" char="." rowspan="1" colspan="1">11.4 ± 1.0</td><td align="char" char="." rowspan="1" colspan="1">< 1.0</td></tr></tbody></table><table-wrap-foot><fn><p>Radioactively labelled compounds were present at a final concentration of 50 µM. IPTG-induced <italic>E. coli</italic> cells harbouring the plasmid pET16b containing the respective nucleotide transporter gene were incubated for 5 min. For calculation of the transport rate control values (induced <italic>E. coli</italic> cells harbouring the plasmid without insert) were subtracted from the measured transport values. Data given in nmol mg total <italic>E. coli</italic> protein<sup>−1</sup> h<sup>−1</sup> are mean of three independent experiments.</p></fn></table-wrap-foot></table-wrap><p>At a substrate concentration of 50 µM <italic>Pam</italic>NTT2 transported CTP most efficiently at a rate of 145.3 nmol mg<sup>−1</sup> protein h<sup>−1</sup>, followed by GTP (95.6 nmol mg<sup>−1</sup> protein h<sup>−1</sup>). <italic>Pam</italic>NTT2-mediated ATP and UTP transport occurred with similar rates, namely 38.5 and 36.2 nmol mg<sup>−1</sup> protein h<sup>−1</sup> respectively (<xref ref-type="table" rid="tbl1">Table 1</xref>). <italic>Pam</italic>NTT3 did not accept nucleotides other than UTP, which is imported at a rate of 11.4 nmol mg<sup>−1</sup> protein h<sup>−1</sup> (<xref ref-type="table" rid="tbl1">Table 1</xref>). <italic>E. coli</italic> cells expressing <italic>Pam</italic>NTT5 transported GTP at a rate of 63.2 nmol mg<sup>−1</sup> protein h<sup>−1</sup>, whereas ATP is transported with a ninefold reduced transport rate (6.98 nmol mg<sup>−1</sup> protein h<sup>−1</sup>). CTP and UTP are not transported by <italic>Pam</italic>NTT5 at substantial rates (<xref ref-type="table" rid="tbl1">Table 1</xref>).</p><p>A more detailed characterization of transport properties, however, requires the analysis of the time linearity of import, as only the time-linear phase of uptake allows for the calculation of apparent affinities and apparent maximal velocities, as well as the identification of effective inhibitors. <italic>Pam</italic>NTT2-mediated ATP import is rapid and occurred linear with time for the first 5 min of incubation (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>). Already after 10 min ATP transport equilibrated and no further accumulation of radioactivity within the bacterial cells occurred (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>). <italic>Pam</italic>NTT3-mediated UTP import is linear with time for about 10 min and saturated after more than 15 min of incubation (<xref ref-type="fig" rid="fig01">Fig. 1B</xref>), whereas ATP import by <italic>Pam</italic>NTT5 is comparably slow and appeared to be linear with time for 10–15 min of incubation (<xref ref-type="fig" rid="fig01">Fig. 1C</xref>). In all cases non-induced control cells hardly imported any labelled nucleotides (<xref ref-type="fig" rid="fig01">Fig. 1A–C</xref>), which is consistent with our previous observations on <italic>E. coli</italic> transformed with expression plasmids harbouring other NTT genes (<xref ref-type="bibr" rid="b41">Tjaden <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b42">1999</xref>; <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>).</p><fig id="fig01" position="float"><label>Fig. 1</label><caption><p>Time dependency of [α-<sup>32</sup>P]-labelled nucleotide uptake into IPTG-induced <italic>E. coli</italic> cells. <italic>E. coli</italic> cells harbouring plasmid pET16b encoding respective NTT gene from <italic>P. amoebophila</italic> (▪), or without insert [control (^)] were incubated in phosphate buffer medium containing 50 µM labelled nucleotide for the indicated time periods.</p><p>A. Time dependency of [α-<sup>32</sup>P]-ATP uptake mediated by recombinant <italic>Pam</italic>NTT2 protein.</p><p>B. Time dependency of [α-<sup>32</sup>P]-UTP uptake mediated by recombinant <italic>Pam</italic>NTT3 protein.</p><p>C. Time dependency of [α-<sup>32</sup>P]-ATP uptake mediated by recombinant <italic>Pam</italic>NTT5.</p></caption><graphic xlink:href="mmi060-1534-f1"/></fig><p>The members of the NTT protein family analysed so far accept a wide array of substrates comprising either ATP and ADP (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b41">Tjaden <italic>et al</italic>., 1998</xref>), all four types of RNA nucleotides (<xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>), or NAD<sup>+</sup> and ADP (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>) respectively. Having identified transported substrates of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 (<xref ref-type="fig" rid="fig01">Fig. 1A–C</xref>, <xref ref-type="table" rid="tbl1">Table 1</xref>) it is possible to identify further putative substrates by competition experiments. For this analysis we studied the effect of various structurally related compounds on <italic>Pam</italic>NTT2- and <italic>Pam</italic>NTT5-mediated ATP transport, and on <italic>Pam</italic>NTT3-mediated UTP transport (<xref ref-type="table" rid="tbl2">Table 2</xref>). NTT2-catalysed ATP transport is inhibited by a 10-fold excess of unlabelled ATP, GTP, CTP and UTP, and inhibition ranged from 32.2% to 57.6% residual activity (<xref ref-type="table" rid="tbl2">Table 2</xref>). All other substrates tested (including TTP, but with the exception of dGTP) did not interfere notably with <italic>Pam</italic>NTT2-mediated ATP import (<xref ref-type="table" rid="tbl2">Table 2</xref>). In contrast, <italic>Pam</italic>NTT3-mediated UTP import is inhibited by unlabelled UTP, UDP, UMP and IDP (42.7–78.2% residual activity), whereas all other substrates did not interfere markedly with UTP uptake by <italic>Pam</italic>NTT3 (<xref ref-type="table" rid="tbl2">Table 2</xref>). <italic>Pam</italic>NTT5-catalysed ATP import is strongly inhibited by unlabelled GTP, GDP, dGTP (13.5–34.3% residual activity) and lesser by ATP, ADP, IDP and GMP (43.3–73.2% residual activity, <xref ref-type="table" rid="tbl2">Table 2</xref>). Interestingly, the presence of NAD<sup>+</sup> in the transport medium resulted in a slight increase of nucleotide uptake mediated by all three transporters (<xref ref-type="table" rid="tbl2">Table 2</xref>). This is probably due to the fact that <italic>E. coli</italic> is able to utilize exogenously supplied NAD<sup>+</sup> via the salvage pathway (<xref ref-type="bibr" rid="b10">Gholson <italic>et al</italic>., 1969</xref>) and that the imported cleavage products most likely improved bacterial fitness.</p><table-wrap id="tbl2" position="float"><label>Table 2</label><caption><p>Effects of various metabolites on [α-<sup>32</sup>P]-UTP (<italic>Pam</italic>NTT3) and [α-<sup>32</sup>P]-ATP (<italic>Pam</italic>NTT2, <italic>Pam</italic>NTT5) uptake into IPTG-induced <italic>E. coli</italic> cells</p></caption><table frame="hsides" rules="groups"><tbody><tr><td align="left" rowspan="1" colspan="1"><graphic xlink:href="mmi060-1534-t2"/></td></tr></tbody></table></table-wrap></sec><sec><title>Substrate affinities and mode of transport of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5</title><p>To interpret the function of a transport protein within the physiological context of a cell, it is important to know its apparent substrate affinities. Therefore, we analysed the change of nucleotide transport velocities of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 in response to different substrate concentrations.</p><p>The apparent affinities of <italic>Pam</italic>NTT2 were 437 µM for ATP, 676 µM for UTP, 156 µM for GTP and 570 µM for CTP. Among these substrates, CTP is transported with the highest apparent maximal velocity (1920 nmol mg<sup>−1</sup> protein h<sup>−1</sup>), which is four to five times faster than for the other substrates (<xref ref-type="table" rid="tbl3">Table 3</xref>). The protonophore CCCP, added at a concentration of 50–250 µM, did not inhibit ATP uptake by <italic>Pam</italic>NTT2 (<xref ref-type="fig" rid="fig02">Fig. 2</xref>), which is consistent with properties of other nucleotide antiporters of the NTT family (<xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>).</p><table-wrap id="tbl3" position="float"><label>Table 3</label><caption><p>Apparent K<sub>M</sub> and V<sub>max</sub> values of the five recombinant NTT proteins from <italic>P. amoebophila</italic> and reported nucleotide concentrations in eukaryotic cells and other bacteria</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">ATP</th><th align="left" rowspan="1" colspan="1">ADP</th><th align="left" rowspan="1" colspan="1">GTP</th><th align="left" rowspan="1" colspan="1">CTP</th><th align="left" rowspan="1" colspan="1">UTP</th><th align="left" rowspan="1" colspan="1">NAD</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Nucleotide concentration in eukaryotic cells (mM)</td><td align="left" rowspan="1" colspan="1">1.05–4.6</td><td align="left" rowspan="1" colspan="1">0.47</td><td align="left" rowspan="1" colspan="1">0.47</td><td align="left" rowspan="1" colspan="1">0.28</td><td align="left" rowspan="1" colspan="1">0.05–0.57</td><td align="left" rowspan="1" colspan="1">1.07</td></tr><tr><td align="left" rowspan="1" colspan="1">K<sub>M</sub> (V<sub>max</sub>) NTT1</td><td align="left" rowspan="1" colspan="1">95 (384)</td><td align="left" rowspan="1" colspan="1">55 (384)</td><td align="left" rowspan="1" colspan="1">128 (12)</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td></tr><tr><td align="left" rowspan="1" colspan="1">K<sub>M</sub> (V<sub>max</sub>) NTT2</td><td align="left" rowspan="1" colspan="1">437 (450)</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">156 (412)</td><td align="left" rowspan="1" colspan="1">570 (1920)</td><td align="left" rowspan="1" colspan="1">676 (501)</td><td align="left" rowspan="1" colspan="1">–</td></tr><tr><td align="left" rowspan="1" colspan="1">K<sub>M</sub> (V<sub>max</sub>) NTT3</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">1320 (321)</td><td align="left" rowspan="1" colspan="1">–</td></tr><tr><td align="left" rowspan="1" colspan="1">K<sub>M</sub> (V<sub>max</sub>) NTT4</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">275 (67)</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">15 (188)</td></tr><tr><td align="left" rowspan="1" colspan="1">K<sub>M</sub> (V<sub>max</sub>) NTT5</td><td align="left" rowspan="1" colspan="1">360 (58)</td><td align="left" rowspan="1" colspan="1">273 (4)</td><td align="left" rowspan="1" colspan="1">22 (98)</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td><td align="left" rowspan="1" colspan="1">–</td></tr><tr><td align="left" rowspan="1" colspan="1">Nucleotide concentration in bacteria (mM)</td><td align="left" rowspan="1" colspan="1">2.7–3.0</td><td align="left" rowspan="1" colspan="1">0.25</td><td align="left" rowspan="1" colspan="1">0.9–1.1</td><td align="left" rowspan="1" colspan="1">0.5–0.7</td><td align="left" rowspan="1" colspan="1">0.89–1.4</td><td align="left" rowspan="1" colspan="1">0.17–1.1</td></tr></tbody></table><table-wrap-foot><fn><p>K<sub>M</sub> values are given in µM, V<sub>max</sub> values (in brackets) are given in nmol mg protein<sup>−1</sup> h<sup>−1</sup>. Data of high affinity transport (< 160 µM) are the mean of four, and data of low affinity transport (> 160 µM) are the mean of eight independent experiments. K<sub>M</sub> and V<sub>max</sub> values for <italic>Pam</italic>NTT1 and <italic>Pam</italic>NTT4 were taken from <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>. (2004)</xref> and <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>. (2004)</xref> respectively. Intracellular nucleotide concentrations were taken from the following references: (<xref ref-type="bibr" rid="b33">Pannbacker, 1967</xref>; <xref ref-type="bibr" rid="b25">Mathews, 1972</xref>; <xref ref-type="bibr" rid="b4">Bochner and Ames, 1982</xref>; <xref ref-type="bibr" rid="b43">Traut, 1994</xref>; <xref ref-type="bibr" rid="b22">Larsson <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b40">Theobald <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b32">Pan <italic>et al</italic>., 2001</xref>; <xref ref-type="bibr" rid="b45">Varela-Gomez <italic>et al</italic>., 2004</xref>).</p></fn></table-wrap-foot></table-wrap><fig id="fig02" position="float"><label>Fig. 2</label><caption><p>Effect of the protonophore CCCP on [α-<sup>32</sup>P]-UTP (<italic>Pam</italic>NTT3) and [α-<sup>32</sup>P]-ATP (<italic>Pam</italic>NTT2, <italic>Pam</italic>NTT5) uptake into IPTG-induced <italic>E. coli</italic> cells. Uptake of [α-<sup>32</sup>P]-ATP mediated by recombinant <italic>Pam</italic>NTT2 or <italic>Pam</italic>NTT5, and of [α-<sup>32</sup>P]-UTP mediated by recombinant <italic>Pam</italic>NTT3 was measured at a substrate concentration of 50 µM and 200 µM respectively. Rates of nucleotide uptake are given as percentage of control rates (non-affected transport = 100%). Data are the mean of three independent experiments.</p></caption><graphic xlink:href="mmi060-1534-f2"/></fig><p><italic>Pam</italic>NTT3 exhibited an apparent affinity for UTP of 1320 µM and a maximal transport velocity of 321 nmol mg<sup>−1</sup> protein h<sup>−1</sup> (<xref ref-type="table" rid="tbl3">Table 3</xref>). In contrast to <italic>Pam</italic>NTT2, CCCP acted inhibitory on <italic>Pam</italic>NTT3. The presence of this protonophore at a concentration of 50 µM inhibited UTP transport by <italic>Pam</italic>NTT3 to 31.4% (SE ± 1.6) of the corresponding control rate (<xref ref-type="fig" rid="fig02">Fig. 2</xref>).</p><p><italic>Pam</italic>NTT5 exhibited highest apparent affinity for GTP (22 µM) and GDP (59 µM), followed by dGTP (121 µM), and ATP (360 µM). Similar to the apparent affinities, the maximal transport velocities catalysed by <italic>Pam</italic>NTT5 were extremely different between the substrates tested. They ranged from only 4 nmol mg<sup>−1</sup> protein h<sup>−1</sup> for ADP <xref ref-type="table" rid="tbl3">Table 3</xref>) to nearly 160 nmol mg<sup>−1</sup> protein h<sup>−1</sup> for GDP. In contrast to <italic>Pam</italic>NTT2, but similar to <italic>Pam</italic>NTT3, CCCP inhibited <italic>Pam</italic>NTT5-mediated nucleotide transport markedly. The presence of 50 µM CCCP inhibited <italic>Pam</italic>NTT5-mediated ATP import to 34.4% (SE ± 2.3) residual activity.</p><p>The inhibitory effect of the protonophore CCCP on NTT-mediated nucleotide movement provides further indications on driving forces required to energize the corresponding transport process. However, we have to keep in mind that certain members of the NTT protein family do not catalyse a counter exchange mode of transport, but unidirectional nucleotide import (<xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>). One often used experimental system to distinguish between counter exchange and unidirectional nucleotide import is the so-called back exchange approach. For this experimental approach <italic>E. coli</italic> expressing either <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 or <italic>Pam</italic>NTT5 were pre-loaded with labelled substrate, external radioactivity was removed by washing with buffer medium, and the cells were finally resuspended in phosphate buffer (control), or transport medium (buffer containing putative counter exchange nucleotides). By analysing the time-course of the loss of radioactive label it is possible to differentiate between counter exchange, or unidirectional mode of transport (<xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>).</p><p><italic>E. coli</italic> cells expressing <italic>Pam</italic>NTT2 and pre-loaded with [α<sup>32</sup>P]-ATP released significant amounts of internal label only after resuspension in buffer medium supplemented with unlabelled ATP compared with the control (<xref ref-type="fig" rid="fig03">Fig. 3A</xref>). We verified by thin-layer chromatography that radioactivity exported in the presence of the unlabelled counter exchange substrates CTP, GTP, ATP and UTP is mainly [α<sup>32</sup>P]-ATP (<xref ref-type="fig" rid="fig04">Fig. 4</xref>).</p><fig id="fig04" position="float"><label>Fig. 4</label><caption><p>Thin-layer chromatography analysis and quantification of exported radioactively labelled nucleotides by <italic>E. coli</italic> cells expressing <italic>Pam</italic>NTT2. <italic>E. coli</italic> cells expressing <italic>Pam</italic>NTT2 were incubated for 3 min in phosphate buffer containing 50 µM radioactively labelled [á-<sup>32</sup>P]-ATP. After removal of external radioactivity, back exchange was initiated by addition of indicated unlabelled nucleotides at a concentration of 500 µM each, and stopped after 2 min.</p><p>A. Thin-layer chromatography analysis of exported nucleotides.</p><p>B. Quantification of exported radioactive ATP and ADP.Spots of exported radioactively labelled ATP (black columns), and ADP (grey columns) separated by thin-layer chromatography were excised, transferred to scintillation vessels, and radioactivity was quantified in a scintillation counter. Data in (A) represent a typical experiment, data in (B) are the mean of four independent experiments. Standard errors are displayed.</p></caption><graphic xlink:href="mmi060-1534-f4"/></fig><fig id="fig03" position="float"><label>Fig. 3</label><caption><p>Back exchange properties of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5. <italic>E. coli</italic> cells synthesizing <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 or <italic>Pam</italic>NTT5 respectively, were pre-loaded with 50 µM [á-<sup>32</sup>P]-labelled nucleotide and back exchange was initiated by resuspension of washed cells in either phosphate buffer medium [control (•)] or phosphate buffer containing putative counter exchange substrate (▪).</p><p>A. <italic>Pam</italic>NTT2 expressing cells were pre-loaded with [á-<sup>32</sup>P]-ATP, unlabelled ATP for back exchange was present at a concentration of 2 mM.</p><p>B. <italic>Pam</italic>NTT3 expressing cells were pre-loaded with [á-<sup>32</sup>P]-UTP, unlabelled UTP for back exchange was present at a concentration of 3 mM.</p><p>C. <italic>Pam</italic>NTT5 expressing cells were pre-loaded with [á-<sup>32</sup>P]-GTP, unlabelled GTP for back exchange was present at a concentration of 2 mM.</p><p>Data represent the mean of three independent experiments. Standard error of the mean was always less than 5% of the given value.</p></caption><graphic xlink:href="mmi060-1534-f3"/></fig><p><italic>Escherichia coli</italic> cells expressing <italic>Pam</italic>NTT3 and pre-loaded with [α<sup>32</sup>P]-UTP did not release additional internal radioactivity after resuspension in buffer medium containing unlabelled UTP when compared with the control (<xref ref-type="fig" rid="fig03">Fig. 3B</xref>). Similarly, <italic>Pam</italic>NTT5-expressing <italic>E. coli</italic> cells pre-loaded with [α<sup>32</sup>P]-GTP did not export internal radioactive label after resuspension in medium containing unlabelled GTP (<xref ref-type="fig" rid="fig03">Fig. 3C</xref>).</p></sec><sec><title>Transcription of NTT genes from <italic>P. amoebophila</italic></title><p>It has been shown previously that the transporter genes <italic>ntt1</italic>, <italic>ntt2</italic> and <italic>ntt4</italic> of <italic>P. amoebophila</italic> are transcribed during multiplication in its amoeba host (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>). In order to verify whether also <italic>ntt3</italic> and <italic>ntt5</italic> of <italic>P. amoebophila</italic> are transcribed, we performed reverse transcriptase PCR experiments (<xref ref-type="fig" rid="fig03">Fig. 3</xref>). For this total RNA purified from amoebae containing intracellular <italic>P. amoebophila</italic> was used as template, and gene-specific primers allowed amplification of <italic>ntt3</italic> and <italic>ntt5</italic> transcript fragments showing the expected sizes (<xref ref-type="fig" rid="fig05">Fig. 5</xref>). The identity of the amplified fragments was confirmed by sequencing.</p><fig id="fig05" position="float"><label>Fig. 5</label><caption><p>Transcription of <italic>ntt3</italic> and <italic>ntt5</italic> during multiplication of <italic>P. amoebophila</italic> within acanthamoebae.</p><p>A. Transcription of <italic>ntt3</italic>.</p><p>B. Transcription of <italic>ntt5</italic>.</p><p>Lane 1, amplification products of cDNA synthesized from whole RNA from amoebae harbouring <italic>P. amoebophila</italic>; lane 2, PCR positive controls using DNA from purified <italic>P. amoebophila</italic> elementary bodies; lane 3, PCR negative controls (no cDNA added); lane 4, PCR using whole RNA from amoebae containing <italic>P. amoebophila</italic> (control for the absence of DNA in the RNA preparation, no reverse transcriptase added); m, molecular-size marker.</p></caption><graphic xlink:href="mmi060-1534-f5"/></fig></sec></sec><sec><title>Discussion</title><p>NTT type nucleotide transport proteins occur in few intracellular bacteria and in all plastids from higher and lower plants analysed so far (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b41">Tjaden <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b46">Winkler and Neuhaus, 1999</xref>; <xref ref-type="bibr" rid="b23">Linka <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>), but they do not share any structural similarity to mitochondrial and peroxisomal adenylate transporters belonging to the mitochondrial carrier (MC) family (<xref ref-type="bibr" rid="b20">Klingenberg, 1989</xref>; <xref ref-type="bibr" rid="b35">Saier, 2000</xref>; <xref ref-type="bibr" rid="b31">Palmieri <italic>et al</italic>., 2001</xref>; <xref ref-type="bibr" rid="b34">Ren <italic>et al</italic>., 2004</xref>). However, like mitochondrial and peroxisomal ATP carriers, NTT type nucleotide transport proteins catalyse a remarkable reaction by carrying nucleotide cargos, which are extremely large and highly charged substrates.</p><p>In the light of the only recently recognized functional diversity of NTT proteins, we suggest to group all NTT proteins known so far into a new family named nucleotide transporter (NTT) family, replacing the original AAA family (<xref ref-type="bibr" rid="b35">Saier, 2000</xref>; <xref ref-type="bibr" rid="b34">Ren <italic>et al</italic>., 2004</xref>). Taking the mode of transport for classification, the nucleotide transporter (NTT) family can be further subdivided into three classes: (i) class I catalyses a nucleotide counter exchange mode of transport. This transport process allows for example to energize anabolic reactions of endosymbiotic (or parasitic) bacteria exhibiting impaired energy metabolism. Typical members of NTT class I are the homologous NTT1 proteins found in the human pathogens <italic>R. prowazekii</italic> and <italic>C. trachomatis</italic>, or in the symbionts <italic>H. obtusa</italic> and <italic>C. caryophilus</italic> residing in paramecia (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b23">Linka <italic>et al</italic>., 2003</xref>). (ii) Class II NTT proteins catalyse an unidirectional, proton-driven nucleotide import and are thus able to compensate lacking <italic>de novo</italic> nucleotide biosynthesis capacity for example in <italic>C. trachomatis</italic> (<xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>). (iii) Finally, class III NTT proteins catalyse NAD<sup>+</sup>/ADP counter exchange (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>) and thus bypass the inability of <italic>P. amoebophila</italic> to synthesize the important electron carrier and cofactor NAD (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>).</p><p>Using this classification scheme, <italic>Pam</italic>NTT2 analysed in this study belongs to the class I NTT proteins. This assignment is based on the observations that (i) <italic>Pam</italic>NTT2 catalyses a nucleotide counter exchange (<xref ref-type="table" rid="tbl2">Table 2</xref>; <xref ref-type="fig" rid="fig03">Figs 3</xref> and <xref ref-type="fig" rid="fig04">4</xref>) and that (ii) transport activity is not dependent upon an existing proton gradient (<xref ref-type="fig" rid="fig02">Fig. 2</xref>). However, in contrast to the rickettsial adenylate translocase, to NTT1 from <italic>C. trachomatis</italic>, and to all plant NTT proteins (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>; <xref ref-type="bibr" rid="b26">Möhlmann <italic>et al</italic>., 1998</xref>; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>), the carrier <italic>Pam</italic>NTT2 does not accept nucleoside diphosphates as substrate, but is able to transport UTP, CTP, ATP and GTP (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>). Thus, <italic>Pam</italic>NTT2 is a class I NTT protein with a previously not recognized substrate specificity. This finding differs from our earlier assumption that <italic>Pam</italic>NTT2 represents an unidirectional class II transporter (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>) and shows that the exact mode of transport can only be revealed by extended biochemical analyses using uncouplers like CCCP and back-exchange experiments.</p><p><italic>Pam</italic>NTT3 belongs to the class II NTT proteins, because this carrier mediates a unidirectional (<xref ref-type="fig" rid="fig03">Fig. 3B</xref>), proton-energized transport of a nucleoside triphosphate (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>; <xref ref-type="fig" rid="fig01">Figs 1B</xref> and <xref ref-type="fig" rid="fig02">2</xref>). However, similar to <italic>Pam</italic>NTT2 within NTT class I, <italic>Pam</italic>NTT3 differs from <italic>Ct</italic>NTT2, the only other member of class II, with respect to its substrate spectrum. While the <italic>Ct</italic>NTT2 carrier imports all RNA nucleotides (UTP, ATP, GTP and CTP; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>), <italic>Pam</italic>NTT3 solely uses a single substrate, namely UTP (<xref ref-type="table" rid="tbl2">Table 2</xref>). <italic>Pam</italic>NTT5 also belongs to the class II NTT proteins, as this carrier mediates a unidirectional, proton-driven nucleotide transport (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>; <xref ref-type="fig" rid="fig01">Figs 1B</xref>, <xref ref-type="fig" rid="fig02">2</xref> and <xref ref-type="fig" rid="fig03">3C</xref>). However, in opposite to <italic>Pam</italic>NTT3 and <italic>Ct</italic>NTT2, this carrier accepts GTP, GDP and ATP as substrates.</p><p>With the exception of <italic>Pam</italic>NTT4, all NTT proteins from <italic>P. amoebophila</italic> contain a number of amino acid residues highly conserved among known NTT proteins (<xref ref-type="supplementary-material" rid="SD1">Fig. S1</xref>). Taking the plastidic ATP/ADP transporter NTT1 from <italic>A. thaliana</italic> as an example, it has been demonstrated that some of these amino acid residues are critical for transport of adenine nucleotides (<xref ref-type="bibr" rid="b44">Trentmann <italic>et al</italic>., 2000</xref>). For example, the lysine residue K446 (counted on the basis of <italic>Pam</italic>NTT1, and corresponding to K527 in NTT1 of <italic>A. thaliana</italic>) is critical for transport of nucleoside triphosphates, but not for nucleoside diphosphate transport (<xref ref-type="bibr" rid="b44">Trentmann <italic>et al</italic>., 2000</xref>). Consistently, <italic>Pam</italic>NTT4 known to be unable to transport ATP, but capable of ADP transport does not exhibit a lysine residue at this position (<xref ref-type="supplementary-material" rid="SD1">Fig. S1</xref>). However, all other NTT proteins from <italic>P. amoebophila</italic> exhibit this conserved lysine residue (<xref ref-type="supplementary-material" rid="SD1">Fig. S1</xref>) and in fact we could demonstrate that <italic>Pam</italic>NTT1, <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 transport nucleoside triphosphates as expected (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>; <xref ref-type="fig" rid="fig01">Fig. 1</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>). These correlations show that the presence of certain critical amino acid residues are more suitable as signposts for NTT function and substrate specificity than the overall similarity of the amino acid sequences between NTT proteins. However, it is still impossible to predict <italic>in silico</italic> the exact specificity and transport modes of a newly discovered NTT protein underlining the requirement of a more detailed analysis of structure–function relationships of representative NTT proteins in the future.</p><p>The availability of the biochemical characteristics of all five NTT proteins from the symbiont <italic>P. amoebophila</italic> allowed us to develop a metabolic interaction scenario for this bacterium with its amoebal host (<xref ref-type="fig" rid="fig06">Fig. 6</xref>). This model is also based on the observation from this and previous studies that all five NTT proteins of <italic>P. amoebophila</italic> are actually transcribed during intracellular multiplication in its amoeba host (<xref ref-type="fig" rid="fig05">Fig. 5</xref>; <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>), indicating that these five NTT proteins are both functional and necessary for the intracellular lifestyle of <italic>P. amoebophila</italic>. Furthermore, with the exception of <italic>Pam</italic>NTT3 all <italic>P. amoebophila</italic> NTT proteins exhibit apparent substrate affinity (K<sub>M</sub>) values for their respective substrates in the submillimolar range (0.015–0.676 mM<italic>,</italic><xref ref-type="table" rid="tbl3">Table 3</xref>). Although the cytosolic nucleotide concentrations of the <italic>Acanthamoeba</italic> host are unknown<italic>,</italic> it is likely that these concentrations are at comparable levels as reported for various other eukaryotic cells (0.05–4.6 mM, <xref ref-type="table" rid="tbl3">Table 3</xref>; <xref ref-type="bibr" rid="b33">Pannbacker, 1967</xref>; <xref ref-type="bibr" rid="b16">Hauschka, 1973</xref>; <xref ref-type="bibr" rid="b4">Bochner and Ames, 1982</xref>; <xref ref-type="bibr" rid="b38">Stitt <italic>et al</italic>., 1989</xref>; <xref ref-type="bibr" rid="b22">Larsson <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b40">Theobald <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b45">Varela-Gomez <italic>et al</italic>., 2004</xref>). According to these data and according to cytoplasmic nucleotide concentrations reported previously for other bacteria (<xref ref-type="table" rid="tbl3">Table 3</xref>; <xref ref-type="bibr" rid="b25">Mathews, 1972</xref>; <xref ref-type="bibr" rid="b43">Traut, 1994</xref>; <xref ref-type="bibr" rid="b32">Pan <italic>et al</italic>., 2001</xref>), and under the prerequisite that the inclusion membrane that surrounds <italic>P. amoebophila</italic> is permeable for all NTT substrates<italic>,</italic> the apparent substrate affinities of all NTT proteins from <italic>P. amoebophila</italic> should be sufficient to transport their substrates under <italic>in vivo</italic> conditions. The high degree of substrate specificity of <italic>Pam</italic>NTT3 for UTP might compensate for its suboptimal affinity (which is lower than reported intracellular UTP concentrations).</p><fig id="fig06" position="float"><label>Fig. 6</label><caption><p>Model of NTT-mediated metabolic interactions between <italic>P. amoebophila</italic> and its <italic>Acanthamoeba</italic> host. This model is based on the deciphered biochemical properties of all nucleotide transport proteins of <italic>P. amoebophila</italic> determined in this and previous studies (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>), the completely sequenced genome of this species (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>), the observed apparent substrate affinity (K<sub>M</sub>) values, and previously reported intracellular nucleotide concentrations. The combined activity of all five NTT proteins (1–5) allows for the uptake of the electron carrier and cofactor NAD<sup>+</sup>, all nucleotides needed for RNA and DNA synthesis, and for the import of energy in form of ATP. PyrG, CTP synthase (GenBank Accession No. CAF23548); NrdAB, ribonucleotide reductase subunits A and B (CAF24072, CAF24073); Ndk, nucleoside-diphosphate kinase (CAF23090); Cmk, cytidylate kinase (CAF23043); Tmk, thymidylate kinase (CAF23797); ThyX, thymidylate synthase complementing protein (CAF22879).</p></caption><graphic xlink:href="mmi060-1534-f6"/></fig><p>Genome sequence analysis of <italic>P. amoebophila</italic> has revealed that it lacks key genes necessary for the biosynthesis of NAD<sup>+</sup> and thus is unable to synthesize NAD<sup>+</sup><italic>de novo</italic> (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>). Therefore, to import NAD during phases of cell growth and division, <italic>Pam</italic>NTT4 catalyses NAD<sup>+</sup> uptake in exchange for ADP (<xref ref-type="fig" rid="fig06">Fig. 6</xref>; <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>). The ADP molecule, required as a counter exchange substrate for the NAD<sup>+</sup> import, can be provided via <italic>Pam</italic>NTT5-catalysed ATP import and the subsequent energy-providing conversion of ATP to ADP by various anabolic reactions (<xref ref-type="fig" rid="fig06">Fig. 6</xref>).</p><p><italic>Protochlamydia amoebophila</italic> is also unable to synthesize nucleotides <italic>de novo</italic> (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>). Therefore, a net nucleotide uptake for the synthesis of RNA and DNA is necessary, which can be catalysed by a concerted action of <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 (<xref ref-type="fig" rid="fig06">Fig. 6</xref>). Within such concerted action, <italic>Pam</italic>NTT3 imports UTP unidirectional (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>; <xref ref-type="fig" rid="fig03">Fig. 3B</xref>), and <italic>Pam</italic>NTT5 imports GTP and ATP unidirectional (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>; <xref ref-type="fig" rid="fig03">Fig. 3C</xref>). The last missing building block for RNA synthesis is CTP, a compound transported with high affinity and high velocity by <italic>Pam</italic>NTT2 (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>). However, <italic>Pam</italic>NTT2-mediated CTP import would depend upon the presence of a suitable nucleoside triphosphate for counter exchange (<xref ref-type="fig" rid="fig06">Fig. 6</xref>) that might be provided by, e.g. <italic>Pam</italic>NTT3 in form of UTP, or <italic>Pam</italic>NTT5 in form of GTP or ATP (<xref ref-type="table" rid="tbl1">Tables 1</xref> and <xref ref-type="table" rid="tbl2">2</xref>). Because the <italic>P. amoebophila</italic> genome contains a gene coding for a CTP synthase (PyrG; <xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>), this bacterium might alternatively synthesize CTP from UTP. During replication <italic>P. amoebophila</italic> needs deoxynucleotides for DNA synthesis. dATP, dGTP, dCTP and dUMP can be generated from the respective RNA dinucleotides by the help of ribonucleotide reductase (both subunits, NrdA and NrdB, are encoded in its genome), but <italic>P. amoebophila</italic> in addition requires TTP. However, none of the five NTT proteins accepts thymidylates as substrates (<xref ref-type="table" rid="tbl2">Table 2</xref>), and according to its genome sequence <italic>P. amoebophila</italic> lacks a thymidylate synthase (ThyA) gene, generally required by the majority of bacteria for thymidylate synthesis. However, <italic>P. amoebophila</italic> possesses instead a gene encoding the thymidylate synthase complementing protein (ThyX), which uses dUMP and methylenetetrahydrofolate as substrates for dTMP formation, and which has been recently biochemically characterized in the related bacterium <italic>C. trachomatis</italic> (<xref ref-type="bibr" rid="b29">Myllykallio <italic>et al</italic>., 2002</xref>; <xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b14">Griffin <italic>et al</italic>., 2005</xref>). Taken together, the substrate specificity of the three nucleotide transporters <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, and the presence of genes encoding a CTP synthase, a ribonucleotide reductase, the alternative thymidylate synthase ThyX, a nucleoside-diphosphate kinase and specific adenylate, guanylate, cytidylate and thymidylate kinases are sufficient to compensate the lack of a complete nucleotide biosynthesis pathway. Our model thus fully explains the auxotrophy of <italic>P. amoebophila</italic> for nucleotides and illustrates how <italic>P. amoebophila</italic> acquires all precursors for RNA and DNA biosynthesis from its amoeba host cell (<xref ref-type="fig" rid="fig06">Fig. 6</xref>).</p><p>Although <italic>P. amoebophila</italic> is capable of endogenous ATP regeneration via oxidative phosphorylation (<xref ref-type="bibr" rid="b17">Horn <italic>et al</italic>., 2004</xref>), this organism possesses a functional ATP/ADP exchange system in form of <italic>Pam</italic>NTT1 (<xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>). The presence of this transport activity resembles the situation found in <italic>C. trachomatis</italic>, which also possesses an ATP/ADP exchanger (<italic>Ct</italic>NTT1; <xref ref-type="bibr" rid="b42">Tjaden <italic>et al</italic>., 1999</xref>) in addition to an endogenous ATP regeneration system via oxidative phosphorylation (<xref ref-type="bibr" rid="b37">Stephens <italic>et al</italic>., 1998</xref>). Nevertheless, the exchange of host ATP for bacterial ADP is an alternative way to acquire energy and is certainly of benefit for example under conditions where substrate usable for oxidative phosphorylation in the bacterium is limiting in the host cytosol. The upregulation of the gene coding for the ATP/ADP antiporter in <italic>C. trachomatis</italic> during the early phase of the developmental cycle indicates that this is the case during the initial differentiation of elementary bodies to reticulate bodies shortly after the infection of a new host cell (<xref ref-type="bibr" rid="b2">Belland <italic>et al</italic>., 2003</xref>). It thus seems likely that also <italic>P. amoebophila</italic> is an energy parasite at least under certain physiological conditions.</p><p>In this context it is worth mentioning that similar to <italic>P. amoebophila</italic> the only distantly related epidemic typhus causing <italic>R. prowazekii</italic> is also unable to synthesize nucleotides <italic>de novo</italic>, and like <italic>P. amoebophila</italic> also harbours five NTT isoforms (<xref ref-type="bibr" rid="b1">Andersson <italic>et al</italic>., 1998</xref>). It will therefore be interesting to study in addition to <italic>Rp</italic>NTT1 (<xref ref-type="bibr" rid="b21">Krause <italic>et al</italic>., 1985</xref>) the biochemical properties of <italic>Rp</italic>NTT2 to <italic>Rp</italic>NTT5 in the near future and to compare the data with features of the NTT proteins found in <italic>P. amoebophila</italic>.</p><p>In conclusion, we showed in this study that <italic>P. amoebophila</italic> exploits its amoebal host by simultaneously expressing a set of five nucleotide transporters of which four differ from all hitherto characterized NTT proteins regarding substrate specificity or mode of transport. The metabolic model inferred from these data illustrates a previously unseen, tight and apparently well balanced coupling of the endosymbiont and host metabolisms. As environmental chlamydiae are capable to infect mammalian cells (<xref ref-type="bibr" rid="b13">Greub <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b6">Collingro <italic>et al</italic>., 2005b</xref>) and as all human pathogenic chlamydiae also rely on functional NTT proteins, these transporters are not only key to the understanding of the biology of these organisms but should also represent attractive targets for the development of highly specific anti-chlamydial drugs.</p></sec><sec sec-type="methods"><title>Experimental procedures</title><sec><title>Cloning of <italic>ntt</italic>3 and <italic>ntt</italic>5, and construction of expression plasmids</title><p><italic>Protochlamydia amoebophila</italic> UWE25 was cultivated in the host <italic>Acanthamoeba</italic> sp. UWC1 (<xref ref-type="bibr" rid="b9">Gautom and Fritsche, 1995</xref>) as previously described (<xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>). Purification of DNA from amoebae containing the symbiont <italic>P. amoebophila</italic> was performed using the FastDNA-Kit (BIO 101, Carlsbad, CA, USA) according to the protocol recommended by the manufacturer. <italic>ntt3</italic> and <italic>ntt5</italic> were amplified with the Extensor Hi-Fidelity PCR Enzyme Mix (ABgene, Epsom, UK) using forward primers introducing a XhoI restriction site instead of the start codon (5′-TGCACCCTC GAGTCACAGACACCAACAGGG-3′ for <italic>ntt3</italic>, and 5′-TGCA CCCTCGAGAAAAATCAACAAAATTCT-3′ for <italic>ntt5</italic>), and reverse primers containing a BamHI restriction site after the stop codon (5′-TTGGGATCCTTATACAGCTTTCTGTTCTTT CAA-3′ for <italic>ntt3</italic>, and 5′-CACTTGGGATCCTTATCCATGGGA AGCTTCTATAAA-3′ for <italic>ntt5</italic>). The resulting amplification products were digested with BamHI and XhoI, gel purified, and inserted into the expression vector pET16b (Novagen, Heidelberg, Germany). The newly constructed plasmids (pNTT3, pNTT5) were transformed into and maintained in <italic>E. coli</italic> XL1Blue (Stratagene, Heidelberg, Germany). The identity of the cloned genes was checked by sequencing.</p></sec><sec><title>Heterologous expression of PamNTT2, PamNTT3 and PamNTT5 in <italic>E. coli</italic></title><p>For heterologous expression of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5 the <italic>E. coli</italic> strain BLR (DE3) was used. Synthesis of recombinant nucleotide transporters was conducted according to previously reported methods (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>). Briefly, <italic>E. coli</italic> cells harbouring the vector pET16b containing <italic>ntt</italic>2, <italic>ntt</italic>3 and <italic>ntt5</italic>[pNTT2 (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>), pNTT3, pNTT5] were induced by IPTG application to express the respective transporter gene. After 1 h cells were sedimented (3000 <italic>g</italic>, 5 min, at 8°C) and resuspended in 50 mM potassium phosphate buffer medium (pH 7.0) to an OD<sub>600</sub> of 5.0, kept at room temperature, and subsequently used for uptake experiments.</p></sec><sec><title>Uptake of radioactively labelled nucleotides into intact <italic>E. coli</italic> cells</title><p>To analyse transport properties of <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, 100 µl of either induced <italic>E. coli</italic> cells harbouring the corresponding constructs, or non-induced cells (control) were added to 100 µl incubation medium (50 mM P<sub>i</sub> buffer, pH 7.0) containing the [α-<sup>32</sup>P]-labelled substrates ATP, ADP, GTP, GDP, dGTP, CTP and UTP. [α-<sup>32</sup>P]-labelled ADP and also GDP was synthesized as described previously (<xref ref-type="bibr" rid="b41">Tjaden <italic>et al</italic>., 1998</xref>). Cells lacking the expression plasmid, but incubated in the presence of IPTG exhibited the same uptake as the control cells (data not shown).</p><p>Uptake was allowed for indicated time spans, at 30°C, and terminated by removal of external substrate (<xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>). For the latter purpose, cells were applied on 0.45 µm nitrocellulose filters, pre-wetted with incubation medium, and set under vacuum. The cells were immediately washed with 3 × 4 ml incubation medium, and filters were transferred into 20 ml scintillation vessels, containing 10 ml water. Radioactivity in the samples was quantified in a scintillation counter (Tricarb 2500, Canberra-Packard, Heidelberg, Germany).</p></sec><sec><title>Back exchange studies and thin-layer chromatography</title><p>To characterize the transport mode of recombinant <italic>Pam</italic>NTT2, <italic>Pam</italic>NTT3 and <italic>Pam</italic>NTT5, back exchange studies were conducted. IPTG-induced <italic>E. coli</italic> cells synthesizing the recombinant NTT were incubated for an appropriate time span in phosphate buffer containing radioactively labelled [α-<sup>32</sup>P]-ATP, GTP or UTP. Subsequently, cells were collected by centrifugation, washed and resuspended in phosphate buffer medium containing indicated excess of non-labelled counter exchange substrate. Back exchange was carried out at 30°C for the indicated time span, terminated by rapid filtration, and quantified of remaining internal radioactivity in a scintillation counter.</p><p>To identify the molecular nature of the radioactively labelled compounds exported by cells expressing <italic>Pam</italic>NTT2 and pre-loaded with [α-<sup>32</sup>P]-ATP, we employed a thin-layer chromatography method (<xref ref-type="bibr" rid="b24">Mangold, 1967</xref>), routinely used in our laboratory (<xref ref-type="bibr" rid="b44">Trentmann <italic>et al</italic>., 2000</xref>; <xref ref-type="bibr" rid="b15">Haferkamp <italic>et al</italic>., 2004</xref>). For this, <italic>E. coli</italic> cells were collected by centrifugation, a 10 µl aliquot of the export supernatant (see above) was loaded onto a 0.5 mm poly(ethylene amine) cellulose thin-layer chromatography plate and dried with a fan. RF values of radioactively labelled nucleotides and phosphate were determined after radioautography and corresponded to values of unlabelled nucleotides visualized under ultraviolet light, and to radioactively labelled standards.</p></sec><sec><title>Transcriptional analysis</title><p><italic>Protochlamydia amoebophila</italic> UWE25 was grown in the host <italic>Acanthamoeba</italic> sp. UWC1 (<xref ref-type="bibr" rid="b36">Schmitz-Esser <italic>et al</italic>., 2004</xref>). For transcriptional analysis amoebae harbouring <italic>P. amoebophila</italic> were harvested by centrifugation (2350 <italic>g</italic>, 5 min, at 4°C), resuspended in TRIzol (Invitrogen Life Technologies, Lofer, Austria), and immediately homogenized using the Bead-Beater Fast Prep FP120 Instrument (BIO 101, Carlsbad, CA, USA). Whole RNA purification was performed according to the recommendations of the manufacturer, followed by a DNase treatment using Deoxyribonuclease I. A control PCR using 16S rRNA gene-targeted primers and Taq DNA polymerase instead of reverse transcriptase was performed to ensure that the RNA preparation was free of DNA, and 1 µg of RNA was subsequently used for cDNA synthesis using the RevertAid First Strand cDNA Synthesis Kit (MBI-Fermentas, St Leon-Rot, Germany) and a gene-specific primer. cDNA was subsequently used as template in standard PCR reactions. Negative controls (no cDNA added) were included in all PCR reactions. A standard PCR cycling program with 40 cycles, and an annealing temperature of 58°C was used for the amplification of the <italic>ntt3</italic> and <italic>ntt5</italic> cDNA. Primers targeting a 588 bp fragment of <italic>ntt3</italic> (forward primer: 5′-CATCCTCACG CAACTGCTGAC-3′, reverse primer: 5′-GCTCTCGGACTT GATGTTTCC-3′), or a 577 bp fragment of <italic>ntt5</italic> (forward primer: 5′-CACAGCAAGTGATGCAGGTGC-3′, reverse primer: 5′-CAGTTGTAAGAGCCAGCCGTC-3′), respectively, were used for cDNA synthesis and PCR reactions. Amplification products were sequenced to ensure that amplification was specific. All experiments were performed in independent triplicates.</p></sec></sec>
|
Helical distribution of the bacterial chemoreceptor via colocalization with the Sec protein translocation machinery
|
<p>In <italic>Escherichia coli</italic>, chemoreceptor clustering at a cell pole seems critical for signal amplification and adaptation. However, little is known about the mechanism of localization itself. Here we examined whether the aspartate chemoreceptor (Tar) is inserted directly into the polar membrane by using its fusion to green fluorescent protein (GFP). After induction of Tar–GFP, fluorescent spots first appeared in lateral membrane regions, and later cell poles became predominantly fluorescent. Unexpectedly, Tar–GFP showed a helical arrangement in lateral regions, which was more apparent when a Tar–GFP derivative with two cysteine residues in the periplasmic domain was cross-linked to form higher oligomers. Moreover, similar distribution was observed even when the cytoplasmic domain of the double cysteine Tar–GFP mutant was replaced by that of the kinase EnvZ, which does not localize to a pole. Observation of GFP–SecE and a translocation-defective MalE–GFP mutant, as well as indirect immunofluorescence microscopy on SecG, suggested that the general protein translocation machinery (Sec) itself is arranged into a helical array, with which Tar is transiently associated. The Sec coil appeared distinct from the MreB coil, an actin-like cytoskeleton. These findings will shed new light on the mechanisms underlying spatial organization of membrane proteins in <italic>E. coli</italic>.</p>
|
<contrib contrib-type="author"><name><surname>Shiomi</surname><given-names>Daisuke</given-names></name><xref ref-type="aff" rid="au1">1</xref><xref ref-type="author-notes" rid="fn1">†</xref></contrib><contrib contrib-type="author"><name><surname>Yoshimoto</surname><given-names>Masayuki</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Homma</surname><given-names>Michio</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Kawagishi</surname><given-names>Ikuro</given-names></name><xref ref-type="aff" rid="au1">1</xref><xref ref-type="aff" rid="au2">2</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib><aff id="au1"><label>1</label><institution>Division of Biological Science, Graduate School of Science, Nagoya University</institution><addr-line>Chikusa-ku, Nagoya 464-8602, Japan</addr-line></aff><aff id="au2"><label>2</label><institution>Institute for Advanced Research, Nagoya University</institution><addr-line>Chikusa-ku, Nagoya 464-8602, Japan</addr-line></aff>
|
Molecular Microbiology
|
<sec><title>Introduction</title><p>The proteins involved in many biological systems are apt to be spatially organized within a cell or even within a smaller compartment rather than to be freely diffusible and encountering each other stochastically. Proper spatial organization of proteins is of vital importance not only in eukaryotes but also in prokaryotes despite their smaller sizes. Bacterial cellular functions that involve localization of proteins include virulence, cell division and chemotaxis. For example, IcsA of <italic>Shigella flexneri</italic> and ActA of <italic>Listeria monocytogenes</italic>, which nucleate actin polymerization and are required for the bacterial movement within a host cell and hence virulence, localize to an old pole of the rod-shaped bacterial cell (<xref ref-type="bibr" rid="b27">Goldberg <italic>et al.</italic>, 1993</xref>; <xref ref-type="bibr" rid="b53">Smith <italic>et al.</italic>, 1995</xref>). CpaE and CpaC, which are required for polar pilus biogenesis in <italic>Caulobactor crescentus</italic>, localize to one pole and their localization is controlled by PodJ (<xref ref-type="bibr" rid="b57">Viollier <italic>et al.</italic>, 2002</xref>). In <italic>Escherichia coli</italic>, the best characterized bacterial species, many proteins have been shown to localize (for reviews, see <xref ref-type="bibr" rid="b40">Nanninga, 1998</xref>; <xref ref-type="bibr" rid="b34">Lybarger and Maddock, 2001</xref>; <xref ref-type="bibr" rid="b47">Shapiro <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="b37">Margolin, 2003</xref>). However, little is known about mechanisms underlying protein localization. For example, the cell division machinery of <italic>E. coli</italic> has to localize to the mid cell position and this localization is regulated by other proteins (the Min system) (<xref ref-type="bibr" rid="b9">Bi and Lutkenhaus, 1991</xref>; <xref ref-type="bibr" rid="b1">Akerlund <italic>et al.</italic>, 1992</xref>), but it remains to be elucidated whether the localization results directly from interactions with other protein(s), including those between the Min proteins and the division machinery, peptidoglycan (cell wall) or phospholipids. Experimental approaches to examine the mechanisms of their localization are limited, because most of these components are essential for cellular viability and/or morphology.</p><p>The chemotactic signalling system of <italic>E. coli</italic> may be amenable to studying mechanisms underlying spatial organization of proteins because the lack of any chemotactic signalling component protein does not affect cell morphology and viability at least under typical laboratory conditions. Moreover, all of the components (Che proteins and the chemoreceptors) have been identified and well characterized in terms of genetics, biochemistry and structural biology (for reviews, see <xref ref-type="bibr" rid="b41">Parkinson, 1993</xref>; <xref ref-type="bibr" rid="b55">Stock and Surette, 1996</xref>; <xref ref-type="bibr" rid="b18">Djordjevic and Stock, 1998</xref>; <xref ref-type="bibr" rid="b3">Armitage, 1999</xref>; <xref ref-type="bibr" rid="b22">Falke and Kim, 2000</xref>). The chemoreceptors (also known as transducers or methyl-accepting chemotaxis proteins (MCPs)) form clusters with the histidine kinase CheA and the adaptor CheW at a cell pole and many other Che proteins target to this polar receptor-kinase cluster (<xref ref-type="bibr" rid="b35">Maddock and Shapiro, 1993</xref>; <xref ref-type="bibr" rid="b54">Sourjik and Berg, 2000</xref>; <xref ref-type="bibr" rid="b50">Shiomi <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="b14">Cantwell <italic>et al.</italic>, 2003</xref>; <xref ref-type="bibr" rid="b4">Banno <italic>et al.</italic>, 2004</xref>). Polar localization and clustering have been thought to be critical for signal amplification (<xref ref-type="bibr" rid="b11">Bray <italic>et al.</italic>, 1998</xref>; <xref ref-type="bibr" rid="b49">Shimizu <italic>et al.</italic>, 2000</xref>; <xref ref-type="bibr" rid="b2">Ames <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="b26">Gestwicki and Kiessling, 2002</xref>; <xref ref-type="bibr" rid="b31">Kim <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="b29">Homma <italic>et al.</italic>, 2004</xref>) and adaptation (<xref ref-type="bibr" rid="b33">Li and Weis, 2000</xref>; <xref ref-type="bibr" rid="b5">Barnakov <italic>et al.</italic>, 2001</xref>; <xref ref-type="bibr" rid="b6">2002</xref>; <xref ref-type="bibr" rid="b32">Levit and Stock, 2002</xref>; <xref ref-type="bibr" rid="b50">Shiomi <italic>et al.</italic>, 2002</xref>; <xref ref-type="bibr" rid="b4">Banno <italic>et al.</italic>, 2004</xref>).</p><p>It has been shown that polar localization of the chemoreceptors depends partially on CheA and CheW (<xref ref-type="bibr" rid="b35">Maddock and Shapiro, 1993</xref>; <xref ref-type="bibr" rid="b52">Skidmore <italic>et al.</italic>, 2000</xref>; <xref ref-type="bibr" rid="b51">Shiomi <italic>et al.</italic>, 2005</xref>). However, it is not known how the chemoreceptor localizes to a cell pole. To target to a pole, a nascent chemoreceptor protein might be inserted into the cytoplasmic membrane (i) at or near a cell pole thereby staying in the vicinity of the insertion point (the <italic>direct</italic> membrane insertion model) or (ii) at random positions thereafter migrating or diffusing away from the insertion point (the <italic>indirect</italic> membrane insertion model) (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>). Both of the mechanisms have been shown to operate in other systems: (i) IcsA of <italic>S. flexneri</italic> employs a <italic>direct</italic> mechanism when targeting directly to an old pole in <italic>E. coli</italic>, which does not have an IcsA homologue (<xref ref-type="bibr" rid="b15">Charles <italic>et al.</italic>, 2001</xref>) and (ii) SpoIVFB of <italic>Bacillus subtilis</italic> utilizes an <italic>indirect</italic> mechanism, being randomly inserted into the cytoplasmic membrane and then diffused to and captured in the outer forespore membrane (<xref ref-type="bibr" rid="b45">Rudner <italic>et al.</italic>, 2002</xref>).</p><fig id="fig01" position="float"><label>Fig. 1</label><caption><title>Polar localization of Tar–GFP.</title><p>A. <italic>Direct</italic> and <italic>indirect</italic> membrane insertion models for polar localization of the chemoreceptors. To target to a cell pole, a nascent chemoreceptor protein might be inserted into the cytoplasmic membrane (i) at or near a cell pole (a <italic>direct</italic> model), or (ii) at random positions thereafter migrating or diffusing away from the insertion point (an <italic>indirect</italic> model). IM, the cytoplasmic (inner) membrane; OM, the outer membrane.</p><p>B. Time-course of polar localization of Tar–GFP. HCB436 (CheAW<sup>+</sup>) or HCB437 (CheAW<sup>−</sup>) cells carrying a plasmid encoding Tar–GFP or Taz1–GFP were observed at indicated time points after the addition of 1 mM arabinose. Numbers above pictures represent harvested time points (min).</p><p>C. Time-lapse observation of polar localization of Tar–GFP. HCB436 cells carrying plasmid encoding Tar–GFP were spotted onto a glass slide covered with 0.5% agarose. Fluorescence was observed with 10 or 60 s intervals.</p></caption><graphic xlink:href="mmi060-0894-f1"/></fig><p>To elucidate the localization mechanism of the chemoreceptor, it is indispensable to identify the site(s) of membrane insertion. It has been reported that the serine chemoreceptor Tsr of <italic>E. coli</italic> is inserted into the cytoplasmic membrane via a SecA-dependent process (<xref ref-type="bibr" rid="b25">Gebert <italic>et al.</italic>, 1988</xref>). SecA is an ATPase that associates with the integral membrane proteins, SecY, SecE and SecG, to constitute a general protein translocase (for reviews, see <xref ref-type="bibr" rid="b19">Driessen <italic>et al.</italic>, 1998</xref>; <xref ref-type="bibr" rid="b39">Mori and Ito, 2001</xref>). Recent observation of SecY–GFP (green fluorescent protein) suggested that the Sec machinery is evenly distributed throughout the cytoplasmic membrane in <italic>E. coli</italic> (<xref ref-type="bibr" rid="b10">Brandon <italic>et al.</italic>, 2003</xref>; <xref ref-type="bibr" rid="b21">Espeli <italic>et al.</italic>, 2003</xref>). It has recently been reported that the Sec translocases of some Gram-positive bacteria show characteristic localization: to a microdomain in <italic>Streptococcus pyogenes</italic> (<xref ref-type="bibr" rid="b44">Rosch and Caparon, 2004</xref>) and to a helical array in <italic>B. subtilis</italic> (<xref ref-type="bibr" rid="b13">Campo <italic>et al.</italic>, 2004</xref>). Thus without any direct evidence it is hard to judge which model can apply to the chemoreceptor localization in <italic>E. coli</italic>: (i) a subset of Sec translocases at a pole might be devoted to insert the chemoreceptors (and possibly other polar membrane proteins) into the membrane, or (ii) the chemoreceptors might be inserted into the membrane at virtually any position throughout a cell.</p><p>Here we report that the aspartate chemoreceptor Tar employs an indirect mechanism for its polar localization: i.e. after induction, Tar–GFP first appeared in lateral membrane regions, clustered in the presence of CheA and CheW, and then seemed to migrate towards a cell pole. Unexpectedly, Tar–GFP formed a helical array when it is located in lateral membrane regions. Further analyses, including observation of GFP–SecE and the mutant versions of MalE–GFP, as well as indirect immunofluorescent microscopy on SecG, suggested that this Tar coil reflects a helical array of the Sec machinery itself, which was shown to be distinct from that of the actin homologue MreB.</p></sec><sec><title>Results</title><sec><title>The aspartate chemoreceptor Tar is not directly inserted into polar membranes</title><p>To visualize the localization pathway of a chemoreceptor, GFP was fused with the C-terminus of the aspartate chemoreceptor Tar (<xref ref-type="bibr" rid="b29">Homma <italic>et al.</italic>, 2004</xref>), and localization of the resulting fusion (named Tar–GFP) was observed sequentially after induction. For tight regulation of expression, the <italic>tar–gfp</italic> gene was placed downstream of the <italic>araBAD</italic> promoter (pBAD24-Tar–GFP). Cells expressing Tar–GFP swarmed only slightly slower than those expressing wild-type Tar (<xref ref-type="bibr" rid="b51">Shiomi <italic>et al.</italic>, 2005</xref>), indicating that the Tar–GFP retains essential receptor function. A chimeric protein named Taz1–GFP that does not localize to a pole was used as a negative control (pBAD24-Taz1–GFP): Taz1 was constructed by replacing the cytoplasmic domain of Tar, which is required for interaction with CheA and CheW and formation of a predicted trimer of the Tar dimers, with the cytoplasmic kinase and phosphatase domains of the osmosensor histidine kinase EnvZ (details will be described elsewhere). It was previously reported that the kinase and phosphatase activities of Taz1 (<xref ref-type="bibr" rid="b56">Utsumi <italic>et al.</italic>, 1989</xref>), which was not fused with GFP, are modulated by the Tar-specific ligand aspartate (<xref ref-type="bibr" rid="b56">Utsumi <italic>et al.</italic>, 1989</xref>; <xref ref-type="bibr" rid="b60">Yang <italic>et al.</italic>, 1993</xref>). We observed that Taz1–GFP is distributed almost evenly throughout the cytoplasmic membrane (<xref ref-type="fig" rid="fig01">Figs 1B</xref> and <xref ref-type="fig" rid="fig02">2F</xref>). HCB436 (CheAW<sup>+</sup>) cells carrying the plasmid encoding Tar–GFP or Taz1–GFP were grown in TG medium. After 3 h, 1 mM arabinose was added to express the GFP fusion proteins (time 0). Cells were harvested at indicated time points (<xref ref-type="fig" rid="fig01">Fig. 1B</xref>). At 20 min after induction, in most cells both Tar–GFP and Taz1–GFP appeared in lateral cytoplasmic membrane regions, where the fusion proteins formed small clusters, whereas in the other cells we hardly detected any fluorescence from Tar–GFP. At 60 min, subpopulations of Tar–GFP formed clusters at apparently random positions in the cytoplasmic membrane and others localized to cell poles. At 80 min, Tar–GFP localized almost completely to cell poles, whereas Taz1–GFP was almost evenly distributed in the cytoplasmic membrane, an expected result as it lacks the highly conserved domain critical for polar localization (D. Shiomi and I. Kawagishi, unpubl. results). It should be noted that the expression levels of Tar–GFP was comparable with chromosome-encoded Tar at 60 min after the addition of 1 mM arabinose. In the absence of CheA and CheW, Tar–GFP also formed clusters at apparently random positions (see Results at 20 and 40 min) and eventually localized to cell poles to a lesser extent (100 min) than in the presence of CheA and CheW. Under our experimental conditions, cell-to-cell variations in the localization patterns of Tar–GFP at each time point except for at 20 min were only limited (data not shown). These results are consistent with an <italic>indirect</italic> model, in which the proper polar localization of Tar is achieved by random insertions into the cytoplasmic membrane followed by the association with CheA and CheW and/or interaction with each other through the <underline>h</underline>ighly <underline>c</underline>onserved <underline>d</underline>omain among MCPs (HCD) and the diffusion or transport of the resulting clusters to cell poles.</p><p>To further examine this possibility, we observed the processes of polar localization of Tar–GFP in a single cell. An hour after the addition of 1 mM arabinose, cells were washed and resuspended in TG medium supplemented with 25 μg ml<sup>−1</sup> rifampicin to prevent further transcription. Cells were further incubated at 30°C for 10 min and then observed by fluorescence microscopy with 1, 10 or 60 s intervals. Time-lapse images of a representative cell are shown in <xref ref-type="fig" rid="fig01">Fig. 1C</xref>. Tar–GFP formed clusters at random positions of the cytoplasmic membrane. At a cell pole, a fluorescent cluster was detected after the addition of rifampicin. Again this observation is consistent with the notion that Tar–GFP clusters at lateral membrane regions and then migrates towards a cell pole. However, we have never seen a cluster travelling from the lateral site to the pole, but this cannot argue against the net migration to the pole: it would be rather hard to imagine that a migrating cluster stays in the same focal plane for a long period. Anyhow, fluorescent spots did not appear to migrate straightforward to the pole. They might move in a random-walk fashion and be trapped when they reach the pole. It should also be noted that this result may not exactly reflect native migration patterns of the chemoreceptor from lateral membrane regions to a pole because cells were attached to a 0.5% agarose layer on a slide glass and therefore nutrients and oxygen must be limited.</p></sec><sec><title>Tar–GFP shows a helical arrangement</title><p>Careful observation of Tar–GFP clusters at lateral cytoplasmic membrane regions (<xref ref-type="fig" rid="fig01">Fig. 1B and C</xref>) led us to suspect that Tar could be organized into a coil-like array in the lateral membrane. The two-dimensional images of lateral clusters of Tar–GFP are reminiscent of those of the bacterial actin homologue MreB that forms a helical structure along the long axis of a cell (<xref ref-type="bibr" rid="b30">Jones <italic>et al.</italic>, 2001</xref>; <xref ref-type="bibr" rid="b48">Shih <italic>et al.</italic>, 2003</xref>). We therefore wanted to know the three-dimensional arrangement of Tar in lateral cytoplasmic membrane regions. Optical dissection of single cells (more than 30 cells for each condition) was carried out and the resulting series of fluorescent images were processed by using a deconvolution software to remove out-of-focus fluorescence and to reconstitute a three-dimensional fluorescent image. First, we examined Tar–GFP in the presence of CheA and CheW (i.e. in strain HCB436) at 60 min after the addition of the inducer arabinose (<xref ref-type="fig" rid="fig02">Fig. 2A–C</xref>). All of the cells checked showed coil- or ring-like fluorescence patterns along their long axes. Similar helical fluorescence patterns were detected even when Tar–GFP was expressed in the absence of CheA and CheW (i.e. in strain HCB437) (<xref ref-type="fig" rid="fig02">Fig. 2D and E</xref>). In contrast, Taz1–GFP did not show any characteristic pattern in either host strain (<xref ref-type="fig" rid="fig02">Fig. 2F</xref>). We previously observed that GFP–CheA forms clusters at a cell pole and lateral cytoplasmic membrane regions when Tar was overexpressed (D. Shiomi and I. Kawagishi, unpubl. results). Under the same condition, we obtained reconstituted three-dimensional images of GFP–CheA and detected coiled fluorescence patterns (<xref ref-type="fig" rid="fig02">Fig. 2G</xref>). These results suggest that Tar and CheA are organized into a helical array in lateral membrane regions. The formation of such an array might involve the HCD-mediated interactions of Tar–GFP with neighbouring Tar–GFP (between dimers) and/or CheA.</p><fig id="fig02" position="float"><label>Fig. 2</label><caption><title>Helical arrays of the chemotaxis machinery in a cell. Cells were subjected to optical sectioning and processing as described in <italic>Experimental procedures</italic>. GFP-tagged proteins were induced by 1 mM arabinose for 60 min. Three-dimensionally reconstructed images from different angles (1, 2 and 3) are shown. Each image was rotated around the long axis of the cell to display the helical array. Unprocessed images are also shown (4, 5, 6).</title><p>A–C. Tar–GFP in HCB436 (CheAW<sup>+</sup>) cells.</p><p>D and E. Tar–GFP in HCB437 (CheAW<sup>–</sup>) cells.</p><p>F. Taz1–GFP in an HCB436 (CheAW<sup>+</sup>) cell.</p><p>G. GFP–CheA in an HCB437 cell expressing Tar and CheW. Scale bars indicate 1 μm.</p></caption><graphic xlink:href="mmi060-0894-f2"/></fig><p>We now hypothesized two possibilities about the nature of this helical arrangement of Tar–GFP: (i) Tar might migrate along a helical track within lateral regions of the cytoplasmic membrane; or (ii) the protein translocation machinery that inserts Tar into the cytoplasmic membrane might be arranged into a helical array. Whichever the case, introduction of Cys residues in the periplasmic region of Tar, which is devoid of Cys residue, might result in a better helical image if disulphide cross-linking leads to higher oligomers and occurs immediately after the translocation. Introduction of a Cys residue at an appropriate periplasmic position (e.g. S36C) of the subunit interface within a dimer results in a disulphide bond between subunits. When another Cys residue is introduced at the external surface of the dimer (e.g. D142C), the double Cys mutant would form cross-linked oligomers (<xref ref-type="fig" rid="fig03">Fig. 3A</xref>). This has been demonstrated for Tar-S36C&D142C (<xref ref-type="bibr" rid="b29">Homma <italic>et al.</italic>, 2004</xref>). Among the double Cys mutant proteins tested, Tar-S36C&A118C was most efficiently cross-linked (H. Irieda, M. Homma, M. Homma and I. Kawagishi, submitted for publication). Immunoblotting analyses of the GFP fusion of the double Cys mutant showed that it is indeed highly cross-linked leaving only a very small amount of the uncross-linked monomer even without the addition of any catalyst for oxidation (<xref ref-type="fig" rid="fig03">Fig. 3B</xref>; note that highly cross-linked products could not get into the gel). In HCB436 cells, Tar-S36C&A118C–GFP showed clearer coiled structures (<xref ref-type="fig" rid="fig03">Fig. 3C</xref>).</p><fig id="fig03" position="float"><label>Fig. 3</label><caption><title>Helical arrays of disulphide-cross-linked Tar–GFP and Taz1–GFP.</title><p>A. Schematic illustration of the inter-dimer cross-linking strategy of Tar–GFP or Taz1–GFP. Two cysteine residues were introduced into the periplasmic domain of the Tar– or Taz1–GFP protein to form higher oligomers with disulphide bonds linking subunits both within and between dimers. IM, the cytoplasmic (inner) membrane; N and C, the N- and C-termini; open and shaded circles, the introduced Cys residues (S36C and A118C).</p><p>B. <italic>In vivo</italic> disulphide cross-linking of the Cys-replaced Tar– and Taz1–GFP proteins. HCB436 cells (ΔMCP ΔCheR ΔCheB) expressing the wild-type (WT) or mutant (S36C&A118C) versions of the fusion proteins (Tar–GFP and Taz1–GFP) were incubated with (+) or without (−) 60 μM Cu(II)(<italic>o</italic>-phenanthroline)<sub>3</sub> for 5 min at 30°C and their whole-cell extracts were subjected to non-reducing (−2ME) and reducing (+2ME) SDS-PAGE followed by immunoblotting with anti-GFP (Molecular Probes).</p><p>C and D. Three-dimensionally reconstituted images of HCB436 cells expressing the S36C&A118C mutant versions of Tar–GFP (C) and Taz1–GFP (D). For each cell, reconstituted images form different angles (1–3) and unprocessed images (4–6) are shown. Scale bars indicate 1 μm.</p></caption><graphic xlink:href="mmi060-0894-f3"/></fig><p>We also constructed and examined Taz1-S36C&A118C–GFP, which lacks HCD, as a negative control. Unexpectedly, however, it also formed a helical array (<xref ref-type="fig" rid="fig03">Fig. 3D</xref>), indicating that HCD is dispensable for the helical array. A possible explanation of this phenomenon is that Taz1-S36C&A118C–GFP might be cross-linked immediately after inserted into the cytoplasmic membrane thereby trapped in the vicinity of the sites of insertion. It was previously shown that the serine chemoreceptor Tsr is inserted into the membrane via the general protein translocation apparatus (the Sec machinery) (<xref ref-type="bibr" rid="b25">Gebert <italic>et al.</italic>, 1988</xref>). We suspect that cross-linking immediately after the insertion would prevent the mutant protein from diffusing away from the Sec machinery, resulting in better helical images. If cross-linking was extremely efficient, a newly inserted Cys derivative of Tar–GFP or Taz1–GFP might be cross-linked with the neighbouring one, which would clog the Sec machinery. If this is the case, overproduction of these proteins would affect cell growth. As expected, the A118C derivative of Tar–GFP or Taz1–GFP retarded cell growth even when the expression level was adjusted roughly equal to the total amount of chromosome-encoded chemoreceptors, whereas the corresponding wild-type versions did not show any significant effect (data not shown). Therefore, we speculate that the helical arrangements of Tar–GFP and Taz1–GFP within a cell may reflect that of the Sec machinery itself.</p></sec><sec><title>Cellular organization of the Sec machinery</title><p>The Sec machinery contains, as essential components, three integral membrane proteins, SecY, SecE and SecG (the SecYEG complex) that forms a translocation pore, and a membrane-peripheral ATPase, SecA. To examine subcellular localization of the Sec machinery, we constructed GFP–SecE, which was expressed in strain HCB436 and three-dimensional images were reconstituted (<xref ref-type="fig" rid="fig04">Fig. 4A–C</xref>). GFP–SecE was also organized into a helical array. This is inconsistent with recent publications reporting that SecY- and SecE–GFP fusion proteins are evenly distributed throughout the cytoplasmic membrane in <italic>E. coli</italic> (<xref ref-type="bibr" rid="b10">Brandon <italic>et al.</italic>, 2003</xref>; <xref ref-type="bibr" rid="b21">Espeli <italic>et al.</italic>, 2003</xref>). This discrepancy could be accounted for by levels of the Sec–GFP fusion proteins: their even distribution might result from the failure of excess Sec–GFP proteins to form a SecYEG complex with their partner Sec proteins as we also observed almost even distribution of GFP–SecE when it was mildly overexpressed (<xref ref-type="fig" rid="fig04">Fig. 4D and E</xref>).</p><fig id="fig04" position="float"><label>Fig. 4</label><caption><title>Helical arrangement of the GFP–SecE fusion protein. Images of more than 30 cells were analysed and representative cells are shown.</title><p>A–C. Three-dimensionally reconstituted images of HCB436 cells expressing GFP–SecE. Cells were harvested 15 min after the addition of 1 mM arabinose and subjected to optical sectioning and processing as described in <italic>Experimental procedures</italic>. For each cell, reconstituted images form different angles (1–3) and unprocessed images (4–6) are shown.</p><p>D and E. Overexpressed GFP–SecE in HCB436 cells. Cells were harvested 30 min after the addition of 1 mM arabinose.</p></caption><graphic xlink:href="mmi060-0894-f4"/></fig><p>We also wanted to observe subcellular localization of the Sec machinery without GFP fusion and at the wild-type stoichiometry. First, we employed indirect immunofluorescence microscopy (IFM) to see localization of SecG, which is other component of the Sec machinery. Immunoblotting of wild-type cells (W3110) with anti-SecG antibody detected a major band of approximately 12 kDa, which should correspond to SecG, with low levels of cross-reacting bands (<xref ref-type="fig" rid="fig05">Fig. 5A</xref>). Wild-type cells were then fixed, treated with anti-SecG antibody and then with the second antibody labelled with Alexa Fluor 488. In many cells, with significant levels of background, diagonal lines were observed (<xref ref-type="fig" rid="fig05">Fig. 5B</xref>). These lines are consistent with the helical arrangement of SecG in a cell.</p><fig id="fig05" position="float"><label>Fig. 5</label><caption><title>Helical arrangement of SecG.</title><p>A. Detection of chromosome-encoded SecG by immunoblotting. W3110 (wild-type) cells were subjected to immunoblotting with anti-SecG antibody. Arrow head indicates bands of SecG. Open triangle on the top indicates increasing amounts applied (from left to right, one-, two- and fourfold).</p><p>B. Detection of chromosome-encoded SecG by IFM. W3110 (wild-type) cells were subjected to IFM with anti-SecG as the first antibody and Alexa Fluor 488-labelled goat anti-rabbit IgG antibody. Typical cells are shown.</p></caption><graphic xlink:href="mmi060-0894-f5"/></fig><p>Second, we took advantage of export-defective mutations of maltose-binding protein (MBP or MalE) to observe subcellular localization of the Sec machinery at the wild-type expression level. The mutant MalE proteins with M18R or M19R substitution in the leader peptide sequence are defective in translocation across the membrane, but not in its initiation step, and their precursors are thought to remain trapped in the Sec machinery (<xref ref-type="bibr" rid="b7">Bassford and Beckwith, 1979</xref>; <xref ref-type="bibr" rid="b8">Bedouelle <italic>et al.</italic>, 1980</xref>). It has been shown that wild-type MalE–GFP is exported to the periplasmic space successfully via the Sec machinery but that the protein is not fluorescent presumably due to misfolding (<xref ref-type="bibr" rid="b23">Feilmeier <italic>et al.</italic>, 2000</xref>). We therefore reasoned that fluorescence of the M18R or M19R derivative of MalE–GFP would reflect cytoplasmic localization of the Sec machinery (<xref ref-type="fig" rid="fig06">Fig. 6A</xref>). Immunoblotting verified that expression levels of the wild-type and mutant versions of MalE–GFP were comparable with each other and that the mutant versions were not processed (<xref ref-type="fig" rid="fig06">Fig. 6B</xref>). Wild-type MalE–GFP was not fluorescent (data not shown) whereas the mutant versions of MalE–GFP formed clusters in the cytoplasmic membrane in strain HCB436 (see <xref ref-type="fig" rid="fig06">Fig. 6C–E</xref>). The reconstituted three-dimensional images of MalE-M19R–GFP clearly showed helical arrangements (<xref ref-type="fig" rid="fig06">Fig. 6C–E</xref>). In this experiment, no component of the Sec machinery is overproduced and therefore the result argues strongly that the Sec machinery is organized into a helical array. We tried to examine whether MalE-M19R–CFP colocalizes with YFP–SecE, but it was unsuccessful probably because the expression level of the latter fusion was difficult to control (data not shown). Instead, we examined colocalization of the coils of MalE and Tar (<xref ref-type="fig" rid="fig06">Fig. 6A and F</xref>). CFP and YFP were fused to MalE-M19R and Tar respectively. MalE-M19R–CFP and Tar–YFP colocalized in the lateral membrane regions (<xref ref-type="fig" rid="fig06">Fig. 6F</xref>). We therefore conclude that the Sec machinery is organized into a coil-like structure in a cell.</p><fig id="fig06" position="float"><label>Fig. 6</label><caption><title>Helical arrangement of the translocation-defective mutant MalE–GFP protein.</title><p>A. Schematic illustration of double labelling of Tar and MalE. YFP and CFP/GFP were fused with the C-terminus of Tar and MalE to yield Tar–YFP and MalE–CFP/GFP respectively. Exported MalE–CFP was not fluorescent whereas the signal sequence mutant versions of MalE–CFP (M18R and M19R) were fluorescent probably because they remain clogged in the Sec machinery. MalE–CFP and MalE*–CFP represent wild-type and mutant versions of MalE–CFP respectively.</p><p>B. Expression of the mutant MalE–GFP proteins. HCB436 cells carrying a plasmid encoding GFP or the wild-type (WT) or mutant (M18R or M19R) version of MalE–GFP were grown in the presence (+) or absence (−) of 1 mM arabinose and their whole-cell lysates were subjected to SDS-PAGE followed by immunoblotting with anti-GFP.</p><p>C–E. Three-dimensionally reconstituted images of HCB436 cells expressing MalE-M19R–GFP. For each cell, reconstituted images form different angles (1–3) are shown. Scale bars indicate 1 μm.</p><p>F. Tar–YFP and MalE-M19R–CFP in strain HCB436. Individual images of Tar–YFP (images 1, 4 and 7; artificially coloured green) and MalE-M19R–CFP [original (2, 5 and 8) and processed (2′, 5′-and 8′) images; artificially coloured red] were merged (images 3, 6 and 9).</p></caption><graphic xlink:href="mmi060-0894-f6"/></fig></sec><sec><title>The Sec coil is distinct from the helical filament of MreB</title><p>It has recently been shown that <italic>E. coli</italic> has cytoskeletal proteins that are homologues of actin (FtsA and MreB) and tubulin (FtsZ) (<xref ref-type="bibr" rid="b20">van den Ent <italic>et al.</italic>, 2001</xref>). FtsA plays a key role in cytokinesis whereas MreB is critical for the rod shape. More recently, it has been shown that MreB forms a helical filament both in <italic>B. subtilis</italic> (<xref ref-type="bibr" rid="b30">Jones <italic>et al.</italic>, 2001</xref>) and <italic>E. coli</italic> (<xref ref-type="bibr" rid="b48">Shih <italic>et al.</italic>, 2003</xref>). The finding of the helical array of the Sec machinery raised a possibility that it may be associated with the MreB coil. To address this issue, we examined whether Tar–YFP colocalizes with CFP–MreB (<xref ref-type="fig" rid="fig07">Fig. 7</xref>). In all of the cells checked, the coils of CFP–MreB and Tar–YFP were not perfectly overlapped with each other. The CFP–MreB coil seemed to have a smaller pitch than the Tar–YFP coil. We therefore conclude that the Sec coil is distinct from the MreB coil.</p><fig id="fig07" position="float"><label>Fig. 7</label><caption><title>Subcellular localization of Tar–YFP with CFP–MreB within the same cells. Tar–YFP and CFP–MreB were expressed in strain HCB436. Images of Tar–YFP (left) and CFP–MreB (middle) were merged.</title></caption><graphic xlink:href="mmi060-0894-f7"/></fig></sec></sec><sec><title>Discussion</title><p>In this study, the pursuit of the mechanisms underlying polar localization of the chemoreceptor Tar leads us to the finding that the general protein translocation machinery (i.e. the SecYEG complex) is organized into a helical array. Sequential observations of cells expressing Tar–GFP supported the <italic>indirect</italic> model for polar localization (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>), in which Tar is inserted into lateral membrane regions and then migrates through a lipid bilayer to a cell pole. Further analyses of the Tar– and Taz1–GFP fusions identified helical arrays of fluorescence, which were deduced to reflect a similar array of the Sec machinery that inserts Tar into the membrane (<xref ref-type="bibr" rid="b25">Gebert <italic>et al.</italic>, 1988</xref>). The helical array of the Sec machinery was confirmed by IFM observation of chromosome-encoded SecG and the observation of GFP–SecE and the translocation-defective mutant versions of MalE–GFP. Previous immunoelectron microscopic analyses detected 80% of the cytoplasmic membrane population of Tsr, the serine chemoreceptor, at cell poles (<xref ref-type="bibr" rid="b35">Maddock and Shapiro, 1993</xref>). In light of our finding, a minor population of Tsr detected at lateral cytoplasmic membrane regions could represent nascent Tsr inserted via the Sec machinery into the cytoplasmic membrane.</p><p>It has recently been reported that the Sec machinery is evenly distributed throughout the cytoplasmic membrane in <italic>E. coli</italic> (<xref ref-type="bibr" rid="b10">Brandon <italic>et al.</italic>, 2003</xref>; <xref ref-type="bibr" rid="b21">Espeli <italic>et al.</italic>, 2003</xref>). However, this could be due to overexpression of Sec proteins because GFP–SecE was evenly distributed throughout the cytoplasmic membrane when mildly overexpressed (<xref ref-type="fig" rid="fig04">Fig. 4D and E</xref>). To visualize the localization of the Sec machinery without violating wild-type cellular levels and stoichiometry of the components, we took advantage of the mutant versions (M18R and M19R) of MalE–GFP, which are thought to be stuck in the Sec machinery (<xref ref-type="bibr" rid="b7">Bassford and Beckwith, 1979</xref>; <xref ref-type="bibr" rid="b8">Bedouelle <italic>et al.</italic>, 1980</xref>). Wild-type MalE–GFP is transported through the Sec machinery but is not fluorescent probably because the GFP part cannot be folded properly in the periplasmic space (<xref ref-type="bibr" rid="b23">Feilmeier <italic>et al.</italic>, 2000</xref>). In contrast, the mutant versions of MalE–GFP were fluorescent, suggesting that these proteins reside in the cytoplasm probably by associating with the Sec machinery. Fluorescence patterns of the M18R (not shown) and M19R versions of MalE–GFP further supported the notion that the Sec machinery is organized into a helical array. We failed to detect significant effects of the mutant MalE–GFP protein on the processing of other exported proteins and cell growth. Considering its fluorescence images, we gather that the mutant MalE–GFP protein does not occupy all of the translocases. Recently, GFP fusions to SecA, SecY and the precursor of a translocation substrate protein of a Gram-positive bacterium <italic>B. subtilis</italic> have been reported to localize with similar helical patterns (<xref ref-type="bibr" rid="b13">Campo <italic>et al.</italic>, 2004</xref>), raising a possibility that such a cellular organization is a common feature of general protein translocases among rod-shape bacteria.</p><p>The size and the subunit stoichiometry of the Sec machinery have not been unambiguously determined. Electron microscopy with negative staining suggested that the Sec machinery may consist of dimeric SecA and four SecYEG complexes (<xref ref-type="bibr" rid="b36">Manting <italic>et al.</italic>, 2000</xref>). In contrast, cryo-electron microscopy suggested that the translocation pore consists of dimeric SecYEG complex (<xref ref-type="bibr" rid="b12">Breyton <italic>et al.</italic>, 2002</xref>). Based on our fluorescence observation, we roughly estimated the average length of the Sec coil to be ∼10 μm assuming that one coil is continuous from one pole to the other. Given the published value (10.5–12 nm) of the diameter of the SecYEG tetramer (<xref ref-type="bibr" rid="b36">Manting <italic>et al.</italic>, 2000</xref>), the maximal number of SecYEG tetramers that can be packed into a helical path would then be 800–1000. Another line of evidence estimates the number of SecY, SecE or SecG monomers per cell to be ∼500 (<xref ref-type="bibr" rid="b38">Matsuyama <italic>et al.</italic>, 1992</xref>) and therefore the number of SecYEG tetramers should be ∼125. Considering that negative staining does not reflect the actual size of a protein and that SecYEG tetramers might not be packed tightly within a helical path, our rough estimation (800–1000 tetramers per helix) may be in good agreement with the predicted size of a cellular pool (∼125 tetramers per cell).</p><p>How can the Sec machinery, a complex of integral membrane proteins, be organized into a helical array? The Sec machinery (i) might associate with some rigid helical support, such as cytoskeletal protein(s) or peptidoglycan (cell wall), or (ii) might itself assemble into a coil. It has been shown that an integral membrane protein subunit of topoisomerase IV (SetB) interacts with MreB and forms a helical array (<xref ref-type="bibr" rid="b21">Espeli <italic>et al.</italic>, 2003</xref>). Although the Sec coil was shown to be distinct from the MreB coil in this study as well as in <italic>B. subtilis</italic> (<xref ref-type="bibr" rid="b13">Campo <italic>et al.</italic>, 2004</xref>), it is still possible that unknown cytoskeletal protein(s) serve to anchor the Sec machinery to form a helical array. It has recently been demonstrated that morphology of cells lacking some penicillin-binding proteins, which are involved in cell wall elongation, is abnormal (<xref ref-type="bibr" rid="b43">de Pedro <italic>et al.</italic>, 2003</xref>) and some penicillin-binding proteins localize with a spot-like pattern like MreB (<xref ref-type="bibr" rid="b17">den Blaauwen <italic>et al.</italic>, 2003</xref>; <xref ref-type="bibr" rid="b46">Scheffers <italic>et al.</italic>, 2004</xref>), raising a possibility that it might form a helical polymer and that <italic>E. coli</italic> might have cytoskeletal proteins other than the identified ones. Another candidate for a helical support may be peptidoglycan as it is newly synthesized as a helix in <italic>B. subtilis</italic> (<xref ref-type="bibr" rid="b16">Daniel and Errington, 2003</xref>). To our knowledge, there is no experimental evidence that the SecYEG complex forms oligomers higher than a tetramer. However, the native complex within the membrane environment might have a property distinct from detergent-solubilized preparations. Moreover, <italic>E. coli</italic> shows lateral heterogeneity of phospholipid distribution in the cytoplasmic membrane and such heterogeneity is involved in cell division (<xref ref-type="bibr" rid="b24">Fishov and Woldringh, 1999</xref>). Therefore, it is possible that such a certain kind of membrane lipids might help the Sec machinery to form a coil even without a rigid anchoring structure. In <italic>B. subtilis</italic>, a decrease in phosphatidylglycerol causes delocalization of SecA (<xref ref-type="bibr" rid="b13">Campo <italic>et al.</italic>, 2004</xref>). It would be intriguing to see how GFP–SecE and MalE-M19R–GFP localize in round-shaped <italic>E. coli</italic> mutants (<italic>rodA</italic> or <italic>mreB</italic>) or mutants in membrane lipid synthesis.</p><p>Physiological significance of the helical array of the Sec machinery is unclear. The helical arrangement of the Sec machinery along the long axis of a cell might have some advantage in distributing translocated proteins. If SecYEG tetramers have to be anchored to some rigid support rather than randomly distributed and freely diffusing, then a helical array would be one of the best solutions to spread proteins efficiently throughout the cytoplasmic membranes as well as the periplasmic space and the outer membrane (<xref ref-type="fig" rid="fig08">Fig. 8</xref>). Proteins that localize to certain parts of the cell, such as a pole and a septum, might be actively transported or passively diffused away towards their destinations.</p><fig id="fig08" position="float"><label>Fig. 8</label><caption><title>Model for helically arranged Sec protein translocases within a cell. The Sec machinery is organized into a helical array and translocates proteins from the cytoplasm to the cytoplasmic membrane and the periplasmic space that migrate. Some integral membrane proteins are inserted into the cytoplasmic membrane via the Sec machinery and may be sorted after insertion to migrate towards their destined regions (e.g. polar or random distribution). Outer membrane proteins are omitted for clarity. IM, the cytoplasmic (inner) membrane; OM, the outer membrane.</title></caption><graphic xlink:href="mmi060-0894-f8"/></fig><p>It is not clear whether a chemoreceptor is actively transported within a lipid bilayer to a cell pole or passively diffused and somehow trapped when it reaches a pole. In either case, the fact that both the Tar- and Taz1–GFP fusions colocalized transiently with the Sec coil indicates that membrane proteins with different destinations must be ‘sorted’ after insertion by the Sec machinery. Such sorting event(s) remain to be characterized but might provide a clue to understand the mechanisms of membrane protein localizations as well as the functions of the Sec machinery.</p></sec><sec sec-type="methods"><title>Experimental procedures</title><sec><title>Bacterial strains and plasmids</title><p>All strains used in this study are derivative of <italic>E. coli</italic> K-12. Strain RP437 is wild-type for chemotaxis (<xref ref-type="bibr" rid="b42">Parkinson and Houts, 1982</xref>). Strains HCB436 (<xref ref-type="bibr" rid="b58">Wolfe and Berg, 1989</xref>) and HCB437 (<xref ref-type="bibr" rid="b59">Wolfe <italic>et al.</italic>, 1987</xref>) lack all of the four chemoreceptors. In addition, the former lacks the methylesterase CheB and the methyltransferase CheR, and the latter lacks all of the cytoplasmic Che proteins, including the histidine kinase CheA and the adaptor CheW.</p><p>The vector plasmid pBAD24 (Ap<sup>r</sup>) (<xref ref-type="bibr" rid="b28">Guzman <italic>et al.</italic>, 1995</xref>) carries the <italic>araBAD</italic> promoter, the <italic>araC</italic> gene, which encodes the positive and negative regulator of the ara<italic>BAD</italic> promoter. Plasmids pEGFP, pECFP and pEYFP, which encode the enhanced green, cyan and yellow fluorescent protein, respectively, were purchased from Clontech.</p></sec><sec><title>Construction of plasmids encoding fluorescent fusion proteins</title><p>For construction of plasmids encoding Tar–YFP under the control of the <italic>nahG</italic> promoter, the AvaI-BsrGI fragment of pEYFP was subcloned between the AvaI and BsrGI sites of pTrc-Tar–GFP (<xref ref-type="bibr" rid="b29">Homma <italic>et al.</italic>, 2004</xref>). A BamHI site was introduced downstream of the <italic>tar–gfp/yfp</italic> gene by polymerase chain reaction (PCR) and the EcoT22I-BamHI fragment was subcloned into pLC113 (Cm<sup>r</sup>) (<xref ref-type="bibr" rid="b2">Ames <italic>et al.</italic>, 2002</xref>), which is compatible with a pBR322-derivative plasmid, to yield pLC113-Tar–GFP/YFP.</p><p>For construction of pBAD24-Tar-S36C&A118C–GFP, the AatII-NdeI fragment of pBAD24-Tar-S36C&A118C (M. Homma and I. Kawagishi, unpublished) was subcloned into pBAD24-Tar–GFP (D. Shiomi and I. Kawagishi, unpublished). The NdeI-HindIII fragment of pBAD24-EnvZ–GFP (D. Shiomi and I. Kawagishi, unpublished) was subcloned into pBAD24-Tar-S36C&A118C–GFP to yield pBAD24-Taz1-S36C&A118C–GFP.</p><p>For construction of pBAD24–GFP–SecE or GFP/CFP–MreB, the <italic>secE</italic> or <italic>mreB</italic> gene of <italic>E. coli</italic> was amplified by PCR using chromosomal DNA of strain RP437 as a template. The forward and reverse primers used were designed to introduce unique BsrGI and EcoRI sites at the 5′ and 3′ ends of the <italic>secE</italic> or <italic>mreB</italic> coding region respectively. The BsrGI-EcoRI fragments were subcloned into pEGFP or pECFP. The NcoI-HindIII fragments from the plasmids were subcloned into pBAD24 to yield pBAD24–GFP–SecE or pBAD24–GFP/CFP–MreB.</p><p>A PstI site was introduced at the 5′ end of the <italic>ecfp</italic> coding region of pECFP by PCR. The PCR product was digested with PstI and EcoRI and the resulting fragment containing the <italic>ecfp</italic> gene was subcloned into pTrcHisB. The NcoI-XbaI fragment of the resulting plasmid was subcloned into pBAD24 to yield pBAD24-CFP(NX). For construction of MalE–CFP, the <italic>malE</italic> gene was amplified from pMalE (S. Banno and I. Kawagishi, unpublished) by PCR to introduce unique EcoRI and XhoI sites at the 5′ and 3′ ends of the <italic>malE</italic> gene respectively. The PCR product was digested with EcoRI and XhoI and the resulting fragment containing the <italic>ecfp</italic> gene was subcloned into pBAD24-CFP(NX) to yield pBAD24-MalE–CFP. The M18R or M19R mutation was introduced by two-step PCR to yield pBAD24-MalE-M18R/M19R–CFP.</p></sec><sec><title>Immunoblotting</title><p>Immunoblotting was carried out as described previously (<xref ref-type="bibr" rid="b50">Shiomi <italic>et al.</italic>, 2002</xref>) using anti-GFP antibody (Molecular Probes).</p></sec><sec><title>Fluorescence microscopy</title><p>Observations of fluorescent proteins were carried out essentially as described previously (<xref ref-type="bibr" rid="b50">Shiomi <italic>et al.</italic>, 2002</xref>). For time-course experiments, cells carrying a plasmid encoding Tar–GFP or Taz1–GFP were grown in TG medium (1% tryptone, 0.5% NaCl, 0.5% glycerol) at 30°C and after 3 h, 1 mM arabinose was added to each culture. Cells were harvested with appropriate intervals (see <italic>Results</italic>), spotted onto a glass slide coated with 0.5% agarose and observed under an inverted fluorescence microscope (Olympus IX71). The fluorescent images were processed by using the software Metamorph 5.0r4 (Molecular Devices) and Photoshop ver.7 (Adobe). For time-lapse experiments, after the addition of 1 mM arabinose, cells were further incubated at 30°C for 1 h, harvested, washed and resuspended in TG medium. Rifampicin (25 μg ml<sup>−1</sup>) was added to inhibit transcription and cells were further incubated for 10 min, harvested and observed with intervals of 1 s, 10 s or 1 min. For GFP–SecE or CFP–MreB, cells were grown at 30°C and after 3 h, 1 or 10 mM arabinose was added respectively. Cells were further incubated for 15 or 30 min, respectively, harvested and observed. For double labelling with MalE-M19R–CFP and Tar–YFP, cells were grown in TG medium supplemented with 0.5 μM sodium salicylate (to induce Tar–YFP) at 30°C for 3 h. Then, 0.1 mM arabinose was added to induce MalE-M19R–CFP. Cells were further incubated for 15 min, harvested and observed. For double labelling with CFP–MreB and Tar–YFP, cells were grown in TG medium supplemented with 0.5 μM sodium salicylate at 30°C for 3 h. Then, 10 mM arabinose was added to induce CFP–MreB. Cells were further incubated for 30 min, harvested and observed.</p><p>For optical sectioning, a fluorescence microscope with piezo drive (Zeiss Axiovert) was used to obtain a series of z-sections with a fixed spacing of 0.2 μm. Each stack of 20–30 sectioned fluorescence images was deconvolved by using CELLscan (Scanalytics). Three-dimensional images were reconstituted by using IPLab (Hamamatsu Photonics). Expressions of fluorescent proteins were carried out under the conditions described above.</p></sec><sec><title>Immnofluorescence microscopy</title><p>Immnofluorescence microscopy was carried out according to the method of <xref ref-type="bibr" rid="b35">Maddock and Shapiro (1993)</xref> with modifications. Cells were harvested and resuspended in MLM and fixed by adding an equal volume of 0.6% folmaldehyde in MLM and by incubating on ice for 2 h. Cells were then washed three times with MLM and resuspended in GTE [50 mM glucose, 20 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 7.0)] supplemented with 2 mg ml<sup>−1</sup> lysozyme. An aliquot was spotted on a cover slip and allowed to dry. PBST (140 mM NaCl, 2 mM KCl, 8 mM Na<sub>2</sub>HPO<sub>4</sub>, 1.5 mM KH<sub>2</sub>PO<sub>4</sub>, 0.05% Tween20, 2% BSA) were then spotted onto the sample. After 15 min of incubation, the sample was treated with the first antibody against SecG (provided by Dr H. Tokuda) for 1 h, washed twice, treated with the second antibody [goat anti-rabbit IgG labelled with Alexa Fluor 488 (Molecular Probes)], washed twice and observed under the fluorescence microscope.</p></sec></sec>
|
<italic>Plasmodium berghei</italic> calcium-dependent protein kinase 3 is required for ookinete gliding motility and mosquito midgut invasion
|
<p>Apicomplexan parasites critically depend on a unique form of gliding motility to colonize their hosts and to invade cells. Gliding requires different stage and species-specific transmembrane adhesins, which interact with an intracellular motor complex shared across parasite stages and species. How gliding is regulated by extracellular factors and intracellular signalling mechanisms is largely unknown, but current evidence suggests an important role for cytosolic calcium as a second messenger. Studying a <italic>Plasmodium berghei</italic> gene deletion mutant, we here provide evidence that a calcium-dependent protein kinase, CDPK3, has an important function in regulating motility of the ookinete in the mosquito midgut. We show that a <italic>cdpk3</italic><sup>–</sup> parasite clone produces morphologically normal ookinetes, which fail to engage the midgut epithelium, due to a marked reduction in their ability to glide productively, resulting in marked reduction in malaria transmission to the mosquito. The mutant was successfully complemented with an episomally maintained <italic>cdpk3</italic> gene, restoring mosquito transmission to wild-type level. <italic>cdpk3</italic><sup>–</sup> ookinetes maintain their full genetic differentiation potential when microinjected into the mosquito haemocoel and <italic>cdpk3</italic><sup>–</sup> sporozoites produced in this way are motile and infectious, suggesting an ookinete-limited essential function for CDPK3.</p>
|
<contrib contrib-type="author"><name><surname>Siden-Kiamos</surname><given-names>Inga</given-names></name><xref ref-type="aff" rid="au1">1</xref><xref ref-type="aff" rid="au2">2</xref></contrib><contrib contrib-type="author"><name><surname>Ecker</surname><given-names>Andrea</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Nybäck</surname><given-names>Saga</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Louis</surname><given-names>Christos</given-names></name><xref ref-type="aff" rid="au2">2</xref><xref ref-type="aff" rid="au3">3</xref></contrib><contrib contrib-type="author"><name><surname>Sinden</surname><given-names>Robert E</given-names></name><xref ref-type="aff" rid="au1">1</xref></contrib><contrib contrib-type="author"><name><surname>Billker</surname><given-names>Oliver</given-names></name><xref ref-type="aff" rid="au1">1</xref><xref ref-type="corresp" rid="cor1">*</xref></contrib><aff id="au1"><label>1</label><institution>Division of Cell and Molecular Biology, Sir Alexander Fleming Building, Imperial College London</institution><addr-line>London SW7 2AZ, UK</addr-line></aff><aff id="au2"><label>2</label><institution>Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology – Hellas</institution><addr-line>Vassilika Vouton, PO Box 1527, 71110 Heraklion, Crete, Greece</addr-line></aff><aff id="au3"><label>3</label><institution>Department of Biology, University of Crete</institution><addr-line>71110 Heraklion, Crete, Greece</addr-line></aff>
|
Molecular Microbiology
|
<sec><title>Introduction</title><p>To complete their life cycle, malaria parasites require three invasive ‘zoite’-stages. (i) Merozoites invade erythrocytes in the bloodstream of the vertebrate host. (ii) Ookinetes are the motile zygotes that form in the blood meal of the mosquito vector, penetrate the peritrophic matrix and the mosquito midgut epithelium and then replicate as an oocyst on the gut wall. (iii) Sporozoites are highly motile parasite stages that emerge from the cyst and penetrate cells of the mosquito salivary gland and various tissues of the vertebrate host, before invading and differentiating inside parenchymal cells of the liver. Malarial invasive life cycle stages differ in morphology, size, target tissues and modes and speed of movement, but the molecular mechanisms driving motility and invasion are nevertheless thought to be very similar (<xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>).</p><p>Much of our molecular understanding of apicomplexan gliding motility comes from studies on tachyzoites of <italic>Toxoplasma gondii</italic> and on malaria sporozoites [recently reviewed by <xref ref-type="bibr" rid="b20">Kappe <italic>et al</italic>. (2004)</xref> and <xref ref-type="bibr" rid="b21">Keeley and Soldati (2004)</xref>]. Motility and invasion require the regulated apical release of adhesive proteins or protein complexes from secretory organelles, such as micronemes, rhoptries and dense granules, which typically form part of the apical complex that characterizes all apicomplexan invasive stages. The actomyosin-based motor complex that powers motility is located between the parasite’s plasma membrane and the underlying inner membrane complex (IMC). The bridge between the motor and the extracellular substrate is provided by a family of micronemal adhesins typified by the thrombospondin-related anonymous protein (TRAP) of the sporozoite. Its translocation to the posterior end of the parasite generates the force for gliding motility and invasion (<xref ref-type="bibr" rid="b32">Sultan <italic>et al</italic>., 1997</xref>). In malaria parasites, each invasive stage uses a different member of the TRAP family, probably reflecting the diverse extracellular substrates and host tissues, with which the malarial life cycle stages interact. In ookinetes the circumsporozoite/TRAP-related protein (CTRP) fulfils this function in the mosquito midgut (<xref ref-type="bibr" rid="b11">Dessens <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b40">Yuda <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b33">Templeton <italic>et al</italic>., 2000</xref>), while erythrocyte invasion is thought to involve the merozoite TRAP homologue, MTRAP (<xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>).</p><p>In marked contrast to the surface adhesins, intracellular components of the motor complex are thought to be highly conserved across all invasive life cycle stages and apicomplexan parasite species (<xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>). The conserved cytoplasmic tails of all TRAP-family adhesins engage the motor indirectly, by binding tetramers of the glycolytic enzyme fructose-1,6-bisphosphate aldolase, which in turn is capable of nucleating short actin filaments (<xref ref-type="bibr" rid="b6">Buscaglia <italic>et al</italic>., 2003</xref>; <xref ref-type="bibr" rid="b19">Jewett and Sibley, 2003</xref>). The force for parasite gliding and invasion is generated by the dynamic interaction of these actin filaments with a class XIV myosin, MyoA, which is essential for gliding (<xref ref-type="bibr" rid="b27">Meissner <italic>et al</italic>., 2002</xref>). MyoA is anchored to the IMC by a MyoA tail-interacting protein (MTIP; <xref ref-type="bibr" rid="b4">Bergman <italic>et al</italic>., 2003</xref>), which has properties of a myosin light chain (<xref ref-type="bibr" rid="b17">Herm-Gotz <italic>et al</italic>., 2002</xref>), and by two glideosome-associated proteins, GAP45 and GAP50 (<xref ref-type="bibr" rid="b15">Gaskins <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>).</p><p>Studies on <italic>T. gondii</italic> tachyzoites point towards an important role of calcium as a second messenger regulating gliding. Microneme secretion, motility and invasion occur independently of extracellular or host cell calcium levels, but require the release of calcium from intracellular stores, most likely the endoplasmic reticulum (<xref ref-type="bibr" rid="b7">Carruthers and Sibley, 1999</xref>; <xref ref-type="bibr" rid="b26">Lovett <italic>et al</italic>., 2002</xref>). Artificially raising cytosolic calcium levels by pharmacological agents, such as calcium ionophores, is sufficient to trigger microneme release in <italic>T. gondii</italic> tachyzoites (<xref ref-type="bibr" rid="b7">Carruthers and Sibley, 1999</xref>) and <italic>Plasmodium falciparum</italic> sporozoites (<xref ref-type="bibr" rid="b14">Gantt <italic>et al</italic>., 2000</xref>). Studying individual gliding tachyzoites, <xref ref-type="bibr" rid="b25">Lovett and Sibley (2003)</xref> observed rapid increases in cytosolic calcium accompanying or immediately preceding bursts of motility <italic>in vitro</italic>, illustrating the key role calcium plays as a second messenger.</p><p>Protein kinases are emerging as important downstream effectors of the calcium signal(s) regulating gliding (<xref ref-type="bibr" rid="b12">Dobrowolski <italic>et al</italic>., 1997</xref>; <xref ref-type="bibr" rid="b8">Carruthers <italic>et al</italic>., 1999</xref>; <xref ref-type="bibr" rid="b37">Wiersma <italic>et al</italic>., 2004</xref>). Candidates include members of the family of calcium-dependent protein kinases (CDPK), one of which, TgCDPK2, was identified as a prominent target for a kinase inhibitor, KT5926 (IC<sub>50</sub> <italic>c.</italic> 100 nM), that blocks tachyzoite motility in <italic>T. gondii</italic> (<xref ref-type="bibr" rid="b22">Kieschnick <italic>et al</italic>., 2001</xref>). Members of the CDPK family have been identified in plants, green algae, ciliates and apicomplexan parasites (see <xref ref-type="bibr" rid="b16">Harper and Harmon, 2005</xref> for review). They characteristically combine within the same protein a serine/threonine protein kinase domain with a regulatory carboxy-terminal calmodulin-like domain composed of four calcium-binding EF hands. The <italic>P. falciparum</italic> genome encodes five typical members of the CDPK family but none of these have so far been implicated in regulating gliding motility. CDPK1 is secreted from asexual blood stage parasites by an unusual, acylation-dependent mechanism (<xref ref-type="bibr" rid="b29">Moskes <italic>et al</italic>., 2004</xref>), but its molecular function remains unknown. Another family member, CDPK4, plays a key role in cell cycle regulation during male gamete formation (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>). CDPK3 was also speculated to have a function in gametocytes, based on the transcriptional upregulation of the <italic>cdpk3</italic> gene in sexual stages of <italic>P. falciparum</italic> (<xref ref-type="bibr" rid="b24">Li <italic>et al</italic>., 2000</xref>). However, most recently a <italic>Plasmodium berghei</italic> mutant with a disrupted <italic>cdpk3</italic> gene has provided strong evidence for an essential function of CDPK3 in the ookinete (<xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>., 2006</xref>). The authors of that study conclude CDPK3 has no fundamental function in gliding, but defining a novel transition point in the midgut invasion process, they suggest that CDPK3 is required for ookinetes to traverse a gel-like layer surrounding the blood meal of <italic>Anopheles stephensi</italic> mosquitoes, in which <italic>cdpk3</italic><sup>–</sup> ookinetes migrating out of the blood meal were concluded to become trapped.</p><p>Studying a similar <italic>cdpk3</italic> gene knock-out mutant in <italic>P. berghei</italic> we here confirm an important function of CDPK3 in malaria transmission. Importantly, and in marked contrast to the previous study, we show that <italic>cdpk3</italic><sup>–</sup> ookinetes are severely compromised in productive gliding motility <italic>in vitro</italic>. The crucial role of CDPK3 in regulating gliding motility provides a satisfactory explanation for the strongly reduced transmission of the <italic>cdpk3</italic><sup>–</sup> mutant. We argue that the developmental defect of <italic>cdpk3</italic><sup>–</sup> ookinetes can be understood within the traditional model of mosquito midgut invasion by the ookinete, and that no revision of this model is thus required.</p></sec><sec><title>Results</title><sec><title>Ookinetes need CDPK3 to penetrate the mosquito midgut epithelium</title><p>We generated a <italic>cdpk3</italic> gene knock-out clone in <italic>P. berghei</italic> by replacing the protein-coding regions corresponding to amino acids 181–543 with a resistance marker (<xref ref-type="fig" rid="fig01">Fig. 1A</xref>). Integration of the targeting construct and deletion of the <italic>cdpk3</italic> gene was verified by diagnostic PCR and Southern blot analysis (<xref ref-type="fig" rid="fig01">Fig 1B and C</xref>). <italic>cdpk3</italic><sup>–</sup> parasites had no detectable defect in asexual erythrocytic development in mice and normal numbers of male and female gametocytes were formed (data not shown). When cultured <italic>in vitro</italic>, <italic>cdpk3</italic><sup>–</sup> gametocytes differentiated into gametes, fertilized, and the zygotes developed into morphologically mature ookinetes. Conversion rates from female gametes to ookinetes were equally high in wild-type (80.3 ± 7.8%, <italic>n</italic> = 3) and <italic>cdpk3</italic><sup>–</sup> parasites (76.0 ± 5.6%, <italic>n</italic> = 3), showing that CDPK3 has no essential function up to the stage of ookinete formation. However, when <italic>A. stephensi</italic> mosquitoes were allowed to feed on mice carrying <italic>cdpk3</italic><sup>–</sup> gametocytes, the number of oocysts present on the midgut epithelium 10 days later was reduced by about 98% compared with wild type (<xref ref-type="table" rid="tbl1">Table 1</xref>), a marked decrease that is in good agreement with the 99% blockade observed by <xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>. (2006)</xref> with a separately derived <italic>cdpk3</italic><sup>–</sup> parasite. We conclude that CDPK3 is important in ookinetes to give rise to oocyst infections. To test this hypothesis, equal numbers of wild-type or <italic>cdpk3</italic><sup>–</sup> ookinetes from <italic>in vitro</italic> cultures were offered to mosquitoes in membrane feeders and oocyst numbers counted 10 days later. Infectivity of <italic>cdpk3</italic><sup>–</sup> ookinetes was reduced by 99% (<xref ref-type="table" rid="tbl1">Table 1</xref>, Experiments 8 and 9). When we re-introduced the <italic>pbcdpk3</italic> gene into the knock-out clone on an episomally maintained plasmid (<xref ref-type="fig" rid="fig02">Fig. 2A</xref>), ookinetes expressed a CDPK3 protein with a carboxy-terminal epitope tag (<xref ref-type="fig" rid="fig02">Fig. 2B</xref>), and their infectivity was restored to wild-type levels (<xref ref-type="fig" rid="fig02">Fig. 2C</xref>). These results confirm that a CDPK3-dependent defect at or after the ookinete stage is responsible for the dramatic reduction in oocyst numbers.</p><table-wrap id="tbl1" position="float"><label>Table 1</label><caption><p>Oocyst counts from <italic>A. stephensi</italic> mosquitoes fed either directly on infected mice (experiments 1–7) or on membrane feeders containing uninfected mouse blood mixed with different numbers of ookinetes from cultures (experiments 8 and 9).</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">Clone</th><th align="left" rowspan="1" colspan="1">Oocysts per mosquito <xref ref-type="table-fn" rid="tf1-1">a</xref></th><th align="left" rowspan="1" colspan="1">Prevalence <xref ref-type="table-fn" rid="tf1-2">b</xref></th><th align="right" rowspan="1" colspan="1"><italic>n</italic><xref ref-type="table-fn" rid="tf1-3">c</xref></th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Experiments 1–7 (fed on infected mice)</td><td align="center" rowspan="1" colspan="1">Wild type</td><td align="char" char="." rowspan="1" colspan="1">375.41</td><td align="char" char="." rowspan="1" colspan="1">97.14 (3.80)</td><td align="right" rowspan="1" colspan="1">7</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"><italic>cdpk3</italic><sup>–</sup></td><td align="char" char="." rowspan="1" colspan="1">8.25</td><td align="char" char="." rowspan="1" colspan="1">77.81 (21.99)</td><td align="right" rowspan="1" colspan="1">8</td></tr><tr><td align="left" rowspan="1" colspan="1">Experiment 8 (fed on 800 ookinetes per µl)</td><td align="left" rowspan="1" colspan="1">Wild type</td><td align="char" char="." rowspan="1" colspan="1">38.96</td><td align="char" char="." rowspan="1" colspan="1">100.00</td><td align="right" rowspan="1" colspan="1">1</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"><italic>cdpk3</italic><sup>–</sup></td><td align="char" char="." rowspan="1" colspan="1">0.15</td><td align="char" char="." rowspan="1" colspan="1">15.00</td><td align="right" rowspan="1" colspan="1">1</td></tr><tr><td align="left" rowspan="1" colspan="1">Experiment 9 (fed on 2200 ookinetes per µl)</td><td align="left" rowspan="1" colspan="1">Wild type</td><td align="char" char="." rowspan="1" colspan="1">148.36</td><td align="char" char="." rowspan="1" colspan="1">100.00</td><td align="right" rowspan="1" colspan="1">1</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"><italic>cdpk3</italic><sup>–</sup></td><td align="char" char="." rowspan="1" colspan="1">0.07</td><td align="char" char="." rowspan="1" colspan="1">10.00</td><td align="right" rowspan="1" colspan="1">1</td></tr></tbody></table><table-wrap-foot><fn id="tf1-1"><label>a</label><p>Geometric mean oocyst numbers per mosquito are given</p></fn><fn id="tf1-2"><label>b</label><p>Prevalence of infections is given as the percentage of mosquitoes infected (standard deviation)</p></fn><fn id="tf1-3"><label>c</label><p><italic>n</italic> gives the number of independent infectious feeds. For each feed 20–30 mosquitoes were dissected</p></fn></table-wrap-foot></table-wrap><fig id="fig01" position="float"><label>Fig. 1</label><caption><title>Targeted disruption of the <italic>cdpk3</italic> gene.</title><p>A. Illustration of the <italic>cdpk3</italic> gene locus and the disruption construct used.</p><p>B. Diagnostic PCR showing absence of a <italic>cdpk3</italic>-specific product and presence of an integration-specific product in <italic>cdpk3</italic><sup>–</sup> clone 7.1.</p><p>C. Southern blot analysis of EcoRI/HindIII-digested genomic DNA, showing presence of the <italic>tgdhfr/ts</italic> resistance gene in clone 7.1 and two bands diagnostic of the intact <italic>cdpk3</italic> locus in wild type. The presence of two bands is due to HindIII restriction sites within the <italic>cdpk3</italic> gene.</p></caption><graphic xlink:href="mmi060-1355-f1"/></fig><fig id="fig02" position="float"><label>Fig. 2</label><caption><title>Functional complementation of the <italic>cdpk3</italic><sup>–</sup> mutant.</title><p>A. Schematic illustration of the complementation plasmid, showing the <italic>pbcdpk3</italic> gene fused to an N-terminal double c-myc epitope tag under the control of 2.7 kb upstream sequence and a generic 3′UTR from the <italic>pbdhfr/ts</italic> gene.</p><p>B. Western blot analysis of protein extracts from 10<sup>5</sup> ookinetes of the <italic>cdpk3</italic><sup>–</sup> clone 7.1 and a CDPK3 myc-complemented parasite population. The blot was first probed with the anti-myc mouse monoclonal antibody 9E10 (Sigma, UK). As a loading control the membrane was then stripped and re-probed with an rabbit polyclonal serum raised against <italic>T. gondii</italic> CDPK2 (<xref ref-type="bibr" rid="b22">Kieschnick <italic>et al</italic>., 2001</xref>) that recognizes CDPK4 in <italic>P. berghei</italic> (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>).</p><p>C. Oocyst numbers counted 10 days after mosquitoes were allowed to feed on blood containing 800 ookinetes per µl. Averages oocyst numbers and standard deviations from two experiments using independent ookinetes culture are shown. Twenty-five midguts were dissected in each replicate experiment.</p></caption><graphic xlink:href="mmi060-1355-f2"/></fig><p>To test whether <italic>cdpk3</italic><sup>–</sup> ookinetes retained the genetic potential to undergo ookinete-oocyst development we microinjected equal numbers of <italic>in vitro</italic> cultured wild-type and <italic>cdpk3</italic><sup>–</sup> ookinetes into the haemocoel of <italic>A. stephensi</italic>. When given direct access to the mosquito in this way, both wild-type and <italic>cdpk3</italic><sup>–</sup> ookinetes matured ectopically into oocysts, gave rise to heavy salivary gland infections within 21 days, and produced salivary gland sporozoites that were infectious to mice (<xref ref-type="table" rid="tbl2">Table 2</xref>). We conclude that <italic>cdpk3</italic><sup>–</sup> ookinetes retain their full developmental potential but have a specific defect in overcoming the physical barrier posed by the midgut epithelium. We next asked whether <italic>cdpk3</italic><sup>–</sup> ookinetes could bind and invade midgut epithelial cells. Equal numbers of wild-type and <italic>cdpk3</italic><sup>–</sup> ookinetes (800 per µl) were fed to <italic>A. stephensi</italic> in a membrane feeding apparatus. Midguts were dissected within 2–4 h and processed for immunostaining to visualize ookinetes associated with the gut epithelium (<xref ref-type="fig" rid="fig03">Fig. 3</xref>). On average, 34.9 wild-type ookinetes (±14.3; <italic>n</italic> = 15 midguts) were associated with the midgut cells. In marked contrast, only 1.1 <italic>cdpk3</italic><sup>–</sup> ookinetes (±0.9; <italic>n</italic> = 14 midguts) had accessed the gut.</p><table-wrap id="tbl2" position="float"><label>Table 2</label><caption><p>Salivary gland sporozoites in <italic>A. stephensi</italic> that were either fed on or injected with <italic>in vitro</italic> cultured ookinetes</p></caption><table frame="hsides" rules="groups"><thead><tr><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th colspan="2" align="center" rowspan="1">Infectious to mice <xref ref-type="table-fn" rid="tf2-2">b</xref></th></tr><tr><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th rowspan="1" colspan="1"/><th colspan="2" rowspan="1"><hr/></th></tr><tr><th rowspan="1" colspan="1"/><th align="left" rowspan="1" colspan="1">Clone</th><th align="left" rowspan="1" colspan="1">Sporozoites in glands <xref ref-type="table-fn" rid="tf2-1">a</xref></th><th align="left" rowspan="1" colspan="1">By bite</th><th align="left" rowspan="1" colspan="1">By injection</th></tr></thead><tbody><tr><td align="left" rowspan="1" colspan="1">Fed on 800 ookinetes per µl</td><td align="left" rowspan="1" colspan="1">Wild type</td><td align="center" rowspan="1" colspan="1">19 800 (±4 900)</td><td align="left" rowspan="1" colspan="1">2/2</td><td align="left" rowspan="1" colspan="1">n.d.</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"><italic>cdpk3</italic><sup>–</sup></td><td align="center" rowspan="1" colspan="1">  0 (±0)</td><td align="left" rowspan="1" colspan="1">0/2</td><td align="left" rowspan="1" colspan="1">n.d.</td></tr><tr><td align="left" rowspan="1" colspan="1">Injected with 600 ookinetes</td><td align="left" rowspan="1" colspan="1">Wild type</td><td align="center" rowspan="1" colspan="1">13 000 (±3 200)</td><td align="left" rowspan="1" colspan="1">2/2</td><td align="left" rowspan="1" colspan="1">n.d.</td></tr><tr><td rowspan="1" colspan="1"/><td align="left" rowspan="1" colspan="1"><italic>cdpk3</italic><sup>–</sup></td><td align="center" rowspan="1" colspan="1">4 800 (±200)</td><td align="left" rowspan="1" colspan="1">2/2</td><td align="left" rowspan="1" colspan="1">4/4</td></tr></tbody></table><table-wrap-foot><fn id="tf2-1"><label>a</label><p>Arithmetic mean sporozoites per mosquito are given (± standard deviation from repeat experiments with different ookinete cultures). For each data point 18–30 pairs of glands were examined</p></fn><fn id="tf2-2"><label>b</label><p>Infectivity to mice was determined by allowing mosquitoes to feed on naïve mice or by injection of isolated sporozoites into a tail vein. Whenever sporozoites were observed in the glands, these were highly infectious, resulting in blood infections from day 4 post infection. Data are given as mice positive/mice infected; n.d., not done</p></fn></table-wrap-foot></table-wrap><fig id="fig03" position="float"><label>Fig. 3</label><caption><p><italic>cdpk3</italic><sup>–</sup> ookinetes fail to associate with the mosquito midgut epithelium. Representative midgut epithelial sheets are shown that were dissected 2 h after mosquitoes had fed on equal numbers of cultured wild-type (A–C) or <italic>cdpk3</italic><sup>–</sup> (D–F) ookinetes. Midguts were fixed, permeabilized and immunostained with an antibody against <italic>A. stephensi</italic> annexin to visualized the epithelium (panels A and D), and against the ookinete surface antigen P28 (panels B and E). All images are stacks of confocal sections. A higher magnification inset in (B) shows an ookinete associated with the midgut epithelium.</p></caption><graphic xlink:href="mmi060-1355-f3"/></fig></sec><sec><title>CDPK3 is required for productive gliding motility of ookinetes</title><p>Recognizing that the reduced infectiousness of <italic>cdpk3</italic><sup>–</sup> ookinetes could result from a fundamental defect in motility, we compared the ability of wild-type and <italic>cdpk3</italic><sup>–</sup> ookinetes to glide on glass slides <italic>in vitro.</italic> We exploited the fact that <italic>P. berghei</italic> ookinetes purified from cultures are usually found in aggregates. We find that their dispersal relies on gliding motility and is sensitive to the anti-cytoskeletal agent cytochalasin D. When cocultured with insect cells, aggregates of wild-type ookinetes dispersed rapidly (<xref ref-type="fig" rid="fig04">Fig. 4A</xref> and <xref ref-type="supplementary-material" rid="SD1">Movie S1</xref>). In marked contrast, <italic>cdpk3</italic><sup>–</sup> ookinetes showed a strongly reduced ability to glide productively and aggregates thus failed to disperse (<xref ref-type="fig" rid="fig04">Fig. 4B</xref> and <xref ref-type="supplementary-material" rid="SD2">Movie S2</xref>). Motility of <italic>cdpk3</italic><sup>–</sup> ookinetes was abnormal, showing frequent flexing, bending, twirling and pendular motions, but only rare bouts of translocation over short distances (<xref ref-type="supplementary-material" rid="SD2">Movie S2</xref>). These observations were confirmed in a quantitative analysis, in which the proportion of <italic>cdpk3</italic><sup>–</sup> ookinetes that dispersed from aggregates was scored and compared with wild type, either in the absence or in the presence of cytochalasin D (<xref ref-type="fig" rid="fig04">Fig. 4C</xref>). Prompted by the reported conservation among the motor components across invasive parasites stages (<xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>), we asked whether sporozoite motility was also CDPK3-dependent. Twenty to 25% of wild-type and <italic>cdpk3</italic><sup>–</sup> sporozoites obtained from ookinete-injected mosquitoes produced typical circular motility traces on glass slides (<xref ref-type="fig" rid="fig04">Fig. 4E</xref>) that were indistinguishable from wild type in shape and size (<xref ref-type="fig" rid="fig04">Fig. 4D</xref>). Motility traces were of similar length, corresponding to on average 1.7 ± 0.5 (<italic>n</italic> = 23) complete revolutions in wild type and 1.9 ± 0.8 (<italic>n</italic> = 23) in the <italic>cdpk3</italic><sup>–</sup> sporozoites. Motility in <italic>cdpk3</italic><sup>–</sup> sporozoites was consistent with their normal infectivity for mice. Together with the absence of a <italic>cdpk3</italic><sup>–</sup> phenotype in asexual parasite development, these data suggest a key role for CDPK3 in regulating productive gliding motility specifically at the ookinete stage.</p><fig id="fig04" position="float"><label>Fig. 4</label><caption><title>Motility of wild-type and <italic>cdpk3</italic><sup>–</sup> ookinetes and sporozoites.</title><p>A. Wild-type ookinetes purified from <italic>in vitro</italic> cultures disperse from aggregates after being incubated with insect cells.</p><p>B. <italic>cdpk3</italic><sup>–</sup> ookinetes fail to show productive gliding motility. Each sequence of images is representative of at least three independent experiments with ookinetes cultured from different infected mice.</p><p>C. Quantification of ookinete dispersal. The proportion of dispersed ookinetes was determined after 24 h of culture in the absence and presence of cytochalasin D. Arithmetic means ± standard deviations from three independent experiments are shown.</p><p>D and E. Motility trails of wild-type (D) and <italic>cdpk3</italic><sup>–</sup> sporozoites (E) obtained from salivary glands of ookinete-injected mosquitoes. Trails were visualized using a monoclonal antibody directed against the circumsporozoite protein, which labels both the sporozoites (arrows) and the trails of surface proteins shed by gliding sporozoites. The upper row of each panel shows corresponding immunofluorescence and phase contrast images of the same sporozoites. The bottom row of each panel illustrates additional representative circumsporozoite trails.</p></caption><graphic xlink:href="mmi060-1355-f4"/></fig></sec></sec><sec><title>Discussion</title><p>Due to their stage-specific expression patterns, the plant-like CDPK of apicomplexan parasites have the potential to translate the ubiquitous second messenger calcium into different, stage-specific cellular responses in the parasite. We have previously identified a specific essential role for CDPK4 in the initiation of cell cycle events in male gametocytes of <italic>P. berghei</italic> (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>). Our current analysis of a <italic>cdpk3</italic><sup>–</sup> deletion mutant shows that this protein kinase has no essential function during asexual and sexual intraerythrocytic parasite development and that fertilization and the differentiation of the zygote into morphologically normal ookinetes do not rely on CDPK3. Furthermore, when ookinetes are given direct access to the mosquito haemocoel, CDPK3 is not required for their further differentiation into oocyst and infectious sporozoites. However, our analysis concurs with recent work studying a similar mutant (<xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>., 2006</xref>) in concluding that <italic>cdpk3</italic><sup>–</sup> ookinetes are severely affected in their ability to establish an infection in the mosquito, because they fail to access the midgut epithelium. Our functional analysis is supported by a complementation experiment, which shows that re-introducing a <italic>cdpk3</italic> allele into the <italic>cdpk3</italic><sup>–</sup> mutant on an episomally maintained plasmid restores infectivity.</p><p>Using video microscopy and a quantitative ookinete spreading assay, we show here for the first time that <italic>cdpk3</italic><sup>–</sup> ookinetes have a severely reduced ability to move productively on glass slides and to disperse from aggregates <italic>in vitro</italic>. This motility defect is sufficient to explain the reduced ability of <italic>cdpk3</italic><sup>–</sup> ookinetes to engage and invade the mosquito midgut epithelium. Importantly, <italic>cdpk3</italic><sup>–</sup> ookinetes are still capable of some bending, twirling and pendular movements and limited translocation, which may account for their residual infectivity of 1–2% of wild type. How CDPK3 regulates ookinete gliding remains to be defined. Gliding in other apicomplexan zoites can be regulated through the apical secretion of microneme proteins that form the bridge between the extracellular substrate and the intracellular motor. Microneme secretion is calcium-dependent in many apicomplexan zoites, including malarial sporozoites (<xref ref-type="bibr" rid="b14">Gantt <italic>et al</italic>., 2000</xref>) and in <italic>T. gondii</italic> tachyzoites it is a precondition for motility and host cell invasion (<xref ref-type="bibr" rid="b8">Carruthers <italic>et al</italic>., 1999</xref>). Our unpublished observations suggest ookinete motility is stimulated in the presence of insect cells and – consistent with the regulatory function of CDPK3 – is inhibited by a chelator of cytosolic calcium. However, whether the secretion of the microneme adhesin CTRP is controlled by cytosolic calcium levels and CDPK3 is unknown. Looking at two micronemal proteins of the ookinete, SOAP and CTRP, we have been unable with currently available reagents to establish a suitable microneme secretion assay to examine this possibility.</p><p>Recent <italic>in vivo</italic> imaging studies showed that ookinetes and sporozoites not only initiate and terminate gliding, but display twirling and bending actions, circular and straight-segment movements, and shape changes, in addition to invasive motility (<xref ref-type="bibr" rid="b13">Frischknecht <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b36">Vlachou <italic>et al</italic>., 2004</xref>; <xref ref-type="bibr" rid="b2">Amino <italic>et al</italic>., 2006</xref>). Furthermore, oocyst-derived sporozoites exhibit chemotaxis towards a heat-stable factor from mosquito salivary glands (<xref ref-type="bibr" rid="b1">Akaki and Dvorak, 2005</xref>) and it is tempting to speculate that ookinetes may be able to similarly direct their motility towards the midgut epithelium. This surprisingly large behavioural repertoire probably requires other layers of regulation in addition to microneme secretion. Potential phosphorylation-dependent control points include (i) the nucleation, stabilization or destabilizing of actin filaments by aldolase or other actin-binding molecules, such as toxofilin, which is subject to dynamic phosphorylation in <italic>T. gondii</italic> (<xref ref-type="bibr" rid="b10">Delorme <italic>et al</italic>., 2003</xref>), or (ii) the activity of myosin, which may involve the myosin light chain-like MTIP protein (<xref ref-type="bibr" rid="b4">Bergman <italic>et al</italic>., 2003</xref>). It will be interesting to examine whether molecular components of the motor complex are substrates for CDPK3. The ookinete-specific essential function for CDPK3 is unexpected in view of the universal role of calcium in zoite motility and considering that all invasive stages of <italic>Plasmodium</italic> are thought to use the same molecular motor (<xref ref-type="bibr" rid="b3">Baum <italic>et al</italic>., 2006</xref>). We speculate that in merozoites and sporozoites other members of the CDPK family may have functions similar to CDPK3 in the ookinete.</p><p>Our quantitative assay showing that productive gliding in <italic>cdpk3</italic><sup>–</sup> ookinetes is fundamentally disturbed is in marked disagreement with a recent report investigating ookinete gliding in a similar mutant, but on a gel-like substrate (Matrigel™) that contains mouse basement membrane proteins (<xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>., 2006</xref>). Gliding of <italic>cdpk3</italic><sup>–</sup> ookinetes on the gel surface is reported (but not quantified) as being normal and is contrasted with the failure of <italic>cdpk3</italic><sup>–</sup> ookinetes to penetrate the gel matrix. Analysing mosquito blood meals 20 h after an infectious feed by transmission electron microscopy, <xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>. (2006)</xref> further show that <italic>cdpk3</italic><sup>–</sup> ookinetes accumulate in an electron-lucent layer surrounding the blood meal. Likening this layer to the Matrigel™ used <italic>in vitro</italic>, a novel, CDPK3-dependent transition point in the midgut invasion process, is proposed, at which ookinetes would have to switch from CDPK3-independent gliding through the blood meal to a different, CDPK3-dependent mode of motility to penetrate the layer surrounding the blood meal, before eventually gaining access to the epithelium. In an attempt to reconcile our data with the previous report, we performed a quantitative estimate of ookinete gliding in the two movies published by <xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>. (2006)</xref>. Wild-type ookinetes migrating in Matrigel™ cover a distance corresponding to on average 5.61 times their own length (standard deviation = 1.25, <italic>n</italic> = 16 ookinetes) in 45 min. During the same period <italic>cdpk3</italic><sup>–</sup> ookinetes on the gel surface, while clearly showing some movement, progress by only 0.83 times their own length (standard deviation = 1.33, <italic>n</italic> = 20 ookinetes), and many ookinetes not included in our estimate fail to disperse from an aggregate on the gel surface, just as we have observed with our <italic>cdpk3</italic> KO clone.</p><p>As we show here that <italic>cdpk3</italic><sup>–</sup> ookinetes have a general motility defect, it is no longer necessary to assume CDPK3 has a more specific function for ookinetes when they penetrate the digested periphery of the blood meal. It is well established that haemolysis and digestion proceed centripetally from the periphery of the blood mass inwards (<xref ref-type="bibr" rid="b9">Clements, 1992</xref>) and immotile objects resistant to digestive enzymes, such as <italic>cdpk3</italic><sup>–</sup> ookinetes, must therefore accumulate in the narrow lysis zone that forms between the undigested blood and the midgut epithelium. Finally, our membrane feeding assays showing that mature cultured <italic>cdpk3</italic><sup>–</sup> ookinetes still fail to infect when brought in immediate intimate contact with the midgut epithelium, rule out any specific function for CDPK3 at the hypothetical transition point between the undigested erythrocyte mass and the peripheral layer as suggested by <xref ref-type="bibr" rid="b18">Ishino <italic>et al</italic>. (2006)</xref>. We conclude that a fundamental motility defect is entirely sufficient to explain the phenotype of <italic>cdpk3</italic><sup>–</sup> ookinetes within the traditional model of ookinete interactions with the blood meal. Future studies should focus on dissecting the exact molecular functions of CDPK3 and other CDPK family members to assess the importance of this group of plant-like protein kinases as targets for pharmacological intervention.</p></sec><sec sec-type="methods"><title>Experimental procedures</title><sec><title><italic>Parasite maintenance</italic>, in vitro <italic>culture and transmission to mosquitoes</italic></title><p>The <italic>P. berghei</italic> ANKA wild-type strain 2.34 and transgenic lines were maintained in phenyl hydrazine-treated Theiler’s Original outbred mice and transmitted by <italic>A. stephensi</italic>, strain SD500, as described previously (<xref ref-type="bibr" rid="b31">Sinden <italic>et al</italic>., 2002</xref>). The course of infections and gametocyte production were monitored on Giemsa-stained blood films. Ookinetes were grown <italic>in vitro</italic> by culturing of gametocyte-infected mouse blood, quantified and then purified by selective lysis of erythrocytes as described previously (<xref ref-type="bibr" rid="b31">Sinden <italic>et al</italic>., 2002</xref>). Macrogamete-to-ookinete conversion rates were determined in 24 h cultures by live immunofluorescence microscopy with a Cy3-conjugated monoclonal antibody against the macrogamete/zygote/ookinete surface antigen p28, as previously described (<xref ref-type="bibr" rid="b34">Tewari <italic>et al</italic>., 2005a</xref>). For transmission experiments, batches of 50 mosquitoes, which had been starved overnight, were either fed directly on anaesthetized infected mice on day 4 of a blood-induced infection, or were allowed to feed on a membrane feeder apparatus loaded with a suspension of cultured ookinetes in blood from an uninfected mouse. All feeds were done for 20 min at 19°C. Unfed mosquitoes were removed the following day. Oocysts were counted on dissected midguts on day 10 after feeding. Alternatively, adult female mosquitoes were infected by microinjection of 60 nl of cultured ookinetes into the thorax, using a Nanoject II hand-held microinjector (Drummond, USA). Wild-type and <italic>cdpk3</italic><sup>–</sup>ookinetes for injection were cultured <italic>in vitro</italic> from infected blood, from which leucocytes had been removed (<xref ref-type="bibr" rid="b31">Sinden <italic>et al</italic>., 2002</xref>). Ookinete numbers in 20–24 h cultures were determined in a haemocytometer, ookinete density adjusted to 10<sup>4</sup> per µl, and cultures back-filled into borosilicate injector needles.</p></sec><sec><title><italic>Deletion and complementation of the</italic> cdpk3 <italic>gene</italic></title><p>A targeting vector for <italic>pbcdpk3</italic> was constructed in plasmid pBS-DHFR, in which polylinker sites flank a <italic>T. gondii dhfr/ts</italic> expression cassette conveying resistance to pyrimethamine. A 608 bp fragment comprising 5′ upstream sequence followed by the first 543 bp of exon1 of <italic>cdpk3</italic> was PCR amplified from <italic>P. berghei</italic> genomic DNA and inserted into KpnI and ApaI restriction sites upstream of the <italic>dhfr/ts</italic> cassette of pBS-DHFR. A 686 bp fragment comprising the last two exons and 3′ flanking region of <italic>pbcdpk3</italic> was then inserted downstream of the <italic>dhfr/ts</italic> cassette. The replacement construct was excised as a KpnI/BamHI fragment and used for the electroporation of cultured <italic>P. berghei</italic> schizonts as described (<xref ref-type="bibr" rid="b28">Menard and Janse, 1997</xref>; <xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>). Following dilution cloning of drug-resistant parasites, genotyping of a <italic>cdpk3</italic><sup>–</sup> clone, named 7.1, was carried out by Southern blot analysis and diagnostic PCR across the predicted integration site in principle as described previously for other gene deletions (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>). A complementation vector was constructed in two steps in plasmid p141, a generic version of the myc-tagging vector p142 described previously (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>). First a 2.7 kb fragment representing the <italic>cdpk3</italic> 5′ upstream intergenic region was PCR-amplified from <italic>P. berghei</italic> genomic DNA using oligonucleotides ol175 (3′-GAGA<underline>GGTACC</underline>GGGAAAATGAT GTAACTATAAGATG, restriction site underlined) and ol176 (3′-CAT<underline>GCTAGC</underline>TTTTACGTATTAAACTATTTCCAAAAT) and inserted into KpnI abd NheI digested p142, giving rise to p142/11. Next the complete <italic>cdpk3</italic> genomic sequence, including all introns, but excluding the TAA stop codon, was amplified from genomic DNA using primers ol177 (3′-CAT<underline>GCTAGC</underline>ATGAATCAATTATGTGTAGAAAG) (restriction site underlined) and ol178 (3′-GAGA<underline>GGGCCC</underline>ATACTT TAGTTTCATCATTTCGCAA) and inserted into the NheI and ApaI restriction sites of 142/11, downstream of the putative <italic>cdpk3</italic> promoter and 5′UTR sequence, in frame with a double c-myc epitope tag, followed by the stop codon and 0.5 kb of the <italic>P. berghei dhfr/ts</italic> 3′UTR. The resulting plasmid, p217, was confirmed by sequencing. p217 was introduced into the <italic>cdpk3</italic><sup>–</sup> clone 7.1 by electroporation and maintained as an episome by selection for the <italic>hdhfr</italic> gene as described previously (<xref ref-type="bibr" rid="b5">Billker <italic>et al</italic>., 2004</xref>).</p></sec><sec><title>Staining of infected midguts</title><p>Midguts from fed mosquitoes were dissected at the indicated times in 1.85% paraformaldehyde in phosphate- buffered saline (PBS), pH 7.3. The blood meal was removed and the midgut epithelia were fixed for 1 h in 3.7% paraformaldehyde, washed once in PBS and then incubated for 30 min in PBS, 0.1% TritonX-100, 5% normal goat serum (NGS, Jackson Immunoresearch). Gut epithelial cells were stained with a mouse monoclonal antibody against mosquito annexin (<xref ref-type="bibr" rid="b23">Kotsyfakis <italic>et al</italic>., 2005</xref>), diluted in PBS, 5% NGS, for at least 1 h at room temperature. Three 10 min washes in PBS, 5% NGS, were followed by an incubation with Alexa488-conjugated goat anti-mouse antibody (Molecular Probes) for 1 h. Following another three washes, ookinetes were visualized with a Cy3-conjugated mouse monoclonal antibody 13.1 directed against the ookinete surface protein P28 (<xref ref-type="bibr" rid="b38">Winger <italic>et al</italic>., 1988</xref>). Stained midgut epithelial sheets were mounted in Vectashield (Vector Laboratories) and analysed by confocal microscopy.</p></sec><sec><title>Motility assays</title><p>Ookinetes purified from <italic>in vitro</italic> cultures were mixed on a microscope slide with <italic>Aedes aegytpi</italic> Mos20 cells in M199 medium (Sigma, UK), supplemented with 10% bovine calf serum. Ten microlitres of cell suspension was placed under Vaseline-rimmed coverslips and digital images recorded every 30 s for 10 min. Wild-type and mutant ookinetes were analysed in parallel. To quantify motility, aggregated ookinetes purified from <italic>in vitro</italic> cultures were seeded in duplicates into cocultures with Mos20 cells in LabTek 8-well chamber slides (Nunc, UK). 10 µM cytochalasin D (Calbiochem, UK) was added to one of the two wells for each sample. After 20–24 h ookinetes were visualized by staining with the Pb70 antibody against an ookinete cytoskeletal protein (<xref ref-type="bibr" rid="b30">Siden-Kiamos <italic>et al</italic>., 2000</xref>). Dispersed ookinetes and those that remained in aggregates were counted separately in randomly selected fields under the microscope. At least 200 ookinetes were counted in each sample. The proportion of all ookinetes that had dispersed is a measure of the motility of the ookinetes, while the sample containing cytochalasin D indicates the random occurrence of single ookinetes, as ookinetes are non-motile under these conditions (I. Siden-Kiamos and C. Louis, unpublished). To examine sporozoite gliding, we visualized motility trails using the 3D11 mouse monoclonal antibody (<xref ref-type="bibr" rid="b39">Yoshida <italic>et al</italic>., 1980</xref>), generously provided by Laurent Renia, against the circumsporozoite protein, using a previously described protocol (<xref ref-type="bibr" rid="b35">Tewari <italic>et al</italic>., 2005b</xref>).</p></sec></sec>
|
Investigation of autism and GABA receptor subunit genes in multiple ethnic groups
|
<p>Autism is a neurodevelopmental disorder of complex genetics, characterized by impairment in social interaction and communication, as well as repetitive behavior. Multiple lines of evidence, including alterations in levels of GABA and GABA receptors in autistic patients, indicate that the GABAergic system, which is responsible for synaptic inhibition in the adult brain, may be involved in autism. Previous studies in our lab indicated association of noncoding single nucleotide polymorphisms (SNPs) within a GABA receptor subunit gene on chromosome 4, <italic>GABRA4</italic>, and interaction between SNPs in <italic>GABRA4</italic> and <italic>GABRB1</italic> (also on chromosome 4), within Caucasian autism patients. Studies of genetic variation in African-American autism families are rare. Analysis of 557 Caucasian and an independent population of 54 African-American families with 35 SNPs within <italic>GABRB1</italic> and <italic>GABRA4</italic> strengthened the evidence for involvement of <italic>GABRA4</italic> in autism risk in Caucasians (rs17599165, <italic>p</italic>=0.0015; rs1912960, <italic>p</italic>=0.0073; and rs17599416, <italic>p</italic>=0.0040) and gave evidence of significant association in African-Americans (rs2280073, <italic>p</italic>=0.0287 and rs16859788, <italic>p</italic>=0.0253). The <italic>GABRA4</italic> and <italic>GABRB1</italic> interaction was also confirmed in the Caucasian dataset (most significant pair, rs1912960 and rs2351299; <italic>p</italic>=0.004). Analysis of the subset of families with a positive history of seizure activity in at least one autism patient revealed no association to <italic>GABRA4</italic>; however, three SNPs within <italic>GABRB1</italic> showed significant allelic association; rs2351299 (<italic>p</italic>=0.0163), rs4482737 (<italic>p</italic>=0.0339), and rs3832300 (<italic>p</italic>=0.0253). These results confirmed our earlier findings, indicating <italic>GABRA4</italic> and <italic>GABRB1</italic> as genes contributing to autism susceptibility, extending the effect to multiple ethnic groups and suggesting seizures as a stratifying phenotype.</p>
|
<contrib contrib-type="author"><name name-style="western"><surname>Collins</surname><given-names>Ann L.</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Ma</surname><given-names>Deqiong</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Whitehead</surname><given-names>Patrice L.</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Martin</surname><given-names>Eden R.</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Wright</surname><given-names>Harry H.</given-names></name><xref ref-type="aff" rid="Aff2">2</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Abramson</surname><given-names>Ruth K.</given-names></name><xref ref-type="aff" rid="Aff2">2</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Hussman</surname><given-names>John P.</given-names></name><xref ref-type="aff" rid="Aff3">3</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Haines</surname><given-names>Jonathan L.</given-names></name><xref ref-type="aff" rid="Aff4">4</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Cuccaro</surname><given-names>Michael L.</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author"><name name-style="western"><surname>Gilbert</surname><given-names>John R.</given-names></name><xref ref-type="aff" rid="Aff1">1</xref></contrib><contrib contrib-type="author" corresp="yes"><name name-style="western"><surname>Pericak-Vance</surname><given-names>Margaret A.</given-names></name><address><phone>+1-919-6842063</phone><fax>+1-919-6840910</fax><email>[email protected]</email></address><xref ref-type="aff" rid="Aff1">1</xref></contrib><aff id="Aff1"><label>1</label>Center for Human Genetics, Duke University Medical Center, Durham, NC USA </aff><aff id="Aff2"><label>2</label>School of Medicine, University of South Carolina, Columbia, SC USA </aff><aff id="Aff3"><label>3</label>Hussman Foundation, Ellicott City, MD USA </aff><aff id="Aff4"><label>4</label>The Center for Human Genetics Research, Vanderbilt University Medical Center, Nashville, TN USA </aff>
|
Neurogenetics
|
<sec id="Sec1" sec-type="introduction"><title>Introduction</title><p>Autism is a neurodevelopmental disorder characterized by severe impairment in social interaction and communication, as well as repetitive behavior. Autism is part of a spectrum of disorders denoted autism spectrum disorders (ASD), which in addition to autism, include Asperger syndrome, childhood disintegrative disorder, and pervasive developmental disorder not otherwise specified (PDD-NOS). Onset of this disorder is typically before 3 years of age. The incidence of autism is estimated at approximately one in 1,000 individuals, with an increased incidence in males [<xref ref-type="bibr" rid="CR1">1</xref>–<xref ref-type="bibr" rid="CR3">3</xref>]. Incidences of approximately two to three in 1,000 are reported when a broader diagnosis includes the entire spectrum of pervasive developmental disorders [<xref ref-type="bibr" rid="CR2">2</xref>, <xref ref-type="bibr" rid="CR4">4</xref>].</p><p>Evidence indicates that this disease has a strong genetic component. Twin studies have shown that monozygote twins have a higher recurrence rate than dizygotic twins. While the actual percentages have varied, recurrence rates have been measured as high as 95% in monozygote twins and 23% in dizygotic twins [<xref ref-type="bibr" rid="CR5">5</xref>–<xref ref-type="bibr" rid="CR7">7</xref>]. Despite this strong evidence for a genetic component, the inheritance appears to be complex, with estimates that more than 15 genes may be involved in its inheritance [<xref ref-type="bibr" rid="CR8">8</xref>, <xref ref-type="bibr" rid="CR9">9</xref>], with different families carrying different combinations of contributing genes. The genes may act independently or interactively, adding to the complexity of determining the genetic contribution to this disorder.</p><p>There are two approaches to identifying genetic contributors to disease. The first is a genome wide search in which linkage or association analysis is used to identify regions of the genome that may contain autism susceptibility genes. The second is the candidate gene approach, which investigates a specific gene or genes for involvement in autism risk. In the candidate gene approach, genes are chosen for study based on either what is known about the gene’s function, its location (for example in a recognized linkage peak), or a combination of both. Several candidates are hypothesized to be involved in autism; however, no single candidate gene has consistently emerged as involved in autism risk.</p><p>One candidate pathway that is hypothesized to be involved in autism is the GABAergic system. Hussman [<xref ref-type="bibr" rid="CR10">10</xref>] suggested that autism is the result of an imbalance of the excitatory glutamatergic and inhibitory GABAergic pathways, resulting in overstimulation in the brain and inability to filter out excess stimuli from environmental and intrinsic sources. This theory is supported by multiple lines of evidence. First, histological, biochemical, and molecular approaches have demonstrated altered levels and distribution of GABA and GABA receptors in peripheral blood and plasma, as well as in the brain, including decreased GABA-A receptors and benzodiazepine binding sites in the hippocampal formation [<xref ref-type="bibr" rid="CR11">11</xref>–<xref ref-type="bibr" rid="CR13">13</xref>]. There are also reported alterations in GABAergic neurons, as demonstrated by the increased packing density of GABAergic interneurons in the CA3 and CA1 subfields, and by the decreased numbers and reduced size of cerebellar GABAergic Purkinje cells [<xref ref-type="bibr" rid="CR14">14</xref>, <xref ref-type="bibr" rid="CR15">15</xref>]. In addition, duplications and isodicentric chromosomes in the region containing the three clustered GABA receptor subunits <italic>GABRB3</italic>, <italic>GABRA5</italic>, and <italic>GABRG3</italic> on chromosome 15q have been associated with autism [<xref ref-type="bibr" rid="CR16">16</xref>, <xref ref-type="bibr" rid="CR17">17</xref>]. As well, evidence for both linkage and allelic association have been reported for this same GABA gene cluster, although the findings have not been consistent across datasets [<xref ref-type="bibr" rid="CR18">18</xref>–<xref ref-type="bibr" rid="CR22">22</xref>]. Investigation of association of GABA receptor subunits outside of the chromosome 15 region has been limited [<xref ref-type="bibr" rid="CR23">23</xref>]. Lastly, mutations have been reported in multiple GABA receptor genes in families with epilepsy [<xref ref-type="bibr" rid="CR24">24</xref>]. Given the high comorbidity of autism with epilepsy and seizures, these data suggest that a similar molecular etiology could exist between the disorders.</p><p>Signaling in the GABAergic system is mediated by receptors for the neurotransmitter GABA. There are 19 known GABA receptor subunits arranged in clusters throughout the genome. Functional pentamers formed by various combinations of these subunits result in receptors of varying properties and sensitivities. The amounts and functional capabilities of individual receptor subunits that form a specific pentamer can affect the amount and quality of signaling in different parts of the brain.</p><p>Previously published work from our laboratory analyzed 14 autosomal GABA receptor genes on four different chromosomes, using 70 SNPs in a Caucasian dataset of 470 families. This analysis revealed allelic association to rs1912960 (<italic>p</italic>=0.01) within gamma-aminobutyric acid (GABA) A receptor, alpha-4 (<italic>GABRA4</italic>) on chromosome 4 [<xref ref-type="bibr" rid="CR23">23</xref>] and a significant interaction with rs2351299 within the neighboring gene gamma-aminobutyric acid (GABA) A receptor, beta-1 (<italic>GABRB1</italic>).</p><p>Despite similar prevalence rates between Caucasian and African-Americans [<xref ref-type="bibr" rid="CR25">25</xref>, <xref ref-type="bibr" rid="CR26">26</xref>], autism studies in African-Americans are rare. Risk alleles may be different between ethnic groups or the same risk alleles may have differential effects in each ethnic group, warranting studies in multiple groups. Evidence that phenotypic factors, including indicators of language development, may be more severe in African-Americans, compared to non-Hispanic Caucasians, [<xref ref-type="bibr" rid="CR27">27</xref>] is consistent with these possibilities and underscores the need to investigate autism in different ethnic groups. In this study, we present a validation of our previous report in an independent dataset of 54 African-American families, as well as confirmation of our previous results in an expanded Caucasian sample of 557 autism families.</p></sec><sec id="Sec2" sec-type="materials|methods"><title>Materials and methods</title><sec id="Sec3"><title>Samples</title><p>All families were drawn from a large multisite study of autism genetics conducted in the southeastern United States. These families are recruited through the Center for Human Genetics (CHG) at Duke University Medical Center (DUMC), the University of South Carolina, and the Center for Human Genetic Research at Vanderbilt (<italic>N</italic>=54 African-American and 557 Non-Hispanic Caucasian families) through support groups, advertisements, and clinical and educational settings. All sites recruited, enrolled, and sampled individuals with autism and family members, per study protocols approved by their respective institutional review boards (IRBs). Written informed consent was obtained from parents and from children who were able to give informed consent. Families were enrolled based on probands meeting the following core inclusion criteria of: (1) probands ranging from three to 21 years of age; (2) a presumptive clinical diagnosis of autism; and (3) an expert clinical diagnosis of autism using DSM-IV criteria [<xref ref-type="bibr" rid="CR28">28</xref>], supported by the Autism Diagnostic Interview-Revised (ADI-R) [<xref ref-type="bibr" rid="CR29">29</xref>] and in some cases, the Autism Diagnostic Observation Schedule (ADOS) [<xref ref-type="bibr" rid="CR30">30</xref>]. Our original ascertainment protocol relied on clinical expertise and the ADI-R. The ADOS was subsequently added as a required diagnostic method, and families that were missing ADOS assessments were updated as available. To assure valid ADI-R results, all participants who met current diagnostic criteria for autism were included only if they had a minimal developmental level of 18 months, as extrapolated from the Vineland Adaptive Behavior Scale score [<xref ref-type="bibr" rid="CR31">31</xref>], or had an IQ equivalent greater than 35 based on scores from a standardized measure of cognitive ability such as an the Wechsler Scale of Intelligence for Children—4th edition [<xref ref-type="bibr" rid="CR32">32</xref>], Differential Abilities Scale [<xref ref-type="bibr" rid="CR33">33</xref>], Mullen Scales of Early Learning [<xref ref-type="bibr" rid="CR34">34</xref>] or Leiter Intelligence Scale-Revised [<xref ref-type="bibr" rid="CR35">35</xref>]. IQ data was derived from medical records or direct assessment. Exclusion criteria for participation in the larger genetics study included: severe sensory problems (e.g., visual impairment or hearing loss), significant motor impairments (e.g., failure to sit by 12 months, or walk by 24 months), or identified metabolic, genetic, or progressive neurological disorders, based on screening by clinical staff. Additional samples are from the Autism Genetic Research Exchange (AGRE). These individuals were qualified using similar methods including ADI-R, ADOS, VABS, and a standardized measure of IQ.</p><p>Thirty-nine African-American families were used in an initial GABA receptor screen. Follow-up analysis of significant findings was performed in 54 African-American families. Analysis of the extended Caucasian dataset included 557 non-Hispanic Caucasian families. One-hundred and five new non-Hispanic Caucasian families were added to the analysis (18 families previously analyzed by Ma et al. [<xref ref-type="bibr" rid="CR23">23</xref>] were newly identified as Hispanic, and were omitted from the current study in an effort decrease heterogeneity in the Caucasian dataset).</p><p>Classification of history of seizure activity in autism patients was based on question 92 from the ADI-R, which queries for both current and lifetime presence of convulsions, seizures, and epilepsy. Caregiver responses to question 92 are coded to indicate no seizure activity, seizure activity with no definitive diagnosis of epilepsy, and seizures with a definite diagnosis of epilepsy. Using lifetime ratings, two groups of families were defined: those in which no seizure activity was reported, and those in which seizure activity was present in at least one autism patient. In addition, question 92 allows for coding of febrile seizures. Families with only febrile seizures were classified as negative for seizure activity and not included in the seizure subset analysis. Both families with positive and negative history of seizure activity were included in our overall dataset. This resulted in a dataset of 41 Caucasian families with a positive history for seizures.</p></sec><sec id="Sec4"><title>Molecular analyses and genotyping</title><p>The analysis of 14 GABA receptor subunit genes was performed in 39 African-American families as previously described [<xref ref-type="bibr" rid="CR23">23</xref>]. Briefly, between three and seven intronic, UTR and synonymous coding SNPs within each gene were identified from Applied Biosystems (ABI, Foster City, CA, USA) Assay on Demand (AoD) products, resulting in 70 SNPs within 14 GABA receptor genes on four autosomes. Genes analyzed were: <italic>GABRA1</italic>, <italic>GABRA6</italic>, <italic>GABRB2</italic>, <italic>GABRG2</italic>, and <italic>GABRP</italic> from chromosome 5; <italic>GABRA2</italic>, <italic>GABRA4</italic>, <italic>GABRB1</italic>, and <italic>GABRG1</italic> from chromosome 4; <italic>GABRB3</italic>, <italic>GABRA5</italic>, and <italic>GABRG3</italic> from chromosome 15; and <italic>GABRR1</italic> and <italic>GABRR2</italic> from chromosome 6.</p><p>Additional SNPs within <italic>GABRA4</italic> and <italic>GABRB1</italic> were analyzed in the extended African-American (<italic>N</italic>=54) and Caucasian (<italic>N</italic>=557) datasets to expand the coverage of variation across this region. Thirty-five SNPs, representative of different linkage disequilibrium (LD) blocks across the two genes (20 in <italic>GABRA4</italic> and 15 in <italic>GABRB1</italic>), were genotyped (Fig. <xref rid="Fig1" ref-type="fig">1</xref>). SNP’s for genotyping were selected from online databases (University of California Santa Cruz <ext-link ext-link-type="uri" xlink:href="http://genome.ucsc.edu">http://genome.ucsc.edu</ext-link> and NCBI dbSNP <ext-link ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=snp">http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=snp</ext-link>) and from resequencing of exons and surrounding areas of both <italic>GABRB1</italic> and <italic>GABRA4</italic> genes.
<fig id="Fig1"><label>Fig. 1</label><caption><p>Approximate locations of SNPs for extended analysis within <italic>GABRA4</italic> and <italic>GABRB1</italic>. <italic>Horizontal arrows</italic> represent transcriptional direction. <italic>Vertical arrows</italic> represent SNPs, as numbered and identified below the diagram. <italic>Pter</italic> indicates the P-terminal end of the chromosome and <italic>cen</italic> indicates the centromere</p></caption><graphic position="anchor" xlink:href="10048_2006_45_Fig1_HTML" id="MO1"/></fig></p><p>SNP genotyping was performed using Taqman allelic discrimination assays (Applied Biosystems). DNA was extracted from whole blood according to established protocols [<xref ref-type="bibr" rid="CR36">36</xref>], and 3 ng of genomic DNA was used per reaction. Amplification was performed on GeneAmp PCR Systems 9700 thermocyclers, with cycling conditions as recommended by Applied Biosystems. Fluorescence was measured using Applied Biosystem’s 7900. Genotype discrimination was conducted using ABI Prism SDS 2.1 software. Quality control, to ensure accurate genotyping, involved two different CEPH DNAs in quadruplicate on each 384 well plate, as well as the presence of samples which were replicated elsewhere in the sample list. Additionally, ≥95% genotyping efficiency is required.</p></sec><sec id="Sec5"><title>Statistical analysis</title><p>To ensure genotyping quality, Pedcheck was run for detection of Mendelian inheritance inconsistency. One affected and one unaffected individual from each family were selected randomly for tests of Hardy–Weinberg equilibrium (HWE), which was assessed using exact tests from the Genetic Data Analysis program [<xref ref-type="bibr" rid="CR37">37</xref>]. Pairwise LD between markers was calculated using graphical overview of linkage disequilibrium (GOLD) [<xref ref-type="bibr" rid="CR38">38</xref>] in the parents of autism cases for both the African-American and Caucasian samples. LD was evaluated in parents to increase the available sample size for analysis and comparison between the two ethnic groups. The pedigree disequilibrium test (PDT) and its extension the genotypic pedigree disequilibrium test (genoPDT) [<xref ref-type="bibr" rid="CR39">39</xref>, <xref ref-type="bibr" rid="CR40">40</xref>] were used to test for association to autism susceptibility.</p><p>The EMDR [<xref ref-type="bibr" rid="CR23">23</xref>, <xref ref-type="bibr" rid="CR41">41</xref>], an extension of the MDR [<xref ref-type="bibr" rid="CR42">42</xref>, <xref ref-type="bibr" rid="CR43">43</xref>], was used to test for potential gene–gene interaction, to identify specific locus combinations of interest for further investigation and validation of previous results. EMDR analysis was performed using seven SNPs, the four in <italic>GABRA4</italic> found to show significant allelic or genotypic association in the Caucasian sample-set, and the three in <italic>GABRB1</italic> found to be significant in the seizure subgroup. One-, two-, and three-way analysis was performed on the Caucasian dataset. For case-control pairs used in EMDR, the proband (or most completely genotyped affected child) from each multiplex and triad family was selected (<italic>n=</italic>470 total) as a case. Controls were generated using the untransmitted alleles from parental genotypes. In this study, a cross-validation option was not utilized. The sample size of the African-American dataset is too small to provide power for EMDR analysis, not allowing us to test for validation of the interaction seen in the paper of Ma et al. within the African-American dataset.</p><p>The haplotype family-based association test (HBAT; [<xref ref-type="bibr" rid="CR44">44</xref>]) was used for haplotype association analysis using the significant SNPs in <italic>GABRA4</italic>.</p></sec></sec><sec id="Sec6" sec-type="results"><title>Results</title><p>Allelic association studies of 70 SNPs across the 14 GABA receptor subunit genes in the 39 African-American screen set of families, revealed association in rs2280073 (<italic>GABRA4</italic>; <italic>p</italic>=0.0053) and hcv2119841 (<italic>GABRB1</italic>; <italic>p</italic>=0.0343), the same two genes identified through allelic association and interaction analysis in the Caucasian dataset [<xref ref-type="bibr" rid="CR23">23</xref>]. Genotypic association analysis revealed the same <italic>GABRA4</italic> SNP, rs2280073 (<italic>p</italic>=0.0262), and marginal significance within <italic>GABRP</italic>, rs1862242 (<italic>p</italic>=0.0471). The remaining SNPs showed no significant association (data not shown).</p><p>Analysis of the screening SNPs and newly identified SNPs within <italic>GABRA4</italic> and <italic>GABRB1</italic> in the Caucasian population (<italic>N</italic>=557), and within the extended African-American population (<italic>N</italic>=54) (Table <xref rid="Tab1" ref-type="table">1</xref>), revealed new SNPs with significant association. In the Caucasian dataset, rs1912960, also significant in the study of Ma et al. (<italic>p</italic>=0.012), showed significant allelic association (<italic>p</italic>=0.0073). Additional significant SNPs were identified in <italic>GABRA4</italic> as well, rs17599165 (<italic>p</italic>=0.0015) and rs17599416 (<italic>p</italic>=0.0040). Genotypic association was also seen in these SNPs (<italic>p</italic>=0.0046, 0.0009, and 0.0043, respectively), as well as in a fourth SNP, also in <italic>GABRA4</italic> (rs7660336, <italic>p</italic>=0.0368). In the African-American dataset, rs2280073 (<italic>p</italic>=0.0287), identified in the smaller African-American dataset above, and rs16859788 (<italic>p</italic>=0.0253), were found to be associated with the allele based test. Genotypic association was also identified in rs16859788 (<italic>p</italic>=0.0412). No SNPs within <italic>GABRB1</italic> were found to be associated with autism in either ethnic group.
<table-wrap id="Tab1"><label>Table 1</label><caption><p>Analysis of <italic>GABRA4</italic> and <italic>GABRB1</italic> in extended Caucasian and African-American datasets</p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="5"><italic>GABRA4</italic></th><th colspan="5">GABRB1</th></tr><tr><th rowspan="2">SNP</th><th colspan="2">Caucasian</th><th colspan="2">African-American</th><th rowspan="2">SNP</th><th colspan="2">Caucasian</th><th colspan="2">African-American</th></tr><tr><th>PDT<sup>a</sup></th><th>Geno PDT<sup>a</sup></th><th>PDT<sup>a</sup></th><th>Geno PDT<sup>a</sup></th><th>PDT</th><th>Geno PDT</th><th>PDT</th><th>Geno PDT</th></tr></thead><tbody><tr><td>RS7678338</td><td>0.9350</td><td>0.9923</td><td>0.2230</td><td>0.4190</td><td>RS1866989</td><td>0.3071</td><td>0.3860</td><td>0.5553</td><td>0.7892</td></tr><tr><td>RS6447517</td><td>0.8826</td><td>0.7034</td><td>0.505</td><td>0.3930</td><td>RS2351299</td><td>0.4529</td><td>0.0822</td><td>0.5775</td><td>0.8614</td></tr><tr><td>RS17599102</td><td>0.8055</td><td>0.8913</td><td>0.7518</td><td>0.8805</td><td>RS10016388</td><td>0.1585</td><td>0.2660</td><td>0.2367</td><td>0.4259</td></tr><tr><td>RS7660336</td><td>0.0833</td><td><bold>0.0368(G/G)</bold></td><td>0.5164</td><td>0.7410</td><td>RS1372496</td><td>0.2362</td><td>0.3740</td><td>0.2482</td><td>0.3715</td></tr><tr><td>RS1512136</td><td>0.9052</td><td>0.9869</td><td>0.2888</td><td>0.3575</td><td>RS3114084</td><td>0.0934</td><td>0.1942</td><td>0.2059</td><td>0.3281</td></tr><tr><td>RS17599165</td><td><bold>0.0015(T)</bold></td><td><bold>0.0009(T/T)</bold></td><td>0.6547</td><td>0.4304</td><td>RS4482737</td><td>0.1495</td><td>0.2504</td><td>1.0000</td><td>1.0000</td></tr><tr><td>HCV1592545</td><td>0.7798</td><td>0.9427</td><td>0.2230</td><td>0.5313</td><td>HCV11353524</td><td>0.1959</td><td>0.3117</td><td>1.0000</td><td>1.0000</td></tr><tr><td>RS7685553</td><td>1.0000</td><td>0.8419</td><td>0.4913</td><td>0.6044</td><td>RS3775534</td><td>0.1893</td><td>0.1831</td><td>0.4913</td><td>0.4913</td></tr><tr><td>RS1912960</td><td><bold>0.0073(C)</bold></td><td><bold>0.0046(C/C)</bold></td><td>0.4111</td><td>0.5110</td><td>HCV2119841</td><td>0.3352</td><td>0.0838</td><td>0.2278</td><td>0.4111</td></tr><tr><td>RS2055943</td><td>0.9671</td><td>0.9434</td><td>0.4927</td><td>0.7607</td><td>RS6287</td><td>0.4045</td><td>0.3571</td><td>0.6171</td><td>0.6984</td></tr><tr><td>RS2280073</td><td>0.1404</td><td>0.0955</td><td><bold>0.0287(G)</bold></td><td>0.1100</td><td>RS6289</td><td>0.9349</td><td>0.5554</td><td>0.8658</td><td>0.9220</td></tr><tr><td>RS16859788</td><td>0.3173</td><td>0.3173</td><td><bold>0.0253(A)</bold></td><td><bold>0.0412(A/A)</bold></td><td>RS6290</td><td>0.4285</td><td>0.1973</td><td>0.3173</td><td>0.4594</td></tr><tr><td>RS17599416</td><td><bold>0.0040(A)</bold></td><td><bold>0.0043(A/A)</bold></td><td>0.8084</td><td>0.8084</td><td>4P0413</td><td>1.0000</td><td>1.0000</td><td>0.7389</td><td>0.7389</td></tr><tr><td>RS3792208</td><td>1.0000</td><td>0.4980</td><td>0.1797</td><td>0.1797</td><td>RS10028945</td><td>0.3272</td><td>0.4584</td><td>1.0000</td><td>0.8179</td></tr><tr><td>RS10517174</td><td>0.9484</td><td>0.0894</td><td>0.7150</td><td>0.8903</td><td>RS3832300</td><td>0.4094</td><td>0.6352</td><td>0.6547</td><td>0.6547</td></tr><tr><td>RS7694035</td><td>0.4337</td><td>0.6266</td><td>0.3657</td><td>0.3657</td><td/><td/><td/><td/><td/></tr><tr><td>RS3792211</td><td>0.9057</td><td>0.8382</td><td>0.6547</td><td>0.2895</td><td/><td/><td/><td/><td/></tr><tr><td>RS2229940</td><td>0.7873</td><td>0.9375</td><td>0.5485</td><td>0.7866</td><td/><td/><td/><td/><td/></tr><tr><td>RS13151759</td><td>0.7529</td><td>0.9326</td><td>0.2367</td><td>0.5759</td><td/><td/><td/><td/><td/></tr><tr><td>RS13151769</td><td>0.4740</td><td>0.7436</td><td>0.1824</td><td>0.4768</td><td/><td/><td/><td/><td/></tr></tbody></table><table-wrap-foot><p>Bold indicates significant values</p><p><sup>a</sup>Associated allele/genotype shown in parenthesis</p></table-wrap-foot></table-wrap></p><p>The majority of pairwise <italic>r</italic><sup>2</sup> values between the significant SNPs were less than 0.3, in both ethnic groups (Table <xref rid="Tab2" ref-type="table">2</xref>). However, a few SNPs have values between 0.3 and 0.35. SNPs, rs17599165, and rs17599416 have <italic>r</italic><sup>2</sup> values of 0.709 in African-Americans and 0.853 in Caucasians, and rs7660336 and rs2280073 have a pairwise <italic>r</italic><sup>2</sup> of 0.907 in Caucasians and 0.905 in African-Americans. Allele frequencies are similar, yet not identical between the two groups. One SNP, however, did have large differences in minor allele frequencies (MAFs) between the two groups, with a MAF of 0.24 in African-Americans but only 0.001 in Caucasians (Table <xref rid="Tab2" ref-type="table">2</xref>). Haplotype analysis, using the four SNPs with significant allelic or genotypic association in the Caucasian families, revealed a significant global test (<italic>p</italic>=0.014) in the Caucasian population, further supporting the involvement of these SNPs or another variant on the haplotype background. The haplotype composed of all risk alleles trends towards significance with autism as a risk haplotype (<italic>p</italic>=0.0763), while the haplotype of all protective alleles resulted in being highly significantly protective (<italic>p</italic>=0.0006; Table <xref rid="Tab3" ref-type="table">3</xref>)
<table-wrap id="Tab2"><label>Table 2</label><caption><p>Minor allele frequencies and linkage disequilibrium in Caucasian and African-American datasets</p></caption><graphic xlink:href="10048_2006_45_Tab1_HTML" id="MO2"/><table-wrap-foot><p>Bold indicates significant values</p><p><sup>a</sup><italic>MAF</italic> minor allele frequency</p></table-wrap-foot></table-wrap><table-wrap id="Tab3"><label>Table 3</label><caption><p>Haplotype frequency and association in Caucasian population</p></caption><table frame="hsides" rules="groups"><thead><tr><th>RS7660336</th><th>RS17599165</th><th>RS1912960</th><th>RS17599416</th><th>Haplotype frequency</th><th><italic>P</italic>-value</th></tr></thead><tbody><tr><td>G</td><td>T</td><td>C</td><td>A</td><td>0.503</td><td>0.0763</td></tr><tr><td>C</td><td>T</td><td>C</td><td>A</td><td>0.277</td><td>0.9214</td></tr><tr><td>C</td><td>T</td><td>G</td><td>A</td><td>0.124</td><td>0.6094</td></tr><tr><td>C</td><td>A</td><td>G</td><td>G</td><td>0.082</td><td><bold>0.0006</bold></td></tr></tbody></table><table-wrap-foot><p>Bold indicates significant values</p></table-wrap-foot></table-wrap></p><p>Subsetting of the <italic>GABRA4</italic> and <italic>GABRB1</italic> data to analyze all families with positive history for seizures revealed no association to <italic>GABRA4</italic>. However, three SNPs within <italic>GABRB1</italic> were found to be both allelically and genotypically associated with autism: rs2351299 (<italic>p</italic>=0.0163 and <italic>p</italic>=0.0189 for PDT and genoPDT, respectively), rs4482737 (<italic>p</italic>=0.0339 and <italic>p</italic>=0.0339), and rs3832300 (<italic>p</italic>=0.0253 and <italic>p</italic>=0.0357). These three SNPs all had pairwise <italic>r</italic><sup>2</sup> values less than 0.1 (data not shown).</p><p>In the Caucasian population, EMDR verified the single locus effect identified through PDT analysis in rs1912960 (<italic>p</italic>=0.024), and identified two different significant two-locus gene–gene effects between <italic>GABRA4</italic> and <italic>GABRB1</italic>, rs1912960 with rs2351299 (<italic>p</italic>=0.004), and rs17599416 with rs2351299 (<italic>p</italic>=0.014). Several three locus effects were also significant [rs7660336, rs1912960, and rs2351299 (<italic>p</italic>=0.012); rs17599165, rs1912960, and rs2351299 (<italic>p</italic>=0.012); rs1912960, rs17599416, and rs2351299 (<italic>p</italic>=0.038); and rs7660336, rs17599416, and rs2351299 (<italic>p</italic>=0.047)] (Table <xref rid="Tab4" ref-type="table">4</xref>).
<table-wrap id="Tab4"><label>Table 4</label><caption><p>EMDR results in Caucasian dataset between <italic>GABRA4</italic> and <italic>GABRB1</italic></p></caption><table frame="hsides" rules="groups"><thead><tr><th colspan="3">Input SNPs</th><th colspan="3">Significant interactions</th></tr><tr><th>Gene</th><th>SNP number</th><th>SNP</th><th/><th>SNPs</th><th><italic>P</italic>-values</th></tr></thead><tbody><tr><td rowspan="4">GABRA4</td><td>1</td><td>RS7660336</td><td>One-way</td><td>3</td><td>0.024</td></tr><tr><td>2</td><td>RS17599165</td><td>Two-way</td><td>3, 5</td><td>0.004</td></tr><tr><td>3</td><td>RS1912960</td><td/><td>4, 5</td><td>0.014</td></tr><tr><td>4</td><td>RS17599416</td><td>Three-way</td><td>1, 3, 5</td><td>0.012</td></tr><tr><td rowspan="3">GABRB1</td><td>5</td><td>RS2351299</td><td/><td>2, 3, 5</td><td>0.012</td></tr><tr><td>6</td><td>RS4482737</td><td/><td>3, 4, 5</td><td>0.038</td></tr><tr><td>7</td><td>RS3832300</td><td/><td>1, 4, 5</td><td>0.047</td></tr></tbody></table></table-wrap></p></sec><sec id="Sec7" sec-type="discussion"><title>Discussion</title><p>We have confirmed the involvement of <italic>GABRA4</italic> in autism through identification of significantly associated SNPs within an independent African-American population. Furthermore, we have strengthened our original findings, including identification of additional associated SNPs and a significant interaction between <italic>GABRA4</italic> and <italic>GABRB1</italic> and in an extended dataset (<italic>N</italic>=557) of Caucasian autism families. The AA dataset contains only 54 families, and in general power can be a problem in small sample sets if an effect is not seen. However, for the <italic>GABRA4</italic> gene, we did find significant results in the AA dataset. Furthermore, studies in small datasets such as the AA dataset can also be less robust than in larger datasets. However, the fact that we find the same gene, <italic>GABRA4</italic>, significant in two different ethnic groups is supportive of the role of <italic>GABRA4</italic> in autism risk.</p><p>The identification of two different two-way interactions between <italic>GABRA4</italic> and <italic>GABRB1</italic> provides additional evidence of the complex interaction of these two genes in autism. The rs1912960 (<italic>GABRA4</italic>) with rs2351299 (<italic>GABRB1</italic>) interaction is between the same two SNPs previously reported in the study of Ma et al., and is still significant in our larger dataset. A second significant two-way interaction was found, also including rs2351299 in <italic>GABRB1</italic> as well as rs17599416 in <italic>GABRA4</italic>, further supporting that interactions between these two genes are involved in autism risk. Although rs2351299 does not have a significant PDT result, it is found in both significant interaction pairs, and in each of the significant three-way interactions. The MDR approach used in these analyses is specifically designed to detect interaction effects both in the presence and absence of main effects.</p><p>It is possible that both of the two-way interactions are being identified due to LD between the two <italic>GABRA4</italic> SNPs (rs1912960 and rs17599416). There is significant correlation between the two SNPs, although the <italic>r</italic><sup>2</sup> value (Table <xref rid="Tab2" ref-type="table">2</xref>) between them is not large (0.32). Given that these do not appear to be causative variants, it is likely that the true variant, yet to be identified, is in LD with these <italic>GABRA4</italic> SNPs. Examination of interaction in the independent dataset of African-American families was not possible due to the limited sample size.</p><p>We also identified variants within <italic>GABRB1</italic> as associated within the autistic population with seizures. One of these variants, rs2351299, also shows a significant interaction with SNPs in <italic>GABRA4</italic> (Table <xref rid="Tab4" ref-type="table">4</xref>). While no effect was seen in <italic>GABRA4</italic> for the seizure subset, the sample size may be too small to conclusively determine its role in seizure status in autism. However, the enhanced findings in <italic>GABRB1</italic> in the seizure subset implicate <italic>GABRB1</italic> as a contributor to genetic risk in these patients.</p><p>Despite the identification of <italic>GABRA4</italic> in both ethnic groups, different SNPs were found to be associated. The identification of distinct SNPs within these populations may indicate differences in allele frequency and linkage disequilibrium within the two racial groups, differences in the haplotypic background in which identical causative variations originated, or differences in the causative variation. SNP rs16859788 for example, which is significant in the African-American group, shows little variation in the Caucasian dataset, therefore, providing no power for detection of an effect in this group. Other SNPs, however, show similar allele frequencies between the two populations.</p><p>Some differences in LD do exist between the two ethnic groups as well; however, the majority of the differences are small. The largest differences in LD are in pairwise values with rs16859788, which appear to mostly be due to the fact that the SNP is practically monoallelic in the Caucasian population. The Caucasian dataset suggests that there is a significant association of SNP haplotypes with autism risk, while the African-American set does not. However, this difference may be due to the small size of the African-American dataset. Therefore, while it appears that minor allele frequency differences can account for the lack of association of rs16859788 in the Caucasian dataset, additional studies are needed to determine whether the other differences in results between the two ethnic groups are due to sample size differences, differential LD with the causative variation in the two populations, or are population-specific risk alleles.</p><p>While we have identified several associated SNPs, we do not predict any of the ones in <italic>GABRA4</italic> to have functional consequences; therefore, it is unlikely that these are primary variants leading to the autism susceptibility. One of the SNPs identified in <italic>GABRB1</italic> in the seizure subset, however, is in the 3′ untranslated region (UTR). Given that multiple GABA receptor subunits combine in varying combinations to form a functional GABA receptor, even minor changes in levels of a particular subunit may alter the makeup of receptors within a particular cell type, and alter the GABAergic signaling. Therefore, variations within potential regulatory regions, such as untranslated regions and promoters, could play an important role. It will be important to look at potential changes that may result from this and other potential <italic>GABRB1</italic> UTR variations, as well as sequence coding and potential regulatory regions to identify the primary variation, or variations leading to altered autism susceptibility.</p><p>In summary, these data show that the GABA receptors are implicated in the etiology of autism in two different ethnic populations and suggest seizures as a stratifying phenotype. Furthermore, these results support our earlier findings, indicating <italic>GABRA4</italic> and <italic>GABRB1</italic> as genes contributing to autism susceptibility, independently, in the case of GABRA4, and through complex interactions with each other.</p></sec>
|
Mahaim Fibre Tachycardia: Recognition and Management
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Sternick</surname><given-names>Eduardo Back</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p> Dr. Gallagher et al [<xref ref-type="bibr" rid="R1">1</xref>] wrote 22 years ago that "the role of Mahaim fibers in the genesis of cardiac arrhythmias in man has been controversial since they were first described " in the late 30's by Dr. Ivan Mahaim [<xref ref-type="bibr" rid="R2">2</xref>]. The very early reports were strictly anatomical studies [<xref ref-type="bibr" rid="R2">2</xref>-<xref ref-type="bibr" rid="R6">6</xref>]. This histopathologic quest did not end yet. Mahaim fibers were supposed to be accessory connections taking off from the His bundle and fascicles (FV-fasciculoventricular) to the right ventricle or from the atrioventricular node (NV-nodoventricular fibers) to the right ventricle. Anderson et al [<xref ref-type="bibr" rid="R7">7</xref>] proposed 2 varieties of NV fibers, one that arises from the transitional zone and the other which inserted from the deep, compact nodal portion of the AV junction. In his pioneering work HJJ Wellens paved the road for clinical electrophysiological investigation. He was the first to study a patient with accessory pathway with decremental properties and long conduction times assuming its relationship with the fibers described long ago by "Mahaim", as reported in his doctoral thesis [<xref ref-type="bibr" rid="R8">8</xref>] in 1971. The term nodofascicular (NF) was applied when the retrograde His bundle potential preceded the ventricular deflection, while nodoventricular pathway would be appropriate when the retrograde His bundle deflection followed the ventricular potential. It took some years to electrophysiologists realize the conceptual mismatch among the "Mahaim" physiology and structure described by Mahaim et al. An important observation was done in 1978 by Becker et al [<xref ref-type="bibr" rid="R5">5</xref>] who found an accessory node associated with a bundle of specialized fibers measuring 1 cm and coursing through the right ventricle, mimicking a second AV conduction system located on the lateral tricuspid annulus. However, that did not change the mainstream concept of NV fibers. During the early 80's many centers started to refer patients with drug refractory tachycardias to surgical treatment. According to the current concepts at that time targeting the A-V node would be the logic strategy for curative treatment of patients with NV/NF fibers. Some courageous electrophysiologists used a new technique consisting of high-energy catheter ablation of the A-V node to treat a patient with "Mahaim" fiber, which yielded complete AV block and persistent preexcitation [<xref ref-type="bibr" rid="R9">9</xref>]. The turning point came in 1988 at the University Hospital of Western Ontario, Canada, when Klein, Guiraudon et al [<xref ref-type="bibr" rid="R10">10</xref>] had decided to extensively freeze the A-V node and upper His bundle region of a 29 year old man and they soon realized that preexcitation did not go away. It became clear for them that his accessory pathway was not linked whatsoever to the A-V node. The next patient was luckier, and had kept intact his A-V node, while his "Mahaim fibers" were successfully severed after ice mapping produced a consistent zone of reversible block in the accessory pathway at the right lateral aspect of the tricuspid annulus. Klein's manuscript was received on August 24, 1987, and published the next year on JACC. Two months later (October 20, 1987) Circulation received a manuscript from Tchou P et al [<xref ref-type="bibr" rid="R11">11</xref>] entitled "Atriofascicular connection or a nodoventricular fiber? Electrophysiologic elucidation of the pathway and associated reentrant circuit". From a single case report we were taught how simple it is to make sure that such pathways arise from the atrium. In recent years catheter ablation techniques have shed more light on the subject. Discrete "Mahaim" potentials that are considered surrogates of pseudo-Mahaim tissue depolarization, are used as an effective target for ablation [<xref ref-type="bibr" rid="R12">12</xref>,<xref ref-type="bibr" rid="R13">13</xref>]. A number of pharmacologic [<xref ref-type="bibr" rid="R14">14</xref>] and histologic data [<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R6">6</xref>,<xref ref-type="bibr" rid="R15">15</xref>,<xref ref-type="bibr" rid="R16">16</xref>], electrophysiologic maneuvers and observations during radiofrequency catheter ablation like heat induced "Mahaim" automaticity [<xref ref-type="bibr" rid="R19">19</xref>,<xref ref-type="bibr" rid="R20">20</xref>] are regarded as evidences of either an ectopic A-V node or remnants of the specialized A-V ring tissue. The NV/NF fibers are now considered a rare item but there are some convincing reports [<xref ref-type="bibr" rid="R21">21</xref>] of narrow and regular QRS tachycardias with ventriculoatrial dissociation. The last variety which is known as fasciculoventricular pathway [<xref ref-type="bibr" rid="R22">22</xref>] seems to play no role in clinical tachycardias but as long as it is very often associated with bypass tracts they should be correctly recognized and not targeted for ablation, avoiding unnecessary damage to the A-V node-His bundle conduction system.</p></sec><sec id="s2"><title>Recognition</title><sec id="s2a"><title>Electrocardiographic Features</title><p>Baseline electrocardiogram of patients with atriofascicular or atrioventricular pathways (pseudo-Mahaim) (<xref ref-type="fig" rid="F1">Figure 1</xref>) are characterized by minimal or no preexcitation. Sometimes the only clue is absence of septal Q wave in leads V5 or V6 [<xref ref-type="bibr" rid="R23">23</xref>]. Some patients show a typical LBBB with normal PR interval. A preexcited ECG is more likely to occur in an atrioventricular decremental pathway [<xref ref-type="bibr" rid="R24">24</xref>]. Precordial transition (R/S >1) usually occurs at V4 or V5 (sometimes V6). Latent preexcitation has recently been reported, in patients with spontaneous LBBB-like antidromic tachycardia, without preexcitation at rest and during atrial pacing [<xref ref-type="bibr" rid="R25">25</xref>]. A high degree of day-to-day variability as far as preexcitation is concern, the "concertina" effect is observed in many patients. Anterograde conduction over atriofascicular fibers yields a typical LBBB pattern with variable axis, superior frontal plane axis being the most commom one (ranging from -25° to -60°), but it is of no help in differentiating it from the atrioventricular pathways. QRS complex is usually larger with anterograde conduction over an atrioventricular pathway, with a slurred QRS onset [<xref ref-type="bibr" rid="R26">26</xref>] due to distal muscular insertion, which can be better appreciated in the r wave of V2 to V4 (>40 msec in atrioventricular pathways) [<xref ref-type="bibr" rid="R23">23</xref>].</p><p>Electrocardiogram of fasciculoventricular pathway is characterized by normal frontal plane axis like an anteroseptal accessory pathway (0° to +75°) [<xref ref-type="bibr" rid="R27">27</xref>] with a subtle preexcitation and normal PR interval. It is also commom to see a short PR interval due to associated enhanced A-V nodal conduction. Atrial pacing do not change the degree of preexcitation. Junctional beats are preexcited and intravenous adenosine yields blocked P waves. Precordial transition (R/S >1) usually occurs at V2 (<xref ref-type="fig" rid="F2">Figure 2</xref>).</p></sec><sec id="s2b"><title>Electrophysiologic Characteristics</title><p>The major findings are the slow conducting and decremental properties that can be assessed during right atrial pacing: AH interval lengthens, HV interval shortens and QRS widens until a steady value is achieved. Faster atrial stimulation does not increase preexcitation, but can prolong AV conduction time until block occurs (<xref ref-type="fig" rid="F3">Figure 3</xref>). Decremental conduction is usually defined as rate dependent prolongation of conduction time in more than 30 msec through the accessory pathway (AP) as measured in the electrograms close to the AP insertion. Atrial extrastimulus testing produce likewise results increasing progressively AV interval and preexcitation up to a steady value. During atrial pacing, achievement of maximal preexcitation is associated with retrograde conduction over the right bundle, His bundle and A-V node and cessation of pacing is usually followed by antidromic tachycardia (<xref ref-type="fig" rid="F4">Figure 4</xref>). The next step is to assess the role of the accessory pathway in the tachycardia circuit: active or bystander. It can be done delivering single late lateral right atrial extrastimuli during preexcited tachycardia, timed not to affect the His region and coronary sinus atrial electrogram at the ostium [<xref ref-type="bibr" rid="R11">11</xref>]. Advancement of QRS complex and atrial activation establishes the diagnosis of an extranodal accessory pathway (proximal atrial insertion) as well its involvement in the tachycardia circuit. If advancement of QRS activation occurs without changing of atrial activation, the presence of an extranodal AP is certain but its participation on the circuit is not. McClelland et al could successfully advance QRS activation with late right atrial extrastimuli in 22 of 23 patients with atriofascicular tachycardia [<xref ref-type="bibr" rid="R13">13</xref>]. Another less elegant maneuver proving AP participation in SVT is by producing catheter-induced RBBB. Assuming an antidromic tachycardia incorporating an atriofascicular (or atrioventricular) pathway retrograde conduction occurs via right bundle-His bundle and A-V node axis. RBBB lengthens the circuit path, tachycardia cycle length due to an increase in ventriculoatrial time. Preexcited A-V nodal reentry as well as antidromic tachycardia with retrograde conduction through another AP would not be affected. The proximal atrial insertion can also be localized with the recording of an accessory pathway potential, because it is usually recorded in the postero-lateral or antero-lateral aspect of tricuspid annulus, away from the A-V node. In the study of Grogin et al [<xref ref-type="bibr" rid="R28">28</xref>] clues as to the presence of a nodoventricular fiber were the inability of a premature atrial stimulus to advance the ventricle and the presence of dual A-V nodal pathways. Scheinman et al [<xref ref-type="bibr" rid="R29">29</xref>] were not agreeable with that statement and finding an "M" potential assumes diagnostic importance in this setting, because the "M" potential accurately localizes the anatomic site of the pathway.</p><p>Ventricular stimulation usually discloses ventriculoatrial conduction from the A-V node. The vast majority of atriofascicular and atrioventricular pathways have only anterograde conduction [<xref ref-type="bibr" rid="R30">30</xref>,<xref ref-type="bibr" rid="R31">31</xref>]. Adenosine injection during sinus rhythm can yield complete AV block or sometimes increase preexcitation [<xref ref-type="bibr" rid="R32">32</xref>]. Verapamil has a more proeminent effect over the A-V node, and can be of help in exposing preexcitation. During antidromic tachycardia adenosine causes prolongation of conduction over the pathway and eventual block, terminating tachycardia.</p><p>There are plenty of electrophysiologic and anatomic data supporting the concept that at least atriofascicular pathways are accessory AV nodes. We have seen a patient with an electrophysiologic profile suggestive of the presence of an atriofascicular pathway without conduction through the AV node. This patient had unexplained syncope with a baseline ECG showing LBBB-like pattern with normal PR interval (<xref ref-type="fig" rid="F5a">Figure 5a</xref>). Atrial pacing disclosed a decremental AP and 1:1 conduction up to 280 msec with an "M" potential in right posteroseptal region. There was no VA conduction. Adenosine yielded transient complete AV block. We decided to ablate the Mahaim-like pathway and as soon as radiofrequency current was delivered, the patient had developed complete AV block (<xref ref-type="fig" rid="F5b">Figure 5b</xref>). Ablation was immediately discontinued and patient resumed preexcitation. The transient escape rhythm after ablation had a normal HV interval but no conduction occurred over the AV node. This scenario is consistent with an ectopic accessory AV node without conduction from the "normal" AV node.</p><p>Associated conditions: Mahaim fibers (AF/ AV) very often occurs in the setting of Ebstein's disease (10 to 40%) and associated accessory pathways are a commom finding (up to 30%). Fasciculoventricular pathways also occurs very often in association with accessory pathways. I have reviewed the cases reported since 1981 [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R22">22</xref>,<xref ref-type="bibr" rid="R23">23</xref>,<xref ref-type="bibr" rid="R33">33</xref>,<xref ref-type="bibr" rid="R34">34</xref>] together with 3 cases from our own laboratory, and I found 6 of 15 patients with associated AP's (40%). Dual AV nodal pathways with AV nodal reentrant tachycardia is much higher of that expected by chance alone (<xref ref-type="table" rid="T1">Table 1</xref>).</p></sec></sec><sec id="s3"><title>Treatment</title><sec id="s3a"><title>Mapping and Catheter Ablation</title><p>Some particular features are unique to Mahaim fibers: Mapping of the atrial insertion by ventricular stimulation is usually not possible because those decremental pathways do not conduct retrogradely. Atrioventricular connections can be located by mapping the site of earliest ventricular activation on the annulus, as with other anterogradely conducting accessory AV pathways. On the contrary, atriofascicular pathways or even the long atrioventricular pathways with distal (nonannular) insertion cannot be mapped in this way. To worse matters these decremental pathways are unusually sensitive to mechanical trauma. Inadvertent knocking of the ablation catheter against the annulus can result in transient abolition of conduction through the pathway from minutes to hours [<xref ref-type="bibr" rid="R34">34</xref>,<xref ref-type="bibr" rid="R35">35</xref>]. The following strategies have been used to overcome those problems:
<list list-type="order"><list-item><p>Searching for the "M" (Mahaim) potential (<xref ref-type="fig" rid="F6">Figure 6</xref>) along the tricuspid annulus is the most commonly used technique. The ablation catheter should be carefully moved along the annulus avoiding bumps on the tissue. We routinely use a long sheath like DAIG<sup>®</sup> SR2 or SR3, which improves stability. The potential may be as large as the His bundle potential or small, narrow with low amplitude. Catheter ablation at a site with "M" potential is likely to be successful (<xref ref-type="table" rid="T1">Table 1</xref>). We [<xref ref-type="bibr" rid="R20">20</xref>] and other authors [<xref ref-type="bibr" rid="R19">19</xref>,<xref ref-type="bibr" rid="R29">29</xref>] have recorded automatic rhythms ("Mahaim" automatic tachycardia-MAT) brought about during radiofrequency current delivery. It is probably due to heat-related automaticity of nodal-like tissue in a similar fashion to junctional rhythm that arises during slow A-V nodal pathway ablation. It seems to represent a hallmark for successful ablation particularly of atriofascicular pathways. MAT in most cases is short-lived (<xref ref-type="fig" rid="F7">Figure 7</xref>), but ocasionally it lasts longer. We have had an early out of hospital recurrence when it was not our policy to completely eliminate such rhythm. In a second procedure we decided to ablated until complete elimination of automatic activity (<xref ref-type="fig" rid="F8">Figure 8</xref>). We have seen MAT during radiofrequency catheter ablation of atriofascicular pathway but not with atrioventricular pathways.</p></list-item><list-item><p>Activation mapping of the earliest local ventricular potential is feasible in short atrioventricular pathways like in fast conducting AV accessory pathways. Atrioventricular decremental pathways with a long course often shows extensive arborization over a wide area of ventricular muscle [<xref ref-type="bibr" rid="R23">23</xref>]. Targeting distal branches is a time consuming task. It is possible to ablate some of them, as assessed by changes in preexcitation pattern, but a complete elimination is very unlikely. Some patients with atriofascicular pathway who underwent ablation at the distal insertion had developed a proarrhythmic [<xref ref-type="bibr" rid="R13">13</xref>,<xref ref-type="bibr" rid="R35">35</xref>] response with facilitation of antidromic tachycardia occurrence due to slow conduction induced by radiofrequency ablation.</p></list-item><list-item><p>Shortest stimulus-QRS interval as assessed by atrial stimulation at a constant pacing rate along the atrial aspect of the annulus was the gold standard mapping method before mapping of "M" potential had been reported. Stimulation sites remote from the atrial insertion of the accessory pathway result in long stimulus-QRS interval due to the amount of interposed atrial tissue. We do not use this method because it is time consuming and very inaccurate because it is difficult to stimulate from many sites at the same distance from the annulus and stimulating atrial tissue requires good contact with the tip, which is not always possible.</p></list-item><list-item><p>Extrastimulus mapping during antidromic tachycardia. Finding an atrial site where the longest coupled premature extrastimulus causes resetting, or assessing the amount of advancement of the QRS following application of a fixed atrial extraestimulus coupling interval. Similar to the previous technique it looks for a site in the atrial annulus with the least interposing tissue separating it from the accessory pathway proximal insertion. Likewise shortest stimulus-QRS technique is an inaccurate and tedious method.</p></list-item><list-item><p>Some authors [<xref ref-type="bibr" rid="R36">36</xref>,<xref ref-type="bibr" rid="R37">37</xref>] reported the judicious use of mechanical trauma in a controlled way and transient AP conduction block to find the AP insertion. The rationale of this method is based on the observation that transient conduction block following bumps of the catheter on the atrial aspect of the annulus is a frequent phenomena in decremental anterograde pathways. Gentle pressure of the tip of a steerable mapping catheter during antidromic tachycardia or atrial pacing would lead to transient block of AP conduction locating the site for ablation. The fact that those AP are more prone to mechanical trauma suggests that they may be thinner and have a more superficial location than others. The main pitfalls are: when conduction block occurs the catheter is on the move and may be away from the AP site, and its relocation will be dependent on resuming of AP conduction; mechanical block can last hours and a second procedure would be needed. I do not favor the use of this method and I'd rather avoid mechanical trauma.</p></list-item><list-item><p>Electroanatomic mapping (noncontact mapping) [<xref ref-type="bibr" rid="R34">34</xref>,<xref ref-type="bibr" rid="R38">38</xref>] can be helpful, while not widely available, in those cases where accessory pathway potential cannot be found and when mechanical trauma precludes adequate mapping. This technique allows the operator to "tag" the exact location of the tip of the catheter and in case of transient conduction block the catheter can be manipulated back to the tagged location for ablation</p></list-item></list></p><p>Radiofrequency current should be applied during atrial pacing to enhance preexcitation and making it easier to assess conduction block at the AP. Stability is improved during atrial pacing as compared with ablation during antidromic tachycardia when catheter is likely to move with tachycardia termination. MAT is a common and expected event and can also cause catheter displacement. Before "M" potential mapping technique became the gold standard mapping technique some authors favored targeting the ventricular insertion to avoid MAT and maintain a better catheter stability [<xref ref-type="bibr" rid="R39">39</xref>].</p><p>Catheter ablation have been very successful particularly when ablating at a site with "M" potential or assessing earliest delta-V interval in atrioventricular decremental pathways (<xref ref-type="table" rid="T1">Table 1</xref>).</p></sec></sec>
|
Suppression of Paroxysmal Atrial Fibrillation by Pacing
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Gupta</surname><given-names>Anoop K</given-names></name><degrees>MD, DM, DNB, FACC</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<p>Atrial Fibrillation (AF) affects approximately five million people world- wide. The incidence of AF increases with aging and more common in males [<xref ref-type="bibr" rid="R1">1</xref>]. Atrial fibrillation is the most common cardiogenic cause of stroke and exacerbates heart failure. Despite the prevalence of AF, it is still one of the most difficult arrhythmia to treat [<xref ref-type="bibr" rid="R2">2</xref>]. The management option of AF ranges from pharmacological therapy, catheter based ablation and surgery. However, non-traditional uses of pacemaker therapy have been reported in recent past.</p><p>Pacing therapies for AF are specific to the nature of the patient's AF [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R4">4</xref>]. Pharmacological management alone is the typical initial course of action. In AF cases where rapid ventricular rates cannot be controlled by pharmacological therapy, ventricular pacing therapy may be combined with radiofrequency catheter ablation of AV node [<xref ref-type="bibr" rid="R5">5</xref>].</p><p>An atrial based pacing mode compared with ventricular pacing, results in better clinical and hemodynamic conditions for patients with ventricular dysfunction [<xref ref-type="bibr" rid="R6">6</xref>]. Rosenqvist M et al [<xref ref-type="bibr" rid="R7">7</xref>] demonstrated that ventricular based pacing is associated with abnormal activation (e.g., pre-excitation) patterns that adversely affect cardiac hemodynamics. Specially, cardiac output and peak ventricular filling rates were found to be significantly lower in ventricular paced groups as compared to atrial based pacing groups. As well, lower peak oxygen uptake levels and oxygen saturation levels were found in ventricular paced groups. Similarly, Vardas et al [<xref ref-type="bibr" rid="R8">8</xref>] compared DDD and AAI pacing modes and concluded that atrial based pacing, at rest, produces a significantly greater cardiac output as compared to the DDD mode. However, no differences were found between the pacing modes during increased heart rates.</p><p>Atrial overdrive pacing therapy for patients with paroxysmal AF is done with the intent of altering or controlling AF [<xref ref-type="bibr" rid="R9">9</xref>]. Initiation of AF can be affected by suppressing the atrial premature beats that typically triggers or by modifying their propagation. The mean heart rate just before the onset of an AF episode is slightly higher than at rest and is very difficult to observe in most instances. This makes AF difficult to predict and monitor using pacing technology. Consequently, continuous overdrive atrial pacing is proposed to be a good course of treatment. The theoretical basis for overdrive pacing is the idea to continuously pace the atrium at higher rate than the patient sinus rate, so that it could alter the intrinsic atrial rate, propagation and suppress automaticity, caused by electrical remodeling, in the diseased fibers [<xref ref-type="bibr" rid="R10">10</xref>].</p><p>The case report by Levine et al [<xref ref-type="bibr" rid="R11">11</xref>] in this issue has discussed the utility of this therapy in a symptomatic patient with additional sinus node dysfunction, who was intolerant to the pharmacological therapy. The patient was marked symptomatic with paroxysmal AF and medical therapy, which got remarkably subsided with DDD pacing and withdrawal of medicine, however, the episodes of paroxysmal AF reduced after atrial fibrillation suppression algorithm.</p><p>The AF suppression algorithm is a newly approved algorithm designed to provide a high percentage of atrial overdrive pacing, which control the atrial rate, either from the sinus node or an ectopic foci. The objective is reduced temporal dispersion of the atrial refractory period combined with overdrive suppression of ectopic beats, a common trigger for atrial tachyarrhythmias. Automatic mode switch (AMS) remains functional with the AF Suppression algorithm on, thus if AF occurs, AMS is activated facilitating management as this provides information as to the frequency and duration of each AMS episode.</p><p>While the algorithm may not be 100% effective in all patients, any reduction in the number of AF episodes is beneficial. The use of a pacemaker containing the AF suppression algorithm should be considered in all patients undergoing device implant for a symptomatic bradycardia, particularly when there is a prior history of atrial fibrillation or they are at risk of AF which is more likely when the indication for pacing is sinus node dysfunction.</p>
|
Preventing Sudden Death And The Use Of Prophylactic Implanted Defibrillators
|
<p> Implanted defibrillators have become mainstream therapy for the prevention of sudden cardiac death from ventricular tachyarrhythmias. A decade of studies has confirmed the superiority of ICDs over antiarrhythmic drug therapy in prolonging the life of patients with a prior history of sustained VT or VF.</p><p>More recent studies have compared ICD therapy to drugs or no antiarrhythmic therapy as 'primary prophylaxis' in patients considered at high risk for sudden death or with prior MIs. In selected patients, ICDs lead to important relative and absolute reductions in mortality in patients with no prior history of sustained VT or VF. Clinicians need to carefully consider these studies in their management of patients with CAD and severe LV dysfunction.</p>
|
<contrib contrib-type="author"><name><surname>Ray</surname><given-names>Indranill Basu</given-names></name><degrees>MBBS (Hons), MD, DNB (Card)</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>The understanding that cardiac death and, in particular, sudden cardiac death from fatal ventricular arrhythmias, is one of the most common causes of death in Western society is now widespread. Following the spectacular successes of thrombolytic, antiischemic, and revascularization therapies in the 1990's, focus has increasingly turned to the care of patients with the chronic consequences of coronary artery disease, chiefly left ventricular dysfunction and heart failure, and the propensity to sudden arrhythmic death. India being at present a nation facing an epidemic of coronary artery disease; would consequently have more & more of its populace in chronic CAD with risk of SCD.</p><p>Although it has been very clearly established that patients with left ventricular dysfunction, with or without symptomatic heart failure, are at high risk for out-of-hospital cardiac arrest, and undocumented but presumably arrhythmic sudden death, preventing such deaths has posed a major therapeutic challenge. First, it is difficult if not impossible to predict, with any reasonable degree of certainty, which particular patients are destined to suffer fatal arrhythmias, and which others are destined to remain clinically stable, or at least be free of serious ventricular arrhythmias. If one could identify such patients, then therapy could, of course, be targeted to only that select proportion whose destiny it is to suffer VT or VF (this represents approximately 40-60% of all patients with moderate to severe left ventricular dysfunction) [<xref ref-type="bibr" rid="R1">1</xref>]. Tests to identify patients at particularly high risk of sudden death have included the ability to induce ventricular tachycardia or fibrillation at invasive electrophysiologic study; the documentation of nonsustained ventricular tachycardia on holter or in-hospital ECG monitoring; the presence of subtle (not visible to the naked eye) ECG abnormalities of depolarization and repolarization using the filtered, signal-averaged ECG, or the presence of microvolt T-wave alternans; and the presence of abnormal autonomic modulation of cardiac function, by the registration of abnormally low heart rate variability (HRV), or depressed baroreceptor sensitivity.</p><p>Although each of these tests is of some prognostic value, they are insufficiently accurate, for practical clinical purposes, to direct therapy. Even amongst patients with poor ejection fraction, past history of myocardial infarction & the presence of ventricular scar are actually believed to be at high risk of sudden cardiac death. However there is no available diagnostic modality by which one can identify the exact population of patient liable to have SCD amongst thousands with these characteristics. This is tragic but true that this is in fact a serious limitation in our capability to triage patients requiring protection from SCD, as the incidence of sudden death amongst all patients with prior myocardial infarction is relatively low.</p><p>A second conceptual and practical problem is our inability to identify the timing, or proximate causes of sudden death from VT/VF. Such events appear to occur "out of the blue", and there are no clearly identifiable factors, which precede sudden cardiac death in most individuals. Although coronary artery disease is the most important etiologic factor leading to life threatening ventricular arrhythmias, angina, other manifestations of myocardial ischemia, sudden worsening of heart failure, or behavioural factors such as stress or exercise are rarely observed to immediately precede sudden death.</p><p>Improved acute and long term therapies have increased survival for patients with myocardial infarction, leading to a relative increase in the number of the patients with chronic coronary disease and left ventricular dysfunction, who are nevertheless stable and not expected to suffer imminent recurrent infarction or progressive heart failure. Such patients usually feel relatively well, and may require some persuasion to consider prophylactic therapy for cardiac arrhythmias, which from a subjective standpoint, can only decrease their quality of life in the short term. In confronting these dilemmas, clinicians through the 1980's and 1990's were optimistic that sudden death could be prevented by the administration of antiarrhythmic drug therapy. This approach had the conceptual benefit of being able to be delivered to a large group of patients at relatively low risk, as a "chemoprophylaxis" of sudden cardiac death. With the spectacular failure of class I drugs, for example with Flecainide following myocardial infarction, attention has turned to drugs that prolong cardiac repolarization (class III drugs). The most extensively studied of these drugs is amiodarone. Several very large trials have examined in detail the potential usefulness of amiodarone in preventing sudden death in high risk patients with coronary artery disease and left ventricular dysfunction, the largest being the EMIAT study (European Myocardial Infarction Arrhythmia Trial) [<xref ref-type="bibr" rid="R2">2</xref>], and the CHF-STAT study (Congestive Heart Failure - Survival Trial of Antiarrhythmic Therapy) [<xref ref-type="bibr" rid="R3">3</xref>]. The CAMIAT study (Canadian Amiodarone Myocardial Infarction Arrhythmia Trial) [<xref ref-type="bibr" rid="R4">4</xref>] also included patients with prior myocardial infarction, most of whom had at least moderate left ventricular dysfunction, as well as frequent ventricular premature beats. All of these studies were randomized placebo-controlled clinical trials. None were able to show neither a statistically significant, nor a clinically meaningful reduction in all cause cardiac mortality. Although meta analyses of amiodarone have suggested a small, and statistically significant reduction in all cause mortality in high risk populations, individual trials in patients with coronary artery disease and left ventricular dysfunction leaves us scant hope that amiodarone will be highly useful in this population of patients. The progressively increasing burden of adverse effects from amiodarone is another barrier to its use. Given the paucity of evidence, there is no good reason to prescribe amiodarone as primary prophylaxis for ventricular tachycardia or fibrillation in patients with coronary artery disease and left ventricular dysfunction, but no symptoms of sustained ventricular arrhythmias.</p><p>Large studies have also examined the potential benefit from new class III antiarrhythmic drugs in the prevention of sudden death following MI or with heart failure, including studies of dofetilide (DIAMOND [<xref ref-type="bibr" rid="R5">5</xref>] and DIAMOND-CHF [<xref ref-type="bibr" rid="R6">6</xref>]), and azimilide (the ALIVE study [<xref ref-type="bibr" rid="R7">7</xref>]). These also failed to show any difference between drug and placebo treated patients in sudden or all cause mortality. It is extremely important to note that beta-blocker therapy is of undoubted benefit in prolonging life in patients following myocardial infarction, particularly those with heart failure or extensive left ventricular dysfunction. Beyond the universal requirement for beta blockers unless absolutely contraindicated, there is however not much room for optimism that antiarrhythmic drugs, at least for the time being, will be even a partial solution to the problem of sudden cardiac death in susceptible coronary populations.</p></sec><sec id="s2"><title>The Implanted Defibrillator</title><p>The implanted defibrillator represents an effective, if intellectually inelegant therapy to prevent death from ventricular arrhythmias. The device, after all, does not prevent such arrhythmias but only treats them after they occur. Shocks are painful and unpleasant, the devices are expensive, a surgical procedure is required for its implantation, and the follow-up of patients can be technically challenging. Where do we stand with respect to the evidence concerning implanted defibrillators and sudden death?</p><p>There is extensive information regarding the <italic>efficacy</italic> of implanted defibrillator therapy. Appropriately tested devices have a 99% or greater probability of successfully restoring a perfusing rhythm in patients with ventricular tachycardia or ventricular fibrillation. Current devices can be implanted with a less than 1% major morbidity or mortality, with a surgical complexity and morbidity very similar to that of pacemaker implantation.</p><p>Studies in patients with a prior history of cardiac arrest, or sustained ventricular tachycardia; what is called the secondary prevention, have demonstrated convincingly that the implanted defibrillator is both effective, and superior to antiarrhythmic drug therapy in preventing all cause mortality in such patients. The AVID [<xref ref-type="bibr" rid="R8">8</xref>], CIDS [<xref ref-type="bibr" rid="R9">9</xref>], and CASH [<xref ref-type="bibr" rid="R10">10</xref>] studies, and their meta-analysis [<xref ref-type="bibr" rid="R11">11</xref>], have shown an approximately 20-30% reduction in all cause mortality in such patients. The greatest relative benefit from defibrillators over antiarrhythmic therapy (primarily amiodarone) occurs in those with the worst left ventricular function, and the elderly [<xref ref-type="bibr" rid="R12">12</xref>]. Since the majority of sudden deaths occur in patients without a prior history of documented sustained ventricular tachycardia or ventricular fibrillation, studies have assessed the usefulness of defibrillators as "primary prophylaxis" of sudden cardiac death. The accumulated evidence from these studies is briefly reviewed below.</p></sec><sec id="s3"><title>Clinical Trials</title><p>Initial trials focused on the selection of patients expected to be at particularly high risk of sudden cardiac death, based on a combination of low ejection fraction, and an additional risk marker for sudden cardiac death.</p><p>The MADIT I study assessed patients with coronary artery disease, poor left ventricular function, and asymptomatic nonsustained ventricular tachycardia, with inducible VT or VF at electrophysiologic study, not suppressible by antiarrhythmic drug therapy [<xref ref-type="bibr" rid="R13">13</xref>]. This study, the first to document a potential benefit from prophylactic ICDs, showed a 54% reduction in mortality in patients implanted with a defibrillator as opposed to those receiving "conventional medical therapy". The weaknesses of this trial included its relatively small size, inadequate therapy with beta blockers and ACE inhibitors, and the clinically impractical sequence of EP study and need for VT induction, followed by attempted VT/VF suppression with procainamide, that was required for risk stratification. Nevertheless, the results from this study led to FDA approval of implanted defibrillators for the particular subset of patients meeting the inclusion criteria for this study.</p><p>The CABG PATCH study randomized patients immediately following successful aortocoronary bypass surgery, if they met the inclusion criteria of a low ejection fraction (<35%), and a positive signal-averaged ECG, to either an implanted defibrillator or control therapy without the ICD [<xref ref-type="bibr" rid="R14">14</xref>]. All devices were attached to the heart by means of epicardial defibrillator patches (which are no longer used during the CABG procedure.</p><p>This study failed to show any benefit whatsoever from the implanted defibrillator, but both defibrillator and no defibrillator patients had a low cardiac mortality (5.9% per year), suggesting that surgical revascularization has a very important protective effect against sudden death.</p><p>The MUSTT study, like the MADIT study, also selected patients with coronary artery disease and ejection fraction of <40% if they had asymptomatic nonsustained VT on holter or in-hospital ECG monitoring, as well as inducible VT at EP study [<xref ref-type="bibr" rid="R15">15</xref>]. They were randomized to either "electrophysiologically guided", or no antiarrhythmic drug therapy. The electrophysiologically guided arm could include antiarrhythmic drugs designed to suppress the inducibility of ventricular tachycardia, those that would render inducible arrhythmias hemodynamically stable, or an implanted defibrillator. The choice between defibrillator versus drug therapy was not randomized.</p><p>Freedom at 5 years from sudden death was significantly lower in the electrophysiologically guided arm (25 vs. 32%, p=0.04) but all cause mortality was not (48 vs. 42%, p=0.06). However, a secondary analysis comparing patients with no antiarrhythmic therapy, the implanted defibrillator, and antiarrhythmic drug therapy, showed some striking trends. The relative risk of death from all causes in the ICD group compared to the no antiarrhythmic therapy group was 0.45 (95%, CI 0.32-0.63) and compared to electrophysiologically guided antiarrhythmic drug therapy was 0.40 (0.27-0.59). Although this is not strictly a randomized therapy assignment outcome, the study was widely and reasonably interpreted as showing superiority of the implanted defibrillator to no antiarrhythmic therapy or antiarrhythmic drug therapy. The observation that the "electrophysiologically guided" strategy was increasingly superior over no antiarrhythmic therapy as ICDs were increasingly frequently used over time, and relatively better in those centers that used ICDs more frequently, lent plausibility to the belief that it was the defibrillator which contributed all of the observed benefit of the antiarrhythmic therapy arm.</p><p>As a consequence of the MUSTT study, most expert bodies, stipulating guidelines for the treatment of ventricular arrhythmias, concluded that patients with coronary artery disease, ejection fraction <40%, and nonsustained VT, if they had inducible ventricular tachycardia at EP study, should preferably be treated with an implanted defibrillator [<xref ref-type="bibr" rid="R16">16</xref>].</p><p>Up until very recently, the database above was sufficiently ambiguous and related to a sufficiently select subgroup of patients (those with all of CAD, low ejection fraction, nonsustained VT, and inducible VT/VF at EP studies), that these recommendations have not been widely adopted in everyday clinical practice.</p><p>The MADIT II study, published in March 2002 [<xref ref-type="bibr" rid="R17">17</xref>], took a simplified approach to the testing of the hypothesis that implanted defibrillators would reduce all cause mortality in at risk populations. The only criteria to identify patients at risk from sudden death were the presence of coronary artery disease, a prior myocardial infarct, and an ejection fraction of <30%. This study randomized a total of 1232 patients to either the ICD (742 patients), or conventional medical therapy (490 patients, a 3:2 ratio). Neither nonsustained VT nor an electrophysiologic study was required for entry into this study.</p><p>The patient population in this study was reasonably representative of a potentially very large group of patients with chronic coronary artery disease and prior MI. The mean age was 65 years, and 70% of patients were either NYHA class II or I. A majority had a remote history of coronary bypass surgery (57%), or coronary angioplasty (44%). In the vast majority, more than 6 months had elapsed since their most recent MI. Interestingly enough the associated drug therapy that most patients in the trial had was sufficiently appropriate as to allow generalizability in this trial. Seventy percent were receiving ACE inhibitors, 70% beta-blockers, and 57% digitalis. Sixty-six percent received statins. About 12% were receiving amiodarone at last contact (presumably most often for atrial fibrillation), and only 9% received calcium channel blockers and 3% received class I antiarrhythmic drugs.</p><p>Patients were followed to a common primary endpoint of death from any cause. The pre-specified mortality efficacy boundary was achieved just over 4 years after the study began, after an average follow-up of 20 months.</p><p>The defibrillator therapy resulted in an increasing mortality benefit over conventional therapy over time, with an aggregate 31% reduction in the risk of death at any time interval, including a relative decrease of mortality of 12%, 28%, and 28% at 1, 2, and 3 years respectively. In absolute terms, this meant a 1%, 6%, and 9% reduction in mortality at 1, 2, and 3 years; in other words, the number needed to treat (NNT) to prevent 1 death by 3 years was approximately 11. This NNT compares very favorably to other cardiovascular therapies in common use, for example beta-blockers (CIBIS 2, NNT = 23), statins (4S, NNT = 28), or ACE inhibitors (SAVE, NNT = 20). There was a slightly higher probability of hospitalization for heart failure in the ICD group (11 per 1000 months), versus the control group (9 per 1000 months, p=0.09).</p><p>Subsequent further subgroup analysis showed that patients with QRS prolongation of >120 msec at baseline received a particularly and dramatically large benefit from the implantation of an ICD, the mortality reduction being from 53% to 21% (a 63% reduction) at 3 years in these patients. This latter observation is consistent with prior demonstration of QRS prolongation on the surface ECG as being particularly potent, simple marker for the probability of all cause mortality and sudden death.</p></sec><sec id="s4"><title>Present Status of Prophylactic ICDs</title><p>The evidence indicating that implanted defibrillators prolong life in patients who are susceptible to sudden cardiac death is compelling. It is important to underline that the trials pertain exclusively to patients with coronary disease (as opposed to those with dilated or other forms of cardiomyopathy), and probably are not applicable to patients immediately after bypass surgery. No study has shown superiority of ICDs over medical therapy in patients with dilated cardiomyopathy, and a large study of ICDs vs. amiodarone (SCD-HeFT) will address this question. With these exceptions, such patients with very poor ventricular function unquestionably benefit from the implantation of a defibrillator, even if they are receiving optimal medical therapy. Although the MADIT and MUSTT trials did not systematically compare the ICD to "best" medical therapy (almost certainly amiodarone), the absence of clear proof that amiodarone is effective, and the toxicity burden from amiodarone (which itself increases progressively over time), suggests that for the moment the defibrillator should be considered clearly superior to amiodarone therapy or no antiarrhythmic therapy in the prevention of sudden and all cause mortality in susceptible populations. Importantly, defibrillators in the MADIT study and other studies were implanted with virtually no perioperative mortality, and a 2.5% incidence of non-fatal adverse events requiring surgical interventions (lead problems or infection) [<xref ref-type="bibr" rid="R17">17</xref>]. The main barrier to more widespread use of prophylactic implanted defibrillators, at least in the Indian context; though it applies as well to the west particularly in countries with government funded health care system, seems to be resource limitations, both with respect to device and implantation costs, and the availability of medical personnel to perform the procedures and follow the patients. In addition, the total number of years added to life, as well as the quality of these added years, is not fully elucidated given the relatively short follow-up time of all of the studies published thus far.</p></sec><sec id="s5"><title>Who is The Appropriate Candidate?</title><p>For the time being, it seems appropriate to at least <italic>consider</italic> a prophylactic implanted defibrillator in all patients with a history of remote myocardial infarction and ejection fraction of <30%, provided they are receiving or have been considered for evidence based pharmacological therapies including beta-blockers, ACE inhibitors, aspirin, statins, and spironolactone as indicated. If revascularization is indicated and feasible, it should be performed. The presence of nonsustained ventricular tachycardia on in-hospital or Holter monitoring probably adds some prognostic significance, although the amount of information contained in this finding is not clear. Performing an electrophysiologic study for risk stratification is probably not required for most such patients. The expectation of treatment benefit is amplified in patients with bundle branch block or QRS >120 msec.</p><p>Following these considerations, it is appropriate and should be considered advisable to at least inform the patient of the treatment options available, unless there are severe co-morbidities, which reduce the expectation of treatment benefit.</p></sec>
|
T Wave Alternans And Ventricular Tachyarrhythmia Risk Stratification: A Review
|
<p>Sudden cardiac death (SCD) is one of the leading causes of mortality in industrialized countries. Thus, identifying patients at high risk of SCD is an important goal. T wave alternans (TWA) is a new method for identifying patients with lethal ventricular tachyarrhythmias, and is dependent on heart rate. The maximal predictive accuracy is achieved at heart rates between 100 and 120 bpm, so that TWA is usually measured during exercise, phamacological stress, or atrial pacing. It has been shown that TWA has high sensitivity and negative predictive value for predicting SCD after myocardial infarction and is also useful for predicting SCD in patients with nonischemic cardiomyopathy. Although the implantable cardioverter defibrillator (ICD) is now the primary therapy for preventing SCD, it is difficult to identify those patients who are susceptible to lethal ventricular tachyarrhythmias for primary prevention. In the prediction of SCD, TWA can be used as a screening test of appropriate patients for further electrophysiological examination and therapy.</p>
|
<contrib contrib-type="author"><name><surname>Takagi</surname><given-names>Masahiko</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Yoshikawa</surname><given-names>Junichi</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<p>Sudden cardiac death (SCD) is a leading cause of mortality and remains a major clinical challenge [<xref ref-type="bibr" rid="R1">1</xref>]. Concerning therapeutic modalities, tremendous progress has been made in the development of the implantable cardioverter defibrillator (ICD). However, progress in identifying patients at high risk of SCD has lagged behind. Large multicenter trials have shown that electrophysiologic study (EPS) may be useful in identifying patients who would benefit from ICD therapy [<xref ref-type="bibr" rid="R2">2</xref>,<xref ref-type="bibr" rid="R3">3</xref>]. Unfortunately, EPS is costly, invasive, and imperfect [<xref ref-type="bibr" rid="R4">4</xref>]. Several noninvasive markers of risk-stratification have been studied and compared with EPS. Left ventricular ejection fraction (LVEF), frequent ventricular premature complexes (VPC) on Holter recording, and ventricular late potentials (LP) are all sensitive, but of low specificity and positive predictive value (PPV) [<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R6">6</xref>]. Measurement of heart rate variability, especially in combination with LVEF, VPC, and LP, has significantly improved risk prediction, but PPV remains low [<xref ref-type="bibr" rid="R6">6</xref>]. A screening procedure that is more sensitive and specific - with high PPV for identifying patients at high risk of developing ventricular tachyarrhythmias - is needed.</p><p>Recently, assessment of repolarization alternans (T wave alternans [TWA]) in the electrocardiogram (ECG) has been suggested as a predictor of susceptibility to lethal ventricular tachyarrhythmias [<xref ref-type="bibr" rid="R7">7</xref>,<xref ref-type="bibr" rid="R8">8</xref>]. Although overt TWA in the ECG is not common [<xref ref-type="bibr" rid="R9">9</xref>], digital signal processing techniques capable of detecting subtle degrees of TWA (microvolt TWA) have shown that TWA may represent an important marker of vulnerability to ventricular tachyarrhythmias. This review discusses the electrophysiologic mechanisms that link TWA to arrhythmogenicity and the recent clinical data for its prognostic efficacy in predicting lethal ventricular tachyarrhythmias.</p><sec id="s1"><title>History of TWA</title><p>TWA was first described in 1908; it is the variation in vector and amplitude of the T wave that occurs on an every-other-beat basis [<xref ref-type="bibr" rid="R10">10</xref>]. In a review by Kalter in 1948, 5 patients were identified with macroscopic TWA - a frequency of 0.08% [<xref ref-type="bibr" rid="R11">11</xref>]. In humans, macroscopic TWA has been associated with increased vulnerability to ventricular tachyarrhythmias under several pathophysiologic conditions such as myocardial ischemia [<xref ref-type="bibr" rid="R12">12</xref>-<xref ref-type="bibr" rid="R14">14</xref>], vasospastic angina [<xref ref-type="bibr" rid="R15">15</xref>,<xref ref-type="bibr" rid="R16">16</xref>], marked electrolyte abnormalities [<xref ref-type="bibr" rid="R17">17</xref>,<xref ref-type="bibr" rid="R18">18</xref>], hypertrophic cardiomyopathy (HCM) [<xref ref-type="bibr" rid="R19">19</xref>], the long QT syndrome [<xref ref-type="bibr" rid="R20">20</xref>-<xref ref-type="bibr" rid="R22">22</xref>], and the Brugada syndrome [<xref ref-type="bibr" rid="R23">23</xref>,<xref ref-type="bibr" rid="R24">24</xref>]. Microscopic TWA was first reported in 1982 [<xref ref-type="bibr" rid="R25">25</xref>]. Thereafter, many studies have led to the development of the method for detecting microvolt TWA and to the establishment of a relationship between TWA and susceptibility to ventricular tachyarrhythmias, in humans [<xref ref-type="bibr" rid="R26">26</xref>,<xref ref-type="bibr" rid="R27">27</xref>].</p></sec><sec id="s2"><title>Pathophysiology of TWA</title><sec id="s2a"><title>Ionic currents and TWA</title><p>There is some evidence that TWA is linked to alternations in cellular calcium homeostasis, which significantly influence the action potential duration (APD) 1982 [<xref ref-type="bibr" rid="R28">28</xref>]. In the failing heart, electrical remodeling is a recurring feature that has been associated with an increased risk of SCD 1982 [<xref ref-type="bibr" rid="R29">29</xref>]. The arrhythmogenesis is due to the functional expression of proteins to control calcium homeostasis.</p><p>Potassium channels may also play an important role in ischemia-induced TWA. The different sensitivity of KATP channel activation during ischemia between epicardium and endocardium may be linked to TWA at the cellular level 1982 [<xref ref-type="bibr" rid="R30">30</xref>-<xref ref-type="bibr" rid="R32">32</xref>].</p></sec><sec id="s2b"><title>Arrhythmogenesis and TWA</title><p>Currently, the hypothesis regarding the arrhythmogenic mechanisms associated with TWA is based on the concept that heterogeneous prolongation and increased dispersion of repolarization produce reentrant ventricular tachyarrhythmias 1982 [<xref ref-type="bibr" rid="R33">33</xref>]. The heterogeneity in dispersion of repolarization results in a 2:1 appearance on the surface ECG and provides conduction block in the areas with prolongation of repolarization, which fractionates the wavefront and facilitates reentry. Shimizu et al. 1982 [<xref ref-type="bibr" rid="R34">34</xref>] studied an experimental model of long QT syndrome utilizing an arterially perfused wedge of canine left ventricular wall. When the preparation was paced at a critical fast rate, there was pronounced alternation of APD of mid-myocardial (M) cells, resulting in a reversal of the transmural repolarization sequence leading TWA in the unipolar ECG . The alternation of APD of M cells observed under long QT conditions may exaggerate transmural dispersion of repolarization and develop torsade de pointes. Pastore et al. 1982 [<xref ref-type="bibr" rid="R35">35</xref>] investigated TWA in Langendorff-perfused guinea pig heart using mapping of epicardial APD during pacing. The critical pacing rate induced concordant TWA, which developed to discordant alternans of APD and increased susceptibility to ventricular tachyarrhythmias.</p></sec></sec><sec id="s3"><title>Measurement of microvolt TWA</title><p>Detection of microvolt TWA has been made possible by the use of advanced signal processing techniques and high-resolution electrodes to reduce noise. A number of beats, generally 128, are sampled and a time series of amplitudes of multiple corresponding points on the T wave are analyzed using a Fast Fourier Transform to generate a power spectrum (<xref ref-type="fig" rid="F1">Figure 1</xref>). Several frequency peaks correspond to respiratory variation, pedaling (if bicycle exercise is performed), and noise. The presence of alternans is indicated by a frequency peak at 0.5 cycles per beat [<xref ref-type="bibr" rid="R36">36</xref>]. The analysis yields two measurements: the alternans magnitude and the alternans ratio. The former represents the magnitude of the alternating variation in T wave morphology compared to the mean T wave; a conventional threshold of 1.9 μV is used for significance. The latter is a measure of the statistical significance of the alternans with respect to the standard deviation of the background noise; it is generally required to be greater than 3, for significance. Furthermore, by definition, TWA must be sustained for more than 1 minute [<xref ref-type="bibr" rid="R37">37</xref>].</p><p>TWA is a rate-dependent phenomenon and microvolt TWA could develop in normal subjects at sufficiently high heart rate. It has been shown that the onset heart rate is relatively low in patients with structural heart disease and history of sustained ventricular tachyarrhythmia [<xref ref-type="bibr" rid="R38">38</xref>]. Kavesh et al. [<xref ref-type="bibr" rid="R39">39</xref>] showed that both TWA and false positive results increase with heart rate. Therefore, an onset heart rate of less than 110 beats per minute is a conventional requisite for positivity.</p><p>The original study of TWA was performed with atrial pacing to increase heart rate [<xref ref-type="bibr" rid="R40">40</xref>]. Either bicycle or treadmill exercise is now available with the use of high-resolution electrodes and advanced noise reduction algorithms. However, noise, premature beats, rapid changes in heart rate, or prominent beat-to-beat variability of RR intervals, may all mask true alternans [<xref ref-type="bibr" rid="R8">8</xref>]. All of these factors - plus the failure to achieve target heart rate - may result in an indeterminate test of TWA.</p></sec><sec id="s4"><title>Clinical studies of TWA for ventricular arrhythmic risk stratification</title><p>The first large clinical study of TWA was performed by Rosenbaum and co-workers [<xref ref-type="bibr" rid="R40">40</xref>], who observed 83 patients undergoing both EPS and TWA measured during atrial pacing. Over the following 20 months, ventricular tachyarrhythmic events occurred in 81% of patients with a significant level of TWA, compared with only 6% of those without. TWA was performed equivalently to EPS as a predictor of ventricular tachyarrhythmic events. Recently, Gold et al. [<xref ref-type="bibr" rid="R41">41</xref>] reported a prospective multicenter study of TWA, measured with bicycle exercise testing, in 313 patients referred for EPS because of syncope or presyncope, cardiac arrest, ventricular tachycardia, or supraventricular tachycardia. Signal-averaged electrocardiography (SAECG) was also performed at the time of TWA. Follow-up was obtained in 290 patients with a mean duration of 297 days. The predictive value of TWA and EPS for ventricular arrhythmic events was comparable, and better than SAECG . The combination of TWA with SAECG appeared to enhance the predictive value for ventricular arrhythmic events and the results of EPS.</p><sec id="s4a"><title>TWA in patients with prior myocardial infarction</title><p>Regarding the prognostic utility of TWA in patients with a prior myocardial infarction (MI), only a few studies have been reported. TWA, SAECG, and LVEF were measured in 102 patients with recent MI (20±6 days after the onset of the MI) [<xref ref-type="bibr" rid="R42">42</xref>]. A positive TWA test showed the highest sensitivity, relative risk, and negative predictive value (NPV) but also the lowest specificity, PPV, and predictive accuracy compared to SAECG and LVEF. With multivariate analysis, the combination of TWA and SAECG was the most significant predictor. Recently, a larger cohort study consisting of 836 patients who underwent TWA testing (2.7±5.4 months after the onset of the MI) revealed that TWA predicted SCD or resuscitated ventricular fibrillation (VF) [<xref ref-type="bibr" rid="R43">43</xref>].</p><p>The sensitivity, NPV, and risk hazard of TWA for predicting SCD or VF were higher than LVEF, SAECG, and the presence of non-sustained ventricular tachycardia (VT); however, the specificity and PPV remained worse. The utility of TWA for prognosis in patients with recent MI needs to be further clarified.</p></sec><sec id="s4b"><title>TWA in patients with cardiomyopathy</title><p>Adachi et al. [<xref ref-type="bibr" rid="R44">44</xref>] reported a study of 58 patients with dilated cardiomyopathy (DCM) who underwent a TWA test. Analysis of recorded ventricular tachyarrhythmias, including non-sustained or sustained VT, revealed that ventricular tachyarrhythmias were more common in patients with a significant level of TWA (the sensitivity, specificity, and predictive accuracy rates of TWA to predict VT were 88%, 72%, and 77%, respectively).</p><p>Klingenheben et al. [<xref ref-type="bibr" rid="R45">45</xref>] studied 107 patients with congestive heart failure, a mean LVEF of 28±7%, and no history of sustained ventricular tachyarrhythmias. During 18 months of follow-up there were no patients in the TWA negative group that experienced an arrhythmic event or SCD. Multivariate Cox regression analysis revealed that TWA was the only independent predictor of arrhythmic events.</p><p> A study of 104 patients with DCM, and with 12 arrhythmic events during 21±14 months demonstrated that TWA in a group of patients with an onset heart rate less than 100 beats per minute was the most significant predictor of arrhythmia-free survival (the sensitivity, specificity, PPV, NPV, and relative risk were 75%, 78.9%, 37.5%, 94.9%, and 7.4, respectively) [<xref ref-type="bibr" rid="R46">46</xref>].</p><p>Studies of TWA in patients with other cardiomyopathies are more scarce. Momiyama et al. [<xref ref-type="bibr" rid="R19">19</xref>] studied 14 patients with HCM. A significant level of TWA was found in 71% of 7 patients at high risk of ventricular tachyarrhythmias, compared with none of the other 7 patients who were at low risk. This result suggests that TWA may be a useful marker for high risk of ventricular tachyarrhythmias in patients with HCM; however, this finding was based on a small number of patients. The role of TWA for prognosis in patients with HCM needs to be elucidated.</p></sec><sec id="s4c"><title>TWA in patients with the long QT and Brugada syndromes</title><p>Macroscopic TWA has been reported in patients with the long QT syndrome [<xref ref-type="bibr" rid="R20">20</xref>-<xref ref-type="bibr" rid="R22">22</xref>,<xref ref-type="bibr" rid="R34">34</xref>]. Prolongation and unstable state of the ventricular action potential may produce the macroscopic TWA and result in the polymorphic VT known as torsade de pointes. The prognostic value of microscopic TWA has not yet been assessed in patients with the long QT syndrome.</p><p>In patients with the Brugada syndrome, some reports have revealed that intravenous administration of class Ic antiarrhythmic drugs induced macroscopic TWA and resulted in VF [<xref ref-type="bibr" rid="R23">23</xref>,<xref ref-type="bibr" rid="R24">24</xref>]. These results suggest that in the Brugada syndrome class Ic antiarrhythmic drugs may accentuate the underlying sodium channel abnormalities, produce an unstable state of repolarization, increase the triggering PVC, and induce VF. On the other hand, Ikeda et al. [<xref ref-type="bibr" rid="R47">47</xref>] reported a low prognostic value of microscopic TWA in patients with the Brugada syndrome.</p></sec></sec><sec id="s5"><title>Limitations of TWA</title><p>There are some technical and electrophysiologic limitations of TWA. The technical limitations are: 1) TWA cannot be measured in patients with atrial fibrillation, which is a common arrhythmia in patients with structural heart disease; and 2) the presence of frequent atrial or ventricular ectopy, excessive motion artifacts, and, in particular, the inability to achieve the target heart rate would render the results of the TWA test indeterminate. An incidence of indeterminate results of up to 25% is present in most of the published studies. The electrophysiologic limitation is that TWA testing may lose much of its predictive power within a few weeks after onset of MI.</p></sec><sec id="s6"><title>Perspectives</title><p>The clinical utility of TWA in evaluating arrhythmic risk stratification appears promising for patients with suggestive ventricular tachyarrhythmias, congestive heart failure or an LVEF of less than 40%, and a recent MI. The ultimate role of the TWA test as a noninvasive predictor of SCD awaits larger-scale prospective studies. In the near future, better definition of the clinical role of TWA test will be achieved with the results of the ongoing ABCD (Alternans Before Cardioverter Defibrillator) trial, in which ICD implantation will be performed in patients with ischemic cardiomyopathy and abnormal EPS and TWA results.</p></sec>
|
The Effects Of Right Ventricular Apical Pacing On Left Ventricular Function. <italic>Stimulation Of The Right Ventricular Apex: Should It Still Be The Gold Standard?</italic>
|
<p>Current pacing practice is undergoing continuous and substantial changes. Initially pacing had an exclusively palliative role, since it was reserved for patients developing complete heart block or severe symptomatic bradycardia. With the appearance of novel pacing indications such as pacing for heart failure and atrial fibrillation, the effect of pacing site on cardiac function has become a critically important issue and a subject for consideration. It seems that the classical pacing site in the right ventricular apex is no longer the gold standard because of possible disadvantageous effects on cardiac function. The aim of this review article is to discuss the effect of right ventricular apical pacing on cardiac function including cellular and hemodynamic changes. We also aim to discuss the role of alternative pacing sites in the light of cardiac function.</p>
|
<contrib contrib-type="author"><name><surname>Szili-Torok</surname><given-names>T</given-names></name><degrees>MD, PhD</degrees></contrib><contrib contrib-type="author"><name><surname>Thornton</surname><given-names>A</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Background</title><p>One of the most challenging tasks of modern pacing is to optimise or at least stabilize cardiac performance. This is clearly dependent on the important factors of chronotropic function, the quality of AV synchrony, the ventricular activation sequence and ventricular pacing site. Over the last 40 years stimulation of the right ventricular apex became the standard method for pacing the ventricles because it is a stable and easily accessible site, and usually provides appropriate sensing and threshold parameters [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R2">2</xref>]. Although it was quickly realised that stimulation of this site leads to an abnormal contraction pattern by bypassing the physiological conduction system, for pacing in patients with life-threatening conditions it still seemed appropriate. However, with the widening of classical pacing indications and the realisation that some patients developed deterioration of left ventricular function and sometimes heart failure after pacing, it became the subject of studies into its effect on left ventricular function.</p><p>The aim of this review is to provide insights into understanding how the pacing site may influence cardiac function. This may have an impact on selection of alternative locations for the right ventricular lead when preservation or even improvement of left ventricular function seems to be important for the patient.</p></sec><sec id="s2"><title>Endocardial activation during right ventricular apical pacing</title><p>During ventricular pacing the impulse conduction occurs predominantly through the working myocardium and the normal conduction system is bypassed. During right ventricular pacing the right ventricle is activated first followed by trans-septal activation of the left ventricle, then the remaining part of the left ventricle is activated. The QRS duration is increased, due to a combination of slow trans-septal activation (local and diffuse), and delayed activation of the remaining part of the left ventricle [<xref ref-type="bibr" rid="R3">3</xref>]. A partial comparison can be drawn between the endocardial activation of patients with right ventricular apical pacing and patients with LBBB, although there are some differences. In patients with LBBB, endocardial breakthrough of right ventricular activation is on the right ventricular septum and this occurs earlier than left ventricular breakthrough [<xref ref-type="bibr" rid="R4">4</xref>]. This means that right ventricular activation in patients with LBBB is normal for most patients [<xref ref-type="bibr" rid="R4">4</xref>]. Since in most patients the right ventricular lead is placed more antero-apically than septally, clearly the endocardial activation during right ventricular apical pacing is different, resulting in a more superior axis and even more intraventricular delay during the activation of the right ventricle. On the other hand the left ventricle is still activated trans-septally in patients with right ventricular apical pacing, however left ventricular breakthrough has a heterogeneous pattern in these patients, resulting in a somewhat different activation pattern to that observed during LBBB. In most patients with LBBB high septal breakthrough is most common [<xref ref-type="bibr" rid="R4">4</xref>], whereas in most patients during RV apical pacing a single site of left ventricular breakthrough close to the ventricular apex is found. The presence of structural heart disease such as previous myocardial infarction may further increase the QRS duration during right ventricular apical pacing [<xref ref-type="bibr" rid="R5">5</xref>]. The effects of pacing site will also be different depending on whether intrinsic AV node conduction and thus fusion of activation patterns, is present [<xref ref-type="bibr" rid="R6">6</xref>,<xref ref-type="bibr" rid="R7">7</xref>].</p></sec><sec id="s3"><title>Effect of right ventricular apical pacing on the structure and function of the heart</title><p>For many years the attention given to optimising the atrio-ventricular delay during dual chamber pacing has delayed an understanding of the importance of the effect of pacing site on global and regional myocardial function. This is despite the fact that diminished ventricular function during pacing at the right ventricular apex has been known for decades from numerous animal and human studies [<xref ref-type="bibr" rid="R8">8</xref>-<xref ref-type="bibr" rid="R10">10</xref>]. Ventricular pacing results in an abnormal sequence of activation and this is associated with decreases in fiber shortening, contractile work, and myocardial blood flow and oxygen consumption in regions activated early and increases in these parameters in those regions with delayed activation [<xref ref-type="bibr" rid="R11">11</xref>-<xref ref-type="bibr" rid="R13">13</xref>]. Studies have also confirmed that the early-activated regions are hypofunctional and the late activated regions are hyperfunctional. These differences are due to regional alterations in effective preload during asynchronous activation. The ventricular wall has to adapt to these changes. The asynchronous activation induces asymmetrical hypertrophy of the left ventricular wall. This asymmetric hypertrophy in itself however does not primarily influence the pump function. Abnormal electrical activation may thus lead to depressed left ventricular function [<xref ref-type="bibr" rid="R12">12</xref>] by the classical theory of "loss of effective muscle mass". According to this theory, during ventricular pacing the ventricle looses a part of its effective muscle mass due to reduced function of the early-activated regions. An alternative or probably additional mechanism is that RV apical pacing results in inferior interventricular coupling, since during RV pacing the right ventricular pressure rises much earlier than that in the LV and this causes significant paradoxical septal motion [<xref ref-type="bibr" rid="R10">10</xref>].</p><p>Experimental animal data have also indicated that right ventricular apical pacing may decrease regional myocardial blood flow within the interventricular septum [<xref ref-type="bibr" rid="R14">14</xref>,<xref ref-type="bibr" rid="R15">15</xref>]. Prolonged compression of the septal arteries, as a result of the altered left ventricular depolarisation sequence was proposed as a possible explanation [<xref ref-type="bibr" rid="R14">14</xref>,<xref ref-type="bibr" rid="R16">16</xref>]. Extravascular compression is also a well-known determinant of coronary blood flow during the basal state or during coronary vasodilation [<xref ref-type="bibr" rid="R17">17</xref>]. There is evidence, that aberrant and delayed depolarisation of the left ventricle can result in augmented intra-myocardial pressure in the septum, thus significantly affect myocardial perfusion in this region [<xref ref-type="bibr" rid="R16">16</xref>]. These animal data have been confirmed by human studies, where atrial and/or AV sequential pacing did not alter coronary flow reserve, however ventricular pacing decreased resting coronary flow velocity in some patients [<xref ref-type="bibr" rid="R18">18</xref>]. Furthermore, long term right ventricular apical pacing results in a high incidence of myocardial perfusion defects on nuclear studies. The magnitude of these defects was linearly proportional to the duration of pacing [<xref ref-type="bibr" rid="R19">19</xref>]. These myocardial perfusion abnormalities are associated with apical wall motion abnormalities and impaired global left ventricular function [<xref ref-type="bibr" rid="R19">19</xref>]. Ventricular pacing also results in regional changes in tissue perfusion and heterogeneity between perfusion and sympathetic innervation [<xref ref-type="bibr" rid="R20">20</xref>]. Both ventricular and dual chamber pacing are associated with increase in cathecolamine activity [<xref ref-type="bibr" rid="R20">20</xref>]. Furthermore, so-called functional mitral regurgitation plays a crucial role in suboptimal hemodynamics. Right ventricular apical pacing causes significant intraventricular conduction delay. Segments of the left and right ventricle contract at different times and the interventricular septal wall contracts abnormally. This phenomenon results in decreased contractility, reduced diastolic filling and prolonged duration mitral regurgitation. The abnormal activation of the ventricles via right ventricular apical pacing may result in multiple abnormalities of cardiac function, which may ultimately affect clinical outcome.</p></sec><sec id="s4"><title>Chronic pacing of the right ventricular apex for symptomatic bradycardia</title><p>Acute clinical studies show that atrial pacing with preserved atrio-ventricular conduction and normal intraventricular conduction ensure better pump function at rest and a low pacing rates than DDD pacing with a restricted AV interval [<xref ref-type="bibr" rid="R21">21</xref>]. Because of the lack of long term follow up these findings were not confirmed by clinical studies. On the other hand evidence has been growing that pacing at alternative pacing sites improves function and influences outcome as well [<xref ref-type="bibr" rid="R22">22</xref>-<xref ref-type="bibr" rid="R24">24</xref>]. The maintenance of atrio-ventricular synchrony is a critically important issue for these patients [<xref ref-type="bibr" rid="R25">25</xref>]. The risk of atrial arrhythmias and thrombo-embolic complications is clearly higher in VVIR paced patients than in AV sequentially or AAI paced patients specifically if AV conduction is preserved [<xref ref-type="bibr" rid="R26">26</xref>]. VA conduction is has been documented in 75% of the patients with SSS and even in more than 50% of the patients with complete AV block [<xref ref-type="bibr" rid="R27">27</xref>]. Therefore maintenance of AV synchrony prevents "pacemaker syndrome". Improved exercise tolerance was also reported. Patients with SSS and VVI pacing are more likely to develop heart failure than with AAI pacing [<xref ref-type="bibr" rid="R26">26</xref>,<xref ref-type="bibr" rid="R28">28</xref>]. One of the key targets of pacing therapy apart from mortality is improved QOL. Initially, DDD pacing seemed to be of most benefit in patients with high degree AV block and preserved sinus node function [<xref ref-type="bibr" rid="R29">29</xref>]. However this hypothesis was not confirmed by recent multicenter study. Patients with SSS had more benefit from DDD pacing compared with VVI pacing, than patients with complete AV block [<xref ref-type="bibr" rid="R30">30</xref>]. Another important issue, that may explain the controversial clinical data is related to the duration of right ventricular apical pacing. It seems that a longer duration of right ventricular pacing results in a more myocardial perfusion abnormalities [<xref ref-type="bibr" rid="R19">19</xref>]. Interestingly, this phenomenon seems to be reversible even after two years pacing [<xref ref-type="bibr" rid="R19">19</xref>,<xref ref-type="bibr" rid="R31">31</xref>]. The duration of right ventricular pacing was also associated with the magnitude of decreased left ventricular ejection fraction. The reduced ejection fraction is mainly caused by an increased left ventricular end-systolic volume.</p></sec><sec id="s5"><title>Left ventricular function after AV node ablation and right ventricular apical pacing for patients with permanent atrial fibrillation: Discordant evolution of subjective and objective parameters</title><p>Patients with permanent atrial fibrillation undergoing AV node ablation and VVIR pacemaker implantation are optimal subjects for studying the effect of right ventricular apical pacing on left ventricular function for two reasons. First of all, these patients have no AV conduction, therefore the effect of AV delay does not influence the results; the effect of pacing site on the function could be studied independently. The other obvious reason is related to the concept of tachycardiomyopathy [<xref ref-type="bibr" rid="R32">32</xref>]. A relatively large proportion of these patients are suffering for a long time from permanent atrial fibrillation. Atrial fibrillation is a common supraventricular arrhythmia, which leads to cardiac dilatation and dysfunction in some patients - tachycardiomyopathy [<xref ref-type="bibr" rid="R32">32</xref>]. Theoretically, and practically in the majority of patients, ablation of the atrio-ventricular node followed by right ventricular apical pacing may result in an improvement of the patient's symptoms as well as in cardiac function because of the advantage of a regular ventricular response and adequate rate control [<xref ref-type="bibr" rid="R32">32</xref>-<xref ref-type="bibr" rid="R35">35</xref>]. However, variable results were reported about the course of patients following AV junction ablation. Although noticeable improvement in QOL has been reported, some other studies reported not improved or sometimes decreases left ventricular function [<xref ref-type="bibr" rid="R34">34</xref>,<xref ref-type="bibr" rid="R36">36</xref>,<xref ref-type="bibr" rid="R37">37</xref>]. An important aspect of these controversial data is that in most available large studies only data on the overall group was reported, despite the obvious fact that some patients deteriorated. After careful analysis of the data extracted from these studies, it seems that during the follow up, objective and subjective parameters showed somewhat of a discordant evolution. Correct interpretation of this data may allow us to develop a better understanding of the natural course of these patients and the reasons for this discordance. After AV node ablation numerous factors are influencing LV function. Some of them act in the direction of improvement, but some of them may cause deterioration. Regularisation and ventricular rate control appear to be the most important factors that may have an impact on improvement. On the other hand right ventricular apical pacing results in disadvantageous cellular changes and worsened hemodynamics. It seems so far, that the net effect of interplay between the beneficial and the worsening factors is unpredictable. The almost uniform improvement in quality of life supports the idea that subjective parameters are more influenced by the beneficial factors, however function react independently. In some patients, concordant with the QOL, function improves, however in others, despite the improvement in QOL, it may deteriorate. Therefore, in symptom control, regularisation and rate control are important factors, but their role in functional changes are not that clear. This variable outcome is of clinical significance as per the important question of Wood, as to whether AV node ablation is applicable to a wider spectrum of patients [<xref ref-type="bibr" rid="R37">37</xref>]. According to our experience and unpublished data we think that the effect of AV node ablation and right ventricular apical pacing on cardiac function is highly dependent on the baseline left ventricular ejection fraction. It seems that patients with preserved left ventricular function will more likely deteriorate their left ventricular function. Therefore, this therapy should be avoided in patients where only symptom control is the goal and who have normal cardiac function. This is in concordance with the concept of tachycardiomypoathy. It seems, that AV node ablation and right ventricular apical pacing is the best for patients with tachycardiomyopathy [<xref ref-type="bibr" rid="R32">32</xref>].</p></sec><sec id="s6"><title>Alternative pacing sites for preservation of left ventricular function: The possible role of high ventricular, left ventricular and biventricular pacing</title><p>As shown in previous trials, pacing of the right ventricular apex (the traditional pacing site) reduces LV function more than pacing the high ventricular septum or at LV sites [<xref ref-type="bibr" rid="R8">8</xref>,<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R38">38</xref>]. Since pacing of these alternate sites results in a narrowed QRS complex as compared to RV apical pacing, synchronisation of electrical activation of the two ventricles seems to have a certain level of importance [<xref ref-type="bibr" rid="R7">7</xref>,<xref ref-type="bibr" rid="R24">24</xref>]. However, in some studies LV pacing without reduction of the QRS duration was more beneficial than biventricular pacing with marked reduction in QRS duration [<xref ref-type="bibr" rid="R10">10</xref>]. The study of Prinzen et al. showed improved hemodynamics with high ventricular septum pacing [<xref ref-type="bibr" rid="R6">6</xref>], while Blanc and colleagues demonstrated improvement with LV pacing compared with RV apical pacing [<xref ref-type="bibr" rid="R39">39</xref>]. Biventricular pacing had a similar effect [<xref ref-type="bibr" rid="R39">39</xref>]. Interestingly enough, LV pacing alone was sufficient to improve LV function [<xref ref-type="bibr" rid="R39">39</xref>]. We can conclude from these studies that the pacing site probably has more influence on LV function than changes in the activation sequence itself. Certainly all of these studies were acute hemodynamic studies; therefore correlation with long-term results is still to be investigated. The mechanism of this action stems from the classical theory of "loss of effective muscle mass". According to this theory, during ventricular pacing the ventricle looses a part of its effective muscle mass due to reduced function of the early activated regions. Studies confirmed that the early-activated regions are hypofunctional and the late activated muscles are hyperfunctional. This results in a substantial degree of asynchrony. An alternative mechanism is that left ventricular pacing results in a superior interventricular coupling. During RV pacing the right ventricular pressure rises much earlier than during LV pacing and causes a significant paradoxical septal motion [<xref ref-type="bibr" rid="R10">10</xref>].</p><p>In conclusion, the ventricular pacing site has a major effect on ventricular function. It influences hemodynamics, initiates cellular changes and alters myocardial perfusion. These changes might result in left ventricular dysfunction in a considerable number of patients. Since the patients who develop dysfunction with pacing cannot be predicted according to our present knowledge, selection of alternative pacing sites, when the preservation and improvement of left ventricular function is important, should be considered during the routine clinical practice.</p></sec>
|
Pacing for the Suppression of Paroxysmal Atrial Fibrillation in an 87-year-old Patient
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Levine</surname><given-names>Paul A</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Wachsner</surname><given-names>Robin</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>El-Bialy</surname><given-names>Adel</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Clinical History</title><p>The patient is a frail 87-year-old woman with symptomatic sinus node dysfunction and documented paroxysmal atrial fibrillation (PAF). She presented with PAF and presyncopal spells. The presyncopal episodes coincided with episodes of sinus arrest occurring upon spontaneous termination of AF. Various antiarrhythmic agents (AA) were utilized including digoxin 0.125 mg/d with serum digoxin level of 1.0 ng/ml, atenolol starting at 25 mg QD, titrating up to 50 mg BID and sotalol 80 MG BID. All of these AA were either ineffective, exacerbated her bradyarrhythmia or caused intolerable side effects. She was also placed on coumadin 2.5 mg alternating with 5.0 mg due to the frequent episodes of AF, with a target INR of 2-3.</p><p>A pacemaker (St. Jude Medical Integrity™ μ DR) was implanted (October 21, 2001) to provide dual-chamber pacing (DDD mode) to manage her symptomatic bradycardic episodes and allow for the safe administration of pharmacologic therapy. The automatic mode switch function (AMS) was enabled to prevent tracking of the high atrial rate by the pacemaker. It also served as a diagnostic marker for recurrent and possibly clinically asymptomatic episodes of PAF. This function allows internal pacemaker diagnostic records to detail pacing and AMS events, enabling their quantification at patient follow-up.</p><p> At five weeks, the pacing system was functioning appropriately, but the patient complained of fatigue. It was rationalized that the fatigue might be an inherent effect of her medications.</p><p>At this time, all medications except atenolol were discontinued and the atenolol was reduced to 25 mg QD, to facilitate control of the ventricular response during AF. Coumadin was continued.</p><p>At her follow-up visit 4 months post-implant, the patient reported a marked decrease in the frequency, but not total elimination of her palpitations. There was total elimination of her near syncopal spells. Standard DDD pacing appeared to have been effective in treating her presyncopal symptoms but at standard settings, episodes of paroxysmal atrial fibrillation continued. The AMS event log recorded 43 episodes, most being triggered by atrial rates > 300 bpm and consistent with AF. Although the cumulative episodes of atrial fibrillation accounted for < 1% of the rhythm activity since her last evaluation, she continued to complain of the palpitations. As she was intolerant of most medications, the AF Suppression™ algorithm integral to her implanted pacing system was enabled and a follow-up visit was scheduled in three months.</p><p>On her return, she reported feeling well with resolution of both palpitations fatigue. The AMS event log data showed no AMS episodes. It was opted to withdraw the low dose of atenolol at this time but continue the coumadin. Future plans include a follow up in six months and if AF remains suppressed, to discontinue coumadin anticoagulation.</p></sec><sec sec-type="discussion" id="s2"><title>Discussion</title><p>The AF Suppression™ algorithm is a newly approved algorithm designed to provide a high percentage of atrial overdrive pacing which usurps the control of the atrial rate, either from the sinus node or an ectopic foci. The objective is reduced temporal dispersion of the atrial refractory period combined with overdrive suppression of ectopic beats, a common trigger for atrial tachyarrhythmias. AMS remains functional with the AF Suppression™ algorithm on, thus if AF occurs, AMS is activated facilitating management as this provides information as to the frequency and duration of each AMS episode. Based on the detected atrial rate, one can infer the mechanism of the atrial arrhythmia with the faster episodes being atrial fibrillation.</p><p>Sinus node dysfunction, AV block and AF progressively increase in frequency with advancing age [<xref ref-type="bibr" rid="R1">1</xref>]. Standard treatment for symptomatic bradyarrhythmias is pacing while rapid heart rhythms are commonly treated pharmacologically [<xref ref-type="bibr" rid="R2">2</xref>,<xref ref-type="bibr" rid="R3">3</xref>]. However, pharmacologic therapy is often a challenge in the very elderly ( > 85) due to alterations in pharmacokinetics and pharmacodynamics. Management of these patients becomes a complex task, particularly when the number of medications prescribed increases and one has to be concerned with adverse drug interactions and side effects [<xref ref-type="bibr" rid="R4">4</xref>]. </p><p>Device therapy is the standard approach to the management of a symptomatic bradycardia. The majority of pacemaker patients are elderly, with more than 85% of pacemaker recipients being at least 64 years old [<xref ref-type="bibr" rid="R5">5</xref>]. Thus, as the technology and features of pacemakers improve, it is important for the clinician to understand how these technological advances can optimize patient care.</p><p>AF is specific arrhythmia associated with age, either by itself or in conjunction with other conduction abnormalities. The prevalence of AF doubles for each decade of life after age 60 [<xref ref-type="bibr" rid="R6">6</xref>]. Although not lethal, patients with AF have twice the risk for death, and if not on anticoagulant therapy, a five-fold risk for stroke. This risk escalates in patients 75 and older [<xref ref-type="bibr" rid="R7">7</xref>]. Quality of life is also compromised as a consequence of the loss of atrial transport and the irregular and often rapid ventricular rates. The initial treatment of AF is pharmacologic designed to either restore sinus rhythm or to control the ventricular response to this rhythm. In those individuals who have the bradycardia-tachycardia syndrome, dizziness and/or syncope are commonly associated with the protracted asystolic pauses following spontaneous termination of a paroxysmal atrial fibrillation episode. Pharmacologic therapy may exacerbate these symptoms by further lengthening the sinus node recovery time following spontaneous termination of an episode as well as exacerbating the underlying bradycardia when in sinus rhythm. A common treatment for PAF associated with bradycardia-tachycardia syndrome is implantation of a permanent pacemaker in combination with drugs, as was done in this case [<xref ref-type="bibr" rid="R8">8</xref>]. Although drug therapy is the initial therapy, the majority of cardiovascular agents, can have deleterious effects in the geriatric population requiring careful attention to management [<xref ref-type="bibr" rid="R9">9</xref>].</p><p>The decision at the time of implantation was to select a device with a commercially available overdrive suppression algorithm although there was no way to determine whether or not it would be successful in this patient prior to its implantation. The addition of the commercially approved AF Suppression™ algorithm proved to be a valuable adjunct to this patient's management allowing us to discontinue a multiplicity of medications. We cannot determine whether it was stabilization of her atrial rhythm or elimination of her polypharmacy that accounted for her clinical improvement but both were made possible by technologic advances in the form of the AF Suppression algorithm.</p><p>While the algorithm may not be 100% effective in all patients, any reduction in the number of AF episodes is beneficial. In this case, standard pacing resulted in a marked improvement for the patient, but episodes of symptomatic and asymptomatic PAF continued. The ability to then enable the AF Suppression™ algorithm provided additional benefit, further improving this patient's quality of life as well as allowing discontinuation of her current antiarrhythmic medications. Although this is an isolated case, the use of a pacemaker containing the AF Suppression™ algorithm should be considered in all patients undergoing device implant for a symptomatic bradycardia, particularly when there is a prior history of atrial fibrillation or they are at increased risk of atrial fibrillation which is more likely when the indication for pacing is sinus node dysfunction.</p></sec>
|
Atrial Remodeling: Evolving Concepts
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Biffi</surname><given-names>Mauro</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Boriani</surname><given-names>Giuseppe</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<p>The term "Electrophysiological remodeling" defines the changes of atrial electrophysiologic properties taking place in atrial myocites during atrial fibrillation and/or following periods of sustained atrial fibrillation (AF).</p><p> Early research was prompted by clinical observations: many patients were seen to get through increasingly frequent and longer paroxisms of AF to persistent AF, until chronic AF eventually ensued, without significant changes of underlying heart disease.</p><p>Indeed, clinical electrophysiology investigations reported peculiar differences between AF patients and patients without atrial arrhythmias. It was felt that the electrophysiologic milieu was the cause of AF [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R2">2</xref>], until a change of perspective occurred in the early 90's, thanks to experimental research.</p><p>In his pioneering work, Wijffels kept goats fibrillating by an implanted device, while measuring atrial refractoriness and its rate adaptation as spontaneous restoration of sinus rhythm occurred. He found that, after several self terminating short bouts of AF, the arrhythmia became considerably longer until a duration of days or weeks was reached. The electrophysiologic companionship of AF pattern transformation consisted of significant shortening of atrial fibrillation cycle, which was paralleled by shortening of atrial ERP and loss of ERP adaptation to heart rate: either flat (no change with shortening cycle length) or inverted (paradoxical shorter ERP at longer cycle length) ERP relationships to cycle length were observed.</p><p>This findings had been already observed in AF patients [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R2">2</xref>]; the novelty was that AF itself was the cause of these abnormal electrophysiologic properties, which could in turn promote further recurrences of AF, and arrhythmia persistence in some cases.</p><p>By the light of this observations, AF may be seen as self-maintaining by creating a functional electrophysiologic substratum, despite the coexistence of only minor structural heart disease or occasional triggers. This principle has been summarised by Maurits Allessie in the aphorism "AF begets AF".</p><p>Following this observations, several studies have addressed the issue of the mechanisms by which atrial electrophysiologic remodeling occurs in different settings: basic electrophysiology, experimental electrophysiology, and clinical electrophysiology. It is nowadays believed that, beyond structural heart disease, changes of the atrial electrophysiologic substratum are responsible for AF recurrence and/or perpetuation.</p><p>The aim of this review will be to assess the extent at which atrial remodeling occurs in the experimental animal and in humans, and its relevance to clinical practice.</p><sec id="s1"><title>Mechanisms of atrial remodeling</title><p>Reviewing literature, one has to consider carefully the investigational setting of reported studies, for significant differences exist in electrophysiologic properties and vulnerability to atrial arrhythmias among species (small mammals atria hardly fibrillate, dogs and goats only under pathologic conditions), so that concepts must lean on solid evidence before moving from experimental to clinical electrophysiology, and eventually to clinical practice.</p><p>On the other hand, the settings of experimental and basic electrophysiology allow precisely controlled conditions, and a deeper knowledge of the mechanisms, up to the ionic/genomic level.</p><p>It is well demonstrated that atrial remodeling is a time-dependent process that develops as an adaptive regulation of cardiac myocites to maintain cell homeostasis against external stressors [<xref ref-type="bibr" rid="R3">3</xref>]. The type and extent of remodeling depends on the strength and the duration of exposure to the "stressor": adaptive responses may thus occur at the ionic/genomic level (fully reversible) at short term, or at the cellular level at mid (hibernation, usually reversible) and long term (apoptosis and fibrosis, irreversible).</p><p>The most common "stressors" of atrial myocites are: tachycardia (high rate of cell depolarisation), and volume/pressure overload (heart failure syndrome).</p><p>The adaptive changes at the ionic level are quite different in the tachycardia-dependent compared to the heart failure setting, as demonstrated by basic electrophysiology, hence we need to focus on different types of "ionic remodeling".</p><sec id="s1a"><title>Tachycardia-dependent remodeling</title><p>High atrial rate causes calcium overload. Chronic calcium overload may cause ultrastructural changes that resemble myocardial ischemia or hibernation, and may ultimately lead to irreversible cell damage. To prevent dangerous calcium overload, atrial myocites decrease L-type Ca++ current (ICa) at first by a short-term adaptation (functional inactivation by voltage- and [Ca++] - dependent mechanisms), in the long term by downregulation of ICa [<xref ref-type="bibr" rid="R4">4</xref>-<xref ref-type="bibr" rid="R7">7</xref>]. Decreased ICa shortens action potential duration (refractoriness) and causes loss of action potential adaptation to heart rate [<xref ref-type="bibr" rid="R6">6</xref>,<xref ref-type="bibr" rid="R7">7</xref>]: these are the hallmark of tachycardia-dependent remodeling.</p><p>The process leading to ICa downregulation implies transcriptional changes at the genomic level, but the signal coupling calcium overload to changes in ion channel expression is unknown. It also remain to be understood whether calcium overload is the only initiating signal in this setting.</p><p>The observations by Wijffels et al [<xref ref-type="bibr" rid="R8">8</xref>] in the experimental animal are in full agreement with this model of atrial remodeling. In patients with lone paroxysmal AF, Capucci et al [<xref ref-type="bibr" rid="R9">9</xref>] demonstrated a strict correlation of atrial refractoriness to AF cycle, and observed progressive shortening of AF cycle during long lasting AF, whereas AF cycle prolonged in self-terminating AF episodes. Acute administration of class 1C drugs was able to prolong AF cycle and terminate long lasting AF episodes [<xref ref-type="bibr" rid="R10">10</xref>]. These data are consistent with basic electrophysiology, and are clues that ionic remodeling occurs early during AF, and sets the background for AF recurrence. In the clinical scenario, early AF recurrence after cardioversion of chronic AF is the most threatened complication. In this setting a typically tachycardia-dependent remodeling occurs: in a study of 101 patients [<xref ref-type="bibr" rid="R11">11</xref>] we identified loss of atrial refractoriness adaptation to heart rate, short atrial refractoriness and AF duration as the only predictors of AF recurrence at logistic analysis.</p><p>In a series of 28 patients, Manios et al [<xref ref-type="bibr" rid="R12">12</xref>] similarly observed that failure of action potential adaptation to heart rate was the only prognostic indicator for AF recurrence. Importantly, they observed that the electrophysiological markers of tachycardia-dependent remodeling recover nearly 24 hours after AF cardioversion. This point raises an important question, that is, whether loss of refractoriness adaptation to heart rate is the cause of AF relapse (which could explain only AF relapse within 24-48 hours) or simply a "marker" of a pathologic atrium, which has already undergone some degree of structural remodeling and thus is prone to AF. Up to now, the question is unresolved.</p><p>Owing to this knowledge, a considerable amount of work has been done to test the hypothesis that preventing calcium overload by Icablockers would prevent tachycardia-dependent remodeling also. Experimental studies found that pre-treatment with verapamil could attenuate [<xref ref-type="bibr" rid="R13">13</xref>] or prevent [<xref ref-type="bibr" rid="R14">14</xref>] rapid pacing-induced shortening of atrial refractoriness at short term (< 24 hours of rapid pacing), respectively in goats and dogs. Following these observations, Daoud [<xref ref-type="bibr" rid="R15">15</xref>] reported that pre-treatment with verapamil reduced short-duration (5.6 3.9 minutes) AF-induced shortening of atrial refractoriness in humans. In the clinical scenario, Tieleman et al [<xref ref-type="bibr" rid="R16">16</xref>] speculated that pre-treatment with verapamil or diltiazem was responsible for reduced AF recurrence in the first month after electrical cardioversion, due to attenuated or rapidly reversed tachycardia-dependent remodeling.</p><p>This hypothesis fits poorly with the electrophysiologic findings by Manios [<xref ref-type="bibr" rid="R12">12</xref>] who demonstrated that atrial remodeling recovers spontaneously nearly 24 hours after cardioversion: the role of verapamil is indeed unclear. Clinical studies have failed to demonstrated any beneficial effect of verapamil in preventing AF recurrence at long term; indeed, a small effect of verapamil pre-treatment could reasonably be expected only in AF of short duration, and does not affect the electrophysiologic changes occurring with long-lasting AF [<xref ref-type="bibr" rid="R17">17</xref>].</p><p>Lee et al [<xref ref-type="bibr" rid="R17">17</xref>] have performed a very important research in chronically instrumented dogs, in which electrophysiologic data were studied at baseline and after creation of a tachycardia-dependent atrial remodeling, comparing a control group with a verapamil-treated group. Atrial refractoriness, dispersion of atrial refractoriness, rate adaptation of atrial refractoriness, atrial conduction velocity AF inducibility, and AF duration were evaluated respectively after 1 day, 1 week, and 6 weeks of rapid atrial pacing. This study demonstrated that verapamil could only attenuate the refractoriness shortening after 1-day pacing, but could not prevent the electrophysiologic remodeling at long term (1 and 6 weeks). Furthermore, verapamil did not affect atrial inducibility, and increased the duration of induced AF. Consistent with this finding is the observation of increased fragmented activity in paroxysmal AF patients treated by verapamil, as reported by Kumagay [<xref ref-type="bibr" rid="R18">18</xref>], which would favour AF persistence. In his study, Lee [<xref ref-type="bibr" rid="R17">17</xref>] reported a significant depression of atrial conduction velocity after 6 weeks rapid pacing: this abnormality had no evidence of recovery during two days of follow-up after pacing had been stopped. A similar finding had been previously reported in 2 studies [<xref ref-type="bibr" rid="R19">19</xref>,<xref ref-type="bibr" rid="R20">20</xref>] of chronic paced dogs, and were related to structural abnormalities occurring along time [<xref ref-type="bibr" rid="R21">21</xref>]. All these data make the point that verapamil use is of little value when high atrial rates persist longer than 24 hours, and thus give an explanation to the results observed in clinical practice.</p><p>While demonstrating ICa reduction by downregulation of L-type Ca++ channels, Yue et al [<xref ref-type="bibr" rid="R6">6</xref>] observed an unchanged inward T-type Ca++ current (ICaT ) even after several weeks of rapid atrial pacing, who could thus represent a continuous "spill" of calcium into atrial cells undergoing high frequency depolarisations. This prompted Fareh et al [<xref ref-type="bibr" rid="R22">22</xref>] to investigate the effect of mibefradil, a pure T-type Ca++ blocker, in a chronic rapid-pacing dog model to prevent the development of an electrophysiologic substrate for AF. After 1 week of rapid pacing, mibefradil-treated dogs did not differ from control dogs (no rapid pacing) in any electrophysiologic parameter, except for a slightly increased heterogeneity of atrial refractoriness, whereas placebo-treated dogs had developed all the characteristic abnormalities of tachycardia-dependent remodeling (easy AF inducibility included). A potential direct electrophysiologic effect of mibefradil was also excluded in this study, thus supporting the evidence that prevention of tachycardia-induced remodeling as the only protective effect of mibefradil. Indeed, the AF-promoting effects of rapid pacing were negatively correlated to mibefradil plasma concentrations, showing a dose-dependent protection against atrial remodeling.</p><p>In a similar study on dogs, Fareh et al [<xref ref-type="bibr" rid="R23">23</xref>] also demonstrated that prevention of tachycardia-dependent remodeling is possible with the T-type Ca++ channel blocker mibefradil, but not with L-type Ca++ channel blockers.</p><p>There is considerable evidence that, in tachycardia-dependent remodeling, calcium overload plays an important role either at short or at long term if high atrial rates persist. In the clinical setting, small benefits may be observed at short term by verapamil, whereas T-type Ca++ channel blockers seem promising to prevent long term remodeling during persistent AF.</p><p>These data support the concept that new therapies should target the development of the substrate for AF, as opposed to "traditional" antiarrhythmic therapy which aims to modify the product of the arrhythmic substrate.</p></sec><sec id="s1b"><title>Heart failure-dependent remodeling</title><p>Heart failure (HF) is the most important clinical syndrome associated to AF [<xref ref-type="bibr" rid="R24">24</xref>]. Heart failure causes important ionic remodeling either at the atrial or at the ventricular level by powerful neurohormonal changes. The study of atrial remodeling in HF is very difficult in humans because of the structural changes caused by underlying myocardial pathology, and because of the confounding effect of drugs acting on the neurohormonal milieu and of antiarrhythmic agents. For this reason most of available data rely upon basic and experimental research, most often on ventricular myocites.</p><p>Heavy alterations in calcium handling are prominent in ventricular myocites: the sarcoplasmic carrier of Ca++ is downregulated in failing myocites [<xref ref-type="bibr" rid="R25">25</xref>], leading to increased cytosolic [Ca++] as a maladaptive attempt to maintain contractility. Sodium-calcium exchanger current (NCX) is instead upregulated in HF, acting to compensate the impaired Ca++ removal due to the sarcoplasmic carrier [<xref ref-type="bibr" rid="R26">26</xref>-<xref ref-type="bibr" rid="R28">28</xref>], to protect contractile function and cellular Ca++ homeostasis. NCX exchanges one Ca++ for three Na+ , thus carrying a net current in the direction of Na+ transport. The net current carried by NCX may cause delayed afterdepolarisation (DAD) in certain electrophysiologic situations, which proved to be arrhythmogenic in HF [<xref ref-type="bibr" rid="R29">29</xref>].</p><p>Danshi li et al. [<xref ref-type="bibr" rid="R30">30</xref>] created a canine model of rapid ventricular pacing-induced HF, and observed for the first time in atrial myocites a decrease in ICa, in Na+ current (Ito), in slow rectifying K+ current (Iks), and an increase in NCX. Heart failure decreased both Iks and ICa 30%, whereas atrial tachycardia decreases ICa 70% without affecting Ik [<xref ref-type="bibr" rid="R6">6</xref>]. The electrophysiologic manifestations were an unchanged action potential duration (APD) at slow heart rates compared to control dogs, whereas an increase of APD at faster high rates was observed.</p><p>Thus, a peculiar type of remodeling with an attenuated rate-adaptation of atrial refractoriness and long APD at high rates is observed in experimental HF. This is quite different from atrial tachycardia-dependent remodeling, which shows decreased APD at any cycle length and flat or reversed adaptation of refractoriness to heart rate.</p><p>The mechanisms of AF onset in the setting of HF must be very different. Structural changes (interstitial fibrosis) related to underlying disease cause conduction abnormalities that predispose to reentrant arrhythmias, increased NCX may promote DADs by transient inward depolarising current, which may be enhanced by high cathecolamine levels and by decreased Iks. It seems then likely that AF may be initiated by triggered activity, in the HF setting.</p><p>If a different sort of atrial remodeling occurs in HF, peculiar interventions for targeting this arrhythmogenic substrate should be sought.</p><p>Danshi Li and co-workers [<xref ref-type="bibr" rid="R31">31</xref>] performed an elegant investigation on 15 dogs with HF induced by rapid ventricular pacing compared to 10 dogs without HF subjected to rapid atrial pacing (RAP), evaluating the effect of dofetilide on vulnerability to AF. It was observed that dofetilide, a pure Ikr blocker, was extremely effective in terminating AF and in preventing AF reinduction in dogs with HF-dependent AF. On the contrary, dofetilide was completely ineffective in dogs with atrial tachycardia-dependent AF. The electrophysiologic findings were striking: atrial refractoriness was shorter and showed less rate adaptation in RAP vs HF dogs; dofetilide increased refractoriness significantly and to a greater extent in HF dogs, whereas it had nonsignificant effects in RAP dogs.</p><p>This finding was related to the increased dependency of atrial refractoriness on Ikr in HF dogs, due to downregulation of Iks occurring in HF dogs.</p><p>Significant clinical results were indeed observed by Torp-Pedersen et al [<xref ref-type="bibr" rid="R32">32</xref>] investigating dofetilide effects in advanced HF patients (class III-IV): a significant proportion of AF patient converted to sinus rhythm by dofetilide, and dofetilide patients had a significantly higher proportion of sinus rhythm persistence during clinical follow up compared to control patients.</p><p>Nonetheless, AF incidence increases with the severity and the duration of HF despite therapeutic interventions.</p><p>Shinigawa et al [<xref ref-type="bibr" rid="R33">33</xref>] studied the electrophysiologic remodeling occurring during atrial tachycardia in a canine model when HF was superimposed, and also after reversal of HF was allowed to take place. They observed that the electrophysiologic changes related to atrial tachycardia were completely attenuated by HF, and that reversal of HF led to normalisation of both ventricular and atrial volume and contractility, with recovery of atrial electrophysiologic abnormalities. However, extensive interstitial fibrosis had developed compared to a group of control dogs, and this invariably caused local conduction slowing and heterogeneity of conduction velocity. This changes were irreversible.</p><p>Interestingly, in the same experimental canine model of HF, Danshi Li et al [<xref ref-type="bibr" rid="R34">34</xref>] reported that angiotensin II causes structural atrial remodeling inducing collagen synthesis by fibroblasts, and that enalapril interferes with signal transduction at the cellular level, thereby reducing the amount of interstitial fibrosis by 30% compared to control dogs or to hydralazine-treated dogs.</p><p>These two studies carry the important notion of targeting the development of the arrhythmogenic substrate understanding the pathophysiologic mechanisms of the disease or of the clinical syndrome.</p><p>It is interesting to note that, either in atrial tachycardia-dependent or in HF-dependent remodeling, both reversible and irreversible changes occur. In either situation, the abnormalities due to ionic remodeling (APD, refractoriness and its adaptation to heart rate) appear to be reversible when the "stressor" is removed, but in the same time structural changes at the cellular level have occurred, which are irreversible. This "structural remodeling" adds slowing and heterogeneity of conduction velocity to the abnormalities of refractoriness in both situations, and may explain the reduced efficacy of medical interventions in patients with persistent or permanent AF.</p><p>Experimental research has clearly demonstrated that "electrophysiologically" different arrhythmias share a common electrocardiographic picture of atrial fibrillation, and that the underlying differences imply peculiar therapeutic interventions.</p><p>All these investigations have brought in the clinical arena a new way of thinking antiarrhythmic therapy: development of new drugs should be guided by the principle of targeting the specific ionic currents depending on the pathophysiologic mechanisms of the arrhythmia, to achieve the optimal ratio between efficacy and safety. Also, the concept of preventing structural remodeling of atrial myocites should be investigated and promptly brought into clinical practice.</p></sec></sec>
|
Left Septal Fascicular Block: Myth Or Reality?
|
<p>Anatomic studies have shown that the left bundle branch divides into three fascicles in most humans. Changes in the 12 lead ECG (electrocardiogram) due to conduction abnormalities of the left anterior fascicle and left posterior fascicle are now part of the standard repertoire of electrocardiographic interpretation. There are no standard criteria for detecting conduction defects involving the third left fascicle, the septal or median fascicle, and the very existence of such defects is still a matter of controversy. The purposes of this article are to review the available evidence on this subject, suggest electrocardiographic criteria for its recognition, and present examples which illustrate that left septal fascicular block does indeed exist as a specific entity. Left septal fascicular block is a polymorphic conduction defect which may explain some previously inadequately understood electrocardiographic abnormalities.</p>
|
<contrib contrib-type="author"><name><surname>MacAlpin</surname><given-names>Rex N</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>More than thirty years ago Rosenbaum, Elizari, and Lazzari [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R2">2</xref>] presented their monograph defining left anterior and left posterior fascicular blocks, or what they termed "the hemiblocks". These divisional blocks involving part of the left bundle branch system have become well-recognized aspects of clinical electrocardiography [<xref ref-type="bibr" rid="R3">3</xref>-<xref ref-type="bibr" rid="R6">6</xref>] Anatomic studies, however, have suggested that the termination of the left bundle branch itself is more complex than just a bifurcation [<xref ref-type="bibr" rid="R7">7</xref>-<xref ref-type="bibr" rid="R14">14</xref>]. Studies of the conduction system going as far back Tawara's work in 1906 [<xref ref-type="bibr" rid="R7">7</xref>] have indicated that as the left bundle branch divides, it gives off fibers to the left endocardial surface of the interventricular septum, as well as to its better recognized anterior and posterior divisions [<xref ref-type="bibr" rid="R7">7</xref>-<xref ref-type="bibr" rid="R14">14</xref>]. In some cases these septal fibers are distributed broadly over the left septal surface like an interconnecting network [<xref ref-type="bibr" rid="R7">7</xref>,<xref ref-type="bibr" rid="R8">8</xref>,<xref ref-type="bibr" rid="R11">11</xref>]. But the elegant histologic reconstructions of the left conduction system by Demoulin and Kulbertus [<xref ref-type="bibr" rid="R12">12</xref>] in 1972 showed that, although there is considerable variability in its anatomy, in most cases the septal fibers can be identified as a distinct third division of the left bundle branch, as shown in <xref ref-type="fig" rid="F1">Figure 1</xref>.</p><p>In 1970 Durrer and his coworkers [<xref ref-type="bibr" rid="R15">15</xref>] described the activation sequence of the ventricles using isolated, perfused human hearts. They found that earliest activation of the left ventricle occurred, 0 to 5 ms after the onset of the left ventricular cavity potential simultaneously in three endocardial regions: an anterior paraseptal area just below the mitral valve, a posterior paraseptal area about one third of the way from apex to base, and the central region of the left septal surface. Right ventricular activation began about 5 ms later near the insertion of its anterior papillary muscle. It was concluded from these studies that these three areas of simultaneous left ventricular activation corresponded roughly to the insertions of the three fascicles that were the terminations of the left bundle branch [<xref ref-type="bibr" rid="R12">12</xref>,<xref ref-type="bibr" rid="R15">15</xref>]. It was also pointed out that the activation waves resulting from the anterior and posterior areas of initial left ventricular activation progress in directions approximately opposite to each other, and hence the resulting potential fields they create tend to cancel each other [<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R16">16</xref>]. This leaves activation of the interventricular septum starting from its left septal surface to dominate the QRS in its first 10 ms, despite a small, opposite, slightly delayed potential from activation of the right septal surface via the right bundle branch [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R15">15</xref>]. This produces initial QRS forces directed to the right and anteriorly, causing initial negative deflections in left-to-right oritented leads, q waves in leads I, V5-V6. These q waves have been termed "septal q waves" [<xref ref-type="bibr" rid="R17">17</xref>]. This model of ventricular activation incorporates a quadrafascicular picture of the ramifications of the His bundle: right bundle branch and three fascicles of the left bundle branch [<xref ref-type="bibr" rid="R15">15</xref>,<xref ref-type="bibr" rid="R18">18</xref>]. It is helpful to have this trifascicular pattern of the left bundle branch divisions in mind when trying to understand the changes that occur in direction of initial QRS forces with left anterior and left posterior fascicular blocks [<xref ref-type="bibr" rid="R2">2</xref>-<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R16">16</xref>,<xref ref-type="bibr" rid="R19">19</xref>]. The term "hemiblock" is therefore no longer felt appropriate in describing these left intraventricular conduction abnormalities [<xref ref-type="bibr" rid="R12">12</xref>,<xref ref-type="bibr" rid="R20">20</xref>].</p><p>The QRS alterations in left anterior and left posterior fascicular blocks are well described not only in the monograph of Rosenbaum, Elizari, and Lazzari [<xref ref-type="bibr" rid="R2">2</xref>], but also in medical articles [<xref ref-type="bibr" rid="R19">19</xref>], and texts of electrocardiography and vectorcardiography [<xref ref-type="bibr" rid="R3">3</xref>-<xref ref-type="bibr" rid="R6">6</xref>,<xref ref-type="bibr" rid="R16">16</xref>]. There are reasons why such agreed upon criteria do not exist for block of the third division of the left bundle branch, left septal fascicular block (LSFB). The anatomy of this fascicle, where it exists as a distinct entity, varies more than that of the other two left fascicles [<xref ref-type="bibr" rid="R12">12</xref>]. These septal fibers usually have many interconnections [<xref ref-type="bibr" rid="R8">8</xref>,<xref ref-type="bibr" rid="R12">12</xref>]. This makes some persons less likely than others to suffer block of this division of the left bundle branch. Septal fibers of the left bundle branch may have the shortest refractory period of conduction fibers originating from this bundle branch [<xref ref-type="bibr" rid="R21">21</xref>,<xref ref-type="bibr" rid="R22">22</xref>]. Various criteria for LSFB have been proposed based on case reports, [<xref ref-type="bibr" rid="R11">11</xref>,<xref ref-type="bibr" rid="R23">23</xref>-<xref ref-type="bibr" rid="R29">29</xref>] on consequences of experimental incision of septal fibers in the dog, [<xref ref-type="bibr" rid="R30">30</xref>,<xref ref-type="bibr" rid="R31">31</xref>] on electrical-anatomical models, [<xref ref-type="bibr" rid="R9">9</xref>] and on computer-based projections of QRS morphology [<xref ref-type="bibr" rid="R32">32</xref>,<xref ref-type="bibr" rid="R33">33</xref>]. It may present with multiple morphologies, and is frequently combined with other conduction abnormalities which obscure its presence [<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R23">23</xref>,<xref ref-type="bibr" rid="R34">34</xref>]. Popular textbooks of electrocardiography tend to ignore the possibility of its existence [<xref ref-type="bibr" rid="R5">5</xref>], to dismiss it as unproven [<xref ref-type="bibr" rid="R6">6</xref>], controversial [<xref ref-type="bibr" rid="R4">4</xref>], or to maintain it cannot be diagnosed [<xref ref-type="bibr" rid="R4">4</xref>,<xref ref-type="bibr" rid="R35">35</xref>]. When it is mentioned, the coverage is brief [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R6">6</xref>,<xref ref-type="bibr" rid="R36">36</xref>].</p><p>One major reason for the multiple morphologies with which LSFB may present is that septal activation is normally accomplished by a double envelopment starting from both left and right septal surfaces as contrasted to the single, endocardial to epicardial direction of activation of the free walls of the two ventricles [<xref ref-type="bibr" rid="R37">37</xref>-<xref ref-type="bibr" rid="R39">39</xref>]. Reference to a diagrammatic depiction of normal ventricular activation in <xref ref-type="fig" rid="F2">2A</xref> illustrates this point. Activation of the right septal surface is initiated by septal branches of the right bundle branch, starting no more than 5 ms after first activation of the left septal surface [<xref ref-type="bibr" rid="R15">15</xref>,<xref ref-type="bibr" rid="R37">37</xref>]. The activation wave traverses the septum in either direction in about 40 ms, so that septal activation is nearly completed within 35 ms of the beginning of ventricular depolarization [<xref ref-type="bibr" rid="R15">15</xref>,<xref ref-type="bibr" rid="R37">37</xref>], except perhaps for parts of the basal septum, which has a relative dearth of Purkinje fibers leading into it [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R11">11</xref>,<xref ref-type="bibr" rid="R40">40</xref>]. By the time instantaneous QRS vectors are dominated by left ventricular free wall activation and have already rotated to the left at 40 ms, septal activation has been almost completed and will have little effect on the direction or magnitude of terminal QRS forces [<xref ref-type="bibr" rid="R37">37</xref>].</p><p>The amount of the septum normally activated via the right bundle branch varies between individuals [<xref ref-type="bibr" rid="R15">15</xref>]. Orientation of the septum also varies, facing to the right, anteriorly and inferiorly in horizontal hearts, and anteriorly, superiorly, and either to the left or right in vertical hearts [<xref ref-type="bibr" rid="R41">41</xref>]. As the direction of initial QRS forces is related to the orientation of the septum [<xref ref-type="bibr" rid="R42">42</xref>], the effect of LSFB on the ECG may vary with septal orientation.</p></sec><sec id="s2"><title>Possible Models of LSFB</title><p>Based on this model of ventricular activation, <xref ref-type="fig" rid="F2">Figure 2</xref> illustrates a range of possibilities with LSFB. In <xref ref-type="fig" rid="F2">Figure 2A</xref> is represented the normal sequence of ventricular activation. In <xref ref-type="fig" rid="F2">Figure 2B</xref>, the delay in left septal activation is only slight, such that the potentials arising from the activation wave proceeding from left-to-right are exactly balanced by those generated by the right-to-left activation wave. This would result in a leftward shift of initial QRS forces, with abolition septal q waves, but with little other change in QRS configuration or duration [<xref ref-type="bibr" rid="R9">9</xref>]. In <xref ref-type="fig" rid="F2">Figure 2C</xref> the delay in left septal activation is greater, resulting in dominant right-to-left and front-to-back septal activation, yielding initial QRS forces directed to the left and posterior. If not counterbalanced by simultaneous anteriorly directed forces from right ventricular activation, this would result in q waves in right precordial leads, particularly V1 and V2, in addition to the absence of septal q waves from left oriented leads, and possibly to increase in magnitude of leftward and posteriorly oriented QRS forces (increased depth of S in right precordial leads) at 30 to 40 ms from the start of ventricular activation. In <xref ref-type="fig" rid="F2">Figure 2D</xref> is depicted what might be expected when left septal activation is blocked in a situation where for some reason right-to-left septal activation is attenuated or delayed (e.g., with some degree of right bundle branch block). In this case septal activation is significantly delayed and has to be initiated from depolarization waves reaching it from left ventricular myocardium which has been previously activated by anterior and posterior fascicles of the left bundle branch. Initial QRS forces will likely be directed leftward, with disappearance of septal q waves. Left-to-right and posterior-to-anterior septal activation will occur later than left ventricular free wall activation, and will be relatively unopposed by any right-to-left septal activation. Depending on the degree of this delay, this may produce only a decrease in depth of right precordial S waves, or might actually result in an anterior shift in mid-to-late QRS forces with an increase in their anterior magnitude, somewhat analogous to the left-superior and right-inferior mid-to-late QRS shifts seen respectively with left anterior and left posterior fascicular blocks. Unfortunately, in some cases this would be masked by the coexisting right bundle branch block (RBBB). Such prominent anterior forces, otherwise unexplained by right ventricular hypertrophy or posterior infarction, have been claimed in anecdotal cases to be the result of just such a LSFB [<xref ref-type="bibr" rid="R11">11</xref>,<xref ref-type="bibr" rid="R12">12</xref>,<xref ref-type="bibr" rid="R23">23</xref>-<xref ref-type="bibr" rid="R27">27</xref>]. That some of these were shown to be associated with RBBB when the conduction defect worsened is consistent with this hypothetical model [<xref ref-type="bibr" rid="R25">25</xref>,<xref ref-type="bibr" rid="R26">26</xref>]. A transient anterior shift of QRS forces has been seen during myocardial ischemia produced by occlusion of the anterior descending artery during balloon angioplasty, possibly due to block of right septal activation [<xref ref-type="bibr" rid="R43">43</xref>]. Other studies have shown that otherwise unexplained prominent anterior forces can be a normal variant without unfavorable clinical implications [<xref ref-type="bibr" rid="R44">44</xref>].</p></sec><sec id="s3"><title>Proposed Criteria for LSFB</title><p>In <xref ref-type="table" rid="T1">Table 1</xref> are listed proposed criteria for diagnosis of LSFB based on the above considerations. The major criterion is based on changes in the first part of the QRS: septal q waves are markedly diminished or lost by shift in initial QRS forces to the left, and in some cases also posteriorly due to loss or frank reversal of initial left-to-right septal activation [<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R29">29</xref>,<xref ref-type="bibr" rid="R32">32</xref>,<xref ref-type="bibr" rid="R34">34</xref>,<xref ref-type="bibr" rid="R46">46</xref>]. This means loss of q waves from I, V5-6, and/or from inferior leads in cases with vertical heart orientation. Q waves may develop in V1-2 especially when anteriorly directed forces of right ventricular free wall activation fail to overbalance the posteriorly directed forces of "right-to-left" septal activation. With intact right bundle branch conduction, this may produce qrS complexes in V1-2, but QS complexes can occur, which mimic the chronic picture of anterior infarction. It must also be differentiated from complete or incomplete LBBB, ventricular pre-excitation, congenitally corrected transposition of the great vessels, effects of left or right ventricular hypertrophy, and orthotopic cardiac transplantation among other conditions. As right septal activation still starts via the right bundle branch no more than 5 ms after the start of left ventricular activation, a moderate delay in activation of its left septal surface would not be expected to increase the QRS duration importantly. Mean frontal plane QRS axis and ventricular activation times in aVL, aVF, and V5-6 should remain normal, as activation of the left ventricular free wall and apex via the left anterior and posterior fascicles is undisturbed. Absence of slurring or delay in R wave upstroke in leads overlying the left ventricle helps to differentiate this conduction defect from various degrees of incomplete LBBB [<xref ref-type="bibr" rid="R46">46</xref>]. Prominent anterior shift in mean QRS vector, manifested as prominent R waves with increased R:S ratio in V1-2, can occur when right septal activation is blocked or otherwise limited, resulting in relatively late, predominantly left-to-right and posterior-to-anterior septal activation. This is more likely to be seen when there is coexistent delay of conduction in the right bundle branch.</p></sec><sec id="s4"><title>Illustrative Examples Of Probable And Possible LSFB</title><sec id="s4a"><title>Absent Septal Q Waves in Otherwise Normal ECGs: LSFB, Pure Septal Infarction, or a Normal Variant?</title><p>Absence of septal q waves in ECGs that are otherwise normal is the simplest way that LSFB might present. In prior work, comparison was made of 92 consecutive cases showing this pattern with age- and gender-matched "controls" with normal tracings including the presence of septal q waves [<xref ref-type="bibr" rid="R47">47</xref>]. No significant difference was found in incidence of overt cardiac disease. Moreover, this did not seem to be an acquired ECG pattern, as the age distribution in cases with normal ECGs except for absence of septal q waves was identical to that of a larger group with normal ECGs including presence of septal q waves. Thus the absence of septal q waves in ECG's that are otherwise completely normal in most instances probably represents a variant of normal found in about 7 percent of normal electrocardiograms. This is consistent with vectorcardiographic studies in normal populations [<xref ref-type="bibr" rid="R48">48</xref>-<xref ref-type="bibr" rid="R50">50</xref>]. However, the identical ECG pattern might result from myocardial infarction limited to the interventricular septum [<xref ref-type="bibr" rid="R51">51</xref>]. In this case, comparison with prior tracings and clinical correlation might be needed for certain differentiation.</p></sec><sec id="s4b"><title>LSFB without Other Fascicular Conduction Defects</title><p>However, such a pattern can result from an aberration in intraventricular conduction [<xref ref-type="bibr" rid="R52">52</xref>]. In <xref ref-type="fig" rid="F3">Figure 3</xref> such an example is shown in a 20 year old woman suffering an adult respiratory distress syndrome following her second stem cell transplant for recurrent acute myelogenous leukemia. She had no signs or symptoms of cardiovascular disease. In this case premature atrial beats were conducted with slight aberrancy, the major result of which was disappearance of q wave from leads II, III, aVF, and V4-V6, leaving septal q waves absent. QRS duration and frontal plane mean QRS axis were unchanged. There was some increase in amplitude of anterior QRS forces at about 30 ms. The aberrant conduction was probably due to LSFB, with that fascicle in this case having the longest refractory period of the four fascicles. Aberrant beats remained otherwise within normal limits, without secondary ST or T changes. Twelve minutes after this ECG she developed atrial fibrillation with a ventricular rate of 195 per minute; at that time all beats showed the same aberrant conduction as was seen only on premature beats in the preceding ECG. If such an aberrant form of intraventricular conduction were to become permanent, without other abnormalities, and without prior tracings, there would be no way to separate the resulting ECG from a normal variant as described in the preceding paragraph, or from the result of a pure septal infarction [<xref ref-type="bibr" rid="R51">51</xref>-<xref ref-type="bibr" rid="R56">56</xref>].</p><p>In my experience, it was not hard to find examples of this type of transient LSFB [<xref ref-type="bibr" rid="R52">52</xref>]. It was seen most commonly in subjects with atrial fibrillation, particularly with rapid ventricular rates. Short R-R intervals would result in disappearance of septal q waves, as shown in the example of <xref ref-type="fig" rid="F4">Figure 4</xref>. This was taken on a 69 year old woman who experienced paroxysmal atrial fibrillation without any prior cardiac history. Presence of prominent septal q waves in inferior leads and V5-V6 was seen in the second ECG taken 12 minutes later, after spontaneous conversion to sinus rhythm. An echocardiogram was normal except for presence of left atrial enlargement. Response of left ventricle to dobutamine stress was normal. Rate-related, transient loss of septal q waves in every beat was also encountered in some cases of supraventricular tachycardia or atrial flutter with 2:1 atrioventricular block [<xref ref-type="bibr" rid="R52">52</xref>]. </p><p><xref ref-type="fig" rid="F5">Figure 5</xref> shows an ECG during supraventricular tachycardia in a 27 year old woman complaining of recurrent palpitations. In every other beat, septal q waves are absent along with slight increase in anteriorly directed forces. This is a rate-related 2:1 LSFB, a form of electrical alternans. Following conversion to sinus rhythm, all beats had septal q waves. She had no clinical or echocardiographic evidence of structural heart disease. Subsequent electrophysiological study showed this to be an atrioventricular nodal re-entrant tachycardia of the usual variety. Attacks were abolished with radiofrequency catheter AV nodal modification.</p><p>In other cases the change in initial QRS direction in aberrantly conducted beats can vary with degree of prematurity of the ectopic atrial beats. <xref ref-type="fig" rid="F6">Figure 6</xref> is an electrocardiogram from an 75 year old man with severe disease of his peripheral, carotid and coronary arteries. His echocardiogram showed a left ventricle with inferior and posterobasal hypokinesis, and an ejection fraction of 45%. Some aberrantly conducted beats lacked septal q waves. In others septal q waves were markedly attenuated. With the most premature ectopic atrial beat which occurred while right precordial leads were being recorded, an initial q appeared in leads V1-V3, suggesting posterior shift of initial QRS forces, which was probably due to a greater degree of LSFB.</p><p>The example shown in <xref ref-type="fig" rid="F7">Figure 7</xref> was from a 90 year old woman with a past history of hypertension, paroxysmal atrial fibrillation, and Parkinsonism,. Her echocardiogram showed left atrial enlargement, but left ventricular size and wall motion were normal despite the presence of electrocardiographic voltage criteria for left ventricular hypertrophy (LVH). Atrial premature beats cause two distinct forms of aberrant QRS complexes. The first type was not associated with discernable change in direction of initial QRS forces, but was associated with development of marked left and superior axis deviation and new broad terminal S in I and slurred terminal R in aVR consistent with left anterior fascicular block (LAFB) combined with RBBB. The second type, in which the ectopic atrial beat occurred slightly earlier, was associated with loss of septal q waves from aVL, V5 and V6, loss of r waves in V1-V3, but no major change in frontal plane mean QRS axis or QRS duration. There was increase in depth of S waves in V1-V4, and in height of R waves in V5 and V6. It would be natural to claim that the latter type of aberration in <xref ref-type="fig" rid="F7">Figure 7</xref> was due to incomplete LBBB, but such is not likely in that the neither ventricular activation time nor total QRS duration was changed, nor was there slurring of R wave upstroke in V5 or V6 in aberrantly conducted beats. This is more likely another example of LSFB. The resulting QRS configuration mimics an anterior infarct, and augments the voltage signs of LVH. Diminution in amplitude or complete loss of septal q waves in the systolic overload pattern of LVH [<xref ref-type="bibr" rid="R57">57</xref>] has been ascribed to augmented left ventricular free wall forces that are not counterbalanced by left-to-right septal forces [<xref ref-type="bibr" rid="R58">58</xref>]. It is more likely that the loss of septal q waves in this situation is the result of a conduction defect, LSFB [<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R29">29</xref>]. This type of aberrancy has been produced with programmed premature electrical stimulation of the right atrium in subjects with known LVH [<xref ref-type="bibr" rid="R29">29</xref>], and has been reported after spontaneous early premature atrial beats [<xref ref-type="bibr" rid="R28">28</xref>]. This manifestation of LSFB may be a major cause of QS complexes that can occur in V1-V2 in some cases of severe LVH [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R29">29</xref>,<xref ref-type="bibr" rid="R59">59</xref>,<xref ref-type="bibr" rid="R60">60</xref>]. It could also contribute to the low specificity of QS complexes in V1-V2 and of absent septal q waves for diagnosing septal or anterior infarction [<xref ref-type="bibr" rid="R51">51</xref>,<xref ref-type="bibr" rid="R59">59</xref>-<xref ref-type="bibr" rid="R63">63</xref>].</p><p>The series of ECGs shown in <xref ref-type="fig" rid="F8">Figure 8</xref> probably also demonstrate this type of LSFB complicating aortic balloon valvuloplasty in a 17 year old boy with severe congenital aortic stenosis. Two preoperative tracings showed LVH with tiny q waves in leads I and V6, and initial r waves in V1-V2. Following the valvuloplasty procedure on 2/6/01, he lost his initial r waves in V1-V2, and q waves were no longer seen in I and V6 (although q waves persisted in inferior leads). QRS duration and ventricular activation time in V5 and V6 did not change. Postoperative troponin I levels peaked at 0.54 ng/ml (normal ≤ 0.1) with normal total creatine kinase (CK) and CK-MB values, so anterior or septal infarction was unlikely to have been the cause of the ECG change. No ST-T changes evolved. The fourth tracing in the series, taken 3 weeks later, showed return of initial r wave in aVL and V2, but QS remained in V1, and clear septal q waves had not returned to V6.</p></sec><sec id="s4c"><title>Combination of Left Septal Fascicular Block with Other Left Fascicular Blocks</title><p>The presence of four fascicles (three left and one right) rather than the traditional three, adds complexity to the possible combinations of fascicular blocks [<xref ref-type="bibr" rid="R9">9</xref>] In <xref ref-type="table" rid="T2">Table 2</xref> are listed the resulting 15 possibilities for mono- and poly-fascicular blocks. This listing would be even more complex if fascicular blocks were further subdivided into incomplete and complete varieties.</p><p>
<xref ref-type="fig" rid="F9">Figure 9</xref> shows three sequential ECGs taken on a patient during a bout of spontaneous angina. This 60 year old woman had mixed angina with a 50 percent diameter stenosis of her proximal anterior descending coronary artery, and an 80 percent ostial right coronary arterial stenosis. Her baseline ECG showed LAFB. On the day of the tracings in <xref ref-type="fig" rid="F9">Figure 9</xref> an episode of spontaneous angina occurred in her cardiologist's office. The three ECGs shown were recorded with the patient supine on an examining table, without moving the electrodes between tracings. As her angina was waxing, her ECG showed LAFB without q waves in lead I, with QS deflections in V1-V2, and ST segment depression that was downslanting in I, nearly horizontal in V6, and upsloping in V4-V5. After recovery from the attack, shown in the third tracing, a q wave in lead I, and initial r waves in V1-V2 had reappeared, and ST depression had disappeared, leaving her ECG exactly as it had been on contemporary but prior tracings. In retrospect, the transient ECG changes had also been associated with decrease in S wave amplitude in V2-V3. The initial ECG of this series looks like LAFB complicated by anterior infarction with lateral subendocardial ischemia. But the transient nature of the QS in right precordial leads and loss of q in lead I could be due to transient LSFB, a conclusion suggested by others describing similar cases [<xref ref-type="bibr" rid="R64">64</xref>,<xref ref-type="bibr" rid="R65">65</xref>]. One could also argue convincingly in this setting that the QRS changes were due to left septal "parietal" block, given the presence of myocardial ischemia at the time [<xref ref-type="bibr" rid="R33">33</xref>], or to severe ischemia causing septal muscle to be transiently electrically inert [<xref ref-type="bibr" rid="R28">28</xref>,<xref ref-type="bibr" rid="R66">66</xref>-<xref ref-type="bibr" rid="R68">68</xref>]. If the changes seen were due to LSFB, this case would illustrate that superimposed on a chronic LAFB, a transient LSFB can be associated with a leftward and sometimes also posterior rotation of initial QRS forces, as would be predicted from the hypothetical model described above. This raises the possibility that a leftward orientation of initial QRS forces in LAFB, deemed inconsistent with a diagnosis of LAFB by Rosenbaum et al. [<xref ref-type="bibr" rid="R69">69</xref>], but seen in a large fraction of cases of this conduction abnormality [<xref ref-type="bibr" rid="R70">70</xref>-<xref ref-type="bibr" rid="R73">73</xref>], could in some cases be due to LAFB combined with involvement of left septal fibers, as suggested by others [<xref ref-type="bibr" rid="R32">32</xref>,<xref ref-type="bibr" rid="R71">71</xref>], particularly when q waves are present in right precordial leads. By reference to <xref ref-type="fig" rid="F1">Figure 1</xref>, it is easily seen how a discrete lesion in the region of the left anterior fascicle could produce such a conduction defect, when the bulk of septal fibers of the left bundle branch originate as a branch of the left anterior fascicle.</p><p>
<xref ref-type="fig" rid="F10">Figure 10</xref> illustrates a case with a chronic pattern consistent with LAFB combined with LSFB in an 83 year old man with severe aortic stenosis, concentric left ventricular hypertrophy, but normal coronary arteries and left ventricular wall motion. Initial QRS forces were directed to the left, inferiorly and possibly posteriorly. QS deflections in V1-V2 mimicked anterior infarction, although there was no clinical or echocardiographic evidence of prior myocardial infarction.</p><p>I have yet to find examples that could convincingly be called left posterior fascicular block (LPFB) combined with LSFB, although such probably exist. Part of the problem is that septal q waves in left facing leads most commonly diminish or disappear when isolated LPFB develops, due to superior and usually leftward shift of initial QRS forces [<xref ref-type="bibr" rid="R2">2</xref>,<xref ref-type="bibr" rid="R9">9</xref>,<xref ref-type="bibr" rid="R16">16</xref>]. This means the addition of LSFB to existing LPFB might not cause any recognizable change in QRS configuration. One can speculate, however, that when LPFB is present, in the absence of intervening infarction, the diminution or disappearance of q waves from inferior leads and of r waves from right precordial leads might indicate the addition of LSFB.</p></sec><sec id="s4d"><title>LSFB Combined with RBBB with or without Other Left Fascicular Blocks</title><p>The presence of RBBB is a special case in the detection of coexisting LSFB. RBBB alone may cause slight decrease in amplitude of initial r waves in V1-2, but does not change the direction of initial QRS forces [<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R29">29</xref>,<xref ref-type="bibr" rid="R46">46</xref>,<xref ref-type="bibr" rid="R74">74</xref>-<xref ref-type="bibr" rid="R76">76</xref>]. Hence, septal q waves are preserved when RBBB occurs [<xref ref-type="bibr" rid="R46">46</xref>,<xref ref-type="bibr" rid="R74">74</xref>,<xref ref-type="bibr" rid="R76">76</xref>]. In the special case of RBBB combined with LSFB, both right ventricular and interventricular septal activation are dependent on delayed left-to-right and back-to-front septal activation via depolarization waves originating from myocardium activated originally through the left anterior and posterior fascicles. Assuming a block of the proximal right bundle branch, there will be no early right-to-left component to septal activation, which must thus occur asynchronously later than the start of depolarization of the free wall of the left ventricle, and is a necessary prelude to depolarization of the right ventricular free wall. Direction of initial QRS forces will depend on the orientation of net potentials generated by the start of left ventricular activation via the left anterior and left posterior fascicles. This might well vary from person to person, but will not be modified by early forces from right ventricular free wall activation. It probably would be directed to the left [<xref ref-type="bibr" rid="R9">9</xref>], but conceivably could point either anteriorly or posteriorly, superiorly or inferiorly. Thus septal q waves would likely be absent from left facing leads, which is not an expected effect of uncomplicated RBBB [<xref ref-type="bibr" rid="R46">46</xref>,<xref ref-type="bibr" rid="R74">74</xref>-<xref ref-type="bibr" rid="R76">76</xref>]. But the presence or absence of initial r waves in right precordial leads would be hard to predict. Their absence when RBBB exists would add weight to a presumptive diagnosis of coexistent LSFB, but their presence would not necessarily rule it out.</p><p>The direction of forces in the middle of the QRS in this setting would be deviated anteriorly by delayed and relatively unopposed left-to-right and back-to-front activation of the interventricular septum. But this would be hard to separate from the anterior and rightward terminal QRS forces due to delayed activation of the right ventricular free wall. Hence, RBBB combined with LSFB might look just like RBBB except for atypical alterations in the direction of initial QRS forces.</p><p>
<xref ref-type="fig" rid="F11">Figure 11</xref> shows ECGs from a 96 year old woman with asthma, hypertrophic cardiomyopathy, and moderate aortic stenosis. Diastolic thicknesses of interventricular septum and posterior left ventricular free wall in 1999 were 13 mm and 12 mm respectively. Systolic anterior motion of the mitral valve was seen intermittently and was associated with a moderate systolic pressure gradient in the aortic outflow tract. Left ventricular wall motion and ejection fraction were normal. She never had hypertension or a clinical myocardial infarction. The two tracings shown are examples of the numerous ECGs she had at two periods of her life. The first one dated in 1995 is typical of her ECGs prior to 1998. It showed high voltage of left ventricular hypertrophy and a typical RBBB. Initial QRS forces were directed to the right, anteriorly and slightly superiorly, resulting in clear septal q waves in left facing leads, and initial r waves in V1 and V2. All ECGs after 1998 were like the second one shown in figure 11 dated in 1999. Initial QRS forces were directed to the left, probably slightly posteriorly, but still slightly superiorly. Septal q waves had disappeared from left facing leads, and in right precordial leads initial r waves had been replaced by q waves. S waves in the mid-precordial leads were deeper. There were no significant changes in her echocardiograms between these two tracings. The change in direction of initial QRS forces responsible for these ECG changes is almost certainly due to the superimposition of LSFB on the RBBB [<xref ref-type="bibr" rid="R9">9</xref>].</p><p>In <xref ref-type="fig" rid="F12">Figure 12</xref> is shown the ECG from a 67 year old woman with a history of hypertension and chronic chest pain. Exercise myocardial perfusion scan was normal, with normal left ventricular wall motion and ejection fraction of 68 percent. Premature atrial beats were conducted with a consistent aberrancy showing shift in initial QRS forces to the left and inferiorly; tiny q waves replaced initial r waves in leads V2 and V3. In these beats there was also left and superior deviation of the middle portion of the QRS, and rightward and anterior deviation of slowly inscribed terminal forces. The aberrant beats were consistent with LAFB combined with RBBB. However the leftward shift of initial forces and resulting loss of q wave in lead I was slightly atypical for LAFB, raising the possibility that it was combined with some degree of LSFB. Development of tiny initial q waves in V2 and V3 can result from uncomplicated LAFB, supposedly due to markedly inferior orientation of initial forces in the face of a heart that lies low in the chest relative to right precordial electrode placement [<xref ref-type="bibr" rid="R5">5</xref>]. Such right precordial q waves are reportedly rarely seen in LAFB in young persons [<xref ref-type="bibr" rid="R4">4</xref>], increasing the likelihood that, when present in the face of initial QRS forces pointing leftward, they could result from a posterior direction of initial forces [<xref ref-type="bibr" rid="R4">4</xref>,<xref ref-type="bibr" rid="R77">77</xref>], related to coexistent LSFB.</p></sec></sec><sec sec-type="conclusions" id="s5"><title>Conclusions</title><p>That the examples illustrated above represent LSFB is predicated on the accuracy of the hypothetical models of altered ventricular activation shown in <xref ref-type="fig" rid="F2">Figure 2B</xref>-<xref ref-type="fig" rid="F2">2D</xref>, on which <xref ref-type="table" rid="T1">Table 1</xref> is based. Proof that this condition is met will require additional investigation, including correlation of surface ECG patterns with electrophysiological studies of interventricular septal activation in intact humans.</p><p>Intraventricular conduction abnormalities consistent with LSFB exist and seem to be more common than has been previously recognized. As mid-late QRS forces may be minimally affected by LSFB, its diagnosis requires particular attention to changes in the direction of the initial forces of ventricular activation. Mild degrees of LSFB can mimic a normal variation [<xref ref-type="bibr" rid="R52">52</xref>]. Thus, comparison of ECGs taken before and after its development are usually needed for diagnosis. When the only clue to infarction limited to the interventricular septum is disappearance of septal q waves [<xref ref-type="bibr" rid="R9">9</xref>], it would be difficult to determine whether their loss was due to the infarction itself or to LSFB caused by the infarction.</p><p>The various electrocardiographic presentations of LSFB can tie together an assortment of otherwise inadequately explained phenomena: 1) the loss of septal q waves by leftward shift of initial QRS forces, transiently following short R-R intervals in some cases, permanently in others; 2) loss of initial r waves in right precordial leads accompanied by loss of septal q waves (which implies posterior as well as leftward orientation of initial QRS forces), either transiently or permanently as can occur with severe LVH in the absence of incomplete or compete LBBB [<xref ref-type="bibr" rid="R59">59</xref>,<xref ref-type="bibr" rid="R60">60</xref>], and in some cases of LAFB or RBBB; 3) some cases of transient right precordial Q waves during myocardial ischemia in the absence of infarction; 4) chronic QS deflections in right precordial leads in the absence of LBBB or infarction in that area; 5) atypical cases of incomplete LBBB in which left ventricular activation time remains normal and the sense of initial septal activation may not be strictly right-to-left [<xref ref-type="bibr" rid="R38">38</xref>,<xref ref-type="bibr" rid="R78">78</xref>].</p><p>This study has not addressed many important questions about the potential clinical significance of LSFB. Does the alteration of activation pattern of the interventricular septum in LSFB produce recognizable changes in its pattern of contraction? A study of this might require an analysis of septal motion more sophisticated than is currently employed in clinical echocardiography or other imaging methods [<xref ref-type="bibr" rid="R79">79</xref>-<xref ref-type="bibr" rid="R82">82</xref>]. Are there prognostic implications to the presence of LSFB? If permanent, it certainly implies the presence of disease involving at least a portion of the left bundle branch system. However, It remains for further study to show whether it is a harbinger of more advanced disease of the conduction system or myocardium, whether there are histopathologic correlations to its presence, and whether it carries prognostic significance similar to that of LAFB or LPFB [<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R5">5</xref>].</p></sec>
|
Pacemaker Lead Displacement: Mechanisms And Management
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Fuertes</surname><given-names>Beatriz</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Toquero</surname><given-names>Jorge</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Arroyo-Espliguero</surname><given-names>Ramon</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Lozano</surname><given-names>Ignacio F</given-names></name><degrees>MD</degrees></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>Pacemaker lead displacements can be defined as any other pacemaker position change, whether the functionality of the pacemaker is affected or not. However, only those displacements that provoke a malfunction in the pacing system are clinically relevant. Chronologically speaking, there are early displacements, which occur within the first six weeks after implantation, and late displacements, after this period of time [<xref ref-type="bibr" rid="R1">1</xref>]. Early displacements are more frequent than late displacements and they usually affect atrial leads. The incidence of early displacements is 1% in VVI pacemakers and 5.2% in DDD pacemakers (3.8% of the cases affecting atrial leads and 1.4% ventricular leads). Acceptable displacement rates should probably be less than 1 percent for ventricular leads and no more than 2 to 3 percent for atrial leads. These values are higher in biventricular pacing devices, related to coronary sinus lead displacement. Early lead displacements are the most frequent cause of reintervention, involving atrial leads in the majority of cases. After the first six weeks, late displacements are remarkable and they can rarely be related to a specific event [<xref ref-type="bibr" rid="R2">2</xref>].</p><p>Long-term stability in epicardial systems, though it may seem different, is similar to that of endocardial systems [<xref ref-type="bibr" rid="R3">3</xref>]. The difference in the number of complications is not significant between both systems. Epicardial leads are implanted using median sternotomy, and sutured to the epicardial surface of the heart. Pacemaker generator is located in the abdominal cavity and the lead tunnelized through the anterior aspect of the diaphragm. Globally speaking, complications rate is estimated among 5% to 13%.</p><p>Among the complications that may arise after the implantation of a permanent endocardial or epicardial cardiac pacing device, or a cardioverter defibrillator, lead displacements, together with lead fractures or retractions, usually take place within the first months after implantation. On the other hand, faults in the isolating material, in the adapter or in the connectors may arise later.</p><p>Dislodgement has been classified as “macrodislodgement” and “micro-dislodgement”. Macrodislodgement is radiographically evident, microdislodgement is not, and they constitute a particular case. They are considered as minimal displacements in the lead tip, not evident in chest radiographs, but enough to produce an increase in capture threshold and eventually a loss of capture while keeping normal lead impedance values or changing the initial values minimally. This is due to the fact that current density, responsible for the myocardial depolarization provoked by the pacemaker, is directly proportional to the current intensity (amperage) and inversely proportional to the distribution area. Since the area size increases proportionally to the radius square, the density will decrease proportionally to that square, this one being the distance between the current distribution center and the myocardium to be depolarized, (that is to say, the sum of the lead radius and the thickness of the fibrosis caused by the lead).</p></sec><sec id="s2"><title>Etiology</title><p>Information about the causes of lead displacement is scarce and it is often difficult to relate lead displacements to a specific etiology. Among some of them, we may name the following:
<list list-type="simple"><list-item><label>Twiddler’s Syndrome</label><p>: a rare complication observed both in patients with implanted pacemakers and defibrillators. First described in 1968, refers to permanent malfunction of a pacemaker due to the patient’s manipulation of the pulse generator. The patient, inadvertently or deliberately, turns and rotates the generator on its long axis and, because of traction, causes the lead displacement. Most patients with such a problem are middle-aged obese women whose surgical pocket is larger than the size of the generator together with the presence of loose subcutaneous tissue [<xref ref-type="bibr" rid="R4">4</xref>-<xref ref-type="bibr" rid="R6">6</xref>]. Ipsilateral phrenic nerve can be stimulated, resulting in diaphragmatic pacing and sensation of abdominal pulsations. Even rhythmic arm twitching has been described [<xref ref-type="bibr" rid="R7">7</xref>], related to pacing the brachial plexus.</p></list-item><list-item><label>Reel’s Syndrome</label><p>: It is similar to the Twiddler’s Syndrome. In this case, the patient rotates the generator on its transverse axis rolling the lead around the generator and provoking a lead displacement. Chest radiography is crucial to diagnose this kind of complication [<xref ref-type="bibr" rid="R8">8</xref>].</p></list-item><list-item><label>Direct trauma over the system</label><p>: this may produce not only a lead or connection fracture but also a system displacement [<xref ref-type="bibr" rid="R9">9</xref>], leading to macro or micro-displacements. Intense respiratory therapy is one of the etiologies that may lead to micro-dislodgment, specifically “Clapping” [<xref ref-type="bibr" rid="R10">10</xref>,<xref ref-type="bibr" rid="R11">11</xref>]. It consists of manual application of rhythmic percussion to the patient’s chest wall to break up and detach the accumulated bronchial wall secretions and requires considerable energy. Other passive techniques, such as vibration or vibrating pressure are recommended instead.</p></list-item></list></p><p>Any significant trauma over the system, because of traction, may produce a micro-displacement resulting in discontinuous capture or even loss of capture with the programmed output voltage. When checking the device, sensing, lead impedance and battery status are usually correct. Leads appear well in place in chest radiographies making difficult the diagnosis of the problem. In some cases, however, the use of fluoroscopy will help us to see that the lead is free in the right ventricle.</p></sec><sec id="s3"><title>Diagnosis</title><p>In general, the first signs of pacemaker’s malfunction will show in hospital. Depending on the pacemaker-dependency of the patient and on the degree of loss of capture, moderate dizziness or lightheadedness episodes and even repeated syncope’s may be reported. Other times, the only symptom reported is that derived from extra cardiac pacing. Patients may experience discomfort in the chest area or have hiccups due to phoenix nerve stimulation, mainly on the right side because of anomalous position of the atrial lead. Hemidiaphragm contraction synchronized with pacemaker pacing may occur, evidenced by the palpation of the upper quadrants of the abdomen or even by mere eye inspection, which can be eliminated or reduced if pacemaker output is programmed to a lower value. Left intercostal pacing is always caused by right ventricle perforation by the lead [<xref ref-type="bibr" rid="R12">12</xref>].</p><p>There are many etiologies [<xref ref-type="bibr" rid="R13">13</xref>] on pacing failures in permanent pacing systems (<xref ref-type="table" rid="T1">Table 1</xref>) but many of them are out of the scope of this work and we will focus from now on those provoked by lead displacements.</p><p>The symptoms referred by the patient will give us the first approach to device malfunction. The most useful and cost-effective tools to detect pacemaker malfunctions related to lead displacement are the ECG and the chest radiography. After that, lead status can be established interrogating the device through the programmer, measuring lead impedance and determining the pacing threshold (<xref ref-type="table" rid="T2">Table 2</xref>).</p><p>The ECG constitutes a valuable tool when evaluating a patient with suspected lead displacement. Loss of capture with adequate sensing points to micro-dislodgment (<xref ref-type="fig" rid="F1">Figure 1</xref>), whereas combination of loss of capture and inadequate sensing suggest lead displacement (<xref ref-type="fig" rid="F2">Figure 2</xref>).</p><p>The ECG allows us to determine permanent or non-permanent capture failures and even the absence of pacing artifact. This finding opens a wide range of likely etiologies that can be limited using other complementary tests. When evaluating the chest X ray, it is crucial to have two views available, lateral and posteroanterior, to determine whether leads have been placed correctly (see <xref ref-type="fig" rid="F3">Figure 3</xref> and <xref ref-type="fig" rid="F4">4</xref>). Passive fixation atrial leads are usually placed in the right appendage, and its trabeculation is important to provide more stability. In special circumstances, it may be interesting to perform an atrial implantation far from the right ventricle, either on the free wall or the interatrial septum. This is used either to avoid interferences in the atrial channel, as it happens in far-field potentials in double-chamber defibrillators, or to reduce the incidence of atrial fibrillation. An active fixation lead is used in these situations. Passive fixation ventricular leads are placed in the apical area, known to be trabeculated and therefore, stable for lead implantation. Although ventricular leads have been exceptionally placed on the right ventricle outflow tract, the presence of a lead tip at that position indicates a clear displacement in most occasions. In posteroanterior chest radiographies, a lead should be placed inside the right ventricle oriented to the apex. In the lateral view, the apical placement should be verified with an anterior orientation. In case of frank displacements, the lead can be sometimes seen completely free in the ventricle, although that is not the case in most occasions. Micro-displacements as short as 1-2 mm cannot be seen in the X ray. It is often more difficult to see displacements in left ventricular pacing electrodes since there is no clear cut anatomical structure to define its position. It is then crucial to have a baseline radiography available in the period 24-48 hours after implantation to use as a reference for subsequent evaluations.</p><p> After the clinical evaluation and first complementary tests have lead us to suspect that a displacement has occurred, the next step is to interrogate the device: a displaced lead is characterized by an abnormal reduction in impedance values and a higher capture threshold. In micro-displacements, usually harder to detect, an increase in capture threshold and a normal impedance, or an unchanged impedance with respect to the previous one, can be observed. In some occasions, we may find important variations in capture threshold and pacing impedance values at different stages depending on the degree of myocardial contact of the displaced lead.</p></sec><sec id="s4"><title>Treatment</title><p>The approach to lead displacement is going to be different depending on time from implantation, patient clinical status, pacemaker dependency, lead displaced (atrial or ventricular, active or passive fixation) and degree of malfunction of the device. In early displacements, reopening the pouch and lead reposition are possible since the distal end of the lead has not been caught and fixed by the endocardial fibrous reaction [<xref ref-type="bibr" rid="R14">14</xref>], allowing its manipulation.</p><p>In late displacements, surgical repositioning is often not feasible. In these cases, a solution is to implant a new lead in the chamber in which displacement has occurred canceling the previous one. Lead repositioning via percutaneous access is a less aggressive option that has provided also good results in cases reported. This technique was initially described to reposition J-shaped passive fixation atrial leads [<xref ref-type="bibr" rid="R15">15</xref>]. A quadripolar catheter with detectable curve must be advanced up to the atrial lead through a femoral vein approach. Then, the femoral catheter is removed while performing a clock-wise turn at the same time. This allow us to pull the electrode curve until it is released and goes back to its normal J position contacting with the higher portion of the right atrium. After these proceedings, it must be checked whether operating parameters are normal.</p><p>Transvenous proceedings are easy, fast (average 15 min.) and provide good results (evaluated through verified stability after one to six months). Authors conclude that can be successfully carried out using a percutaneous catheter through the femoral vein. They are straightforward, safe and no new surgical intervention is needed, even in patients with dilated right atrium, a factor that may negatively affect lead stability. Adverse effects of this technique are derived from the femoral vein approach. For this reason, percutaneous lead reposition should always be considered in most patients before performing an open surgery intervention.</p><p>However, prevention and not treatment is the most important aspect when dealing with lead displacements. Several facts have to be taken into account in order to avoid lead displacements:
<list list-type="alpha-lower"><list-item><p>Surgical implantation must be adequate from a technical point of view, creating a small surgical pocket to avoid generator displacements that may put pressure on the lead and displace it. Lead fixation to vein access and generator suturing to the fascia will help to keep the system stable [<xref ref-type="bibr" rid="R16">16</xref>].</p></list-item><list-item><p>Using active fixation leads, particularly in those patients whose chambers could have no proper trabeculation as it is the case in dilated cardiomyopathy or atrial dilatation [<xref ref-type="bibr" rid="R17">17</xref>].</p></list-item><list-item><p>Always verify the position of the leads in the period 24-48h after implantation using chest X ray in posteroanterior and lateral views.</p></list-item><list-item><p>Consider subpectoral implantation in aged obese patients whose flabby subcutaneous tissue may allow for progressive generator displacement inside the subcutaneous pouch and eventually progressive lead displacement. Our group previously described [<xref ref-type="bibr" rid="R18">18</xref>]. the utility of placing the pulse generator beneath the pectoral muscle and securing it to the fascia, to avoid recurrent Twiddler’s Syndrome. Technology development and improvement as well as reduction in generators’ size and weight have contributed to reduce this complication.</p></list-item><list-item><p>Introduce the generator in a Dacron pouch: although it is not a generally accepted method, it should be taken into account in cases in which pouch erosion, migration or manipulation have been observed.</p></list-item></list></p></sec><sec sec-type="conclusions" id="s5"><title>Conclusion</title><p>In spite of its importance, studies on the incidence and handling of lead displacement are surprisingly scarce. From reported cases related to this problem and its etiologies, we can conclude that, even using state of the art electrode technology and the most adequate surgical technique, lead displacement is a complication that has to be kept in mind for an early approach whenever feasible. Clinical evaluation of the patient will suggest the diagnosis. Together with ECG and chest X ray, a correct diagnosis can be made in the majority of cases. Device interrogation through corresponding programmer is helpful in doubtful situations, as micro-dislodgements. Our best tool to prevent lead displacement is the use of careful and adequate surgical implantation techniques. Percutaneous techniques could also allow us to reposition leads in a less aggressive way maintaining an adequate success rate and reducing the number of complications. When this is not feasible, surgical approach is necessary to manage the problem.</p></sec>
|
Tilt Table Test: State of The Art
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Baron-Esquivias</surname><given-names>Gonzalo</given-names></name><degrees>MD, PhD</degrees><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Martinez-Rubio</surname><given-names>Antoni</given-names></name><degrees>MD, FESC, FACC</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<p>The loss of consciousness has been a subject of wonder and uncertainty in humans, and for this reason it has been the object of medical investigation since the beginning of time. Even actually, it is certainly an unresolved clinical problem. Many centuries ago, complicated exorcisms and remedies were used on these unfortunate patients, who upon regaining consciousness would find themselves soaked in miraculous liquids, ingesting curative potions, and often on the way to be burned accused of being possessed. In the seventeen century, physicians began to relate loss of consciousness and haemodinamic changes. William Harvey was perhaps the first to describe a circulatory response (vasovagal reaction) during a phlebotomy in the year 1628: “…Yet it fear or any other cause, or something do intervene through passion of the mind, so that the heart do beat more faintly, the blood will be no means pass through but drop after drop… ” [<xref ref-type="bibr" rid="R1">1</xref>]. During the nineteenth century, loss of consciousness was the object of studies and research, and the vagally mediated cardioinhibition, as a primary cause, was noted by Foster who proposed that profound bradycardia diminished cerebral perfusion to a level inadequate to maintain consciousness ” [<xref ref-type="bibr" rid="R2">2</xref>]. At this time, it was reported the first use of the tilt-table test ” [<xref ref-type="bibr" rid="R3">3</xref>]. Commonly referred to as fainting or loss of consciousness, from last century the preferred medical term is syncope, which itself is derived from the Greek term “syncoptein” meaning “to cut short”. Syncope is defined as the sudden loss of consciousness and postural tone with spontaneous recovery. In 1907, Gowers was the first person to use the term vasovagal syncope [<xref ref-type="bibr" rid="R4">4</xref>]. In 1918 was published the work in which Cotton and Lewis described for the first time the clinical characteristics that are still used today to recognize the syncopal reaction [<xref ref-type="bibr" rid="R5">5</xref>]. However, it was not till 1932 when Lewis described this reaction as being characterized by a combination of bradycardia, hypotension, and syncope, and he coined the term vasovagal syncope for the first time [<xref ref-type="bibr" rid="R6">6</xref>].</p><p>From the first studies with patients suffering loss of consciousness, it was observed that they all suffered a haemodynamic collapse which, though poorly understood initially, was analyzed with the diagnostic methods of the time period. In the past, those patients with recurrent syncopes were often subjected to a long, tedious, expensive, and all too often fruitless routine series of tests aimed at disclosing a potential aetiology. It was not until 1957 that one of the pioneering works on the use of the tilt-table test on such patients was published. In this work with university students, Weissler et al. tilted the individuals to 60° and observed 20% positive (abnormal) results. They induced syncope in others by oral administration of sodium nitrate 10-15 minutes before the test [<xref ref-type="bibr" rid="R7">7</xref>]. After studying patients with syncope of unknown aetiology and based on the known induction of a vasovagal response in young adults [<xref ref-type="bibr" rid="R8">8</xref>,<xref ref-type="bibr" rid="R9">9</xref>] as well as the results of aerospace studies [<xref ref-type="bibr" rid="R10">10</xref>-<xref ref-type="bibr" rid="R15">15</xref>], a group of English cardiologists conducted by Kenny and Sutton in 1986 published the first research paper on tilt testing [<xref ref-type="bibr" rid="R16">16</xref>]. In this Landmark study, they used the tilt-table test at an angle of 40° for 60 minutes on 15 patients with syncope of unknown aetiology, and on 10 control individuals. Nine of 15 patients (67%) and 1 control developed syncope during tilt.</p><p>From the time of this historic work to the present day, more than 830 papers have been published in which the keyword is <italic>vasovagal syncope</italic> and more than 975 with the term <italic>tilt-table test</italic>.</p><p>There is no question that this technique has helped for the understanding of the vasovagal syncope and its pathophysiology. However, many authors currently question this technique due to the many concerns that still need to be resolved:
<list list-type="order"><list-item><p>Are clinical and induced vasovagal syncope equivalent?</p></list-item><list-item><p>Is a tilt-table test indicated in all patients with vasovagal syncope?</p></list-item><list-item><p>Is there a standardized protocol?</p></list-item><list-item><p>What is the result of a tilt-table test?</p></list-item><list-item><p>Is the tilt-table test useful for anything other than diagnostics?</p></list-item><list-item><p>Exist complications of the test?</p></list-item></list></p><sec id="s1"><title>Correlation Between Clinical Syncope And Induced Syncope</title><p>In Medicine, many doubts exist regarding induced phenomena. Physicians should question the correlation between clinically spontaneous and induced results. Furthermore, is the clinical phenomenon reported by the patient equivalent to the induced? In the case of vasovagal syncope, it seems to be clear that they are equivalent. We can base this conclusion on three fundamental facts:
<list list-type="alpha-upper"><list-item><p>Spontaneous syncope and induced syncope in the tilt-table test are associated with similar premonitory signs and symptoms. The most characteristic premonitory symptoms that are almost always described by patients are nausea, redness, perspiration, and abdominal discomfort. Clinical signs include marked pallor, bilateral mydriasis, and loss of postural tone [<xref ref-type="bibr" rid="R17">17</xref>-<xref ref-type="bibr" rid="R19">19</xref>].</p></list-item><list-item><p>The temporal sequence of changes in blood pressure and heart rate during induced syncope parallel those seen with spontaneous syncope. Though earlier data from European studies placed doubt on this conclusion [<xref ref-type="bibr" rid="R20">20</xref>], it has been shown that in patients suffering from syncope, there was a demonstrable and reproducible neurovegetative imbalance [<xref ref-type="bibr" rid="R21">21</xref>,<xref ref-type="bibr" rid="R22">22</xref>]. This was manifested as an increase in central sympathetic stimulation preceding syncope [<xref ref-type="bibr" rid="R23">23</xref>] associated with a very acute decrease in cardiac output and a later drop in mean arterial pressure, as a result of the parasympathetic nervous activity and peripheral resistance [<xref ref-type="bibr" rid="R24">24</xref>] measured by means of microvascular Doppler with a decrease in subcutaneous blood flow [<xref ref-type="bibr" rid="R25">25</xref>].</p></list-item><list-item><p>Plasma catecholamine levels measured before and after spontaneous and induced syncope are very similar.Above all, an increase in circulating catecholamines has been shown both in spontaneous syncope in the upright position [<xref ref-type="bibr" rid="R26">26</xref>-<xref ref-type="bibr" rid="R29">29</xref>] and in induced syncope with the tilt-table test [<xref ref-type="bibr" rid="R30">30</xref>], with maintained levels of norepinephrine and a five-fold increase in epinephrine levels [<xref ref-type="bibr" rid="R31">31</xref>], and with intervention of endogenous opioids and differences between experimental protocols [<xref ref-type="bibr" rid="R32">32</xref>].</p></list-item></list></p></sec><sec id="s2"><title>Who Should Undergo Tilt-Table Testing?</title><p>The ACC expert consensus document for tilt-table testing was published in 1996 with this summary of principals indications [<xref ref-type="bibr" rid="R17">17</xref>] divided into three general categories depending its warranty:</p><p><italic>Tilt table testing is warranted:</italic>
<list list-type="order"><list-item><p>If recurrent syncope or single syncopal episode in a high risk patient have occurred, whether or not the medical history is suggestive of neurally mediated (vasovagal) origin, and</p><p>1.1: No evidence of structural cardiovascular disease.</p><p>1.2: Structural cardiovascular disease is present, but other causes of syncope have been excluded by appropriate testing.</p></list-item><list-item><p>For further evaluation of patients in whom an apparent cause has been established (e.g. asystole, atrioventricular block), but in whom demonstration of susceptibility to neurally mediated syncope would affect treatment plans.</p></list-item><list-item><p>Parts of the evaluation of exercise-induced or exercise-associated syncope.</p></list-item></list></p><p><italic>Reasonable differences of opinion exist regarding utility of tilt table testing for:</italic>
<list list-type="order"><list-item><p>Differentiating convulsive syncope from seizures.</p></list-item><list-item><p>Evaluating patients (especially the elderly) with recurrent unexplained falls.</p></list-item><list-item><p>Assessing recurrent dizziness or presyncope.</p></list-item><list-item><p>Evaluating unexplained syncope in the setting of peripheral neuropathies or dysautonomies.</p></list-item><list-item><p>Follow-up evaluation to asses therapy of neurally mediated syncope.</p></list-item></list></p><p><italic>Tilt table test is not warranted after:</italic>
<list list-type="order"><list-item><p>Single syncopal episode, without injury and not in a high risk setting with clear-cut vasovagal clinical features.</p></list-item><list-item><p>Syncope in which an alternative specific cause has been established and in which additional demonstration of a neurally mediated susceptibility would not alter treatment plans.</p></list-item></list></p><p><italic>Potential emerging indications are:</italic>
<list list-type="order"><list-item><p>Recurrent idiopathic vertigo</p></list-item><list-item><p>Recurrent transient ischemic attacks</p></list-item><list-item><p>Chronic fatigue syndrome</p></list-item><list-item><p>Sudden Infant Death Syndrome (SIDS)</p></list-item></list></p><p>Later, and following the same kind of recommendations (class I, II and III), the European Society of Cardiology [<xref ref-type="bibr" rid="R33">33</xref>] published its guidelines in 2001. They recommended:</p><sec id="s2a"><title>Indications</title><sec id="s2a1"><title>Class I:</title><p><italic>Tilt testing is indicated for diagnostic purposes:</italic>
<list list-type="order"><list-item><p>In cases of unexplained single syncopal episodes in high risk settings (e.g. occurrence of, or potential risk for, physical injury or with occupational implications), or recurrent episodes in the absence of organic heart disease, or, in the presence of organic heart disease, after cardiac causes of syncope have been excluded.</p></list-item><list-item><p>When it will be of clinical value to demonstrate susceptibility to neurally-mediated syncope to the patient.</p></list-item></list></p></sec><sec id="s2a2"><title>Class II:</title><p><italic>Tilt testing is indicated for diagnostic purposes:</italic>
<list list-type="order"><list-item><p>When an understanding of the haemodynamic pattern in syncope may alter the therapeutic approach.</p></list-item><list-item><p>For differentiating syncope with jerking movements from epilepsy.</p></list-item><list-item><p>For evaluating patients with recurrent unexplained falls.</p></list-item><list-item><p>For assessing pre-syncope or dizziness.</p></list-item></list></p></sec><sec id="s2a3"><title>Class III:</title><p><list list-type="order"><list-item><p>For assessment of treatment.</p></list-item><list-item><p>A single episode without injury and not in a high risk setting.</p></list-item></list></p></sec></sec><sec id="s2b"><title>Diagnosis</title><sec id="s2b1"><title>Class I:</title><p><list list-type="order"><list-item><p>In patients without structural heart disease, tilt testing can be considered diagnostic, and no further test needs to be performed when spontaneous syncope is reproduced.</p></list-item><list-item><p>In patients with structural heart disease, arrhythmias or other cardiac causes should be excluded prior to considering positive tilt test results as evidence suggesting neurally mediated syncope.</p></list-item></list></p></sec><sec id="s2b2"><title>Class II:</title><p> The clinical meaning of abnormal responses other than induction of syncope is unclear.</p></sec></sec></sec><sec id="s3"><title>Is There A Standarised Protocol?</title><p>Since the first published work [<xref ref-type="bibr" rid="R16">16</xref>], numerous papers have been written on the methodology of this test that have increased our understanding of the pathophysiology of vasovagal syncope. Various protocols have been proposed in the literature and are explained below:</p><sec id="s3a"><title>Basal or Westminster Protocol:</title><p>In 1991, a group at the Westminster Hospital of London published their methodology showing the greater specificity of the test performed on a table with footboard support at an angle of 60° for 45 minutes without intravenous cannulation or infusion of vasoactive substances. They described a sensitivity of 75% and a specificity of 93% [<xref ref-type="bibr" rid="R34">34</xref>]. Doubts exist about the duration that should be used during tilt testing. To address this question, Stein et al analyzed data from 11 published studies using this protocol and performed tilt testing on 213 patients for 30 to 60 minutes. They demonstrated that diagnostic accuracy is not greatly influenced by duration of the test after 30 minutes, varying from 60% to 84% [<xref ref-type="bibr" rid="R35">35</xref>]. Furthermore, this protocol has been used less and less, in spite of what was initially described, since the true sensitivity appears to be lower. In a study of more than 1,000 consecutive patients, our group found a sensitivity of 24.4% but with a significant decrease with age. Thus, the rate of positive responses in young subjects (<20 years old) was 41% but it decreased to only 19% in subjects older than 60 years (p<0.05) [<xref ref-type="bibr" rid="R36">36</xref>].</p></sec><sec id="s3b"><title>Isoprenaline Protocol:</title><p>Mainly proposed by North American authors, this method consists of an initial phase of tilting without drug administration and a later phase in which isoprenaline is administered by intravenous infusion at different rates, using an angle of 70-80°. Between the various suggested protocols using this drug, one of the most commonly used is the published by Almquist [<xref ref-type="bibr" rid="R37">37</xref>]. In this, following a basal phase of 10 minutes of tilting at 80°, progressive doses from 1 to 5 mg/min of isoprenaline are infused. This method results in a sensitivity of 87% and a specificity of 85%. Other authors have criticized this protocol reporting lower sensitivity (75%) and lack of specificity [<xref ref-type="bibr" rid="R38">38</xref>]. This, in addition to secondary effects later described, led to the use of lower doses. Morillo et al [<xref ref-type="bibr" rid="R39">39</xref>] advocate this, and achieved a sensitivity of 61% and a specificity of 93%. Other authors have compared high and low doses of this substance and achieved better results with low doses (2 mg/min), with a sensitivity of 64% and specificity of 88.9% [<xref ref-type="bibr" rid="R40">40</xref>]. With the intention to reduce the long necessary time for this technique, a protocol using this substance was recently published in which the basal phase is eliminated, using only the isoprenaline phase [<xref ref-type="bibr" rid="R41">41</xref>].</p></sec><sec id="s3c"><title>Edrophonium Protocol:</title><p>After a 45-minute basal phase, a 10 mg intravenous bolus of this substance has been used and tilting is prolonged for another 20 minutes [<xref ref-type="bibr" rid="R42">42</xref>]. These authors found a high rate of positive responses at two different angles (60° and 80°). Since this study used a femoral arterial catheter, the results may be questioned and certainly no other group has reported experience with this protocol.</p></sec><sec id="s3d"><title>Clomipramine Protocol:</title><p>A group in Greece recently reported the use of this protocol without a basal phase, infusing 5 mg of this substance intravenously, and holding the table for 20 minutes [<xref ref-type="bibr" rid="R43">43</xref>,<xref ref-type="bibr" rid="R44">44</xref>]. This protocol resulted in a sensitivity of 80% and a specificity of 95.4%.</p></sec><sec id="s3e"><title>Nitroglycerine Protocol (Italian Protocol):</title><p>In 1994 Raviele et al. studied in 40 patients with syncope and 25 controls with previous negative tilt-table test results, a new protocol at 80° using intravenous nitroglycerine. They reported a sensitivity of 53% and a specificity of 92% [<xref ref-type="bibr" rid="R45">45</xref>]. Several years later, this author used a 60° angle for 45 minutes followed by sublingual administration of 300 mg. nitroglycerine followed by 20 minutes of further tilt testing to analyze the response of 235 patients and 35 controls. This method yielded a sensitivity of 51% and a specificity of 94% [<xref ref-type="bibr" rid="R46">46</xref>]. Later works were published [<xref ref-type="bibr" rid="R47">47</xref>-<xref ref-type="bibr" rid="R51">51</xref>] until the definitive publication of the Italian protocol [<xref ref-type="bibr" rid="R52">52</xref>], which recommends 20 minutes of tilting at 60° without drugs and without cannulation of peripheral veins. This was followed by an infusion of 400 mg of nitroglycerine sprayed sublingually, followed by 15 minutes of tilting. This protocol yielded a sensitivity of 62% and a specificity of 92%. We recently reported our experience with this protocol based on data of 426 patients (sensitivity of 67.5%; specificity of 85.7%) [<xref ref-type="bibr" rid="R53">53</xref>].</p><p>In view of all these data, we may conclude that there is not a standardized protocol for tilt test. However, for selection of a tilt test protocol, one should take into consideration that while the protocol without drugs (Westminster) has shown the highest specificity, the low sensitivity of this method suggests as necessary to use a drug induction phase. North American authors have preferred to use isoprenaline while most Europeans use nitroglycerine. Both strategies are valid since when both protocols are compared (randomly) similar results are achieved [<xref ref-type="bibr" rid="R54">54</xref>,<xref ref-type="bibr" rid="R55">55</xref>].</p></sec><sec id="s3f"><title>General recommendations independent of protocol:</title><p>The test should be performed in a quiet room with the lights slightly dimmed. Although the broad majority of groups report to perform the test during the morning, obviously, it may be undertaken during afternoon or at the evening. However, circadian patterns might influence the results of the test. It is important that the patient has fasted at least 2 hours before the test. In order to reduce the probability of a vasovagal response to venous cannulation, it has been proposed that patients remain in a supine position for 20 to 45 minutes before the test. Based on specificity reduction observed with cannulation [<xref ref-type="bibr" rid="R56">56</xref>], some protocols avoid its use and thus, supine position should be reduced to 5 minutes in those cases. Arterial blood pressure should be measured continuously. It is, therefore, recommended to use beat-to-beat non-invasive blood pressure measurement systems. Intra-arterial cannulation has sometimes been used, but this can modify the specificity of the test in children and the elderly [<xref ref-type="bibr" rid="R33">33</xref>]. Intermittent measurements with a sphygmomanometer have been widely used. Although it is not the most desirable method, the price of continuous measurement devices makes it very commonly used in clinical practice. When performing the test, it should be possible to tilt the patient to the desired angle smoothly and rapidly, and it should also be possible to return quickly to a supine position (<10 sec.) when the test is terminated to avoid the consequences of a very prolonged loss of consciousness [<xref ref-type="bibr" rid="R33">33</xref>]. Our group has shown that the duration of time for changing to Trendelenburg position can have a negative influence on the duration of asystoles observed in some patients during the test (<xref ref-type="fig" rid="F1">Figure 1</xref>). For this reason, we recommend that the necessary time from head-up tilt to Trendelenburg should be as short as possible [<xref ref-type="bibr" rid="R57">57</xref>]. Tilt tables should be equipped with footboard supports. An experienced nurse or technician should be continuously present during the test. The need to have a physician continuously present during the test has not been well established. However, since the risk for the patient during the test is very low, it seams to us as sufficient that physicians should be nearby and immediately available if a problem arises.</p></sec></sec><sec id="s4"><title>What Are The Results Of A Tilt-Table Test?</title><p>When a patient undergoes a tilt-table test, two main responses may be observed. The first is a normal or negative response, which is defined as slight fluctuations in systolic and diastolic blood pressure and in heart rate, without any other abnormalities. The other response is the abnormal or positive response, which is characterized by a loss of consciousness following various haemodynamic patterns. Although vasovagal syncope is the diagnostic objective of the test, from the first studies it has been observed that there are others different positive responses (<xref ref-type="table" rid="T1">Table 1</xref>) [<xref ref-type="bibr" rid="R58">58</xref>].</p><p>Based on changes in arterial blood pressure and heart rate preceding the vasovagal reaction, different patterns have been recognized. Overall, two of these are seen most commonly [<xref ref-type="bibr" rid="R33">33</xref>]. The typical or classic pattern is characterized by a rapid and complete initial reflex phase of adaptation to the upright position with a stabilization of blood pressure and heart rate (which suggests normal baroreflex function) that lasts until the rapid onset of the vasovagal reaction (<xref ref-type="fig" rid="F2">Figure 2</xref>). Patients exhibiting this kind of pattern tend to be young and healthy. They have a history of multiple syncopal episodes over many years. In many cases the first syncopal episode occurred at the age of 10 to 15 years. Secondary trauma is uncommon in those patients. This pattern represents a hypersensitive autonomic nervous system that responds excessively to various stimuli. On the other hand, a different pattern is often observed which is characterized by an inability to achieve adaptation to the upright position. This results in a progressive drop in arterial blood pressure and heart rate, which occurs until symptoms appear. This is the so-called hypotensive pattern, in which the cause of symptoms appears to be the inability to adapt to external influences (hyposensitive autonomic nervous system function). Patients who show this kind of response tend to be elderly and have (many) associated diseases. They have a short history of syncope with few episodes per patient. The syncopal episodes begin later in life, suggesting that they occur as a result of some underlying progressive dysfunction. This pattern is similar to the pattern observed in patients with autonomic nervous system dysfunction and suggests the existence of a superimposition of the typical vasovagal syncope and other more complex pathological conditions of the autonomic nervous system. The tilt-table test can be useful in differentiating between these two syndromes.</p><p>The typical or classic positive response has been initially classified as vasodepressor, cardioinhibitory, and mixed. The occasional appearance of asystoles during this typical response prompted to Sutton et al in 1992 to use the details of the haemodynamic response to the test to propose a classification scheme which has been recently modified and it has become an easy and useful tool [<xref ref-type="bibr" rid="R59">59</xref>,<xref ref-type="bibr" rid="R60">60</xref>] (<xref ref-type="table" rid="T2">Table 2</xref>).</p><p>The decision of when to terminate the test will influence the type of response that is observed. For this reason, it is recommended that in order to ensure correct classification of the response, the test should be terminated at the same moment that consciousness and postural tone are lost (when syncope occurs). A premature interruption will result in underestimation and a delayed termination will result in an overestimation of the cardioinhibitory response and will expose the patient to the consequences of a prolonged loss of consciousness. However, there has not been a consensus on this subject, and many physicians feel that it is sufficient to terminate the test when a sustained drop in blood pressure is observed simultaneously with symptoms.</p></sec><sec id="s5"><title>Is The Tilt Test Useful For Anything Other Than Diagnostics?</title><p>Evaluation of the response to treatment: Many authors have attempted to show the benefits of the various proposed treatments for vasovagal syncope evaluating their ability to produce a negative result on a patient who had previously presented a positive response to the test (ACC Class II recommendation in 1996 [<xref ref-type="bibr" rid="R17">17</xref>]. However, this has been criticized [<xref ref-type="bibr" rid="R33">33</xref>,<xref ref-type="bibr" rid="R61">61</xref>,<xref ref-type="bibr" rid="R62">62</xref>]. In order to use the test for evaluation of therapeutic options, two conditions need to be fulfilled: the test and a positive response to the test must be highly reproducible and must have high predictive value of later outcome. Reproducibility of the tilt-table test has been widely studied. Overall, reproducibility of an initial negative response (85%-94%) is greater than it is for a positive response (31%-92%). Furthermore, data from controlled studies show that approximately 50% of patients with a positive basal test result will present a negative test if it is repeated under treatment or with a placebo [<xref ref-type="bibr" rid="R63">63</xref>]. Additionally, short-term studies are not predictive of the long-term outcome under therapy (such as definitive stimulation). Thus, actual data show that the use of this test to evaluate the effectiveness of different treatments is subject to significant limitations [<xref ref-type="bibr" rid="R33">33</xref>].</p><sec id="s5a"><title>Usefulness as therapy:</title><p>Based on the above mentioned tendency of response negativization after repeated test, and due to the absence of any fully effective treatment for vasovagal syncope, various authors have proposed the use of the tilt-table test as a form of therapy for vasovagal syncope, the so-called “tilt training” [<xref ref-type="bibr" rid="R64">64</xref>,<xref ref-type="bibr" rid="R65">65</xref>]. These authors propose the use of the upright position supported against the wall for short and repeated time periods on successive days to train the body to withstand the upright position, thus significantly reducing the recurrence of syncope.</p></sec></sec><sec id="s6"><title>Complications</title><p>The tilt-table test is a safe procedure, and the occurrence of complications is very low. Although extremely prolonged asystolic pauses have been reported, the occurrence of these asystoles during a positive response cannot be considered a complication since this is an end-point of the test [<xref ref-type="bibr" rid="R57">57</xref>,<xref ref-type="bibr" rid="R60">60</xref>]. A rapid return to a supine position (or better to a Trendelemburg position), as soon as syncope occurs, is usually enough to prevent or limit the consequences of a prolonged loss of consciousness, however brief resuscitation maneuvers are occasionally required [<xref ref-type="bibr" rid="R66">66</xref>]. Complications have not been reported in relation to the Westminster protocol or with the use of nitroglycerine, but minor side effects occur commonly and include palpitations with the use of isoprenaline and headaches with nitroglycerine. However, using isoprenaline protocols, high doses of the drug are well tolerated in only 62% of patients, must be reduced in 33% and discontinued in 4.2%. Side effects include tachycardia (45%), nausea (35%), chest pain (2.2%), arrhythmias (6%), and other effects (10.3%) [<xref ref-type="bibr" rid="R67">67</xref>]. Vasospasm has been described several times with the use of this substance [<xref ref-type="bibr" rid="R68">68</xref>]. Life-threatening ventricular arrhythmias have been described with isoprenaline in the presence of ischemic heart disease or sinus dysfunction. Atrial fibrillation can be induced during or after a positive tilt test and is usually self limiting.</p></sec><sec sec-type="conclusions" id="s7"><title>Conclusions</title><p>Vasovagal syncope is the most frequent form of syncope of unknown aetiology. Since 1986, the tilt-table test has become an important tool in the diagnosis of vasovagal syncope. Before this test was introduced, several patients were submitted to a multitude of tests although most of them did little more than waste of resources. The similarity between clinical and induced vasovagal syncope during the test shows its value. The tilt-table test yields a range of positive results from 60% to 70%, with a specificity greater than 85%. The results are also highly reproducible (data similar to provocation diagnostic tests such as the treadmill test) when drug-based protocols are used. This, with the absence of serious complications and very small number of side effects, makes probably the nitroglycerine protocol as the actually most recommended test for clinical practice.</p></sec>
|
<italic>To Pace Or Not To Pace!</italic> — Prevention Of Atrial Fibrillation After Coronary Artery Bypass Surgery
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Vora</surname><given-names>Amit</given-names></name><degrees>MD, DM</degrees></contrib><aff>Consultant cardiologist and Electrophysiologist, Mumbai</aff>
|
Indian Pacing and Electrophysiology Journal
|
<p>Atrial fibrillation (AF) is a very undesirable, but unfortunately a common arrhythmia following coronary artery bypass graft (CABG) surgery, occurring in up to 40% of patients [<xref ref-type="bibr" rid="R1">1</xref>]. There is an increase in hospital stay [<xref ref-type="bibr" rid="R2">2</xref>] and adds to the overall cost of the surgery. Atrial fibrillation occurrence may identify a subset of patients with reduced survival [<xref ref-type="bibr" rid="R3">3</xref>]. Prevention of AF therefore would have a significant positive impact on patients undergoing CABG surgery.</p><p>The pathophysiology of post-operative AF is incompletely understood and appears multi-factorial. There are studies to suggest clinical factors like advanced age, prior history of AF [<xref ref-type="bibr" rid="R4">4</xref>], post-operative withdrawal of beta-blocker or an angiotensin-converting enzyme (ACE) inhibitor, chronic obstructive lung disease predicting post-operative AF [<xref ref-type="bibr" rid="R5">5</xref>]. P-wave duration, increase in P wave dispersion [<xref ref-type="bibr" rid="R6">6</xref>], postoperative low cardiac output, preoperative larger left atrium (LA) and LA appendage area, a lower LA ejection fraction, post-bypass atrial systolic dysfunction and abnormal relaxation of the left ventricle, higher preoperative heart rate, increased pulmonary capillary wedge pressure [<xref ref-type="bibr" rid="R7">7</xref>], also predicts postoperative AF. It has been implicated that endogenously released adenosine has a role for triggering early (< 48 hr) post-CABG AF [<xref ref-type="bibr" rid="R8">8</xref>]. Another issue that received attention was the role of on-pump or off-pump bypass surgery, wherein no significant influence of either on the incidence of postoperative AF could be definitively established [<xref ref-type="bibr" rid="R9">9</xref>]. There are two opinions about the change in the autonomic tone brought about by dissection of the anterior epicardial fat or the aortic fat pad influencing the incidence of post CABG AF [<xref ref-type="bibr" rid="R10">10</xref>].</p><p>It is crucial to identify who would be at risk of developing AF, post CABG surgery, so as to institute appropriate preventive measures. Pharmacological approach has been widely used wherein preoperative initiation of beta-blocker therapy like oral metoprolol and continuation of the same postoperatively has been very effective in preventing AF [<xref ref-type="bibr" rid="R11">11</xref>]. Amiodarone, sotalol and propafenone have also been studied. Atrial Fibrillation Suppression Trial (AFIST) [<xref ref-type="bibr" rid="R12">12</xref>] showed that oral amiodarone prophylaxis in combination with beta-blockers prevents atrial fibrillation and reduces the risk of cerebrovascular accidents. Similarly in patients with chronic obstructive pulmonary disease, where beta-blockers would be contraindicated, early prophylactic amiodarone was found to reduce AF incidence [<xref ref-type="bibr" rid="R13">13</xref>]. Amiodarone and sotalol were found to have similar efficacy and safety in reducing postoperative AF. However, in patients undergoing more complex surgery, postoperative AF occurred more frequently with sotalol than with amiodarone [<xref ref-type="bibr" rid="R14">14</xref>].</p><p>Based on the mechanism of postoperative AF, it seems likely that overcoming the slow atrial conduction with reduction in the dispersion of atrial refractoriness and suppression of the atrial ectopy should prevent AF. With these considerations atrial overdrive pacing to prevent post CABG AF has been evaluated with a number of randomized, controlled trials. These trials compared controls with either single site atrial pacing i.e. right atrial pacing (RAP) or left atrial pacing (LAP) and biatrial pacing (BAP) or only RAP and BAP or only LAP and BAP. Bachmann’s bundle pacing as an alternative to two sites, biatrial pacing has also been studied. Different pacing modalities (AAT, AAI, DDD, atrial overdrive algorithms) and different stimulation rates are initiated after the surgery and continued for a period of 3-4 days. Earlier studies with pacing highlighted the limitations including sensing and capture problems, increase in atrial ectopy with demand (AAI) pacing and reported no advantage with pacing. However, most recent studies seem to favor prophylactic biatrial pacing to reduce the incidence of post CABG AF [<xref ref-type="bibr" rid="R15">15</xref>]. Two studies, one by Goette et al [<xref ref-type="bibr" rid="R16">16</xref>] and the other, AFIST II [<xref ref-type="bibr" rid="R17">17</xref>] show no impact on post-operative AF. Goette et al studied the role of Bachmann's bundle (BB) pacing, and required early termination of pacing in 16% of patients because of rise in thresholds. The AFIST II evaluated the prophylactic use of a hybrid intravenous and oral amiodarone regimen, atrial septal pacing, or both strategies in post-cardiothoracic surgery (inclusive of valve surgeries) patients. A hybrid intravenous and oral amiodarone regimen was effective at decreasing the incidence of post-operative AF, but atrial septal pacing was found to be ineffective. However, patients receiving both amiodarone and pacing had significantly lower AF rates than those receiving placebo with or without pacing.</p><p>The present study involving 120 patients by Massoud E. et al [<xref ref-type="bibr" rid="R18">18</xref>] in the present issue is one more study evaluating the role of biatrial pacing in prevention of AF after CABG surgery. This is a prospective, randomized, double blinded-placebo controlled study comparing biatrial pacing with LA pacing and no pacing. The baseline characteristics especially age, LV function, use of beta-blockers, perfusion time and cross-clamp time were well matched in the three groups. Atrial fibrillation occurred in 17.5% patients with bi-atrial pacing, compared to 30% in the LA pacing (p=0.04) and 45% in the control group (p=0.02). Consequently the duration of stay in the intensive care unit and the total hospital stay was significantly reduced in the biatrial-pacing group. The complications were no different between the groups. The time to occurrence and duration of AF was also similar in the three groups. They conclude two sites, biatrial pacing as well tolerated and effective in preventing AF and thereby shorten hospital stay.</p><p>It seems use of oral beta-blockers preoperatively and continuing the same postoperatively definitely prevents AF occurrences in the post-operative patients. Amiodarone therapy is especially rewarding when a hybrid intravenous and oral loading regime is employed as suggested by AFIST II trial. A meta-analysis of smaller pacing studies support the role of biatrial pacing. Few questions remain unanswered. Is pacing a cost-effective strategy; should it be prophylactically used in all patients; is there at all a role of single chamber (RA or LA) pacing and single site biatrial pacing (Bachmann bundle or atrial septal pacing)? A larger, multi-center study with a cost effective analysis, over and above the use of beta-blocker and amiodarone is likely to settle the role of prophylactic pacing. Till such time we need to pre-select the high-risk group and offer biatrial pacing.</p>
|
Role of Implantable Cardioverter Defibrillators in the Treatment of Hypertrophic Cardiomyopathy
|
<p>Hypertrophic cardiomyopathy (HCM) is an important cardiovascular disease with sudden cardiac death as the most devastating presentation. Implantable cardioverter defibrillators (ICD) are the optimal therapy for prevention of sudden death from ventricular tachycardia or fibrillation of any cause. While there is no controversy with implanting ICDs in patients who have already survived a cardiac arrest, identifying high-risk patients for primary prevention in this disease remains a challenge. Implanting ICDs in patients with HCM is an important clinical consideration since many individuals could achieve normal or near-normal lifespans with this protection.</p>
|
<contrib contrib-type="author"><name><surname>Ghosh</surname><given-names>Joydeep</given-names></name><degrees>MD, FACC</degrees><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Francis</surname><given-names>Johnson</given-names></name><degrees>MBBS, MD, DM</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Maron</surname><given-names>Barry J</given-names></name><degrees>MD, FACC</degrees><xref ref-type="aff" rid="aff3">‡</xref></contrib>
|
Indian Pacing and Electrophysiology Journal
|
<p>Hypertrophic cardiomyopathy (HCM) is the most common genetically inherited cardiac disorder with an estimated prevalence of 1 in 500 in the general population [<xref ref-type="bibr" rid="R1">1</xref>]. The disease is characterised by marked heterogeneity with respect to clinical manifestations, natural history and prognosis. Sudden cardiac death (SCD) is the most the devastating consequence of the disease and HCM is now recognized as most frequent cause of sudden cardiac death in pre-adolescent and adolescent children, as well as young athletes. This is not to suggest, however that patients who have reached midlife or even beyond are immune from SCD; in fact, no age is considered safe from this point-of-view [<xref ref-type="bibr" rid="R2">2</xref>,<xref ref-type="bibr" rid="R3">3</xref>].</p><p>The causes of sudden death in HCM remain controversial. While earlier theories outflow-tract obstruction, addressedthe current consensus favors primary ventricular arrhythmias, while recognising that the pathogenesis of such arrhythmias is multifactorial and involve complex interactions between a combination of factors. Given the high liklihood of prevention of sudden death from ventricular tachycardia or ventricular fibrillation by implantable cardioverter-defibrillators (ICD) in multiple primary and secondary prevention trials of ischemic and non-ischemic dilated cardiomyopathy, it is a reasonable expectation that such results can be replicated in patients with HCM. The difference between HCM and coronary artery disease is that risk-stratification in HCM with such variable natural history is exceedingly difficult and there exists no definitive consensus on precisely which patients should receive an ICD for primary prevention. On the other hand the ICD is strongly indicated for any patient, including those with HCM, who has survived a documented cardiac arrest i.e., secondary prevention [<xref ref-type="bibr" rid="R4">4</xref>, <xref ref-type="bibr" rid="R5">5</xref>].</p><p>It is important to remember that unlike other patients with ischemic or non-ischemic cardiomyopathy and ICD implants who have a limited life-span owing to advanced congestive heart failure and age, high risk HCM patients are usually younger and asymptomatic. Consequently, the potential of the ICD to prolong life is much greater in the HCM population, and therefore it is vitally important that patients be identified correctly for this therapy. In order to create a high risk profile, all HCM patients should undergo an initial comprehensive ambulatory risk stratification assessment, including detailed personal and family history and physical examination, 12-lead ECG, echocardiogram, ambulatory Holter monitoring and exercise testing. The currently recognised major risk factors for SCD in HCM include: unexplained syncope, (particularly when exertional or recurrent), family history of HCM-related sudden death, identification of high-risk mutant genes, frequent, multiple or prolonged episodes of non-sustained ventricular tachycardia on Holter monitoring, abnormal blood pressure response to exercise and extreme degrees of left ventricular hypertrophy (maximum left ventricular wall thickness of 30 mm or more), particularly in adolescents and young adults [<xref ref-type="bibr" rid="R6">6</xref>]. Invasive electrophysiology study as a means of risk stratification has been largely abandoned in HCM patients, primarily because of the unknown clinical significance of commonly induced ventricular tachyarrhythmias in the laboratory.</p><p>Controversy exists as to how many major risk factors are required in a particular patient to be considered deserving of an ICD, and some of this is uncertainty is due to the fact that the magnitude of risk conferred by each risk marker may not be the same. While the presence of multiple risk factors may convey an increased risk, a single risk factor may be adequate in a given patient to justify an ICD implant. ICD trials in HCM have documented their efficacy for sensing and terminating arrhythmias. A recent study by Maron, et al. showed appropriate discharges triggered by VT in 23% of patients when followed over 3 years, with an average discharge rate of 5% per year for primary prevention and 11% per year for secondary prevention. About 60% of the patients with appropriate interventions received multiple discharges [<xref ref-type="bibr" rid="R7">7</xref>].</p><p>There are certain technical considerations for the electrophysiologist implanting ICD's in this population that deserve mention. The first relates to whether a single- or dual-chamber system should be employed. Dual chamber devices have the benefit of AV synchrony with pacing, although for the most part pacing as a treatment modality in HCM to reduce symptoms and outflow gradient was largely abandoned after the M-PATHY trial [<xref ref-type="bibr" rid="R8">8</xref>]. More importantly, dual chamber devices provide superior SVT discrimination. The latter is particularly important in this population which has a high incidence of atrial fibrillation, though newer technology (for morphology, stability and onset discrimination) may provide relatively good discrimination even with single lead systems. The single-chamber “shock box” device is probably a more appropriate option for young patients who do not appear susceptible to atrial fibrillation and in whom survival depends largely on effective ventricular arrhythmia termination. A single chamber ICD is also indicated in all patients with chronic atrial fibrillation, while patients with paroxysmal atrial fibrillation would probably benefit more from dual chamber devices with mode-switch capability. It should be emphasized that atrial fibrillation may be poorly tolerated in HCM patients who already have severe diastolic dysfunction since LV filling is critically dependent on effective atrial systole. A dual chamber device with rate-drop response is probably a better option for a HCM patient with recurrent episodes of neurocardiogenic syncope, although trials of pacing for neurocardiogenic syncope have so far provided ambiguous results. An excellent update on selection of ICDs for each patient has been published recently [<xref ref-type="bibr" rid="R9">9</xref>].</p><p>Lead fractures and device recalls are particularly relevant events that affect young HCM patients who will likely spend their whole lives with the ICD system. Unfortunately, these are impossible to predict in advance, and lead removal, in particular for older leads, is fraught with difficulties [<xref ref-type="bibr" rid="R10">10</xref>]. Whenever possible, adding a new lead should be preferred to extracting an old one. Though active fixation leads that are less than 6-months old are usually removed without difficulty, passive leads require laser excision and should only be performed in a center with considerable experience with such procedures.</p><p>The number of HCM patients who meet criteria for an ICD implant is increasing. It behoves the cardiovascular community to offer these patients all major advances in treatment so that they can continue to lead productive lives, given their preserved systolic function and often minimum symptoms. Many of the current uncertainties about patient selection will be resolved soon with further clinical trials, and in the future more HCM patients will live longer lives with the aid of more technologically advanced ICD systems.</p>
|
Atrial Fibrillation and Pacing Algorithms
|
<p>Pacing prevention algorithms have been introduced in order to maximize the benefits of atrial pacing in atrial fibrillation prevention. It has been demonstrated that algorithms actually keep overdrive atrial pacing, reduce atrial premature contractions, and prevent short-long atrial cycle phenomenon, with good patient tolerance. However, clinical studies showed inconsistent benefits on clinical endpoints such as atrial fibrillation burden. Factors which may be responsible for neutral results include an already high atrial pacing percentage in conventional DDDR, non-optimal atrial pacing site and deleterious effects of high percentages of apical ventricular pacing. Atrial antitachycardia pacing (ATP) therapies are effective in treating spontaneous atrial tachyarrhythmias, mainly when delivered early after arrhythmia onset and/or on slower tachycardias. Effective ATP therapies may reduce atrial fibrillation burden, but conflicting evidence does exist as regards this issue, probably because current clinical studies may be underpowered to detect such an efficacy. Wide application of atrial ATP may reduce the need for hospitalizations and electrical cardioversions and favorably impact on quality of life. Consistent monitoring of atrial and ventricular rhythm as well as that of ATP effectiveness may be extremely useful for optimizing device programming and pharmacological therapy.</p>
|
<contrib contrib-type="author"><name><surname>Terranova</surname><given-names>Paolo</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff1">*</xref><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Severgnini</surname><given-names>Barbara</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Valli</surname><given-names>Paolo</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Dell'Orto</surname><given-names>Simonetta</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Greco</surname><given-names>Enrico Maria</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><aff id="aff1"><label>*</label>U.O. Cardiologia e UTIC, Azienda Ospedaliera "S. Paolo" - Polo Universitario, Department of Medicine, Surgery and Odontoiatry, University of Milan, Italy</aff><aff id="aff2"><label>†</label>U. O. di Cardiologia, Presidio Ospedaliero "Causa Pia Ospedaliera Uboldo", Cernusco sul Naviglio, Azienda Ospedaliera di Melegnano, Milano, Italy</aff>
|
Indian Pacing and Electrophysiology Journal
|
<sec sec-type="" id="s1"><title>Introduction</title><p>Atrial fibrillation is surely the most common arrhythmia in clinical practice and nowadays its incidence is increasing mainly due to the progressive ageing of population. Atrial fibrillation represents also the cardiac rhythm disorder that causes the highest number of hospitalizations [<xref ref-type="bibr" rid="R1">1</xref>] and is associated with higher mortality [<xref ref-type="bibr" rid="R2">2</xref>], major clinical complications such as heart failure, acute myocardial infarction, stroke [<xref ref-type="bibr" rid="R3">3</xref>], and impaired quality of life [<xref ref-type="bibr" rid="R4">4</xref>]. Atrial fibrillation is frequently associated with ventricular tachyarrhythmias. It has been calculated that 20% of the patients with conventional indication for cardioverter defibrillator implantation had atrial fibrillation before implant and that during the lifespan of the device more than 50% may develop atrial fibrillation [<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R6">6</xref>]. Antiarrhythmic drugs have been widely used for cardioversion and maintenance of sinus rhythm, but they showed a limited and usually temporary efficacy [<xref ref-type="bibr" rid="R7">7</xref>]. As a consequence, several non-pharmacological therapies, including physiological pacing, pacing prevention algorithms, anti-tachycardia pacing therapies and low-energy internal cardioversion have been introduced to treat drug refractory patients and have been implemented in multifunction implantable devices [<xref ref-type="bibr" rid="R8">8</xref>].</p></sec><sec sec-type="" id="s2"><title>Pacing prevention algorithms</title><p>Antiarrhythmic benefits of atrial and dual-chamber pacing versus single-chamber ventricular pacing in reducing atrial fibrillation recurrences and in preventing progression to permanent atrial fibrillation have been clearly demonstrated in large prospective trials enrolling either sinus node disease patients or unselected patients candidate for pacemaker implantation [<xref ref-type="bibr" rid="R9">9</xref>-<xref ref-type="bibr" rid="R11">11</xref>]. It has also been suggested that availability of rate responsiveness may maximize the effect of physiological pacing [<xref ref-type="bibr" rid="R12">12</xref>]. Pacing prevention algorithms have been introduced in order to maximize the preventive benefits of atrial pacing. Specific mechanisms involved in atrial fibrillation prevention by pacing algorithms include premature atrial contraction suppression or conditioning, with fewer chances to initiate atrial fibrillation, short-long atrial cycle prevention and maintenance of a high degree of exit block from all natural subsidiary atrial pacemakers. Several algorithms have been introduced by the different manufacturers during the last years. To summarize, they may be classified as follows: 1) dynamic sinus rhythm overdrive, 2) premature atrial beat reaction (short-long cycle prevention and ectopy overdrive), 3) post-tachycardia overdrive to prevent early recurrence of atrial fibrillation, and 4) prevention of inappropriate rate fall after exercise. The atrial pacing preference algorithm was the first evaluated, before being implemented in market released devices, as a temporary software, named "consistent atrial pacing", which could be downloaded into a conventional dual-chamber pacemaker via telemetry using a custom research telemetry device. It included a diagnostic which could be interpreted by a special Microsoft Excel spreadsheet. Diagnostic data were available also when the algorithm was switched off (suspended). The reliability and effectiveness of such algorithm was evaluated in a prospective randomized pilot study in which 61 patients with brady-tachy syndrome were enrolled and implanted with a rate responsive dual-chamber pacemaker [<xref ref-type="bibr" rid="R13">13</xref>]. After downloading of the algorithm, patients were randomly assigned to algorithm programming "on" or "suspended", with crossover after 1 month. Consistent atrial pacing induced an increase in atrial pacing percentage from 77 to 96%, and was associated with an 80% reduction of premature atrial contractions' number. Atrial fibrillation recurrences were not reduced in the overall population, but they significantly decreased by 42% when considering the patients in whom the atrial pacing percentage was < 90% during the algorithm "suspended" period. Algorithm tolerance was good with no severe adverse events. The atrial rate stabilization algorithm has been evaluated (alone or combined with consistent atrial pacing) in 16 patients with brady-tachy syndrome [<xref ref-type="bibr" rid="R14">14</xref>]. With regard to the effects on atrial fibrillation burden, 11 patients (69%) were found to benefit significantly from the consistent atrial pacing or atrial rate stabilization algorithms (reduction > 50% of atrial fibrillation burden). In detail, 7 patients were responders to both algorithms, 2 to consistent atrial pacing only, and 2 to atrial rate stabilization only.</p><p>The ability of the post-mode switching overdrive algorithm to stabilize the atrial rate after termination of treated atrial arrhythmias has been studied in 15 patients with structural heart disease and documented atrial and ventricular arrhythmias, receiving a Jewel AF device [<xref ref-type="bibr" rid="R15">15</xref>]. The algorithm was active in 41% of 600 spontaneous atrial tachyarrhythmia episodes. It was capable of driving and stabilizing the atrial rhythm in the presence of slow spontaneous atrial rhythm or premature atrial beats with normal atrioventricular conduction. In case of premature atrial beats with any degree of atrioventricular block, the algorithm stabilized the ventricular rate. The authors concluded that the algorithm is reliable and might be of benefit for atrial arrhythmia treatment.</p><p>Efficacy of preventive pacing algorithms has been confirmed by the preliminary results of the AF Therapy Study [<xref ref-type="bibr" rid="R16">16</xref>]. In 97 patients with a history of paroxysmal atrial fibrillation of at least 1 year, who had experienced at least three episodes during the last 3 months and were refractory to drug therapy, activation of four different preventive algorithms was associated with a significant improvement in all pre-selected endpoints: atrial fibrillation burden, average sinus rhythm duration, mean time in-between atrial arrhythmia episodes and number of patients free of atrial arrhythmias lasting > 1 min. All benefits were observed either in patients with or without conventional indication for pacing. Similarly, in the ADOPT trial [<xref ref-type="bibr" rid="R17">17</xref>], in which 320 patients were enrolled in a parallel design with randomization to algorithms on vs. off, symptomatic atrial fibrillation burden was reduced by 25% in the treatment group. On the contrary, atrial fibrillation episode number, quality of life and hospitalizations did not differ between the two randomization arms. An important criticism of ADOPT trial was that only symptomatic AF was used as an end-point.</p><p>Also, a retrospective analysis by the ADOPT investigators presented in Heart Rhythm 2004 suggested that less ventricular pacing is associated with less AF recurrence (although the results did not reach statistical significance). Concern about the actual effectiveness of prevention algorithms in improving clinical outcome has been raised by the conflicting results [<xref ref-type="bibr" rid="R18">18</xref>,<xref ref-type="bibr" rid="R19">19</xref>] of the most recently published studies in which arrhythmia burden was usually selected as the main endpoint. To explain that, the interaction with other critical factors has been claimed, first of all the atrial pacing site. It is well known that atrial pacing from the right atrial appendage is associated with a prolonged P wave duration with unfavorable effects on atrial conduction and refractoriness. Clinical studies suggest that combining prevention algorithms with interatrial septal pacing may lead to better clinical outcome [<xref ref-type="bibr" rid="R20">20</xref>]. Furthermore, the hemodynamic negative effects of unnecessary ventricular pacing in dual-chamber pacing systems may counterbalance the preventive role of pacing algorithms. Blanc et al. [<xref ref-type="bibr" rid="R21">21</xref>] demonstrated that pacing algorithms could reduce the atrial arrhythmia burden only in patients with ventricular pacing percentage < 70%. In the MOST study [<xref ref-type="bibr" rid="R22">22</xref>], in patients with sinus node disease, atrial fibrillation recurrences had reverse relationship with ventricular pacing percentage. Moreover, the OASES trial [<xref ref-type="bibr" rid="R23">23</xref>] showed that the AF Suppression algorithm was even more effective when the atrial lead was implanted in the low atrial septum. In patients with right atrial appendage leads, the algorithm reduced AF burden as measured by mode-switches by 49%, whilst in patients with low atrial septal leads, the AF burden was reduced by 70%. However, whereas the ADOPT and OASES trials showed that this AF suppression algorithm reduces AF, trials testing other algorithms have not shown a similar efficacy. The ATTEST trial [<xref ref-type="bibr" rid="R24">24</xref>] evaluated the efficacy of overdrive pacing, and high-rate pacing algorithms in patients with AF and an indication for pacing. There were no statistically significant differences in recurrence of symptomatic AF or in AF burden. ATP terminated 54% of atrial tachycardia episodes, but high-frequency pacing could not terminate AF. Another subsequent trial, quite similar to OASES one, the ASPECT [<xref ref-type="bibr" rid="R25">25</xref>], showed that pacing from the atrial septum resynchronizes the atrium and provides an antiarrhythmic effect that enhances the efficacy of the suppression algorithms.</p><p>It is very important to recognize the differences between these trials. The ADOPT [<xref ref-type="bibr" rid="R17">17</xref>] and OASES trials [<xref ref-type="bibr" rid="R23">23</xref>] tested the efficacy of a single algorithm to prevent AF, whereas ATTEST [<xref ref-type="bibr" rid="R24">24</xref>] tested the efficacy of 3 preventive pacing algorithms and 2 pacing algorithms designed to terminate atrial tachyarrhythmias. Moreover, the follow-up in ATTEST trial was 3 and not 6 months. Another matter with the OASES and the ASPECT trials was the crossover design. Because the beneficial effect of atrial pacing may persist, this design may have limited the observed efficacy of the pacing algorithms. Thus the difference in results between these trials may reflect differences in study design or differences in the efficacy of the algorithms used. Finally, due to high variability of atrial fibrillation recurrence patterns [<xref ref-type="bibr" rid="R26">26</xref>], published studies may be underpowered to demonstrate the true benefits of prevention algorithms.</p><p>In summary, 2 trials have shown that AF suppression algorithms enhances the antiarrhythmic effects of atrial pacing in patients with AF and an indication for pacing. Thus, minimizing ventricular pacing and placing the atrial lead in the septum may probably enhance the efficacy of these algorithms.</p></sec><sec sec-type="" id="s3"><title>Atrial antitachycardia pacing to treat atrial tachyarrhythmias</title><p>It has been demonstrated that rapid atrial pacing delivered for treating atrial tachycardia or atrial flutter may be effective in restoring sinus rhythm in 60-90% of patients [<xref ref-type="bibr" rid="R27">27</xref>]. Maximal effectiveness can be usually obtained by delivering antitachycardia pacing at a rate that is slightly greater than the atrial arrhythmia rate. Delivering of some extrastimuli following a rapid pacing train may be more efficacious than overdrive atrial pacing at the same pacing cycle length in terminating atrial flutter [<xref ref-type="bibr" rid="R28">28</xref>]. High-frequency pacing may change atrial tachycardia in transient atrial fibrillation with later sinus rhythm restoration. Appropriate detection of atrial tachyarrhythmias and efficacy of pacing therapies in patients receiving a dual defibrillator have been evaluated in large series. Adler et al. [<xref ref-type="bibr" rid="R29">29</xref>] reported on 537 patients with ventricular arrhythmia implanted with the Medtronic 7250 dual defibrillator (Medtronic Inc., Minneapolis, MN, USA) who were enrolled in the Worldwide Jewel AF Study and followed on average for 1 year. Seventy-four percent had a documented history of prior atrial tachyarrhythmias. Seventy-one percent of the patients had atrial therapies enabled at some time during the follow up, allowing collection and analysis of 3500 atrial episodes from 167 patients. In the 7250 Dual Defibrillator Italian Registry [<xref ref-type="bibr" rid="R30">30</xref>], 105 patients were enrolled. Implant indication was represented by combination of ventricular and atrial tachyarrhythmias in 52% of patients, ventricular tachyarrhythmias only in 33%, and atrial tachyarrhythmias only in 14%. During a mean follow-up of 6 months, 863 treated atrial episodes were collected and analyzed. Gold et al. [<xref ref-type="bibr" rid="R31">31</xref>], on behalf of the Worldwide Jewel AF-Only Investigators, studied 146 patients with recurrent drug refractory atrial fibrillation without prior ventricular tachyarrhythmias, who were implanted with a dual defibrillator and were followed on average for 1 year. During the follow-up 4913 treated episodes were available for stored electrogram analysis and therapy efficacy evaluation. Atrial tachyarrhythmia detection was very good in all the studies with a positive predictive value of atrial detection ranging from 91 to 99%. The most common reasons of inappropriate detection were represented by far-field R-wave oversensing during sinus rhythm or cluster of premature atrial contractions. Reprogramming of atrial sensitivity allowed in some case avoidance of inappropriate detection due to far-field R-wave oversensing. Sensitivity of atrial tachycardia and fibrillation detection and validation of continuous detection of atrial tachyarrhythmias have been specifically addressed by Swerdlow et al. [<xref ref-type="bibr" rid="R42">42</xref>] who performed 80 Holter recordings with telemetered atrial electrograms in 58 patients, implanted because of combination of atrial and ventricular arrhythmias. Continuous detection of atrial tachyarrhythmia could be demonstrated in 96% of patients with spontaneous episodes. Atrial sensitivity for arrhythmia detection was 100%. It is worthwhile to stress that the atrial arrhythmia detected at very early onset was very commonly a well organized atrial tachycardia. Among 2380 episodes in the worldwide series, 63% were classified as atrial tachycardia (mean atrial cycle 278 ± 56 ms) and 37% as atrial fibrillation (mean cycle 204 ± 35 ms). In the Italian Registry, among 863 atrial episodes, 53% were automatically classified by the device as atrial tachycardia and 47% as atrial fibrillation. After revision of the stored data, among 843 appropriately detected episodes, 55% were clinically classified as atrial fibrillation and 45% as atrial tachycardia. In the AF-Only group, among 3116 episodes treated by antitachycardia pacing, 67% were classified as atrial tachycardia.</p><p>Efficacy of antitachycardia pacing therapy in the three series is reported in <xref ref-type="table" rid="T1">Table 1</xref> [<xref ref-type="bibr" rid="R29">29</xref>-<xref ref-type="bibr" rid="R31">31</xref>]. The efficacy was estimated in two ways: 1) crude estimate, defined as the proportion of successful terminations out of the total number of treated episodes; 2) adjusted estimate, using the generalized estimate equation method [<xref ref-type="bibr" rid="R30">30</xref>] which allows adjusting estimate for multiple episodes within a patient through a correlation structure between episodes and patients.</p><p>In the worldwide study [<xref ref-type="bibr" rid="R29">29</xref>-<xref ref-type="bibr" rid="R31">31</xref>], a subanalysis performed to compare the efficacy of burst+ vs. ramp therapy for atrial tachycardia did not find any significant difference between the two therapies. An history of atrial flutter was the only independent predictor of pacing efficacy for atrial tachycardia, while no independent predictors could be identified for atrial fibrillation. No comparison was feasible between burst+ or ramp and high-frequency burst because the last was usually applied after unsuccessful delivery of burst+ or ramp therapy.</p><p>Pooling all antitachycardia pacing therapies (burst+, ramp and 50 Hz burst), their efficacy consistently increased as far as atrial arrhythmia cycle lengthened. As a matter of fact, in the Italian Registry, while the relationship between efficacy of atrial burst+ and ramp vs. atrial cycle length was very high (r2 = 0.85, p < 0.001), the relationship between efficacy of 50 Hz burst and atrial cycle length was very poor, with a trend toward a reverse correlation (r2 = 0.31, p < 0.05). Such a difference could be explained by taking into account the mechanisms of action of different antitachycardia pacing techniques. During overdrive pacing, the wavefront of pacing stimulus enters the reentrant pathway in the antidromic and orthodromic direction. The antidromic wavefront blocks by collision the arrhythmia wavefront, while the orthodromic wavefront may either reset the tachycardia or stop it, when early enough to be blocked in a refractory area. Slower tachycardias, due to a wider excitable gap, may be more easily terminated in this way. Termination of atrial tachyarrhythmias by high-frequency pacing is typically preceded by tachycardia acceleration, which becomes unable to be sustained, hence resulting in arrhythmia termination and sinus rhythm restoration. It has been hypothesized that high-frequency pacing induces a new faster reentrant circuit which is unable to sustain itself [<xref ref-type="bibr" rid="R34">34</xref>]. This mechanism looks independent of the arrhythmia cycle.</p><p>Looking at individual patients, random and wide distribution of median atrial cycles at onset during the follow-up was observed in the majority of them. A subgroup showed a narrow Gaussian distribution along either a fast (200 ms) or a slow band (250 ms). Antiarrhythmic drugs have been demonstrated to be capable of modifying atrial cycle profile. Antiarrhythmics may modify the electrophysiological properties of the atria and the arrhythmia organization pattern by lengthening the mean atrial arrhythmia cycle and by widening the temporal excitable gap [<xref ref-type="bibr" rid="R35">35</xref>-<xref ref-type="bibr" rid="R37">37</xref>]. Dijkman and Wellens [<xref ref-type="bibr" rid="R38">38</xref>] demonstrated that atrial arrhythmias in defibrillator patients with structural heart disease, receiving class III antiarrhythmic drugs, frequently had longer cycle lengths than atrial fibrillation. In fact, among 600 spontaneous episodes, atrial fibrillation was diagnosed in 19%, fast polymorphic atrial tachycardia in 20%, fast monomorphic atrial tachycardia in 57%, and slow atrial tachycardia in 4%. Drug-induced atrial cycle length changes may impact on atrial antitachycardia pacing therapy efficacy. In spite of that, in the AF-Only series there were no drugs that were independent predictors of therapy efficacy.</p><p>Finally, increasing the delay from arrhythmia detection to therapy delivery was associated with a significant reduction in efficacy. Optimal delay has been identified in less than 5 min in the Worldwide Study and in less than 1 min in the Italian Registry. That was probably due to the fact that the majority of atrial tachyarrhythmias accelerated in few minutes after onset. Considering that slower arrhythmias can be more easily treated by pacing techniques, a short delay in therapy delivery should be recommended. The efficacy of 50 Hz burst on atrial fibrillation may be a matter of debate. Wide local capture of atrial fibrillation has been documented [<xref ref-type="bibr" rid="R39">39</xref>]. Data from canine studies suggested that high-frequency pacing could accelerate the local atrial fibrillation cycle length and destabilize the reentrant rhythm so that destabilized atrial fibrillation could be converted to sinus rhythm in some cases. Anyway, in humans termination of persistent atrial fibrillation by high-frequency pacing could never be demonstrated [<xref ref-type="bibr" rid="R40">40</xref>,<xref ref-type="bibr" rid="R41">41</xref>]. High-frequency pacing may terminate induced atrial fibrillation during electrophysiological study with a 33% efficacy rate [<xref ref-type="bibr" rid="R34">34</xref>] and atypical atrial flutter with a 60% efficacy rate [<xref ref-type="bibr" rid="R41">41</xref>]. In selected cases, local capture of new-onset atrial fibrillation and entrainment of a relatively large area of the atria [<xref ref-type="bibr" rid="R42">42</xref>] could have destabilized the arrhythmia by reducing the number of wavelets of the random reentry, hence preventing arrhythmia sustenance. Anyway, considering that arrhythmia classification is based on atrial activity at atrial lead site, misclassification of atrial tachycardia as atrial fibrillation cannot not be excluded. Furthermore, some episodes may have terminated spontaneously after therapy delivery [<xref ref-type="bibr" rid="R43">43</xref>].</p><p>The ASPEN-ICD and ASSIST trials will probably give more data about these new perspectives. They will test the algorithm in patients with ICDs. The former will determine if the first episodes of AF can be prevented by the algorithm, whereas the latter will test if this algorithm prevents AF in patients who have a history of atrial arrhythmia.</p><p>Safety of antitachycardia pacing therapy was excellent since no ventricular arrhythmia induction was observed in any case after antitachycardia pacing delivery. These data suggest that, on a large scale, these algorithms would presumably be safe, and that it would be necessary to better determine how effective atrial ATP is, and whether it would reduce hispitalizations and/or electrical or pharmacological cardioversions and thereby improve quality-of-life.</p></sec><sec sec-type="" id="s4"><title>Conclusions</title><p>As stated in the discussed trials, pacing prevention algorithms should probably be better and further evaluated, before having a definitive and comprehensive answer on their utility. Clinical studies showed inconsistent clinical benefits of the algorithms, although they reduce atrial fibrillation triggers such as premature atrial complexes, thus preventing short-long cycle phenomenon. Factors which may be responsible for neutral results include: 1) high atrial pacing percentage in conventional DDDR, 2) non-optimal atrial pacing site, 3) deleterious effects of high percentages of ventricular pacing, and 4) inappropriate study design and endpoint selection. Variability of atrial arrhythmia recurrence patterns and onset mechanisms suggest individual programming of prevention algorithms by using data stored in the device memory. Atrial antitachycardia pacing is an effective tool to treat atrial tachyarrhythmias and it may stop nearly 50% of arrhythmia episodes. In particular, all previously described data are more related to atrial tachycardia reduction than to atrial fibrillation decrease. Delivery of atrial therapies early after arrhythmia onset and on more organized arrhythmias may improve success rate. Associated use of antiarrhythmic drugs, mainly propafenone and flecainide, may further increase effectiveness by lengthening atrial arrhythmia cycle.</p><p>Effective antitachycardia pacing therapies may reduce atrial fibrillation burden, but conflicting evidence does exist as regards this issue, probably because current clinical studies may be underpowered to detect such an efficacy. Consistent monitoring of atrial and ventricular rhythm as well as that of antitachycardia pacing effectiveness may be extremely useful for optimizing device programming and pharmacological therapy. At present the efficacy of pacing algorithms to prevent AF is by no means something that has been proved beyond reasonable doubt, so that more large-scale prospective trials are needed to answer this question. Our information is currently based on small trials and any attempt to extrapolate these data to the general population will probably produce inaccurate results. Clinical endpoints should be further investigated in future studies.</p></sec>
|
Quality of Life in Patients with Implantable Cardioverter Defibrillators
|
<sec id="st1"><title>Background</title><p>The implantable cardioverter defibrillator (ICD) is a life saving device for individuals with life threatening ventricular arrhythmias. There is no doubt that it is a cost effective therapy in various congenital and acquired arrhythmogenic disorders. Nevertheless, shock delivery may be painful and frightening which causes psychological distress and deterioration of perceived quality of life.</p></sec><sec id="st2"><title>Methods</title><p>A systematic meta-analysis on studies reporting quality of life in patients implanted with ICDs was done using professional databases. Related articles and references of the relevant articles were also searched for suitable studies.</p></sec><sec id="st3"><title>Results</title><p>Thirty studies with a total of 3412 patients on implantable defibrillators were identified. Five of them were large randomised studies with a total of 1680 patients, while 25 were non-randomised studies. Medical Outcome Study 36-item Short Form health survey (SF -94 36) was the most common instrument used for assessment of quality of life. Only one of the 5 major randomised trial reported worsening of quality of life after implantation of a defibrillator. In the subgroup of patients receiving shocks, three out of the five trials reported worsening of quality of life.</p></sec><sec id="st5"><title>Summary</title><p>Most of the randomised studies showed either neutral or better quality of life in patients on implantable defibrillators. In the subset of patients receiving shocks, worsening of quality of life was found in most randomised studies. Therefore, activation of antitachycardia pacing should be performed in every ICD-patient in order to miminze painful shocks and consequent deterioration of quality of life.</p></sec>
|
<contrib contrib-type="author"><name><surname>Francis</surname><given-names>Johnson</given-names></name><degrees>MBBS, MD, DM, FCSI</degrees><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Johnson</surname><given-names>Beena</given-names></name><degrees>MBBS, DCH, MD, FIACAM</degrees><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Niehaus</surname><given-names>Michael</given-names></name><degrees>MD</degrees><xref ref-type="aff" rid="aff3">‡</xref></contrib><aff id="aff1"><label>*</label>Associate Professor of Cardiology, Medical College Calicut, Kerala, India</aff><aff id="aff2"><label>†</label>Consultant in Psychological Medicine, Baby Memorial Hospital, Calicut, Kerala, India</aff><aff id="aff3"><label>‡</label>Professor of Cardiology, Dept. of Cardiology and Angiology, Hannover Medical School, Carl-Neuberg-Str. 1, 30625 Hannover, Germany</aff>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>Implantable cardioverter - defibrillators (ICD) have proven their value as life saving devices [<xref ref-type="bibr" rid="R1">1</xref>-<xref ref-type="bibr" rid="R3">3</xref>]. As the number of patients implanted with these devices rises world-wide, concerns about the quality of the life prolonged by the ICDs become more and more relevant. Patients on ICD are prone for deterioration of quality of life (QOL) due to worsening of the pre-existing cardiac disorder as they survive longer with the support of the device. Unexpected and often painful shocks can either be perceived as instances of life regained or as potential threats to survival by different patients. Systematic attempts at evaluation of QOL in ICD recipients have been there over the past decade. This paper aims at a comprehensive analysis of the literature on QOL in ICD recipients.</p></sec><sec sec-type="methods" id="s2"><title>Methods</title><p>A systematic accumulation of available data on QOL in patients with ICDs. PubMed searches were conducted using the search words implantable defibrillator, ICD, quality of life and QOL. The retrieved results were checked to identify relevant studies. Further studies were sought by searching for the related articles and the references of the retrieved article. Reports of major ICD trials were screened for data on quality of life.</p></sec><sec id="s3"><title>Results</title><p>Thirty studies on patients with ICDs reported data on QOL (<xref ref-type="table" rid="T1">Table 1</xref>) [<xref ref-type="bibr" rid="R4">4</xref>-<xref ref-type="bibr" rid="R33">33</xref>]. Of the total 4653 patients in these 30 studies, 3412 had ICDs implanted. The earliest study in 1993 reported on 57 patients [<xref ref-type="bibr" rid="R4">4</xref>. The largest available data is from the PainFREE Rx II trial with 634 patients on ICD [<xref ref-type="bibr" rid="R33">33</xref>]. The next largest study was the Antiarrhythmics Versus Implantable Defibrillators (AVID) trial with 416 patients on ICD with an equal number on amiodarone for comparison [<xref ref-type="bibr" rid="R1">1</xref>-<xref ref-type="bibr" rid="R23">23</xref>]. 22 studies had 50 or more patients on ICD while only seven had over hundred patients on ICD. Fourteen studies were published in the last decade while sixteen were in the current decade. Twelve studies used Medical Outcome Study 36-item Short Form health survey (SF - 36) for assessment of quality of life.</p></sec><sec id="s4"><title>Instruments for Assessment of QOL</title><p>SF - 36 was the most common instrument used for QOL assessment across the studies. It comprises of a 36-item short-form questionnaire constructed to survey health status in the Medical Outcomes Study [<xref ref-type="bibr" rid="R33">33</xref>]. SF - 36 is a generic instrument and not disease specific. There are eight multi-item scales, four each for physical and mental health assessment. The physical scales are: 1) Physical functioning, 2) Role physical, 3) Bodily pain and 4) General health. The scales for assessing mental health are: 1) Vitality, 2) Social functioning, 3) Role emotional and 4) Mental health.</p><p>Various other scales have also been used in the studies, like Ferrans and Powers Quality of Life Index [<xref ref-type="bibr" rid="R7">7</xref>,<xref ref-type="bibr" rid="R18">18</xref>,<xref ref-type="bibr" rid="R30">30</xref>], Quality-of-life Profile for the Chronically Ill [<xref ref-type="bibr" rid="R13">13</xref>], Quality of Life Index: Cardiac Version [<xref ref-type="bibr" rid="R1">1</xref>] and specifically developed quality of life questionnaire for ICD patients [<xref ref-type="bibr" rid="R10">10</xref>]. The latter two can be considered as disease specific measurements of QOL.</p></sec><sec id="s5"><title>Major Randomized Trials</title><p>Of the thirty studies available, five were large randomized trials (<xref ref-type="table" rid="T2">Table 2</xref>) with 1680 patients on ICD altogether. Three of them were primary prevention trials [<xref ref-type="bibr" rid="R17">17</xref>,<xref ref-type="bibr" rid="R27">27</xref>,<xref ref-type="bibr" rid="R33">33</xref>] while two of them were secondary prevention trials [<xref ref-type="bibr" rid="R21">21</xref>,<xref ref-type="bibr" rid="R23">23</xref>]. One primary prevention trial reported worsening of QOL after ICD while another reported neutral results and the third and most recent one reported improvement in QOL. This disparity was present among the secondary prevention trials as well, where one reported improvement while another reported worsening of QOL. Disparity was less between the trials in those patients receiving ICD shocks. Three trials reported worsening of QOL while one each reported neutral and better QOL in patients receiving shocks.</p></sec><sec id="s6"><title>Summary of QOL Data from Major Randomized Trials</title><sec id="s6a"><title>A] Primary Prevention Trials</title><sec id="s6a1"><title>1. CABG Patch Trial</title><p>Patients with impaired left ventricular function and positive signal averaged electrocardiogram undergoing CABG were randomly assigned to ICD versus no antiarrhythmic treatment in the CABG Patch trial [<xref ref-type="bibr" rid="R17">17</xref>]. All ICD-patients were implanted with patch electrodes during thoracotomy for CABG application. SF-36 was used to evaluate the QOL in 490 (68%) of the 719 patients available at 6 month follow up. Of these, 262 patients had ICD while 228 were controls. Patients in the ICD group had lower levels of psychological well being than the control group, especially those who experienced shock delivery of the ICD. The study was terminated when statistical analysis suggested that there was no significant difference in the primary outcome of total mortality. The lack of effect of ICDs in reducing mortality has been attributed to better survival in patients who have undergone revascularisation by CABG.</p></sec><sec id="s6a2"><title>2. AMIOVIRT Trial</title><p>Amiodarone versus implantable cardioverter-defibrillator (AMIOVIRT) trial included one hundred three patients with non-ischemic dilated cardiomyopathy (NIDCM), left ventricular ejection fraction < or = 0.35, and asymptomatic non sustained ventricular tachycardia (NSVT) [<xref ref-type="bibr" rid="R27">27</xref>]. They were randomized to receive either amiodarone or an ICD. The authors described no statistically significant survival advantage after one year or three years. QOL was also similar in both treatment groups. This trial was also neutral towards change of QOL in those receiving ICD shocks. Quality of Well Being Schedule and the State Trait Anxiety Inventory were the instruments used for assessment in this study.</p></sec><sec id="s6a3"><title>3. PainFREE Rx II Trial</title><p>Pacing Fast Ventricular Tachycardia Reduces Shock Therapies (PainFREE Rx II) trial was the largest trial reporting data on QOL as well as the latest [<xref ref-type="bibr" rid="R33">33</xref>]. This trial compared the safety and utility of empirical anti-tachycardia pacing (ATP) with shocks for fast ventricular tachycardia in patients receiving ICD. Of 634 ICD patients 313 were randomized to empirical ATP while 321 patients received shocks as the initial therapy of spontaneous fast ventricular tachycardia. ATP was effective in 229 of 284 episodes in the ATP arm. QOL was measured with the SF-36 and found to be improved in patients with fast ventricular tachycardia in both arms but even more in the ATP arm. Of the 131 patients with fast ventricular tachycardia, 98 completed SF-36 both at baseline and at 1 year. The ATP arm had significant improvement in 5 subscales - physical functioning, role physical, bodily pain, social functioning and role emotional, while the shock arm had improvement in only bodily pain score. None of the subscale measurements were significantly reduced at 1 year in either arm.</p></sec></sec><sec id="s6b"><title>B] Secondary Prevention Trials</title><sec id="s6b1"><title>1. The Canadian Implantable Defibrillator Study (CIDS)</title><p>CIDS [<xref ref-type="bibr" rid="R21">21</xref>] compared antiarrhythmic therapy versus ICD in patients resuscitated from cardiac arrest, and those with sustained ventricular tachycardia with syncope or ventricular tachycardia of more than 150 beats per minute with near syncope and left ventricular ejection fraction < or = 0.35 or syncope with inducible ventricular tachycardia.</p><p>The total mortality was lower in the ICD group, but the difference did not reach statistical significance. (P=0.09). QOL evaluation was done as a secondary endpoint in the study with the Rand Corporation's 38-item Mental Health Inventory (MHI) and the Nottingham Health Profile (NHP) for 317 English-speaking participants. Perceived QOL was found to be improved more in the ICD group. Significant improvement from baseline to the 6- and 12-month follow-up assessment was seen for 7 of the 10 variables assessed. Quality of life did not improve in those patients who received > or =5 shocks from their device. Still they were comparable to the antiarrhythmic group in QOL. There was no difference between those who received no shock and those who received 1-4 shocks from their ICD. None of the quality-of-life variables improved with time in the antiarrhythmic therapy group.</p></sec><sec id="s6b2"><title>2. The Antiarrhythmics Versus Implantable Defibrillator (AVID) Trial </title><p>AVID Trial [<xref ref-type="bibr" rid="R1">1</xref>] compared antiarrhythmic drug therapy with ICD in survivors of cardiac arrest and those with sustained ventricular tachycardia with syncope, or hemodynamically unstable sustained ventricular tachycardia with ejection fraction < or = 40%. The study was prematurely terminated when a statistically significant reduction in mortality was demonstrated in the ICD group. Perceived QOL was evaluated as a secondary endpoint using SF-36 and Quality of Life Index: Cardiac Version. Of the 800 eligible patients with a 1-year survival, 416 patients were in the ICD group and 384 patients in the antiarrhythmic group. Perceived QOL related was low in both groups, related to adverse symptoms of the particular therapy. The occurrence of one or more shocks was associated with significant reduction in QOL in the ICD group.</p></sec></sec></sec><sec id="s7"><title>Data from non-randomised studies</title><p>Of the 30 studies evaluated in this report, 25 were non-randomised (<xref ref-type="table" rid="T3">Table 3</xref>). Sixteen of these trials had 50 or more patients on ICD. Five of these trials reported improvement in QOL in ICD patients while 6 reported neutral results and five reported worsening of QOL. IICD implantation did not worsen QOL in recipients.</p><p>The earliest study which reported QOL data from ICD patients was by Luderitz et al [<xref ref-type="bibr" rid="R4">4</xref>]. Fifty-five of their 57 patients stated that it was worth having an ICD implanted. This study also documented that patients receiving more than 5 shocks had a higher anxiety level. The latest of the non-randomised studies was by Sears et al [<xref ref-type="bibr" rid="R32">32</xref>]. They noted that patients with high positive health expectations and high optimism reported better mental health and social functioning at follow-up assessment. These two factors may be targeted in future studies to improve QOL in ICD recipients.</p><p>The largest of the non-randomised studies actually reported on dual chamber defibrillators for restoration of sinus rhythm in atrial tachyarrhythmias. This study by Newman et al [<xref ref-type="bibr" rid="R25">25</xref>] had 267 patients implanted with DDD ICD. A total of 150 patients completed SF-36 evaluation. Significant improvement was noted in role-physical, physical functioning, vitality, mental health and social functioning scales. No deterioration of QOL was noted in those who received shocks.</p><p>In the next largest non-randomized study by Kamphuis et al [<xref ref-type="bibr" rid="R22">22</xref>] 133 patients were on ICD and 35 on other modalities of treatment. No significant difference in QOL was noted between the two groups. Physical function, mental health and social function improved with time in both treatment groups. In contrast, Stankoweit et al [<xref ref-type="bibr" rid="R15">15</xref>] who studied 132 patients on ICD noted reduced well being in 61% of them.</p></sec><sec id="s8"><title>Conclusions</title><p>Distressing aspects of ICD shocks are lack of warning, multiple shocks, and progressively increased sensations with multiple shocks. Three of the major randomised studies reported worsening of QOL in those patients receiving ICD shocks, especially when patients were receiving five or more shocks. Despite the AMIOVIRT Trial [<xref ref-type="bibr" rid="R27">27</xref>] did not find any difference in the QOL between those receiving ICD shocks, the main issue regarding quality of life is the occurrence of shocks. The PainFREE Rx II Trial [<xref ref-type="bibr" rid="R33">33</xref>] demonstrates very clearly that QOL can be improved by programming of antitachycardia pacing even in patients with fast VTs. This trial also documented the efficacy of antitachycardia pacing.</p><p>With regard to a potential deterioration of perceived QOL caused by delivery of ICD shocks for fast VTs, we would recommend to program ATP attempts in all ICD patients regardless of the cycle length of the index arrhythmia.</p></sec>
|
Infra-His Block in a Normal Heart
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Alasti</surname><given-names>Mohammad</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Alizadeh</surname><given-names>Abolfath</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Fazelifar</surname><given-names>Amirfarjam</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Mohammad Ali</surname><given-names>Abolfath</given-names></name><degrees>MD</degrees></contrib><aff> Department of Pacemaker and Electrophysiology, Rajaie Cardiovascular Medical Center, Vali-asr Avenue, Tehran, IRAN</aff>
|
Indian Pacing and Electrophysiology Journal
|
<p>A 55 year old man with history of palpitation was referred for electrophysiologic study. Baseline ECG, physical examination and transthoracic echocardiographic study were normal. Electrophysiologic study revealed normal AH and HV intervals. Pacing of right atrium with a cycle length of 300 msec showed 2:1 AV block. AH interval was 252 msec and the block was infra-his (<xref ref-type="fig" rid="F1">Figure 1</xref>). With continual of right atrial pacing, one to one AV conduction with increasing AH interval to 282 msec and QRS widening (LBBB pattern) were being observed. HV intervals during 2:1 block and during 1:1 AV conduction were normal. What is the mechanism? Is it an abnormal finding in this patient?</p><p>During 2:1 AV conduction, AH interval was 252 msec and conducting system below His was in absolute refractory period so the impulse could not propagate and infra-his block occurred. After increasing of AH interval to 282, the conducting system below His had enough time to enter the relative refractory period. So AH conduction delay permitted transition from 2:1 AV conduction to 1:1 AV conduction. Infra-his block can be produced at paced cycle lengths of less than 400 msec in normal His-purkinje system and is not considered abnormal in this case.</p>
|
Transient unexpected improvement of AV conduction: What is the mechanism?
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Friedman</surname><given-names>Meir</given-names></name><degrees>MD</degrees></contrib><contrib contrib-type="author"><name><surname>Schweitzer</surname><given-names>Paul</given-names></name><degrees>MD</degrees></contrib><aff> Beth Israel Medical Center, New York, NY</aff>
|
Indian Pacing and Electrophysiology Journal
|
<p>This ECG was recorded from a 93 year old patient with a previously documented third degree AV block and an underlying LBBB. The twelve lead ECG demonstrates sinus bradycardia at a rate of 52 beats per minute, a PR interval of 230 milliseconds, and a left bundle branch block (<xref ref-type="fig" rid="F1">Figure 1</xref>). In <xref ref-type="fig" rid="F2">figure 2</xref>, a rhythm strip of leads V1, II and V5 shows sinus rhythm with AV block. Note that only P waves in the T wave of the paced beat are conducted. The non-conducted P waves are followed by ventricular paced beats at an escape interval of 1200 milliseconds.</p><p>What is the mechanism of AV conduction in this patient? The following possibilities should be considered:
<list list-type="order"><list-item><p>Supernormal conduction</p></list-item><list-item><p>Wedensky phenomenon</p></list-item><list-item><p>Gap phenomenon</p></list-item><list-item><p>Peel back effect</p></list-item></list></p><p>In the sixties and seventies Pick et al [<xref ref-type="bibr" rid="R1">1</xref>], Schamroth [<xref ref-type="bibr" rid="R2">2</xref>] and others [<xref ref-type="bibr" rid="R3">3</xref>] suggested supernormal conduction as the most likely mechanism of unexpected transient improvement of atrio-ventricular conduction defects. This electrophysiological phenomenon was defined as better than expected conduction in patients with depressed conduction during a short interval in the ventricular cycle [<xref ref-type="bibr" rid="R1">1</xref>]. However, there is no definitive proof of supernormal conduction in humans.</p><p>The next possible explanation is the Wedensky effect. This mechanism was first documented on the nerve-muscle preparation and was defined as a "stronger" stimulus, where in the case of AV block, a ventricular premature or paced beat, is followed by transient antegrade conduction by decreasing the refractoriness of the AV conduction [<xref ref-type="bibr" rid="R2">2</xref>]. Similarly to supernormal conduction, its existence in humans remains controversial.</p><p>According to Moe et al [<xref ref-type="bibr" rid="R4">4</xref>] and others [<xref ref-type="bibr" rid="R5">5</xref>], there are alternative explanations for electrocardiographic abnormalities suggestive of supernormal conduction or Wedensky phenomenon. The first of such explanations is the Gap phenomenon. Experimental and clinical studies [<xref ref-type="bibr" rid="R4">4</xref>,<xref ref-type="bibr" rid="R5">5</xref>] showed that atrial premature beats with longer coupling intervals were blocked and earlier ones conducted. The explanation of Gap Phenomenon is conduction delay in the proximal part of the AV conduction system, causing recovery of its more distal portion. The gap phenomenon is an unlikely mechanism in our patient because the conducted P wave occurred after the paced beat rather than following a premature beat.</p><p>A more likely mechanism is the "peeling back" effect. It is assumed that pre-excitation of the AV node by a ventricular or junctional beat shortens the absolute refractory period of the AV or the His-Purkinje system and allows conduction of a supraventricular impulse [<xref ref-type="bibr" rid="R4">4</xref>].</p><p>The final question is the site of AV block. Because the patient has a left bundle branch block the possibility of infranodal block has to be considered. However, without His bundle recording, the site of AV block remains uncertain.</p>
|
Sudden Cardiac Death in Athletes - What Can be Done?
|
<p>Sudden death in athletes is a rare event but brings with it an impact that goes beyond sport. There are many causes of sudden death during exercise. While the responsibility of preventing or treating them lays with us physicians, preparticipation screening is largely ineffective and impractical. Definitive, large scale prospective research is required in order to design the most cost-effective system for screening of athletes. In the meanwhile rapid access to defibrillators by trained personnel remains the best possible approach to abort sudden death.</p>
|
<contrib contrib-type="author"><name><surname>Ghosh</surname><given-names>Joydeep</given-names></name><degrees>MD, FACC</degrees></contrib><aff>Assistant Professor of Medicine, Maimonides Medical Center, 4802 Tenth Avenue, Brooklyn, NY 11219, USA</aff>
|
Indian Pacing and Electrophysiology Journal
|
<p>With the World Cup in soccer - arguably the most passionate sporting event of our times - only a few days away, it is appropriate to look at the lessons we have learned over the past few years about sudden death in competitive sport.</p><p>In 490 BC Phidippides ran 26.2 miles from Marathon to Athens, delivering the news of the Greek victory over the Persians (the first 'marathon' in recorded history), but then immediately collapsed and died. This is probably the first recorded incident of sudden death of an athlete. More recently the death of some high-profile athletes have focused public interest on this infrequent event and current data reveal that in the United Sates alone, about 100 athletes - ranging from schoolchildren to professionals - die every year while engaged in sporting activity [<xref ref-type="bibr" rid="R1">1</xref>]. In India accurate data are not available but recently media attention was focused on a Brazilian soccer player who died while playing for his adopted club. Across societies around the world athletes are considered to be the pinnacle of health and given high status; death of a young, well-trained athlete on the field represents a particularly tragic event that leads to utter dismay within the family and the community and often a barrage of negative media attention.</p><p>The paradox of exercise is that while in the long-term it generally attenuates the risk of sudden cardiac death, the relative risk is actually increased during the time of exercise and up to 30 minutes following it [<xref ref-type="bibr" rid="R2">2</xref>]. This is especially so in patients with hypertrophic cardiomyopathy, congenital anomalous coronary artery abnormalities, and inherited ion channelopathies like long QT syndrome and catecholaminergic polymorphic ventricular tachycardia, or Arrhythmogenic Right Ventricular Cardiomyopathy, and it is no surprise that patients with these conditions have a risk of sudden death with sport that is well above normal. Other conditions implicated in sudden death in athletes include accessory pathways with antegrade conduction properties, bicuspid aortic valve, myocarditis, ruptured aortic aneurysm, Marfan syndrome, mitral and tricuspid valve prolapse and ingestion of stimulants like cocaine and anabolic steroids. In addition, sudden death on the field can also be caused by certain activities like trauma to the chest (commotio cordis) or to the neck (causing asystole from carotid sinus reactivity) [<xref ref-type="bibr" rid="R3">3</xref>].</p><p>Given the myriad causes listed above, the question we have to answer is whether pre-participation screening works. Since the early eighties, Italy has been in the forefront of efforts to identify athletes perceived to be at risk and withdraw them from competitive sports, with a preparticipation screening program based on 12-lead ECG, a family and personal history and physical examination [<xref ref-type="bibr" rid="R4">4</xref>]. Data presented recently [<xref ref-type="bibr" rid="R5">5</xref>] reveal a striking decline in mortality in the population studied (12-35 year olds, in the Veneto region) with the incidence of sudden death declining to one-tenth of what it was at the beginning of the study. Most of this was attributed to fewer cases of athletes dying from cardiomyopathies that were detectable on ECG. To the best of my knowledge, there are no other prospective studies (of any magnitude) that look at the effect of preparticipation screening in athletes.</p><p>Whether the result of this study is applicable to other countries is of course, questionable. Northern Italy has an unusually high incidence of ARVD in its homogenous population, and unless one is dealing with similar circumstances the results may not be reproducible. In the United States currently there is no advocacy for routine ECG's, and the diagnostic tool used most frequently is a detailed history and physical examination, with particular attention paid to syncope, palpitations, chest pain or sudden death in the family. ECG, stress testing and echocardiograms are used only if there is a suspicion of high risk from the history elicited. Interpretations of 'quick-screen' echocardiograms, while being effective in diagnosing HCM or ARVC in a short period of time, may be particularly challenging given the similarities between "athletes heart" and hypertrophic cardiomyopathy.</p><p>With regard to India and other developing countries, such preparticipation screening will need an enormous increase in the Government's commitment to public health given the huge population of people that needs to be screened and the low incidence of disease that cause sudden death (a notable exception may be hypertrophic cardiomyopathy, which has been reported to have an incidence of 1 in 500 in the Western population). Unfortunately I doubt that this 'high investment, low return' policy will find many takers in a country like India, though saving even one young adult or child's life ought to be seen as making an investment in the future. The other significant problem with screening is identifying false positives, with the unnecessary psychological burden it places on athletes and their families [<xref ref-type="bibr" rid="R6">6</xref>]. If any kind of screening policy is indeed adopted, we need physicians who are well-trained to identify the diseases they are seeking, and who know how to proceed with borderline cases without causing unnecessary alarm.</p><p>If preparticipation screening is far away in a country like India, the least we can do is provide external defibrillators in as many athletic events as possible, and train paramedics to use them properly. In particular they must be able to recognize cardiac arrest quickly, and deliver defibrillation promptly. A successful outcome depends primarily on time to defibrillation, with about a 10% per-minute decrease in survival [<xref ref-type="bibr" rid="R5">5</xref>].</p><p>In the meanwhile, enjoy the World Cup. Strict compliance to FIFA guidelines at all venues in Germany regarding availability of rapid access defibrillation will hopefully ensure that no gifted athlete dies suddenly on the field.</p>
|
Diastolic And Systolic Right Ventricular Dysfunction Precedes Left Ventricular Dysfunction In Patients Paced From Right Ventricular Apex
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Dwivedi</surname><given-names>SK</given-names></name><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Bansal</surname><given-names>Sandeep</given-names></name><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Puri</surname><given-names>Aniket</given-names></name><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Makharia</surname><given-names>MK</given-names></name><xref ref-type="aff" rid="aff2">†</xref></contrib><contrib contrib-type="author"><name><surname>Narain</surname><given-names>VS</given-names></name><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Saran</surname><given-names>RK</given-names></name><xref ref-type="aff" rid="aff1">*</xref></contrib><contrib contrib-type="author"><name><surname>Hasan</surname><given-names>M</given-names></name><xref ref-type="aff" rid="aff3">‡</xref></contrib><contrib contrib-type="author"><name><surname>Puri</surname><given-names>VK</given-names></name><xref ref-type="aff" rid="aff4">§</xref></contrib><aff id="aff1"><label>*</label>Professor of Cardiology, Department of Cardiology, King George's Medical University, Lucknow, India.</aff><aff id="aff2"><label>†</label>Senior Resident, Department of Cardiology, King George's Medical University, Lucknow, India.</aff><aff id="aff3"><label>‡</label>Ex-Professor and Head, Department of Cardiology, King George's Medical University, Lucknow, India.</aff><aff id="aff4"><label>§</label>Professor and Head, Department of Cardiology, King George's Medical University, Lucknow, India.</aff>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>Single chamber right ventricular apical pacing (VVI) continues to be the predominant pacing mode in the developing countries primarily due to economic constraints, despite the well-known advantages of dual chamber pacing. Pacemaker induced myocardial dysfunction has been recognized for a long time and the effect of single chamber right ventricular (RV) apical pacing (VVI) on left ventricular (LV) functions has been studied previously, yet there remain some unexplored areas such as the chronology of events to occur [<xref ref-type="bibr" rid="R1">1</xref>-<xref ref-type="bibr" rid="R8">8</xref>]. Majority of studies have looked into LV dysfunction while effects on the right ventricle (RV) have not been studied thoroughly. Furthermore there is little data on the extent and the time frame in which these changes occur. We therefore, aimed to evaluate systolic and diastolic functions of LV and RV in consecutive patients undergoing VVI permanent pacemaker implantation (PPI) over a period of 6 months using trans-thoracic echocardiography.</p></sec><sec sec-type="materials|methods" id="s2"><title>Material and Methods</title><p>All consecutive patients undergoing VVI pacemaker implantation having a class I indication according to AHA guidelines [<xref ref-type="bibr" rid="R9">9</xref>] were included in the study. A written informed consent was taken for inclusion in the study. Patients undergoing PPI implantations with other modes of pacing were excluded from the study.</p><sec id="s2a"><title>Clinical evaluation</title><p>A thorough clinical evaluation and resting 12-lead electrocardiogram (ECG) were done at admission, on day 7 after PPI and on follow up at 1 month and 6 months. Patients were discharged with a programmed pacing rate of 70/mt. At follow up visits, patients were evaluated for any evidence of a new onset dyspnea attributable to pacemaker syndrome, in the absence of any recognizable recent myocardial infarction, anaemia, respiratory illness and new LV ejection fraction of less than 50% [<xref ref-type="bibr" rid="R10">10</xref>,<xref ref-type="bibr" rid="R11">11</xref>]. The left ventricular function was evaluated by 2D echocardiography. At the end of 6 months follow up, a 24 hour ambulatory ECG was done to document the duration of paced and sinus rhythm in the previous 24 hours. A patient with paced rhythm of more than 90% of all ventricular complexes was considered to be on persistent pacing mode.</p></sec><sec id="s2b"><title>Echocardiography protocol</title><p>Baseline echocardiographic evaluation was done within 24 hours of hospital admission with the temporary pacemaker rate being kept at 70/mt. In the event of intact atrio-ventricular (AV) conduction, echocardiography was done on right ventricular temporary pacing at a rate 10 beats higher than the intrinsic sinus rate. Echocardiography was repeated at day 7, 1 month and 6 months after implantation. Each measurement was taken in three consecutive paced beats not preceded by a sinus P wave and the average was considered as the representative value.</p><p>LV end diastolic and systolic volumes (LVEDV and LVESV) were calculated using modified Simpson's method. Left ventricular internal diameter in diastole (LVIDd) and systole (LVIDs) and fractional shortening (FS) were measured in parasternal long axis view using the M-mode. The right ventricular internal diameter in diastole (RVIDd) and systole (RVIDs) were taken in the parasternal long axis view and their mean was taken as the RVID. Left atrial size was measured in systole (LAs) and diastole (LAd) on M-line in parasternal long axis view. Left sided systolic time intervals; LV pre ejection period (LVPEP) and LV ejection time (LVET) were calculated using Doppler aortic flow. Right sided systolic time intervals, RV pre ejection period (RVPEP), and RV ejection time (RVET) were calculated using Doppler pulmonary flow. The LV diastolic parameters were calculated in apical 5-chamber view keeping the Doppler cursor at a point where both aortic outflow and mitral inflow could be obtained. The LV isovolumic relaxation time (LV-IVRT) was taken as a period from end of aortic flow to beginning of mitral inflow. The acceleration and deceleration time of mitral inflow (MV-AT and MV- DT) were measured from the onset of the mitral inflow to its peak and from peak to return to baseline respectively with the Doppler cursor placed just beyond the mitral annulus in apical four-chamber view. The RV isovolumic relaxation time (RV-IVRT) was calculated using two views. First, the RV electromechanical systole was calculated in parasternal short axis view using Doppler pulmonary flow. This was taken as the period from Q-wave on ECG to the completion of pulmonary flow. An average of three such readings was taken. Then the tricupid valve Doppler inflow tracing was taken in apical four chamber view in the manner similar to mitral inflow. The RV electromechanical systole was marked on this. This point was taken as beginning of the RV diastole and the period from this point to the beginning of tricuspid inflow was taken as RV-IVRT. The tricuspid acceleration and deceleration time (RV-AT and RV-DT) were measured in a manner similar to mitral AT and DT. The systemic and pulmonary cardiac outputs were calculated using velocity time integral (VTI) method across aortic and pulmonary Doppler flow respectively. Colour Doppler was used to conform and quantify AV valvar regurgitation, using area of the regurgitant jet relative to the size of LA [<xref ref-type="bibr" rid="R10">10</xref>,<xref ref-type="bibr" rid="R12">12</xref>].</p></sec><sec id="s2c"><title>Statistical analysis</title><p>Student 't' test was used for statistical evaluation. The paired 't' test was used to compare the baseline values with the values at day 7, 1 month and 6 months respectively. A 'p' value of <0.05 was considered to be significant. Parameters compared were LVIDd, LVIDs, LVEDV, LVESV, RVID, LVEF, LVFS, LVPEP, LVET, RVPEP, RVET, LAd, LAs, systemic and pulmonic cardiac output, LV-IVRT, LV- AT& DT, RV-IVRT, RV- AT & DT. The effect of confounding parameters as age (< 50 versus > 50 years), sex (male versus female), indication (complete heart block versus Sick Sinus Syndrome), LV dysfunction (LVEF < 50% versus > 50%) and the requirement of pacing (intermittent versus regular) were also studied using two way 't' test.</p></sec></sec><sec id="s3"><title>Results</title><p>Forty-eight patients underwent VVI pacing. Three of the 48 patients had poor echo window and were excluded from the final analysis. Over the 6 months follow up, 2 patients died. One patient had a haemorrhagic stroke and the other having documented coronary artery disease died suddenly. Two patients were lost to follow up after discharge but postal inquiries showed that they had no fresh symptoms. Data could therefore be obtained for all 45 patients at baseline and day 7, in 44 patients at 1 month and 41 patients at 6 months.</p><p>Baseline characteristics of the patients are given in <xref ref-type="table" rid="T1">Table 1</xref>. The study population consisted of 66.7% males with a mean age of 65.6±11.8 years (range 35-90years). Thirty four (75.6%) patients had complete heart block (CHB) and remaining (24.4%) had sick sinus syndrome (SSS). The mean LVEF of the group was 61.8±10.4%. Seven patients had LVEF of < 50% with 5 having dilated cardiomyopathy with normal coronaries and 2 having coronary artery disease, 1 patient had complete heart block secondary to inferior wall MI persisting beyond 3 weeks and 5 patients had co-existing coronary artery disease (CAD) of varying duration.</p><p>Syncope did not recur in any patients over the 6 month period. Eight patients who had presented on admission with dyspnoea experienced initial relief with drug therapy but dyspnoea worsened in 4 of these patients. Over the 6 months follow up period new onset dyspnoea occurred in 7 patients. In 4 patients, it was considered as a part of pacemaker syndrome while in 3 it was attributed to progressive LV dysfunction. New onset persistent atrial fibrillation developed in 9 patients, it was noted at the end of 1 month in 2 patients and at 6 months in remaining 7 patients. Nine patients were adjudged to be having intermittent pacing showing <90% paced complexes on 24 hours ambulatory ECG monitoring at 6 months.</p><p>Cardiac dimensions, volumes, systolic functions and systolic time intervals are shown in (<xref ref-type="table" rid="T2">Table 2</xref>). The LVEDV and LVESV increased significantly by 6 months over their corresponding baseline values (p<0.05). The right ventricular dimension (RVID) showed an earlier increase from baseline (p<0.05) at day 7 and progressively increased in the 6 months follow up period. Significant increase in left atrial diastolic diameter (LAd) was observed at 6 months (p<0.05). The left atrial systolic diameter (LAs) also increased but not statistically significant (p>0.05). The LVEF decreased progressively from baseline (61.82 ± 10.36%) and was statistically significant at 6 months (52.52 ± 12.11%, p<0.05). Systemic and pulmonic resting cardiac output did not change significantly over the six months follow up period. The LVPEP increased significantly by 6 months (p<0.001) while LVET decreased over the same period (p<0.05). The RVEP increased significantly by 1 month (p<0.001) and continued to increase progressively thereafter. The RVET decreased significantly by 6 months (p<0.02).</p><p>Among the diastolic parameters (<xref ref-type="table" rid="T3">Table 3</xref>), LV-IVRT increased significantly at 1 month (p<0.001) and continued to increase progressively till 6 months. The mitral valve DT also increased similarly becoming statistically significant by 1 month (p<0.01). The right ventricular parameters showed a change much earlier than corresponding LV parameters. RV-IVRT showed a significant increase as early as day 7 (p<0.01) and continued to do so during the 6 month follow up period. The RV-DT also showed a significant increase by 1 month (p<0.02). The ventricular filling acceleration time (AT) did not change significantly during the entire period in either of the two ventricles. New valvar regurgitation was seen in 7 patients. Of these, 4 had mild mitral regurgitation (MR) while 3 had mild tricuspid regurgitation (TR). Two of these patients had both MR and TR co-existent. One patient had pre-existing MR and TR, which worsened after pacemaker implantation.</p><p>The changes observed were largely uniform and independent of age, sex, indication for pacing and pre-existing LV dysfunction. Only persistent pacing, as identified by >90% paced beats in a 24 hour holter, was associated with a significantly higher decrease in LVEF (p<0.02) in persistently paced group. This group also showed a higher LVESV (p<0.005), greater RV dysfunction as suggested by higher RVPEP/RVET ratio (p=0.024), and a tendency to have a larger LAd (p=0.063).</p></sec><sec id="s4"><title>Discussion</title><p>In the present study, the first significant change to appear in myocardial functions following VVI pacing was in diastolic properties of RV. This was manifest by an increase in RV-IVRT at one week, followed by an increase in RV dimensions and the appearance of RV systolic abnormality i.e. increased RV-PEP and RV-PEP/RV-ET ratio at one month. The LV diastolic parameters were significantly altered at one month, which were manifest by an increase in IVRT and DT, followed by LV systolic dysfunction which appeared at 6 months with an increase in LV-PEP and LV-PEP/LV-ET ratio. The reduction in LVEF also manifest at 6 months without change in the resting cardiac output.</p><p>Our findings suggest two inter-related sequence of events. Firstly, it appears that changes in cardiac haemodynamics start appearing as early as one-week after VVI pacing, which are subtle and are limited to right ventricle only. Left ventricular dysfunction occurs later and this follows RV dysfunction, as reported in previous studies [<xref ref-type="bibr" rid="R4">4</xref>,<xref ref-type="bibr" rid="R6">6</xref>-<xref ref-type="bibr" rid="R8">8</xref>]. Secondly, diastolic abnormalities are first to appear which are followed by appearance of systolic abnormalities of the respective ventricles. Our study is unique in highlighting the sequence in change in the LV and RV diastolic and systolic function in a serial fashion in VVI paced patients.</p><p>There is often a debate whether LV dysfunction after long term ventricular pacing is real or is dependent on pre-existing disease. Mohan et al4 in their study found no significant difference in resting LVEF in long term follow up of AAI and VVI paced patients of sinus node dysfunction without structural heart disease. Similarly, Anderson et al [<xref ref-type="bibr" rid="R6">6</xref>], in their study of sick sinus syndrome patients, found no difference in incidence of clinical congestive heart failure or LV dimensions between AAI and VVI pacing modes on long term follow up. On the other hand, Faerestrand et al [<xref ref-type="bibr" rid="R13">13</xref>] found an increase in end diastolic volume in VVI paced patients after 3 months. Likewise, Fehrsson et al7 found greater resting LV volumes and lesser EF in patients paced on VVI rather than with physiological pacing mode. In most of these studies, stress has been on systolic function of LV after long term pacing while few have looked into the serial follow up of these patients. Data regarding the diastolic function is conflicting. LeClercq et al [<xref ref-type="bibr" rid="R14">14</xref>] in a study of 11 patients found that filling rates improved with AAI pacing as compared to DDD pacing, and similar Doppler data was provided in another report by Rosenqvist et al [<xref ref-type="bibr" rid="R15">15</xref>]. However data from Vardas et al [<xref ref-type="bibr" rid="R16">16</xref>] found no difference in the the diastolic Doppler indices between AAI and DDD modes of pacing. Moreover, the systolic functions of right ventricle as well as diastolic functions of both the ventricles have not been studied comprehensively.</p><p>The true mechanism of cardiac dysfunction is still unclear. It has been shown that pacing from RV apex induces an electrical sequence of depolarisation, which results in reduced stroke output of both the ventricles [<xref ref-type="bibr" rid="R17">`7</xref>,<xref ref-type="bibr" rid="R18">18</xref>]. RV apical pacing induces a left bundle branch block pattern and studies have shown that alteration in normal activation sequence adversely affects LV function as demonstrated by decrease in the global ejection fraction [<xref ref-type="bibr" rid="R19">19</xref>]. This is supported by finding that near normal activation sequence as seen in RV outflow tract pacing results in a narrowed QRS and improved cardiac output [<xref ref-type="bibr" rid="R20">20</xref>]. Karpawich and Mital, in a report, have also shown a normalized LV function with septal pacing despite a loss in atrial contribution as compared to atrial pacing [<xref ref-type="bibr" rid="R21">21</xref>]. Resultant chronic reduction in stroke output may mediate a compensatory mechanisms leading to cardiac chamber enlargement in an effort to maintain requisite stroke volume by Starling's law. There is evidence that delayed depolarisation of the LV can result in augmented intra-myocardial pressure in the septum and affect myocardial perfusion. Kolletis et al [<xref ref-type="bibr" rid="R22">22</xref>] have shown that atrial or AV sequential pacing does not alter coronary flow reserve while ventricular pacing decrease resting coronary flow velocity in some patients. Furthermore, long term RV apical pacing also results in myocardial perfusion defects on nuclear studies [<xref ref-type="bibr" rid="R23">23</xref>]. The magnitude of these defects is shown to be proportional to the duration of pacing and are associated with apical wall motion abnormalities and impaired global LV function [<xref ref-type="bibr" rid="R23">23</xref>].</p><p>Beside this, animal experimental studies have shown myofibrillar disarray [<xref ref-type="bibr" rid="R24">24</xref>], redistribution of myocardial fibre strain and blood flow [<xref ref-type="bibr" rid="R25">25</xref>] and adverse histopathological changes [<xref ref-type="bibr" rid="R26">26</xref>] on long term pacing in canine hearts which may also contribute to systolic dysfunction on RV apical pacing. Since repolarization sequence is determined by depolarization sequence, myocardial relaxation is also likely to be non-homogeneous and diastolic abnormalities are likely to appear. The diastolic filling parameters on radionuclide left ventriculography, such as time to peak filling rate and negative dp/dt are significantly altered in ventricular paced animal models [<xref ref-type="bibr" rid="R27">27</xref>] as well as in human hearts on atrioventricular pacing [<xref ref-type="bibr" rid="R28">28</xref>].</p><p>However, these mechanisms are not enough to explain significant reduction in systolic performance of the left ventricle as adjudged by gross reductions of the LVEF on long term pacing. It is hypothesized that left ventricular diastolic and systolic functions are deranged in many conditions secondary to involvement of right ventricle. Isolated right ventricular infarction, atrial septal defect with right sided chronic volume over load, cor pulmonale and chronic obstructive lung disease result in altered LV diastolic as well as systolic functions secondary to RV dysfunction [<xref ref-type="bibr" rid="R29">29</xref>-<xref ref-type="bibr" rid="R32">32</xref>]. LV dysfunction in paced patients may be a consequence of RV dysfunction which occurs earlier as is shown in the present study. However, more studies are required to have a firm explanation to this hypothesis.</p><sec id="s4a"><title>Study Limitations</title><p>Echocardiographic evaluation of RV size may not be a reliable and reproducible method with inherent problems of inter and intra observer variability. However, since every patient was serving as their own control, and a change in the parameters could be attributed to permanent pacing. The study was designed to include consecutive patients undergoing single chamber VVI pacing and those having a baseline ejection fraction of less than 50% were also included, however in the final analysis on comparison with patients having ejection fraction greater than 50% there was no difference in the observed parameters. A critically timed atrial contraction can change the filling characteristics of ventricle and alter the measured diastolic and systolic parameters. Though, we chose to evaluate only paced complexes, it may not have completely removed the impact of atrial contribution to ventricular filling during the intervening sinus rhythm.</p><p>In conclusion, the present study shows that right ventricular apical pacing has a major effect on systolic and diastolic ventricular functions. On serial evaluation, right ventricle dysfunction is the first abnormality that occurs in right ventricular apical paced patients, followed by LV dysfunction which appears later and on both sides, diastolic dysfunction precede the systolic dysfunction.</p></sec></sec>
|
Single-chamber Versus Dual-chamber Implantable Cardioverter Defibrillators: Do We Need Physiologic Pacing in The Course?
|
Could not extract abstract
|
<contrib contrib-type="author"><name><surname>Budeus</surname><given-names>Marco</given-names></name></contrib><contrib contrib-type="author"><name><surname>Buck</surname><given-names>Thomas</given-names></name></contrib><contrib contrib-type="author"><name><surname>Wieneke</surname><given-names>Heinrich</given-names></name></contrib><contrib contrib-type="author"><name><surname>Erbel</surname><given-names>Raimund</given-names></name></contrib><contrib contrib-type="author"><name><surname>Sack</surname><given-names>Stefan</given-names></name></contrib><aff> Department of Cardiology, West-German Heart Centre, University of Duisburg-Essen, Germany</aff>
|
Indian Pacing and Electrophysiology Journal
|
<sec id="s1"><title>Introduction</title><p>The main function of implantable cardioverter defibrillator (ICD) therapy is to preserve life by terminating life-threatening tachycardias like ventricular fibrillation (VF) and ventricular tachycardia (VT). Many studies showed the benefit for ICD in primary and secondary prevention of sudden cardiac death in patients with different cardiac diseasess [<xref ref-type="bibr" rid="R1">1</xref>-<xref ref-type="bibr" rid="R9">9</xref>]. In addition ICD is superior to antiarrhythmic therapy for preventing sudden cardiac death [<xref ref-type="bibr" rid="R1">1</xref>,<xref ref-type="bibr" rid="R3">3</xref>,<xref ref-type="bibr" rid="R5">5</xref>,<xref ref-type="bibr" rid="R7">7</xref>,<xref ref-type="bibr" rid="R8">8</xref>]. One side effect of ICD therapy is the painful shock needed to terminate the life-threatening tachycardias. Different studies showed a reduction of the ICD therapy by an additional antiarrhythmic therapy [<xref ref-type="bibr" rid="R10">10</xref>-<xref ref-type="bibr" rid="R14">14</xref>].</p><p>Patients having a reduced left ventricular function had worsening of their cardiac function with higher percentage of ventricular pacing [<xref ref-type="bibr" rid="R15">15</xref>-<xref ref-type="bibr" rid="R18">18</xref>]). Physiologic pacing or no pacing has to be preferred in patients with lower cardiac function and ICD implantation because of worsening of their cardiac function by ventricular pacing. But the accompanying antiarrhythmic therapy (e.g. sotalol, amiodarone, β-blockers) can cause an AV-block or a chronotropic incompetence resulting in a higher percentage of ventricular pacing. Therefore patients with accompanying antiarrhythmic therapy for preventing painful shocks should be paced physiologically or not paced at all to prevent a worsening of their cardiac function.</p><p>The aim of the retrospective investigation was to compare DDD(R) versus VVI(R) pacing on subjective (NYHA classification) and objective [brain natriuretic peptide (BNP), 6 minute walk test, echocardiography] parameters in patients after a required upgrading of a single to a dual chamber ICD for physiologic pacing.</p></sec><sec sec-type="methods" id="s2"><title>Methods</title><p>We examined 124 patients with different heart diseases and implanted single chamber ICD since 1998 in our study. The ICD was implanted because of a VT (42 patients), VF (59 patients) or a primary indication (23 patients) due to the MADIT II criteria [<xref ref-type="bibr" rid="R4">4</xref>]. The ICD had to be implanted at least six months before the inclusion into our study without a pacing indication at the time of the implant. The ICD was programmed to 50 ppm in the VVI mode at the beginning of our retrospective study. We used the interrogable data of the ICD to look for the percentage of ventricular pacing every three months.</p><p>We monitored every patient who had an increase of NYHA classification for the indication of upgrading the ICD since 2000 with echocardiography and ECG. The indication for a resynchronization therapy was an intrinsic QRS ≥ 120ms and an asynchrony on echocardiography.</p><p>The indication for upgrading to a DDD-ICD was a chronotropic incompetence and a high percentage of ventricular pacing (>70%) corresponding with an increase of NYHA classification. Patients (12 patients) with a chronotropic incompetence and >70% ventricular pacing were programmed to VVIR pacing. When the patients (12 patients) were still in the increased NYHA classification the ICD was upgraded to a DDD-ICD.</p><p>Patients were excluded in case of an indication for resynchronization therapy (16 patients), second (no patient) or third (no patient) degree AV block, an inability to walk (no patient) or a life expectancy below 6 months (no patient). The inclusion criterion was an upgradation from single to dual-chamber ICD because of the indication of pacing like chronotropic incompetence. In addition the medication had to be unchanged for 3 months.</p><p>Demographic data, medications and medical history were gathered by a patient interview and the review of their medical record at baseline and after six and twelve months. The DDD-ICD was programmed to 50 ppm with rate-responsive pacing, an AV-interval which was 20 ms longer than the intrinsic AV-interval and an AV hysteresis for reducing ventricular pacing (DDDR-50; n = 12). The interrogable data of the ICD were analysed for the percentage of ventricular pacing every three months.</p><p>The subjective (NYHA classification) and objective parameters [brain natriuretic peptide (BNP), 6 minute walk test, echocardiography] were evaluated at baseline (within one week before upgrading to a dual-chamber ICD) after six and twelve months in DDD(R)-pacing which were routine clinical practice in our clinic. The 6-minute walk tests were performed randomly during working hours at baseline, after 6 and 12 months. Patients were instructed to walk as far as possible within 6 minutes with a running wheel, with standardized encouragement and breaks when necessary. At the time of echocardiographic evaluation the physician was blinded for the ICD pacing mode. BNP measurement (Triage Meter Plus®, Biosite GmbH, Willich, Germany) was performed randomly during working hours at baseline, after 6 and 12 months.</p><sec id="s2a"><title>Echocardiography</title><p>The physicians were blinded to the mode of the ICD during the study. Biplanar left ventricular end-diastolic and end-systolic cavity volumes were calculated using Simpson's rule [<xref ref-type="bibr" rid="R19">19</xref>] from paired apical four-chamber and apical long-axis echocardiographic images of a minimum of five cardiac cycles; mean values of each variable were estimated. Biplanar ejection fractions were calculated as End-diastolic volume - End systolic volume / End-diastolic volume x 100% [<xref ref-type="bibr" rid="R22">22</xref>]. Pulsed Doppler analysis of mitral inflow included measurements for maximal E and A velocities, E/A ratio. The mean of five measurements was taken as the result. Doppler colour flow mapping was used to identify the presence or absence of mitral valve regurgitation. Gain settings were optimized by reducing the gain to the point where background noise disappeared. The direction of the MR jet was assessed from both parasternal and apical views, and the area of the largest clearly definable colour flow disturbance was traced in each view as an index of the severity of mitral valve regurgitation [<xref ref-type="bibr" rid="R21">21</xref>]. Left atrial size was measured by M-mode and two-dimensional echocardiography in all patients with VVI(R) pacing and 6 months after DDD(R) pacing using the Phillips ultrasonic device (3.5 MHz; model Sonos 5500, Philips Medical System, Andover, Massachusetts, USA).</p></sec><sec id="s2b"><title>Statistics</title><p>All data are presented as mean values ± standard deviation and as percentages for categorical variables. Data sets were tested with regard to normal distribution. For comparison between baseline and 6 months follow-up, the two-sided Friedman ANOVA test was used for NYHA classification, echocardiographic parameters, 6-minute walk test, BNP and ventricular pacing. A measurement of the linear association between two variables was evaluated using Pearson correlation coefficient. A multivariate Cox regression analysis was performed on variables found to be significant predictors (p < 0.1) with an univariate analysis for upgrading of the ICD. All statistical tests were two-tailed. A P value < 0.05 was considered as statistically significant. SPSS 12.0 for Windows was used as the statistical package.</p></sec></sec><sec id="s3"><title>Results</title><p>The ICD had to be upgraded because of chronotropic incompetence in 12 patients. These patients had a higher incidence of amiodarone therapy (<xref ref-type="table" rid="T1">Table 1</xref>). Defibrillator systems were manufactured by Biotronik GmbH & Co (Berlin, Germany), Guidant Corp. (St. Paul, MN), Medtronic Inc. (Minneapolis, MN), (alphabetical order). The ICD was implanted because of a VT (8 patients), VF (2 patients) or a primary indication (2 patients) due to the MADIT II criteria [<xref ref-type="bibr" rid="R4">4</xref>] without indication for a DDD-ICD at implantation. The clinical characteristics of the patients are shown in <xref ref-type="table" rid="T1">Table 1</xref>.</p><p>The patients received an additional antiarrhythmic therapy in the follow-up because of paroxysmal atrial fibrillation with inappropriate shocks (3 patients) and repeated VTs with appropriate shocks (9 patients). Paroxysmal atrial fibrillation was no longer observed during the further follow-up after the additional antiarrhythmic therapy with amiodarone in the three patients with inappropriate shocks.</p><p>The increase of NYHA classification occurred 11.6 ± 2.3 months after receiving an additional antiarrhythmic therapy caused by a sinus bradycardia and programming rate-responsive pacing function. The heart rate decreased from 59.0 ± 12.7 to 31.2 ± 5.2 (P < 0.0001) due to the antiarrhythmic therapy with amiodarone. The ICD had to be upgraded after a mean follow-up of 25.8 ± 13.7 months after ICD implant and nearly 85% ventricular pacing in all patients. The programmed AV-interval was 220.0 ± 10.4 ms.</p><p>A retrograde ventricular-atrial conduction was excluded with the dual chamber ICD as reason for the impaired effect of VVIR-and VVI-pacing.</p><sec id="s3a"><title>Subjective parameters</title><p>The NYHA classification increased significantly from 1.5 ± 0.4 at implant to 2.6 < 0.8 nearly 12 months with >70% ventricular pacing after receiving amiodarone therapy. After 6 months (1.6 ± 0.6 vs. 2.6 < 0.8; P ± 0.0001) and after 12 months (1.4 ± 0.3 vs. 2.6 < 0.8, P < 0.0001) with DDD(R) pacing NYHA classification decreased significantly.</p></sec><sec id="s3b"><title>Objective parameters</title><p>After 6 months with DDD(R) pacing the BNP (410.4 ± 297.0 vs. 312.3 ± 213.6 pg/ml, P < 0.014) and 6 minute walk test (324.6 ± 93.3 vs. 374.7 ± 113.2 m, P < 0.013) improved in comparison to VVI(R) pacing. The 6-minute walk test (392.4 ± 91.4 vs. 324.6 ± 93.3 m, P < 0.001), and BNP (234.1 ± 73.5 vs. 410.4 ± 297.0 pg/ml, P < 0.001) improved further after 12 months.</p><p>The left ventricular endsystolic volume, left ventricular ejection fraction, left atrial size (and A-wave also improved significantly after 6 and 12 months with DDD(R) pacing (<xref ref-type="table" rid="T2">Table 2</xref>). The ventricular pacing was significantly reduced from 84.2 ± 18.1% to 1.1 ± 1.7 % (P < 0.0001) after 12 months with DDD(R) pacing. During the study the diuretic therapy was reduced in 8 (75%) patients. This reduction included a reduction of furosemide (67.5 ± 30.1 vs. 40.0 ± 28.9 mg, p < 0.054) and spironolactone (65.6 ± 22.9 vs. 37.5 ± 24.9 mg, p < 0.018). In the other four patients the medication was unchanged.</p></sec><sec id="s3c"><title>Correlation</title><p>We observed positive correlations between the reduction of ventricular pacing and improvement at the 6 minute walk test (r = 0.84, P < 0.001), NYHA classification (r = 0.79, P < 0.001), the improvement of the BNP (r = 0.62, P < 0.031), the left ventricular endsystolic volume (r = 0.69, P < 0.019) and left ventricular ejection fraction (r = 0.92, P < 0.001).</p><p>In the multivariate regression analysis amiodarone treatment (odds ratio 31.6; 95% Cl 4.26-122.05; p < 0.0001) and a ventricular pacing > 10% (odds ratio 129.7; 95% Cl 19.28-315.73; p < 0.0001) were independent parameters for chronotropic incompetence corresponding with the upgrading to a DDD-ICD.</p></sec></sec><sec id="s4"><title>Discussion</title><p>In the present study we observed a significant improvement of subjective (NYHA classification) and objective (BNP, 6 minute walk test, echocardiography) parameters after 12 months in DDD(R)-pacing compared with VVI(R)/VVI-pacing. The improvements correlated with a reduction of ventricular pacing. The additional implantation of an atrial lead and a programmed long AV-interval and AV hysteresis caused a reduction of ventricular pacing. We believe that the improvement was caused by atrioventricular synchrony and a larger part by reduction of ventricular pacing.</p><sec id="s4a"><title>Comparison of VVI vs DDD pacing</title><p>All patients received a single chamber ICD for primary or secondary prevention of sudden cardiac death. The ICD had to be upgraded because of nearly 100% ventricular pacing corresponding with an increase of NYHA classification in our study. Former studies also observed an unanticipated necessity of pacing during the follow-up [<xref ref-type="bibr" rid="R22">22</xref>,<xref ref-type="bibr" rid="R24">24</xref>]. We showed an improved of NYHA-classification, BNP, 6 minute walk test, left ventricular ejection fraction and left ventricular endsystolic volume after 6 and 12 months with DDD(R)-pacing compared with VVI(R)-pacing. The improvements were significantly correlated with the reduction of ventricular pacing. On the other side left atrial contraction was improved with physiologic pacing which became apparent by an increased A wave and the reduction of left atrial size.</p><p>The DAVID Trial confirmed the disadvantage of DDD(R)-pacing regarding a higher mortality and hospitalization for congestive heart failure whether ventricular pacing was necessary or not [<xref ref-type="bibr" rid="R15">15</xref>]. Because of the high incidence of ventricular pacing (55.7%) the left ventricular function worsened in the DAVID Trail due to programmed rate-responsive pacing at 70/min for DDD pacing [<xref ref-type="bibr" rid="R15">15</xref>-<xref ref-type="bibr" rid="R18">18</xref>].</p><p>In our study the improvements were achieved by a reduction of ventricular pacing. The histological alterations [<xref ref-type="bibr" rid="R25">25</xref>,<xref ref-type="bibr" rid="R26">26</xref>] as a result of ventricular pacing were shown by myofibril disarray [<xref ref-type="bibr" rid="R25">25</xref>,<xref ref-type="bibr" rid="R26">26</xref>] followed by an impaired cardiac function DDD pacing [<xref ref-type="bibr" rid="R15">15</xref>-<xref ref-type="bibr" rid="R18">18</xref>].</p></sec><sec id="s4b"><title>Echocardiography</title><p>The advantage of physiologic pacing was shown by different echocardiographic parameters. By means of synchrony of atrial and ventricular contraction a reduction of left ventricular endsystolic volume and left atrial size was accomplished DDD pacing [<xref ref-type="bibr" rid="R27">27</xref>]. In addition the increase of the A wave suggested an improved left atrial contraction with physiologic pacing DDD pacing [<xref ref-type="bibr" rid="R28">15</xref>,<xref ref-type="bibr" rid="R29">29</xref>]. As a result of these improvements we observed an increase of the left ventricular ejection fraction like former studies DDD pacing [<xref ref-type="bibr" rid="R27">27</xref>-<xref ref-type="bibr" rid="R31">31</xref>].</p><p>VVI pacing results in an impairment of the left ventricular function because of a loss of atrioventricular synchrony in patients with heart failure [<xref ref-type="bibr" rid="R27">27</xref>,<xref ref-type="bibr" rid="R29">29</xref>,<xref ref-type="bibr" rid="R31">31</xref>]. or by the pacemaker syndrome [<xref ref-type="bibr" rid="R32">32</xref>]. The loss of the atrioventricular synchrony is reversible in time due to physiologic pacing [<xref ref-type="bibr" rid="R28">28</xref>]. These parameters can be evaluated by echocardiography, which is an appropriate examination for the course of patients with pacemaker or ICD.</p></sec></sec><sec id="s5"><title>Limitation</title><p>The examination of a small study sample reduced the statistical significance of our results. In addition this was a short-term observational study and long-term outcomes are unknown in our trial. But similar results can be found in former studies with larger patients sample size [<xref ref-type="bibr" rid="R15">15</xref>-<xref ref-type="bibr" rid="R18">18</xref>,<xref ref-type="bibr" rid="R22">22</xref>-<xref ref-type="bibr" rid="R24">24</xref>,<xref ref-type="bibr" rid="R27">27</xref>-<xref ref-type="bibr" rid="R31">31</xref>]. The influence of DDD(R) pacing on the incidence of atrial or ventricular tachyarrhythmias can not be estimated because the observation period between DDD(R) and VVI(R) pacing was too short. We could not fully achieve the blinding of the echocardiographer because the additional lead was visible in the right atrium and the difference between VVI pacing and AV synchrony was also apparent.</p></sec><sec id="s6"><title>Conclusions</title><p>The results of our study suggest that physiologic pacing improved subjective (NYHA classification) and objective (BNP, 6 minute walk test, echocardiography) parameters of patients with a new pacing indication. Physiologic pacing improves atrial and ventricular function in comparison to VVI(R) pacing. The reduction of ventricular pacing was the reason for these improvements. Even though the attention has been directed towards the cardiac resynchronization therapy for the improvement of heart failure we could achieve a significant enhancement of heart failure due to a reduction of ventricular pacing with increased physiologic pacing with a long AV-interval. In addition the indication for upgrading to a dual chamber ICD was accomplished in a small group of patients (10%). Thus in the most cases the implantation of a single chamber ICD was adequate. But this problem is going to accompany the clinical work. Amiodarone seems to be the risk factor for the upgrading of the ICD according to high incidence of ventricular pacing because of its antitachycardia effect.</p></sec>
|
Checkpoint control of mitotic exit—do budding yeast mind the GAP?
|
<p>Cell cycle checkpoints can delay mitotic exit in budding yeast. The master controller is the small GTPase Tem1, with inputs from a proposed guanine nucleotide exchange factor (GEF), Lte1, and a GTPase-activating protein (GAP), Bub2/Bfa1. In this issue, Fraschini et al. (p. 335) show that GAP activity of Bub2/Bfa1 appears to be dispensable for inactivation of Tem1 in cells. Their results call into question the GTP/GDP switch model for Tem1 activity, as have other results in the past. The paper also focuses attention on the two spindle pole bodies as potential sites for regulation of Tem1.</p>
|
<contrib contrib-type="author"><name><surname>Cooper</surname><given-names>John A.</given-names></name></contrib><contrib contrib-type="author"><name><surname>Nelson</surname><given-names>Scott A.</given-names></name></contrib>
|
The Journal of Cell Biology
|
<p>During the cell cycle, budding yeast are able to monitor DNA replication and repair, as well as mitotic spindle assembly and position. If one of these processes has not finished, the cell senses this and delays exit from mitosis, providing extra time to remedy the situation (<xref rid="fig1" ref-type="fig">Fig. 1</xref>). The master controller for the decision to exit appears to be a small G-protein, Tem1 (<xref rid="bib16" ref-type="bibr">Shirayama et al., 1994</xref>). Tem1 appears to collect inputs from various sensors that monitor these processes, integrate that information, and then notify the mitotic exit network (MEN), a cascade of signaling proteins, when it is safe to go ahead and finish mitosis. For a recent review of this area, see <xref rid="bib15" ref-type="bibr">Seshan and Amon (2004)</xref>.</p><fig id="fig1" position="float"><label>Figure 1.</label><caption><p>
<bold>Stages in progression through mitosis.</bold> T/B is Tem1 complexed with Bub2/Bfa1, and the drawing illustrates the timing of their location on SPBs, with respect to spindle position and mitotic exit. In this case, failure of the spindle to move into the neck is what activates the checkpoint. The diffuse cytoplasmic pools of the components are not indicated, but they are likely to be important, as discussed in the text. Over time during anaphase, Tem1 accumulates on the D-SPB, relative to the M-SPB, while Bub2/Bfa1 does the opposite (Molk et al., 2004).</p></caption><graphic xlink:href="jcb1720331f01"/></fig><p>A switch model for Tem1 has been proposed, with the GTP-bound state as ON and promoting mitotic exit and the GDP-bound state as OFF and having no effect on mitotic exit (<xref rid="fig2" ref-type="fig">Fig. 2 A</xref>). The ON to OFF transition is proposed to be accelerated by a GTPase-activating protein (GAP), the heterodimer of Bub2 with Bfa1 (<xref rid="bib8" ref-type="bibr">Geymonat et al., 2002</xref>). In cells, Bub2/Bfa1 clearly inhibits mitotic exit, via Tem1 (<xref rid="bib1" ref-type="bibr">Bardin et al., 2000</xref>; <xref rid="bib2" ref-type="bibr">Bloecher et al., 2000</xref>; <xref rid="bib13" ref-type="bibr">Pereira et al., 2000</xref>), and, in vitro, Bub2/Bfa1 increases GTP hydrolysis by Tem1 (<xref rid="bib8" ref-type="bibr">Geymonat et al., 2002</xref>). A new study by Fraschini and colleagues in this issue (p. 335) challenges the idea that Bub2/Bfa1 acts as a GAP on Tem1 in cells, based on the finding that the GAP activity of Bub2/Bfa1 appears to be dispensable for inhibiting mitotic exit.</p><fig id="fig2" position="float"><label>Figure 2.</label><caption><p>
<bold>Schematics of possible models for how Tem1 controls mitotic exit.</bold> (A) GTP/GDP switch model. (B) Effector model.</p></caption><graphic xlink:href="jcb1720331f02"/></fig><p>This new study also focuses attention on the mother-bound spindle pole body (M-SPB) as a potential location for interactions that control Tem1 activity, whatever they may be. Several previous observations implicate the daughter-bound SPB (D-SPB) as potentially important. Passage of the D-SPB through the neck appears to be a critical event that sets the cellular clock ticking on the course for mitotic exit (<xref rid="bib11" ref-type="bibr">Molk et al., 2004</xref>). During the course of a normal cell cycle, Tem1 accumulates on the D-SPB, along with active MEN components (<xref rid="bib11" ref-type="bibr">Molk et al., 2004</xref>). Bub2/Bfa1 first accumulates and then is lost from the D-SPB, as one might expect an inhibitor to behave.</p><p>In this new work, <xref rid="bib7" ref-type="bibr">Fraschini et al. (2006)</xref> found that a Myc-tagged version of Bub2 was anomalously localized to both SPBs throughout the cell cycle, in contrast to the normal behavior of untagged Bub2 or HA-tagged Bub2, which disappear from the M-SPB over time. Expression of this Myc-tagged Bub2 produced inhibition of mitotic exit, in an appropriate sensitized background. Genetically, the effect was dominant and worked through Tem1. The investigators logically presumed that Bub2-Myc/Bfa1 was exerting excessive GAP activity on Tem1 at the M-SPB, but, surprisingly, they found that the GAP activity of Bub2-Myc/Bfa1 was undetectable in vitro, even with Tem1 as the substrate. Thus, Bub2-Myc displayed loss of function in terms of GAP activity coupled with gain of function in terms of Tem1 inhibition. In addition, Bub2-Myc can inhibit mitotic exit when the spindle assembly checkpoint is activated, shown in previous work by the same group (<xref rid="bib6" ref-type="bibr">Fraschini et al., 1999</xref>). At face value, the results argue that Bub2/Bfa1 inhibits Tem1 in cells by a biochemical mechanism other than acceleration of GTP hydrolysis.</p><p>The authors then deliberately targeted Bub2's GAP activity by mutating a conserved Arg residue in the proposed catalytic site. This mutant, Bub2-R85A, also had undetectable GAP activity in vitro, as expected. However, in cells, Bub2-R85A was not able to inhibit Tem1 and thereby delay mitotic exit when called upon by a checkpoint, in contrast to Bub2-Myc. At this point, one might defend the GAP hypothesis by simply proposing that Bub2-R85A has much less GAP activity than Bub2-Myc, and that the low activity levels of both proteins happen to be undetectable in this particular biochemical assay. On the other hand, in cells, Bub2-R85A showed an important difference compared with Bub-Myc. Bub2-R85A did not recruit Bfa1 to the SPB, while Bub2-Myc did. Bub2 and Bfa1 are both necessary to inhibit Tem1, so the R85 residue of Bub2 may simply be necessary for Bub2 to interact with Bfa1 at the SPB, and thus the heterodimer cannot function. Bub2-R85A did localize to and persist at the M-SPB, as did Bub2-Myc but not wt Bub2. Based on these results, the authors suggest that Bub2's GAP activity might be required for Bub2/Bfa1 to leave the M-SPB, in support of the hypothesis that Bub2/Bfa1 persistence at the M-SPB can inhibit Tem1 and delay mitotic exit.</p><p>Other previous results, some of which are also admittedly negative, question whether the GTP/GDP switch model explains the cellular action of Tem1. First, a Tem1 point mutation analogous to Ras Q61L, which should be locked in the GTP state and therefore constitutively active, had no obvious effect on cell growth (<xref rid="bib16" ref-type="bibr">Shirayama et al., 1994</xref>). However, this mutation remains to be tested in a setting where mitotic exit is delayed. Second, an apparent guanine nucleotide exchange factor (GEF) domain can be found in the sequence of Lte1, a protein that promotes mitotic exit in cells by activating Tem1 (<xref rid="bib1" ref-type="bibr">Bardin et al., 2000</xref>). However, removing Lte1's GEF domain has little or no effect on Lte1's ability to promote mitotic exit (<xref rid="bib10" ref-type="bibr">Jensen et al., 2002</xref>; <xref rid="bib20" ref-type="bibr">Yoshida et al., 2003</xref>). Third, Tem1's intrinsic rates of GTP hydrolysis and release are high (∼0.1–0.2/min; <xref rid="bib8" ref-type="bibr">Geymonat et al., 2002</xref>), so that increasing them may not be useful.</p><p>What alternative mechanisms might one consider? Strong evidence, especially genetic analysis, shows that Bub2/Bfa1 and Lte1 antagonize each other, that they work through Tem1, and that each one is important for cells to thrive, at least under conditions in the wild. For example, at low temperatures, Lte1 is essential for growth and was named as such—low temperature essential (<xref rid="bib19" ref-type="bibr">Wickner et al., 1987</xref>). What, then, is the active state of Tem1, the state that drives mitotic exit? A mutant lacking Bub2 and Lte1 is viable, under optimal lab conditions, so free Tem1 should be sufficient for mitotic exit (<xref rid="bib9" ref-type="bibr">Hofken and Schiebel, 2002</xref>; <xref rid="bib17" ref-type="bibr">Stegmeier et al., 2002</xref>). Perhaps Bub2/Bfa1 sequesters Tem1 in an inactive state, while free Tem1 and Lte1-bound Tem1 bind effectors to activate the MEN (<xref rid="fig2" ref-type="fig">Fig. 2 B</xref>). The possibility that the Tem1/Bub2/Bfa1 complex has an independent inhibitory effect on the MEN has not been excluded. Finally, the recent discovery of phosphorylation of Tem1 provides a new factor to consider (<xref rid="bib18" ref-type="bibr">Wang and Ng, 2006</xref>).</p><p>Localization of components has helped to formulate and test models (<xref rid="fig1" ref-type="fig">Fig. 1</xref>). As noted above, several results argue that the D-SPB may be a crucial site for regulation. In addition, the activator Lte1 is confined to the bud, positioning it to activate the Tem1 of the D-SPB as the spindle enters the neck (<xref rid="bib1" ref-type="bibr">Bardin et al., 2000</xref>). The D-SPB often strikes the cortex, where Lte1 is heavily concentrated, but this event does not correlate with the timing of mitotic exit and thus may be incidental (<xref rid="bib11" ref-type="bibr">Molk et al., 2004</xref>). The Lte1 in the bud cytoplasm may be the form that interacts with Tem1 (<xref rid="bib4" ref-type="bibr">Castillon et al., 2003</xref>).</p><p>The results of <xref rid="bib7" ref-type="bibr">Fraschini et al. (2006)</xref> suggest that the mother-bound SPB (M-SPB) may also be important, in that the presence of Bub2-Myc/Bfa1 at the M-SPB inhibited mitotic exit, via Tem1. In support of this idea, when mitotic exit is delayed by a checkpoint, Bub2/Bfa1 persists at the M-SPB (<xref rid="bib14" ref-type="bibr">Pereira et al., 2001</xref>). In addition, a novel kinase, Kin4, which is confined to the mother cortex, has been found to inhibit mitotic exit via Bub2/Bfa1 (<xref rid="bib5" ref-type="bibr">D'Aquino et al., 2005</xref>; <xref rid="bib12" ref-type="bibr">Pereira and Schiebel, 2005</xref>). One may need to consider every cellular location of Tem1 as a potentially important place where regulation of Tem1 activity can occur, because Tem1 molecules exchange rapidly between SPB and cytoplasm (<xref rid="bib11" ref-type="bibr">Molk et al., 2004</xref>). The sum of the inhibiting and activating effects on Tem1 in all its cellular pools may be what tips the balance. On the other hand, scaffolding mechanisms may activate Tem1 and promote its interactions with downstream effectors more effectively in certain locations. A computational analysis of the system should be helpful at this point, perhaps necessary, given the complexity of the signaling pathways (<xref rid="bib3" ref-type="bibr">Bosl and Li, 2005</xref>).</p><p>In sum, the mechanism for controlling the timing of mitotic exit may not be the obvious one suggested by the protein sequences and biochemical activities in vitro. The addition of protein biochemistry to the toolbox of the yeast cell biologist is helping the field to test molecular mechanisms for mitotic exit in new ways, with some unexpected results. Our understanding of what happens in the cell has many gaps to be filled, including the role of the GAP at hand.</p>
|
An ELISA-based procedure for assaying proteins in digests of human leukocytes and cell lines, using specifically selected peptides and appropriate antibodies
|
<sec><title>Background</title><p>We describe the application of an ELISA-based assay (the Peptidomatrix) that can be used to simultaneously identify and quantitate a number of proteins in biological samples. The biological sample (blood component, biopsy, culture or other) is first lysed to release all the proteins, without any additional separation. The denatured proteins in the sample are then digested in bulk with the desired proteolytic enzyme(s). The peptides in the digest are then assayed by appropriate antibodies, using a competition ELISA protocol.</p></sec><sec><title>Results</title><p>As an example of its use, the present paper applies the Peptidomatrix to the assay of four membrane proteins MDR1 (P-glycoprotein or ABCB1), MRP1 (ABCC1), BCRP/MXR (ABCG2) and the alpha subunit of the Na, K_ATPase (ATP1A1), present in a number of cell lines and in human lymphocytes. We show that we can detect and quantitate these proteins, using a series of peptide-antibody pairs, and that we can differentiate between cell lines or cell preparations that express the target proteins and those that do not.</p></sec><sec><title>Conclusion</title><p>We have devised a simple, ELISA-based proteomics assay that enables the quantitation of designated proteins in a cell or tissue sample, and that can be used in any laboratory, with minimal specialized equipment.</p></sec>
|
<contrib id="A1" equal-contrib="yes" contrib-type="author"><name><surname>Braitbard</surname><given-names>Ori</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" equal-contrib="yes" contrib-type="author"><name><surname>Bishara-Shieban</surname><given-names>Janette</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Glickstein</surname><given-names>Hava</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Kott-Gutkowski</surname><given-names>Miriam</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Pace</surname><given-names>Umberto</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Rund</surname><given-names>Deborah G</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A7" corresp="yes" contrib-type="author"><name><surname>Stein</surname><given-names>Wilfred D</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
Proteome Science
|
<sec><title>Background</title><p>The revolution in biology initiated by the Genome Project is being further stimulated by research aimed at the elucidation of the proteome, the complement of proteins expressed by an organism [<xref ref-type="bibr" rid="B1">1</xref>]. Proteomics aims to develop methods for providing a total accounting of the proteins present in a biological sample, with all the valuable insights that will flow from achieving this aim [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B7">7</xref>] Much proteomics research employs techniques such as 2D gel electrophoresis for the separation of the protein mixtures followed by the use of HPLC and mass spectroscopy technology for the identification of the target proteins [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. However, there is a definite need for simpler and less costly methods that can identify a limited number of proteins in a biological sample, as needed in small clinical or research laboratories.</p><p>In this paper we describe an ELISA-based assay, the Peptidomatrix, based on a procedure which has been developed to identify and quantitate proteins in biopsies and other biological samples[<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. Since the principle of the assay is the use of peptides derived from a tryptic digest of the sample, it can be used on samples that have undergone denaturation. Thus, an advantage of the Peptidomatrix is that the procedure does not require that the target protein be present in its native form. In addition, no prior isolation and purification of the protein target is required for setting up the assay. All that is needed is knowledge of the sequence of the protein (or of the mRNA that codes for it).</p><p>The biological sample (blood fraction, biopsy, culture or other) is first lysed to release all the proteins, without any additional separation. The denatured proteins in the sample are then digested in bulk with the desired proteolytic enzyme(s). The peptides in the digest are then assayed by appropriate antibodies, using a competition ELISA protocol.</p><p>The Peptidomatrix assay is based on competition between a peptide derived from a proteolytic digest of the sample and an identical synthetic peptide, which has been pre-bound to the ELISA plate, for an appropriate antibody[<xref ref-type="bibr" rid="B10">10</xref>].</p></sec><sec><title>Results</title><p>The Peptidomatrix assay uses peptides that are chosen as being (i) specific for a target protein and (ii) present amongst the products of tryptic digestion of that protein. We subjected the membrane protein transporters MDR1 (or ABCB1 i.e. P-glycoprotein 1), MXR (or BCRP i.e. ABCG2) and MRP1 (ABCC1) the alpha chain of Na, K-ATPase (ATP1A1) to a virtual tryptic digestion and from the products of digestion, we selected the peptides of length 7 to 15 amino-acids. Each one of these peptides was analyzed using the BLAST program (see methods). A desirable peptide contains only matches that are 5 amino acids or shorter, and a minimal number of them. The peptides chosen are listed in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>The peptides and antibodies used throughout this study. Note that all these peptides have a cysteine at their N terminus, which has been added for conjugating them to a carrier protein for the immunization.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Peptide name</td><td align="left">Sequences</td><td align="left">Location</td><td align="left">Serum from rabbit</td></tr></thead><tbody><tr><td align="left">MDR1 P1</td><td align="left"><bold>C</bold>QDVSWFDDPK</td><td align="left">799–808</td><td align="left">#64840, #64841</td></tr><tr><td align="left">MDR1 P3</td><td align="left"><bold>C</bold>SEIDALEMSSNDSR</td><td align="left">646–659</td><td align="left">#65158, #64541</td></tr><tr><td align="left">MDR1 P494</td><td align="left">MPNTLEGNVT<bold>K</bold></td><td align="left">1027–1036</td><td align="left">C494 (Commercial monoclonal antibody)</td></tr><tr><td></td><td></td><td></td><td></td></tr><tr><td align="left">MXR P1</td><td align="left"><bold>C</bold>VGTQFIR</td><td align="left">178–184</td><td align="left">#65158, #64541</td></tr><tr><td align="left">MXR P2</td><td align="left"><bold>C</bold>LAEIYVNSSFYK</td><td align="left">332–343</td><td align="left">#64795, #64850</td></tr><tr><td align="left">MXR P3</td><td align="left"><bold>C</bold>EISYTTSFCHQLR</td><td align="left">366–378</td><td align="left">#64795, #64850</td></tr><tr><td align="left">MXR P4</td><td align="left"><bold>C</bold>LFIHYISGYYR</td><td align="left">454–465</td><td align="left">#64853, #64851</td></tr><tr><td align="left">MXR P5</td><td align="left"><bold>C</bold>NDSTGIQNR</td><td align="left">418–426</td><td align="left">#64853, #64851</td></tr><tr><td></td><td></td><td></td><td></td></tr><tr><td align="left">MRP1 P1</td><td align="left"><bold>C</bold>PSDLLQQR</td><td align="left">1511–1518</td><td align="left">#A0151, #A0152</td></tr><tr><td align="left">MRP1 P2</td><td align="left"><bold>C</bold>DLWSLNK</td><td align="left">240–246</td><td align="left">#A0151, #A0152</td></tr><tr><td></td><td></td><td></td><td></td></tr><tr><td align="left">NaK ATPase P1</td><td align="left"><bold>C</bold>IPFNSTNK</td><td align="left">478–485</td><td align="left">#64845, #64846</td></tr><tr><td align="left">NaK ATPase P2</td><td align="left"><bold>C</bold>PTTPEWVK</td><td align="left">74–81</td><td align="left">#64845, #64846</td></tr><tr><td align="left">NaK ATPase P3</td><td align="left"><bold>C</bold>TGTLTQNR</td><td align="left">368–375</td><td align="left">#64855, #64856</td></tr><tr><td align="left">NaK ATPase P4</td><td align="left"><bold>C</bold>YEPAAVSE</td><td align="left">11–07</td><td align="left">#64855, #64856</td></tr></tbody></table></table-wrap><p>The positions of these peptides in the primary sequence of each of the four proteins are shown in Fig <xref ref-type="fig" rid="F1">1</xref>. Polyclonal antibodies were generated in rabbits for all of these peptides except in the case of the peptide denoted as P494 for which a commercial monoclonal antibody (C494) was available.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Structure of the proteins detected by the Peptidomatrix</bold>. The ATP binding site for the ABC transporters and the location of the peptides selected for the Peptidomatrix are indicated.</p></caption><graphic xlink:href="1477-5956-4-14-1"/></fig><p>The Peptidomatrix assay, as described in the Introduction, is a competition assay (Figure <xref ref-type="fig" rid="F2">2</xref>): Peptides are bound to plastic wells. Antibodies specific to the peptide are then added in solution and allowed to bind to the attached peptide in the presence or the absence of a sample digest. A calibration curve is generated in parallel with known quantities of free synthetic peptide. The concentration of soluble peptide in the sample is then measured by interpolation.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>The procedure of the 'Peptidomatrix' assay</bold>. The ELISA Peptidomatrix is a competition assay: Peptides are bound to the plastic wells and they are subsequently exposed to antibodies generated against them. A labeled secondary antibody is used to detect the bound primary antibodies. In the assay itself the antibodies are mixed with soluble synthetic peptide in known amounts, to generate a calibration curve. The peptide in solution competes with the bound peptide and will affect the signal. As shown schematically in the figure: Panel A, A': No free peptide is added, thus the signal will be the highest. Panel B, B': A small amount of free peptide is added and the signal will be intermediate. Panel C, C': A large amount of competing peptide is added, thus no antibody binds to the well and the signal is minimal.</p></caption><graphic xlink:href="1477-5956-4-14-2"/></fig><p>Figure <xref ref-type="fig" rid="F3">3</xref> shows a typical experiment in which peptide P494 (from MDR1) and monoclonal antibody C494 were used to detect and quantify the peptide in CEM cells, in digests prepared from either wild-type cells (WT) or from multidrug resistant cells (Col1000). In panel A the calibration curve depicting the ELISA signal as function of free synthetic peptide is shown as a solid line. This curve is a descending hyperbola, since the more free peptide is added the lower the signal. In parallel, aliquots of the cell digest were assayed in the same way. Also in this case, the more peptide present in the digest, the lower the signal will be. In Figure <xref ref-type="fig" rid="F3">3A</xref> the signals from digests of WT and Col1000 CEM cells are shown (solid and empty triangles). The concentration of peptide in the digest is calculated by interpolation on the calibration curve. A simple calculation (see Materials and Methods) yields an estimate of the concentration of peptide (in ng/ml) present in the digest. Panel B of Fig <xref ref-type="fig" rid="F3">3</xref> summarizes the results of several experiments such as those shown in panel A. In each experiment, we added increasing amounts of the tryptic digest of the wild type or drug resistant CEM cells, estimated the amounts of peptide in the sample and plotted these against the amount of added digest. The amount of peptide detected in the drug resistant cells increases with increasing amounts of digest while the level in the wild type CEM cells remains close to zero.</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Detection of the MDR1 peptide P494</bold>. In panel A calibration curves are plotted as optical density (OD450) vs. peptide concentration. The curve is the calculated regression fit, based on the hyperbolic equation detailed in the Materials and Methods section. The OD values of 50 μl of cell digest (approximately 4.5 million cells) from wild type and drug-resistant CEM cells are shown as solid and empty triangles respectively. For the sake of clarity we have added arrows to indicate the position of these points on the curve. In panels B the amounts of peptide (in nanograms) for a range of digest volumes (from 6.25 μl to 50 μl) are shown. These values have been calculated using the regression curve of panels A.</p></caption><graphic xlink:href="1477-5956-4-14-3"/></fig><p>Fig <xref ref-type="fig" rid="F4">4</xref> depicts the means and standard errors of 15 experiments performed as in Fig <xref ref-type="fig" rid="F3">3</xref>, with the data now presented as ng of peptide detected per millions of cells. There is a very clear difference between the data set for the drug resistant Col1000 cells and the parent CEM cells. The plot for the Col1000 cells is not linear, presumably because the wells hold only a limited amount of the bound peptide. Fitting these data to a hyperbola (i.e. Michaelis-Menten equation) gave an estimate of the initial slope of 19.7 ± 2.5 ng peptide per million cells and a half-saturation value of 5.2 ± 1.7 million cells. With a molecular weight of 1158.31 for the peptide, this slope translates to 10.2 million P-glycoprotein molecules per Col1000 cell.</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Average (± SE) of 15 experiments detecting P494 in CEM cells</bold>. The abscissa values have been recalculated in millions of cells, in order to deduce the number of PgP molecules per cell (see text for details).</p></caption><graphic xlink:href="1477-5956-4-14-4"/></fig><p>Interestingly, titration curves such as Fig <xref ref-type="fig" rid="F3">3A</xref> enable another estimate to be made of the number of P-glycoprotein molecules per Col1000 cell. The average of the half-saturation values for the 15 calibration curves performed to generate Fig <xref ref-type="fig" rid="F4">4</xref> was 0.678 ± 0.0678 (SE, n = 15) μg/ml peptide. Since each well receives 100 μl of peptide solution, there is 68 ± 6.8 ng peptide in the average well at which the antibody is one-half saturated. In Fig <xref ref-type="fig" rid="F4">4</xref>, half-saturation is reached at the peptide present in 5.2 ± 1.7 million cells. Equating these two values, one can conclude that there are 68/5.2 or 13.0 ± 4.4 ng peptide per million cells (or 6.7 million P-glycoprotein molecules per Col1000 cell). This and the estimate of 19.7 ± 2.5 ng per well found from the initial slope of the plot of Fig <xref ref-type="fig" rid="F4">4</xref> are less than 1 SE unit apart.</p><p>We chose to apply the Peptidomatrix protocol to the Na, K-ATPase in order to provide a "calibrating protein" that would be present in every animal cell type and would enable the Peptidomatrix user to normalize data in terms of this protein. It is also a fitting control in this case since this protein share a large similarity in structure and cellular location with the other target proteins. Fig <xref ref-type="fig" rid="F5">5</xref> depicts such an experiment in which, on the same ELISA plate, samples of a digest of leukemia cells (from patient ME) were assayed for MDR1 using the P3 peptide and the appropriate polyclonal antibody and, in parallel, were assayed for the Na, K-ATPase using the P4 peptide and appropriate polyclonal antibody. Parts A and B of this figure show the peptide concentration obtained (with the procedure of Fig <xref ref-type="fig" rid="F3">3</xref>) with these two proteins separately, while Fig <xref ref-type="fig" rid="F5">5C</xref> shows the level of MDR1 P3 plotted against that of Na/K-ATPase P4. The regression through the points has a slope of 0.65, indicating that on a mole-for-mole basis, these cells membranes contain three molecules of the sodium pump for every two MDR1 molecules.</p><fig position="float" id="F5"><label>Figure 5</label><caption><p><bold>Simultaneous detection of Na/K ATPase and MDR1 proteins in cells from a leukemia patient</bold>. In Panel A and B the measured amounts of the peptides PgP P3 and ATPase P4 (in ng/well) are plotted as function of the amounts of cells. In panel C the same calculated amounts of peptides are plotted one against the other, allowing the calculation of the ATPase to PgP ratio over a range of cell amounts.</p></caption><graphic xlink:href="1477-5956-4-14-5"/></fig><p>Fig <xref ref-type="fig" rid="F6">6</xref> shows a summary compilation of data from this and five other experiments where this internal calibration protocol was applied to white blood cell samples taken from five different patients. In all these experiments, duplicate samples of cells from four or five patients were assayed on the same ELISA plate for both proteins. The ratio of P-glycoprotein polypeptide to the Na, K-ATPase polypeptide was computed for each patient sample and this ratio normalized to that of a particular reference sample (the cells from patient VR, mean ratio of PgP P3/ATPase P4 = 0.714 ± 0.237) for displaying in Fig <xref ref-type="fig" rid="F6">6</xref>. The amount of MDR1 is significantly higher (at p < 0.05) in the sample from patient SR than in those from patients VR and SH, and the normal donor, but not in that from ME. In one experiment of the series, a sample of the Col1000 drug-resistant cells was included on the ELISA plate. The ratio of MDR1 P3 peptide to ATPase P4 peptide in this sample was 13 times higher than that for the reference VR cells depicted in Fig <xref ref-type="fig" rid="F6">6</xref>. These cells, just like other cell lines, seem to have a very low level of Na, K-ATPase, which may not be a good normalizing control in this case.</p><fig position="float" id="F6"><label>Figure 6</label><caption><p><bold>Ratio of MDR1 P3 to Na/K ATPase P4 in white blood cell preparations from leukemia patients and normal controls</bold>. The average of at least 5 independent experiments is shown here. The amounts of the proteins were determined with the Peptidomatrix protocol and the ratios calculated in each experiment. In order to combine the experiments the data were normalized for one of the samples (VR, taken as reference).</p></caption><graphic xlink:href="1477-5956-4-14-6"/></fig><p>To test the validity of the Peptidomatrix protocol we compared the results obtained with an independent method of testing for MDR levels: a clinically established functional method based on the uptake of the dye Rhodamine 123 (rho123), used on clinical samples. In cells that express functional PgP, rho123 is rapidly and actively pumped out of the cell and, within a short period of time, marked differences in intracellular fluorescence are seen between cells that have or have not extruded rho123 via this ATP-dependent pump[<xref ref-type="bibr" rid="B11">11</xref>]. PgP-reversing drugs, like Verapamil, can be used in the rho123 retention assay to preferentially block PgP function[<xref ref-type="bibr" rid="B12">12</xref>].</p><p>The clinical samples were either fresh samples or frozen samples of white blood cells. All the lymphocyte preparations were from peripheral blood or from leukopheresis treatment. The frozen samples had been kept at -70°C for up to six years and the Rhodamine uptake assay had been performed and recorded at the time of the collection.</p><p>Multidrug sensitive CEM cells (wt) and resistant CEM cells (col) served as controls. Because counting cells in the frozen samples was not possible, the parameter for the number of cells that was used was the concentration of protein. The protein concentration was measured using the Bradford method before the lysis of the cells and precipitation of the protein. The MDR protein levels in the samples were determined by the Peptidomatrix using the MDR P3 peptide and are expressed as ng peptide per mg protein.</p><p>Table <xref ref-type="table" rid="T2">2</xref> shows a comparison of the results obtained from the Peptidomatrix with those from the functional analysis. Care was taken to ensure that the dates of the samples that had been tested by the functional assay and the frozen samples tested by the Peptidomatrix assay were matched, because of the known variation of the expression of the protein in the same patient at different times.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Summary of presence of P-glycoprotein in patient samples, as detected with the Peptidomatrix and the Rhodamine uptake assay.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="left">Patient</td><td align="left">Type of leukemia</td><td align="left">Peptidomatrix results (ng peptide/mg protein)</td><td align="left">Peptidomatrix results (Grade)</td><td align="left">Status by FACS (Grade)</td></tr></thead><tbody><tr><td></td><td align="left">CEM wt</td><td></td><td align="center">7</td><td align="left">Negative</td><td></td></tr><tr><td></td><td align="left">CEM col1000</td><td></td><td align="center">114</td><td align="left">Positive</td><td></td></tr><tr><td align="left">1</td><td align="left">AP</td><td align="left">AML</td><td align="center">72.5</td><td align="left">Positive (low)</td><td align="left">Positive</td></tr><tr><td align="left">2</td><td align="left">DA</td><td align="left">T-ALL</td><td align="center">26.37</td><td align="left">Positive (low)</td><td align="left">Negative</td></tr><tr><td align="left">3</td><td align="left">GB</td><td align="left">t-AML</td><td align="center">150</td><td align="left">Positive (high)</td><td align="left">Positive</td></tr><tr><td align="left">4</td><td align="left">KA</td><td align="left">T-ALL</td><td align="center">0</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">5</td><td align="left">RE</td><td align="left">AML</td><td align="center">144</td><td align="left">Positive (high)</td><td align="left">Positive</td></tr><tr><td align="left">6</td><td align="left">WE</td><td align="left">AML</td><td align="center">12.2</td><td align="left">Positive (low)</td><td align="left">Positive (low)</td></tr><tr><td align="left">7</td><td align="left">SA</td><td align="left">AML</td><td align="center">7</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">8</td><td align="left">SN</td><td align="left">AML</td><td align="center">90.93</td><td align="left">Positive (low)</td><td align="left">Positive</td></tr><tr><td align="left">9</td><td align="left">SY</td><td align="left">CLL</td><td align="center">209</td><td align="left">Positive (high)</td><td align="left">Positive</td></tr><tr><td align="left">10</td><td align="left">WR</td><td align="left">APL</td><td align="center">4.9</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">11</td><td align="left">YN</td><td align="left">AML</td><td align="center">7.9</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">12</td><td align="left">CM</td><td align="left">AML</td><td align="center">6</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">13</td><td align="left">DM</td><td align="left">CML</td><td align="center">62.27</td><td align="left">Positive (low)</td><td align="left">Positive</td></tr><tr><td align="left">14</td><td align="left">GG</td><td align="left">AML</td><td align="center">4.8</td><td align="left">Negative</td><td align="left">Negative</td></tr><tr><td align="left">15</td><td align="left">SG</td><td align="left">AML</td><td align="center">56.4</td><td align="left">Positive (low)</td><td align="left">Positive</td></tr><tr><td align="left">16</td><td align="left">OZ</td><td align="left">AML</td><td align="center">38.5</td><td align="left">Positive (low)</td><td align="left">Positive (low)</td></tr></tbody></table><table-wrap-foot><p>The table compares the results that we have from the clinical functional method, Rhodamine 123 uptake and the Peptidomatrix assay (see text). The Rhodamine 123 uptake assay is based on the differential dye retention in the absence and presence of Verapamil, an inhibitor of Pgp. The MDR1 protein levels in the samples were determined by the Peptidomatrix using the MDR P3 peptide and are expressed as ng peptide per mg protein. Multidrug sensitive CEM cells (wt) and resistant CEM cells (col1000) served as controls. To allow a comparison with the semiquantitative data of the other assay a gradation of MDR1 levels was defined by comparison to the control cells: The samples that had levels similar to the wt cells ± 10 % (up to 8 ng/mg protein) were graded as negative, the samples that had levels similar to the col1000 cells (100–125 ng/mg) were graded as positive, the samples that had levels higher (> 125 ng/mg) or lower then the col1000 cells (8–100 ng/mg) were graded as high-positive or low-positive respectively. The data shown are averages of duplicates, the difference between the duplicates were below 5%.</p></table-wrap-foot></table-wrap><p>The results in Table <xref ref-type="table" rid="T2">2</xref> are summarized in qualitative terms (negative, low positive, positive and high positive) and their correlation was tested by using Pearson's correlation.</p><p>The Pearson's correlation between the Peptidomatrix assay and the rhodamine uptake assay was calculated as 0.79.</p><p>In order to validate the quantification of PgP by another method that is independent of the Peptidomatrix assay, PgP was also quantified by the ATPase activity assay and the results obtained by the two methods were compared.</p><p>The determination of the number of PgP molecules based on ATPase activity was performed on purified membranes obtained from CEM- col cells, using the ATPase activity associated with PgP, using both the basal activity and the verapamil (10 μM)-stimulated activity. In one typical experiment the basal ATPase activity was 23.2 mM Pi/μg protein/hour and the verapamil-stimulated activity was 82.7 mM Pi/μg protein/hour. As detailed in the Methods section the final volume of the liberated Pi determination was 260 μl. These data were converted to number of molecules/mg protein using the following turnover numbers: 2.9 s<sup>-1 </sup>for the basal activity and 9.9 s<sup>-1 </sup>for the verapamil-stimulated activity respectively[<xref ref-type="bibr" rid="B13">13</xref>]. The final calculated values were 3.48 × 10<sup>14 </sup>molecules PgP/mg protein and 3.63 × 10<sup>14 </sup>molecules/mg protein for the stimulated and the basal activity. The data were obtained from quadruplicate samples. To measure the PgP amount by Peptidomatrix an inhibition curve with synthetic peptide (PgP P3) was run alongside an analogous curve run with dilution of a tryptic digest of the same purified membranes used above. The Kd's (half saturation points) of the two curves were 2.87 × 10-2 μg/ml peptide and 3.11 μl digest respectively. Given that the concentration of the digest (determined prior to digestion) is 25 mg/ml, the molecular weight of the peptide is 1111 daltons, and the volume of the ELISA reaction is 100 μl, the amount of PgP in the membranes comes to 3.03 × 10<sup>14 </sup>molecules PgP/mg protein, a value very close to the one obtained from the ATPase assay. In a series of experiments using different preparations of membranes, the values for the number of molecules of Pgp per mg membrane protein obtained by the two methods were 5.49 ± 1.14 × 10<sup>14 </sup>(standard error n = 4) and 5.46 ± 1.73* 10<sup>14 </sup>(standard error n = 4) for the ATPase (combining the data from basal and verapamil stimulated) and Peptidomatrix respectively.</p><p>In fig <xref ref-type="fig" rid="F7">7</xref> Peptidomatrix data are shown for two additional membrane proteins, MRP1, studied in a drug sensitive cell line (VP16) and the parent MCF7 cell line (Fig <xref ref-type="fig" rid="F7">7A</xref> and <xref ref-type="fig" rid="F7">7B</xref>) and MXR, studied in drug-sensitive (AdVp) and wild type MCF7 cell lines (Fig <xref ref-type="fig" rid="F7">7C</xref>). The Peptidomatrix assay shows that the parental strain is low in MRP while the drug-resistant line is rich in this protein (Fig <xref ref-type="fig" rid="F7">7B</xref>).</p><fig position="float" id="F7"><label>Figure 7</label><caption><p><bold>Detection of the proteins MRP 1 and MXR</bold>. Panel A and B show an experiment detecting the protein MRP1 in wild type and transformed MCF7 cells, using the MRP P1 peptide and one of the specific antisera. Again a calibration curve and a calculated peptide amount vs. cell amount are shown. In panel C an experiment detecting the MXR protein in wild type and transformed MCF7 cells is shown. Similarly to Figure 3A we have marked the readings of wt and resistant cell digests with a solid and an empty triangle respectively.</p></caption><graphic xlink:href="1477-5956-4-14-7"/></fig><p>From the initial slope of the line drawn through the empty circles, it can be calculated that there are some 1.69 ± 0.38 million molecules of the MRP1 protein per single MCF-7 Vp 16 cell. The wild-type cell line had little if any of the MRP1 protein, by this assay.</p></sec><sec><title>Discussion</title><p>The recent advances in proteomics research have fostered the emergence of many different techniques for identifying and quantitating proteins from cells and tissues. In order to apply these scientific advances to the world of routine clinical diagnostics there is a need for a simple and easy-to-use method. We present here a possible answer to this need in the form of the Peptidomatrix, and describe, as an example, its use for identifying, and assaying quantitatively, four membrane proteins present in digests of wild type and multidrug-resistant cell lines and in white blood cells of leukemia patients. The assay is ELISA-based and uses antibodies directed against small peptides that have been selected as being (i) likely to be liberated by tryptic digestion of cells and (ii) unique to the particular protein under investigation. We found that many of the peptides that we synthesized on the basis of these criteria, as being suitable for use in the Peptidomatrix protocol, did give us useful peptide-antibody combinations. Most of the peptides that were designed for use with the proteins P-glycoprotein, MRP1, and MXR gave satisfactory differences in signal when wild type and drug resistant cell strains were compared (Figs <xref ref-type="fig" rid="F3">3</xref>–<xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F6">6</xref>–<xref ref-type="fig" rid="F7">7</xref>), suggesting that they were identifying a protein that was present in the drug resistant cells but not in the parental strains. Such a test could not be made for the peptide/antibody pairs made against the ubiquitous Na, K-ATPase. The protein for which we have most substantial data is MDR1 (also known as P-glycoprotein or ABCB1) that confers multi-drug resistance to cells in culture and is associated with drug resistance in cancer patients. For this protein, the peptide P494, the epitope of the monoclonal antibody C494, allows the specific identification of the protein in cell digests from a drug-resistant line of human T-lymphoblastic leukemia cells, the Col1000 cells. Fig <xref ref-type="fig" rid="F3">3</xref> exemplifies the competition assay that we use in the Peptidomatrix protocol. The data from individual experiments can be transformed into plots of peptide vs. number of cells. Fig <xref ref-type="fig" rid="F4">4</xref> summarized the results of 15 experiments using the Peptidomatrix protocol on these cells. Fitting the data by a linear hyperbola enables the initial slope of the curve to be determined. This initial slope is in units of nanograms of peptide present per million cells. Since the molecular weight of the peptide is known, the data can be translated into a number that is equivalent to the amount of P-glycoprotein molecules per cell. The number that we calculated is 10.2 million P-glycoprotein molecules per Col1000 cell. We wished to compare these values with other similar quantifications reported in the literature.</p><p>Although there have been very many studies of the expression of the MDR1 mRNA levels and of the protein in drug resistant cells, the latter using western blots or FACS analysis, we have been able to find only two quantitative studies that record the amount of P-glycoprotein present in such cells. One study[<xref ref-type="bibr" rid="B14">14</xref>] used the FACS technique and an antibody depletion method on three lines of mouse 3T3 cells. The highly resistant line (N3-2400) was found to have 14.2 × 10<sup>6 </sup>molecules of PgP/cell[<xref ref-type="bibr" rid="B15">15</xref>], a value quite close to the one that we detected in the highly resistant CEM cell. The other quantitative study of P-glycoprotein molecules per cell is by Maynadie and colleagues[<xref ref-type="bibr" rid="B16">16</xref>]. These authors also used a FACS technique, but calibrated the FACS signal with a set of beads labeled with a fluorescent dye[<xref ref-type="bibr" rid="B17">17</xref>]. They measured P-glycoprotein in a sensitive and drug resistant erythroleukemic cell line (the K562 and K562-ADR cells). These cells had 291 and 303,481 P-glycoprotein molecules per cell, respectively. It must be pointed out that the K562-ADR cells are indeed resistant to doxorubicin, but much less so than the CEM cells that we used in this study and the 3T3 line mentioned previously).)[<xref ref-type="bibr" rid="B18">18</xref>]. Maynadie and colleagues also reported data on human leukemic cells and found very variable numbers for their P-glycoprotein content, ranging from 117 to 10,947 (mean 2509 ± 2805) molecules per cell, in the cells from the 25 donors studied[<xref ref-type="bibr" rid="B16">16</xref>]. Our results seem to confirm the large individual variability in the amount of PgP in white blood cells. Comparison between the Peptidomatrix method and a clinically established functional method, Rhodamine 123 uptake, shows a high correlation (Pearson's correlation of 0.79) between the two methods (Table <xref ref-type="table" rid="T2">2</xref>). Peptidomatrix testing of MDR 1 protein in AML patients showed that 66% of them express the protein. This fits roughly with estimates in the literature, which show that 35–80 % of AML patients express the protein. This provides support for the reliability of Peptidomatrix as a diagnostic tool for identifying the presence or absence of the P-gp protein in patients.</p><p>The results obtained with Peptidomatrix were compared with several other accepted biochemical methods. Western blotting and RT-PCR, which provide only qualitative results, were fully consistent with the Peptidomatrix results (data not shown). We determined the number of P-glycoprotein molecules per cell, by measuring the ATPase activity of this protein and combining this with the known turnover number of the enzyme. This provided a quantitative result, enabling direct comparison with the Peptidomatrix. The specificity of the ATPase method for P-gp was ensured by including in the assay mixture inhibitors of all the irrelevant ATPases. The two procedures, Peptidomatrix and ATPase activity were in excellent accord. Thus, as far as can be assessed, the Peptidomatrix protocol provides a reliable method for identifying and quantitating the MDR 1 that was used in the present study.</p><p>An important point to discuss here is the general applicability of the Peptidomatrix and its ability to detect low-abundance proteins. In this study we worked with 4 different proteins, which, although similar in structure and cellular location have different sequences and distributions. The Peptidomatrix procedure, run with the appropriate peptide-antibody pair is clearly able to detect all of them. In experiments (not shown) in which we mismatched protein, peptide and antibody, we generally obtained background values and we could not build a calibration curve, indicating that the specificity of the system, at least for the range of proteins that we examined, is quite good. Virtually all the peptides chosen, listed in Table <xref ref-type="table" rid="T1">1</xref>, were active and specific when tested, a fact that clearly indicates the broad applicability of the method. As for the sensitivity of the method we must say that there is room for improvement. The PgP experiments shown in Figure <xref ref-type="fig" rid="F3">3</xref> and <xref ref-type="fig" rid="F4">4</xref> indicate an LOD in the submicromolar range (10<sup>-7</sup>–10<sup>-8 </sup>M). This is a very high value, which may not always be adequate, although in the context of the PgP field it may be. However, a look at the data in Figure <xref ref-type="fig" rid="F7">7B</xref> shows that the LOD for MRP using the MRP I peptide and cognate antibody is in the range of 10<sup>-14 </sup>M, which clearly indicates that there is the potential for an adequately sensitive assay.</p><p>The proteins we studied were chosen merely on the grounds of this laboratory's familiarity with them. We offer the Peptidomatrix procedure, however, as a general solution to the problem of finding a simple and easily set up method for such assays in many biological research and clinical situations. All it requires is selecting suitable peptides in the proteins to be assayed and preparing antibodies against them. Once these are available, reliable estimates of the amounts of the specified protein can be made from digests of a variety of biological material.</p></sec><sec><title>Conclusion</title><p>We have devised a simple, ELISA-based proteomics assay that enables the quantitation of designated proteins in a cell or tissue sample, and that can be used in any laboratory, with minimal specialized equipment.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Peptide selection</title><p>The peptides on which the ELISA is based are chosen using a three-fold screening process: First, a list is made of all the peptides that are likely to be present in a tryptic digest of the protein to be identified. This can be done using a simple word processing program or more specialized software. Next, peptides are selected that have lengths of between 7 and 15 amino acids. Each peptide from this selection is checked for its uniqueness amongst all the polypeptide chains that comprise the human proteome, using the BLAST program (parameters: PAM 30 Gap Costs Existence 5, Extension 2). The main criterion for this stage of selection is the presence of matches to other known proteins that are 5 amino acids or shorter. From this limited list, we chose the more hydrophilic peptides as being those that would be most likely to be good antigens [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>] A number of these selected peptides were ultimately chosen for production of antibodies, and then used in the protocol described below</p><p>The peptides chosen for this investigation are shown in Table <xref ref-type="table" rid="T1">1</xref>.</p></sec><sec><title>Antibodies</title><p>The polyclonal antibodies used throughout this study were generated in rabbits by Affinity Bioreagents, Inc. (ABR, Golden, CO). The antigen for immunization was also prepared by ABR, including the synthesis of the peptides, the conjugation to a carrier and the injection to rabbits. The rabbits were bled once before immunization and 3 or 4 times after immunization. The titers were recorded and the various bleedings kept and used for the development of the immunoassay.</p><p>Monoclonal antibody C494 was purchased from Dako (Glostrup, Denmark)</p></sec><sec><title>Cells</title><p>CEM (WT; Col1000.) These cells are described in Kohler and Stein [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>CEM cell lines were grown in suspension in a 5% CO<sub>2 </sub>atmosphere at 37°C using RPMI-1640 medium (Biological Industries, Kibbutz Beit Ha'emek, Israel) supplemented with 10 % fetal calf serum, 2 mM glutamate, 100 U ml<sup>-1 </sup>penicillin, 100 μg ml<sup>-1 </sup>streptomycin, 250 μg ml<sup>-1 </sup>amphotericin. The human lymphoblastoid cell line CEM Col1000 was derived from the line CEM ADR5000 by growing the original cells over four weeks in medium with 1000 ng ml<sup>-1 </sup>colchicine. They were maintained in this concentration of colchicine.</p><p>MCF7 cells are human breast cancer cells. The two drug-resistant lines, VP-16, and AV, were generated by transforming them with plasmids expressing MRP1 and MXR respectively. The cells were grown in DMEM medium supplemented with fetal calf serum and antibiotics as described above. The growth medium of the VP-16 line was supplemented with 4 μM etoposide while the AV line was grown in the presence of 5 μg/ml verapamil and 3 μg/ml adriamycin.</p></sec><sec><title>Peripheral blood lymphocytes</title><p>Peripheral blood mononuclear cells were separated over Ficoll gradients. Ficoll separation was performed in most cases on the same day, and when this was not possible, whole blood was kept at 4°C and separated the following day. Cells were suspended in RPM1 1640 with 10% fetal calf serum until analysis, generally on the same day or up to 48 h later.</p></sec><sec><title>Sample preparation</title><p>Pellets of 20 million CEM or white blood cells (or 5 million MCF7 cells) were dissolved by re-suspending the sample in 250 μl 2% SDS in PBS and incubated at room temperature for 10 minutes. The protein was precipitated by adding 4 volumes of methanol/acetic acid pH 4.0 (15 ml methanol/40 μl glacial acetic acid) and incubating overnight at -20°C. The protein was then pelleted by centrifugation at 13,000 rpm, in an Eppendorf centrifuge (16,000 g) for 20 min at 4°C. The pellets were washed twice with ethanol and air-dried before re-suspension.</p><sec><title>Protein digestion</title><p>To the protein pellet were added 70 μl of a solution containing 0.03 M Tris HCl, 2 mM CaCl<sub>2</sub>, and 10 μg/ml DNAase I. After incubation for 10 minutes at 37°C, 130 μl of a solution containing 2 M urea, 0.05% SDS and 0.02% NaN<sub>3 </sub>were added. After incubation for 10 additional minutes at 37°C, 20 μl of a stock solution of TPCK-treated Trypsin (SIGMA) (40 mg/ml) was added and the reaction mix was incubated for 20 hr at 37°C. The trypsinolysis was stopped by boiling the digest for 10 minutes. In parallel, a stock of digestion buffer, containing all the components but no protein samples, was prepared, to be used as diluent in the assay.</p></sec></sec><sec><title>ELISA</title><p>Maxisorp ELISA plates (NUNC, Denmark) were coated overnight at 4°C, with 100 μl of a0.1–2 μg/ml solution of the relevant synthetic peptide in 0.1 M carbonate buffer pH 9.6 and blocked 2 hours at room temperature with 200 μl blocking buffer (3% BSA/0.05% Tween 20 in phosphate buffered saline, PBS: 0.01 M phosphate buffer, 150 mM NaCl, pH 7.2). Serial binary dilutions of the synthetic peptide (from 5 to 0.078 μg/ml) and a blank sample were prepared in 50% digestion buffer containing 1 × PBS, 3% BSA, 0.05% Tween 20 and the desired immune serum diluted 1:3000. When monoclonal antibody C494 was used it was diluted to 0.1 μg/ml. These solutions were used for the calibration curve. The experimental samples contained 50 % cell digest, antibody as above and no additional peptide. 100 μl of these solutions were added to the wells and incubated at room temperature 1–3 hours. The wells were washed 4 times with 1 × PBS/0.05% Tween 20 and then 100 μl of horseradish peroxidase(HRP)-conjugated secondary Ab (anti-mouse or anti-rabbit) diluted 1:10,000 or 1:20,000 in blocking buffer were added in each well. After incubation for one hour at room temperature and washing, as above, the bound HRP conjugate was detected by adding 100 μl of tetramethyl benzidine. The peroxidase reaction was stopped after 5 minutes by the addition of 50 μl 0.5 M H<sub>2</sub>SO<sub>4</sub>. Optical densities at 450 nm were measured using an ELISA reader.</p></sec><sec><title>Data determination</title><p>The data from the Elisa plate reader are fed into a data analysis template in the program Sigmaplot (SPSS, Chicago, IL). A plot of the OD450 vs. the concentration of the antibody or the concentration of the free peptide is drawn. The plot is then fitted, using a regression program, to a hyperbola fitting an ascending (Equation. 1) or a descending (Equation. 2) hyperbolic 3-parameter equation, as described in the text:</p><p><inline-graphic xlink:href="1477-5956-4-14-i1.gif"/></p><p><inline-graphic xlink:href="1477-5956-4-14-i2.gif"/></p><p>where Amax is the calculated maximal amplitude of the curve, D the predicted minimum of the ELISA readings, corresponding essentially to the background signal, S the concentration of the antibody or the attached peptide (in equation 1) or the free peptide (in equation 2) and Kd is the concentration that gives one-half of the shift between maximum and minimum readings.</p></sec><sec><title>Rhodamine uptake</title><p>Peripheral blood mononuclear cells, prepared as described above, were stained with rhodamine-123 using the method outlined by Chaudhary and Roninson [<xref ref-type="bibr" rid="B22">22</xref>]. We used 150 ng/ml of rhodamine-123 for staining, for 15 min at 37°C with an efflux time of 2 h; 10 mM verapamil was used as an MDR1 inhibitor. Cells were counterstained with propidium iodide immediately prior to analysis to identify dead cells (if any were present), which were removed from the analysis by gating. We defined a positive result as one in which the Kolmogorov-Smirnov (abbreviated K-S) D value [<xref ref-type="bibr" rid="B23">23</xref>] was equal to or greater than 0.15 [<xref ref-type="bibr" rid="B24">24</xref>-<xref ref-type="bibr" rid="B26">26</xref>].</p></sec><sec><title>Plasma membranes preparation</title><p>The cells were harvested in a table top centrifuge at 1000 rpm × 5 min, and washed twice with 20 mM Hepes pH = 7.4/0.9% NaCl buffer. The cells were then resuspended in lysis buffer (10 mM Hepes-Tris pH = 7.4, 2 mM DTT, 5 mM EDTA, 5 mM EGTA) containing protease inhibitors (cocktail from Sigma) at a concentration of 30 millions/ml. The cells were incubated in ice for 15 minutes and all subsequent procedures were performed at 4°C. The cells were homogenized in a Teflon-glass homogenizer (40 strokes), until 70% disruption. The lysis of the cells was examined under the light microscope. The homogenate is centrifuged for 10 minutes at 3000 rpm (300 g) in a tabletop centrifuge to spin down nuclei and unlysed cells. Subsequently the mitochondria were removed by centrifugation at 5500 rpm (4000 g) in a Sorvall centrifuge (rotor SS34) for an additional 10 minutes. The supernatant was then transferred to an ultracentrifuge tube and centrifuged at 35,000 rpm (114,000 g) in a 70 Ti rotor for 45 min. In this third and final centrifugation the plasma membranes were sedimented. The supernatant was discarded, the membrane pellet was resuspended in 500–1000 μl of lysis buffer and homogenized by aspiration 5 times through a 22 gauge syringe. The isolated membranes were stored at -70°C in aliquots.</p></sec><sec><title>ATPase assay</title><p>The PgP-associated ATPase was determined colorimetrically as the vanadate-sensitive release of inorganic phosphate from ATP hydrolysis (REF). 1 μg membrane protein were incubated in ATPase buffer (50 mM KCl, 2.5 mM MgSO4, 0.5 mM EGTA, 3 mM DTT, 3 mM ATP, 2 mM oubain, 3 mM sodium azide, 25 mM Tris/HCl pH 7.4) and verapamil (10 μM), if desired. The reaction was carried out in 96 well microtiter plates in a final volume of 60 μl in the presence or the absence of 0.25 mM sodium orthovanadate. The microtiter plate was incubated at 37°C for one hour and the reaction was terminated by adding 200 μl of stop solution (0.2% ammonium molybdate, 1.3% sulfuric acid, 0.9% SDS and 1% ascorbic acid). After an additional 30 minutes of incubation the absorbance at 620 nm was read in a microplate reader. ATPase activity was expressed in mmol Pi/hr/mg protein.</p></sec></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>HG and J B-S initially developed the Peptidomatrix procedure. J B-S worked out the method for preparations of tryptic peptides from biological material. J B-S, OB, M K-G and HG performed the experimental work and contributed to the discussions. OB, WDS and UP wrote the manuscript. DGR and UP provided intellectual guidance. DGR provided the patient material. WDS initiated the project, developed the mathematical analyses used, and directed the research.</p></sec>
|
Carney complex (CNC)
|
<p>The Carney complex (CNC) is a dominantly inherited syndrome characterized by spotty skin pigmentation, endocrine overactivity and myxomas. Skin pigmentation anomalies include lentigines and blue naevi. The most common endocrine gland manifestations are acromegaly, thyroid and testicular tumors, and adrenocorticotropic hormone (ACTH)-independent Cushing's syndrome due to primary pigmented nodular adrenocortical disease (PPNAD). PPNAD, a rare cause of Cushing's syndrome, is due to primary bilateral adrenal defect that can be also observed in some patients without other CNC manifestations or familial history of the disease. Myxomas can be observed in the heart, skin and breast. Cardiac myxomas can develop in any cardiac chamber and may be multiple. One of the putative CNC genes located on 17q22-24, (<italic>PRKAR1A</italic>), has been identified to encode the regulatory subunit (R1A) of protein kinase A. Heterozygous inactivating mutations of <italic>PRKAR1A </italic>were reported initially in 45 to 65 % of CNC index cases, and may be present in about 80 % of the CNC families presenting mainly with Cushing's syndrome. PRKAR1A is a key component of the cAMP signaling pathway that has been implicated in endocrine tumorigenesis and could, at least partly, function as a tumor suppressor gene. Genetic analysis should be proposed to all CNC index cases. Patients with CNC or with a genetic predisposition to CNC should have regular screening for manifestations of the disease. Clinical work-up for all the manifestations of CNC should be performed at least once a year in all patients and should start in infancy. Cardiac myxomas require surgical removal. Treatment of the other manifestations of CNC should be discussed and may include follow-up, surgery, or medical treatment depending on the location of the tumor, its size, the existence of clinical signs of tumor mass or hormonal excess, and the suspicion of malignancy. Bilateral adrenalectomy is the most common treatment for Cushing's syndrome due to PPNAD.</p>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Bertherat</surname><given-names>Jérôme</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
Orphanet Journal of Rare Diseases
|
<sec><title>Disease name/synonyms</title><p>Carney complex.</p><p>The complex of cardiac myxomas, endocrine overactivity and spotty pigmentation.</p></sec><sec><title>Definition/diagnostic criteria</title><p>The Carney complex (CNC) was first described in 1985 by J. Aidan Carney, as the combination of myxomas, spotty pigmentation and endocrine overactivity [<xref ref-type="bibr" rid="B1">1</xref>]. It is defined by the association of multiple endocrine neoplasia and cardiocutaneous manifestations. Patients previously characterized as LAMB (lentigineses, atrial myxoma, mucocutaneous myxoma, blue nevi) or NAME (nevi, atrial myxoma, myxoid neurofibroma, ephelide) could be considered as having Carney complex. Numerous organs may be involved in CNC and the manifestations vary greatly among patients. Some of them are quite specific, such as primary pigmented nodular disease (PPNAD), while others show little specificity, such as thyroid nodes or blue nevi. It is generally assumed that a patient presenting with two or more of the manifestations listed in Table <xref ref-type="table" rid="T1">1</xref> would be diagnosed as having Carney complex. This table lists the most frequent features of CNC and their estimated frequency [according to references [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B8">8</xref>] and our personal observations]. The incidence of each manifestation depends on its presentation and might not reflect true prevalence. For instance, according to autopsy studies PPNAD is a constant feature in CNC patients [<xref ref-type="bibr" rid="B8">8</xref>], however, reports of Cushing syndrome in the literature indicate that only 25 to 45 % of CNC patients have PPNAD. It has been established that at least two of these manifestations need to be present to confirm the diagnosis of CNC. If the patient has a germline <italic>PRKAR1A </italic>mutation and/or a first-degree relative affected by CNC, a single manifestation is sufficient for the diagnosis.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Main features and diagnostic criteria of Carney complex. This table lists the most frequent features of CNC and their estimated frequency [according to references 1, 3–5, 8 and our personal observations]. The incidence of each manifestation depends on its presentation and might not reflect true prevalence. For instance, according to autopsy studies PPNAD is a constant feature in CNC patients [8], however, reports of Cushing syndrome in the literature indicate that only 25 to 45 % of CNC patients have PPNAD and our personnal experience suggests that about 60 % of CNC patients systematically investigated have PPNAD. It has been established that at least two of these manifestations (except ovrian cyst) need to be present to confirm the diagnosis of CNC. If the patient has a germline <italic>PRKAR1A </italic>mutation and/or a first-degree relative affected by CNC, a single manifestation is sufficient for the diagnosis.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Main features of Carney complex</bold></td><td align="left"><bold>(%)</bold></td></tr></thead><tbody><tr><td align="left">Primary Pigmented Nodular Adrenocortical Disease (PPNAD)</td><td align="left">25–60</td></tr><tr><td align="left">Cardiac myxoma</td><td align="left">30–60</td></tr><tr><td align="left">Skin myxoma</td><td align="left">20–63</td></tr><tr><td align="left">Lentiginosis</td><td align="left">60–70</td></tr><tr><td align="left">Multiple blue nevus</td><td></td></tr><tr><td align="left">Breast ductal adenoma</td><td align="left">25</td></tr><tr><td align="left">Testicular tumors (LCCSCT: Large-Cell Calcifying Sertoli Cell Tumor) (in male)</td><td align="left">33–56</td></tr><tr><td align="left">Ovarian cyst (in female)</td><td align="left">20–67</td></tr><tr><td align="left">Acromegaly</td><td align="left">10</td></tr><tr><td align="left">Thyroid tumor</td><td align="left">10–25</td></tr><tr><td align="left">Melanotic schwannoma</td><td align="left">8–18</td></tr><tr><td align="left">Osteochondromyxoma</td><td align="left"><10</td></tr></tbody></table></table-wrap></sec><sec><title>Epidemiology</title><p>CNC is a rare disease. About 500 patients have been registered by the NIH-Mayo Clinic (USA) and the Cochin center (France) [<xref ref-type="bibr" rid="B2">2</xref>]. Cumulative reports from these centers, plus information from the Cornell center in New York, indicate that there are about 160 index cases of CNC presently known [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec><title>Clinical description</title><p>The manifestations of CNC can be numerous and vary between patients. Even in the same kindred, phenotypic variability can be observed. The estimated frequencies of these manifestations are listed in Table <xref ref-type="table" rid="T1">1</xref>. Endocrine, dermatologic and cardiac anomalies are the main manifestations of the disease.</p><sec><title>Skin lesions</title><p>The lentiginosis is observed in most patients and is so characteristic that can makes the diagnosis. It appears as small (2 to 10 mm) brown to black macules typically located around the upper and lower lips, on the eyelids, ears and the genital area. Multiple blue nevi and junctional or compound nevi may also be observed in CNC, as well as cutaneous myxomas. The skin myxomas present as non-pigmented subcutaneous nodules. Myxomas can also be located in the ear canal.</p></sec><sec><title>Endocrine tumors</title><p>The following types of endocrine gland tumors are observed in CNC patients: growth hormone (GH)-secreting pituitary adenomas (acromegaly), thyroid adenomas or carcinomas, testicular tumors (large-cell calcifying Sertoli cell tumors), and ovarian cysts. Adrenocorticotropic hormone (ACTH)-independent Cushing's syndrome due to PPNAD is observed in 25 to 30 % of patients with CNC. PPNAD is a rare disease observed mostly in patients with CNC. The disease was named after the macroscopic appearance of the adrenals that is characterized by the small pigmented micronodules observed in the cortex (Figure <xref ref-type="fig" rid="F1">1</xref>) [<xref ref-type="bibr" rid="B7">7</xref>]. The disease is usually bilateral with primary involvement of both adrenals. Cushing's syndrome due to PPNAD is observed in children and young adults, with a peak during the second decade of life [<xref ref-type="bibr" rid="B8">8</xref>]. It is rare, but can occur before the age of 4-yr and it is rarely diagnosed after the age of 40-yr. Diagnosis of Cushing's syndrome due to PPNAD is often difficult because hypercortisolism can develop progressively over years. In contrast, a large and rapid burst of cortisol excess can be observed in some patients, which might spontaneously regress. In some cases of PPNAD, clearly cyclic forms of hypercortisolism have been documented [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. PPNAD can also be diagnosed by systematic screening in patients with CNC investigated for other clinical manifestations of the complex or after familial screening. Despite the unusual time course of Cushing's syndrome observed in some patients with PPNAD, clinical signs are quite similar to those observed in patients presenting with other causes of hypercortisolism. Urinary cortisol is increased in most patients at the time of diagnosis of PPNAD, but its level can be highly variable. The circadian rhythm of cortisol secretion is usually completely abolished. As with ACTH-independent Cushing's syndrome due to other causes, patients with PPNAD have low plasma levels of ACTH and show no stimulation of cortisol or ACTH secretion after corticotropin-releasing hormone (CRH) injection. In addition, dexamethasone fails to suppress cortisol secretion, even after high dose administration. Pathological investigation reveals that adrenal glands from patients with PPNAD are usually normal in size and weight (between 4 to 17 g) [<xref ref-type="bibr" rid="B8">8</xref>]. In keeping with this finding, adrenals appear normal on computed tomography (CT)-scan in one out of three patients (Figure <xref ref-type="fig" rid="F1">1</xref>). In the other patients, micronodules can be visible and, more rarely, macronodules (>1 cm diameter) in one or both glands. Iodocholestrol scintigraphy, when performed, usually shows a bilateral uptake despite ACTH suppression by endogenous hypercortisolism.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Macroscopic and CT-scan findings in primary pigmented nodular adrenocortical disease (PPNAD) A: Macroscopic appearance of the adrenal gland in PPNAD. The cut surfaces show multiple pigmented micronodules. The periadrenal fat is also visible around the adrenal capsule. B: Adrenal CT-scan in PPNAD. A micronodule is visible on the external part of the left adrenal on the CT-scan shown (see red arrow).</p></caption><graphic xlink:href="1750-1172-1-21-1"/></fig><p>Acromegaly due to a pituitary GH-secreting tumor is not very frequent, but most patients with CNC present with a mild increase in GH, and sometimes in prolactin (PRL) secretion. Alterations in the rhythm of GH secretion are frequently observed.</p><p>Thyroid tumors are most often benign, non-toxic adenomas, mostly of follicular type. Some patients present with papillary carcinoma that can be multiple and sometimes quite aggressive.</p><p>Testicular tumors are easily detected by ultrasound investigation as bilateral microcalcifications. Ovarian cysts and cystadenoma have been observed in CNC patients.</p></sec><sec><title>Cardiac myxomas</title><p>Cardiac myxoma is an important manifestation of CNC. It may be the cause of the high rate (16 %) of sudden death historically reported in CNC families [<xref ref-type="bibr" rid="B11">11</xref>], thus underlying the importance of its early diagnosis. In the past, underdiagnosis of cardiac myxomas may have accounted for the majority of deaths due to CNC. In contrast with sporadic myxoma, they can develop in any cardiac chamber and may be multiple. Cardiac myxoma can be the cause of stroke due to embolism and cardiac deficiency. It is therefore important to screen regularly (by ultrasound) patients with CNC for the presence of cardiac myxoma. In difficult cases, trans-esophageal ultrasound and cardiac magnetic resonance imaging (MRI) can be very helpful.</p></sec><sec><title>Other tumors</title><p>Various other tumors, some of them quite specific for CNC, can be observed. Melanotic schwannoma is a rare tumor and occurs mainly in CNC. It is a pigmented tumor that can be misdiagnosed as a melanoma. This tumor can be observed in any periphal nerve and can be, in rare cases, malignant. Breast ductal adenomas, breast myxomas, and osteochondromyxoma are among the tumors also observed in CNC.</p></sec></sec><sec><title>Etiology/genetics of CNC</title><p>The first description of CNC included 40 patients [<xref ref-type="bibr" rid="B1">1</xref>], among them 10 were familial cases, leading to the hypothesis of a genetic origin, at least in a subset of patients. CNC seems to be a genetically heterogeneous disease and linkage analysis has shown that at least two loci are involved: 2p16 and 17q22-24 [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. The <italic>CNC1 </italic>gene, located on 17q22-24, has been identified as the regulatory subunit (R1A) of the protein kinase A (<italic>PRKAR1A</italic>) [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. PRKAR1A is a key component of the cAMP signaling pathway that has been implicated in endocrine tumorigenesis. Heterozygous inactivating mutations of <italic>PRKAR1A </italic>have been detected in about 45 to 65 % of CNC families [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. In CNC patients with Cushing's syndrome the frequency of <italic>PRKAR1A </italic>mutations is about 80 %, suggesting that families with PPNAD are more likely to carry a 17q22-24 defect [<xref ref-type="bibr" rid="B5">5</xref>]. Interestingly, patients with isolated PPNAD and no familial history of CNC may also carry a germline <italic>de novo </italic>mutation in <italic>PRKAR1A </italic>[<xref ref-type="bibr" rid="B6">6</xref>]. In the tumors of CNC patients loss of heterozygosity (LOH) at 17q22-24 may be observed, suggesting that <italic>PRKAR1A </italic>is a tumor suppressor gene. Somatic mutation of <italic>PRKAR1A </italic>in a patient with PPNAD already carrying a germline mutation may lead to inactivation of the wild type allele [<xref ref-type="bibr" rid="B6">6</xref>]. However, inactivation of the remaining wild type allele by genetic alteration does not appear to be a constant step in PPNAD and CNC tumor development [<xref ref-type="bibr" rid="B5">5</xref>]. In a mice transgenic model with heterozygous inactivation of <italic>PRKAR1A</italic>, tumors may develop without allelic loss [<xref ref-type="bibr" rid="B4">4</xref>]. This suggests that the classic model of tumor suppressor gene with a germline inactivating first allelic alteration, followed by a second genetic hit leading to inactivation of the remaining wild type allele, might to some extent be applicable to PRKAR1A. It is also possible that in PPNAD, a general polyclonal expansion might be stimulated by haploinsufficiency due to the first germline defect; a second genetic hit would then lead to inactivation of the wild type allele and further stimulate tumorigenesis and the development of adrenocortical nodules.</p><p>A putative <italic>CNC2 </italic>gene located at the 2p16 locus remains to be determined [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. Somatic alterations of the 2p16 region have been reported in CNC tumors, even in patients with mutation of the <italic>CNC1 </italic>gene (<italic>i.e. PRKAR1A </italic>located on 17q22-24) [<xref ref-type="bibr" rid="B15">15</xref>]. These alterations are usually gene amplifications, suggesting that the gene located at 2p16 is a potential oncogene. Considering the genetics in isolated PPNAD, it should be mentioned that the clinical manifestations in a subgroup of very young PPNAD patients may differ from those in older patients with CNC. In these patients the classical pathological finding of pigmented nodules may be absent although micronodules are present [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B16">16</xref>]. In this subgroup of very young PPNAD patients, Cushing's syndrome may occur between birth and the age of 5-yrs. The main reason for differentiating this group of PPNAD or PPNAD-like patients is the lower rate of germline inactivating mutation.</p></sec><sec><title>Management including treatment</title><p>Patients with CNC or with a genetic predisposition to CNC (<italic>i.e. PRKAR1A </italic>germline mutation carriers) should have regular screening for manifestations of the disease. At present, it is not possible to produce evidenced-based recommendations for the screening schedule. However, there is general agreement on the following recommendations. Clinical work-up for all the manifestations of CNC should be performed at least once a year in all patients and should start in infancy. Cardiac myxomas are one of the main features of CNC and should be diagnosed early through screening performed at least once a year by cardiac ultrasound. In patients with a history of cardiac myxoma, screening should be performed every 6 months. Screening for cardiac myxoma by ultrasound should start during the first 6 months. In contrast, screening for the other manifestations (only by clinical examination) should be performed in patients under 5 years-old. For children, pubertal staging and growth rate should be monitored. Biological and hormonal work-up should include: measurement of the levels of blood glucose, urinary cortisol, plasma and/or salivary circadian variations of cortisol, plasma ACTH, GH, PRL, Insulin-Like-Growth-Factor I, and ionograms. Imaging investigations should include: adrenal CT-scan (in cases with biological evidence of Cushing's syndrome), thyroid ultrasound (in cases with abnormal palpation), testicular or ovarian ultrasound, pituitary MRI, and spine MRI when the clinical signs suggest schwannoma. These imaging investigations do not always need to be performed each year; the necessity will be determined according to the history and previous imaging results, as well as the present clinical data and results of biological investigations.</p><p>A cardiac myxoma requires surgical removal. Treatment of the other manifestations should be discussed and may include follow-up, surgery, or medical treatment depending on the location of the tumor, its size, the existence of clinical signs of tumor mass or hormonal excess, and the suspicion of malignancy. Malignancy in CNC is mostly observed in thyroid nodes (fine-needle investigation is helpful for diagnosis) and melanotic schwannoma (estimated rate of malignancy: 10 %). Bilateral adrenalectomy is the most common treatment for Cushing's syndrome due to PPNAD. Some rare cases have been treated by O, p'-dichlorodiphenyldichloroethane (DDD), ketoconazol, or unilateral adrenalectomy. In the few patients in whom overt Cushing's syndrome did not recur after unilateral adrenalectomy, alterations in the rhythm of cortisol secretion can be observed on long term follow-up, demonstrating that despite apparent cure the disease is indeed bilateral [<xref ref-type="bibr" rid="B17">17</xref>].</p></sec><sec><title>Genetic counseling</title><p>Genetic analysis should be proposed to all CNC index cases. When a <italic>PRKAR1A </italic>mutation is identified, a genetic analysis should be proposed to first degree relatives. This is best performed in a multidisciplinary genetic center, in keeping with the regulations specific to each country. Psychological assessment and management of the patients might be important when performing genetic screening in asymptomatic patients. The identification of mutation carriers should lead to the same follow-up and management as that described for CNC patients. Therefore, it is suggested that genetic screening of first degree relatives is performed at the same time as the first cardiac ultrasound. When performing genetic screening in an asymptomatic child, the parent should be given information about the fact that a positive screening with identification of a <italic>PRKAR1A </italic>mutation should lead to regular screening and follow-up for CNC manifestations. At present, no recommendation has been made for prenatal diagnosis and this should be discussed by a multidisciplinary team in keeping with the regulation specific to each country. There are no reports of prenatal diagnosis of CNC in Europe.</p></sec><sec><title>Unresolved questions</title><p>No <italic>PRKAR1A </italic>gene mutations were found in some families and the 17q22-24 locus was excluded. The other gene(s) responsible for CNC remain to be identified. In CNC kindred without the <italic>PRKAR1A </italic>mutation, a large gene deletion or alteration (not detected by the commonly used direct sequencing or DHPLC methods) may be responsible for the condition.</p><p>A mutation of the <italic>perinatal myosin heavy chain </italic>gene has been reported in patients with trismus-pseudocamptodactyly syndrome. In some of the cases, manifestations of this syndrome overlap with those observed in CNC [<xref ref-type="bibr" rid="B18">18</xref>]. However, the phenotype differs from CNC and at present it does not seem that this gene is involved in kindreds with the classic diagnostic criteria of CNC [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>The subgroup of very young infants with isolated PPNAD, no other personal or familial history of CNC and no <italic>PRKAR1A </italic>mutation may have a separate disease.</p><p>Some rare manifestations of the disease (such as osteochondromyxoma) have not yet been extensively described.</p><p>The mechanism of <italic>PRKAR1A </italic>mutation-induced tumorigenesis is currently under extensive investigation and is an important field of research.</p></sec>
|
MATER protein expression and intracellular localization throughout folliculogenesis and preimplantation embryo development in the bovine
|
<sec><title>Background</title><p><italic>Mater </italic>(Maternal Antigen that Embryos Require), also known as <italic>Nalp5</italic> (NACHT, leucine rich repeat and PYD containing 5), is an oocyte-specific maternal effect gene required for early embryonic development beyond the two-cell stage in mouse. We previously characterized the bovine orthologue <italic>MATER </italic>as an oocyte marker gene in cattle, and this gene was recently assigned to a QTL region for reproductive traits.</p></sec><sec><title>Results</title><p>Here we have analyzed gene expression during folliculogenesis and preimplantation embryo development. <italic>In situ </italic>hybridization and immunohistochemistry on bovine ovarian section revealed that both the transcript and protein are restricted to the oocyte from primary follicles onwards, and accumulate in the oocyte cytoplasm during follicle growth. In immature oocytes, cytoplasmic, and more precisely cytosolic localization of MATER was confirmed by immunohistochemistry coupled with confocal microscopy and immunogold electron microscopy. By real-time PCR, MATER messenger RNA was observed to decrease strongly during maturation, and progressively during the embryo cleavage stages; it was hardly detected in morulae and blastocysts. The protein persisted after fertilization up until the blastocyst stage, and was mostly degraded after hatching. A similar predominantly cytoplasmic localization was observed in blastomeres from embryos up to 8-cells, with an apparent concentration near the nuclear membrane.</p></sec><sec><title>Conclusion</title><p>Altogether, these expression patterns are consistent with bovine MATER protein being an oocyte specific maternal effect factor as in mouse.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Pennetier</surname><given-names>Sophie</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Perreau</surname><given-names>Christine</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Uzbekova</surname><given-names>Svetlana</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Thélie</surname><given-names>Aurore</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Delaleu</surname><given-names>Bernadette</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Mermillod</surname><given-names>Pascal</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" corresp="yes" contrib-type="author"><name><surname>Dalbiès-Tran</surname><given-names>Rozenn</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Developmental Biology
|
<sec><title>Background</title><p>Preimplantation embryo development is largely dependent on maternal transcripts and proteins synthesized during oogenesis. Maternal factors are able to support the first cleavages, while blastocyst formation involves both maternal and embryonic factors. Over the last years, oocyte-restricted maternal effect genes have been the focus of much attention due to their specific expression profile and crucial function in early embryo development. They are predominantly expressed in oocyte, remain present in early embryos and then are degraded at the time of maternal-to-embryo transition (MET), without compensation by embryonic transcription. Functional studies based on knock-out mouse models have demonstrated their essential role in preimplantation embryo development, whereas functions in the oocyte itself have not been elucidated until this day.</p><p><italic>Mater </italic>(Maternal Antigen that Embryos Require) is one such oocyte-specific maternal effect genes and was first identified in mouse [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The transcript and protein expression profiles have been investigated during oogenesis and preimplantation embryo development. <italic>Mater </italic>transcript and protein are first detected in oocyte from primary follicles and accumulate during oocyte growth. The transcript level decreases after fertilization as shown by ribonuclease protection assay or DNA array [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]). The protein remains abundant until the morula stage and is still present in blastocysts [<xref ref-type="bibr" rid="B3">3</xref>]. <italic>Mater</italic>-null females present a normal phenotype regarding folliculogenesis, ovulation and fertilization, but their embryos do not develop beyond the 2-cell stage coincident with the maternal-to-embryo transition. <italic>Mater </italic>precise function remains to be elucidated, although the global transcription decrease described in two-cell embryos lacking MATER may suggest a role in embryonic genome activation [<xref ref-type="bibr" rid="B2">2</xref>]. In our previous work, we identified the bovine orthologue of <italic>Mater</italic>. The longest open reading frame encodes a putative protein of 1098 amino acids (121 kDa). As its human counterpart, bovine MATER includes the 3 domains characteristics of the Nacht, Leucine rich repeat and Pyrin domain containing (NALP) family: a N-terminal Pyrin domain, followed by a NACHT domain and twelve C-terminal leucine rich repeats of the ribonuclease inhibitor subtype (LRR-RI). It is localized within a QTL region for reproductive traits [<xref ref-type="bibr" rid="B5">5</xref>]. <italic>MATER </italic>transcript pattern, i.e. its tissue distribution and its disappearance in blastocyst, appeared in agreement with bovine <italic>MATER </italic>also being an oocyte-specific maternal effect gene and suggested a conserved function as in mouse.</p><p>To reinforce this hypothesis, we needed to refine transcript expression during folliculogenesis, and mostly to characterize the expression and localization of the protein. In this study, we show that bovine MATER transcript and protein are expressed in the oocyte as early as the primary follicle stage and accumulate during folliculogenesis. The protein localizes in the cytosol of immature oocytes. It remains abundant in the cytoplasm of preimplantation embryos until the blastocyst stage and is mostly degraded after hatching.</p></sec><sec><title>Results</title><sec><title>Antibody characterization</title><p>First, we checked our antipeptide serum and purified antibody for sensitivity and specificity by western blotting (Fig. <xref ref-type="fig" rid="F1">1</xref>). Under normal exposure, a single intense band was detected at the MATER protein predicted molecular weight (121 kDa) with as few as 10 oocytes. No such band was detected in a protein extract from cumulus cells in huge excess (see the amount of TUBULIN), although faint bands were observed at various molecular weights. Specificity was further confirmed by the absence of MATER signal in oocytes using preimmune serum.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Characterization of anti-MATER serum and purified antibody by Western blot</bold>. Detection of MATER in immature oocytes (IO) but not in cumulus cells (Cc) with anti-peptide serum or purified antibody, nor with the preimmune serum. Molecular weight is indicated on the left.</p></caption><graphic xlink:href="1471-213X-6-26-1"/></fig></sec><sec><title>MATER expression throughout folliculogenesis</title><p>The expression of <italic>MATER </italic>transcript during folliculogenesis was analysed using <italic>in situ </italic>hybridization onto ovarian sections. As previously described [<xref ref-type="bibr" rid="B6">6</xref>], <italic>MATER </italic>expression was restricted to the oocyte within bovine ovary. In this experiment, a signal could be detected in oocyte as early as the primary follicle stage and at a higher level in oocyte from antral follicle. Hybridization with the corresponding sense probe was used as negative control (Fig. <xref ref-type="fig" rid="F2">2</xref>).</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Ovarian localization of bovine <italic>MATER </italic>transcript by <italic>in situ </italic>hybridization</bold>. Bright field (A, D) and dark field (B, C, E, F) photomicrographs of adjacent ovarian sections showing primary follicles (A-C, arrows), and an antral follicle (D-F) hybridized with either antisense (A, B, D, E) or sense (C, F) <italic>MATER</italic>-[<sup>35</sup>S] riboprobes. Scale bar: 100 μm.</p></caption><graphic xlink:href="1471-213X-6-26-2"/></fig><p>To further characterize <italic>MATER </italic>gene expression during folliculogenesis, we studied its expression pattern at the protein level. Using anti-MATER serum, the protein was exclusively detected in oocyte and no staining was observed in granulosa or theca cells. MATER was detected in oocyte of primary follicles at a low level, with a very distinctive pattern: the staining localized to a single dot (Fig. <xref ref-type="fig" rid="F3">3A</xref>) which was specifically and reproductively observed. As folliculogenesis proceeds, staining intensity increases in oocyte from preantral to small antral and antral follicles (Fig. <xref ref-type="fig" rid="F3">3B–D</xref>), suggesting an accumulation of the protein. Interestingly, on sections that went through the nucleus in pre-antral and early antral follicles, the protein appeared excluded from the germinal vesicle (Fig. <xref ref-type="fig" rid="F3">3B, C</xref>).</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Ovarian localization of bovine MATER protein by immunohistochemistry</bold>. Adjacent ovarian sections showing primary follicle (A, E), pre-antral follicle (B, F), early antral follicle (C, G) and over 1 mm antral follicle (D, H) incubated with either anti-MATER serum (A-D) or preimmune serum (E-H) followed by biotinylated horse anti-rabbit IgG antibody. Scale bar: 100 μm. Inset in A is the same oocyte shown at higher magnification.</p></caption><graphic xlink:href="1471-213X-6-26-3"/></fig></sec><sec><title>MATER expression during preimplantation embryo development</title><p>Evolution of MATER messenger RNAs during during oocyte maturation, fertilization and preimplantation embryo development was followed by real time PCR. Triplicate reactions were run onto four independent samples at each stage. The level relative to immature oocytes is shown (Fig. <xref ref-type="fig" rid="F4">4A</xref>). Detection level decreased by 72% during maturation, then remained stable during fertilization. It decreased progressively during the cleavage stages, up to 10% of its initial level in 5 to 8 cell-embryos. In morulae and blastocysts, amplicons were produced only in one or two samples; the level of polyadenylated transcript was less than 0.5% of its initial value. Western blotting was then performed to characterize MATER protein expression (Fig. <xref ref-type="fig" rid="F4">4B</xref>). MATER was abundant during oocyte maturation, and after fertilization until the expanded blastocyst stage. Most protein was then degraded after hatching. This pattern was reproducibly observed with two independent oocyte and embryo collections. By contrast, our positive control β-TUBULIN increased in hatched blastocysts (Fig. <xref ref-type="fig" rid="F4">4B</xref>).</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Expression of bovine <italic>MATER </italic>in oocytes and preimplantation embryos</bold>. Developmental stages are immature oocytes (IO), <italic>in vitro </italic>matured oocytes (MO), <italic>in vitro </italic>cultured zygote (1C), 2-, 4-, 5- to 8- cell embryos, morula (mo), expanded blastocysts (EB) and hatched blastocysts (HB). (A) Quantification of the messenger RNA using real-time PCR. Relative level to immature oocytes is shown (mean ± SEM). In late stage embryos, transcript level is indicated but too low to be seen on the histogram. Different superscripts indicate a significant difference (P ≤ 0.05). (B) Detection of the protein by Western blot; α-TUBULIN is shown as a positive control.</p></caption><graphic xlink:href="1471-213X-6-26-4"/></fig></sec><sec><title>MATER protein localization in oocyte and preimplantation embryos</title><p>Immunohistochemistry onto ovary sections suggested that MATER could be predominantly cytoplasmic. To refine protein localization, we performed confocal microscopy analysis on oocytes from 3 to 8 mm follicles and preimplantation embryos, after immunofluorescent staining using purified antipeptide (Fig. <xref ref-type="fig" rid="F5">5</xref>). Firstly, we observed oocytes at germinal vesicle (GV) and nuclear envelope break down (NEBD) stages, as discriminated by LAMIN A/C staining (Fig. <xref ref-type="fig" rid="F5">5A</xref>). In most GV oocytes, MATER protein was predominantly localized in the cytoplasm (Fig. <xref ref-type="fig" rid="F5">5A</xref>, left panel); however, a few oocytes (less than 10%) presented a distinct pattern, with staining within both the cytoplasm and the nucleus (Fig. <xref ref-type="fig" rid="F5">5A</xref>, middle panel). In oocytes characterized by degradation of LAMIN A/C (Fig. <xref ref-type="fig" rid="F5">5A</xref>, right panel), MATER was distributed throughout the oocyte, only excluded from the chromatin area as revealed by Hoechst staining. Then we analysed mature oocytes, zygotes and preimplantation embryos (Fig. <xref ref-type="fig" rid="F5">5B</xref>). MATER was predominantly detected in the cytoplasm, with the protein apparently concentrating in the cortical region of mature oocytes, zygotes and embryos at least until the 8-cell stage. In zygotes to 8-cell embryos, increased staining around the chromatin area (as revealed by Hoechst staining, not shown) suggested that MATER also accumulated around the nuclear membrane. In expanded blastocysts, MATER was distributed evenly between inner cell mass and trophectoderm. A dramatic decrease was observed in hatched blastocysts, as previously observed by Western blotting. As negative control, no staining was revealed using rabbit IgG as primary antibody (not shown).</p><fig position="float" id="F5"><label>Figure 5</label><caption><p><bold>Localization of MATER protein in bovine oocytes and preimplantation embryos by confocal microscopy</bold>. Oocytes and embryos were incubated with anti-MATER peptide purified antibody (1:250) and revealed by green staining. Scale bar: 50 μm. (A) Oocytes at GV or NEBD stage. Lamins A/C are stained in red (middle row). Chromatin is stained in blue with Hoechst 33258 (lower row). (B) immature oocytes (IO), <italic>in vitro </italic>matured oocytes (MO), <italic>in vitro </italic>cultured zygote (1C), 2-, 4-, 8- cell embryos, morula (mo), expanded blastocysts (EB) and hatched blastocysts (HB).</p></caption><graphic xlink:href="1471-213X-6-26-5"/></fig><p>To refine MATER protein subcellular localization, transmission electron microscopy analysis was performed on ultrathin sections of immature oocytes using purified antipeptide. Immunogold particules were diffusely present in the cytosol at a low density. MATER protein was not detected in the nucleus, nor in organelles including mitochondria (Fig. <xref ref-type="fig" rid="F6">6</xref>). As a negative control, no particles were observed when omitting the primary antibody (not shown).</p><fig position="float" id="F6"><label>Figure 6</label><caption><p><bold>Subcellular localization of MATER protein in bovine immature oocytes by transmission electron microscopy</bold>. Immunogold particles are pointed at by arrows. (A) region including a mitochondria (B) a detail of the nuclear membrane, separating the nucleus (below) and the cytoplasm (above).</p></caption><graphic xlink:href="1471-213X-6-26-6"/></fig></sec></sec><sec><title>Discussion</title><p>In this study, we characterized the bovine <italic>MATER </italic>gene and its expression throughout folliculogenesis and in vitro preimplantation embryo development. We demonstrated that both the transcript and protein were present in oocyte from primary follicle and accumulated thereafter. During preimplantation embryo development, MATER messengers were hardly detected after the morula stage, while the protein remained present until the blastocyst stage before rapid degradation after hatching. In immature oocytes and in embryos, the protein was predominantly cytoplasmic.</p><sec><title>MATER gene expression throughout bovine folliculogenesis and preimplantation embryo development</title><p>In a previous study, we demonstrated that <italic>MATER </italic>transcript was restricted to the oocyte within bovine ovary [<xref ref-type="bibr" rid="B6">6</xref>]. Here, we have refined our expression analysis. By <italic>in situ </italic>hybridization and immunohistochemistry experiments, we detected <italic>MATER </italic>transcript and protein in oocytes as early as the primary follicle stage (Fig. <xref ref-type="fig" rid="F2">2</xref> and <xref ref-type="fig" rid="F3">3</xref>), as previously described in mouse [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. This shows that at least some transcripts support immediate translation, but it does not exclude that others are stored within the oocyte cytoplasm. Such an early protein synthesis may appear premature since functional studies did not reveal a deleterious effect of <italic>Mater </italic>absence at this early stage, but rather in the 2-cell embryo [<xref ref-type="bibr" rid="B2">2</xref>]. Indeed, ZAR1, another mouse oocyte-specific maternal effect protein, was also detected in mouse primary follicles [<xref ref-type="bibr" rid="B7">7</xref>]. Several scenarios may reconcile these apparent discrepancies. The first hypothesis is that MATER indeed plays a role in the oocyte, but that another gene may be redundant at this stage, at least in <italic>Mater</italic>-null females. Obvious candidates are genes of the NALP family, and especially oocyte restricted members such as <italic>NALP9 </italic>in bovine [<xref ref-type="bibr" rid="B8">8</xref>] or the <italic>Nalp9 </italic>or <italic>Nalp4 </italic>genes in mouse [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. To our knowledge, there are no functional data on these genes from knock-out models or other targeted inhibition experiments. This early expression may also be indicative of an upstream role of MATER in a cascade of events, as demonstrated for the mouse oocyte-specific homeobox gene <italic>Nobox </italic>[<xref ref-type="bibr" rid="B11">11</xref>]. Alternatively, the protein may be synthesized despite not being required at this stage, simply because specifically repressing the gene would be more complex for the cell than support minimal expression. Overall, we observed that immunostaining of MATER protein increased in parallel with follicle development, from primary to large antral. This must involve a growing level of translation; in addition, it is possible that once synthesized, the protein remains stable for a long period of time and persists in the oocyte during folliculogenesis. Neo-synthesized proteins would then add to those already present, and early synthesis would allow for the accumulation of a largest stockpile of MATER.</p><p>By real-time PCR, bovine MATER transcripts showed a massive deadenylation and/or degradation during maturation (Fig <xref ref-type="fig" rid="F4">4A</xref>). Following fertilization, the messengers slowly decreased in embryos until the 5- to 8-cell stage; it was hardly detected in morulae and blastocysts, confirming our previous report [<xref ref-type="bibr" rid="B6">6</xref>]. We did not evidence neo-synthesis or polyadenylation of MATER transcripts neither transiently nor after the MET. By contrast, we have demonstrated by Western-blot and immunofluorescence coupled to confocal microscopy that the protein is present at all stages, and displays a sharp decrease only after hatching (Fig. <xref ref-type="fig" rid="F4">4</xref> and <xref ref-type="fig" rid="F5">5</xref>). These distinct patterns for the transcript and the protein provide valuable information. First, later persistence of the protein as compared to the messenger is consistent with MATER being a fairly stable protein. This, together with the accumulation of MATER during folliculogenesis, suggests that at least some protein detected in early embryos is of maternal origin. Second, degradation of MATER protein after hatching appears temporally controlled and substrate specific, since most proteins are rather synthesized at this stage following the MET. The bovine protein in fact persists longer than its mouse counterpart, which was mostly degraded by the expanded blastocyst stage. This is consistent with active degradation of MATER involving factors synthesized from embryonic transcripts in both species. The ubiquitin-proteasome pathway is one possible mechanism, and ubiquitin transcripts were actually shown to increase in bovine blastocysts [<xref ref-type="bibr" rid="B12">12</xref>]. On the other hand, ubiquitin cross-reactive structures were detected preferentially in the trophoblast over the the inner cell mass [<xref ref-type="bibr" rid="B13">13</xref>] whereas here we observe a simultaneous degradation of MATER in both structures at hatching (Fig. <xref ref-type="fig" rid="F5">5</xref>). Overall, the pattern of expression of bovine MATER, i.e. non reactivation at MET and degradation of the protein before implantation, supports the hypothesis that it is a maternal effect gene as its mouse orthologue.</p></sec><sec><title>MATER protein intracellular localization in bovine oocytes and preimplantation embryos</title><p>Data from immunohistochemistry on ovary sections suggested that MATER protein was stored in the cytoplasm of oocyte during folliculogenesis (Fig. <xref ref-type="fig" rid="F3">3B, C</xref>). Intracellular localization of MATER was further investigated in oocytes and early embryos by immunofluorescence coupled to confocal microscopy analysis (Fig. <xref ref-type="fig" rid="F5">5</xref>). In most immature oocytes, MATER was predominantly localized in the cytoplasm with an accumulation around the nuclear membrane and was hardly detected in the germinal vesicle. A few oocytes displayed a different pattern of protein distribution, where MATER was detected both in the cytoplasm and the germinal vesicle (Fig. <xref ref-type="fig" rid="F5">5A</xref>). After NEBD, the protein was uniformly distributed throughout the cytoplasm and was only excluded from the chromatin area (Fig. <xref ref-type="fig" rid="F5">5A</xref>). Such transient nuclear localization of MATER was not described in mouse oocytes. We propose two hypotheses consistent with our observations. It may be that MATER passively diffuses to the germinal vesicle as the nuclear membrane starts to disaggregate, even though lamins are still detected. Alternatively, we can speculate that MATER actively translocates to the nucleus just before NEBD. MATER sequence does not display a nuclear localization signal as analyzed by the PredictNLS server [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. But neither was such signal identified in the mouse oocyte specific variant of DNA (cytosine-5)-methyl-transferase 1 (Dnmt1o), whose temporarily controlled nucleocytoplasmic trafficking was nevertheless demonstrated: this protein localizes to the cytoplasm of oocytes and preimplantation embryos with exclusion from all nuclei except those of 8-cell embryos [<xref ref-type="bibr" rid="B16">16</xref>]. MATER might be involved in the short burst of transcription that has been suggested to occur in bovine oocyte-cumulus complexes just before chromatin condenses [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B19">19</xref>]</p><p>Then, we followed MATER localization after oocyte maturation, fertilization and during preimplantation embryo development (Fig. <xref ref-type="fig" rid="F5">5B</xref>). Localization was predominantly cytoplasmic in embryos at least until the 8 cell stage. We noticed that the protein appeared to concentrate in the cortical region of mature oocytes and embryos up to 8-cells, as previously observed in mouse preimplantation embryos [<xref ref-type="bibr" rid="B3">3</xref>]. A similar peripheral concentration had also been reported for the ovary-specific 2',5'-oligoadenylate synthetase-like protein (OAS1D) in metaphase II oocytes [<xref ref-type="bibr" rid="B20">20</xref>], and for DNMT1o in mature oocytes, 2- and 4-cell embryos [<xref ref-type="bibr" rid="B16">16</xref>]. The authors suggested that DNMT1o may be sequestered near the plasma membrane through interaction with annexin V, a calcium-sensitive phospholipid binding protein [<xref ref-type="bibr" rid="B16">16</xref>]. MATER includes protein-protein interacting domains and identifying potential partners may help elucidating this intriguing distribution.</p><p>In GV oocytes, MATER localization was refined by immunogold coupled to transmission electron microscopy analysis. The protein was detected in the cytosol but not in organelles (Fig. <xref ref-type="fig" rid="F6">6</xref>). This pattern differs from MATER distribution in mouse, where the protein was detected in the nucleus close to nuclear pores and in mitochondria [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>Based on the presence of a leucine-rich repreat domain homologous to a porcine ribonuclease inhibitor, mouse MATER has been suggested to play a role in mRNA stability. Altogether, our observations on bovine MATER distribution are consistent with this hypothesis. In oocytes, MATER would stabilize transcripts from the moment when they are exported from the nucleus and during their storage in the oocyte cytoplasm up until the time when there are recruited for translation in the oocyte or after fertilization. Through a similar peri-nuclear accumulation in early embryos, MATER could protect the rare transcripts produced from the embryonic genome at these stages. Finally, MATER is degraded after hatching, once the embryo synthesizes large amount of messengers and proteins. This model is consistent with the phenotype of embryos from <italic>Mater </italic>null female mice: transcripts required for development beyond the 2-cell stage (e.g. in genome activation) would not be stabilized, resulting in developmental block.</p></sec></sec><sec><title>Conclusion</title><p>We have demonstrated that MATER transcript and protein are specifically expressed in oocyte from the primary follicle onwards in bovine ovary. The protein is stored in the cytoplasm of growing oocytes and persists during maturation, fertilization and in embryos until the expanded blastocyst stage. It is mostly degraded after hatching. Therefore <italic>MATER </italic>transcript and protein expression patterns are consistent with a maternal effect function conserved in bovine. Challenging functional approaches are now required to confirm this hypothesis.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Oocyte collection and in vitro embryo production</title><p>Ovaries from adult cows (unless otherwise specified) were collected at a slaughterhouse and the oocyte-cumulus complexes were aspirated from 3–6 mm follicles, selected based on morphological criteria and washed in saline solution (for details see [<xref ref-type="bibr" rid="B21">21</xref>]). Denuded immature and <italic>in vitro </italic>matured oocytes were obtained as previously described [<xref ref-type="bibr" rid="B22">22</xref>]. Briefly, oocyte-cumulus complexes were denuded by mechanical treatment either before or after <italic>in vitro </italic>maturation in TCM199 (Sigma, Saint Quentin Fallavier, France) supplemented with 10 ng/ml epidermal growth factor and 10% fetal calf serum for 24 hr at 39°C in water-saturated air with 5% carbon dioxide. For <italic>in vitro </italic>fertilization, groups of 50 intact <italic>in vitro </italic>matured oocyte-cumulus complexes were transferred into 500 μl fertilization medium containing 10<sup>6 </sup>motile spermatozoa. 24 hr later, presumptive zygotes were denuded. Groups of 25 zygotes were transferred into 25 μl droplets (under paraffin oil) of modified synthetic oviduct fluid supplemented with 5% fetal calf serum. Embryos were cultured at 39°C in a water-saturated atmosphere with 5% CO<sub>2</sub>/5% O<sub>2</sub>/90% N<sub>2</sub>. Groups of 10 embryos were collected over preimplantation development period: zygote and 2-cell embryos (day 2), 4-cell, 5 to 8-cell embryos (day 3), morulae (day 5), expanded blastocysts (day 6) and hatched blastocysts (day 8). All oocyte and embryo samples were stored frozen at -80°C. Two and four independent sets of oocytes and embryos were collected for immunological and real-time PCR analyses respectively.</p></sec><sec><title>In situ hybridization</title><p>Ovaries from 6 month old calves were embedded in Tissue-Tek medium (Sekura, Bayer Diagnostics, France), frozen in liquid nitrogen and then serially sectioned (10 μm) with cryostat. The sections were fixed in 4% paraformaldehyde. A 391 bp <italic>MATER </italic>PCR fragment (corresponding to nt 2819–3209 of the coding sequence) was subcloned in Dual Promoter pCRII plasmid (Invitrogen, Cergy-Pontoise, France). Sense and antisense probes were generated using Riboprobe combined system SP6/T7 (Promega, Charbonnières, France) and labelled with [<sup>35</sup>S]-UTP. Hybridization and washed were performed as described elsewhere [<xref ref-type="bibr" rid="B23">23</xref>]. Sections were counterstained with hematoxylin.</p></sec><sec><title>Real-time PCR</title><p>After adding 10 pg of luciferase mRNA (Promega) as an exogenous standard, total DNAse-treated RNA was extracted from 4 independent pools of 10 oocytes or embryos at each stage using the Picopure RNA isolation kit (Alphelys, Plaisir, France). Reverse transcription was performed using oligo(dT)<sub>15 </sub>primers (Promega) during 50 min at 37°C by mouse Moloney leukaemia virus reverse transcriptase (Invitrogen, cergy Pontoise, France). Target cDNA were then quantified by real-time PCR using iQ SYBR green supermix (Bio-Rad, Marnes la Coquette, France) in a MyCycler system (Bio-Rad) using specific primers for <italic>MATER </italic>(forward 5'-GCTGGAGGCGTGTGGACTG; reverse 5'-GGTCTGTAGATTAGAGGTGGGATGC) and <italic>luciferase </italic>(forward 5'-TCATTCTTCGCCAAAAGCACTCTG; reverse 5'-AGCCCATATCCTTGTCGTATCCC). A 3-step protocol was repeated for 40 cycles (95°C for 30 sec, 60°C for 30 sec, 72°C for 20 sec), followed by acquisition of the melting curve. cDNA amount equivalent to 0.05 oocyte or embryo was used in triplicate reactions. The standard curve was deduced from five serial dilutions (100 fg to 0.01 fg) of a plasmid including the target sequence included in each run. In each sample, the median value of PCR triplicates was considered (0 when not detected), and MATER data was normalized with exogenous luciferase to correct for loss of RNA; the MATER/luciferase value was then compared to this same ratio in immature oocytes (mean of four oocyte collections). Data are presented as mean ± SEM. After analysis of variance, first between immature oocytes and other stages, then between stages starting from mature oocyte onwards, differences were considered statistically significant at the 95% confidence level (P ≤ 0,05) using Tukey comparison test.</p></sec><sec><title>Immunohistochemistry</title><p>A keyhole limpet hemocyanin (KLH)-coupled MATER peptide (C terminus, aa 1084–1098) was used to immunize rabbits to produce the primary antibody (Eurogentec, Angers, France). Bovine ovaries were fixed in Bouin solution (50% saturated picric acid, 3.7% formaldehyde, 5% acetic acid), embedded in paraffin and serially sectioned (10 μm). The ovarian sections were immersed in antigen unmasking solution (Abcys, Paris, France) and warmed for 6 min in a microwave. After removing endogenous peroxydase activity in 0,3% H<sub>2</sub>O<sub>2 </sub>and blocking with horse serum, ovarian sections were incubated either with primary anti-MATER serum or rabbit preimmune serum (1:10000) overnight at 4°C, followed by incubation for 1 hr at room temperature with secondary biotinylated horse anti-mouse/rabbit/goat IgG antibody (1:200) (Abcys). Avidin, biotinylated horseradish peroxydase (HRP) and diamino-benzidene (DAB) were applied according to the manufacturer's instructions using Vectastain elite ABC kit (Abcys). Sections were counterstained with hematoxylin.</p></sec><sec><title>Western blot</title><p>Pools of 10 oocytes or embryos, or cumulus cells, were frozen and thawed three times, and then the proteins were separated on 7.5 or 10% SDS-PAGE gels and transferred onto a nitrocellulose membrane. After blocking with 5% dry milk in tris buffered saline, the blot was incubated with anti-MATER serum (1:10000) or pre-immune serum (1:10000) or purified antibody (1:2000) overnight at 4°C under agitation, and subsequently with secondary HRP-conjugated goat anti-rabbit IgG antibody (1:5000) for 1 hr 30 min at room. The signal was detected using an enhanced chemiluminescence kit according to the manufacturer's instructions (Amersham Biosciences, Orsay, France). The membranes were stripped and blotted with an anti-α-TUBULIN monoclonal antibody (Sigma). Experiment was performed on two independent oocyte and embryo collections.</p></sec><sec><title>Confocal scanning laser microscopy</title><p>Oocytes and embryos were fixed for 20 min in 4% paraformaldehyde. After treatment in phosphate buffered saline 1X/Triton 0.2% followed by blocking in 5% goat serum, they were subsequently incubated with either rabbit anti-MATER antibody purified onto a peptide affinity column by the manufacturer (1:250) or rabbit IgG overnight at 4°C, and then for 1 hr at room temperature with secondary goat anti-rabbit IgG antibody conjugated to fluoprobe 488 (1:100) or fluorescein isothiocyanate (1:400) for immature oocytes (both from Interchim, Montluçon, France). For LAMIN staining, immature oocytes were incubated with mouse anti-LAMIN A/C antibody (1:200; Ozyme, Saint Quentin Yvelines, France) for 2 hr at room temperature and with secondary Texas Red-conjugated goat anti-mouse antibody (1:400; Sigma), for 1 hr at room temperature. Chromatin was stained using Hoechst 33258 (Sigma). Mowiol was used as mounting medium.</p></sec><sec><title>Transmission electron microscopy (TEM)</title><p>Immature oocytes were fixed in 4% paraformaldehyde for 20 min at 4°C. Subsequently, they were washed in phosphate buffer 0.1 M, incubated in ammonium chloride 0.05 M in phosphate buffer for 30 min at room temperature and washed in phosphate buffer. Then, oocytes were dehydrated in ethanol and embedded in LRWhite resin (London Resin Company, Theale, England). Ultra-thin sections (80 nm) were placed on 200 mesh formvar-coated nickel grids and blocked in 10% goat serum and 1% BSA. The grids were then incubated with purified rabbit anti-MATER antibody (1:1000) overnight at 4°C, washed with PBS/0,1% BSA/1% goat serum and incubated with colloidal gold-labelled (18 nm) secondary goat anti-rabbit antibody (1:10, (Immunotech, Marseille, France) for 2 hr at room temperature. After several washing in PBS and deionised water, sections were counterstained with 4% uranyl acetate and lead citrate. Specificity of immunostaining was checked by incubation of sections with PBS/0,1% BSA/1% goat serum. Sections were observed with a CM10 electron microscope (Philips, Eindhoven, Netherlands).</p></sec></sec><sec><title>Authors' contributions</title><p>SP participated in experimental design and carried out most experimental work. CP collected biological material, carried out in vitro embryo production, and participated in immunohistochemistry. BD carried out the TEM experiments. SU was involved in biological sample collection, Western blot, and provided input in writing the manuscript. AT performed oocyte/embryo collection and real-time PCR. PM provided expertise in experimental design and analysis. RDT designed and supervised the study, and participated in experimental work (oocyte collection, Western blot). SP and RDT wrote the manuscript.</p></sec>
|
Genomic analysis of <italic>Xenopus </italic>organizer function
|
<sec><title>Background</title><p>Studies of the <italic>Xenopus </italic>organizer have laid the foundation for our understanding of the conserved signaling pathways that pattern vertebrate embryos during gastrulation. The two primary activities of the organizer, BMP and Wnt inhibition, can regulate a spectrum of genes that pattern essentially all aspects of the embryo during gastrulation. As our knowledge of organizer signaling grows, it is imperative that we begin knitting together our gene-level knowledge into genome-level signaling models. The goal of this paper was to identify complete lists of genes regulated by different aspects of organizer signaling, thereby providing a deeper understanding of the genomic mechanisms that underlie these complex and fundamental signaling events.</p></sec><sec><title>Results</title><p>To this end, we ectopically overexpress Noggin and Dkk-1, inhibitors of the BMP and Wnt pathways, respectively, within ventral tissues. After isolating embryonic ventral halves at early and late gastrulation, we analyze the transcriptional response to these molecules within the generated ectopic organizers using oligonucleotide microarrays. An efficient statistical analysis scheme, combined with a new Gene Ontology biological process annotation of the <italic>Xenopus </italic>genome, allows reliable and faithful clustering of molecules based upon their roles during gastrulation. From this data, we identify new organizer-related expression patterns for 19 genes. Moreover, our data sub-divides organizer genes into separate head and trunk organizing groups, which each show distinct responses to Noggin and Dkk-1 activity during gastrulation.</p></sec><sec><title>Conclusion</title><p>Our data provides a genomic view of the cohorts of genes that respond to Noggin and Dkk-1 activity, allowing us to separate the role of each in organizer function. These patterns demonstrate a model where BMP inhibition plays a largely inductive role during early developmental stages, thereby initiating the suites of genes needed to pattern dorsal tissues. Meanwhile, Wnt inhibition acts later during gastrulation, and is essential for maintenance of organizer gene expression throughout gastrulation, a role which may depend on its ability to block the expression of a host of ventral, posterior, and lateral fate-specifying factors.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Hufton</surname><given-names>Andrew L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Vinayagam</surname><given-names>Arunachalam</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Suhai</surname><given-names>Sándor</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" corresp="yes" contrib-type="author"><name><surname>Baker</surname><given-names>Julie C</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Developmental Biology
|
<sec><title>Background</title><p>The organizer is the primary patterning center during early vertebrate gastrulation. As might be expected for a tissue with such capabilities, the organizer is complex. Studies in multiple species, including frogs and mice, have shown that the organizer has distinct regions that induce head and trunk, and these abilities decisively change as development proceeds. At the molecular level, the organizer's inductive properties are mediated by factors that inhibit the BMP, Wnt, and Nodal signaling pathways. BMP inhibitors, including the secreted molecule Noggin, can induce a partial secondary axis that lacks a head and notochord. However, BMP inhibition alone cannot sustain the expression of most organizer genes past late gastrula without the addition of Wnt inhibitors, such as Dkk-1 [<xref ref-type="bibr" rid="B1">1</xref>]. Furthermore, Wnt inhibitors alone cannot induce secondary structures, but when combined with Noggin can induce a complete secondary axis, including properly patterned head and trunk tissues. Therefore, inhibition of both pathways generates the complete spectrum of molecules required for total organizer function and maintenance, illustrating that regional differences in organizer activity are created by the mixes of inhibitors present and active within particular regions (reviewed in [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>]).</p><p>Research has identified a host of genes that, under the control of the organizer, pattern different aspects of the embryo during gastrulation. As these studies collectively build an ever more complicated tangle of genetic interactions, it is imperative that we begin knitting together our gene-level knowledge into genome-level signaling models. A global analysis can identify comprehensive sets of genes that respond to different aspects of organizer signals i.e. head versus trunk, thus giving us a complete toolbox in which to study the molecular mechanisms regulating organizer function within different contexts and through developmental time. Mapping these genome-level patterns of organizer regulation will allow us to fill-out the current models of gastrula patterning with a greater degree of detail. With these goals in mind, microarray experiments hold particular promise.</p><p>Several <italic>Xenopus </italic>microarray-based experiments have been published in recent years as genomic tools have become available. A series of papers have used two-condition comparisons to identify genes up- or down-regulated by a particular process, starting with the cDNA arrays produced by the Brivanlou lab [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>] and recently using the more comprehensive cDNA arrays developed in the Cho and Ueno labs [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. These methods have been effective in producing new lists of candidate genes, and in two cases have been used to identify genes with new overexpression or morpholino knockdown phenotypes [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. In addition to this two-condition design, studies in other organisms have shown that microarray experiments that employ multiple conditions can be used to cluster genes based on their expression patterns across the samples, and that within these clusters, genes of common function will often group together [<xref ref-type="bibr" rid="B11">11</xref>]. This method has been applied fruitfully to the study of specific events in the early development of invertebrates. Some notable examples include the <italic>Drosophila </italic>studies of dorsal-ventral patterning [<xref ref-type="bibr" rid="B12">12</xref>] and mesoderm formation [<xref ref-type="bibr" rid="B13">13</xref>], where in each case the microarray data was able to subdivide genes based on their roles in these processes. In fact, this type of analysis has recently been applied to the <italic>Xenopus </italic>model; thirty-seven different tissue types were profiled using cDNA arrays creating a broad view of gene expression across development [<xref ref-type="bibr" rid="B14">14</xref>]. The resulting cluster data successfully grouped genes with common molecular functions and identified many new tissue specific genes. Moreover, a study by Wessely et al. used an innovative macroarray technique to describe the suites of genes that underlie organizer formation prior to gastrulation, suggesting that genomic methods do have much to offer to early developmental studies [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>In this paper, we present a genomic view of the signaling processes that underlie certain aspects of organizer function during gastrulation. To this end, we simulated either trunk or head organizer induction by expressing Noggin and/or Dkk-1 within the ventral mesoderm and subsequently analyzed these tissues with oligonucleotide arrays. From this data, we identify genes whose expression levels respond to organizer signaling and then cluster these genes based on their pattern of response. By combining these cluster results with a new GO biological process annotation of the <italic>Xenopus </italic>genome, we are able to rapidly identify clusters that are highly enriched for known gastrula patterning genes. These patterns accurately predict the expression patterns of unknown genes within the clusters: of 19 genes that show a specific pattern during gastrulation, all show organizer enrichment or exclusion in accord with our cluster predictions. Moreover, the cluster patterns allow us to make biological conclusions about the genomic mechanisms that underlie organizer signaling, shedding new light on the divisions between the head and trunk organizer programs.</p></sec><sec><title>Results and discussion</title><sec><title>Creating ectopic organizers with separate functions using Noggin and Dkk-1</title><p>In order to describe and separate the genomic expression changes induced by the two main organizing activities, BMP inhibition and Wnt inhibition, we ectopically overexpressed one or both of these activities in ventral mesoderm, and compared these samples to endogenous dorsal and ventral mesoderm at early and late gastrula stages. Two well-studied organizer secreted factors, Noggin [<xref ref-type="bibr" rid="B16">16</xref>] and Dkk-1 [<xref ref-type="bibr" rid="B17">17</xref>], were used to ectopically inhibit BMP and Wnt signaling, respectively. Four different mixtures were injected ventrally into 4-cell embryos: <italic>noggin </italic>and <italic>eGFP </italic>(anti-BMP); <italic>noggin</italic>, <italic>dkk-1</italic>, and <italic>eGFP </italic>(anti-BMP and anti-Wnt); <italic>dkk-1 </italic>and <italic>eGFP </italic>(anti-Wnt); and <italic>eGFP </italic>alone (Figure <xref ref-type="fig" rid="F1">1B</xref>). <italic>eGFP </italic>mRNA was used to trace targeting. Plasmid DNA was used for both Noggin and Dkk-1 to ensure that these molecules were only expressed after the start of zygotic transcription (mid-blastula transition), thereby mimicking the endogenous regulation of these genes. The ventrally injected embryos were grown to early stage 10, sorted for appropriately targeted eGFP florescence, and then bisected between the dorsal and ventral halves at either stage 10 (early gastrula) or stage 11.5 (late gastrula) (Figure <xref ref-type="fig" rid="F1">1A</xref>). For the <italic>noggin </italic>and/or <italic>dkk-1 </italic>injected embryos, only the ventral halves of the embryos were saved, eliminating endogenous organizer tissues. For the embryos injected with only <italic>eGFP</italic>, both the ventral and dorsal halves were saved, creating separate ventral (<bold>Ven</bold>) and dorsal (<bold>Dor</bold>) samples. These five conditions (Figure <xref ref-type="fig" rid="F1">1C</xref>) were each generated twice at both stage 10 (early gastrula) and 11.5 (late gastrula), creating twenty total tissue samples (Table <xref ref-type="table" rid="T1">1</xref>). For each batch of injections, remaining sorted embryos were allowed to develop through tailbud stages in order to validate the phenotypes induced by our constructs (Table <xref ref-type="table" rid="T1">1</xref>). Total RNA was then isolated from the twenty tissue samples and hybridized to Affymetrix oligonucleotide arrays (see methods).</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Generating tissue samples with different aspects of organizer activity</bold>. (A) shows an overview of injection and embryo sorting procedure used to produce samples for microarray analysis. (B) shows the four injection mixtures below their respective tailbud phenotypes. Embryos ventrally injected with these mixtures were bisected at either stg. 10 or stg. 11.5 to produce the tissue conditions in (C).</p></caption><graphic xlink:href="1471-213X-6-27-1"/></fig><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Secondary axis phenotypes are improved by sorting for proper targeting</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Clutch</bold></td><td align="left"><bold>Stage</bold></td><td align="left"><bold>Sample</bold></td><td align="left"><bold>Branched NT</bold></td><td align="left"><bold>Normal</bold></td></tr></thead><tbody><tr><td align="left">1</td><td align="left">10</td><td align="left">Nog</td><td align="left">17 (100%)</td><td align="left">0</td></tr><tr><td align="left">1</td><td align="left">10</td><td align="left">Nog+Dkk</td><td align="left">9 (100%)</td><td align="left">0</td></tr><tr><td align="left">1</td><td align="left">10</td><td align="left">Dkk</td><td align="left">0 (0%)</td><td align="left">11</td></tr><tr><td align="left">1</td><td align="left">10</td><td align="left">eGFP Con.</td><td align="left">0 (0%)</td><td align="left">19</td></tr><tr><td colspan="5"><hr></hr></td></tr><tr><td align="left">2</td><td align="left">10</td><td align="left">Nog</td><td align="left">16 (100%)</td><td align="left">0</td></tr><tr><td align="left">2</td><td align="left">10</td><td align="left">Nog+Dkk</td><td align="left">12 (100%)</td><td align="left">0</td></tr><tr><td align="left">2</td><td align="left">10</td><td align="left">Dkk</td><td align="left">0 (0%)</td><td align="left">11</td></tr><tr><td align="left">2</td><td align="left">10</td><td align="left">eGFP Con.</td><td align="left">0 (0%)</td><td align="left">14</td></tr><tr><td colspan="5"><hr></hr></td></tr><tr><td align="left">3</td><td align="left">11.5</td><td align="left">Nog</td><td align="left">19 (100%)</td><td align="left">0</td></tr><tr><td align="left">3</td><td align="left">11.5</td><td align="left">Nog+Dkk</td><td align="left">13 (100%)</td><td align="left">0</td></tr><tr><td align="left">3</td><td align="left">11.5</td><td align="left">Dkk</td><td align="left">0 (0%)</td><td align="left">15</td></tr><tr><td align="left">3</td><td align="left">11.5</td><td align="left">eGFP Con.</td><td align="left">0 (0%)</td><td align="left">13</td></tr><tr><td colspan="5"><hr></hr></td></tr><tr><td align="left">4</td><td align="left">11.5</td><td align="left">Nog</td><td align="left">16 (94%)</td><td align="left">1</td></tr><tr><td align="left">4</td><td align="left">11.5</td><td align="left">Nog+Dkk</td><td align="left">22 (100%)</td><td align="left">0</td></tr><tr><td align="left">4</td><td align="left">11.5</td><td align="left">Dkk</td><td align="left">0 (0%)</td><td align="left">16</td></tr><tr><td align="left">4</td><td align="left">11.5</td><td align="left">eGFP Con.</td><td align="left">0 (0%)</td><td align="left">15</td></tr></tbody></table><table-wrap-foot><p>Sorted embryos assayed at late neurula for the presence of a branched neural tube (NT).</p></table-wrap-foot></table-wrap><p>In order to maximize our ability to detect organizer-related expression changes in the microarray data, it was essential that we appropriately control for other sources of biological variability. First, all samples within each replicate were generated from a single clutch of embryos, thereby controlling for genetic heterogeneity in <italic>Xenopus laevis </italic>laboratory populations, which creates observable molecular and morphological differences between clutches. Arima <italic>et al</italic>., using a similar approach, greatly reduced their false detection rate compared to earlier experiments that pooled several clutches in each sample [<xref ref-type="bibr" rid="B7">7</xref>]. Second, we reduced variability caused by phenotypic penetrance differences by pre-selecting embryos that showed proper injection targeting. Phenotypic penetrance variation was a concern since both <italic>noggin </italic>and combinations of <italic>noggin </italic>and <italic>dkk-1 </italic>have previously been observed to induce secondary axes in about 50% of ventrally injected embryos [<xref ref-type="bibr" rid="B17">17</xref>]. We tested whether this low penetrance could be improved by ensuring accurate targeting to the ventral mesoderm. To this end, we coinjected our constructs with <italic>eGFP </italic>and selected only embryos that contained eGFP fluorescence within the ventral mesoderm at early stage 10. In these sorted embryos, Noggin and Noggin+Dkk-1 were able to induce secondary axes in nearly all the eGFP-sorted embryos (Table <xref ref-type="table" rid="T1">1</xref>). Therefore, this sorting protocol (Figure <xref ref-type="fig" rid="F1">1A</xref>) was used on all the batches generated for microarray analysis and eliminates the concern of phenotypic variability in response to the injected molecules.</p><p>These manipulations produce five different tissue conditions, at two gastrula stages, that each contain functionally distinct organizing capabilities (Figure <xref ref-type="fig" rid="F1">1 B&C</xref>). Ventral mesoderm expressing Noggin (<bold>Nog</bold>) is sufficient to induce secondary trunk tissues. Ventral mesoderm expressing both Noggin and Dkk-1 (<bold>Nog+Dkk</bold>) is sufficient to induce a full secondary axis, including a properly patterned ectopic head. Ventral mesoderm expressing Dkk-1 (<bold>Dkk</bold>) cannot induce any ectopic tissues. Endogenous dorsal tissues (<bold>Dor</bold>) include full organizer capabilities, while endogenous ventral tissue (<bold>Ven</bold>) is used as a baseline control sample throughout the microarray analysis. These samples are designed to contain much informational redundancy in that many gastrula patterning genes should respond to more than one of these conditions. For example, head inducing genes are likely to be up-regulated in both <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions, when compared to the <bold>Ven </bold>condition. Furthermore, all conditions were collected at both stage 10 and 11.5, bolstering our power to detect organizer-regulated genes, and allowing us to compare early and late responses to organizer signaling.</p></sec><sec><title>Reliable differences in genomic transcription observed in response to Noggin and Dkk-1</title><p>To visualize the extent to which each of our experimental treatments induced changes in the population of genes on the array, we used scatter plots and regression analysis. Figure <xref ref-type="fig" rid="F2">2A</xref> and <xref ref-type="fig" rid="F2">2B</xref> show comparisons of the averaged stage 10 <bold>Nog+Dkk </bold>or <bold>Dor </bold>samples, versus the averaged stage 10 <bold>Ven </bold>samples (scatter plots for the remaining conditions are shown in <xref ref-type="supplementary-material" rid="S1">Additional file 1 A–F</xref>). The majority of the 15,491 probe sets show highly similar expression levels. Nonetheless, there are clearly differentially expressed genes in both comparisons, which in these cases are weighted towards genes that are overexpressed in the <bold>Nog+Dkk </bold>or <bold>Dor </bold>samples. Moreover, there is a greater spread in the <bold>Dor </bold>vs <bold>Ven </bold>plot, which is not surprising since these samples are from distinct endogenous sources. Figure <xref ref-type="fig" rid="F2">2C</xref> shows a summary of the R-squared values, a regression measure of the correlation between each condition and its stage-matched <bold>Ven </bold>condition. The three conditions that contain axis inducing capabilities (<bold>Nog</bold>, <bold>Nog+Dkk</bold>, <bold>Dor</bold>) show more variation from <bold>Ven </bold>than the <bold>Dkk </bold>conditions, which cannot induce axial tissues. Moreover, the stage 11.5 samples, which have had more time for change to occur, show consistently greater divergence from their stage-matched <bold>Ven </bold>samples than the stage 10 samples. Together these results confirm that organizer signaling differences are driving detectable changes in gene expression within the samples.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Expression variation among the samples</bold>. (A) and (B) show scatter plot comparisons of the mean <bold>Nog+Dkk </bold>or <bold>Dor </bold>stg. 10 log<sub>2 </sub>expression values vs the mean stg. 10 <bold>Ven </bold>log<sub>2</sub>expression values. Probe sets measuring two known organizer genes, <italic>otx2 </italic>(green) and <italic>gsc </italic>(red), are labeled within the plots (<italic>otx2 </italic>probe sets: Xl.1268.1.S1_at, Xl.3004.1.A1_at, Xl.11672.1.A1_at, and XlAffx.1.11.S1_at; <italic>gsc </italic>probe sets: Xl.801.1.A1_at, Xl.801.1.S1_at, and Xl.801.1.S1_s_at). (C) R-square regression value summaries for each experimental condition compared to the stage-matched ventral condition. (D) Scatter plot comparison of the mean <bold>Nog </bold>and <bold>Ven </bold>clutch-1 log<sub>2 </sub>expression values vs. the mean <bold>Nog </bold>and <bold>Ven </bold>clutch-2 log<sub>2 </sub>expression values. Note that the clutch variation clearly exceeds differences between the experimental conditions (A-C). Contributing to this variation, many genes show a shift in the average expression between the two clutches, but retain a similar pattern in response to the experimental conditions. As an example, log<sub>2 </sub>expression values for <italic>otx5-A </italic>(Xl.3452.1.A1_at) are shown in (E). After standardizing the average log<sub>2 </sub>expression of both clutches, on a gene-by-gene basis, the comparison from (D) was repeated in (F), eliminating most clutch variation.</p></caption><graphic xlink:href="1471-213X-6-27-2"/></fig><p>To further assess our ability to detect organizer-induced expression changes, we marked probes that correspond to two well-known organizer genes, <italic>otx2 </italic>[<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>] and <italic>goosecoid </italic>(<italic>gsc</italic>) [<xref ref-type="bibr" rid="B20">20</xref>], within the previous scatter plots (Figure <xref ref-type="fig" rid="F2">2 A&B</xref>). Both of these genes are present more than once on the array, thereby also providing a sampling of the consistency among different probe sets. All of the four <italic>otx2 </italic>probes and two of the three <italic>gsc </italic>probes appear to be up-regulated in both the <bold>Nog+Dkk </bold>and the <bold>Dor </bold>samples, as expected. However, in the <bold>Nog+Dkk </bold>plot these inductions appear relatively subtle, highlighting the need for solid statistical methods to select genes that are genuinely altered.</p><p>Because of our initial concerns about population heterogeneity in <italic>Xenopus laevis</italic>, we also analyzed the variation between the different clutches used to generate our replicates. To visualize this variation we averaged together the stage 10 <bold>Nog </bold>and <bold>Ven </bold>samples from clutch 1 and compared them by scatter plot to the similarly averaged samples from clutch 2 (Figure <xref ref-type="fig" rid="F2">2D</xref>), effectively reducing experimentally induced variation, while leaving clutch specific variation (single array comparisons are also shown in <xref ref-type="supplementary-material" rid="S1">Additional file 1 G–I</xref>). Clearly, the differences between the clutches exceed the differences between the experimental conditions. However, when we look at the expression patterns of single genes (Figure <xref ref-type="fig" rid="F2">2E</xref>), we see that both clutches show a similar pattern of expression across the five samples, while the basal level of expression differs between the two clutches. This clutch to clutch shift can be observed in the expression of many genes, however the magnitude and direction of the shift is different for each gene, suggesting that this is not merely an array to array normalization problem. If we subtract out these differences in basal expression, by standardizing the average expression for each gene between the two clutches, we find that these basal shifts explain the majority of the clutch to clutch variation (Figure <xref ref-type="fig" rid="F2">2F</xref>). Similar patterns were observed in the stage 11.5 data (data not shown). While these clutch differences are striking, organizer signaling induced variation can be easily filtered away from the clutch variation. Replicate averaging removed this source of variation from the scatter plot comparisons between the experimental samples (Figure <xref ref-type="fig" rid="F2">2 A&B</xref>), and statistical tests can now be conducted by making direct comparisons only between clutch matched samples.</p></sec><sec><title>A successful three-step method that enriches for organizer-related molecules</title><p>One of the goals of this study was to functionally subdivide – on a genomic level – genes that are regulated by different aspects of organizer signaling. To this end, we employed a three step computational process to ensure solid statistical analysis and to group genes into clusters that share similar roles during gastrulation.</p><p>First, to select genes altered by organizer signaling we used the rank products (RP) method [<xref ref-type="bibr" rid="B21">21</xref>]. For each comparison between an experimental sample and its clutch-matched <bold>Ven </bold>control, this method ranks genes by their log-ratio expression difference. Then, the rank numbers from the two replicate sample pairs are multiplied together, producing a 'rank product' score for each gene. From this rank product, a permutation based false detection rate (FDR) can be calculated. This method only makes direct comparisons between clutch-matched samples, eliminating concerns about clutch variation. Moreover, since significance calculations are based on ranks and not absolute values, the method does not assume normality. Up- and down-regulated genes are tested for separately, creating a total of eight statistical tests. Table <xref ref-type="table" rid="T2">2</xref> summarizes the number of genes that passes each test at a 10% FDR. For further analysis, we selected all genes that were significantly altered by overexpression of Noggin and/or Dkk-1, creating a broad list of genes that are likely to be involved in gastrula patterning. At 10% FDR, 188 probe sets are identified that are up- or down-regulated in the <bold>Nog</bold>, <bold>Nog+Dkk</bold>, or <bold>Dkk </bold>conditions, when compared to <bold>Ven</bold>, at either stage 10 or 11.5 (<xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Probe sets declared altered vs Ven by RP test at 10% FDR</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Condition</td><td align="left">Stage</td><td align="left">Increased</td><td align="left">Decreased</td></tr></thead><tbody><tr><td align="left"><bold>Nog</bold></td><td align="left">10</td><td align="left">25</td><td align="left">0</td></tr><tr><td align="left"><bold>Nog+Dkk</bold></td><td align="left">10</td><td align="left">47</td><td align="left">4</td></tr><tr><td align="left"><bold>Dkk</bold></td><td align="left">10</td><td align="left">19</td><td align="left">0</td></tr><tr><td align="left"><bold>Dor</bold></td><td align="left">10</td><td align="left">95</td><td align="left">54</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left"><bold>Nog</bold></td><td align="left">11.5</td><td align="left">62</td><td align="left">10</td></tr><tr><td align="left"><bold>Nog+Dkk</bold></td><td align="left">11.5</td><td align="left">46</td><td align="left">34</td></tr><tr><td align="left"><bold>Dkk</bold></td><td align="left">11.5</td><td align="left">25</td><td align="left">47</td></tr><tr><td align="left"><bold>Dor</bold></td><td align="left">11.5</td><td align="left">109</td><td align="left">80</td></tr></tbody></table></table-wrap><p>Second, hierarchical clustering was performed on this list of 188 probe sets. Replicates were averaged before clustering, and stage 10 and stage 11.5 samples were standardized separately so that the clustering was driven by expression differences between the sample conditions, not the stages. An overview of the entire cluster is shown in Figures <xref ref-type="fig" rid="F3">3</xref>; <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref> lists the probe set names and their relevant annotations in the same order as the clustergram, allowing referencing to the specific gene identities. The results of the RP method tests at the 10% FDR level are summarized next to the clustergram; a key in Figure <xref ref-type="fig" rid="F3">3</xref> explains the colors used to denote these results. By definition each gene has at least one positive test in the first three columns, but many genes passed more than one test. The clustering of these genes divides them into approximately four main groups, with each showing a distinctive pattern of RP test results (Figure <xref ref-type="fig" rid="F3">3</xref>). The first three groups, containing 119 of the 188 probe sets, are composed of genes that are generally up-regulated compared to <bold>Ven</bold>, while the last group contains the genes that are generally down-regulated compared to <bold>Ven</bold>.</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Clustering genes regulated by organizer signaling</bold>. (A) Key to the hierarchical cluster format used throughout the paper. The clustergram shows the standardized expression intensity for the ten experimental conditions, after replicates have been averaged. To the right of the clustergram, the RP method results at 10% FDR are summarized in four columns representing the comparisons of <bold>Nog, Nog+Dkk</bold>, <bold>Dkk</bold>, or <bold>Dor </bold>to <bold>Ven</bold>. The colors found in the row for each gene represent the tests passed by that gene. Two colors in one column indicate that a gene passed the column's test at both stages. (B) Hierarchical cluster of all the selected genes. The far right shows the hierarchical cluster tree, followed by the clustergram, then RP results. Black ticks between the cluster tree and the clustergram mark every tenth gene, allowing referencing to <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref> for the gene identities. The list break into four main clusters, labeled with red, yellow, orange, and magenta bars. Expanded views of clusters 3 and 4 can be found in Figures 4 and 5, respectively.</p></caption><graphic xlink:href="1471-213X-6-27-3"/></fig><p>Third, in order to rapidly screen the identified clusters for possible enrichments of developmental processes, we developed a machine annotation method to adopt Biological Process GO annotations from other annotated genomes (see methods). Each of the four main groups from the cluster was then tested for statistically significant enrichments of Biological Process terms (<p = 0.05 after multiple test correction) as compared to the population of genes on the array using the EASE method [<xref ref-type="bibr" rid="B22">22</xref>]. The first two clusters show no significant enrichments, while clusters 3 is enriched for 35 terms and cluster 4 is enriched for 27 terms. The top ten terms from each, by p-value, are shown in Table <xref ref-type="table" rid="T3">3</xref>. Both clusters share several terms that describe different developmental processes (neurogenesis, morphogenesis, development, organogenesis), and terms related to transcriptional regulation (regulation of transcription from Pol II promoter; regulation of transcription, DNA-dependent; transcription from Pol II promoter). Together these terms indicate strong enrichment for processes involved with development. Moreover, there is a strong correlation in these clusters between the RP test results from the <bold>Nog+Dkk </bold>and the <bold>Dor </bold>conditions, as one would expect from genes that are endogenously relevant to organizer signaling.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Enriched biological process GO terms</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Enriched GO Biological Process Term</bold></td><td align="left"><bold>p-value</bold></td></tr></thead><tbody><tr><td align="left" colspan="2"><bold>Cluster 1</bold></td></tr><tr><td align="left">No significant enrichments</td><td></td></tr><tr><td align="left" colspan="2"><bold>Cluster 2</bold></td></tr><tr><td align="left">No significant enrichments</td><td></td></tr><tr><td align="left" colspan="2"><bold>Cluster 3</bold></td></tr><tr><td align="left">neurogenesis</td><td align="left">9.37E-05</td></tr><tr><td align="left">development</td><td align="left">1.03E-04</td></tr><tr><td align="left">organogenesis</td><td align="left">1.48E-04</td></tr><tr><td align="left">regulation of transcription from Pol II promoter</td><td align="left">1.49E-04</td></tr><tr><td align="left">imaginal disc development</td><td align="left">1.79E-04</td></tr><tr><td align="left">morphogenesis</td><td align="left">1.91E-04</td></tr><tr><td align="left">transcription from Pol II promoter</td><td align="left">2.99E-04</td></tr><tr><td align="left">regulation of transcription, DNA-dependent</td><td align="left">5.46E-04</td></tr><tr><td align="left">regulation of transcription</td><td align="left">6.78E-04</td></tr><tr><td align="left">regulation of nucleobase, nucleoside, nucleotide and nucleic acid metabolism</td><td align="left">7.43E-04</td></tr><tr><td align="left" colspan="2"><bold>Cluster 4</bold></td></tr><tr><td align="left">organogenesis</td><td align="left">5.15E-11</td></tr><tr><td align="left">morphogenesis</td><td align="left">8.04E-11</td></tr><tr><td align="left">development</td><td align="left">8.59E-11</td></tr><tr><td align="left">pattern specification</td><td align="left">3.99E-08</td></tr><tr><td align="left">neurogenesis</td><td align="left">1.24E-07</td></tr><tr><td align="left">brain development</td><td align="left">2.56E-07</td></tr><tr><td align="left">central nervous system development</td><td align="left">6.26E-07</td></tr><tr><td align="left">regulation of transcription from Pol II promoter</td><td align="left">1.96E-06</td></tr><tr><td align="left">transcription from Pol II promoter</td><td align="left">4.80E-06</td></tr><tr><td align="left">regulation of transcription, DNA-dependent</td><td align="left">1.04E-05</td></tr></tbody></table></table-wrap></sec><sec><title>Cluster 3 and cluster 4 exclusively contain organizer or ventral-posteriorizing functions, respectively</title><p>In order to both verify our GO annotation descriptions and to provide a deeper understanding of the biological processes behind these clusters, we searched the literature for publications relating any of these genes to early development. Fitting the GO predictions, among clusters 1 and 2 we found only a handful of known developmental regulators:<italic>fibroblast growth factor receptor 2 </italic>[<xref ref-type="bibr" rid="B38">38</xref>], <italic>secreted frizzled-related sequence protein 2 </italic>(<italic>sfrp2</italic>) [<xref ref-type="bibr" rid="B39">39</xref>], and <italic>mix.3 </italic>(<italic>mixer</italic>) [<xref ref-type="bibr" rid="B13">13</xref>] (<xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>). Furthermore, clusters 3 and 4, which the GO annotations identified as significantly enriched for developmental terms, contain many published developmental regulators. Cluster 3 is densely populated with genes that have known roles in organizer function (Figure <xref ref-type="fig" rid="F4">4</xref>). Of the 31 unique genes in this cluster, 22 (71%) have described roles in organizer function. Of the 9 remaining genes, only 2 have described roles in early development that are not clearly organizer related, and both of these genes are distinct outliers to the cluster group. Moreover, cluster 4 is largely comprised of genes involved in gastrula and neurula stage patterning of ventral, lateral, and posterior tissues – all tissue types that are repressed by organizer activity (Figure <xref ref-type="fig" rid="F5">5</xref>). Of the 52 unique genes in this cluster, 31 (60%) have a described role in the patterning of these tissues. 27 are known or predicted transcriptional regulators, of which 17 encode homeobox proteins, highlighting the importance of these genes in developmental patterning. This cluster also contains a small sub-cluster of genes with an interesting and unexpected pattern. These genes are induced at stage 10 in the three ectopic organizer conditions,<bold>Nog</bold>, <bold>Nog+Dkk</bold>, and <bold>Dkk</bold>, but are strongly down-regulated in the stage 11.5 <bold>Dor </bold>condition. Within this sub-cluster is the G-protein coupled receptor gene <italic>X-msr (Xangio1)</italic>, a known marker of paraxial mesoderm [<xref ref-type="bibr" rid="B23">23</xref>], as well as two unstudied genes encoding a paralogous G-protein coupled receptor (Xl.34.1.S1_at), and a ras-like protein (Xl.13019.1.S1_at), suggesting that perhaps this cluster contains components of an unknown signaling process. Overall, a combination of GO annotations and literature validation has revealed two clusters of genes within our dataset that represent specific and discrete functions occurring during gastrula: organizer function and dorsal patterning (cluster 3); and ventral, lateral, and posterior patterning (cluster 4).</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Cluster 3 is enriched for genes involved in organizer function</bold>. This figure shows an enlarged view of all of the genes in cluster 3, from Figure 3. Each row is annotated with the probe set number and matching gene name. Genes with names in blue have a described role in organizer function. Genes with names in red have no described function during gastrula stage development. Genes with names in green have a published role or expression pattern that is not organizer related. † gene name was assigned by protein sequence homology using the NCBI Homologene database.</p></caption><graphic xlink:href="1471-213X-6-27-4"/></fig><fig position="float" id="F5"><label>Figure 5</label><caption><p><bold>Cluster 4 is enriched for genes involved in ventral, lateral, and posterior patterning</bold>. This figure shows an enlarged view of all the genes in cluster 4, from Figure 3. Each row is annotated with the probe set number and matching gene name. Genes with names in blue have a described role or specific expression pattern in ventral, lateral, or posterior tissues. Genes with names in red have no described function during gastrula stage development. Genes with names green have a published role or expression pattern that is not ventral, lateral, or posterior related. † gene name was assigned by protein sequence homology using the Homologene database.</p></caption><graphic xlink:href="1471-213X-6-27-5"/></fig></sec><sec><title>Cluster 3 and cluster 4 reveal 12 new genes with organizer-related expression patterns</title><p>Since this method of clustering clearly enriches for known organizer-related genes, our next step was to test whether our clustering method could also successfully identify unknown genes involved in these processes. To that end, we selected genes without known gastrula-stage expression patterns from our two developmentally enriched clusters and analyzed each gene by whole mount <italic>in situ </italic>hybridization. We obtained clones for twenty-one genes: six clones from the organizer gene rich cluster 3, and fifteen clones from the ventral/posterior/lateral gene rich cluster 4. <italic>In situ </italic>results are shown in Figure <xref ref-type="fig" rid="F6">6</xref>. Of the 12 clones that showed clear patterns during gastrulation, all clones from cluster 3 showed organizer enrichment, and all clones from cluster 4 showed organizer exclusion. Moreover, there appear to be finer correlations between some of the <italic>in situ </italic>patterns and the location of the genes within the cluster. For example, both <italic>Xl.13537.1.A1_at </italic>and <italic>Xl.15738.1.S1_at </italic>show expression that starts in the dorsal lip and then migrates anteriorly with the developing head tissues, displaying patterns that are similar to their immediate flanking neighbors in the cluster, <italic>gsc </italic>[<xref ref-type="bibr" rid="B20">20</xref>] and <italic>otx2 </italic>[<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>]. Overall, these clusters faithfully classify transcripts by their gastrulation expression.</p><fig position="float" id="F6"><label>Figure 6</label><caption><p><bold>Clusters 3 and 4 faithfully predict expression patterns for unknown genes</bold>. Genes found in clusters 3 and 4 that lacked described gastrula stage expression patterns were analyzed by whole mount <italic>in situ </italic>hybridization. Unknown genes from cluster 3 that showed a specific pattern are enriched in organizer tissues (orange box), and unknown genes from cluster 4 that showed a specific pattern are excluded from organizer tissues (magenta box). Each tested gene is labeled with its name and the Affymetrix probe set number. Genes marked "no pattern" showed no staining, or a non-specific staining pattern that was similar to sense controls. Genes marked "no gastrula pattern," showed no pattern during gastrula stages, but did show specific patterns at later stages that are not shown here. Each photo is labeled with the developmental stage of the embryo in the bottom left corner, and the orientation in the bottom right corner. veg: vegetal view, dorsal faces up. dor: dorsal view, anterior faces up.</p></caption><graphic xlink:href="1471-213X-6-27-6"/></fig><p>The majority of the unknown genes within these clusters have no previous associations with development and no known function. However, several have previous links that implicate them in developmental patterning, and our expression patterns are consistent with these previous findings. A clone related to <italic>LRIG2 </italic>was recently identified in a microarray screen for neural specific genes, and showed an identical expression pattern to our clone (NIBB clone XL098p21 [<xref ref-type="bibr" rid="B9">9</xref>]). <italic>AGTRL1 </italic>and <italic>frzb3 </italic>each have close paralogs in <italic>Xenopus </italic>that have described expression patterns identical to our genes: <italic>X-msr </italic>[<xref ref-type="bibr" rid="B23">23</xref>] and <italic>sizzled </italic>[<xref ref-type="bibr" rid="B24">24</xref>] respectively. Both of these described genes pass the RP method test and cluster near their undescribed paralog (Figure <xref ref-type="fig" rid="F5">5</xref>). Additionally, two genes for which we described new ventral-posterior expression patterns, <italic>zinc finger protein 503 </italic>and <italic>XARP</italic>, produce proteins that have previous associations with posterior patterning. Zinc finger protein 503 belongs to a family of NocA-like zinc-finger proteins that have been implicated in zebrafish hindbrain patterning [<xref ref-type="bibr" rid="B25">25</xref>-<xref ref-type="bibr" rid="B27">27</xref>], and XARP is the <italic>Xenopus </italic>ortholog of mammalian axin 2, which plays an important role in the transduction of the canonical Wnt signaling cascade [<xref ref-type="bibr" rid="B28">28</xref>], a key posteriorizing signal during gastrulation.</p></sec><sec><title>Relaxing stringency on genes similarly expressed within Nog+Dkk and Dor conditions reveals seven new organizer-related molecules</title><p>In our initial clustering results, most known gastrula patterning genes were similarly induced or repressed in the two conditions with complete organizer function, <bold>Nog+Dkk </bold>and <bold>Dor</bold>. Furthermore, all new expression patterns identified from clusters 3 and 4 were consistent with our expectations, indicating that the clusters were highly reliable (Figure <xref ref-type="fig" rid="F6">6</xref>). Based on these observations, we hypothesized that among genes with similar expression in the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions we should be able to relax our statistical test and identify additional unknown organizer-related genes. To that end, we selected for genes with a RP score of less than 0.0006 in both the <bold>Nog+Dkk </bold>and the <bold>Dor </bold>conditions, when compared to <bold>Ven</bold>, at either stage 10 or stage 11.5. The test was repeated for both up-regulated and down-regulated genes and then all positive genes were merged, creating a new list of 220 probe sets (<xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>). Because the two tests that we require each gene to pass are not independent, there is no clear way to calculate a FDR for this list of genes.</p><p>Since this second list of genes was specifically required to be similarly up- or down-regulated in the <bold>Nog+Dkk-1 </bold>and <bold>Dor </bold>conditions, hierarchical clustering of these genes produces a clustergram with less pattern variety (overview in Figure <xref ref-type="fig" rid="F7">7</xref> and probe information in <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>). Broadly, this cluster retains most of the genes from the first list's clusters 3 and 4, while eliminating most of the genes from clusters 1 and 2 (Figure <xref ref-type="fig" rid="F3">3</xref>). Additionally, this new list captures 119 new probe sets not represented in the first. Many of these are known gastrula patterning genes, including important organizer regulators such as <italic>Xnot </italic>[<xref ref-type="bibr" rid="B29">29</xref>] and <italic>cerberus </italic>[<xref ref-type="bibr" rid="B30">30</xref>], and ventral fate inducers that are repressed by organizer signaling, such as <italic>vent-2 </italic>[<xref ref-type="bibr" rid="B31">31</xref>], <italic>vox-1 </italic>(<italic>Xbr-1</italic>) [<xref ref-type="bibr" rid="B32">32</xref>], <italic>Xvex-1 </italic>[<xref ref-type="bibr" rid="B33">33</xref>]. Hence, our second set of RP criteria does indeed find more organizer-regulated genes, and it also eliminates the clusters of genes in the first list which have no clear relevance to gastrula patterning.</p><fig position="float" id="F7"><label>Figure 7</label><caption><p><bold>Relaxed statistical criteria select for additional genes with organizer-related expression patterns</bold>. The cluster results for the second set of RP criteria, which required correlated expression in the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions, identifies additional unknown genes with organizer-related expression patterns. The hierarchical tree is on the far left, followed by the clustergram, and then the summary of the RP results at 10% FDR. Colors and columns are same as described in the Figure 3A. Black ticks between the cluster tree and the clustergram mark every tenth gene, allowing referencing to <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref> for the gene identities. Genes in this list that lacked described gastrula stage expression patterns were analyzed by whole mount <italic>in situ </italic>hybridization. The top of the cluster contains genes that were repressed in the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions; unknown genes within this group are excluded from organizer tissues (magenta box). The bottom of the cluster contains genes that were activated in the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions; unknown genes within this group are enriched in organizer tissues (orange box). Each tested gene is labeled with its name and the Affymetrix probe set number. Genes marked "no pattern" showed no staining, or a non-specific staining pattern that was similar to sense controls. Genes marked "no gastrula pattern," showed no pattern during gastrula stages, but did show specific patterns at later stages that are not shown here. Each photo is labeled with the developmental stage of the embryo in the bottom left corner, and the orientation in the bottom right corner. veg: vegetal view, dorsal faces up. dor: dorsal view, anterior faces up. † gene name was assigned by protein sequence similarity using Homologene.</p></caption><graphic xlink:href="1471-213X-6-27-7"/></fig><p>From this list of genes we selected fifteen additional clones to test by <italic>in situ </italic>hybridization, again predicting either organizer-enriched or organizer-excluded expression (Figure <xref ref-type="fig" rid="F7">7</xref>). Of the seven genes that showed clear patterns during gastrula stages, once again, the microarray data correctly predicts organizer enrichment or organizer-exclusion for all. Of these seven genes, four have no known developmental function, and the remaining three, HES5, HES6, and dlx2, have described functions during neurulation [<xref ref-type="bibr" rid="B34">34</xref>-<xref ref-type="bibr" rid="B36">36</xref>].</p></sec><sec><title>Noggin and Dkk-1 regulate newly identified genes in patterns that validate the microarray data</title><p>To further assure that we were genuinely identifying genes regulated by organizer signaling, we selected several of the unknown genes identified by the clustering results and analyzed their expression patterns in stage 10.5 embryos overexpressing Noggin and/or Dkk-1. <italic>LRIG2 </italic>(Figure <xref ref-type="fig" rid="F8">8 A–D</xref>), and <italic>ARHGEF3 </italic>(Figure <xref ref-type="fig" rid="F8">8 E–H</xref>) both showed strong ectopic ventral expression in embryos overexpressing Noggin+Dkk-1, but only <italic>LRIG2 </italic>showed ectopic expression with Noggin alone, matching the microarray data. <italic>HES6 </italic>expression (Figure <xref ref-type="fig" rid="F8">8 I–L</xref>) was successfully cleared from ectopic ventral regions by Noggin+Dkk-1 and Dkk-1, but not Noggin alone, again recapitulating the microarray data. <italic>Frzb3 </italic>expression was disrupted by Noggin and Noggin+Dkk-1 overexpression, as the microarray data reported (Figure <xref ref-type="fig" rid="F8">8 M, N</xref>). Moreover, in some cases <italic>Frzb3 </italic>appeared slightly expanded by Dkk-1; the microarray recorded a weak induction by Dkk-1 at stage 10, but not stage 11.5 (Figure <xref ref-type="fig" rid="F8">8 O</xref>). The expression patterns of <italic>Xl.13826.1.A1_at </italic>and <italic>RASL11B </italic>were not clearly altered by ectopic Noggin and/or Dkk-1 (data not shown). However, <italic>Xl.13826.1.A1_at </italic>produces an extremely faint stain, and <italic>RASL11B </italic>exhibits a fair amount of spottiness and variability among wild type embryos, making it difficult to convincingly identify ectopic pattern alterations in either.</p><fig position="float" id="F8"><label>Figure 8</label><caption><p><bold>Noggin and Dkk-1 regulate the expression of newly identified genes</bold>. 4-cell embryos were ventrally injected with the <italic>noggin </italic>and/or <italic>dkk-1 </italic>concentrations described in Figure 1, and tested by <italic>in situ </italic>hybridization for patterns of ectopic induction or repression. In each case the observed <italic>in situ </italic>patterns confirm the microarray patterns (A-D) <italic>LRIG2 </italic>shows ectopic ventral expression in both the Noggin and the Noggin+Dkk-1 overexpressing embryos, but not in the Dkk-1 embryos. (E-H) <italic>ARHGEF3 </italic>shows ectopic ventral expression only in the Noggin+Dkk-1 overexpressing embryos. (I-L) <italic>HES6 </italic>shows ectopic ventral repression in the Noggin+Dkk-1 and the Dkk-1 overexpressing embryos. (M-P) <italic>Frzb3 </italic>expression is ectopically repressed by Noggin or Noggin+Dkk-1 overexpression. (Q-T) <italic>Xl.3529.1.A1 </italic>expression is ectopically repressed by Noggin+Dkk-1 and Dkk-1 overexpression. (U-X) <italic>Gadd45g </italic>is ectopically induced only in the Noggin+Dkk-1 condition. All embryos are between stages 10.5 and 11 and are shown in vegetal view with dorsal side facing up.</p></caption><graphic xlink:href="1471-213X-6-27-8"/></fig><p>From the genes identified by the second clustering effort, <italic>Xl.3529.1.A1 </italic>showed disruption by Noggin+Dkk-1 and Dkk-1, but not Noggin, as expected (Figure <xref ref-type="fig" rid="F8">8 Q–T</xref>), and <italic>gadd45g </italic>only shows induction by the combination of Noggin+Dkk-1 (Figure <xref ref-type="fig" rid="F8">8 U–X</xref>). On the array, <italic>gadd45g </italic>has no inductions that break 10% FDR cutoff, but it was selected by our second set of RP criteria because of correlated inductions in both the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions. This again confirms that the more focused set of RP criteria is successfully identifying genuine organizer-regulated genes that fell below the more stringent 10% FDR cutoff.</p></sec><sec><title>Head and trunk organizing genes show distinct responses to Noggin and Dkk-1</title><p>In addition to allowing us to identify new organizer-regulated genes, the cluster patterns provide us with a broad overview of the genomic response to organizer signaling, allowing us to compare these patterns with current models of organizer function. A close inspection of cluster 3 shows that the genes seem to subdivide in two groups that represent the head and trunk organizers (Figure <xref ref-type="fig" rid="F4">4</xref>). The top two-thirds of the cluster is dense with genes that can induce head identity (<italic>otx2 </italic>[<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B19">19</xref>], <italic>otx5 </italic>[<xref ref-type="bibr" rid="B37">37</xref>], <italic>frzb-1 </italic>[<xref ref-type="bibr" rid="B38">38</xref>]) and includes some genes that are sufficient to induce secondary axes (<italic>gsc </italic>[<xref ref-type="bibr" rid="B20">20</xref>], <italic>chordin </italic>[<xref ref-type="bibr" rid="B39">39</xref>]). The bottom of the cluster contains genes that are strongly induced by Noggin alone, and indeed many of these genes produce proteins that are implicated in processes crucial to trunk formation: three Fox proteins, A4, D3, and D5, which pattern or induce different aspects of trunk tissues (reviewed in [<xref ref-type="bibr" rid="B40">40</xref>]); Zic3, a potent neural inducer [<xref ref-type="bibr" rid="B41">41</xref>]; and Frizzled 7, a key player in convergent extension [<xref ref-type="bibr" rid="B42">42</xref>]. A key difference between these sub-clusters is their differing response to Noggin in the absence of Dkk-1. The head sub-cluster shows some weak induction by Noggin at stage 10, but this induction never crosses the 10% FDR cutoff, and no measurable induction remains for these genes by late gastrulation (Figure <xref ref-type="fig" rid="F4">4</xref>). In contrast, the trunk sub-cluster shows strong early induction by Noggin, and most genes show some induction in late gastrula embryos, although all are below the 10% FDR cutoff.</p><p>To confirm these cluster sub-divisions and better visualize how the genes from the head and trunk sub-clusters differed in their response to organizer signaling, we performed <italic>in situ </italic>hybridization on genes from each cluster in <italic>noggin </italic>and <italic>noggin+dkk-1 </italic>injected embryos during both early and late gastrulation (Figure <xref ref-type="fig" rid="F9">9 A–J</xref>'). Two genes from the head sub-cluster, <italic>Xlim-1 </italic>and <italic>otx2</italic>, show similar weak ectopic induction by both Noggin and Noggin+Dkk-1 at stage 10.5, but by stage 11.5 the induction is much stronger and more widespread in the Noggin+Dkk-1 overexpressing embryos (Figure <xref ref-type="fig" rid="F9">9 A–L</xref>). Indeed, for <italic>Xlim-1</italic>, ectopic expression induced by Noggin alone never expands past the immediate border of the blastopore lip region (Figure <xref ref-type="fig" rid="F9">9 A–F</xref>). For these head organizer genes Dkk-1 seems to play a critical role in maintaining induction throughout gastrulation. We also tested <italic>frzb-1</italic>, a member of the head sub-cluster, which, in contrast to most of the head cluster genes, is not significantly induced by Noggin or Noggin+Dkk-1 on the array, although weak inductions are recorded in each condition. In accord with these measurements, we can see some weak ectopic staining during early gastrulation, however these induction disappears completely by stage 11.5 (Figure <xref ref-type="fig" rid="F9">9 M–R</xref>).</p><fig position="float" id="F9"><label>Figure 9</label><caption><p><bold>The head and trunk sub-cluster genes show distinct responses to organizer signaling</bold>. Three genes were selected from each sub-cluster, and tested by <italic>in situ </italic>hybridization at stage 10.5 (top row) and 11.5 (bottom row), in embryos ventrally injected with <italic>noggin </italic>or <italic>noggin+dkk-1</italic>. Black arrowheads mark ectopic staining. <italic>Xlim-1 </italic>(A-F) and <italic>otx2 </italic>(M-R) expression are similarly induced at stage 10.5 by Noggin and Noggin+Dkk-1, but by stage 11.5 Noggin+Dkk-1 induction is clearly much stronger and more widespread. For <italic>Xlim-1</italic>, ectopic expression induced by Noggin+Dkk-1 is observed migrating away from the blastopore lip region, but never for Noggin alone. (M-R) <italic>Frzb-1 </italic>expression is ectopically induced only by the combination of Noggin+Dkk-1, not Noggin alone, and neither can sustain expression into late gastrulation. For the three trunk genes, <italic>FoxD5b </italic>(<italic>XFD-12'</italic>) (S-X), <italic>FoxA4a </italic>(<italic>pintallavis</italic>) (Y-D'), and <italic>Xsox-2 </italic>(E'-J') ectopic induction is similar in both intensity and spread in Noggin and Noggin+Dkk-1 overexpressing embryos at stage 10.5 and 11.5. A-L and S-J' vegetal view; M-R animal view. Dorsal faces up in all pictures. (K') Otx2 expression was assayed by real-time RT-PCR, in stage 11.5 ventral tissues injected with the same mixtures used to create the microarray samples. Weak, but significant (p = 0.043 by one-sided <italic>t</italic>-test), induction in Dkk-1 overexpressing tissues are seen compared to ventral, supporting the weak Dkk-1 inductions seen on the microarray. Error bars show the standard error calculated from two biological replicates.</p></caption><graphic xlink:href="1471-213X-6-27-9"/></fig><p>From the trunk genes, we tested two trunk patterning genes <italic>foxD5b </italic>(<italic>XFD-12'</italic>) and <italic>foxA4a </italic>(<italic>pintallavis</italic>), as well the general neural inducer <italic>Xsox-2</italic>. <italic>FoxA4a </italic>is a close paralog, and likely pseudoallele, of <italic>FoxA4b </italic>(<italic>XFKH1</italic>) which is among the trunk sub-cluster genes (Figure <xref ref-type="fig" rid="F4">4</xref>). Both genes show the same developmental expression pattern (reviewed in [<xref ref-type="bibr" rid="B40">40</xref>]). <italic>FoxA4a </italic>is captured by our second set of RP criteria, and groups among the trunk cluster genes found by the first gene list (<xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>). For these three trunk genes, the ectopic expression intensity and territory is more similar between the Noggin and Noggin+Dkk-1 overexpressing embryos, than for the head cluster genes (Figure <xref ref-type="fig" rid="F9">9 S–J</xref>'). Together these results indicate that our clustering results are genuinely subdividing genes into groups with different responses to organizer signaling, and these patterns give us a better understanding of how organizer function is generated at the genomic level.</p><p>Another key difference between the head and trunk sub-cluster genes are the weak Dkk-1 inductions observed for several head organizer genes. Conflicting with the array data, Dkk-1 injected embryos did not show ectopic staining for <italic>frzb-1 </italic>and <italic>otx2 </italic>by <italic>in situ </italic>hybridization (data not shown). To test whether these inductions might simply be below the level of detection in our <italic>in situ </italic>analysis, we used real-time RT-PCR to analyze <italic>otx2 </italic>expression in response to ectopic Noggin and/or Dkk-1. Embryos were injected, sorted, and dissected by the same protocol used to generate the array samples, creating two new sets of tissue samples at stage 11.5. Real-time RT-PCR analysis shows that Dkk-1 does indeed induce <italic>otx2 </italic>in these samples, with an average induction of 1.65-fold over ventral (Figure <xref ref-type="fig" rid="F9">9K</xref>'; p = 0.043 by one-sided <italic>t</italic>-test). This is in line with the inductions seen on the array; the four <italic>otx2 </italic>microarray probe sets show an average induction of 1.99-fold over ventral. This supports the reliability of the array data, and shows that Dkk-1 does have some ability to weakly induce organizer genes in the ventral tissues of whole embryos without ectopic anti-BMP signaling.</p></sec></sec><sec><title>Conclusion</title><p>In these studies we used our existing knowledge of organizer function as leverage in the design and analysis of a genomic visualization of gastrula stage signaling. To this end, we ectopically expressed Noggin and Dkk-1 to induce different aspects of organizer activity, and then analyzed the genomic expression consequences of these activities at two different gastrula stages. The resulting data provide us with two valuable sources of information. First, our analysis generates clusters that are strongly enriched for known gastrula patterning genes. Using these enrichments we were able to predict with high confidence the expression patterns of unknown genes within the clusters. Second, our results provide a genome-level view of the transcriptional response patterns to organizer signaling, which helps us to understand the separate roles played by BMP and Wnt inhibition during organizer function, and also defines suites of genes that share similar response patterns.</p><sec><title>A genomic view of organizer signaling refines our understanding of organizer function</title><p>Overall our data reveals that most gastrula-patterning genes respond to organizer signaling according to a few distinct patterns. These patterns divide genes into approximately four groups: a) Trunk patterning genes that are primarily induced by Noggin; b) Head and general organizer genes that require both Noggin and Dkk-1 for strong induction, and in some cases show weak induction by Dkk-1 alone; c) A small number of genes that are primarily repressed by Noggin; d) A wide range of ventral and posterior genes that are primarily repressed by Dkk-1. These broad patterns help to elucidate the genomic networks that underlie organizer signaling and provide a deeper functional understanding of different phenotypes induced by BMP and Wnt inhibition.</p><p>The group of genes primarily induced by Noggin appears to be sufficient to explain the secondary trunk tissues induced by BMP inhibition, a phenomenon that has remained somewhat mysterious in light of previous reports that did not observe sustained induction of organizer genes by BMP inhibition [<xref ref-type="bibr" rid="B1">1</xref>]. This cluster contains a set of Fox genes that can induce and pattern various aspects of trunk identity [<xref ref-type="bibr" rid="B40">40</xref>], as well as a strong neural inducer, Zic3 [<xref ref-type="bibr" rid="B41">41</xref>], and Frizzled 7, a part of the non-canonical Wnt signaling pathway that regulates convergent extension movements [<xref ref-type="bibr" rid="B42">42</xref>]. Noggin strongly induces these genes at stage 10, and maintains induction of these genes into late gastrulation. This late induction is seen weakly on the microarray, but presents clear ectopic staining by <italic>in situ </italic>hybridization (Figure <xref ref-type="fig" rid="F9">9 S–J</xref>'). Indeed, in concert with previous reports, we observe that Noggin induction of head and general organizer molecules is comparatively weak, becoming undetectable on the microarray by stage 11.5, clearly explaining Noggin's inability to induce head structures on its own. In contrast, Wnt inhibition by Dkk-1 can weakly induce several organizer genes at 11.5, including secreted BMP and Wnt inhibitors like <italic>chordin </italic>and <italic>frzb-1</italic>, but has no similar inductive capabilities on the trunk cluster genes we identified. Hence, we have isolated a group of trunk-inducing genes that share a distinctly regulatory pattern that is less sensitive to Wnt signaling than other organizer factors. These genes may be the key to understanding Noggin's ability to induce secondary tissues in the absence of Wnt inhibition.</p><p>In addition to helping us understand BMP inhibition's induction of trunk tissues, the patterns seen here reveal a clear difference in the role Wnt and BMP inhibition play in repressing gene expression. Overexpression of Dkk-1 represses a wide-range of ventral, posterior, and lateral genes, more in fact than the combination treatment of Noggin+Dkk-1. Although the essential role Wnt signaling plays in posteriorizing embryos is well appreciated [<xref ref-type="bibr" rid="B43">43</xref>], the sheer number and variety of genes repressed by Dkk-1 is quite surprising, especially in contrast to Noggin, which can only significantly repress three genes, <italic>sizzled</italic>, <italic>frzb3</italic>, and <italic>Xhox3</italic>. Interestingly, none of these three genes can be repressed by Dkk-1, highlighting the functional separation of BMP and Wnt inhibition. In fact, Dkk-1 is even a better repressor of well-described BMP targets like <italic>msx1 </italic>(Nog -1.06 fold; Dkk -1.58 fold at stage 11.5) [<xref ref-type="bibr" rid="B44">44</xref>]. Together these results indicate that during gastrulation Wnt inhibition is largely responsible for blocking the spread of activities from outside of the dorso-anterior domain, thereby preventing the ventralization and posteriorization of the organizer. Moreover, for many genes, this activity appears to be essentially independent of BMP inhibition. Taken together, these results help to explain the critical role anti-Wnts play in maintaining organizer gene expression as gastrulation progresses.</p><p>Although individually Noggin and Dkk-1 display distinct and separate activities during gastrulation, the combination of Noggin and Dkk-1 generates tissue expressing both suites of organizer genes: the trunk patterning genes induced early by Noggin, and the head organizer genes weakly induced late by Dkk-1. Furthermore this combinatorial induction is more than additive; many of these genes are more strongly induced during both early and late gastrulation, overcoming the temporal restrictions that are evident with either activity alone. This duality of signaling via BMP and Wnt inhibitory actions has been previously indicated using single gene approaches: the competence of ectopic BMP inhibition to induce secondary axes is known to end abruptly at stage 10 [<xref ref-type="bibr" rid="B45">45</xref>,<xref ref-type="bibr" rid="B46">46</xref>], and it can only maintain organizer gene inductions in the presence of anti-Wnt signals [<xref ref-type="bibr" rid="B1">1</xref>]. Our data extends these observations to a genomic level, describing suites of genes that underlie these observed phenotypes. These patterns clearly demonstrate distinct temporal roles for Noggin and Dkk-1 in the establishment of the complete organizer.</p><p>Overall, analysis of the differences in the expression changes induced by Noggin and Dkk-1 supports a model where BMP inhibition plays a largely inductive role during early developmental stages, thereby initiating the suites of genes needed to pattern dorsal tissues. Meanwhile, Wnt inhibition acts later during gastrulation, and is essential for maintenance of organizer gene expression throughout gastrulation, a role which may depend on its ability to block the expression of a host of ventral, posterior, and lateral fate specifying transcription factors. By observing these genomic mechanisms behind known developmental phenomenon we are helping to move toward a network understanding of organizer function.</p></sec><sec><title>Statistical approach vastly enriches for genes expressed in organizer related patterns during gastrulation</title><p>In order to detect genes that were regulated by organizer signaling, we employed the rank products (RP) method [<xref ref-type="bibr" rid="B21">21</xref>], which we believe represents a significant advance over previous statistical testing methods. Common <italic>t</italic>-test based methods test each gene on the array separately, and then must make a multiple test correction for the tens of thousands of genes tested, greatly reducing their ability to detect significant change. In contrast, the RP method uses a simple ranking strategy that looks at all genes relative to one another. In essence, this is a much more realistic model of genomic regulation, since the expression levels of all genes are interconnected. Hence, we benefit from the sheer number of genes on the array; genes that rank themselves near the top of a list with tens of thousands of members in more than one replicate are highly likely to be significantly altered. By using the RP method for our statistical tests we were able to produce acceptable FDR rates with only two replicates, freeing us to complete more conditions.</p><p>Next, we classified the identified genes based on their pattern of response to the different conditions. Our experiment contains a panel of conditions that represent distinct aspects of organizer signaling, helping to focus our hierarchical clustering results on biologically relevant gastrula stage processes. We found that genes known to be involved in organizer function clustered tightly together (Figure <xref ref-type="fig" rid="F3">3</xref>, cluster 3) and genes that induce ventral, posterior, and lateral fates clustered separately (Figure <xref ref-type="fig" rid="F3">3</xref>, cluster 4). Moreover, genes that have no clear relationships to development were largely segregated from the known gastrula patterning genes (clusters 1 and 2, Figure <xref ref-type="fig" rid="F3">3</xref>), highlighting the value of combining a statistical test with hierarchical clustering.</p><p>Overall, this strategy was highly successful at enriching for known organizer genes and predicting the expression patterns of unknown genes. From the first broad set of RP criteria (Figure <xref ref-type="fig" rid="F3">3</xref>), 70% of the cluster 3 genes already have described organizer-related functions, and 60% of the cluster 4 genes are known to play a role in the patterning of lateral, posterior, or ventral fates, functions that oppose the dorsalizing and anteriorizing influences of the organizer (Figure <xref ref-type="fig" rid="F4">4</xref> &<xref ref-type="fig" rid="F5">5</xref>). Using these enrichments of known gastrula patterning genes, we were able to predict the expression patterns of unknown genes within the clusters with remarkable accuracy. In all cases, genes that showed a specific <italic>in situ </italic>pattern during gastrula stages were either organizer-expressed or organizer-excluded in a manner consistent with their cluster. Once we account for these new patterns, we find that 87% of the genes in cluster 3 are functionally important to, or at least have increased expression in, the organizer. Similarly, 73% of the genes in cluster 4 have functions that oppose organizer activity, or expression that is organizer-excluded. These numbers are conservative since we did not test every unknown gene within these clusters, and some of our negative <italic>in situ </italic>results probably resulted from clones that produced poor probes (<italic>in situ </italic>false negatives), rather than from microarray measurement errors (microarray false positives). Moreover, the total lack of <italic>in situ </italic>patterns contrary to our expectations suggests that our clustering predictions were highly reliable. Overall, these studies reveal new restricted gastrula expression patterns for 19 genes, 11 of which lacked any previous associations with early development, providing a list of candidates for future functional studies.</p><p>While our data did provide a powerful enrichment for gastrula patterning genes, it is clear the some key genes were missed in our analysis. For instance BMP-4, the ventral signal that is inhibited by Noggin and other BMP inhibitors, did show lower expression in the <bold>Dor </bold>samples, but the difference was not declared significant by our statistical test. Additionally, several key ventralizing molecules that have previously been shown to be repressed by BMP inhibition, <italic>vent-2</italic>, <italic>vox-1 </italic>(<italic>Xbr-1</italic>), and <italic>Xvex-1</italic>, were not captured by our main clustering list, although they were captured by our second less stringent gene list (<xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>) [<xref ref-type="bibr" rid="B31">31</xref>-<xref ref-type="bibr" rid="B33">33</xref>]. Indeed, for these genes we do observe weak repression by Noggin activity alone (mean = -1.26 fold at stage 11.5). Together, these data indicate that there are real expression differences relevant to organizer signaling that fall below our statistical cutoffs. Additionally, our data identifies many genes that were significantly different between the <bold>Dor </bold>and <bold>Ven </bold>samples, but were not affected by Noggin or Dkk-1 overexpression, and as such were not analyzed in this paper. These include several genes important to early developmental patterning such as <italic>X</italic>w<italic>nt-8</italic>, <italic>Xnr3</italic>, and <italic>siamois </italic>[<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>].</p></sec><sec><title>Adoption of GO annotation from other organisms allows rapid identification of developmental gene enrichments</title><p>In this paper we have used machine assigned annotations to help identify clusters enriched for development functions, and to give objective measures of significance to these enrichments. This has helped us to address a cardinal challenge in the analysis of microarray data: the need to sift through long lists of genes and glean common functional themes, a laborious and subjective process. This challenge has been a primary motivator in the development of systematic gene annotation schemes such as the Gene Ontology (GO) [<xref ref-type="bibr" rid="B47">47</xref>]. The <italic>Xenopus </italic>genome has not been directly annotated using these systems, but methods have been published that allow the adoption of gene annotations from related genes in other organisms, including the method previously described in Vinayagam <italic>et al</italic>. [<xref ref-type="bibr" rid="B48">48</xref>]. For this paper we used a modification of this method to generate biological process GO annotations for the <italic>Xenopus </italic>genome. Not surprisingly, the metrics produced in building this annotation suggest that biological process terms are more difficult than molecular function terms to map between organisms based on protein sequence similarity. Regardless, after employing a strict statistical cutoff to select only terms assigned with high confidence, we found that this annotation provided a useful method to rapidly identify clusters that were enriched for developmental processes, and the p-values proved that we were receiving highly significant enrichments.</p></sec><sec><title>Future directions</title><p>Our results provide new candidate genes for functional studies, and describe the transcriptional response to organizer signaling on a genomic scale. Currently, we are testing the new organizer-regulated genes for specific functions during gastrulation using traditional overexpression techniques. More broadly, we hypothesize that the suites of commonly regulated genes described by our clusters may share similar regulatory mechanisms. To explore this possibility we are investigating methods that combine our microarray patterns with regulatory element prediction algorithms, using the <italic>Xenopus tropicalis </italic>genome to provide flanking sequence for each gene. As these resources grow we hope to begin building a rigorous, network-based model of gastrula patterning. Lastly, we have presented and discussed only a small portion of the data generated by our experiment. The raw data from our arrays has been made available through the NCBI Gene Expression Omnibus (GEO) database [<xref ref-type="bibr" rid="B49">49</xref>] (GSE3368), with the hope of providing utility for researchers beyond the conclusions in this paper.</p></sec></sec><sec sec-type="methods"><title>Methods</title><sec><title>Embryology and overexpression</title><p>Female adult <italic>Xenopus </italic>laevis were ovulated by injection of human chorionic gonadotropin, and eggs were fertilized <italic>in vitro </italic>[<xref ref-type="bibr" rid="B50">50</xref>]. After treatment with 2.5% cysteine (pH 8.0) embryos were reared in 1/3 MR. For microinjections, embryos were placed in 2.5% Ficoll in 1/3 MR and injected at 4-cell stage into the marginal zone of one ventral blastomere. Each embryo received 5 nl of a solution containing a combination of <italic>noggin </italic>(pCS2noggin) and/or <italic>dkk-1 </italic>(pCSdkk-1 [<xref ref-type="bibr" rid="B17">17</xref>]) plasmid and <italic>eGFP </italic>mRNA (concentrations in Figure <xref ref-type="fig" rid="F1">1</xref>), in sterile RNase-free water. Plasmid constructs were linearized at NotI sites and purified by incubation with 0.1μg/μl proteinase K and 0.5% SDS, followed by phenol:chloroform:isoamyl alcohol extraction and sodium acetate/ethanol precipitation. <italic>eGFP </italic>mRNA was prepared using the Sp6 mMessage mMACHINE kit (Ambion) from the CS2P eGFP X/P plasmid. Embryos produced for microarray analysis were sorted at early stage 10 for eGFP fluorescence that was located ventral-marginal, using the emerging blastopore lip to differentiate dorsal and ventral hemispheres. For stage 10 samples, embryos were bisected when the blastopore lip had spread about 50% around the embryo, again using the blastopore lip to distinguish the dorsal and ventral halves. For stage 11.5 embryos there is no reliable morphological indicator of dorsal-ventral polarity, so embryos were cut using eGFP fluorescence as a marker of ventral identity. Embryos were manually devitellinized, followed by bisection with a scalpel blade. Bisected embryo halves were dropped into a microfuge tube resting in liquid nitrogen within a minute of cutting. Ten embryo halves were collected for each sample condition, and stored at -80°C until RNA extraction. Each batch of five conditions was conducted in a single clutch. Sorted embryos that were not bisected were allowed to develop to score secondary axis induction.</p></sec><sec><title>Total RNA isolation from gastrula stage embryos</title><p>Ten half-embryos were digested for 1 hr at 42°C in 1.2 mL of lysis buffer (0.5% SDS, 5 mM EDTA, 50 mM Tris pH 7.5, 50 mM NaCl) with 0.2 mg/mL proteinase K. Each sample was then extracted with an equal volume of acid phenol, and then phenol:choloform:isoamyl alcohol (25:24:1) using Eppendorf phase lock gel (PLG) tubes. Supernatants were precipitated with 120μl 3 M sodium acetate (pH 5.2) and 3 mL ethanol at -20°C for 1 hr. Pellets were spun down at 20800 g for 15 min, washed with 70% ethanol, and resuspended in 300μl RNase-free water. Solutions were then precipitated again by adding 60μl of 7.5 M lithium chloride, 50 mM EDTA, and incubating at -20°C overnight. The next day pellets were spun down, washed, and resuspended in 90μl RNase-free water. Solutions were DNased for 20 min at 37°C in 60μl of 1X DNase buffer, 2.5 mM DTT with 3μl RNase inhibitor and 2.4μl DNase (4.8U). Mixtures were then extracted with phenol:choloroform:isoamyl alcohol and PLG tubes, and supernatants were precipitated with 15μl 3 M sodium acetate and 375μl ethanol at -20°C for 1 hr. After spinning down and washing, pellets were resuspended in 11μl RNase-free water. 1μl was gel analyzed; 1μl was analyzed by a spectrophotometer. For each batch of ten half-embryos this procedure produced 10–30μg of total RNA.</p></sec><sec><title>Microarray hybridization</title><p>RNA was biotin labeled and hybridized to Affymetrix GeneChip<sup>® </sup><italic>Xenopus laevis </italic>Genome Arrays according to the manufacturer's protocols (Affymetrix, Santa Clara, CA). The arrays contain 15,611 probe sets which represent approximately 14,400 transcripts.</p></sec><sec><title>Microarray data analysis</title><p>Raw probe intensities (.cel files) from each oligonucleotide array were processed by the RMA algorithm [<xref ref-type="bibr" rid="B51">51</xref>], implemented by RMAExpress [<xref ref-type="bibr" rid="B52">52</xref>]. This method performs background correction and quantile normalization, followed by calculation of a PM-only log<sub>2 </sub>expression measure for each set of 16 probes. Raw .cel files as well as the processed log<sub>2 </sub>expression data can be freely downloaded from the NCBI GEO database as series GSE3368 [<xref ref-type="bibr" rid="B53">53</xref>]. Control sequences were filtered out leaving 15,491 probe sets. Log<sub>2</sub>-expression values from these genes were then tested for significant alteration between conditions by the rank products (RP) method, implemented by a perl script distributed by the method's author [<xref ref-type="bibr" rid="B21">21</xref>]. Subsets of genes selected by RP test criteria were subjected to hierarchical clustering by dChip [<xref ref-type="bibr" rid="B54">54</xref>,<xref ref-type="bibr" rid="B55">55</xref>], using the following options: replicates were averaged before clustering; all genes were standardized to each other, and stages were standardized separately; cluster calculation used Pearson correlation and centroid linkage.</p></sec><sec><title><italic>Xenopus laevis </italic>GO annotation</title><p>The Gene Ontology (GO) terms were annotated using the strategy proposed by Vinayagam <italic>et al</italic>. [<xref ref-type="bibr" rid="B48">48</xref>]. The method extracts possible GO terms for uncharacterized sequences by running BLAST against GO-mapped protein databases. Subsequently, suitable GO terms were predicted using a combination of multiple Support Vector Machines as well as a voting scheme devised for the purpose. Each prediction is associated with a confidence value to assess its reliability. Previously, this method was optimized only for molecular function GO terms.</p><p>We extended this approach to predict biological process terms and validated the prediction quality with 13 model organisms. Our data shows that biological process terms correlate less tightly with protein sequence similarity than molecular function. This is reflected in our dataset with more negative samples (terms inappropriate to the sequence) than positive ones (appropriate terms). Furthermore, the validation result shows a relatively poor correlation of the precision and accuracy values against the number of votes. However, at higher thresholds (more number of votes), a significant number of biological process GO terms were predicted with good precision (<xref ref-type="supplementary-material" rid="S5">Additional file 5</xref>). Thus, considering only annotations with higher confidence values helps us to avoid misleading terms.</p><p>We applied this new biological process prediction approach to annotate <italic>Xenopus laevis </italic>contig sequences produced by The Institute of Genome Research (TIGR) <italic>Xenopus laevis </italic>Gene Index (XGI) [<xref ref-type="bibr" rid="B56">56</xref>], using the 39,558 contig sequences (excluding singletons) corresponding to XGI Release 8.0 (May 12, 2004). Our annotation system predicted GO terms for 15,649 of these sequences. After selecting only those terms predicted with confidence values of 70% or above, in order to remove uncertain predictions, 10,151 contigs remain with at least one biological process GO term. The annotations were then mapped to Affymetrix probe sets using TIGR's Resourcerer database [<xref ref-type="bibr" rid="B57">57</xref>]. At this point, some of the Affymetrix probe sets still held enormous numbers of GO terms. To ensure that we were only using the best available information for each gene, we restricted our annotation to ten terms for each probe set, selecting only the ten with the highest confidence values in cases where this limit was exceeded (<xref ref-type="supplementary-material" rid="S4">Additional file 4</xref>). The EASE program was used to search gene clusters for enriched terms. Duplicate genes were removed prior to analysis. P-values were calculated using the Bonferroni corrected EASE score, which conservatively corrects for multiple testing [<xref ref-type="bibr" rid="B22">22</xref>].</p></sec><sec><title>Whole mount <italic>in situ </italic>hybridization</title><p>Antisense probes were generated for <italic>otx2 </italic>and <italic>foxA4a </italic>(<italic>pintallavis</italic>) from pXOT30.1 [<xref ref-type="bibr" rid="B58">58</xref>] and Pintallavis/64T [<xref ref-type="bibr" rid="B59">59</xref>] plasmids, respectively. For the remaining genes chosen from the microarray clusters, publicly available clones were selected from the Unigene clusters used to build the probe sets, and then ordered from either NIBB [<xref ref-type="bibr" rid="B60">60</xref>] or IMAGE via Open Biosystems [<xref ref-type="bibr" rid="B61">61</xref>]. The following clones were used: Xl.3374.1.A1_at, XL456p16ex; Xl.13826.1.A1_at, XL051f16; Xl.13537.1.A1_at, XL027n24; Xl.15738.1.S1_at, XL159b23; Xl.19933.1.S1_at, IMAGE:4969205; Xl.12235.1.A1_at, XL218o09; Xl.610.1.S1_at, XL146e16; Xl.13019.1.S1_at, XL146o05; Xl.16672.1.S1_at, XL057a15; Xl.2466.1.A1_at, XL063j24; Xl.34.1.S1_at, XL064h24; Xl.620.1.S1_s_at, IMAGE:4404876; Xl.22607.1.S1_at, XL031h14; Xl.14812.1.S1_at, IMAGE:4959067; Xl.23634.1.S1_s_at, XL512p03ex; Xl.3435.1.A1_at, XL051c23; Xl.11129.1.A1_at, XL061l02; Xl.10684.1.A1_at, XL016j10; Xl.5940.1.A1_at, XL023i06; Xl.8933.1.A1_at, XL048m08; Xl.13925.1.A1_at, XL061i19; Xl.1643.2.S1_a_at, XL086e24; Xl.20670.1.S1_at, IMAGE:6643750; Xl.11148.1.A1_at, XL202e23; Xl.12869.1.A1_at, XL081f23; Xl.15362.1.A1_at, XL039i15; Xl.10415.1.A1_at, XL069c09; Xl.3529.1.A1_at, XL197l22; Xl.15745.1.A1_at, XL094f20; Xl.15270.1.A1_at, XL040d20; Xl.2208.1.A1_at, XL056a18; Xl.22609.1.S1_at, XL102p06; Xl.24155.1.A1_at, XL101m04; Xl.16395.1.S1_at, XL142j09; Xl.16206.1.A1_at, XL013l05; Xl.209.1.S1_at, XL060d01.</p><p>Whole mount <italic>in situ </italic>hybridization was performed as previously described [<xref ref-type="bibr" rid="B62">62</xref>]. In addition to the antisense <italic>in situ </italic>analyses, sense probes were also generated for the following clones, and used to conduct negative control <italic>in situ </italic>analyses: XL456p16ex, XL051f16, IMAGE:4969205, XL146o05, XL057a15, IMAGE:4404876, XL512p03ex, XL023i06, XL048m08, IMAGE:6643750, XL202e23, XL039i15, XL069c09, XL197l22, XL094f20, XL040d20, XL056a18, XL102p06, XL101m04, XL142j09, XL013l05, XL060d01. In all cases, the antisense <italic>in situ </italic>patterns were not present in the sense controls. In fact, all of the sense probes produced a similar pattern of background staining that varied only in intensity, which was characterized by faint general animal staining during blastula and gastrula stages, followed by diffuse general neural and neural crest staining during neurula stages.</p></sec><sec><title>Real-time RT-PCR</title><p>Two additional set of samples were collected exactly as described for the microarray experiments at stage 11.5. Real-time RT-PCR was performed using an iCycler™ machine and iScript™ one-step RT-PCR kit with SYBR<sup>® </sup>green (Bio-Rad, Hercules, CA). <italic>ODC </italic>and <italic>otx2 </italic>primer sequences, and anneal, extension and acquisition temperatures were used as described in Heasman <italic>et al</italic>. 2000 [<xref ref-type="bibr" rid="B63">63</xref>]. PCR was performed with a 30s annealing, 12s extension, and 30s acquisition. For each measurement the <bold>Dor </bold>sample was loaded at 100%, 50%, and 10% dilutions and used to define a standard curve; each condition is reported as a proportion of the <bold>Dor </bold>expression. Two technical replicates were conducted for each measurement and averaged. RT- controls were run and were negative in each case. Within each clutch, <italic>Otx2 </italic>expression was standardized by the <italic>ODC </italic>expression.</p></sec></sec><sec><title>Authors' contributions</title><p>ALH and JCB conceived and designed the <italic>Xenopus </italic>experiments and their analysis, and wrote the paper. ALH performed all the <italic>Xenopus </italic>experiments and analyzed the data. AV and SS provided the biological process GO annotation, and wrote the methods describing their generation.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p><bold>Supplementary scatter plots</bold>. (A-F) Shows scatter plot comparisons of the conditions not shown in Figure <xref ref-type="fig" rid="F2">2</xref>. Log<sub>2</sub>-expression values were averaged between replicates and then plotted against the mean log<sub>2</sub>-expression of the stage-matched <bold>Ven </bold>condition. (A) <bold>Nog </bold>vs <bold>Ven</bold>, stage 10. (B) <bold>Dkk </bold>vs <bold>Ven</bold>, stage 10. (C) <bold>Nog </bold>vs <bold>Ven</bold>, stage 11.5. (D) <bold>Nog+Dkk </bold>vs <bold>Ven</bold>, stage 11.5. (E) <bold>Dkk </bold>vs <bold>Ven</bold>, stage 11.5. (F) <bold>Dor </bold>vs <bold>Ven</bold>, stage 11.5. Probe sets measuring two known organizer genes, <italic>otx2 </italic>(green) and <italic>gsc </italic>(red), are labeled within the plots (<italic>otx2 </italic>probe sets: Xl.1268.1.S1_at, Xl.3004.1.A1_at, Xl.11672.1.A1_at, and XlAffx.1.11.S1_at; <italic>gsc </italic>probe sets: Xl.801.1.A1_at, Xl.801.1.S1_at, and Xl.801.1.S1_s_at). (G-I) Shows scatter comparisons of selected single arrays, further illustrating the relative amounts of clutch variation and experimental variation. (G) <bold>Dkk </bold>vs <bold>Ven</bold>, clutch 3 stage 11.5. (H) <bold>Dor </bold>vs <bold>Ven</bold>, clutch 3 stage 11.5. (I) <bold>Ven </bold>clutch 3 vs <bold>Ven </bold>clutch 4, stage 11.5. Note that the R-square value in (I) is less than (G) and (H), showing greater clutch variation than experimental variation.</p></caption><media xlink:href="1471-213X-6-27-S1.PDF" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p><bold>Genes regulated by ectopic organizer signaling</bold>. This tab delimited table contains information about the all of the genes which passed at least one RP method test above the 10% FDR cutoff in the <bold>Nog</bold>, <bold>Nog+Dkk</bold>, or <bold>Dkk </bold>conditions, when compared to <bold>Ven</bold>, at either stage. Figure <xref ref-type="fig" rid="F3">3</xref> shows the results of hierarchical clustering of this list; genes are listed in the same order as the cluster. Columns: List Number, Affymetrix Probe Set, Unigene ID, Gene Title, Gene Symbol. Names and symbols were assigned by the Affymetrix NetAffx database [<xref ref-type="bibr" rid="B64">64</xref>].</p></caption><media xlink:href="1471-213X-6-27-S2.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional File 3</title><p><bold>Genes that show similar regulation in the full organizer conditions</bold>. This tab delimited table contains information about the genes selected for our second list. Each gene was required to show either up-regulation or down-regulation that attained a RP test score less than 0.0006 in both the <bold>Nog+Dkk </bold>and <bold>Dor </bold>conditions, when compared to <bold>Ven</bold>. Genes are listed in the same order as the hierarchical cluster shown in Figure <xref ref-type="fig" rid="F7">7</xref>. Columns: List Number, Affymetrix Probe Set, Unigene ID, Gene Title, Gene Symbol. Names and symbols were assigned by the Affymetrix NetAffx database [<xref ref-type="bibr" rid="B64">64</xref>].</p></caption><media xlink:href="1471-213X-6-27-S3.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S4"><caption><title>Additional File 4</title><p><bold>GO Biological Process annotation of the <italic>Xenopus laevis </italic>genome</bold>. This tab delimited table contains the machine-generated GO Biological Process annotation used in this paper. Columns: Affymetrix Probe Set, Confidence Value, Biological Process ID, Biological Process Term.</p></caption><media xlink:href="1471-213X-6-27-S4.txt" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S5"><caption><title>Additional File 5</title><p><bold>Precision and accuracy of the GO Biological Process annotation</bold>. The accuracy and precision of the annotation test data are plotted against the number of votes. See Vinayagam <italic>et al</italic>. [<xref ref-type="bibr" rid="B48">48</xref>] for a description of the method used to produce these measures.</p></caption><media xlink:href="1471-213X-6-27-S5.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
The contractile vacuole in Ca<sup>2+</sup>-regulation in <italic>Dictyostelium</italic>: its essential function for cAMP-induced Ca<sup>2+</sup>-influx
|
<sec><title>Background</title><p>cAMP-induced Ca<sup>2+</sup>-influx in <italic>Dictyostelium </italic>is controlled by at least two non-mitochondrial Ca<sup>2+</sup>-stores: acidic stores and the endoplasmic reticulum (ER). The acidic stores may comprise the contractile vacuole network (CV), the endosomal compartment and acidocalcisomes. Here the role of CV in respect to function as a potential Ca<sup>2+</sup>-store was investigated.</p></sec><sec><title>Results</title><p>Dajumin-GFP labeled contractile vacuoles were purified 7-fold by anti-GFP-antibodies in a magnetic field. The purified CV were shown for the first time to accumulate and release Ca<sup>2+</sup>. Release of Ca<sup>2+ </sup>was elicited by arachidonic acid or the calmodulin antagonist W7, the latter due to inhibition of the pump. The characteristics of Ca<sup>2+</sup>-transport and Ca<sup>2+</sup>-release of CV were compared to similarly purified vesicles of the ER labeled by calnexin-GFP. Since the CV proved to be a highly efficient Ca<sup>2+</sup>-compartment we wanted to know whether or not it takes part in cAMP-induced Ca<sup>2+</sup>-influx. We made use of the LvsA<sup>-</sup>-mutant expected to display reduced Ca<sup>2+</sup>-transport due to loss of calmodulin. We found a severe reduction of cAMP-induced Ca<sup>2+</sup>-influx into whole cells.</p></sec><sec><title>Conclusion</title><p>The contractile vacuoles in <italic>Dictyostelium </italic>represent a highly efficient acidic Ca<sup>2+</sup>-store that is required for cAMP-induced Ca<sup>2+</sup>-influx.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Malchow</surname><given-names>Dieter</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Lusche</surname><given-names>Daniel F</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Schlatterer</surname><given-names>Christina</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>De Lozanne</surname><given-names>Arturo</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Müller-Taubenberger</surname><given-names>Annette</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib>
|
BMC Developmental Biology
|
<sec><title>Background</title><p>The contractile vacuole (CV) network of <italic>Dictyostelium </italic>consists of tubes and bladders. It transiently fuses with the plasma membrane to expel water and ions and thereby serves as an efficient osmoregulatory organelle [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The CV-system is also assumed to be involved in Ca<sup>2+</sup>-transport since it contains a PMCA-type Ca<sup>2+</sup>-ATPase (PAT1), calmodulin [<xref ref-type="bibr" rid="B3">3</xref>] and a vacuolar proton pump that establishes a proton gradient for Ca<sup>2+</sup>-transport [<xref ref-type="bibr" rid="B4">4</xref>]. PAT1 is localized to the CV and the plasma membrane and is upregulated under conditions of Ca<sup>2+</sup>-stress [<xref ref-type="bibr" rid="B5">5</xref>]. Downregulation of PAT1 by antisense RNA reduced vesicular Ca<sup>2+</sup>-uptake.</p><p>We are interested in the characterization of Ca<sup>2+</sup>-stores that are involved in cAMP-induced Ca<sup>2+</sup>-influx. Previously, it has been shown that acidic Ca<sup>2+</sup>-stores and an IP<sub>3</sub>-sensitive store participate in this regulation [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B11">11</xref>]. Acidic means that the stores are equipped with a V-type H<sup>+</sup>-ATPase. Acidic vesicular Ca<sup>2+</sup>-stores in <italic>Dictyostelium </italic>comprise the CV-system, endosomes and acidocalcisomes [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. In the present study we focus on the contribution of the CV-system to intracellular Ca<sup>2+ </sup>regulation.</p><p>It has been shown previously that GFP-tagged dajumin labels the entire CV whereas the endosomal compartments are devoid of the label [<xref ref-type="bibr" rid="B14">14</xref>]. By contrast, drainin, a peripheral membrane protein involved in discharge of the bladder, is found only on the bladder [<xref ref-type="bibr" rid="B15">15</xref>]. We used dajumin-GFP expressing cells to obtain a fraction enriched in CV membranes and used this fraction to measure Ca<sup>2+</sup>-transport. Ca<sup>2+</sup>-transport activity increased with enhanced purity of the CV. We also employed a LvsA minus strain which lacks the gene for the protein large volume sphereA (<italic>lvsA</italic>). Besides its involvement in cytokinesis [<xref ref-type="bibr" rid="B16">16</xref>] it is known that the LvsA-protein is localized to the CV. This association with the CV occurs only during the discharge phase of the vacuole. In the lvsA mutant calmodulin was lost from the CV-membranes and the CV became disorganized, unable to discharge its contents [<xref ref-type="bibr" rid="B17">17</xref>]. We found that vesicular Ca<sup>2+</sup>-transport in the lvsA-mutant was diminished and that cAMP-induced Ca<sup>2+</sup>-influx was drastically reduced, indicating that functional CV are absolutely required for the cAMP-dependent Ca<sup>2+</sup>-influx.</p></sec><sec><title>Results</title><sec><title>Distribution of CV in vesicular fractions</title><p>We used differentiated cells 4 to 5 hrs after starvation for preparation of Ca<sup>2+</sup>-transporting vesicles because cAMP-induced Ca<sup>2+</sup>-influx is present at that time and endosomal content is low (see below). Cells labeled with dajumin-GFP as a marker for the CV-system or with calnexin-GFP cells as a marker for the endoplasmic reticulum (ER) are shown in Figure <xref ref-type="fig" rid="F1">1</xref>. Dajumin-GFP allows to visualize the dynamics of the bladder by formation of irregular ventricles and ducts (A). The ER is prominently labeled in the perinuclear region and close to the plasma membrane (B). The cells were lysed by passage through nuclepore filters. Vesicular fractions were separated by differential centrifugation and assayed for Ca<sup>2+</sup>-transporting activity. The dajumin-GFP label was detected in vesicular fractions with the majority being present in the fast sedimenting fraction P<sub>0 </sub>(Table <xref ref-type="table" rid="T1">1</xref>). By contrast, most of the ER occurred in P<sub>1</sub>, whereas the lightest fraction P<sub>2 </sub>contained only a quarter of both organelles (Table <xref ref-type="table" rid="T1">1</xref>). Plasma membranes, as shown previously, sedimented in P<sub>1 </sub>[<xref ref-type="bibr" rid="B18">18</xref>]. Ca<sup>2+</sup>-transport activity was strikingly concentrated in P<sub>0</sub>. The presence of endosomes was measured with RITC-dextran. In two independent experiments 38 ± 6% of the label was associated with P<sub>0</sub>, 62 ± 6% with P<sub>1 </sub>and none was present in P<sub>2</sub>. However, the amount of endosomes of 4 hour starved cells was low and barely detectable. This result is expected since the cells develop in the absence of nutrients. If endosomes accumulate Ca<sup>2+ </sup>their contribution to Ca<sup>2+</sup>-transport was therefore considered to be insignificant under the present experimental conditions.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Characterization of crude vesicular fractions. Vesicles were obtained by differential centrifugation of lysed cells as described in Methods. The amount (percent) of CV (± S.D.) in each fraction was determined according to the dajumin-GFP label present, while that of the ER by the presence of calnexin-GFP label. Total Ca<sup>2+</sup>-transport activity obtained from 4 × 10<sup>8 </sup>cells is shown for each fraction. Data are means ± SD. The number of independent experiments is given in brackets.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Fraction</td><td align="center">Sedimentation</td><td align="center">CV</td><td align="center">ER</td><td align="center" colspan="2">Ca<sup>2+</sup>-uptake</td></tr><tr><td></td><td></td><td></td><td></td><td colspan="2"><hr></hr></td></tr><tr><td></td><td align="center">(g)</td><td align="center">%</td><td align="center">%</td><td align="center">nmol</td><td align="center">%</td></tr></thead><tbody><tr><td align="center">P<sub>0</sub></td><td align="center">3.800</td><td align="center">42.4 ± 4.0 (4)</td><td align="center">33.5 ± 3.7 (3)</td><td align="center">205 ± 42 (3)</td><td align="center">67</td></tr><tr><td align="center">P<sub>1</sub></td><td align="center">12.000</td><td align="center">33.6 ± 7.2 (4)</td><td align="center">43.0 ± 2.8 (3)</td><td align="center">76 ± 24 (3)</td><td align="center">25</td></tr><tr><td align="center">P<sub>2</sub></td><td align="center">40.000</td><td align="center">24.0 ± 10.6 (4)</td><td align="center">23.5 ± 1.3 (3)</td><td align="center">25 ± 6 (3)</td><td align="center">8</td></tr></tbody></table></table-wrap><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold><italic>Dictyostelium </italic>wild-type Ax2 cells expressing dajumin-GFP or calnexin-GFP</bold>. (A): The contractile vacuole system of a <italic>Dictyostelium </italic>cell is visualized by expression of dajumin-GFP using live cell confocal microscopy (upper panels). Dynamics of the bladder is indicated by formation of irregular ventricles and ducts. Lower panels show the corresponding bright field images. (B): A <italic>Dictyostelium </italic>cell expressing calnexin-GFP to visualize the endoplasmic reticulum is shown in the upper panels. Note that the perinuclear region is intensively labeled as observed previously for calnexin-GFP and antibody-stained preparations, whereas a vacuole in the left part of the cell is devoid of any label. Time of depicted frames is indicated in seconds. Bar corresponds to 10 μm.</p></caption><graphic xlink:href="1471-213X-6-31-1"/></fig></sec><sec><title>Purification of CV with magnetic antibody-beads</title><p>Since P<sub>0 </sub>was the richest source of CV we incubated P<sub>0 </sub>isolated from dajumin-GFP labeled cells with anti-GFP-magnetic beads and loaded the mixture onto a column. During loading and washing the beads were retained on the column by applying a magnetic field. Elution was performed in its absence. As shown in Table <xref ref-type="table" rid="T2">2</xref> almost all of the dajumin-label was present in the eluate (P<sub>0</sub>E) as well as most of Ca<sup>2+</sup>-uptake. The flow through (P<sub>0</sub>F) still contained some Ca<sup>2+</sup>-transport activity the percentage of which, however, exceeded only slightly the remaining dajumin-GFP-label.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Purification of CV and ER fractions by antibody. Dajumin-GFP labeled CV present in P<sub>0 </sub>were bound to anti-GFP-microbeads and separated from the unbound vesicles in a magnetic field as detailed in Methods. P<sub>0</sub>F represents the flow through, P<sub>0</sub>E is the eluate containing CV. Calnexin-GFP labeled ER present in P<sub>1</sub>E was purified in the same way. P<sub>1</sub>F: flow through, P<sub>1</sub>E: eluate containing ER. As described in Methods, the measured values of Ca<sup>2+</sup>-uptake were corrected for the same initial [Ca<sup>2+</sup>]<sub>ev</sub>. Data are means ± SD. The number of independent experiments is given in brackets.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Fraction</td><td align="center">GFP-label %</td><td align="center">Ca<sup>2+</sup>-uptake %</td><td align="center">Specific activity of Ca<sup>2+</sup>-ATPase (%)</td></tr></thead><tbody><tr><td align="left">P<sub>0</sub>F (Dajumin)</td><td align="center">12.4 ± 3.9 (3)</td><td align="center">15.6 ± 9.7 (3)</td><td align="center">18 ± 5.4 (3)</td></tr><tr><td align="left">P<sub>0</sub>E (Dajumin)</td><td align="center">87.6 ± 3.9 (3)</td><td align="center">84.4 ± 9.7 (3)</td><td align="center">109 ± 9.3 (3)</td></tr><tr><td colspan="4"><hr></hr></td></tr><tr><td align="left">P<sub>1</sub>F (Calnexin)</td><td align="center">13.8 ± 4.2 (4)</td><td align="center">13.4 ± 8.3 (4)</td><td align="center">8.1 ± 6.3 (4)</td></tr><tr><td align="left">P<sub>1</sub>E (Calnexin)</td><td align="center">86.2 ± 4.2 (4)</td><td align="center">86.6 ± 8.3 (4)</td><td align="center">87 ± 66 (4)</td></tr></tbody></table></table-wrap><p>The selectivity of the GFP-antibody beads for GFP was tested by assaying P<sub>0 </sub>from unlabeled Ax2-cells, the parent strain of dajumin-GFP expressing cells, for binding to the anti-GFP-beads. In three independent experiments 54 ± 31 μg (n = 6) protein was associated with P<sub>0</sub>E in the controls whereas 192 ± 60 μg (n = 6) protein was present in P<sub>0</sub>E from dajumin-GFP labeled cells. This result shows that the GFP-antibody specifically targets the GFP-tagged protein and that 28% of unspecific binding does occur.</p><p>In order to avoid contamination of P<sub>0 </sub>and subsequent fractions with plasma membranes due to possible association of plasma membranes and CV by cortical actin filaments we preincubated the cells (all experiments of Table <xref ref-type="table" rid="T2">2</xref>) with latrunculin B. Latrunculin B, a unique marine toxin, inhibits actin polymerization and disrupts microfilament organization [<xref ref-type="bibr" rid="B19">19</xref>]. In the presence of latrunculin B the cells rounded up and were completely immotile before they were pressed through the filters. The characteristics of Ca<sup>2+</sup>-transport, its inhibition by drugs as well as Ca<sup>2+</sup>-release in P<sub>0</sub>E were in the same range as in preparations performed without latrunculin B treatment. Latrunculin B treatment reduced the protein content of P<sub>0</sub>E by a factor of 1.26 but there was no statistically significant difference in the specific activity of the Ca<sup>2+</sup>-ATPase.</p><p>A specific marker for the CV as compared to the plasma membrane is the V-type H<sup>+</sup>-ATPase. Antibodies to the 37/42 kDa subunit revealed that the proton pump was markedly enriched in P<sub>0</sub>E (Fig. <xref ref-type="fig" rid="F2">2</xref>). A mitochondrial contribution to Ca<sup>2+</sup>-fluxes was prevented by inclusion of mitochondrial inhibitors in all vesicular Ca<sup>2+</sup>-transport measurements.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Enrichment of V-type H<sup>+</sup>-ATPase in P<sub>0</sub>E</bold>. Western blot analysis with antibodies to V-type H<sup>+</sup>-ATPase was carried out as described in Methods using fractions of cells expressing dajumin-GFP. 7 μg of protein per lane of each fraction were applied.</p></caption><graphic xlink:href="1471-213X-6-31-2"/></fig></sec><sec><title>Purification of ER with magnetic antibody-beads</title><p>Since P<sub>1 </sub>was the richest source of ER (Table <xref ref-type="table" rid="T1">1</xref>) we incubated P<sub>1 </sub>isolated from calnexin-GFP labeled cells with anti-GFP-magnetic beads and separated ER-vesicles from other vesicles by a magnetic field. As shown in Table <xref ref-type="table" rid="T2">2</xref> almost all of the calnexin-GFP-label was found in the eluate (P<sub>1</sub>E) as well as most of Ca<sup>2+</sup>-uptake activity. The flow through (P<sub>1</sub>F) also contained some Ca<sup>2+</sup>-transport activity.</p><p>The results of Table <xref ref-type="table" rid="T2">2</xref> thus show that by using the anti-GFP-antibody beads a 6-7-fold enrichment of either CV or ER can be obtained. Furthermore, both fractions enriched either in CV or ER vesicles, displayed a similar enhancement of Ca<sup>2+</sup>-transport. This result demonstrates that vesicles of the CV of <italic>Dictyostelium </italic>are functional in Ca<sup>2+</sup>-homeostasis.</p></sec><sec><title>Properties of CV and ER fractions with respect to Ca<sup>2+</sup>-regulation</title><p>Since both purifications had given a strong enrichment of the GFP-label as well as in Ca<sup>2+</sup>-transport in the eluate (84–87%) and since in control experiments 28% unlabeled protein was associated with the eluate, we decided to test for the properties of CV and ER with respect to Ca<sup>2+</sup>-transport and Ca<sup>2+</sup>-release. As shown in Figure <xref ref-type="fig" rid="F3">3</xref> P<sub>0</sub>E, the eluate of dajumin-GFP-labeled cells containing CV, transported Ca<sup>2+ </sup>efficiently. Note, that the slope is fairly steep and uptake of Ca<sup>2+ </sup>is settled in about 8 min. In three independent experiments we found that starting at an extravesicular Ca<sup>2+</sup>-concentration ([Ca]<sup>2+</sup><sub>ev</sub>) of 0.53 ± 0.02 μM uptake ceased to 0.10 ± 0.05 μM [Ca]<sup>2+</sup><sub>ev </sub>in the presence of 32 ± 12 μg protein. In agreement with previous results obtained for acidic Ca<sup>2+</sup>-stores the calmodulin antagonist W7 as well as arachidonic acid (AA) caused Ca<sup>2+</sup>-release in P<sub>0</sub>E. The response to W7 (30–100 μM) amounted to 21.4 ± 10.2 nmol Ca<sup>2+</sup>/mg protein (n = 4) and to AA (10–20 μM) 23.9 ± 7.2 nmol Ca<sup>2+</sup>/mg protein (n = 4).</p><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Ca<sup>2+</sup>-uptake and Ca<sup>2+</sup>-release in P<sub>0</sub>E</bold>. P<sub>0</sub>E was obtainedby GFP-antibody purification. Ca<sup>2+</sup>-uptake was induced by 0.4 mM ATP and measured with Fluo-3 as described in Methods. A Ca<sup>2+</sup>-calibration pulse was followed by further Ca<sup>2+</sup>-uptake. W-7 and AA induced Ca<sup>2+</sup>-release, 4-BrA23187 elicited further Ca<sup>2+</sup>-release. One out of three independent experiments is shown.</p></caption><graphic xlink:href="1471-213X-6-31-3"/></fig><p>In Figure <xref ref-type="fig" rid="F4">4</xref> Ca<sup>2+</sup>-uptake by P<sub>1</sub>E, the eluate of calnexin-GFP-labeled cells containing ER, is shown. In contrast to P<sub>0</sub>E P<sub>1</sub>E was less responsive to AA. In three out of four experiments no Ca<sup>2+</sup>-release was measured. The one that responded to AA had a lower purification factor with respect to GFP fluorescence compared with the other three, 3.8 versus 7.8 ± 0.96 (n = 3). IP<sub>3</sub>, most of the times, was inefficient to elicit Ca<sup>2+</sup>-release. In general, Ca<sup>2+</sup>-uptake was slower in P<sub>1</sub>E as compared to P<sub>0</sub>E starting at 0.56 ± 0.15 μM [Ca]<sup>2+</sup><sub>ev </sub>and ceased to extravesicular Ca<sup>2+</sup>-concentrations of about 0.36 ± 0.08 μM in the presence of 27 ± 12 μg protein (n = 4), indicating that the CV is more relevant in maintaining a basal cytosolic Ca<sup>2+</sup>-concentration of 50 to 60 nM [<xref ref-type="bibr" rid="B10">10</xref>] than the ER.</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Ca<sup>2+</sup>-uptake and Ca<sup>2+</sup>-release in P<sub>1</sub>E</bold>. P<sub>1</sub>E was obtainedby GFP-antibody-purification. Ca<sup>2+</sup>-measurements were performed as described in the legend to Fig. 1. One out of three independent experiments is shown.</p></caption><graphic xlink:href="1471-213X-6-31-4"/></fig><p>In addition, we tested for inhibition of Ca<sup>2+</sup>-uptake. Table <xref ref-type="table" rid="T3">3</xref> shows that concanamycin A (CMA), a powerful inhibitor of V-type H<sup>+</sup>-ATPases [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>] inhibited P<sub>0</sub>E as expected for a compartment that uses a proton gradient for Ca<sup>2+</sup>-transport and a Ca<sup>2+</sup>-transport ATPase PAT1. Under our experimental conditions BHQ, an inhibitor of Ca<sup>2+</sup>-transport ATPases, inhibited Ca<sup>2+</sup>-uptake only weakly despite its reported property to inhibit cAMP-induced Ca<sup>2+</sup>-influx in <italic>Dictyostelium </italic>[<xref ref-type="bibr" rid="B22">22</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. BHQ was more efficient in inhibiting Ca<sup>2+</sup>-uptake into the ER-fraction. The latter was also sensitive to thapsigargin (Tg) a blocker of SERCA-type Ca<sup>2+</sup>-ATPases. In both fractions, P<sub>0</sub>E and P<sub>1</sub>E, Ca<sup>2+</sup>-uptake was inhibited by low concentrations (30–60 μM) of Sr<sup>2+ </sup>by about 50%.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Inhibition of Ca<sup>2+</sup>-uptake activity. The purified CV-fraction P<sub>0</sub>E and the purified ER-fraction P<sub>1</sub>E were tested for sensitivity of Ca<sup>2+</sup>-uptake to Tg (22 μM), BHQ (200 μM), CMA (40 μM) and Sr<sup>2+ </sup>(30–60 μM) as indicated. nd: not determined. Data are means ± SD. The number of independent experiments is given in brackets.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Fraction</td><td align="center">Tg %</td><td align="center">BHQ %</td><td align="center">CMA %</td><td align="center">Sr<sup>2+ </sup>%</td></tr></thead><tbody><tr><td align="center">P<sub>0</sub>E (CV)</td><td align="center">nd</td><td align="center">25 ± 13 (3)</td><td align="center">57 ± 6 (4)</td><td align="center">52 ± 13 (3)</td></tr><tr><td align="center">P<sub>1</sub>E (ER)</td><td align="center">45 ± 23 (4)</td><td align="center">72 ± 4 (3)</td><td align="center">nd</td><td align="center">43 ± 6 (4)</td></tr></tbody></table></table-wrap></sec><sec><title>Is the CV involved in cAMP-induced Ca<sup>2+</sup>-influx?</title><p>Given our results that the CV is a Ca<sup>2+</sup>-store that transports Ca<sup>2+ </sup>most efficiently at low Ca<sup>2+</sup>-concentrations we wanted to know whether the store is also involved in receptor-mediated Ca<sup>2+</sup>-flux. Although it was previously shown that NBD-Cl and CMA inhibited the cAMP mediated Ca<sup>2+</sup>-influx [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B22">22</xref>] these results only proved that acidic stores as a whole were involved. We made use of the LvsA<sup>-</sup>-mutant whose CV-function is abolished with respect to fusion of CV with the plasma membrane and is disturbed by a loss of calmodulin. Previously, we have shown that the calmodulin antagonist W7 inhibits Ca<sup>2+</sup>-transport of acidic Ca<sup>2+</sup>-stores [<xref ref-type="bibr" rid="B24">24</xref>]. A localized loss of calmodulin confined to the CV would have the same consequences. Therefore, it was interesting to know whether or not Ca<sup>2+</sup>-influx was also disturbed in the mutant. As shown in Table <xref ref-type="table" rid="T4">4</xref> there was no receptor-mediated Ca<sup>2+</sup>-influx in the LvsA minus strain at physiological cAMP-concentrations (0.1–10 μM) in contrast to the parent strain DH1. Only higher concentrations of cAMP yielded Ca<sup>2+</sup>-influx that was reduced as compared to DH1 (Table <xref ref-type="table" rid="T4">4</xref>).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Stimulus-induced Ca<sup>2+</sup>-influx in LvsA<sup>-</sup>and DH1. Ca<sup>2+</sup>-influx was determined with a Ca<sup>2+</sup>-sensitive electrode before and after addition of cAMP or AA at an extracellular Ca<sup>2+</sup>-concentration of 2–3 μM. Data are means ± SD. The number of independent experiments is given in brackets.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center" colspan="2">Ca<sup>2+</sup>-influx pmol/10<sup>7 </sup>cells</td></tr><tr><td align="center">Stimulus (μM)</td><td align="center">LvsA<sup>-</sup></td><td align="center">DH1</td></tr></thead><tbody><tr><td align="center">cAMP 0.1</td><td align="center">< 0.1 (7)</td><td align="center">nd</td></tr><tr><td align="center">1</td><td align="center">< 0.1 (7)</td><td align="center">100 ± 59 (3)</td></tr><tr><td align="center">10</td><td align="center">1 ± 5 (9)</td><td align="center">203 ± 107 (3)</td></tr><tr><td align="center">20 – 50</td><td align="center">50 ± 1 (2)</td><td align="center">nd</td></tr><tr><td align="center">AA 10</td><td align="center">< 0.1 (7)</td><td align="center">70 ± 13 (3)</td></tr></tbody></table></table-wrap><p>Besides cAMP, arachidonic acid (AA) was shown to elicit Ca<sup>2+</sup>-influx. AA acts by eliciting Ca<sup>2+</sup>-release from Ca<sup>2+</sup>-stores which is thought to induce a capacitative Ca<sup>2+</sup>-influx [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B23">23</xref>]. In the LvsA minus mutant AA did not evoke Ca<sup>2+</sup>-influx in contrast to the parent strain (Table <xref ref-type="table" rid="T4">4</xref>). Instead Ca<sup>2+</sup>-efflux occurred. In seven independent experiments 6–10 μM AA induced an efflux of Ca<sup>2+ </sup>of 276- ± 153 pmol/10<sup>7</sup>cells.</p><p>Analysis of vesicular Ca<sup>2+</sup>-transport in the mutant revealed that transport activity was present in all three fractions of P<sub>0</sub>, P<sub>1 </sub>and P<sub>2 </sub>albeit at a somewhat lower extent than in Ax2 (Table <xref ref-type="table" rid="T5">5</xref>). Therefore, the small amount of cAMP-induced Ca<sup>2+</sup>-influx observed in the LvsA minus strain cannot be due to the lack of vesicular Ca<sup>2+</sup>-transport in the mutant.</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Vesicular Ca<sup>2+</sup>-transport of LvsA<sup>-</sup>. The specific Ca<sup>2+</sup>-transport activity of vesicular fractions of LvsA<sup>- </sup>and Ax2 is shown. Data are means ± SD. The number of independent experiments is given in brackets.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center" colspan="2">Ca<sup>2+</sup>-uptake nmol/mg</td></tr><tr><td></td><td align="center">LvsA<sup>-</sup></td><td align="center">Ax2</td></tr></thead><tbody><tr><td align="center">P<sub>0</sub></td><td align="center">42 ± 19 (5)</td><td align="center">68 ± 9.8 (3)</td></tr><tr><td align="center">P<sub>1</sub></td><td align="center">29 ± 17 (3)</td><td align="center">36 ± 7.9 (3)</td></tr><tr><td align="center">P<sub>2</sub></td><td align="center">15 ± 3 (3)</td><td align="center">23 ± 2 (3)</td></tr></tbody></table></table-wrap></sec></sec><sec><title>Discussion</title><p>In this study we have shown that one of the acidic compartments of <italic>Dictyostelium</italic>, the CV-system, is required for cAMP-induced Ca<sup>2+</sup>-influx in intact cells and that isolated CV-vesicles mediate an essential part of Ca<sup>2+</sup>-transport as well as Ca<sup>2+</sup>-release in response to AA. Previously, it was only known that acidic Ca<sup>2+</sup>-stores were involved. The use of dajumin-GFP-labeled CV allowed a 7-fold enrichment of CV by a specific method utilizing antibodies directed against the GFP-tag. Ca<sup>2+</sup>-transport was enhanced 5- to 6-fold. This result is due to Ca<sup>2+</sup>-transport activity of another Ca<sup>2+</sup>-store that was retained in the flow through (P<sub>0</sub>F). Enriched CV displayed a potent and efficient Ca<sup>2+</sup>-transport that resulted in a low basal extravesicular Ca<sup>2+ </sup>concentration. Moreover, the Ca<sup>2+ </sup>taken up could be released by the calmodulin antagonist W-7 and by AA. Both compounds have been shown to be potent regulators of the acidic Ca<sup>2+</sup>-stores [<xref ref-type="bibr" rid="B23">23</xref>,<xref ref-type="bibr" rid="B24">24</xref>]. In addition, Ca<sup>2+</sup>-uptake was sensitive to CMA, an inhibitor of V-type H<sup>+</sup>-ATPases. Moreover, the V-type proton pump was concentrated in P<sub>0</sub>E. A similar enrichment was also obtained for the ER using calnexin-GFP-labeled cells. These Ca<sup>2+</sup>-stores differed from those of the CV by a greater sensitivity to BHQ, a lower responsiveness to AA and a higher basal extravesicular Ca<sup>2+</sup>-concentration following Ca<sup>2+</sup>-uptake.</p><p>In order to examine whether the Ca<sup>2+</sup>-transport occurring in CV is also involved in cAMP-mediated Ca<sup>2+</sup>-influx we made use of a mutant, LvsA minus. In this mutant the CV becomes disorganized during the discharge phase of the vacuole and calmodulin dissociates from the CV-membranes. The CV can swell but seems unable to discharge. The LvsA-protein that belongs to the family of BEACH-proteins transiently binds to the CV. We found that the mutant was strongly impaired in receptor-mediated Ca<sup>2+</sup>-influx when physiological cAMP-concentrations of 0.1–1 μM and even higher concentrations of 10 μM cAMP were applied. A significant Ca<sup>2+</sup>-influx was apparent only at elevated cAMP concentrations of 20 to 50 μM. Even so, influx was inhibited by about 75% as compared to the parent strain. Other compounds that induce Ca<sup>2+</sup>-influx in <italic>Dictyostelium </italic>are long chain fatty acids like AA. They act independently of the cAMP-receptor and of cell differentiation [<xref ref-type="bibr" rid="B7">7</xref>]. Similarly to cAMP, AA did not cause Ca<sup>2+</sup>-influx in the LvsA-minus-strain, instead Ca<sup>2+</sup>-efflux occurred. These results demonstrate that in order to activate capacitative Ca<sup>2+</sup>-influx an intact CV-system is absolutely necessary.</p><sec><title>What is the role of the LvsA-protein in Ca<sup>2+</sup>-influx?</title><p>The function of the LvsA-protein in <italic>Dictyostelium </italic>is not yet known. BEACH-proteins belong to several classes and only a few proteins have been characterized. The FAN-protein mediates TNF-activation via the MAP-kinase pathway [<xref ref-type="bibr" rid="B25">25</xref>]. Neurobeachin conveys anchoring of the regulatory subunit of protein kinase A (PKA) [<xref ref-type="bibr" rid="B26">26</xref>]. In both cases, the BEACH-protein participates in signal transduction that could lead to altered Ca<sup>2+</sup>-regulation in <italic>Dictyostelium</italic>. Elevated cAMP-levels in <italic>Dictyostelium </italic>were shown to affect both, basal [Ca<sup>2+</sup>]<sub>i </sub>and stimulus-induced [Ca<sup>2+</sup>]<sub>i</sub>-transients [<xref ref-type="bibr" rid="B27">27</xref>].</p><p>Apart from this, it was shown that the loss of the LvsA-protein was accompanied by a loss of calmodulin from the CV-membranes. Since we have shown that the calmodulin antagonist W7 inhibits Ca<sup>2+</sup>-transport of acidic Ca<sup>2+</sup>-stores [<xref ref-type="bibr" rid="B24">24</xref>] we conclude that Ca<sup>2+</sup>-uptake by the CV is inhibited in the absence of calmodulin. This could be an explanation of the severe reduction of cAMP-mediated Ca<sup>2+</sup>-influx measured in the LvsA<sup>-</sup>-mutant strain.</p></sec></sec><sec><title>Conclusion</title><p>Among the acidic Ca<sup>2+</sup>-stores the contractile vacuole represents an efficient Ca<sup>2+ </sup>transport organelle that sequesters Ca<sup>2+ </sup>to low extravesicular Ca<sup>2+ </sup>concentrations and releases Ca<sup>2+ </sup>in response to arachidonic acid. Furthermore, it is essential for the occurrence of receptor-mediated Ca<sup>2+ </sup>influx to physiological concentrations of cAMP.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Chemicals</title><p>N-(6-aminohexyl)5-chloro-1-naphthalene sulfonamide (W-7) 2,5-di(tert-butyl)-1,4-hydroquinone (BHQ) and concanamycin A (CMA) were from Fluka (Buchs, Switzerland). Fluo-3 was obtained from MobiTec (Göttingen, Germany), inositol 1,4,5-triphosphate and thapsigargin from Alexis (Grünberg, Germany) and arachidonic acid from Sigma (München, Germany). The μMACS GFP-tagged protein isolation kit was purchased from Miltenyi Biotec (Bergisch Gladbach, Germany).</p></sec><sec><title>Cells and culture</title><p>The strains were grown in liquid culture as discribed by Sonnemann et al. [<xref ref-type="bibr" rid="B28">28</xref>]. The dajumin-GFP [<xref ref-type="bibr" rid="B14">14</xref>] and calnexin-GFP [<xref ref-type="bibr" rid="B29">29</xref>] strains were raised in the presence of 50 μg/ml G418. The LvsA-minus-strain [<xref ref-type="bibr" rid="B16">16</xref>] was grown with 100 μg/ml ampicillin and 60 μg/ml streptomycin and its parent strain DH1 additionally with 20 μg/ml uracil. Differentiation was induced by washing the cells twice in ice cold Sørensen phosphate buffer (17 mM KH<sub>2</sub>/Na<sub>2</sub>H, PO<sub>4</sub>, pH 6.0). Cells were shaken on a rotary shaker at 23°C, 150 rpm at 2 × 10<sup>7 </sup>cells/ml until use.</p><sec><title>Preparation of vesicles</title><p>20 ml 2 × 10<sup>7</sup>cells/ml were shaken in Sørensen phosphate buffer containing 5 mM EGTA for 2–4 h, washed once in ice-cold 20 mM Hepes buffer, pH 7.2, resuspended at 2 × 10<sup>8 </sup>cells/ml, and lysed by passage through nuclepore filters. Immediately, 3% sucrose, 50 mM KCl, 1 mM MgCl<sub>2</sub>, 10 μg/ml leupeptin, 1 μg/ml aprotinin and 1.25 mM dithiothreitol were added (final concentration). After centrifugation for 5 min at 150 × g in order to remove unbroken cells, the supernatant was further fractionated by centrifugation for 5 min at 3800 × g. The sediment (P<sub>0</sub>) was resuspended in 1 ml 10 mM Tris buffer, pH 7.8, 3% sucrose, 50 mM KCl, 1 mM MgCl<sub>2</sub>, 20 μg/ml leupeptin and 2 μg/ml aprotinin (TKS-buffer). The supernatant was centrifuged at 12.000 × g for 20 min. The sediment (P<sub>1</sub>) was resuspended in 0.9 ml TKS-buffer. The supernatant was further centrifugated at 40.000 × g for 30 min. The fluffy sediment was resuspended in about 0.6 ml TKS-buffer and designated P<sub>2</sub>.</p></sec></sec><sec><title>Purification of vesicles with anti-GFP-magnetic microbeads</title><p>400 μl P<sub>0 </sub>suspended in 10 mM Tris-buffer, pH 7.8, 50 mM KCl, 20 μg/ml leupeptin and 2 μg/ml aprotinin (TK-buffer) was incubated for 30 min at 4°C with 100 μl anti-GFP-microbeads and then transferred to a Mini MACS-separation column exposed to a magnetic field. The column had been rinsed with 500 μl TK-buffer. The flow through was centrifuged for 5 min at 3.800 × g. The sediment was resuspended in 150 μl of TKS-buffer and designated P<sub>0</sub>F. In the meantime the column was washed with 500 μl TK-buffer, 500 μl TK-buffer containing 0.1 M NaCl and 500 μl TK-buffer containing 0.3 M NaCl. In case the column was running inefficiently the contents were transferred to a new column by pressing 500 μl TK-buffer through the old column in the absence of a magnetic field onto the new column positioned in the field. The final elution was performed in the same way after the column had been washed free of salt with TK-buffer. The washing solutions usually contained negligible amounts of the GFP-label. The purified eluate was centrifuged for 5 min at 3.800 × g, resuspended in about 350 μl TKS-buffer and designated P<sub>0</sub>E.</p><p>In most experiments the cells were treated with 8 μM latrunculin B for 20 min before lysis to disrupt F-actin filaments. Under this condition the cells rounded up and became immotile. Nonetheless, the dajumin-GFP-label remained associated with P<sub>0 </sub>and purification did not change significantly.</p></sec><sec><title>Ca<sup>2+</sup>-transport</title><p>Ca<sup>2+</sup>-transport was measured essentially as described [<xref ref-type="bibr" rid="B23">23</xref>]. In brief, about 60 μl fraction was added to 10 mM Hepes, pH 7.2, 50 mM KCl, 3% sucrose, 6 μg/ml antimycin A, 6 μg/ml oligomycin A, 100 μM NaN<sub>3</sub>, 2 mM MgCl<sub>2</sub>, and about 0.5 μM Fluo-3 in a total volume of 1 ml. Ca<sup>2+</sup>-uptake was initiated by addition of 400 μM ATP. Fluo-3 fluorescence was monitored at 505 nm excitation and 526 nm emission with a fluorimeter (Perkin Elmer 650-10S, Überlingen, Germany). Since [Ca<sup>2+</sup>]<sub>ev </sub>at the beginning of the experiment is not known and Ca<sup>2+</sup>-ATPase activity is increasing with increasing Ca<sup>2+</sup>-concentrations the results shown in Table <xref ref-type="table" rid="T2">2</xref> were corrected for the same initial [Ca<sup>2+</sup>]<sub>ev </sub>in order to relate the outcome of different experiments to each other. This was done by using a calibration curve at different [Ca<sup>2+</sup>]<sub>ev </sub>obtained with either P<sub>0 </sub>or P<sub>1</sub>, respectively, and adjusting the measured value to a [Ca<sup>2+</sup>]<sub>ev </sub>of 1 μM. The Ca<sup>2+</sup>-ATPase activities of both, P<sub>0 </sub>and P<sub>1</sub>, increased by a factor of about 3.3 from 0.5 μM to 1.5 μM [Ca<sup>2+</sup>]<sub>ev</sub>. Statistical data are expressed throughout as the mean ± S.D.</p></sec><sec><title>Immunoblotting</title><p>Seven microgram of protein per lane were resolved by SDS-PAGE in 12.5% gels, blotted and immunoblotted with a V-type H<sup>+</sup>-ATPase antibody (kindly provided by Margaret Clarke) preferentially reacting with the 37/42 kD subunits of the proton pump.</p></sec><sec><title>Other measurements</title><p>cAMP-induced Ca<sup>2+</sup>-influx was measured as described [<xref ref-type="bibr" rid="B7">7</xref>]. The centrifugation time of the cells (LvsA<sup>- </sup>and DH1) during every washing step was shortened without any loss of cells. Protein concentrations were determined with the Coomassie protein assay reagent (Pierce) using bovine serum albumin as standard. Endosomal content was determined by fluid phase uptake of RITC-dextran as described, except that labeling was performed in phosphate buffer pH 6.0 [<xref ref-type="bibr" rid="B18">18</xref>].</p></sec></sec><sec><title>Abbreviations</title><p>AA: arachidonic acid; BHQ: 2,5-di(tert-butyl)-1,4-hydroquinone; CV: contractile vacuole; CMA: concanamycin A; ER: endoplasmic reticulum; Tg: thapsigargin</p></sec><sec><title>Authors' contributions</title><p>DM performed the organelle experiments and in collaboration with CS and DFL designed the experiments and wrote the manuscript. DFL measured receptor-mediated Ca<sup>2+</sup>-influx and responses of living cells to AA; CS cultured the strains and prepared the figures. ADL contributed the strains LvsA<sup>- </sup>and DH1 and participated in the preparation of the manuscript as did AMT, who also contributed the GFP-strains and Figure <xref ref-type="fig" rid="F1">1</xref>.</p></sec>
|
Identifying patients at risk of nursing home admission: The Leeds Elderly Assessment Dependency Screening tool (LEADS)
|
<sec><title>Background</title><p>Discharge from hospital to a nursing home represents a major event in the life of an older person and should only follow a comprehensive functional and medical assessment. A previous study identified 3 dependency scales able to discriminate across outcomes for older people admitted to an acute setting. We wished to determine if a single dependency scale derived from the 3 scales could be created. In addition could this new scale with other predictors be used as a comprehensive tool to identify patients at risk of nursing home admission.</p></sec><sec sec-type="methods"><title>Methods</title><p>Items from the 3 scales were combined and analysed using Rasch Analysis. Sensitivity and specificity analysis and ROC curves were applied to identify the most appropriate cut score. Binary logistic regression using this cut-off, and other predictive variables, were used to create a predictive algorithm score. Sensitivity, specificity and likelihood ratio scores of the algorithm scores were used to identify the best predictive score for risk of nursing home placement.</p></sec><sec><title>Results</title><p>A 17-item (LEADS) scale was derived, which together with four other indicators, had a sensitivity of 88% for patients at risk of nursing home placement, and a specificity of 85% for not needing a nursing home placement, within 2 weeks of admission.</p></sec><sec><title>Conclusion</title><p>A combined short 17-item scale of dependency plus other predictive variables can assess the risk of nursing home placement for older people in an acute care setting within 2 weeks of admission. This gives an opportunity for either early discharge planning, or therapeutic intervention to offset the risk of placement.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Slade</surname><given-names>Anita</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Fear</surname><given-names>Jon</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Tennant</surname><given-names>Alan</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
BMC Health Services Research
|
<sec><title>Background</title><p>The National Service Framework for older people within the UK highlighted the need for a single assessment process to determine the most appropriate setting for ongoing care [<xref ref-type="bibr" rid="B1">1</xref>]. To date, professionals are often faced with a large assortment of scales to choose from, identifying measures to aid them in this decision was seen as a priority. As the setting of care is largely determined by the extent of dependency and, for example nursing needs, then clearly measures of dependency will be important in this process. A previous study examined the use of 7 outcome scales and other predictive factors (e.g. presence of pressure sores) in order to identify which scales were predictive of outcome when the patient had recently entered an acute hospital setting[<xref ref-type="bibr" rid="B2">2</xref>]. Out of the 7 scales examined only four scales, The Modified Barthel Index (MBI) the Abbreviated Mental Test (AMTS), the Northwick Park Dependency Scale (NPDS), and the Caregiver Strain Index (CSI) were found to discriminate across outcomes as defined by placement, for example to a nursing home [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B7">7</xref>].</p><p>Mindful of the fact that using several different scales can be time consuming for staff and stressful for the patient we wished to determine whether it was possible to create a screening tool to identify dependency consistent with a nursing home placement from three of these health status scales (MBI, AMTS, NPDS). The CSI was not included as not all patients have a carer, and the three other scales were, in theory, measuring an underlying construct of dependency. If combining some of the original items, together with other key predictive variables into a new scale, could create an algorithm to act as a screen for such risk, this may provide an economical way of assessing the likely need for a nursing home admission.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Objective</title><p>To assess if a screening tool (the Leeds Elderly Assessment Dependency Scale (LEADS) could be developed from three previously identified scales which discriminate for nursing home placement [<xref ref-type="bibr" rid="B2">2</xref>]. To test if this scale, along with other key predictive variables (identified in previous study), would be sensitive and specific to predicting the need for a nursing home placement within two weeks of admission to acute wards for the Care of the Elderly.</p></sec><sec><title>Participants</title><p>Patients were recruited on admission to the Care of the Elderly wards in a multi-site acute hospital trust. A random sample of every fourth patient admitted to three such wards, together with all patients requiring a comprehensive assessment were included in the study. Full details of recruitment and patient characteristics are given elsewhere [<xref ref-type="bibr" rid="B2">2</xref>].</p></sec><sec><title>Outcome</title><p>Outcome was defined in terms of discharge destination i.e. whether the person was discharged to home, home with family/carer, sheltered housing, residential or nursing home care (in the initial study). In this paper we are primarily concerned with differentiating between nursing home placement and the 'other' placements, as nursing home placement has considerable impact on patients, their families, and on service provision.</p></sec><sec><title>Statistical Methods</title><p>A three stage approach was used to develop the screening tool. (Figure <xref ref-type="fig" rid="F1">1</xref>)</p><sec><title>Stage 1: Rasch analysis, developing the LEADS</title><p>The three scales were combined and the development of a single shorter scale was explored using Rasch analysis [<xref ref-type="bibr" rid="B8">8</xref>]. The Rasch model is the current standard for the development of unidimensional scales (e.g. of impairment or dependency) delivering metric quality outcomes in health care [<xref ref-type="bibr" rid="B9">9</xref>]. Briefly, data collected from scales completed by clinical staff, which are intended to be summated into an overall score are tested against the expectations of the Rasch measurement model. The model defines how responses to items should be if measurement (at the metric level) is to be achieved. This was considered the most appropriate model for identifying items that could measure the underlying latent trait (level of disability) into a short assessment tool as it is the only measurement model delivering a metric transformation of ordinal scale [<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B11">11</xref>]. A previous study constructed short form scales using this method without any loss of validity, and found it to be a better method for item reduction than more classical forms of item reduction e.g. principal components analysis [<xref ref-type="bibr" rid="B7">7</xref>]. Wolfe in his paper discussed different mechanisms for combining scales [<xref ref-type="bibr" rid="B10">10</xref>]. The method chosen for the current analysis was the common person equating method. That is the same patient is assessed, at the same time, and assigned values (by clinicians) to each item on the three scales. In looking for potential items to discard, items that were identified as redundant by the Rasch analysis (that is they had high negative residuals, equating to high item-total correlations in classical analysis), were discarded. Also, items that failed to follow the expected probabilistic relationship of a valid scale (misfitting items), were also discarded. Finally, items that showed bias for external factors such as age and gender (Differential Item Functioning were also removed) [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>The most appropriate cut off point was determined by the sensitivity, specificity, likelihood ratio and ROC curves [<xref ref-type="bibr" rid="B14">14</xref>]. Deeks and Altman suggest that likelihood ratios have more powerful properties making them more appropriate in clinical use than sensitivity and specificity alone [<xref ref-type="bibr" rid="B15">15</xref>]. Likelihood ratios is the ratio of the probability of finding people who will need nursing home placement to the probability of predicting those patients who will not be at risk of nursing home placement using a defined cut-off score. For comparison, ROC curves are also presented [<xref ref-type="bibr" rid="B16">16</xref>].</p></sec><sec><title>Stage 2: Binary Logistic Regression</title><p>Following the construction of the LEADS, a binary logistic regression analysis was used to identify the screening cut-off score derived from the LEADS in combination with the other variables [<xref ref-type="bibr" rid="B17">17</xref>]. As the numbers going into a nursing home were expected to be relatively small, it was expected that this would cause some problems in interpretation. King and Zeng identified the difficulties in analysing rare event analysis in logistic regression and advocate that all the rare event cases are used and random sample of the remainder [<xref ref-type="bibr" rid="B18">18</xref>]. Consequently this strategy was adopted here, with repeated random samples (in the event of 50 cases) taken from the remainder (the non-nursing home cases) in a one-to -two ratio of those placed in nursing home and non-nursing home cases. The exp (B) values from the binary logistic regression for the indicator variables, together with the LEADS cut-off score were used in an algorithm to create an overall algorithm score for the risk of nursing home placement [<xref ref-type="bibr" rid="B19">19</xref>]. (See Appendix 1)</p></sec><sec><title>Stage 3: Sensitivity, specificity and likelihood ratios</title><p>The predictive score for nursing home, identified by the algorithm was then examined for sensitivity and specificity to identify the best predictive cut-off score for the total algorithm. The aim was to maximise sensitivity, and minimise false positives and again, for comparison, ROC curves are presented [<xref ref-type="bibr" rid="B20">20</xref>].</p></sec></sec><sec><title>Ethics and Consent</title><p>Ethical committee approval was obtained from the Leeds Teaching Hospital NHS Trust. Patients were asked to sign a written consent form. If they were unable to give consent due to cognitive, visual or communication problems, a relative or carer was asked to consent on their behalf.</p></sec><sec><title>Software</title><p>Statistical analysis was undertaken with SPSS version 11.5 and RUMM 2020 [<xref ref-type="bibr" rid="B21">21</xref>,<xref ref-type="bibr" rid="B22">22</xref>].</p></sec></sec><sec><title>Results</title><p>Five hundred and forty nine patients were recruited into the full study, of whom 258 were assessed on each of the three scales to be analysed as potential contributions to the screening tool, and discharged to various destinations [<xref ref-type="bibr" rid="B2">2</xref>]. The mean age of these 258 patients was 83.8 years (SD 5.5) and their mean length of stay was 31.2 days (SD 31.6). Seven out of ten (70.2%) were female.</p><sec><title>Stage 1: Reducing the item set and producing a single scale: using Rasch Analysis</title><p>Items from the 3 scales were merged and examined with the Rasch model. Items that showed misfit or interdependency were discarded, as were those showing bias for age and gender. Reducing the item set in this way gave a 17-item scale with a unidimensional construct of dependency that incorporated both cognitive and functional ability items. The new scale, called the Leeds Elderly Assessment Dependency Screening Tool (LEADS) included 7 items from the MBI, 3 from the AMTS and 7 from the NPDS (Table <xref ref-type="table" rid="T1">1</xref>). The overall fit to the Rasch model of this common 17-item scale was good, with Item Fit of -0.414 (SD 1.013), Person Fit of -0.241 (SD 0.559) and Item-trait interaction of (0.004) (Bonferroni correction significance used 0.0006) [<xref ref-type="bibr" rid="B23">23</xref>].</p><p>Enteral feeding (NPDS 8.3) was the item with the highest negative location (-5.799 logits). This means that the majority of people did not require enteral feeding. In contrast the stairs item (MBI 3) had the highest positive location (+5.031 logits), suggesting that the majority of people found stairs difficult, and independence in this activity was difficult for this group to achieve (Table <xref ref-type="table" rid="T1">1</xref>).</p><p>Distribution of people and items for the 17-item scale is good, as shown by the person separation index of 0.944, this indicates that the scale is able to discriminate between several different groups of patients [<xref ref-type="bibr" rid="B24">24</xref>]. A Principal components analysis of the residuals showed, with a non-significant Bartlett's test, that no patterns remained in the residuals, thus supporting the unidimensionality of the 17-item scale. The distribution of people and item thresholds can be seen in Figure <xref ref-type="fig" rid="F2">2</xref>. Each person is shown located on the logit metric scale and thus the raw score from the LEADS can be transformed into a linear metric number [<xref ref-type="bibr" rid="B25">25</xref>]. The metric logit location scores from the LEADS (ranging from approximately -6 to + 7 logits) were converted to the range of 0 – 39 (i.e. the same as the original raw score from the contributing items, to facilitate ease of interpretation).</p><p>The mean score for those discharged to a nursing home was found to be 15.07 (SD 4.33), while those discharged home without a carer was 23.69 (SD 8.17) (Table <xref ref-type="table" rid="T2">2</xref>). There is a significant difference of the LEADS score by discharge destination (Kruskal-Wallis sig. <0.001) (Data was found to be bimodal) and this is shown in Figure <xref ref-type="fig" rid="F3">3</xref>. As scores for destinations other than nursing home were similar (and had overlapping confidence intervals) we grouped all these destinations into an overall 'other' category. The ROC curve and sensitivity and specificity of different cut points identified the score of 19 as being able to maximise sensitivity (88%) and minimise specificity (61%). With an area under of the curve of 0.81 (SE .036) sig 0.000 (CI 0.738–0.881) this shows that the LEADS score has good predictability as a test for nursing home admission. (Figure <xref ref-type="fig" rid="F4">4</xref>)</p></sec><sec><title>Stage 2: Predicting the need for nursing home placement: Binary logistic regression</title><p>Using the nursing home and the combined 'other' groups as a dependent variable, a binary logistic regression was used to identify predictors for patients at risk of a nursing home placement. There were a disproportionate number of cases between these groups (233:25) and thus, five random samples were selected from the 'other' group and added to the nursing home group to create repeated samples for analysis. The results were consistent for all samples and thus the results presented are the sample that gave the best predictive model.</p><p>The best model to predict nursing home placement included the LEADS (with its cut score at 19) together with respite care on admission, communication difficulties on admission, family or patient wishes for placement, and pressure sores (grade 1 or above). The -2 log likelihood statistic (27.227) is analogous to the error sum of squares in multiple regression and is an indicator to how much unexplained information there is after the model is fitted (Table <xref ref-type="table" rid="T3">3</xref>). The main model (including all of the variables) is explaining approximately 75% of the variance in predicting nursing home need (chi square (<0.01)); this is significantly better than the best model from the five samples which does not use these predictors.</p><p>The coefficients derived from this analysis were then used as the basis of the full algorithm for predicting nursing home placement (Table <xref ref-type="table" rid="T4">4</xref>). Exp (B) indicates the change in odds resulting from a unit change in the predictor. For example having a score of less than 19 in the LEADS increases the odds of needing a nursing home admission sixty-five times. Each of these values were included in the algorithm (see Appendix 1).</p></sec><sec><title>Stage 3: Sensitivity, specificity of the final algorithm</title><p>Initially the random sample, which gave the best result for the LEADS was used to determine the best cut score for the full algorithm score based upon the exponentiated values of the logistic regression, and ranging from 0 to 358. This gave a cut score of 244, with sensitivity of 0.88 and specificity of 0.96, and a positive predictive value of 0.92. The likelihood ratio (22.44) was maximised at this value. In addition ROC Curves (figure <xref ref-type="fig" rid="F5">5</xref>) showed that the area under the curve was 0.972 (SE .016) and this was significant p < 0.000 CI 0.940 to 1.004.</p><p>The algorithm was then re-tested on the original full data set of those discharged to home or sheltered accommodation, to residential care, or nursing home care., Thus 258 patients were used to test its sensitivity, specificity and predictive power. Examination of the ROC curve identified 244 as the score that maximised sensitivity (88%) and minimised specificity (85%) (Table <xref ref-type="table" rid="T5">5</xref>). The area under the curve for the algorithm score was 0.921 (SE 0.019, sig. 0.000, CI 0.883 to 0.959). Therefore the algorithm score is an excellent predictive test and with a likelihood ratio of 6.04 this indicates that given an algorithm score of less than 244 there is a moderate increase in the likelihood that patients will require nursing home admission. It identified 34 patients needing nursing home that subsequently went elsewhere and identified 3 as 'other' who eventually went to a nursing home. Of the 34 who were predicted as needing nursing home and went elsewhere, 22 (64.7%) returned home, 3 (8.8%) returned home with a carer, 7 (20.6%) went to residential accommodation and 2 (5.9%) went to sheltered housing.</p></sec></sec><sec><title>Appendix 1 - Algorithm</title><p>The algorithm for predicting nursing home placement using SPSS based on minimum Exp B</p><p>Compute predscrn = 358</p><p>if (LEADS score le 19) predscrn = predscrn -65</p><p>if (familypat wish = 1(no)) predscrn = predscrn -34</p><p>if (communication difficuties = 1 (yes)) predscrn = predscrn -112</p><p>if (grade 1+ pressure sore = 1 (yes)) predscrn = predscrn -18</p><p>if (respite care on admission = 1 (yes)) predscrn = prdescrn -128</p><p>Excel spreadsheet with the algorithm and scoring for the LEADS is available from:</p><p><email>[email protected]</email> or <email>[email protected]</email></p></sec><sec><title>Discussion</title><p>Currently, people are faced with a bewildering variety of potential measures for use in assessing outcome in an acute elderly setting. Having previously identified three scales that discriminated between people going to a nursing home as against other outcomes, we have now shown that it is possible, through Rasch analysis, to extract items that work well together and measure the underlying dependency trait. Clinicians may still wish to use the original scales for clinical purposes but in terms of measurement, the 17-item LEADS scale and associated algorithm has been shown to be a powerful tool in predicting patients at risk of nursing home placement and those likely to go to other types of care or home.</p><p>The false positive rate in the final analysis may be viewed as a major weakness in the approach. Some patients improved such that they could go home, or into other institutional settings. This is a valid comment and the majority of mismatch between the indicative and final placement was for those patients who went home. It is important to remember this data was collected within two weeks of admission. Thus the algorithm, as well as providing a common equitable means of assessment, can act as an early warning system for risk of institutional care. Early identification of those patients at risk enables interventions to be instigated early on in their admission, potentially reducing the risk of nursing home placement. Given the parsimony of the scale there is nothing to prevent repeated measurements during the patients stay in hospital, so providing a monitoring system for the continuing risk of institutional placement.</p><p>There are a number of weaknesses to the study. The low number of patients subsequently entering a nursing home was always going to be a cause of concern. However, we accommodated this, as best as possible, by sampling from the other group and comparing the results. The assessment was also only undertaken once within two weeks of admission. Additional work needs to be carried out using repeated assessments and to look at the changes in sensitivity and specificity over time in order too determine if there is an optimum time to maximise these parameters. Due to the low numbers of those entering a nursing home, we had to use this group in the development of the algorithm cut point, as well as its validation. This is likely to overestimate its predictive value, although we did try to offset this as far as possible by validating the algorithm on the full data set, rather than the developmental sample. Finally, as with all models developed on a particular set of data, these results need replicating on other elderly acute samples to support conclusions about the predictive validity of the screening tool.</p></sec><sec><title>Conclusion</title><p>Using selective items from three separate scales, previously shown to be discriminative for nursing home placement, together with other key indicators, enables those working in an acute setting, within two weeks of admission, to identify 85 % of patients at risk of needing nursing home placement. The resulting LEADS scale and four indicator variables can easily be administered by any health care professional and the risk algorithm lends itself to a simple spreadsheet calculation.</p></sec><sec><title>Abbreviations</title></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>AS, AT and JF conceived the study question and design.</p><p>AS was the principal data analyst.</p><p>AS wrote the provisional drafts of the manuscript with AT and JF reviewing the manuscripts.</p><p>AS was responsible for data collection and fieldwork.</p><p>All authors contributed to the critical evaluation of the methods, analysis and writing.</p><p>AT is the guarantor of the study.</p><p>All authors read the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1472-6963/6/31/prepub"/></p></sec>
|
BAC-FISH refutes report of an 8p22–8p23.1 inversion or duplication in 8 patients with Kabuki syndrome
|
<sec><title>Background</title><p>Kabuki syndrome is a multiple congenital anomaly/mental retardation syndrome. The syndrome is characterized by varying degrees of mental retardation, postnatal growth retardation, distinct facial characteristics resembling the Kabuki actor's make-up, cleft or high-arched palate, brachydactyly, scoliosis, and persistence of finger pads. The multiple organ involvement suggests that this is a contiguous gene syndrome but no chromosomal anomalies have been isolated as an etiology. Recent studies have focused on possible duplications in the 8p22–8p23.1 region but no consensus has been reached.</p></sec><sec sec-type="methods"><title>Methods</title><p>We used bacterial artificial chromosome-fluorescent <italic>in-situ </italic>hybridization (BAC-FISH) and G-band analysis to study eight patients with Kabuki syndrome.</p></sec><sec><title>Results</title><p>Metaphase analysis revealed no deletions or duplications with any of the BAC probes. Interphase studies of the Kabuki patients yielded no evidence of inversions when using three-color FISH across the region. These results agree with other research groups' findings but disagree with the findings of Milunsky and Huang.</p></sec><sec><title>Conclusion</title><p>It seems likely that Kabuki syndrome is not a contiguous gene syndrome of the 8p region studied.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Kimberley</surname><given-names>Kendra W</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Morris</surname><given-names>Colleen A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Hobart</surname><given-names>Holly H</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Medical Genetics
|
<sec><title>Background</title><p>Kabuki syndrome [OMIM 147920] is a multiple congenital anomaly/mental retardation syndrome of unknown etiology. The syndrome was first described independently by Niikawa <italic>et. al</italic>. and Kuroki <italic>et. al</italic>. in 1981 [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. There is no definitive laboratory diagnostic test for Kabuki syndrome, thus diagnosis is made based on phenotypic presentation and by ruling out other known syndromes. The typical characteristics of Kabuki syndrome are: varying degrees of mental retardation, postnatal growth retardation, characteristic dysmorphic facial features, congenital heart disease, brachydactyly, scoliosis, and persistence of prominent finger pads.</p><p>The dysmorphic facies are characterized by long palpebral fissures with eversion of the lateral part of the lower eyelids, arched eyebrows, broad, depressed nasal tip, prominent ears, cleft or high-arched palate, and dental/occlusal abnormalities. Individuals with Kabuki syndrome have relative strengths in verbal and reasoning abilities and weakness in visuospatial construction; adaptive behavior is characterized by good social interactions and communication but poor motor and personal living skills [<xref ref-type="bibr" rid="B3">3</xref>]. The multiple organ involvement suggests that Kabuki syndrome is a contiguous gene syndrome but no etiology has been determined. There have been several cases of Kabuki syndrome with various chromosomal anomalies but none have had any cytogenetic anomaly in common [<xref ref-type="bibr" rid="B4">4</xref>-<xref ref-type="bibr" rid="B6">6</xref>]. Milunsky and Huang [<xref ref-type="bibr" rid="B7">7</xref>] recently reported finding a duplication of 8p22–8p23.1 in six unrelated patients with Kabuki syndrome. They also reported finding a submicroscopic inversion in all six patients and in two mothers of Kabuki syndrome children. However, studies conducted by Miyake <italic>et. al</italic>. [<xref ref-type="bibr" rid="B5">5</xref>], Schoumans <italic>et. al</italic>. [<xref ref-type="bibr" rid="B6">6</xref>], Miyake <italic>et. al</italic>. [<xref ref-type="bibr" rid="B8">8</xref>], Hoffman <italic>et. al</italic>. [<xref ref-type="bibr" rid="B9">9</xref>], Engelen <italic>et. al</italic>. [<xref ref-type="bibr" rid="B10">10</xref>], Sanlaville <italic>et. al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>], and Turner <italic>et. al</italic>. [<xref ref-type="bibr" rid="B12">12</xref>] failed to find evidence of the duplications reported by Milunsky and Huang. We investigated eight patients with Kabuki syndrome and six normal controls to determine if either an inversion or a duplication in 8p22–8p23.1 exists.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Clinical evaluation</title><p>Eight individuals with Kabuki syndrome identified through the University of Nevada School of Medicine Genetics Program consented to participate in the study. Medical records were reviewed and physical examinations were completed. The diagnosis was based upon the presence of the typical pattern of malformations and dysmorphic features reviewed by Matsumoto and Niikawa [<xref ref-type="bibr" rid="B13">13</xref>].</p></sec><sec><title>Cytogenetics</title><p>Blood was collected from patients diagnosed with Kabuki syndrome, their parents, and normal controls according to approved Institutional Review Board procedures. Blood lymphocytes were cultured using standard cytogenetic techniques with thymidine used for synchronization. The lymphocytes were dropped onto clean, wet slides to obtain well spread metaphase chromosomes and interphase nuclei. The slides were allowed to air dry and were then baked for 1 ½ hours at 90°C. Analysis of metaphase chromosomes occurred by using GTG-banding standard techniques.</p></sec><sec><title>Fluorescent <italic>in-situ </italic>hybridization (FISH)</title><p>Ten bacterial artificial chromosome (BAC) clones that both Milunsky and Huang [<xref ref-type="bibr" rid="B7">7</xref>] and Miyake <italic>et. al</italic>. [<xref ref-type="bibr" rid="B8">8</xref>] reportedly used (RP11-11P7, RP11-140K14, RP11-122N11, RP11-235F10, RP11-112G9, RP11-252K12, RP11-31B7, RP11-92C1, RP11-23H1, and RP11-141K9) were obtained from the Children's Hospital Oakland Research Institute (Oakland, CA) human genomic library. The BACs were grown and the DNA harvested using standard methods. The ends of the BAC DNA insert were sequenced at the Nevada Genomics Center at the University of Nevada, Reno, NV. The sequences were compared to published NCBI genomic sequences using the BLAST program to confirm the location of the BAC DNA on the human genome.</p><p>The BAC DNA was then labeled separately in both Spectrum Red and Spectrum Green using a Nick Translation Kit (Vysis, Downers Grove, IL). The labeled DNA was co-precipitated with Human Hybloc (Cot-1) DNA (Applied Genetics Laboratories, Melborne, FL) and tested on normal control slides to confirm labeling success.</p><p>To detect inversions, three-color FISH was performed by applying probes labeled in Spectrum Red and Spectrum Green. The third color, yellow, was achieved by mixing equal parts of red and green labeled probe. Prepared slides of both Kabuki patients and normal controls were serially dehydrated in ethanol and allowed to air dry. Labeled BAC probe was applied. The slides were coverslipped and sealed with rubber cement. Co-denaturation of probe and target DNA occurred at 77.5°C for 6 minutes followed by hybridization at 37°C overnight. The coverslips were removed and the slides washed in 50% formamide/2X SSC at 50°C for 15 minutes followed by a wash in 2X SSC at 37°C for 8 minutes to remove any mismatched probe. The slides were air dried, counterstained with DAPI/Antifade, coverslipped, and sealed. The slides were then examined by fluorescence microscopy and scored for probe number and probe order.</p></sec></sec><sec><title>Results and discussion</title><p>All eight individuals met the clinical criteria for a diagnosis of Kabuki syndrome. They all had the Kabuki syndrome distinctive facial features (long palpebral fissures with everted lateral lower lids, arched eyebrows, short nasal septum, and prominent ears) and the typical physical findings (hypotonia, prominent fingertip pads, and joint laxity). The additional clinical features are summarized in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Kabuki syndrome clinical features. The clinical features of the eight Kabuki patients included in this study.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Patient</td><td align="center">1</td><td align="center">2</td><td align="center">3</td><td align="center">4</td><td align="center">5</td><td align="center">6</td><td align="center">7</td><td align="center">8</td></tr><tr><td align="center">Age</td><td align="center">8 y</td><td align="center">10.7 y</td><td align="center">9 y</td><td align="center">26 y</td><td align="center">11.5 y</td><td align="center">9 y</td><td align="center">11.5 y</td><td align="center">13 mo</td></tr><tr><td align="center">Sex</td><td align="center">M</td><td align="center">M</td><td align="center">M</td><td align="center">F</td><td align="center">F</td><td align="center">F</td><td align="center">F</td><td align="center">F</td></tr><tr><td align="center">Age of Dx</td><td align="center">6 y</td><td align="center">1 mo</td><td align="center">9 mo</td><td align="center">8 mo</td><td align="center">11.5 y</td><td align="center">9 y</td><td align="center">5 y</td><td align="center">7 mo</td></tr></thead><tbody><tr><td align="left" colspan="9">Development</td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center">Single Words</td><td align="center">24 mo</td><td align="center">5 y</td><td align="center">24 mo</td><td align="center">19 mo</td><td align="center">30 mo</td><td align="center">12 mo</td><td align="center">36 mo</td><td align="center">No</td></tr><tr><td align="center">Walking</td><td align="center">16 mo</td><td align="center">18 mo</td><td align="center">22 mo</td><td align="center">16 mo</td><td align="center">3.5 y</td><td align="center">12 mo</td><td align="center">30 mo</td><td align="center">No</td></tr><tr><td align="center">Hearing Loss</td><td align="center">No</td><td align="center">No</td><td align="center">No</td><td align="center">No</td><td align="center">Yes</td><td align="center">No</td><td align="center">No</td><td align="center">No</td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="left" colspan="9">Growth%</td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center">L</td><td align="center">40%</td><td align="center">30%</td><td align="center"><3%</td><td align="center">50%</td><td align="center"><3%</td><td align="center">25%</td><td align="center"><5%</td><td align="center">75%</td></tr><tr><td align="center">W</td><td align="center">60%</td><td align="center">80%</td><td align="center"><3%</td><td align="center">80%</td><td align="center"><3%</td><td align="center">98%</td><td align="center">70%</td><td align="center">25%</td></tr><tr><td align="center">OFC</td><td align="center">30%</td><td align="center">98%</td><td align="center"><3%</td><td align="center">5%</td><td align="center">2%</td><td align="center">50%</td><td align="center">20%</td><td align="center"><2%</td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="left" colspan="9">Anomalies</td></tr><tr><td colspan="9"><hr></hr></td></tr><tr><td align="center">Palate</td><td></td><td></td><td></td><td></td><td align="center">Cleft palate</td><td></td><td></td><td align="center">Lip pit</td></tr><tr><td align="center">Cardiac *</td><td align="center">BAV</td><td align="center">DAR</td><td align="center">ASD</td><td align="center">AS</td><td align="center">BAV</td><td></td><td></td><td align="center">Mitral insufficiency</td></tr><tr><td align="center">Renal</td><td></td><td></td><td></td><td align="center">Pelvic kidney</td><td></td><td></td><td></td><td></td></tr><tr><td align="center">Premature thelarche</td><td></td><td></td><td></td><td></td><td></td><td align="center">+</td><td align="center">+</td><td></td></tr><tr><td align="center">Scoliosis</td><td></td><td></td><td></td><td></td><td></td><td></td><td align="center">+</td><td></td></tr><tr><td align="center">Other *</td><td align="center">Crypt.</td><td></td><td></td><td align="center">GER,<break/>Ant. anus</td><td align="center">GER,<break/>Strabismus,<break/>CDH,<break/>Nystagmus</td><td align="center">Strabismus</td><td align="center">Seizure disorder,<break/>Hemivertebra</td><td align="center">GER,<break/>IgA and IgG deficiency</td></tr></tbody></table><table-wrap-foot><p>* AS, aortic stenosis; BAV, bicuspid aortic valve; ASD, atrial septal defect; DAR, dilated aortic root; GER, gastroesophageal reflux; Crypt., cryptorchidism; CDH, congenital dysplasia of hip; Ant. anus, anterior placement of the anus.</p></table-wrap-foot></table-wrap><sec><title>Cytogenetics</title><p>Five or more cells were analyzed for each patient following U.S. clinical cytogenetic standards. No cytogenetic abnormalities were noted by either G-band analysis of at least the 500-band level resolution or by targeted high resolution analysis of 8pter of the Kabuki patients.</p></sec><sec><title>BAC-FISH</title><sec><title>Metaphase FISH</title><sec><title>Normal controls</title><p>The BAC probes were tested against normal controls to confirm the probe location and signal number. Except BAC RP11-122N11, all BAC probes showed two equally strong signals on all metaphases and interphase nuclei. BAC RP11-122N11 showed four or five signals on all control metaphases and interphases and was rejected for use in this study. Interestingly, this BAC was reported by Milunsky and Huang as being inverted in their Kabuki patients [<xref ref-type="bibr" rid="B7">7</xref>]. Hollox <italic>et. al</italic>. [<xref ref-type="bibr" rid="B14">14</xref>] and Sanlaville <italic>et. al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>] have mapped this BAC to the middle of the β-defensin-gene cluster, a gene cluster that is polymorphic for copy number, corresponding to gene SPAG11 at 8p23.1. This BAC also matched with multiple loci on multiple chromosomes when the sequenced ends were compared to known human genomic sequences using the NCBI BLAST program.</p></sec><sec><title>Kabuki patients</title><p>The remaining nine BAC probes were tested on eight Kabuki syndrome probands. All probes yielded two equally strong signals on both metaphase chromosomes and interphase nuclei indicating no deletion or duplication of these probes. This supports the recent findings of three genome-wide array CGH screenings that did not reveal any pathological findings in Kabuki patients, indicating that microduplications or microdeletions are less likely to be the cause of Kabuki syndrome [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B9">9</xref>].</p></sec></sec><sec><title>Inversion FISH</title><p>Three-color FISH was performed on interphase nuclei of normal controls to confirm the order of the BAC probes across the region. The order obtained agreed with the NCBI published sequence. Three-color FISH was then performed on interphase nuclei of all eight Kabuki probands. Forty-five interphase nuclei were scored on each patient to determine the probe order. A nucleus was considered scorable as long as the signals were in a relatively straight line and all three signals could clearly be seen. Milunsky and Huang reported a submicroscopic inversion between BACs RP11-122N11 and RP11-235F10 [<xref ref-type="bibr" rid="B7">7</xref>]. Since BAC RP11-122N11 was not used in this study due to the multiple numbers of signals seen in the normal controls, the BACs flanking RP11-122N11 and RP11-235F10 in sequence order were used for detecting an inversion. No inversions were seen between RP11-235F10 and RP11-140K14 (figure <xref ref-type="fig" rid="F1">1</xref>) or RP11-235F10 and RP11-112G9 (not pictured). No other inversions were detected across the region.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>Inversion FISH</bold>. Image of three-color FISH used to detect an inversion in the region of a previously reported submicroscopic inversion. Probe order: PR11-11P7 (yellow), RP11-140K14 (red), and RP11-235F10 (green).</p></caption><graphic xlink:href="1471-2350-7-46-1"/></fig></sec></sec></sec><sec><title>Conclusion</title><p>To date, we have been unable to find evidence of any inversions, duplications, or deletions in the 8p22–8p23.1 region as reported by Milunsky and Huang [<xref ref-type="bibr" rid="B7">7</xref>]. Our results are consistent with those reported by Miyake <italic>et. al</italic>. [<xref ref-type="bibr" rid="B5">5</xref>], Schoumans <italic>et. al</italic>. [<xref ref-type="bibr" rid="B6">6</xref>], Miyake <italic>et. al</italic>. [<xref ref-type="bibr" rid="B8">8</xref>], Hoffman <italic>et. al</italic>. [<xref ref-type="bibr" rid="B9">9</xref>], Engelen <italic>et. al</italic>. [<xref ref-type="bibr" rid="B10">10</xref>], Sanlaville <italic>et. al</italic>. [<xref ref-type="bibr" rid="B11">11</xref>], and Turner <italic>et. al</italic>. [<xref ref-type="bibr" rid="B12">12</xref>]. The discrepancy that exists among the groups could be attributed to two possible scenarios: 1) the patients studied by Milunsky and Huang represent either a variant of Kabuki syndrome or a different syndrome of similar phenotypic characteristics (phenocopy), or 2) the reported duplication represents a normal variant in the 8p region as this region is known to undergo genetic rearrangement due to olfactory receptor-gene clusters causing unequal crossovers [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B11">11</xref>]; none of these genetic rearrangements have been associated with any clinical pathology. Further studies, possibly with collaboration between the groups, are required to determine why the discrepancies in the research results exist. However, it now seems less likely that Kabuki syndrome is caused by a contiguous gene rearrangement than by a mutation in a regulatory gene.</p></sec><sec><title>Competing interests</title><p>The authors declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>KK participated in creating the experimental design, harvesting the BAC DNA, the sequence alignment, the FISH studies, and writing the draft of the manuscript. CM performed the clinical diagnosis of Kabuki syndrome patients. HH helped with the experimental design, culturing the BAC clones, and reading the karyotypes.</p><p>All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2350/7/46/prepub"/></p></sec>
|
Evidence for the association of the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes with Type 1 Diabetes: a case control study
|
<sec><title>Background</title><p>Type 1 diabetes (T1D) is a chronic, autoimmune and multifactorial disease characterized by abnormal metabolism of carbohydrate and fat. Diminished carnitine plasma levels have been previously reported in T1D patients and carnitine increases the sensitivity of the cells to insulin. Polymorphisms in the carnitine transporters, encoded by the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes, have been involved in susceptibility to two other autoimmune diseases, rheumatoid arthritis and Crohn's disease. For these reasons, we investigated for the first time the association with T1D of six single nucleotide polymorphisms (SNPs) mapping to these candidate genes: slc2F2, slc2F11, T306I, L503F, OCTN2-promoter and OCTN2-intron.</p></sec><sec sec-type="methods"><title>Methods</title><p>A case-control study was performed in the Spanish population with 295 T1D patients and 508 healthy control subjects. Maximum-likelihood haplotype frequencies were estimated by applying the Expectation-Maximization (EM) algorithm implemented by the Arlequin software.</p></sec><sec><title>Results</title><p>When independently analyzed, one of the tested polymorphisms in the <italic>SLC22A4 </italic>gene at 1672 showed significant association with T1D in our Spanish cohort. The overall comparison of the inferred haplotypes was significantly different between patients and controls (χ<sup>2 </sup>= 10.43; p = 0.034) with one of the haplotypes showing a protective effect for T1D (rs3792876/rs1050152/rs2631367/rs274559, CCGA: OR = 0.62 (0.41–0.93); p = 0.02).</p></sec><sec><title>Conclusion</title><p>The haplotype distribution in the carnitine transporter locus seems to be significantly different between T1D patients and controls; however, additional studies in independent populations would allow to confirm the role of these genes in T1D risk.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Santiago</surname><given-names>Jose Luis</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Martínez</surname><given-names>Alfonso</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>de la Calle</surname><given-names>Hermenegildo</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Fernández-Arquero</surname><given-names>Miguel</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Figueredo</surname><given-names>M Ángeles</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" corresp="yes" contrib-type="author"><name><surname>de la Concha</surname><given-names>Emilio G</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Urcelay</surname><given-names>Elena</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Medical Genetics
|
<sec><title>Background</title><p>Type 1 diabetes (T1D) is a multifactorial autoimmune T-cell-mediated disease resulting from selective destruction of the insulin producing β cells in the pancreatic islets, leading to an absolute insulin deficiency. The risk of developing T1D is determined by a complex interaction between multiple genetic and environmental factors. Although susceptibility to disease is strongly associated with alleles in the major histocompatibility complex (MHC) [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>], there are more than 20 putative T1D susceptibility regions identified by linkage and association studies [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. At present, several non-MHC susceptibility loci with modest genetic effects have been clearly defined. However, it is well known that many non-MHC loci predisposing to T1D remain as yet undefined [<xref ref-type="bibr" rid="B5">5</xref>].</p><p>Type 1 diabetes is a chronic degenerative disease, with altered metabolism characterized by hyperglycemia and ketoacidosis and T1D patients depend on exogenous insulin to sustain life. The role of the carnitine system in cell metabolism is mainly known in the mitochondria, where the interaction between fatty acid and glucose metabolism is fundamental for cell energy production [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. However, carnitine not only contributes to the transport of activated long-chain fatty acids into mitochondria for β-oxidation, but it also increases the sensitivity of cells to insulin [<xref ref-type="bibr" rid="B8">8</xref>]. Decreased plasma carnitine levels have been reported in patients with type 2 diabetes [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B11">11</xref>] and some studies have investigated the carnitine status in T1D, finding similar results [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B15">15</xref>].</p><p>Adequate carnitine levels are required for normal lipid metabolism and are important for energy metabolism [<xref ref-type="bibr" rid="B16">16</xref>]. One important component of the carnitine system is the plasma membrane carnitine transporters, named organic cation transporters (OCTN1 and OCTN2) encoded by the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes, respectively. Both genes map to the cytokine gene cluster on chromosome 5q31 and show 88% homology and 77% identity in their sequences. Despite OCTN1 and OCTN2 are considered as carnitine transporters, only OCTN2 is a high-affinity human carnitine transporter, while the carnitine transport activity of OCTN1 is very low [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. In fact, a recent study has reported that the main substrate of this transporter is the ergothioneine, an intracellular antioxidant with metal ion affinity, which is transported one hundred times more efficiently than carnitine [<xref ref-type="bibr" rid="B18">18</xref>]. OCTN2 is widely expressed in many adult tissues, among them in pancreas, and it participates, at least in part, in proton/organic cation antiport at the renal apical plasma membrane level [<xref ref-type="bibr" rid="B19">19</xref>].</p><p>Recent reports performed associations of some polymorphisms within the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes with two other autoimmune complex diseases (rheumatoid arthritis and Crohn's disease) [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>]. The purpose of this study was to investigate the influence of the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes in type 1 diabetes risk in the Spanish population. Six SNPs along these genes were considered good markers to map this region: slc2F2 (rs3792876) and slc2F11 (rs 2306772), which are SNPs in the <italic>SLC22A4 </italic>gene, were originally associated with RA susceptibility [<xref ref-type="bibr" rid="B20">20</xref>]. Other polymorphisms within the same linkage disequilibrium (LD) block (slc2F1) and slc2F2 showed lack of association with T1D [<xref ref-type="bibr" rid="B22">22</xref>]. We have studied two additional SNPs in this gene: T306I (rs272893) and L503F (rs1050152, SNP located in exon 9 of <italic>SLC22A4</italic>). The OCTN2-promoter (rs2631367) is a transversion (-207G>C) disrupting a heat shock element in the promoter region of the <italic>SLC22A5 </italic>gene and it has been described, together with L503F, as etiologic variant in Crohn disease [<xref ref-type="bibr" rid="B21">21</xref>]. Finally, we analyzed an intronic SNP in the <italic>SLC22A5 </italic>gene: the OCTN2-intron (rs274559) in order to define haplotypes within these genes.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Patients</title><p>We studied 295 unrelated Spanish white T1D patients (149 men and 146 women) diagnosed according to the criteria of the American Diabetes Association (ADA) and 508 healthy controls recruited among blood donors. Both groups ethnically matched from the Madrid area. The age at onset for the T1D patients range from 1 to 55 years old (median age at onset 15 years) and all subjects were insulin-dependent at the time to study. The protocol followed the principles expressed in the Declaration of Helsinki and it was approved by the Hospital Ethics Committee.</p></sec><sec><title>SNP genotyping</title><p>SNPs slc2F2 (rs3792876), slc2F11 (rs2306772), T306I (rs272893), L503F (rs1050152) and OCTN2-intron (rs274559) were genotyped by TaqMan Assays on Demand under conditions recommended by manufacturer (Applied Biosystems), with identification numbers: C__3170428_10, C__3170458_1_, C__3170445_1_, C__3170459_10 and C__1173605_1, respectively. For the SLC22A5 promoter -207G>C (rs2631367), a TaqMan Assay by Design was performed.</p></sec><sec><title>Statistical analysis</title><p>Differences in allele or genotype frequencies for each marker were calculated by Chi-square, or Fisher's exact test when necessary. Associations were estimated by the odds ratio (OR) with 95% confidence interval (CI). Statistical analysis used Epi Info v. 6.02 (CDC Atlanta USA).</p><p>Maximum-likelihood haplotype frequencies were estimated by applying the Expectation Maximization (EM) algorithm implemented by the Arlequin software [<xref ref-type="bibr" rid="B23">23</xref>], with number of iterations set at 5000 and initial conditions at 50, with an epsilon value of 10<sup>-7</sup>. This software yields the estimated frequency of each haplotype, but it does not output the expected haplotypes for each individual. The frequency data were transformed into absolute numbers multiplying the frequencies presented in Table <xref ref-type="table" rid="T3">3</xref> by the total number of haplotypes in each group (patients and controls). Then, these values were introduced into contingency tables to calculate Chi-square and p-values.</p></sec></sec><sec><title>Results</title><p>Six SNPs were studied in order to check the role of <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes in T1D predisposition (see figure <xref ref-type="fig" rid="F1">1</xref>). Slc2F2, Slc2F11, T306I and L503F map in the <italic>SLC22A4 </italic>gene and two additional SNPs, OCTN2-promoter and OCTN2-intron, are located in the <italic>SLC22A5 </italic>gene. The analysis of the control cohort showed complete linkage disequilibrium (LD) between Slc2F11 and slc2F2, and T306I was also found in complete LD with OCTN2-intron. Both SNPs (slc2F11 and T306I) were not considered in the subsequent case-control study, because they do not supply additional information. These polymorphisms conformed to Hardy-Weinberg equilibrium.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Schematic representation of the chromosomal region 5q31 with the relative position of the polymorphisms studied.</p></caption><graphic xlink:href="1471-2350-7-54-1"/></fig><p>The association of the intronic slc2F2 polymorphism in the <italic>SLC22A4 </italic>gene with rheumatoid arthritis (RA) was originally reported in a Japanese population [<xref ref-type="bibr" rid="B20">20</xref>]. This SNP was even described as an etiological variant in Japan, being the mutant homozygous genotype (slc2F2*TT) associated with RA. However, this association could not be replicated either in British [<xref ref-type="bibr" rid="B24">24</xref>] or in Spanish populations [<xref ref-type="bibr" rid="B25">25</xref>]. Table <xref ref-type="table" rid="T1">1</xref> shows the phenotype and genotype distribution of this marker in the Spanish T1D patient and control cohorts, and lack of significant association with the disease was observed. The allelic distribution was similar to that previously observed in Spanish RA patients, (minor allele frequency slc2F2*T was 7.5% in RA and 9.5% in T1D patients). We tested the carrier rate of this allele (CT+TT vs. CC) to find a possible association with T1D, but a negative result was again obtained: OR = 1.30 (0.87–1.94); p = 0.17.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Genotype and phenotype frequency distribution for OCTN1 markers.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">Slc2F2</td><td align="center">T1D (%)</td><td align="center">Controls (%)</td></tr></thead><tbody><tr><td align="center">Alleles</td><td align="center">2n = 590</td><td align="center">2n = 1016</td></tr><tr><td align="center">C</td><td align="center">534 (90.5)</td><td align="center">938 (92.3)</td></tr><tr><td align="center">T</td><td align="center">56 (9.5)</td><td align="center">78 (7.7)</td></tr><tr><td align="center">Genotypes</td><td align="center">n = 295</td><td align="center">n = 508</td></tr><tr><td align="center">CC</td><td align="center">240 (81.4)</td><td align="center">432 (85)</td></tr><tr><td align="center">CT</td><td align="center">54 (18.3)</td><td align="center">74 (14.6)</td></tr><tr><td align="center">TT</td><td align="center">1 (0.3)</td><td align="center">2 (0.4)</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="center">L503F</td><td align="center">T1D (%)</td><td align="center">Controls (%)</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="center">Alleles</td><td align="center">2n = 590</td><td align="center">2n = 970</td></tr><tr><td align="center">C</td><td align="center">299 (50.7)</td><td align="center">545 (56.2)</td></tr><tr><td align="center">T</td><td align="center">291 (49.3)</td><td align="center">425 (43.8)</td></tr><tr><td align="center">Genotypes</td><td align="center">n = 295</td><td align="center">n = 485</td></tr><tr><td align="center">CC*</td><td align="center">69 (23.4)</td><td align="center">156 (32.2)</td></tr><tr><td align="center">CT</td><td align="center">161 (54.6)</td><td align="center">233 (48.0)</td></tr><tr><td align="center">TT</td><td align="center">65 (22.0)</td><td align="center">96 (19.8)</td></tr></tbody></table><table-wrap-foot><p>*OR = 0.64 (0.46–0.91); p = 0.009</p></table-wrap-foot></table-wrap><p>Regarding the L503F (1672G>C) polymorphism, it was described by Peltekova et al. [<xref ref-type="bibr" rid="B21">21</xref>] as a functional variant associated with Crohn's disease, together with the OCTN2-promoter (-207G>C) variant. In fact, the authors found that the odds ratios conferred by the allele 1672*T, by the allele -207*C or by the TC minihaplotype were all similar. Both polymorphisms were also in strong linkage disequilibrium in our population (D' = 0.86) [<xref ref-type="bibr" rid="B26">26</xref>]. However, their etiological role could not be verified in Spanish Crohn's patients and they were not described as causative polymorphisms, but as genetic markers of risk or protection haplotypes. In our diabetic population, the allele 1672*T increased disease predisposition [OR = 1.25 (1.01–1.54); p = 0.034] and the 1672*CC genotype showed a protective effect (Table <xref ref-type="table" rid="T1">1</xref>), but no significant differences between patients and controls were found for alleles and genotypes of the OCTN2-promoter (-207G>C) and of the other SLC22A5 intronic variant (Table <xref ref-type="table" rid="T2">2</xref>).</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Distribution of OCTN2 polymorphisms.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center">OCTN2 Promoter</td><td align="center">T1D (%)</td><td align="center">Controls (%)</td></tr></thead><tbody><tr><td align="center">Alleles</td><td align="center">2n = 590</td><td align="center">2n = 1016</td></tr><tr><td align="center">C</td><td align="center">327 (55.4)</td><td align="center">533 (52.5)</td></tr><tr><td align="center">G</td><td align="center">263 (44.6)</td><td align="center">483 (47.5)</td></tr><tr><td align="center">Genotypes</td><td align="center">n = 295</td><td align="center">n = 508</td></tr><tr><td align="center">CC</td><td align="center">91 (30.8)</td><td align="center">150 (29.5)</td></tr><tr><td align="center">CG</td><td align="center">145 (49.2)</td><td align="center">233 (45.9)</td></tr><tr><td align="center">GG</td><td align="center">59 (20.0)</td><td align="center">125 (24.6)</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="center">OCTN2 Intron</td><td align="center">T1D (%)</td><td align="center">Controls (%)</td></tr><tr><td colspan="3"><hr></hr></td></tr><tr><td align="center">Alleles</td><td align="center">2n = 590</td><td align="center">2n = 1016</td></tr><tr><td align="center">A</td><td align="center">370 (62.7)</td><td align="center">636 (62.6)</td></tr><tr><td align="center">G</td><td align="center">220 (37.3)</td><td align="center">380 (37.4)</td></tr><tr><td align="center">Genotypes</td><td align="center">n = 295</td><td align="center">n = 508</td></tr><tr><td align="center">AA</td><td align="center">115 (39.0)</td><td align="center">194 (38.2)</td></tr><tr><td align="center">AG</td><td align="center">142 (48.1)</td><td align="center">248 (48.8)</td></tr><tr><td align="center">GG</td><td align="center">38 (12.9)</td><td align="center">66 (13.0)</td></tr></tbody></table></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>The main haplotypes in region of OCTN1 and OCTN2 genes.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td></td><td></td><td></td><td align="center" colspan="2">T1D (2n = 590)</td><td align="center" colspan="2">Controls (2n = 970)</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td colspan="2"><hr></hr></td><td colspan="2"><hr></hr></td><td></td><td></td></tr><tr><td align="center">Slc2F2</td><td align="center">L503F</td><td align="center">OCTN2 Promoter</td><td align="center">OCTN2 Intron</td><td align="center">Frequency</td><td align="center">Haplotype</td><td align="center">Frequency</td><td align="center">Haplotype</td><td align="center">OR</td><td align="center">p</td></tr></thead><tbody><tr><td align="center">C</td><td align="center">T</td><td align="center">C</td><td align="center">A</td><td align="center">0.47853</td><td align="center">282</td><td align="center">0.42690</td><td align="center">414</td><td align="center">1.23</td><td align="center">0.05</td></tr><tr><td align="center">C</td><td align="center">C</td><td align="center">G</td><td align="center">G</td><td align="center">0.27276</td><td align="center">161</td><td align="center">0.29604</td><td align="center">287</td><td align="center">0.89</td><td align="center">0.33</td></tr><tr><td align="center">C</td><td align="center">C</td><td align="center">G</td><td align="center">A</td><td align="center">0.06522</td><td align="center">38</td><td align="center">0.10018</td><td align="center">97</td><td align="center">0.62</td><td align="center">0.02</td></tr><tr><td align="center">C</td><td align="center">C</td><td align="center">C</td><td align="center">A</td><td align="center">0.07388</td><td align="center">44</td><td align="center">0.08832</td><td align="center">86</td><td align="center">0.83</td><td align="center">0.33</td></tr><tr><td align="center">T</td><td align="center">C</td><td align="center">G</td><td align="center">G</td><td align="center">0.09492</td><td align="center">56</td><td align="center">0.07732</td><td align="center">75</td><td align="center">1.25</td><td align="center">0.22</td></tr></tbody></table><table-wrap-foot><p>Overall comparison between patients and controls: χ<sup>2 </sup>= 10.43; p = 0.034</p></table-wrap-foot></table-wrap><p>We continued our study by analyzing the inferred OCTN1-OCTN2 haplotypes, since haplotypes define better a DNA fragment where a susceptibility or protective gene could be located. Table <xref ref-type="table" rid="T3">3</xref> shows the haplotypes estimated by the Expectation-Maximization (EM) algorithm implemented by the Arlequin software with a frequency over 1%. The overall comparison of haplotypes in a 5 × 2 contingency table rendered a significant difference between patients and healthy controls (χ<sup>2 </sup>= 10.43; p = 0.034), and being a unique omnibus comparison, this result does not need ulterior correction. These results evidenced one protection CCGA [OR = 0.62 (0.41–0.93); p = 0.02] and another risk CTCA [OR = 1.23 (1.00–1.52); p = 0.05] haplotypes, although they did not withstand correction for multiple testing (Table <xref ref-type="table" rid="T3">3</xref>). The effect of the latter is secondary to the protective haplotype, as evidenced when this one is eliminated from the comparison and no significant predisposition effect was observed [OR = 1.16 (0.93–1.44); p = 0.18]; however, when the eliminated haplotipe is CTCA, a significant protection effect is still found for the CCGA haplotype [OR = 0.67 (0.44–1.02); p = 0.047). Moreover, as other haplotypes carrying the 1672*C allele do not show any influence in T1D predisposition, the protective effect of the 1672*C allele (p = 0.034) is due to this CCGA haplotype.</p><p>To complete the study in our population, we analyzed children and adolescents with type 1 diabetes fixing the cut-off age in 15 years old (median age at onset in our cohort). No differences between young patients and either adult patients or controls were found for any isolated variant or for the inferred haplotypes (data not shown). Finally, the frequencies of both, polymorphisms and inferred haplotypes, between male and female T1D patients were similar (data not shown).</p></sec><sec><title>Discussion</title><p>The chromosomal region 5q31 contains several genes involved in immune and inflammatory responses and the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes were associated with two autoimmune diseases (rheumatoid arthritis and Crohn's disease) [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>]. Additionally, several genomewide scans for type 1 diabetes have identified susceptibility loci on different chromosomes, including the region 5q [<xref ref-type="bibr" rid="B27">27</xref>,<xref ref-type="bibr" rid="B28">28</xref>]. Moreover, carnitine has been described to increase sensitivity of cells to insulin [<xref ref-type="bibr" rid="B8">8</xref>], which could hypothetically lend cells more prone to an autoimmune attack. For all these reasons we considered that these genes could be candidates to modify the susceptibility to another autoimmune disease, as type 1 diabetes.</p><p>We tested six polymorphisms in the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes and in five of them no significant independent association with type 1 diabetes could be found (Tables <xref ref-type="table" rid="T1">1</xref> and <xref ref-type="table" rid="T2">2</xref>). The power of this study considering a relative risk of 1.5 was 75% for Slc2F2, 86% for L503F and 99% for OCTN2-promoter and OCTN2-intron, as calculated by an UCLA Department of Statistics sofware [<xref ref-type="bibr" rid="B29">29</xref>]. Therefore, we can exclude the studied <italic>SLC22A5 </italic>markers as causative candidates for T1D in Spanish patients, but in the case of Slc2F2 our results are not conclusive, albeit a recently published study does not find association of Slc2F1 and Slc2F2 <italic>SLC22A4 </italic>polymorphisms within a well-powered T1D cohort [<xref ref-type="bibr" rid="B22">22</xref>], in agreement with our data.</p><p>The overall distribution of the estimated haplotypes was significantly different between patients and healthy controls. Being ours the first study carried out in T1D, replication studies in different T1D populations would be necessary to firmly establish the role of these genes in autoimmune diabetes. We proved by using a stepwise procedure that the association with T1D of the protection haplotype CCGA is primary. The etiological polymorphisms previously found associated with increased susceptibility to RA (slc2F2*T) and to Crohn's disease (L503F*T) do not display a causative role by themselves in our population.</p><p>Low carnitine plasma levels have been found in children and adolescents with type 1 diabetes [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. For this reason, we decided to conclude the study with an age-stratified analysis. No significant differences in the association of each polymorphism were observed when grouping by age at onset or gender. In the study of the inferred haplotypes again no differences were found between any group of patients and when we compared each group with controls. Therefore, it seems that the protective effect of the inferred haplotype is independent of both sex and age at disease onset.</p><p>A previous report about disequilibrium blocks at 5q31 in European-derived population [<xref ref-type="bibr" rid="B30">30</xref>] demonstrated limited haplotype diversity, concordantly with our data (5 haplotypes found out of the 2<sup>4 </sup>= 16 theoretically possible). One of the defined blocks (92 kb long) corresponds to the carnitine transporter genes locus. These authors indicated that the chromosomal region 5q31 is divided into discrete blocks displaying complete linkage disequilibrium (LD). However, some degree of LD extending beyond the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes could imply that most probably these genes, or other in LD with them, are responsible for the reported effect on T1D risk.</p></sec><sec><title>Conclusion</title><p>In conclusion, the overall comparison of haplotypes within the chromosomal region where the carnitine transporter genes map seems to be different between type 1 diabetes patients and healthy controls in our Spanish population. The reduced carnitine plasma levels found in both type 1 and type 2 diabetes patients [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>] could be explained by a mere increase in the activity of the carnitine transporters to supply the higher energetic cellular demand mainly provided by lipid metabolism in young diabetes patients. However, our results with the 1672*C allele and with the inferred haplotypes, the effect of carnitine sensitizing cells to insulin and potentially rendering them more amenable to an immune attack, and also the LD-block data described by Daly [<xref ref-type="bibr" rid="B30">30</xref>], support the role of the OCTN genes in T1D risk. Additional studies in independent populations and in both type 1 and type 2 diabetes patients will be needed to confirm the putative influence of the <italic>SLC22A4 </italic>and <italic>SLC22A5 </italic>genes in these diseases.</p></sec><sec><title>Abbreviations</title><p>LD Linkage Disequilibrium</p><p>MHC Major Histocompatibility Complex</p><p>OCTN Organic Cation Transporter</p><p>SNP Single Nucleotide Polymorphism</p><p>SLC22 Solute Carrier Family 22</p><p>T1D Type 1 Diabetes</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>JLS carried out the genotyping of the patients and a great part of the controls, participated in the statistical analysis and drafted the manuscript.</p><p>AM carried out a part of the genotyping of control samples and participated in the statistical analysis.</p><p>HdlC made the diagnosis and collaborated in collection of samples.</p><p>MAF participated in the genotyping and collection of samples.</p><p>MFA participated in the coordination of the study and helped to collect the DNA samples.</p><p>EgdlC coordinated the study and critically revised the manuscript.</p><p>EU conceived of the study, participated in the statistical analysis and wrote the major part of the manuscript.</p><p>All authors read and approve the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2350/7/54/prepub"/></p></sec>
|
Scrapie infectivity is quickly cleared in tissues of orally-infected farmed fish
|
<sec><title>Background</title><p>Scrapie and bovine spongiform encephalopathy (BSE) belongs to the group of animal transmissible spongiform encephalopathy (TSE). BSE epidemic in the UK and elsewhere in Europe has been linked to the use of bovine meat and bone meals (MBM) in the feeding of cattle. There is concern that pigs, poultry and fish bred for human consumption and fed with infected MBM would eventually develop BSE or carry residual infectivity without disease. Although there has been no evidence of infection in these species, experimental data on the susceptibility to the BSE agent of farm animals other than sheep and cow are limited only to pigs and domestic chicken. In the framework of a EU-granted project we have challenged two species of fish largely used in human food consumption, rainbow trout (<italic>Oncorhynchus mykiss</italic>) and turbot (<italic>Scophthalmus maximus</italic>), with a mouse-adapted TSE strain (scrapie 139A), to assess the risk related to oral consumption of TSE contaminated food. In trout, we also checked the "in vitro" ability of the pathological isoform of the mouse prion protein (PrP<sup>Sc</sup>) to cross the intestinal epithelium when added to the mucosal side of everted intestine.</p></sec><sec><title>Results</title><p>Fish challenged with a large amount of scrapie mouse brain homogenate by either oral or parenteral routes, showed the ability to clear the majority of infectivity load. None of the fish tissues taken at different time points after oral or parenteral inoculation was able to provoke scrapie disease after intracerebral inoculation in recipient mice. However, a few recipient mice were positive for PrP<sup>Sc </sup>and spongiform lesions in the brain. We also showed a specific binding of PrP<sup>Sc </sup>to the mucosal side of fish intestine in the absence of an active uptake of the prion protein through the intestinal wall.</p></sec><sec><title>Conclusion</title><p>These results indicate that scrapie 139A, and possibly BSE, is quickly removed from fish tissues despite evidence of a prion like protein in fish and of a specific binding of PrP<sup>Sc </sup>to the mucosal side of fish intestine.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Ingrosso</surname><given-names>Loredana</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Novoa</surname><given-names>Beatriz</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Valle</surname><given-names>Andrea Z Dalla</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Cardone</surname><given-names>Franco</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Aranguren</surname><given-names>Raquel</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Sbriccoli</surname><given-names>Marco</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Bevivino</surname><given-names>Simona</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Iriti</surname><given-names>Marcello</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Liu</surname><given-names>Quanguo</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A10" contrib-type="author"><name><surname>Vetrugno</surname><given-names>Vito</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A11" contrib-type="author"><name><surname>Lu</surname><given-names>Mei</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A12" contrib-type="author"><name><surname>Faoro</surname><given-names>Franco</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A13" contrib-type="author"><name><surname>Ciappellano</surname><given-names>Salvatore</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A14" contrib-type="author"><name><surname>Figueras</surname><given-names>Antonio</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A15" corresp="yes" contrib-type="author"><name><surname>Pocchiari</surname><given-names>Maurizio</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Veterinary Research
|
<sec><title>Background</title><p>Transmissible spongiform encephalopathy (TSE) or prion diseases are fatal human and animal neurological disorders with a worldwide distribution. Human TSE diseases include sporadic, genetic, iatrogenic and variant Creutzfeldt-Jakob disease (CJD), Gerstmann-Sträussler-Scheinker disease and sporadic or familial fatal insomnia. Animal counterparts are scrapie in sheep and goats, bovine spongiform encephalopathy (BSE), transmissible mink encephalopathy, and chronic wasting disease of mule deer and elk. There are strong evidences that among human TSE diseases only variant CJD is caused by the consumption of BSE-contaminated meat products [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The critical pathogenetic event in TSE diseases is the conformational change of the physiological host prion protein (PrP<sup>c</sup>) into an insoluble form (PrP<sup>Sc</sup>) able to provoke the pathognomonic brain lesions and death. Transgenic mice devoid of PrP<sup>c </sup>are unable to sustain TSE infection after experimental inoculation demonstrating the key role of PrP<sup>c </sup>in the pathogenesis of these diseases [<xref ref-type="bibr" rid="B3">3</xref>].</p><p>The PrP gene is highly preserved among mammals [<xref ref-type="bibr" rid="B4">4</xref>], and sequences of prion-like cDNAs have been described in other vertebrate classes including birds [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>], reptiles [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>], amphibians [<xref ref-type="bibr" rid="B10">10</xref>], and fish [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>]. The presence of proteins "similar to" PrP<sup>c </sup>(stPrP, [<xref ref-type="bibr" rid="B12">12</xref>]) in fish has raised concern about a possible transmission of TSE agents to humans through consumption of farmed fish since mammalian MBM (meat and bone meal) and other mammalian products were historically fed to farmed fish [<xref ref-type="bibr" rid="B14">14</xref>]. The distribution of stPrP in trout organism was also studied through the use of newly described monoclonal antibodies which show that the protein is preferentially distributed in brain, optic nerve and spinal cord in contrast to its absence (or presence at undetectable level) outside the nervous system, including the intestine [<xref ref-type="bibr" rid="B15">15</xref>].</p><p>The passage of TSE agents between animals of different species is usually impaired by what is called the species barrier, i.e. the difficulty to establish clinical disease into the new host even after a prolonged incubation period. Infectivity, however, might be present without clinical presentation of disease, and tissues from first attempted transmission might be infectious when re-inoculated in susceptible animals [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>The need to give an answer to public concern about safety of food possibly contaminated with TSE agents prompted us to set up an experiment that uses fish as recipient of a scrapie agent (mouse-adapted 139A strain). Both "in vitro" and "in vivo" approaches were devised in an attempt to draw a pattern of risk related to human consumption of fish products. The 139A mouse-adapted TSE scrapie strain was chosen because of its ability to cross the species barrier in different species of rodents [<xref ref-type="bibr" rid="B17">17</xref>], while trout (<italic>Oncorhynchus mykiss</italic>) and turbot (<italic>Scophthalmus maximus</italic>) for their large use in aquaculture food industry. We also challenged the "in vitro" ability of PrP<sup>Sc </sup>to cross the intestinal tissues of rainbow trout when 139A was added to the mucosal side of everted or statically perfused intestine.</p></sec><sec><title>Results and discussion</title><p>Our major aim was to investigate whether tissues of orally infected fish would retain some residual infectivity (i.e., not due to replication of TSE agents). Thus, turbot and trout were forced-fed with the 139A strain of scrapie and sacrificed from 1 to 90 days after challenge. Muscle, intestine, and brain were taken at different time points (1, 15, 30, 60 and 90 days post inoculation) and inoculated into recipient mice for the measurement of residual infectivity (Table <xref ref-type="table" rid="T1">1A</xref>).</p><p>No behavioral or swimming abnormalities were observed in orally inoculated trout and turbot and histological examination of tissues from inoculated fish did not show any pathological mark nor prion protein was detected by immunohistochemical examination (data not shown). None of the recipient mice developed scrapie disease. No PrP<sup>Sc </sup>was detected in the brain of the mice injected with turbot tissues (Table <xref ref-type="table" rid="T1">1A</xref>). However, PrP<sup>Sc </sup>was detected in the brain of one from the 8 mice inoculated with the intestine of trout taken one day after force-feeding inoculation, suggesting that the trout intestine contained some residual infectivity.</p><p>We therefore investigated whether trout intestine binds PrP<sup>Sc </sup>and transfers it to the serosal side where it may spread to the lymphoid tissues and, eventually into the CNS. Using everted trout intestine immersed in a solution containing 139A we observed that PrP<sup>Sc </sup>slightly absorbs to the mucosal intestinal layer as shown by the low, yet detectable signal at the western blot (Figure <xref ref-type="fig" rid="F1">1C</xref>). When statically perfused intestine was used in order to study the possible role of pyloric caeca in PrP<sup>Sc </sup>absorption, it was possible to detect the presence of PrP<sup>Sc </sup>by immunohistochemistry in the <italic>stratum compactum </italic>both in the trout intestine and in the pyloric caeca (Figure <xref ref-type="fig" rid="F2">2c, 2f</xref>). On the other hand, the absence of signal for PrP<sup>Sc </sup>at the western blot in the solution fluxed into the serosal side of the everted intestine (Figure <xref ref-type="fig" rid="F1">1B</xref>) excludes, at least in this experimental setting, an active secretion of PrP<sup>Sc </sup>from one side to the other side of the intestinal tract.</p><p>Such data are in agreement with the recent finding that stPrP is not detectable in the intestine of Rainbow trout [<xref ref-type="bibr" rid="B15">15</xref>], which renders unlikely an active uptake of exogenous PrP<sup>Sc</sup>, but does not exclude an unspecific binding of PrP<sup>Sc </sup>to the fish intestinal mucosa.</p><p>In a second experiment, we inoculated trout and turbot with the same strain of scrapie as above by parenteral routes to increase the possibility of a successful infection (Table <xref ref-type="table" rid="T1">1B</xref>). Brain and spleen, as the most likely tissues for PrP<sup>Sc </sup>accumulation, were taken 15 and 90 days (the longest survival time of fish in a close circuit water tanks) after inoculation and residual infectivity was bio-assayed in mice. None of the recipient mice developed scrapie disease during their lifespan. However, PrP<sup>Sc </sup>was observed at the western blot in the brain of a few recipient mice inoculated with tissues taken from turbot or trout (Table <xref ref-type="table" rid="T1">1B</xref>). These data show that 15 days after inoculation only about one LD<sub>50 </sub>of scrapie-infectivity is present in the spleen of turbot, and much less in the brain of turbot or in the spleen of trout. Almost no infectivity (only 1 in 37 recipient mice were PrP<sup>Sc </sup>positive) was however detected 90 days after inoculation of fish tissues, suggesting that the infectivity measured at 15 days post inoculation was likely related to the residual inoculum. As the fishes were not kept more than 90 days, it is not possible to exclude that infectivity might develop at a later time point, after an initial clearance phase of the inoculum, though this possibility is not likely.</p><p>In our work, we did not evaluate whether mammalian TSE agents may establish infection in fish. Nonetheless, the "lesion profile" curves based upon the severity of spongiform changes in brain areas of mice injected with 139A before and after passage in fish (Figure <xref ref-type="fig" rid="F3">3</xref>), show a similar pattern suggesting that the 139A strain did not adapt to fish in the 3 months period examined. A statistical comparison of scores between the two curves, done area by area by the Mann-Whitney's test, showed a significant difference only in the dorsal medulla area (p = 0.045). No statistically significant difference was observed in the superior colliculus, hypothalamus, cingulate, and adjacent cortices. Permutation tests performed to overcome problems due to the small sample size, confirmed this pattern. The overall lower scores observed in the brain of mice inoculated with fish tissues could merely represent the fact that these mice were not symptomatic when sacrificed while control scrapie infected mice showed evident clinical signs of disease.</p></sec><sec><title>Conclusion</title><p>These data show that about 4 million LD<sub>50 </sub>of 139A given by forced feeding were readily removed from fish intestine (both trout and turbot) in the first 24 hours after infection and that infection never reached the brain, the spleen or the muscles. This suggests that scrapie is quickly removed from fish tissues despite the presence of a cellular prion-like protein [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B18">18</xref>] and a prion protein-like gene in fish [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>]. With all the cautions due to the difference between the 139A and the BSE strains, and that in this experiment fishes were observed for no more than 90 days after infection, it is tentatively possible to assume that the consumption of fish fed with BSE-infected MBM should not pose any substantial threat to public health.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>"In vitro" experiments</title><p>Rainbow trout (<italic>Oncorhynchus mykiss</italic>) from 50 to 200 g of body weight were kept at 15°C for one week and starved 24 hours before the experiments. After sacrifice with a lethal dose of MS 222 (200 mg/l) the whole intestine was extracted, carefully everted, and canulated according to published procedures [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Briefly, canulated intestine was submerged in 25 ml of physiological solution (145 mM NaCl, 5 mM KCl, 2 mM MgSO<sub>4</sub>·7H<sub>2</sub>O, 7 mM CaCl<sub>2</sub>·2H<sub>2</sub>O, 10 mM NaHCO<sub>3</sub>, 5 mM D-Glucose, pH 7.2) containing scrapie mice brain homogenate (139A) to a final concentration of 5 mg/ml under constant agitation by air pump (sample A). The same solution, but without infected brain material (sample B), was fluxed through the canula inside the serosal side of the intestine by a peristaltic pump (1 ml/min) and collected for 1 h. After perfusion, the fish intestinal mucosa was abraded with a glass blade, homogenized in PBS (1:4 vol/vol), and further analysed for PrP<sup>Sc </sup>detection by Western blot (sample C). Aliquots of 30 μl from samples A, B (sample B was lyophilised and resuspended in 1% wt/vol sarcosyl solution), and C were processed for PrP<sup>Sc </sup>detection [<xref ref-type="bibr" rid="B21">21</xref>], using the mouse monoclonal antibody SAF83 (1:10000). This antibody was selected after screening different antibodies (SAF70, SAF32, 6H4, 8G8, SAF83, SAF84) for their ability to detect PrP<sup>Sc </sup>on the fish mucosa homogenate. SAF70 and SAF84 showed cross reactivity with the fish mucosa, and 6H4 and 8G8 showed very low ability to detect PrP<sup>Sc</sup>. On the other hand, SAF32 and SAF83 were equally able to detect PrP<sup>Sc </sup>with no cross reactivity with the fish mucosa.</p><p>Trout gastrointestinal tract, from oesophagus to anus, was submerged in sample A (see above) to be statically perfused. Samples were carefully washed and dissected in 10 mm segments. Cross sections, 1 mm thick, were randomly excised from each segment of the gastrointestinal tract, fixed in 0.8% glutharaldehyde, 3.0% formaldehyde in 0.01 phosphate buffer saline (PBS), dehydrated, embedded in hydroxyethyl-methacrylate (Kulzer Histo-Technik 8100, Germany) and serially cut. For immunostaining, sections were subjected to antigen retrieval (98% formic acid for 15 min or alternatively to microwave pretreatment at 600 W for 60 sec in PBS), before incubation with the primary antibody SAF83 at different dilutions. After incubation with a goat-anti mouse secondary antibody conjugated with 6 nm colloidal gold, the slides were rinsed and left to dry for one day before silver staining. Silver enhancement was performed with R-GENT kit for light microscopy (Aurion, The Netherlands). All samples were counterstained with 0.1% toluidine blue observed with epifluorescence (excitation 520–550 nm).</p></sec><sec><title>"In vivo" experiment</title><p>Trout (<italic>Oncorhynchus mykiss</italic>) and turbots (<italic>Scophthalmus maximus</italic>) of 8 cm length were obtained from a commercial farm and acclimatized to laboratory conditions for 3 weeks before oral or parenteral infection with the mouse-adapted 139A strain of scrapie. The inoculum was prepared from a pool of 70 g of 139A scrapie-infected mouse (C57/BL) brains. Brain tissue was homogenized in sterile PBS to obtain a 10% (wt/vol) suspension, centrifuged at 913 × g (4°C per 15 minutes), and supernatant collected. Infectivity titre of this suspension was calculated by end-point titration as previously described [<xref ref-type="bibr" rid="B22">22</xref>], using groups of 6–10 animals for each dilution. The amount of infectivity, calculated by the Reed and Muench method, was 10<sup>8.9 </sup>LD<sub>50 </sub>per gram of tissue. Control homogenate was prepared with the same procedure using brain tissue from healthy mice. One group of turbots (n = 20) or trout (n = 20) was force-fed with 0.05 ml of 139A scrapie-infected (10<sup>6.6 </sup>LD<sub>50</sub>) or control mouse brain homogenates. Another group of turbots (n = 30) or trout (n = 30) was simultaneously inoculated by the intracerebral (0.03 ml, 10<sup>6.4 </sup>LD<sub>50</sub>), intraperitoneal (0.1 ml, 10<sup>6.9 </sup>LD<sub>50</sub>), and intramuscular (0.1 ml, 10<sup>6.9 </sup>LD<sub>50</sub>) routes with either 139A or control brain homogenates. After inoculation, fish were maintained in close circuit water tanks under bio-security conditions to avoid water contamination with the scrapie agent.</p><p>Samples from brain, muscle, and intestine were taken from 3 orally-infected and 1 control fish for each species at 1, 15, 30, 60 and 90 days post-infection. Moreover, brain and spleen were taken from parenterally infected fish at 15 and 90 days post-infection. Collected samples were divided in two parts: one was immediately frozen at -80°C for measuring scrapie infectivity by mouse bioassay, the other was fixed for histological or immunohistochemical examination. Histopathological examination of fish tissues was routinely performed on every sample collected. Briefly, samples were fixed in Davidson and processed in an automatic tissue processor, embedded in paraffin and cut in 5 μm sections, deparaffinized, rehydratated and then stained with haematoxylin-eosin. The immunohistochemical detection of the prion protein in fish tissue sections was performed as described [<xref ref-type="bibr" rid="B23">23</xref>].</p><p>For the bioassay of fish tissues in mice, each group of fish samples, belonging to the same time point, was homogenized using separate sets of instruments to avoid accidental cross-contamination. Homogenisers were then accurately decontaminated (2 cycles of autoclave at 134°C for 1 hour followed by extensive washing and a third cycle of autoclave at 120°C for 20 minutes) before re-use on another group of samples. Twenty μl of 10% tissue homogenate were injected intracerebrally in groups of 6–11 C57/BL male mice for infectivity bioassay. Animals showing clinical signs (not necessarily related to scrapie) or moribund were sacrificed, their brains removed and divided in two halves. One half was immediately frozen at -70°C for PrP<sup>Sc </sup>analysis, the other half was fixed in formalin for histological examination. PrP<sup>Sc </sup>was purified as described [<xref ref-type="bibr" rid="B24">24</xref>] and detected by western blotting using monoclonal antibody SAF84 (SpiBio) diluted 1:5000. Mice were housed at the animal facility of the Istituto Superiore di Sanità (ISS) under the supervision of the Service for Biotechnology and Animal Welfare of the ISS who warrants the adherence to the national and international regulations on animal welfare.</p></sec><sec><title>Histological analysis and lesion profiles of mice brains</title><p>Formalin-fixed brains were decontaminated with 98% formic acid for 1 h, stained with haematoxylin-eosin and scored for the intensity of vacuolar degeneration (ranging from 0, no vacuolar change, to 5, severe vacuolar change) in nine standard grey matter areas for building up the 'lesion profiles' [<xref ref-type="bibr" rid="B25">25</xref>]. Sections were coded and blind examined. In Figure <xref ref-type="fig" rid="F3">3</xref>, the lesion profile of 139A after passage in fish was a mean of 4 animals when brains for histology were available (2 mice inoculated with turbot spleen at 15 days, 1 mouse inoculated with trout spleen at 15 days, and 1 turbot brain taken 90 days after parenteral inoculation). We pooled these data together because none of these mice showed clinical signs and were sacrificed at about the same time after inoculation (221, 204, 215, and 221 days, respectively). The reference curve of 139A not passaged in fish was obtained pooling data from 6 mice. Immunohistochemistry was performed as described [<xref ref-type="bibr" rid="B23">23</xref>].</p></sec></sec><sec><title>Abbreviations</title><p>PrP: prion protein</p><p>PrP<sup>c</sup>: cellular prion protein</p><p>PrP<sup>Sc</sup>: pathological prion protein</p><p>stPrP: "similar to" PrP</p><p>TSE: Transmissible Spongiform Encephalopathy</p><p>CJD: Creutzfeldt-Jakob disease</p><p>GSS: Gerstmann-Sträussler-Scheinker disease</p><p>BSE: bovine Spongiform encephalopathy</p><p>MBM: bovine meat and bone meals</p><p>anti-HA: anti-haemagglutinin</p></sec><sec><title>Authors' contributions</title><p>LI contributed to the design of the study, did mice inoculation, analysis of data and drafted the manuscript. BN, RA established the protocol for fish infection and sampling of fish tissues, they also performed it. AZDV, FF, and MI contributed to the "in vitro" experimental work and to the protocols setting. MS did the neuropathology of the mice brains and contributed to analysis of data. SB, ML, QL, and VV did western blot analysis of mouse brains and contributed to monitoring of mice, collection and analysis of data. FC did coordination and collection of clinical, biochemical and histological data from the mouse bioassay. SC contributed to the design of the study set up the protocol for the "in vitro" experiments. AF contributed to the design of the "in vivo " experimental work and the histological studies on fish tissues. MP participated in the design and coordination of the study, interpretation of data and drafting the manuscript. All the authors read and approved the final manuscript.</p></sec>
|
Invasive aspergillosis causing small bowel infarction in a patient of carcinoma breast undergoing chemotherapy
|
<sec><title>Background</title><p>To report a 45 year old lady presenting with proximal jejunal gangrene due to invasive Aspergillosis. The patient was undergoing adjuvant chemotherapy for advance carcinoma of breast (Stage IV).</p></sec><sec sec-type="methods"><title>Methods</title><p>The patient was referred to our surgical emergency for acute abdominal symptoms for 6 hours. Histopathology revealed bowel wall necrosis and vascular invasion by Aspergillus Fumigatus. Postoperative recovery was uneventful and the patient received Amphotericin-B (1 mg/kg/day) for invasive aspergillosis. Invasive pulmonary aspergillosis was confirmed by isolating Aspergillus Fumigatus from bronchoalveolar lavage and by a positive circulating galactomannan test (ELISA Assay).</p></sec><sec><title>Results</title><p>Detailed history revealed dry cough and two episodes of haemoptesis for 2 weeks. Haemogram and counts revealed anemia and neutropenia. Plain X – ray of the abdomen showed multiple air fluid levels and ultrasound of the abdomen revealed distended bowel loops. On exploration small bowel was found to be gangrenous. The patient was successfully managed by supportive treatment and conventional intravenous Amphotericin-B for 2 weeks. The lady was discharged one week after completion of antifungal therapy and one month later she underwent toilet mastectomy. The lady came to follow up for 1 year and she is currently under hormone therapy.</p></sec><sec><title>Conclusion</title><p>With the emergence of new and powerful immunosuppressive, anticancer drugs and potent antibiotics the survival of transplant and critically ill patients has remarkably increased but it has shown a significant rise in the incidence of invasive opportunistic fungal infections. We conclude hat the diagnosis of invasive gastrointestinal aspergillosis may be considered in a neutropenic patient with acute abdominal symptoms.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Chaudhary</surname><given-names>Amit</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Jain</surname><given-names>Vinod</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Dwivedi</surname><given-names>Rama S</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Misra</surname><given-names>Samir</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
Journal of Carcinogenesis
|
<sec><title>Background</title><p>Despite the use of new and effective drugs, the disseminated invasive aspergillosis often remains lethal in neutropenic patients [<xref ref-type="bibr" rid="B1">1</xref>]. There is a great need to create high index of clinical investigation and develop better diagnostic tools to detect invasive aspergillosis before it causes irreversible damage. In the present study we report a 45 year old lady with severe neutropenia (400/cumm) presenting with acute abdomen. Invasive gastrointestinal aspergillosis leading to widespread proximal bowel infarction was found on exploration. The diagnosis was established by isolating Aspergillus fumigatus from bronchoalveolar lavage and by a positive circulating galactomannan test (ELISA). The patient received conventional intravenous amphotericin-B for 2 weeks.</p></sec><sec sec-type="methods"><title>Methods</title><p>A 45 year old lady admitted in intensive care unit was referred to our surgical emergency unit for severe acute pain and distention of the abdomen for 6 hours. She was undergoing chemotherapy in clinical oncology department for locally advance carcinoma (Stage IV) of breast and four days back she was shifted to the intensive care unit when she developed severe neutropenia after fourth cycle CAF regimen (Cyclophosphamide 500 mg/m<sup>2</sup>, Adriamycin 50 mg/m<sup>2 </sup>and 5-FU 500 mg/m<sup>2</sup>). There was no metastasis. Detailed history revealed that the patient had developed dry cough and fever 7 days back. The patient had absolute constipation. On examination the abdomen was distended and there was guarding and rebound tenderness. Hematocrit and counts revealed anemia (Hb 7 gm%) and neutropenia (400/cumm).</p><p>Plain radiograph of chest showed patchy infiltrates in both the lung fields suggestive of pneumonitis and X-ray abdomen showed distended bowel loops and multiple air fluid levels. There was no gas under the diaphragm. Ultrasound of the abdomen could not add much to the findings of plain X-ray and showed distended bowel loops and minimal interbowel fluid. Emergency exploratory laprotomy was performed under general anesthesia. About 30 cm of proximal jejunum 15 cm distal to duodeno-jejunal junction was found to be gangrenous. Primary resection and anastomosis was done and the specimen (resected bowel segment with adjoining mesentery) was sent for histopathology. After surgery the patient was shifted to postoperative ICU where she developed fever which was not responding to any antibiotic.</p><p>On the basis of preoperative X-ray chest, neutropenia and high index of clinical suspicion broncho-alveolar lavage was performed which revealed pulmonary aspergillosis. Serum galactomannan test (ELISA) was also positive for Aspergillus. The diagnosis of invasive pulmonary aspergillosis was made and the patient was put on conventional intravenous Amphotericin B (1 mg/kg/day) for 2 weeks. Histopathology of the specimen was available on fourth postoperative day and revealed transmural ischemic necrosis of the bowel and widespread invasion and blockade of arterioles due to Aspergillus fungal forms (Fig. <xref ref-type="fig" rid="F1">1</xref>). The diagnosis of invasive gastrointestinal aspergillosis secondary to pulmonary Aspergillosis was made and antifungal treatment was continued. The patient responded to intravenous antifungal treatment and was discharged after 2 week of antifungal treatment. The lady underwent toilet mastectomy and was under follow up for 1 year. Postoperative chemotherapy was also given after improvements in the counts and presently she is under hormone therapy.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Histopathology analysis demonstrates vascular invasion by disseminated aspergillosis. (Arrows showing fungal forms).</p></caption><graphic xlink:href="1477-3163-5-18-1"/></fig></sec><sec><title>Results and discussion</title><p>In our patient the diagnosis was made by exclusion and the decision for bronchi-alveolar lavage was taken which came out positive for aspergillosis. This evidence along with positive galactomannan test was enough to support the diagnosis of invasive Aspergillosis and intravenous Amphotericin B was started. The patient showed dramatic response and fever subsided on the next day. After recovery patient underwent modified radical mastectomy and postoperative chemotherapy and hormone therapy were also given. The patient was alive and well at the end of 1 year follow up.</p><p>Invasive gastrointestinal aspergillosis is an opportunistic infection that characteristically affects the immunocompromised host, resulting in high degree of morbidity and mortality [<xref ref-type="bibr" rid="B2">2</xref>]. The incidence of invasive aspergillosis has risen considerably during the past decade due to more intensive chemotherapy, organ transplantation, intensive care, and aggressive surgical interventions [<xref ref-type="bibr" rid="B3">3</xref>]. Extrapulmonary aspergillosis is rare but can involve gastrointestinal tract, heart, kidney, central nervous system, liver and pancreas etc [<xref ref-type="bibr" rid="B4">4</xref>]. Almost one-quarter of those with invasive pulmonary aspergillosis develop disseminated infection with Aspergillus sp. via hematogenous spread [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. Among these patients the reported incidence of gastrointestinal involvement has been 41 – 47% [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><p>Patients with bowel involvement are rarely diagnosed prior to autopsy because the typical symptoms of abdominal pain, obstruction and gastrointestinal bleeding are usually overshadowed by the septic manifestation of systemic infection. Wingrad et al [<xref ref-type="bibr" rid="B7">7</xref>] documented a case of aspergillosis as an unusual cause of perforated appendicitis in 1982. In our patient gastrointestinal invasion was secondary to hematogenous spread of fungal forms leading to ischemic necrosis of bowel. The diagnosis was made on the basis of high clinical suspicion followed by bronchoalveolar lavage and positive serum galactomannan test. This diagnosis should be considered when a neutropenic patients presents with abdominal pain and distention with fever.</p><p>There has been an increased understanding of the immunology of Aspergillus infection, paving the way to novel immune augmentation strategies in animal models that merit evaluation in phase I clinical trials [<xref ref-type="bibr" rid="B8">8</xref>]. Chest CT scan [<xref ref-type="bibr" rid="B8">8</xref>] and new techniques like PCR assays based on real time technologies that are able to quantify Aspergillus DNA are promising [<xref ref-type="bibr" rid="B9">9</xref>]. Considerable progress has been made in understanding the genetics of Aspergillus Fumigates and molecular techniques for manipulation of the fungus have been developed but still invasive Aspergillus's is associated with a high mortality rate ranging from 30% to 90% [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>Repeated serum assays to detect Aspergillus antigenemia and meticulous search for other aspergillosis localizations like pulmonary, should continue until the diagnosis allows administration of early antifungal therapy. Though in our patient we used conventional antifungal treatment, combined antifungal therapy or treatment with amphotericin-B at high doses significantly reduces mortality as compared with the conventional treatment [<xref ref-type="bibr" rid="B11">11</xref>].</p></sec><sec><title>Conclusion</title><p>Even with more effective and safer therapies, substantial progress in quelling aspergillosis may not occur until the early diagnostic tools are developed and incorporated into strategies to start earlier treatment. We suggest that a very high index of suspicion of invasive gastrointestinal aspergillosis may be maintained in immunocompromised patients presenting with acute abdominal symptoms.</p></sec>
|
Are the opportunities to prevent alcohol related liver deaths in the UK in primary or secondary care? A retrospective clinical review and prospective interview study
|
<sec><title>Background</title><p>Deaths from liver cirrhosis have increased at least 8 fold since the 1970's in the UK and further increases are anticipated, whereas in the rest of Europe liver deaths are decreasing. In the UK, we urgently need strategies to detect those who misuse alcohol and are at risk of developing alcoholic liver disease before they get to that point. One potential strategy is to screen admissions to hospital with alcohol related conditions for evidence of alcohol misuse.</p><p>Surprisingly, there has been no research into the important question of where the opportunities are to detect those who misuse alcohol – primary or secondary care. We attempted to answer this firstly by conducting a retrospective analysis of the medical notes of 94 patients diagnosed with alcohol induced liver cirrhosis between 1<sup>st </sup>January 1995 and 31<sup>st </sup>December 2000 at Southampton General Hospital with the purpose of identifying admissions to hospital prior to a diagnosis of alcoholic liver disease. In the second part of the study, we interviewed patients with alcoholic liver disease about their contact with health services.</p></sec><sec><title>Results</title><p>Before diagnosis of alcoholic liver disease, 33% (31/94) of the patients had had an admission to hospital for an alcohol related condition. There was a mean of 7 years and 1 month (SD 6 years 3 months) between the first alcohol-related admission and presentation with alcoholic liver disease (in those who had had admissions). The commonest reason for alcohol related admission was falls/fractures/injuries, followed by non-variceal gastro-intestinal bleeds. Patients with alcoholic liver disease who were interviewed had seen their General Practitioner on average at least 2 times per year.</p></sec><sec><title>Conclusion</title><p>Most patients who develop alcohol-induced cirrhosis do not have an admission to hospital with an alcohol related condition before developing alcoholic liver disease. Therefore, if we screen patients admitted to hospital with alcohol related conditions for evidence of alcohol misuse, we could potentially detect around a third of those at risk of developing cirrhosis. Although secondary care has an important role to play in detecting those at risk, the main opportunity for detection is in primary care.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Verrill</surname><given-names>Clare</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Smith</surname><given-names>Stewart</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Sheron</surname><given-names>Nick</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
Substance Abuse Treatment, Prevention, and Policy
|
<sec><title>Background</title><p>Deaths from liver cirrhosis have increased at least 8 fold since the 1970's and further increases are anticipated whereas in the rest of Europe liver deaths are decreasing [<xref ref-type="bibr" rid="B1">1</xref>]. Rates of cirrhosis are directly influenced by alcohol consumption in a population and alcohol consumption doubled in the UK between 1960 and 2002 [<xref ref-type="bibr" rid="B2">2</xref>]. The UK Government Alcohol Strategy reported that alcohol causes at least 22,000 deaths and costs the UK £15 billion every year [<xref ref-type="bibr" rid="B3">3</xref>]. Data from the General Household Survey shows that 27% of UK men and 17% of women drink more than the government recommended maximum of 21 and 14 units/week [<xref ref-type="bibr" rid="B4">4</xref>]. Of these, 5% of men and 3% of women drink more than 50 or 35 units/week – the level where liver disease starts to become a major health risk. Many of these individuals will develop significant liver disease (alcoholic hepatitis/cirrhosis) – a clinical process which is usually entirely silent with no signs or symptoms [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>] and the majority of these patients will not have established alcohol dependency [<xref ref-type="bibr" rid="B7">7</xref>]. Of those that present with a sudden variceal haemorrhage or decompensated liver disease with ascites, one third will die within a month, a further third will die within a few months and only one third will survive long term [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>], figures which have not improved in the UK over the last 35 years [<xref ref-type="bibr" rid="B8">8</xref>]. With such high mortality figures when patients present for the first time with cirrhosis, the emphasis must be on detection and treatment of these individuals before they reach this point.</p><p>The Royal College of Physicians published a report in 2001 entitled 'Alcohol – Can the NHS (National Health Service) Afford it' [<xref ref-type="bibr" rid="B10">10</xref>] suggesting various strategies to combat the UK's growing alcohol problem. It recommended screening those who are admitted to hospital with an alcohol related condition, for example, head injuries or drug overdose for evidence of alcohol misuse. The best way of screening was thought to be some kind of questionnaire, for example, AUDIT (Alcohol Use Disorders Identification Test) which identifies hazardous drinkers. It also suggested that it may even be necessary to screen all admissions to hospital, whatever the reason. The UK Government subsequently published an 'Alcohol Harm Reduction Strategy for England' in 2004 [<xref ref-type="bibr" rid="B11">11</xref>] which was criticised for failing to address the health issues.</p><p>Brief interventions are an effective way to detect and treat those with hazardous drinking patterns and are usually used in primary care, although can be used in other settings, for example, Accident and Emergency Departments. They involve 5–10 minutes of advice given on an opportunistic basis when patients attend for other problems. Brief intervention works in half of subjects [<xref ref-type="bibr" rid="B12">12</xref>] and when liver disease is diagnosed at least half stop drinking. 90% of people visit their General Practitioner (GP) in a 5 year period [<xref ref-type="bibr" rid="B13">13</xref>] meaning that primary care potentially plays a pivotal role in detection and treatment of alcohol misusers.</p><p>So where are the opportunities to detect hazardous and harmful alcohol drinkers, in primary (General Practice) or secondary (hospital) care? This is the question we attempted to answer in this study firstly using a retrospective clinical review of patients with alcohol-induced liver cirrhosis and secondly by interviewing patients with alcoholic liver disease about their contact with primary and secondary care before being diagnosed with alcoholic liver disease. Surprisingly, we have not been able to find any previous studies addressing this very important question.</p><p>In this study we show that most patients who develop alcohol-induced liver cirrhosis do not have an admission to hospital for an alcohol related condition before developing alcoholic liver disease, but that they do have contact with primary care.</p></sec><sec sec-type="methods"><title>Methods</title><p>Retrospective analysis of hospital records and a prospective interview study in patients presenting with alcohol related liver disease. Ethical approval was obtained for both parts from the local ethics committee (Part 1 – Isle of Wight, Portsmouth and South East Hampshire Local Ethics Committee ref no 06/Q1701/19, Part 2 – Southampton and South West Hampshire Ethics Committee (B) ref No 400/02/S/W) and conformed to the principles embodied in the Declaration of Helsinki.</p><sec><title>1) Retrospective clinical review of patients with alcohol-induced liver cirrhosis</title><p>All patients who had a liver biopsy between 1<sup>st </sup>January 1995 and 31<sup>st </sup>December 2000 at Southampton General Hospital were identified from the pathology department computer system. The reports of these biopsies were reviewed and any containing a diagnosis of cirrhosis (including incipient or imminent cirrhosis) and a histological picture suggestive of alcohol were flagged up for the study (n = 109). It is our policy to make a histological diagnosis in patients with significant liver failure via a transjugular biopsy, but restricting the study population to biopsy proven cirrhosis would have excluded some subjects with less significant liver disease. This was a deliberate decision to ensure a homogeneous study population. The clinical history was reviewed to corroborate that there was a history of excess alcohol. Any without this history were excluded from the study on the basis that they were probably non-alcoholic fatty liver disease and any with any other form of significant liver co-morbidity eg hepatitis C were excluded, leaving 104 patients. All our patients undergo serological testing for viral hepatitis. The clinical notes were reviewed and complete data was obtained on 94 patients (in 4 the notes could not be found and in 6 only partial data could be obtained and therefore these patients were excluded from data analysis).</p><p>The socio-demographic and clinical characteristics of this group were as follows: 52 male, 42 female; mean age 49 years 2 months; 69 had ascites at the time of biopsy (25 did not); 45 had varices at the time of biopsy (24 did not and 25 were unknown); 10 were Childs-Pugh Grade A, 42 were Grade B, 40 were Grade C and 2 were unknown.</p><p>The date at which each patient was first identified as having alcoholic liver disease was recorded and this was defined as either presenting with alcoholic hepatitis or cirrhosis, or having a biopsy showing alcohol-induced cirrhosis. Any admissions to hospital prior to this diagnosis were identified and were further divided into incidental admissions (no apparent relation to alcohol, for example, elective operations) and alcohol related admissions (including drug overdose, non-variceal gastro-intestinal bleeds, pancreatitis, fractures, head injuries etc).</p></sec><sec><title>2) Interview study of patients with alcoholic liver disease</title><p>Patients with alcoholic liver disease were recruited from the liver wards and liver outpatient clinics at Southampton University Hospitals NHS Trust. 45 subjects were identified and interviewed as part of a wider study into alcohol. Part of this study used a locally developed questionnaire to determine contact with health care services regarding alcohol intake and alcohol related health problems. Specifically the questions asked were <italic>'when was the first time, if ever, that you came to the attention of a health-care professional as a result of excessive alcohol intake?'</italic>, <italic>'how many times have you seen your GP in the last five years'</italic>, <italic>'have you ever been admitted to hospital/seen in Accident and Emergency for alcohol related problems in the past? If yes give details and provide number of days spent in hospital due to alcohol'</italic>. Subjects were not considered suitable for interview if they were experiencing hepatic encephalopathy or severe withdrawal symptoms.</p><p>Data were recorded and analysed using Microsoft Excel and SPSS.</p></sec></sec><sec><title>Results</title><sec><title>1) Retrospective clinical review of patients with alcohol-induced liver cirrhosis</title><p>Overall, 38 patients out of 94 total (40%) had had an admission to hospital for any reason before being diagnosed with alcoholic liver disease and the average number of admissions per subject (in those who had had an admission) was 2.8 (SD 2.1). This included 31 patients (33%) who had had an admission to hospital for a reason that might have been classed as alcohol related before being diagnosed with alcoholic liver disease and the average number of alcohol related admissions was 2.3 (SD 2.1). 67% had therefore had no admission to hospital for an alcohol related condition.</p><p>The mean length of time between the first admission to hospital for any reason (incidental or alcohol related) and being diagnosed with alcoholic liver disease was 10 years and 4 months (SD 7 years 0 months). The mean length of time between the first alcohol related admission to hospital and being diagnosed with alcoholic liver disease was 7 years and 1 month (SD 6 years 3 months). The reasons for alcohol related admissions are shown in table <xref ref-type="table" rid="T1">1</xref>. 17% of patients (16/94) died within 3 months of their first presentation with alcoholic liver disease. Only 6 of these patients had had a prior admission to hospital with an alcohol related condition.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Reasons for alcohol related admissions prior to developing alcoholic liver disease as determined from a retrospective analysis of medical notes in 94 patients with alcohol-induced cirrhosis (total number of admissions).</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Reason for admission</td><td align="left">Total number of admissions</td></tr></thead><tbody><tr><td align="left">Fall/fracture/injury (including head injury)</td><td align="left">16</td></tr><tr><td align="left">Gastro-intestinal bleed (non-variceal)</td><td align="left">12</td></tr><tr><td align="left">Drug overdose</td><td align="left">11</td></tr><tr><td align="left">Loss of consciousness/seizure</td><td align="left">7</td></tr><tr><td align="left">Abdominal pain due to eg gastritis, peptic ulcer disease</td><td align="left">6</td></tr><tr><td align="left">Alcohol overdose</td><td align="left">4</td></tr><tr><td align="left">Oral malignancy</td><td align="left">2</td></tr><tr><td align="left">Pneumonia</td><td align="left">2</td></tr><tr><td align="left">Confusion/Wernicke's encephalopathy</td><td align="left">2</td></tr><tr><td align="left">Oedema</td><td align="left">1</td></tr><tr><td align="left">TB</td><td align="left">1</td></tr></tbody></table></table-wrap></sec><sec><title>2) Interview study of patients with alcoholic liver disease</title><p>Patients presented after an average of 16 years of heavy drinking to health services. In response to the question <italic>'how many times have you seen your GP in the last five years'</italic>, those who were still drinking had seen their GP a mean of 9 times per year (16–44 age group) or 13 times per year (45–64 age group). Those who were abstinent had seen their GP a mean of 2 times per year (16–44 age group) or 6 times per year (45–64 age group).</p><p>In response to the question <italic>'Have you ever been admitted to hospital/seen in Accident and Emergency for alcohol related problems in the past?' </italic>there was a mean of 3 admittances/attendances per subject (range 0–17). Subjects had spent a mean of 23 days as an inpatient.</p></sec></sec><sec><title>Discussion</title><p>Retrospective clinical review of patients with alcohol-induced cirrhosis showed that before being diagnosed with alcoholic liver disease, most of the patients (67%) did not have an admission to hospital with an alcohol related condition during which they could have been screened for alcohol misuse. This therefore has implications for targeting of resources as most of these patients would need to have been detected in primary care. If there was an effective means to screen 100% of all admissions to hospital, then out of this cohort of patients a maximum of 40% could have been detected as hazardous drinkers. There was a mean of approximately 2 opportunities for detection ie admissions to secondary care in those who did present as such. Slightly more admissions were found in the interviews where the patients with alcoholic liver disease had attended a mean of 3 times per subject.</p><p>16 of the 94 patients died within 3 months of their first presentation with alcoholic liver disease. This is extremely important because these patients probably had no warning of the harm they were causing their liver and the first time they became aware of it, they had presented with a fatal condition. These patients had no chance to stop drinking and change the course of their illness because it was too late. If we can identify these patients and give them advice about their drinking before they develop cirrhosis, this at least gives them a chance to stop drinking. Simply telling someone they have alcoholic liver disease is enough to stop patients drinking or make them reduce their drinking to safe levels in just over half of patients in our experience (see below). Of the16 patients, only 6 had had a prior admission to hospital with an alcohol related condition, and therefore again, the main opportunity for detection and treatment would have been in primary care.</p><p>In those who presented to secondary care with an alcohol related condition, the mean time difference between the first presentation and detection of alcoholic liver disease was approximately 7 years. We know that around half of subjects respond positively to a brief invention [<xref ref-type="bibr" rid="B12">12</xref>] and similar results were seen in a cohort of heavy drinkers screened with gamma GT resulting in a significant reduction in liver morbidity [<xref ref-type="bibr" rid="B14">14</xref>]. Therefore if we can detect patients before they develop end stage liver disease we are likely to be able to reduce their alcohol consumption and thus avoid the development of cirrhosis in a substantial proportion. An audit recently carried out in Southampton showed that of 124 patients with alcoholic liver disease who survived one year after development of alcoholic liver disease, 54% were abstinent or drinking within liver safe levels (N. Sheron, personal communication). Patients were not given specific therapy for alcohol dependence, and thus figures represent the response of medical advice about their liver disease.</p><p>The first part of the study used only patients with cirrhosis who had had a biopsy, which was a deliberate attempt to study the patients with the most severe liver disease. All patients in Southampton with evidence of significant alcohol related liver disease are biopsied at some stage, so we felt this was a good way to detect patients with cirrhosis and that a large proportion would not be missed. Because only patients with alcohol-induced cirrhosis were selected, patients with non-cirrhotic but otherwise significant liver disease were not included because of this, for example those with alcoholic hepatitis and fibrosis and could lead to suggestion of selection bias. We feel that this actually strengthens our study as we used the patients who developed the most severe liver disease and most were not picked up by the health care system.</p><p>In the interview part of the study, patients had been drinking heavily for an average of 16 years before they presented to health services. In other words, there is a long 'latent' period in which we can detect patients who are drinking harmfully or hazardously during which we could reduce or stop their drinking. Most patients who develop alcoholic liver disease are not physically alcohol dependent [<xref ref-type="bibr" rid="B7">7</xref>].</p><p>A survey of drinking in adults conducted in the UK by the Office for National Statistics in 2004 showed that 11% of male drinkers and 8% of female drinkers had had a discussion with either their GP, someone else at the GP surgery, a doctor elsewhere or another medical person elsewhere about their drinking of alcohol in the last year [<xref ref-type="bibr" rid="B15">15</xref>]. Of the 11% of male drinkers who had this discussion, 9% of the discussions had been with their GP or someone else at the surgery, and only 2% had been with a doctor elsewhere or another medical person elsewhere and similar results had been obtained for women.</p><p>This study suggests that there is a place for detection of harmful and hazardous drinkers by screening admissions to secondary care, but we cannot rely on this alone as most patients would not be picked up this way. Primary care therefore probably has the biggest role to play in detection of alcohol misusers. The FAST test was specifically developed to detect hazardous drinking and not just dependency, it takes 13 seconds and has a sensitivity and specificity of around 90% [<xref ref-type="bibr" rid="B16">16</xref>]. The wider use of this test followed by appropriate stepped interventions along the lines of those used in Copenhagen have the potential to reduce liver deaths. Detailed strategies for detection and management of heavy drinkers in primary care have been previously published [<xref ref-type="bibr" rid="B17">17</xref>]. In order to detect alcohol misusers in General Practice, there needs to be screening for alcohol intake and liver disease, but until we do that, deaths will not reduce.</p></sec><sec><title>Conclusion</title><p>We have increasing rates of liver cirrhosis in the UK, directly related to increasing alcohol consumption. Strategies to identify those at risk of developing alcoholic liver disease, including cirrhosis are urgently needed. Patients who misuse alcohol can be identified using questionnaires, which can be administered in primary or secondary care. The Royal College of Physicians have recommended as one strategy to screen admissions to hospital with alcohol related conditions for evidence of alcohol misuse.</p><p>In this study, we showed that most patients who develop alcohol-induced cirrhosis are not admitted to secondary care with an alcohol related condition (or any condition) before they develop alcoholic liver disease. Thus the opportunities to detect them as alcohol misusers in secondary care before they develop alcoholic liver disease are limited. Patients with alcoholic liver disease who were interviewed had on average at least 2 episodes of contact with primary care per year and 90% of people visit their General Practitioner in a 5 year period, meaning that primary care is ideally placed for detection of those who misuse alcohol.</p><p>In summary, there are opportunities to detect patients at risk of developing alcoholic liver disease in secondary care but this study suggests that the majority of detection needs to be done in primary care. In this study, patients had been drinking heavily for an average of 16 years before presenting to health services and therefore there is ample opportunity to detect and treat those at risk of developing alcoholic liver disease before they cause significant damage to their health.</p></sec><sec><title>Competing interests</title><p>The author(s) declare they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>CV carried out the retrospective clinical analysis part of the study and drafted the manuscript.</p><p>SS carried out the interview part of the study.</p><p>NS was the senior advisor on the study and helped draft the manuscript.</p></sec>
|
Globalization and local response to epidemiological overlap in 21st century Ecuador
|
<sec><title>Background</title><p>Third World countries are confronted by a complex overlay of two sets of health problems. Traditional maladies, including communicable diseases, malnutrition, and environmental health hazards coexist with emerging health challenges, including cardiovascular disease, cancer, and increasing levels of obesity. Using Ecuador as an example, this paper proposes a conceptual framework for linking epidemiologic overlap to emerging social structures and processes at the national and global levels.</p></sec><sec><title>Discussion</title><p>Epidemiologic trends can be seen as part of broader processes related to globalization, but this does not imply that globalization is a monolithic force that inevitably and uniformly affects nations, communities, and households in the same manner. Rather, characteristics and forms of social organization at the subnational level can shape the way that globalization takes place. Thus, globalization has affected Ecuador in specific ways and is, at the same time, intimately related to the form in which the epidemiologic transition has transpired in that country.</p></sec><sec><title>Summary</title><p>Ecuador is among neither the poorest nor the wealthiest countries and its situation may illuminate trends in other parts of the world.</p><p>As in other countries, insertion into the global economy has not taken place in a vacuum; rather, Ecuador has experienced unprecedented social and demographic change in the past several decades, producing profound transformation in its social structure. Examples of local represent alternatives to centralized health systems that do not effectively address the complex overlay of traditional and emerging health problems.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Waters</surname><given-names>William F</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
Globalization and Health
|
<sec><title>Introduction: epidemiologic transition and globalization</title><p>This paper begins with the premise that global public health is not at its core only a medical issue but is, rather, embedded in social, cultural, political, and economic structures and processes. Moreover, changes in those structures and processes involve the evolution of patterns of health and wellness, which can be described in terms of epidemiologic transition and overlap. While this transition is part of broader processes related to globalization, globalization is not necessarily an essentially monolithic force that inevitably, invariably, and uniformly affects nations, communities, and households in the same manner. Rather, local specificities and forms of organization can and do shape the way that both globalization and the epidemiologic transition take place. Thus, globalization has affected Ecuador in specific ways and is, at the same time, intimately related to the form in which the epidemiologic transition has transpired in that country.</p><p>Globalization has been viewed from a variety of perspectives and is at the center of overlapping debates. One debate focuses on the fundamental nature of globalization: is it essentially a narrowly-defined economic and financial process of integration of national economies into an international economy, or does it also include more broadly-defined interweavings of political, technological, and cultural processes? This debate is framed by a broader issue: has globalization benefited most people in the world or not? A different debate concerns the relationship between globalization, public health, and the epidemiologic transition [<xref ref-type="bibr" rid="B1">1</xref>]. In this context, globalization affects public health in a variety of ways because it has unleashed profound changes that have redefined how institutions at many levels–nation states, government agencies, transnational corporations, multilateral organizations, non-governmental organizations, public and private health care providers, community-based and other affinity-based organizations, communities, and households–operate and interact with one another.</p><p>At the same time, the world is currently in the midst of an epidemiologic transition, defined as:</p><p>the evolutionary changes in different societal settings from a situation of high mortality, high fertility, short life expectancy, young age structure, and predominance of communicable diseases; especially in the young, to one of low mortality, low fertility, increasing life expectancy, aging, and predominance of degenerative and man-made diseases, especially among the middle and old ages [[<xref ref-type="bibr" rid="B2">2</xref>]: 5].</p><p>The epidemiologic transition incorporates the demographic transition (the change from high mortality and fertility to low mortality and fertility) as well as evolving patterns in the causes of morbidity and mortality. At the heart of the epidemiologic transition is a shift in the determinants of mortality and morbidity, whereby infectious and communicable diseases are supplanted by chronic and non-communicable conditions. This transformation is not uniform, however; it transpires in different ways and different times among and within different societies, and at different velocities. Thus, the transition experienced by presently industrialized countries in the past differs significantly from the experience of underdeveloped countries at present. Moreover, presently underdeveloped countries follow different patterns of transition [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>]. As discussed below, one difference between past and present experience is that in countries like Ecuador, increasing rates of chronic and non-communicable disease associated with increasing longevity and a gradually aging population are experienced by continued high levels of infectious and communicable disease. Moreover, as discussed below, patterns of morbidity and mortality differ among socioeconomic groups due in large part to differences in their relationship to globalizing forces.</p><p>While the global reach of economic and non-economic processes is undeniable, globalization encompasses more than the redefinition of relationships between and among nation states, transnational corporations, and international organizations, as both critics and defenders of globalization often assert. Almost always left out of the analysis are the differences in the effect of these relationships on communities and other forms of local organization and more importantly, how those forces are shaped at the local level. The view that this paper proposes is that local actors are not necessarily relegated to the role of passive recipients of immutable global forces, and that the economic, social, and cultural impacts of globalization are not uniform among or within countries. Moreover, globalization has produced discontent as people and money are subjected to new patterns of mobility, while externally-imposed conditions are confronted by struggling nation states.</p><p>In other words, although much of the Third World still faces poverty and inequality [<xref ref-type="bibr" rid="B5">5</xref>], the impact of globalization is neither monolithic nor uniform, and local response is not only possible, but actually offers viable options to economic and political domination and cultural homogenization. In this view, for instance, local collective capacity in Ecuador continues to represent an effective counterweight to global forces such that globalization can, in effect, be shaped at the local level [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. This is so in part because local culture remains a vital force despite homogenizing influences and can even be brought to bear in order to assert and reassert local values and practices [<xref ref-type="bibr" rid="B8">8</xref>]. More dramatic, perhaps, but no less relevant, is that the effects of globalization have been actively resisted throughout the world, including Latin America [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Local, regional, and national resistance to unpopular measures in Ecuador [<xref ref-type="bibr" rid="B11">11</xref>] has strengthened the indigenous movement as it confronts transnational capital so that grass roots democracy has been strengthened [<xref ref-type="bibr" rid="B12">12</xref>]. At the local level, for instance, public health can be put at the service of real people at the local level, and in addition, communities can and do participate in developing and implementing health care that meets their needs.</p></sec><sec><title>Epidemiologic overlap: a global process</title><p>Just as economic, political, social, and cultural relationships are emerging throughout the world, patterns of morbidity and mortality are also undergoing complex patterns of epidemiologic transition that vary among and within countries [<xref ref-type="bibr" rid="B2">2</xref>]. But the particular path that epidemiological transition takes in a given case is closely related to social, economic, political, and cultural systems and processes that are, in turn, being redefined by globalization. Of particular relevance are the interrelationships among poverty, inequality, and health [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. These interrelationships are particularly germane in contemporary Ecuador [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>] and throughout Latin America [<xref ref-type="bibr" rid="B17">17</xref>-<xref ref-type="bibr" rid="B21">21</xref>].</p><p>The basic model of epidemiologic transition posits that mostly because of enhanced scientific understanding leading to the germ theory of disease and systematic improvements in sanitation infrastructure, four groups of what Omran [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>] called "traditional" health problems began to recede in industrialized countries in the 19<sup>th </sup>and early 20<sup>th </sup>centuries: (1) communicable diseases, including respiratory illnesses and tuberculosis, diarrheal diseases, vaccine preventable diseases, and vector-borne diseases such as malaria and dengue; (2) poor health outcomes in mothers and infants related to reproduction and childbirth; (3) nutritional deficiencies; and (4) illnesses related to poor sanitation, especially water-borne pathogens in public water supplies and deficient sewage disposal. These problems are exacerbated by health care systems that lack the resources and capacity to attend to more than the most basic health problems. According to the basic model, the "traditional" conditions are gradually supplanted by a different set of "modern" health problems: (1) cardiovascular diseases, (2) malignancies due to cancer, (3) stress and other mental disorders, (4) diseases related to aging (such as Alzheimer's disease), (5) accidents (both traffic and occupational), and (6) emerging and re-emerging diseases and conditions, including overweight and obesity, diabetes, and hypertension. These conditions are exacerbated by health care delivery that is inadequate because of poor coverage, urban bias, limited outreach, poorly trained health care professionals, overly centralized operation, and an emphasis on curative rather than preventive care [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>].</p><p>The conception of the epidemiologic transition represents less a theoretical construct than a descriptive model, which was not intended to be and should not be taken as an extension of modernization theory as postulated beginning in the 1960s [<xref ref-type="bibr" rid="B22">22</xref>], according to which, development is thought of as a series of stages through which all societies pass [<xref ref-type="bibr" rid="B23">23</xref>]. Rather, the model describes a variety of global and national processes that have shaped the evolution of health conditions throughout the world and in different historical moments. The simultaneous expression of morbidity and mortality due to "traditional" and "modern" health conditions obliges us to reevaluate the basic model of epidemiologic transition in light of diverse social and economic conditions. First, "traditional" diseases have not disappeared from industrialized countries, and a panoply of new and re-emerging infectious diseases pose new threats. Second, underdeveloped countries like Ecuador continue to experience high prevalence rates of infectious and communicable diseases, but at the same time, increasing rates of chronic and non-transmissible diseases associated with later phases of the epidemiologic transition [<xref ref-type="bibr" rid="B24">24</xref>]. Consequently, on one hand, well-documented general trends in global public health can be observed. For example, chronic diseases now account for 59 percent of the 57 million deaths reported worldwide (about half of these attributable to cardiovascular disease) and 46 percent of the global burden of disease [<xref ref-type="bibr" rid="B25">25</xref>]. At the same time, though, chronic diseases have become increasingly prevalent in underdeveloped countries and less prevalent in industrialized countries. On the other hand, traditional health problems in the former remain highly prevalent. For example, about 60 percent of all deaths among children under the age of five in the world are associated with malnutrition, and Vitamin A and iodine deficiencies continue to take heavy tolls in underdeveloped countries [<xref ref-type="bibr" rid="B26">26</xref>].</p><p>In other words, evolving health profiles in industrialized and underdeveloped countries suggest that the epidemiologic transition involves more than the gradual replacement of one set of diseases with another and that the epidemiologic transition can be more accurately described as a double epidemiologic overlap, one internal and one global [<xref ref-type="bibr" rid="B27">27</xref>]. The first overlap is represented by the continued high prevalence rates of both "traditional" and "modern" diseases in countries like Ecuador. But the burden of disease (which includes mortality and morbidity) is not uniformly distributed within the population. Rather, differences within countries can be attributed to inequalities related to socioeconomic factors such as income, occupation, ethnicity, level of education, and rural/urban residence. The second overlap comes about because as a product of globalization, the health profile of different groups of residents in underdeveloped and industrialized countries overlap. In both cases, the wealthy experience relatively lower rates of disease because of access to globalized health services (within or outside their own borders), information, healthy diets, and protection from environmental and occupational risks. At the same time, the rural and urban poor in both cases experience higher rates of both traditional communicable and infectious diseases (many of which are related to poor sanitary conditions, unhealthy housing, and ineffective control of vectors) and modern diseases, which are exacerbated by limited access to health care and failed health care policies.</p><p>The second overlap is a product of increasing integration into global markets, for example, in the production and processing of export-oriented agricultural commodities (much of it involving non-traditional products like cut flowers, tropical fruit, and temperate vegetables). This process connects the rural and urban poor in Ecuador (whose own consumption consists of increasingly more processed foods of poor nutritional quality) with new forms participation in global supermarkets by residents of the industrialized countries [<xref ref-type="bibr" rid="B28">28</xref>]. But consumption patterns vary within populations: those typical of the tiny affluent elite in Ecuador are similar to those of their northern counterparts–but in a lagged fashion. Among the imported consumer items available at high cost in elite supermarkets in urban Ecuador are imported processed, canned, and frozen items. These items represent a unique form of prestigious consumption because they reflect the same kind of expensive, flexible, and niche-driven consumption in industrialized countries. Moreover, among the Ecuadorian elites, health behaviors and health status now approximate patterns found in the industrialized countries. This is not a coincidence, because these segments have the same level of health care, which is secured (and often paid for through private insurance) either in local, private clinics and hospitals that are indistinguishable from those in industrialized nations, or in facilities actually located in the industrialized countries, especially in the southern United States.</p><p>The epidemiologic transition model proposed by Omran [<xref ref-type="bibr" rid="B4">4</xref>] takes into account these complexities and variations, which are found among and within countries. Thus, the "western" variation experienced by the presently industrialized countries has played out in five stages: (1) an age of pestilence and famine that occurred through the early 19<sup>th </sup>century; (2) an age of receding pandemics beginning in the 19<sup>th </sup>and early 20<sup>th </sup>centuries; (3) an age of increasing degenerative, stress and man-made conditions that is still underway in some places and populations; (4) an age of declining cardiovascular mortality, ageing, lifestyle modifications, and emerging and resurgent diseases, now clearly observable in the United Stages and other industrialized countries; and (5) a future stage of "aspired quality of life, with paradoxical longevity and persistent inequalities" [[<xref ref-type="bibr" rid="B4">4</xref>]: 102]. This analysis also points out that contemporary social structures in the western transition model are characterized by generally improved living conditions, improved sanitation, small family size, and enhanced education and participation among women; while curative and preventive health care is organized at national and subnational levels and health insurance is available for individuals, groups (via employment and managed care plans) and entire nations (as in Great Britain). On the other hand, during the fourth stage of the transition, some residents of industrialized countries may experience limited access to health care, increased cost, and over-specialization of health services [[<xref ref-type="bibr" rid="B4">4</xref>]: 104].</p><p>In contrast, countries in Latin America and the Caribbean have followed a different, non-western model; for example, Ecuador, Peru, Paraguay, and the Dominican Republic typify the "lower intermediate" variation of the non-western model. According to this model, countries like Ecuador experienced the traditional diseases described above in the early 20<sup>th </sup>century (until about 1940), when they began the process of epidemiologic transition, followed by epidemiologic overlap. The co-existence of traditional and modern health conditions is compounded by poor health care because of health systems and medical training that function poorly in the face of multiple new demands. This "triple health burden" [[<xref ref-type="bibr" rid="B4">4</xref>]: 106] distinguishes the epidemiologic transition in countries like Ecuador from that in countries like the United States [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>].</p></sec><sec><title>Ecuador: globalization and health as poverty and inequality</title><p>Ecuador's role in the global economy is very small; its GDP of about 19 billion dollars amounts to less than one tenth of Wal-Mart's annual sales. Nevertheless, Ecuador is still intimately linked to processes of globalization in at least six ways. First, transnational companies (including the two largest banks in the world, Citibank and Bank of America) operate in Ecuador. Second, while Ecuador continues to export traditional commodities (especially oil, bananas, coffee, and cocoa), it has also aggressively embarked upon the export of non-traditional products, mostly agricultural–notably, cut flowers [<xref ref-type="bibr" rid="B29">29</xref>]. Third, Ecuadorian workers produce for a global market, both at home and as transnational migrants [<xref ref-type="bibr" rid="B30">30</xref>]. Fourth, it is signatory to the World Trade Organization's most recent agreements, which govern global trade and finance and is actively engaged in different regional trade agreements. Fifth, it is heavily indebted to transnational banking institutions and multilateral lenders, which have imposed strict conditions related to their loan portfolios. For instance, an agreement signed with the IMF in 2000 contained 167 loan conditions that involved, for example, the privatization of potable waters systems, a new oil pipeline contract, layoffs of some public employees and wage cuts for others, and increases in the price of basic commodities like cooking oil [<xref ref-type="bibr" rid="B31">31</xref>]. Sixth, while autochthonous culture remains vibrant, imported culture floods local markets in the form of language, food, dress, and music.</p><p>Insertion into the global economy does not occur in a domestic vacuum, though; Ecuador has experienced unprecedented social and demographic change in the past several decades, producing profound transformation in its social structure, as reflected in the contribution to total GDP by agriculture, industry, and services. (See Table <xref ref-type="table" rid="T1">1</xref>.) Employment patterns have shifted in parallel fashion; only eight percent of the economically active population now works in agriculture, 24 percent in industry, and 68 percent in the service sector [<xref ref-type="bibr" rid="B32">32</xref>].</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Distribution of gross domestic product by sector. Ecuador, 1965–2004. Percent.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="left">Agriculture</td><td align="left">Industry</td><td align="left">Services</td></tr></thead><tbody><tr><td align="left">1965</td><td align="left">27</td><td align="left">22</td><td align="left">50</td></tr><tr><td align="left">1988</td><td align="left">15</td><td align="left">36</td><td align="left">49</td></tr><tr><td align="left">2004</td><td align="left">7</td><td align="left">40</td><td align="left">63</td></tr></tbody></table><table-wrap-foot><p>SOURCE: [33:182; 34: 296].</p></table-wrap-foot></table-wrap><p>These changes are closely associated with permanent rural-urban migration. Ecuadorian society was largely rural and agrarian through the mid-20<sup>th </sup>century, but 63.2 percent of its population was urban in 2001, and the figure is projected to reach 69.4 percent by 2015. While Quito and Guayaquil have grown dramatically–largely because of rural-urban migration–small and intermediate cities have grown even more quickly in many cases. Urban growth in Ecuador is further fueled by cyclical and temporary immigration by the rural poor in order to supplement meager rural household income with sporadic or temporary incomes derived from the informal urban sector [<xref ref-type="bibr" rid="B35">35</xref>].</p><p>Problems related to rural poverty are generally not resolved by migration, though; they are merely urbanized. Thus, urban unemployment nearly doubled from 9.2 percent in March 1998 to 17 percent in July 1999 and only returned to 9.3 percent by December 2005. In addition, underemployment (mostly in informal microenterprises) stood at 49.2 percent at the end of 2005. Consequently, poverty and indigence (or extreme poverty) expanded beginning in 1990, as shown in Table <xref ref-type="table" rid="T2">2</xref>, and levels remain essentially unchanged today. This trend mirrors stagnant and declining real wages, which have only recently risen above those of several decades ago [<xref ref-type="bibr" rid="B36">36</xref>,<xref ref-type="bibr" rid="B37">37</xref>].</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Poverty and Indigence in Ecuador, 1995–2001. Percent.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center" colspan="4"><bold>Poverty</bold></td><td align="center" colspan="4"><bold>Indigence</bold></td></tr></thead><tbody><tr><td></td><td align="left"><bold>1995</bold></td><td align="left"><bold>1998</bold></td><td align="left"><bold>2000</bold></td><td align="left"><bold>2001</bold></td><td align="left"><bold>1995</bold></td><td align="left"><bold>1998</bold></td><td align="left"><bold>2000</bold></td><td align="left"><bold>2001</bold></td></tr><tr><td align="left"><bold>Rural</bold></td><td align="left">75.8</td><td align="left">82.0</td><td align="left">84.1</td><td align="left">77.5</td><td align="left">33.9</td><td align="left">46.1</td><td align="left">58.2</td><td align="left">50.5</td></tr><tr><td align="left"><bold>Urban</bold></td><td align="left">42.4</td><td align="left">48.6</td><td align="left">60.3</td><td align="left">51.6</td><td align="left">10.6</td><td align="left">13.0</td><td align="left">30.3</td><td align="left">24.7</td></tr><tr><td align="left"><bold>Total</bold></td><td align="left">55.0</td><td align="left">62.6</td><td align="left">68.8</td><td align="left">60.8</td><td align="left">20.0</td><td align="left">26.9</td><td align="left">40.3</td><td align="left">33.8</td></tr></tbody></table><table-wrap-foot><p>SOURCE: [37: 50].</p></table-wrap-foot></table-wrap><p>Crisis-driven poverty is also reflected in the distribution of resources and consumption. As an agrarian society, Ecuador was historically characterized by concentrated land ownership. Today, inequality in an increasingly urban, service-driven society is reflected in income and living conditions. In 1988, the wealthiest quintile of the population earned 50.6 percent of total income, while the poorest quintile earned 3.9 percent. But in 2004, the gap was even wider: the wealthiest quintile earned 62.3 percent of the population, while the poorest quintile earned only 1.7 percent [<xref ref-type="bibr" rid="B36">36</xref>]. Not surprisingly, the Gini coefficient of income inequality increased from 0.49 in 1995 to 0.57 in 1999 and 0.62 in 2001 (following dollarization of the economy), returning to 0.42 in 2003. Similarly, the Gini coefficient of consumption inequality has changed little, decreasing from 0.41 in 1995 to 0.38 for 2003–2004 [<xref ref-type="bibr" rid="B38">38</xref>].</p><p>These differences are closely related to gaps in living conditions. For example, in 2000, 77 percent of the population in the wealthiest income decile had access to a private flush toilet, compared to only 12 percent of people in the poorest income decile. Similar patterns are observed when comparing urban to rural areas; in 2002, 80 percent of urban Ecuadorians had access to improved sanitation while only 59 percent of rural residents did [<xref ref-type="bibr" rid="B36">36</xref>]. Access to clean water is a fundamental aspect of public health, and Table <xref ref-type="table" rid="T3">3</xref> shows enormous breaches between rural and urban residents and between the wealthy (top decile) and poor (bottom decile).</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Access to a source of clean water. Ecuador, 1999 and 2002. Percent.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>Poorest decile</bold></td><td align="center"><bold>Wealthiest decile</bold></td><td align="center"><bold>Total 1999</bold></td><td align="center"><bold>Total 2002</bold></td></tr></thead><tbody><tr><td align="left"><bold>Urban</bold></td><td align="center">56.2</td><td align="center">90.8</td><td align="center">75.3</td><td align="center">92</td></tr><tr><td align="left"><bold>Rural</bold></td><td align="center">42.3</td><td align="center">49.1</td><td align="center">46.3</td><td align="center">77</td></tr><tr><td align="left"><bold>Rural dispersed</bold></td><td align="center">11.2</td><td align="center">26.3</td><td align="center">18.5</td><td align="center">–</td></tr></tbody></table><table-wrap-foot><p>Source: [39:238].</p></table-wrap-foot></table-wrap><p>Over a decade ago, poor living conditions were shown to be associated with adverse health outcomes among the poor in Ecuador [<xref ref-type="bibr" rid="B40">40</xref>]. Perhaps most dramatically, the ratio of the poor/non-poor risk of dying is more than 4 to 1 for Ecuadorian women and almost 3 to 1 for men [[<xref ref-type="bibr" rid="B41">41</xref>]: Statistical annex, table 7]. Gaps between urban and rural residents and by level of educational attainment further illustrate these relationships. Table <xref ref-type="table" rid="T4">4</xref> provides data on two sensitive indicators of health and development and suggests that substantial gaps in health outcomes remain, based on rural/urban residence, level of education, and province of residence (which reflects, among other things, race and ethnicity).</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Health and development disparities, Ecuador. Rates of fertility and infant mortality.</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="left"><bold>Urban areas</bold></td><td align="left"><bold>Rural areas</bold></td><td align="left"><bold>No education or primary</bold></td><td align="left"><bold>Secondary education or more</bold></td><td align="left"><bold>Lowest provincial rate</bold></td><td align="left"><bold>Highest provincial rate</bold></td></tr></thead><tbody><tr><td align="left">Fertility rate per woman 15–49</td><td align="left">2.8</td><td align="left">4.3</td><td align="left">5.6</td><td align="left">1.9</td><td align="left">2.7</td><td align="left">4.7</td></tr><tr><td align="left">Infant mortality rate per 1,000 live births</td><td align="left">22.0</td><td align="left">40.0</td><td align="left">51.0</td><td align="left">11.0</td><td align="left">26.0</td><td align="left">34.0</td></tr></tbody></table><table-wrap-foot><p>SOURCE: [39: 241].</p></table-wrap-foot></table-wrap><p>Health inequalities, understood as gaps in both access to care and outcomes, were exemplified by the rapid spread of cholera in 1991 from the port city of Callao, Peru through virtually the entire continent. Cholera struck almost exclusively in urban neighborhoods and poor rural communities, where morbidity and mortality were due to unsafe drinking water and inadequate sanitation, as well as consumption of unwashed or uncooked foodstuffs [<xref ref-type="bibr" rid="B42">42</xref>] and lack of timely and effective treatment. After Peru, Ecuador had the highest prevalence rate (450.9 per 100,000) and the most cases (46,284) in the first year (1991), and total cases exceeded 93,000 through 2000 [[<xref ref-type="bibr" rid="B39">39</xref>]: 310–311]. But exposure, morbidity, and mortality due to the disease were unevenly distributed: the poorest neighborhoods, particularly on the Coast, were heavily affected, while populations with access to safe supplies of treated public water were not. Cholera was present in relatively isolated highland indigenous communities, where mortality rates due to the disease were six times the national average [<xref ref-type="bibr" rid="B43">43</xref>].</p></sec><sec><title>Epidemiologic overlap in Ecuador</title><p>Table <xref ref-type="table" rid="T5">5</xref> reflects the evolution of causes of death in Ecuador. It can be seen that of the 15 leading causes of death, nine (other heart disease, cerebrovascular diseases, diabetes mellitus, hypertensive diseases, aggression, isquemic heart disease, traffic accidents, malignant tumors, and self-inflicted injuries) can be classified as modern conditions. It can be noted in passing that the prominence of the "other heart disease" category has two explanations. First, as the population gradually ages and enters the final stages of the epidemiologic transition, heart disease will become more prevalent. Second, however, this particular cause of death is often ascribed when accurate information is lacking, particularly when people die of causes that are either poorly treated or not treated at all, when no autopsy is conducted, and when underlying causes leading to heart failure are never established.</p><table-wrap position="float" id="T5"><label>Table 5</label><caption><p>Principal causes of death. Ecuador, 2004 (per 10,000 inhabitants).</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Cause</bold></td><td align="center"><bold>Total</bold></td><td align="center"><bold>Males</bold></td><td align="center"><bold>Females</bold></td></tr></thead><tbody><tr><td align="left">1. Other heart disease</td><td align="center">3.1</td><td align="center">3.1</td><td align="center">3.1</td></tr><tr><td align="left">2. Pneumonia</td><td align="center">2.3</td><td align="center">2.5</td><td align="center">2.1</td></tr><tr><td align="left">3. Cerebrovascular diseases</td><td align="center">2.3</td><td align="center">2.4</td><td align="center">2.2</td></tr><tr><td align="left">4. Diabetes Mellitus</td><td align="center">2.1</td><td align="center">1.8</td><td align="center">2.3</td></tr><tr><td align="left">5. Hypertensive diseases</td><td align="center">1.9</td><td align="center">2.0</td><td align="center">1.8</td></tr><tr><td align="left">6. Aggression</td><td align="center">1.8</td><td align="center">3.2</td><td align="center">0.3</td></tr><tr><td align="left">7. Isquemic Heart Disease</td><td align="center">1.8</td><td align="center">2.1</td><td align="center">1.4</td></tr><tr><td align="left">8. Perinatal infections</td><td align="center">1.5</td><td align="center">1.7</td><td align="center">1.2</td></tr><tr><td align="left">9. Traffic accidents</td><td align="center">1.4</td><td align="center">2.2</td><td align="center">0.7</td></tr><tr><td align="left">10. Liver diseases</td><td align="center">1.3</td><td align="center">1.7</td><td align="center">0.8</td></tr><tr><td align="left">11. Malignant tumors, stomach</td><td align="center">1.1</td><td align="center">1.3</td><td align="center">1.0</td></tr><tr><td align="left">12. Chronic lower respiratory infections</td><td align="center">0.7</td><td align="center">0.8</td><td align="center">0.6</td></tr><tr><td align="left">13. Self-inflicted injuries</td><td align="center">0.6</td><td align="center">0.9</td><td align="center">0.4</td></tr><tr><td align="left">14. Septicemia</td><td align="center">0.5</td><td align="center">0.5</td><td align="center">0.5</td></tr><tr><td align="left">15. Respiratory tuberculosis</td><td align="center">0.5</td><td align="center">0.7</td><td align="center">0.3</td></tr></tbody></table><table-wrap-foot><p>SOURCE: [44].</p></table-wrap-foot></table-wrap><p>It should be noted that the epidemiologic transition in Ecuador occurred in the context of generally improved health outcomes, as measured by classic indicators; life expectancy at birth increased from 58.8 years (1970–1975) to 70.8 years in the 2000–2005 period, the infant mortality rate decreased from 87 per 1,000 live births in 1970 to 24 in 2001, and measles vaccination rates for one-year-olds increased from only 60 percent as recently as 1990 to 99 percent in 2001. Many of the changes are related to the gradual aging of the population; while 4.9 percent of Ecuadorians were over the age of 65 in 2001, the projection for 2015 is 6.6 percent. These are relatively low proportions (that of Uruguay is more than twice that of Ecuador), but it portends an important change in the future, as the presently bottom-heavy age pyramid gradually shifts upward.</p><p>The effects of the epidemiologic transition in Ecuador can also be seen in Table <xref ref-type="table" rid="T6">6</xref>, which provides data on morbidity as measured through hospital discharges. While it is true that these data probably underestimate less serious illnesses that do not require attention at a hospital (or for which many poor people would be unwilling or unable to pay), they nevertheless portray the relative contribution to the total burden of disease in the country. The poor state of health among Ecuadorian women is reflected by the fact that conditions related to pregnancy and child birth represent the top three causes of morbidity for men and women, and about 18 percent of the total. Panel A also shows that at least nine of the top ten causes of morbidity are traditional conditions that would be observed in the earlier stages of the epidemiologic transition. (Attributing fractures as a cause of morbidity to either the traditional or modern category is problematic).</p><table-wrap position="float" id="T6"><label>Table 6</label><caption><p>Principal causes of morbidity: hospital discharges, Ecuador, 2003. Rates per 10,000 inhabitants.</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>A. Total</bold></td><td align="left">Rate</td></tr></thead><tbody><tr><td align="left">1. Other complications from pregnancy and birth</td><td align="left">10.4</td></tr><tr><td align="left">2. Other pregnancies terminating in abortion</td><td align="left">4.2</td></tr><tr><td align="left">3. Other maternal conditions related to the fetus, amniotic cavity, and possible problems with birth</td><td align="left">3.4</td></tr><tr><td align="left">4. Diarrhea and gastroenteritis, presumably infectious</td><td align="left">3.2</td></tr><tr><td align="left">5. Colelitiasis and colecystitis</td><td align="left">2.9</td></tr><tr><td align="left">6. Pneumonia</td><td align="left">2.7</td></tr><tr><td align="left">7. Other traumas</td><td align="left">2.5</td></tr><tr><td align="left">8. Diseases of the appendix</td><td align="left">2.3</td></tr><tr><td align="left">9. Fractures</td><td align="left">1.4</td></tr><tr><td align="left">10. Other infectious intestinal diseases</td><td align="left">1.3</td></tr><tr><td colspan="2"><hr></hr></td></tr><tr><td align="left"><bold>B. Males</bold></td><td></td></tr><tr><td colspan="2"><hr></hr></td></tr><tr><td align="left">1. Other traumas</td><td align="left">5.7</td></tr><tr><td align="left">2. Diarrhea and gastroenteritis, presumably infectious</td><td align="left">5.5</td></tr><tr><td align="left">3. Pneumonia</td><td align="left">4.6</td></tr><tr><td align="left">4. Diseases of the appendix</td><td align="left">3.9</td></tr><tr><td align="left">5. Hernia</td><td align="left">3.2</td></tr><tr><td align="left">6. Fractures</td><td align="left">3.1</td></tr><tr><td align="left">7. Colelitiasis and colecystitis</td><td align="left">2.6</td></tr><tr><td align="left">8. Hyperplasia of the prostate</td><td align="left">2.1</td></tr><tr><td align="left">9. Other infectious intestinal diseases</td><td align="left">2.1</td></tr><tr><td align="left">10. Other respiratory problems in the perinatal period</td><td align="left">2.0</td></tr><tr><td colspan="2"><hr></hr></td></tr><tr><td align="left"><bold>C. Females</bold></td><td></td></tr><tr><td colspan="2"><hr></hr></td></tr><tr><td align="left">1. Other complications from pregnancy and birth</td><td align="left">15.1</td></tr><tr><td align="left">2. Other pregnancies terminating in abortion</td><td align="left">6.1</td></tr><tr><td align="left">3. Other maternal conditions related to the fetus, amniotic cavity, and possible problems with birth</td><td align="left">4.9</td></tr><tr><td align="left">4. Colelitiasis and colecystitis</td><td align="left">3.1</td></tr><tr><td align="left">5. Diarrhea and gastroenteritis, presumably infectious</td><td align="left">2.3</td></tr><tr><td align="left">6. Pneumonia</td><td align="left">1.9</td></tr><tr><td align="left">7. Diseases of the appendix</td><td align="left">1.7</td></tr><tr><td align="left">8. Mioma of the uterus</td><td align="left">1.3</td></tr><tr><td align="left">9. Diabetes mellitus</td><td align="left">1.1</td></tr><tr><td align="left">10. Other problems of the urinary tract</td><td align="left">1.1</td></tr></tbody></table><table-wrap-foot><p>SOURCE: [44].</p></table-wrap-foot></table-wrap><p>Data disaggregated by gender reveal that diabetes appears as an important cause of morbidity in women. In addition, the "other heart conditions" category probably represents further underestimates of chronic disease prevalence, including diabetes, which is asymptomatic in its early stages. (At the same time, screening for diabetes among asymptomatic persons at potential risk is nearly inexistent in Ecuador.) Moreover, diabetes is closely associated with overweight and obesity, which is increasing in Ecuador because of changing socioeconomic conditions related to urbanization, occupational structure, diet, and physical activity. Similarly, among men, conditions of the prostate appear as a leading cause of morbidity in Ecuador. This category probably signals increasing prevalence of cancer in men and women. Prevalence data for cancer is incomplete at best, since services of screening and early detection are rarely available to the bulk of the population.</p><p>General improvements in the indicators of public health and changing patterns of morbidity and mortality were not equally distributed within the population, however. Several studies confirm that health conditions vary by social group within the population. Regarding "traditional" health conditions:</p><p>• A national survey conducted in the mid-1980s found significant differences among social classes in the prevalence of infant and child malnutrition [<xref ref-type="bibr" rid="B45">45</xref>]. More recent studies confirm that these differences persist [<xref ref-type="bibr" rid="B15">15</xref>,<xref ref-type="bibr" rid="B43">43</xref>,<xref ref-type="bibr" rid="B46">46</xref>], and nationwide data for 2004 clearly demonstrate that chronic malnutrition (stunting) in children is closely related to poverty, residence in rural and highland areas, and indigenous ethnicity [<xref ref-type="bibr" rid="B47">47</xref>].</p><p>• Vitamin A deficiency continues to place some segments of the population at risk, particularly households in the highlands, indigenous households, rural households, and households in which the mother has no formal education or in which children are underweight or stunted [<xref ref-type="bibr" rid="B48">48</xref>].</p><p>• Chagas disease, a preventable vector-borne disease, is endemic in the Oriente and in the Guayas River basin. Between 120,000 and 200,000 Ecuadorians are infected and between 2.2 and 3.8 million live under the risk of transmission of the disease [<xref ref-type="bibr" rid="B49">49</xref>].</p><p>On the other hand, the "modern" health problems identified by Omran are highly prevalent [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>].</p><p>• The prevalence of overweight and obesity is now an epidemic only recently recognized. As of 2004, 40.4 percent of women were overweight (BMI of between 25 and 29.9) and 14.1 percent were obese (BMI over 29.9). At greatest risk are the urban poor because of factors associated with urbanization including changing diets, lifestyles, and occupational structure [<xref ref-type="bibr" rid="B50">50</xref>-<xref ref-type="bibr" rid="B52">52</xref>]. Overweight and obesity represents a critical feature of public health because it is associated with diabetes, heart disease, hypertension, and some forms of cancer [<xref ref-type="bibr" rid="B47">47</xref>].</p><p>• A study of the rural area around Borbón on the northwest coast found that cardiovascular diseases were the primary cause of death among adults, and that arterial hypertension, which was uncontrolled in most cases, was a major cause of mortality [<xref ref-type="bibr" rid="B53">53</xref>].</p><p>The situation of cancer merits special mention because it is not only an emerging disease in Ecuador, but because outcomes (both access to care and outcomes) reflect class-based differences. This is a particularly important factor in the case of diseases that may have low death rates when timely screening and treatment are available, but where death rates are high when early detection is not available. The few available studies reflect trends associated with cancer mortality rates.</p><p>• Uterine cancer has declined dramatically in industrialized countries, but more slowly in Latin America. Rates have changed little in Ecuador, however [<xref ref-type="bibr" rid="B54">54</xref>].</p><p>• Cancers related to occupational and environmental conditions pose additional risks for disease. For example, men and women who live around oil fields in the Amazonian provinces of Sucumbios, Orellana, Napo, and Pastaza face elevated risks of cancers of the stomach, rectum, skin, soft tissue, and kidney. In addition, women have increased risk of cancers of the cervix and lymph nodes; and children under the age of 10 have a higher risk of haematopoietic cancers [<xref ref-type="bibr" rid="B55">55</xref>].</p><p>• The age-adjusted incidence for cervical cancer is approximately 48 and mortality is approximately 19 per 100,000 [<xref ref-type="bibr" rid="B56">56</xref>]. This form of cancer is mainly associated with the human papilloma virus, but also to other factors, including poor diet, low life expectancy, barriers to health care, and low birth weight children. Protective factors include low fertility and delayed age at first childbirth. Incidence and mortality rates for cervical cancer also remain high (as compared to significant declines in urbanized countries) because of lack of prevention and control measures (particularly screening), which can reduce both mortality and incidence by 90 percent. Even when screening is available, inadequate collection and analysis of the samples and incomplete follow-up of women after testing further endangers poor women in particular. In sum, existing programs are "piecemeal, lack both organization and quality control, and have failed to meet their objectives" [<xref ref-type="bibr" rid="B56">56</xref>].</p><p>• While the prevalence of lung cancer is not particularly high, outcomes are poorer than expected because of the poor quality of care for those who are screened and treated; outpatient evaluation "is an efficient, slow, and potentially dangerous process in cases in which the probability of a cancer diagnosis is high" [[<xref ref-type="bibr" rid="B57">57</xref>]:167].</p><p>These data suggest that within Ecuador, the epidemiologic transition plays out differently among different populations, so that the non-western model displayed for the country as a whole must be interpreted as an essentially polarized variant in which particularly vulnerable segments of the population (rural, highland, indigenous and Afro-Ecuadorian, and the urban poor) continue to experience a protracted period of overlap.</p><p>In addition, part of the explanation the persistence of gaps in health outcomes lies in the Ecuadorian health care system. Despite important changes in the system in the past decade, the poor, including those who are either unemployed or the nearly half of the population who work in the informal sector (including peasant farmers), primarily use facilities operated by the Ministry of Health (MOH) while employees in the formal sector have access to facilities operated by the Social Security System (IESS). These facilities include rural health posts, regional hospitals that provide both ambulatory care and a limited number of beds, and larger tertiary hospitals. But the quality of service in public facilities has declined due to funding shortfalls. Moreover, the quality of care in MOH and IESS facilities is not the same; in rural areas, Social Security clinics provide better care than Ministry of Health clinics [<xref ref-type="bibr" rid="B58">58</xref>]. In either case, health care in the public sector is largely curative rather than preventive, and given poor living conditions and stagnant incomes, as well as the institution of user fees, most of the rural and urban poor are unlikely to be screened for cardiovascular conditions such as high blood pressure, those associated with overweight and obesity (especially diabetes), and cancers (such as prostate, cervical, and colorectal) that are largely asymptomatic until critical stages are reached.</p><p>Private facilities include modest local clinics that may be operated by a single physician, as well as state-of-the-art hospitals that provide roughly the same level of care as the best facilities in the world. Such facilities are largely accessible only to Ecuadorians who either have private insurance coverage or can pay the costs out-of-pocket.</p></sec><sec><title>Local alternatives to epidemiologic overlap and globalization</title><p>Public spending for health care in Ecuador reflects the enormous gap between what is needed and what is actually provided. While health inequalities, understood in terms of access and outcomes, remain the hallmark of the Ecuadorian health care system, alternatives have been proposed and implemented at the local level. The rural poor are astute in their ability to assess the causes of poverty and realistic approaches to overcoming it [<xref ref-type="bibr" rid="B7">7</xref>]. Furthermore, as long practiced throughout Latin America, social medicine recognizes the multiple interrelationships between public health and socioeconomic conditions, critically assesses the "premise that societal arrangements of power and property powerfully shape the public's health," and acknowledges the role of external forces, especially the effects of "neoliberal economic policies, such as the North American Free Trade Agreement (NAFTA), which result in economic austerity plans, environmental degradation, and growing intra-and interregional disparities in health" [[<xref ref-type="bibr" rid="B59">59</xref>]: 1989]. Social medicine also includes a strong notion of social justice [<xref ref-type="bibr" rid="B60">60</xref>].</p><p>Local participation optimizes the likelihood of sustainability, particularly since experience shows that in Ecuador, community-based assessments and participation shift responsibility to the communities. The community-based approach represents a practical and viable alternative to planning, implementing, and evaluating actions that respond to local needs, especially in partnership with local NGOs and universities [<xref ref-type="bibr" rid="B61">61</xref>]. The importance of local control is officially recognized in Ecuador, which like many other countries has undertaken a process of decentralization supported by legislation and regulation. The basic tenet of this transformation is the assignation of responsibilities–and funds–to local and provincial jurisdictions. But not all local authorities have the capacity or experience to manage health systems and other sources of funds, especially taxes, are often lacking at the local level [<xref ref-type="bibr" rid="B62">62</xref>].</p><p>In spite of the obstacles, experiences in local planning and implementation of health care services–successful, partially successful, and even ultimately unsuccessful–suggest that alternatives to inefficient, centralized services may represent at least a partial solution that not only can successfully address pressing health problems, but also empower local populations.</p><p>With respect to financing, examples of decentralized health insurance include the following.</p><p>• While the national plan for universal health insurance has stagnated, local examples suggest that when local capacity and political will are present, health coverage can be enhanced substantially. In mid-2005, the Metropolitan District of Quito launched its Metropolitan Health Insurance program. Beginning with only 79 affiliates, the program had 5,000 July 2005 and 12,200 by January, 2006. The system has integrated existing groups as well as individuals and provides for services in 40 clinics. Affiliates pay $3.00 per month, for which they receive services up to a value of $1,000. Preventive care is provided, including prenatal care and growth monitoring of children under the age of five, as well as surgery, other curative care, and hospitalization. The goal of the program is to cover 25,000 by the end of 2006 out of a target population of 300,000 [<xref ref-type="bibr" rid="B63">63</xref>].</p><p>• In Guayaquil, the Program of Popular Insurance was inaugurated in January 2006 and in its first week covered 50,000 of 135,000 potential beneficiaries. It provides for health care in 45 centers [<xref ref-type="bibr" rid="B64">64</xref>].</p><p>• Community-based health insurance is combined with the provision of health services in subcenters (<italic>Jambi Huasi</italic>) in the provinces of Cotopaxi, Tungurahua, Cañar, Azuay, Pichincha, Guayas, and Napo. Support is provided by local and international NGOs, universities, multilateral organizations including the World Bank. One analysis [<xref ref-type="bibr" rid="B65">65</xref>] concludes that membership in prepaid health plans was limited, but that this system represents a potentially important vehicle for developing local capacity. An important aspect of the <italic>Jambi Huasi </italic>system is that it protects cultural and linguistic features of local communities by combining western and traditional medical treatment. For example, in the largely indigenous town of Otavalo, nearly 10,000 people had used the <italic>Jambi Huasi </italic>services by 1998, and about half used traditional healers. Quechua-language services provided in the clinic and in the field increased awareness of reproductive health issues, with the result that contraceptive rate increased from 10 percent to 40 percent, while both infant and maternal mortality rates declined [<xref ref-type="bibr" rid="B66">66</xref>].</p><p>• A decentralized, private health plan was less successful. The Pedro Vicente Maldonado Hospital, located in the semitropical region of western Pichincha Province, offered low-cost, prepaid health insurance. For thirty dollars per year, adults could receive five consultations, two emergency room visits, seven days of hospitalization, a 25 percent discount in the cost of surgery, all prenatal exams, all costs related to childbirth, care for newborns, two dental visits, preventive care for diabetes and hypertension, a 50 percent discount in the purchase of medicines, a 50 percent discount in the cost of X-rays, a 25 percent discount on all exams, all costs related to the treatment of snake bite and related to stabilizing traumas, and a 50 percent discount in ambulance fees. A similar program was available to children for an annual fee of 15 dollars. This plan ultimately failed, though, because few local residents enrolled in the plan.</p><p>Examples of local health systems and local participation in addressing specific health problems also suggest that response at this level is a viable alternative:</p><p>• Under the leadership of an indigenous mayor (now nationally prominent) a collective approach to public health in the northern highland town of Cotacachi began with the formation of a broad-based health committee in 1996. A commission with representation from the public health and education sectors as well as local organizations planned a health survey, trained interviewers, and conducted a diagnostic survey based on problems identified by the community. Cotacachi has since developed its own plan to meet the Millennium Development Goals [<xref ref-type="bibr" rid="B67">67</xref>].</p><p>• Community health campaigns supported by public and private alliances are increasingly common. For example, in Cotacachi, a recent campaign supported by a local hospital, a local foundation, and local communities provided a variety of services (dental, preventive care for hypertension and other conditions, prenatal care, cancer screening, and vaccinations) to nearly 3,500 people [<xref ref-type="bibr" rid="B68">68</xref>].</p><p>• A community-based surveillance system was critical in eliminating yaws in Esmeraldas Province [<xref ref-type="bibr" rid="B69">69</xref>].</p><p>• A gender-based approach to community development has been employed to empower poor urban women in Guayaquil, including the establishment of their own health center [<xref ref-type="bibr" rid="B70">70</xref>].</p></sec><sec><title>Summary</title><p>In the first years of the millennium, the Ecuadorian health care system is at a crossroads. From a policy perspective, it is apparent that the government is ill-prepared to assume the responsibility for planning and financing care for an expanding elderly population. In July 2004, pensioners instituted a hunger strike when demands for increases in payments were ignored, and when the government did react, its proposal was to increase the national sales tax. The pensioners took such extreme measures (there were 17 fatalities) because the majority receive less than 100 dollars per month, and a substantial portion receive less than 50 dollars a month.</p><p>The epidemiologic overlap places the country in a double bind; not only is the risk of infectious and communicable diseases inadequately addressed, but the opportunities for timely screening and treatment of chronic and non-transmissible diseases and other modern health problems are extremely limited. For example, the rates of cancer incidence and mortality present challenges that can only grow in the future. First, it is difficult to interpret existing data. Rates of morbidity and mortality associated with different forms of cancer, for example, are almost certainly underestimated due to low levels of screening and correct diagnosis. Particularly among the poor, it seems very likely that a high proportion of cases go undetected. Second, even when cancer is detected, barriers to care (whether economic, cultural, or logistical) mean that a high proportion of cases very likely present at advanced or aggressive stages of the disease, meaning that rates of survival would be lower than expected. Studies in the United States reveal that barriers to care have just that effect among Latinos [<xref ref-type="bibr" rid="B71">71</xref>] even when objectively, screening and treatment services are much more widely available than in Ecuador.</p><p>In human terms, this means that in Ecuador and elsewhere in the Third World (among the poor in particular) men and women are sick and even dying without knowledge of their conditions and without access to even the most rudimentary screening and treatment services. Many forms of cancer are relatively easily treated in their early stages, but in many of these forms (including cervical, colorectal, and prostate cancers) early stages are asymptomatic. More effective screening programs are required, especially since as the population continues to age, prevalence rates can be expected to rise.</p><p>Nevertheless, the major obstacles to effective screening programs for cancer, cardiovascular disease, diabetes, and other "modern" conditions are poverty and inequality, which are problems that globalization does not address, and to the extent that attention is paid to economic and financial integration through enhancing the export sector, the effects may even be negative.</p><p>Fortunately, there is a long history and tradition of social participation in Ecuador that represent the potential basis for shaping the forces of both globalization and epidemiologic overlap. It is interesting to note that demands for decentralization as the most appropriate public policy response (in health and other sectors), come from both sides of the political spectrum. This unusual convergence is more apparent than real, though, since conservative models of decentralization focus on weakening the State while participatory, community-based alternatives are based on democratic principles of local participation as well as efficiency and effectiveness [<xref ref-type="bibr" rid="B72">72</xref>]. The removal of three successive democratically-elected presidents (Mahaud, Bucaram, and Guitierrez) in less than seven years only exacerbated systems of political patronage that have impeded the development of a coherent approach to the challenges presented by epidemiologic transition and overlap within the broader context of globalization.</p><p>Local control of health care is by no means a panacea. Economies of scale are limited or absent, and human resources are unevenly distributed. Local authorities are not by definition more committed to listening to local voices or addressing local needs (or less corrupt) than national authorities. Nevertheless, their ability to shape national policy (for example in participating in global alliances for solving health problems) and organize services represents a viable alternative.</p><p>Important challenges lie ahead. For example, few countries in the world have adequately addressed looming problems associated with modern health conditions. In the coming years, diabetes and related conditions will become so prevalent that it will no longer be possible to ignore them. In addition to the conditions mentioned above, those related to aging, such as Alzheimer's disease–and mental conditions in all age groups–will also be of increasing concern. The ability of local authorities (probably in new alliances with international organizations, national authorities, and even the private sector) to deal with these problems will be a major concern in the mid-21<sup>st </sup>century.</p></sec><sec><title>Competing interests</title><p>The author(s) declares that he has no competing interests.</p></sec>
|
Evaluating the effectiveness of a radiation safety training intervention for oncology nurses: a pretest – intervention – posttest study
|
<sec><title>Background</title><p>Radiation, for either diagnosis or treatment, is used extensively in the field of oncology. An understanding of oncology radiation safety principles and how to apply them in practice is critical for nursing practice. Misconceptions about radiation are common, resulting in undue fears and concerns that may negatively impact patient care. Effectively educating nurses to help overcome these misconceptions is a challenge. Historically, radiation safety training programs for oncology nurses have been compliance-based and behavioral in philosophy.</p></sec><sec sec-type="methods"><title>Methods</title><p>A new radiation safety training initiative was developed for Memorial Sloan-Kettering Cancer Center (MSKCC) adapting elements of current adult education theories to address common misconceptions and to enhance knowledge. A research design for evaluating the revised training program was also developed to assess whether the revised training program resulted in a measurable and/or statistically significant change in the knowledge or attitudes of nurses toward working with radiation. An evaluation research design based on a conceptual framework for measuring knowledge and attitude was developed and implemented using a pretest-intervention-posttest approach for 15% of the study population of 750 inpatient registered oncology nurses.</p></sec><sec><title>Results</title><p>As a result of the intervention program, there was a significant difference in nurse's cognitive knowledge as measured with the test instrument from pretest (58.9%) to posttest (71.6%). The evaluation also demonstrated that while positive nursing attitudes increased, the increase was significant for only 5 out of 9 of the areas evaluated.</p></sec><sec><title>Conclusion</title><p>The training intervention was effective for increasing cognitive knowledge, but was less effective at improving overall attitudes. This evaluation provided insights into the effectiveness of training interventions on the radiation safety knowledge and attitude of oncology nurses.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Dauer</surname><given-names>Lawrence T</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Kelvin</surname><given-names>Joanne F</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Horan</surname><given-names>Christopher L</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>St Germain</surname><given-names>Jean</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Medical Education
|
<sec><title>Background</title><p>Devine and Doyle[<xref ref-type="bibr" rid="B1">1</xref>] identified barriers to effective radiation therapy treatment for carcinoma that included staff's fears and misconceptions associated with radiation. Akers[<xref ref-type="bibr" rid="B2">2</xref>] points out that healthcare personnel, particularly those of childbearing age, are concerned about occupational exposures as they relate to fertility and pregnancy. These fears stem primarily from misconceptions and misunderstandings of radiation and the lack of knowledge of the effects of radiation[<xref ref-type="bibr" rid="B3">3</xref>]. Fear of radiation is highly communicable[<xref ref-type="bibr" rid="B2">2</xref>] and can negatively affect patient care. This concern is especially relevant to cancer patients who often search out or receive education about the risks and benefits of radiation treatments. Prior to entering the hospital, patients have generally come to accept the principle that the benefit of radiation treatment far outweighs the possible risks.</p><p>An understanding of radiation safety principles and their application in practice is critical for all oncology nurses. However, misconceptions about radiation are common, causing undue fears and concerns that may negatively impact patient care. Jankowski[<xref ref-type="bibr" rid="B4">4</xref>] has reported that nurses' fears about their exposure to radiation can be greatly reduced through education.</p><p>A multidisciplinary team of medical professionals including nurse leaders from radiation oncology and radiology, a nurse educator, clinical nurse specialists from inpatient units that commonly house patients admitted for radiation treatment, and radiation safety staff at MSKCC has developed a successful educational intervention addressing radiation safety knowledge and attitudes of nursing staff. A systematic evaluation of the efficacy of training methods for radiation safety education in oncology nursing has not been documented prior to this study.</p><p>This study evaluated potential changes in nursing knowledge and attitudes with regard to the radiation safety program. The presumed cause of any possible differences in pretest versus posttest results was considered to be the training intervention[<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>] (Figure <xref ref-type="fig" rid="F1">1</xref>). Two different dependent variables were evaluated. The first was the cognitive knowledge that nurses display about the required radiation safety program. The second was the personal attitudes of nurses with regard to radiation and the radiation safety program at the hospital.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Design</title><p>Pretest-intervention-posttest designs are uniquely appropriate for investigating the effects of educational innovations[<xref ref-type="bibr" rid="B7">7</xref>] and are commonly used in educational research [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. Strict experimental designs suggest the use of a two-group pretest-intervention-posttest design with a control group that receives no training intervention and a group that receives the training. In the occupational training environment required to meet regulatory compliance requirements, it is often not possible to allow differential services (i.e. different levels of training) for the staff. This was the case for radiation safety training at our cancer center where it was decided not to have a control group because the withholding of training from the control group would represent a differential service inconsistent with a worker's 'right to know' about the potential hazards. In this case, a one-group (no control) pretest-intervention-posttest experimental design was utilized. The absence of a control group was not considered a significant threat to the internal validity of the experiment because the likelihood that extraneous factors account for the change was small[<xref ref-type="bibr" rid="B5">5</xref>]. It was assumed that in the absence of the intervention (i.e. the specific radiation safety training programs) there were minimal or nonexistent outside variables that would have significantly changed a nurse's cognitive knowledge or attitude with regard to radiation safety regulations and policies over the pretest to posttest interval timeframes. It was concluded that the use of an experimental design without a control group was justified.</p></sec><sec><title>Setting</title><p>The study was carried out in 2004 for inpatient staff registered nurses working with oncology patients at MSKCC, a National Cancer Institute designated comprehensive cancer center in New York City.</p></sec><sec><title>Sample</title><p>All inpatient staff registered nurses (750 nurses) were considered as the study population. Participants were recruited by nurse educators and nurse leaders. A total of 15% of the registered nurses (i.e. 113 nurses) completed the study pretests and posttests. All nurses received the same training information.</p></sec><sec><title>Intervention</title><p>The intervention consisted of a multifaceted set of improvements (Figure <xref ref-type="fig" rid="F1">1</xref>) including: nursing procedure revisions, a core concepts video, two types of inservice training, and enhanced "radiation precaution" signs and labels.</p><p>Radiation related nursing procedures were revised to maintain a consistent format and were edited to include only information that was considered to be essential information for nursing care situations. The procedures were validated by nursing and radiation safety experts for both scope and clarity.</p><p>A twelve-minute digital video was developed for incorporation into nursing orientation training and annual mandatory training for all nurses. Video topics were selected specifically to address the perceived knowledge gaps as well as to directly address fears of radiation. The video cast included a narrator, a professional male actor chosen by the nurse members of the multidisciplinary team, and representatives of the nursing and radiation safety staff community depicting various radiation safety precaution actions and discussions. An effort was made to include as many staff members as possible to heighten interest in the movie, help validate the material, and instill a sense of ownership. The video was digitized to facilitate presentation during training workshops as well as making the video available on the MSKCC intranet. Hyperlinks from the nursing procedures website allowed nurses to review the video at any time from virtually anywhere in the hospital.</p><p>Two types of inservice training sessions were held. The first was an interactive, hands-on workshop developed specifically for nursing leadership (managers, clinical nurse specialists, nurse practitioners, and nurse educators) with the goal of increasing their ability to problem solve in the area of radiation and radioactive precautions. The second was a didactic session for nursing staff on units that regularly housed patients requiring radioactive precautions. This inservice focused on the various radiation treatments and the associated precautions most likely to be experienced by nurses.</p><p>Highly visible "radiation precaution" door signs and chart labels were developed to specifically detail the required actions by nursing staff for maintaining the safety of patients and hospital staff while providing care for patients. The improved signs and labels better differentiated among the precautions required for permanent implants, temporary implants, and radiopharmaceutical therapy than the signs previously used.</p></sec><sec><title>Instruments</title><p>Two instruments were designed for this evaluation. The first was designed to measure cognitive knowledge of radiation and radiation protection practices. The second was designed to measure attitudes of nurses with regard to radiation.</p><p>The present research relied upon a criteria-based multiple-choice/true-false test of cognitive knowledge [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B16">16</xref>]. The same instrument was utilized for both pretest and posttest evaluations. The cognitive test included 15 questions with 4 choices for each question. Face validity for the knowledge instrument was assessed by a local group radiation safety specialists and nursing leaders. Each question was scrutinized to ensure that it represented an accurate measure of desired parameters. Content validity was assessed through the use of radiation safety subject matter experts to ensure that proper topical coverage had been afforded by the overall test. [See <xref ref-type="supplementary-material" rid="S1">additional file 1</xref>].</p><p>The cognitive questions addressed the following areas of knowledge: background radiation dose, annual limit, when to wear a badge, how to find exposure records, declared pregnant limit, radiosensitivity of the fetus, external beam treatment, seed implant treatment, visitor precautions, temporary implant treatment, permanent implant treatment, systemic radioiodine precautions, systemic radioiodine contamination, monitoring patient dose rate, and contamination cleanup protocols.</p><p>A Likert-scaled attitude evaluation included 9 questions. Face validity was assessed by a local group of radiation safety specialists and nursing leaders. [See <xref ref-type="supplementary-material" rid="S1">additional file 1</xref>]. The instrument addressed the following attitudinal areas: I feel that radiation safety policies are clear, I know whom to contact for information, I know what steps to take, I can explain precautions well, I feel safe, policies are based on regulations, I am monitored, there is oversight, and I feel safe to have a child.</p></sec><sec><title>Data analysis</title><p>Data were captured from the response sheets using the ReMark Office OMR software[<xref ref-type="bibr" rid="B17">17</xref>]. Two levels of data analysis were performed on the aggregate of pretest and posttest responses. The first used descriptive statistics and the second employed hypothesis evaluation statistics. The significance alpha level was chosen for all statistical tests to be 0.05, the most typical value for social research[<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B18">18</xref>].</p></sec></sec><sec><title>Results</title><sec><title>Descriptive findings</title><p>Descriptive statistics including the number of tests graded, number of graded items, total score possible, maximum score, minimum score, median, mean, variance, and standard deviation were calculated for both the pretest and posttest cognitive scores (Table <xref ref-type="table" rid="T1">1</xref>). The mean score on the cognitive test rose from 58.9% in the pretest to 71.2% in the posttest.</p><p>The frequency distribution and the cumulative description of the cognitive test scores are shown in Figures <xref ref-type="fig" rid="F2">2</xref> and <xref ref-type="fig" rid="F3">3</xref>. Figure <xref ref-type="fig" rid="F2">2</xref> displays the results of the pretest which show a typical Gaussian result with scores centered around 60%. Figure <xref ref-type="fig" rid="F3">3</xref> displays the results of the posttest and demonstrates scores skewed to the right (i.e. shifted toward higher scores).</p><p>To assist in describing the attitude evaluation responses, a weighted average Likert score was calculated for the pretest and posttest attitude questions (Figure <xref ref-type="fig" rid="F4">4</xref>). For 8 out of 9 of the questions, the attitudes were generally positive, i.e. the weighted Likert score was greater than 3, for both the pretest and posttest responses. Responses to the question on oversight were generally negative, (i.e. the weighted Likert score was less than 3, for both the pretest and posttest. Figure <xref ref-type="fig" rid="F4">4</xref> also shows that the weighted Likert score was higher in all cases for the posttest evaluation when compared to the lower scores in the pretest evaluation.</p></sec><sec><title>Hypothesis evaluation findings</title><p>The t-test was utilized to test if as a result of the intervention program there would be no significant difference in nurse's cognitive knowledge as measured with the cognitive test instrument. Table <xref ref-type="table" rid="T2">2</xref> summarizes the results of the t-test analysis performed on the pretest and posttest data. The mean score for the 113 pretest results was 58.9% with a variance of 157.1. The mean score for the 113 posttest results was 71.6% with a variance of 385.7. The calculated t-value was 5.56, with a degree of freedom of 224 and a p = 1.85 × 10<sup>-7</sup>. This evaluation suggested that as a result of the intervention program, there was a significant difference in nurse's cognitive knowledge as measured with the instrument from pretest to posttest (i.e. nurses scored better on the posttest).</p><p>To evaluate the significance of differences on individual questions from pretest to posttest, t-test evaluations were performed on the results of each question. Table <xref ref-type="table" rid="T3">3</xref> lists the results of these t-tests. This evaluation suggested that as a result of the intervention program, there was a significant difference in nurse's cognitive knowledge on eight questions, as measured with the instrument from pretest to posttest. The evaluation also suggested that any differences observed in the remaining questions were not statistically significant.</p><p>A one-way chi-square test analysis for each of the attitude evaluation responses was performed and resulted in p-values that were all less than 0.05 (Table <xref ref-type="table" rid="T4">4</xref>) demonstrating that the responses of each pretest and posttest attitude question differed significantly from chance and represented a measure of the attitudes of the nurses with regard to radiation safety.</p><p>Because the attitude evaluations were not the result of chance alone, the degree of difference between pretest attitudes and posttest attitudes was therefore evaluated. As observed in Figure <xref ref-type="fig" rid="F4">4</xref>, the posttest weighted Likert result was higher than the pretest weighted Likert result for all questions. The two-way chi-square test was used to evaluate the significance of these differences. Table <xref ref-type="table" rid="T5">5</xref> summarizes the results of the two-way chi-square tests. For each question, the table lists the calculated two-way chi-square test statistic, the degrees of freedom for each question, the p-value associated with these two-way chi-square test statistics, and whether or not the p-value was significant.</p><p>There was a significant relationship between the nurses' attitudes toward the radiation safety program and participation in the intervention program for the attitude survey questions concerning the following: I feel that radiation safety policies are clear, I know what steps to take, I can explain precautions well, I feel safe, and policies are based on regulations. Although the weighted average Likert score was higher in the posttests, the differences in pretest and posttest responses were not statistically significant for the following attitude survey questions: I know whom to contact for information, I am monitored, there is oversight, and I feel safe to have a child.</p></sec></sec><sec><title>Discussion</title><p>For the cognitive test, the pretest mean score was 58.9%; the posttest mean score was 71.6%; and the calculated t-value was significantly larger than the critical t-value. The evaluation suggested that as a result of the intervention program, there was a significant difference in nurses' cognitive knowledge.</p><p>The change in percent correct for each individual cognitive question from pretest to posttest (see Table <xref ref-type="table" rid="T3">3</xref>) demonstrated that nurses scored better on thirteen of the fifteen questions. The changes were statistically significant for eight of these questions. Although the performance on the remaining seven questions did not change significantly, in almost all cases, greater than 75% of the nurses chose the correct answer on the pretest. This suggests that about 3 out of 4 nurses possessed a fundamental knowledge of these seven question areas prior to the intervention. In both the pretest and the posttest, less than half of the nurses correctly answered the question on the numerical annual limit of radiation exposure, suggesting that the intervention failed to significantly improve knowledge in this area.</p><p>Two-way chi-square test analyses of the attitude evaluation results demonstrated that there was a significant relationship between the radiation safety training intervention and the increase in agreement with 5 out of 9 of the attitude questions (Table <xref ref-type="table" rid="T5">5</xref>). For the attitude evaluation survey, the Likert scale responses demonstrated that for 8 out of 9 of the questions, the attitudes were generally positive for both the pretest and posttest evaluations. The question on oversight (I feel that I will be called if I receive higher than normal exposures) was the only choice in which nurses generally disagreed with the statement. The weighted Likert score was higher, shifted to an improved, more agreeable attitude, in all cases for the posttest evaluation when compared to the lower scores in the pretest evaluation.</p></sec><sec><title>Conclusion</title><p>Research on the outcomes of educational improvement interventions can be utilized to strengthen the theoretical basis for required regulatory training as well as to validate interventions. The radiation safety training intervention resulted in a statistically significant increase in the cognitive test scores. Therefore, it can be concluded that the elements of the training intervention successfully increased the general knowledge of radiation safety principles associated with oncology practices. It can also be concluded that while nursing knowledge levels significantly increased for more than half of the knowledge areas covered by the cognitive test, the measured knowledge levels did not comprehensively increase in all areas.</p><p>The pretest results of the attitude measures demonstrated that oncology nurses displayed a generally positive attitude with regard to radiation and radiation safety, even before the training intervention. Based on the data analysis of both the pretest and posttest attitude measures, nursing attitudes became more positive after the interventions. A statistically significant increase in survey responses was observed in 5 of the 9 of the questions. Therefore, it can be concluded that there was a relationship between the training intervention and the increase in attitude responses for some but not all of the areas.</p><p>When viewed together, the significant increase in cognitive knowledge and the mixed results of the attitude evaluation suggest that the training interventions were more successful at increasing knowledge and less successful at changing attitudes.</p><p>This study was limited by scope in both the specific nature of the training subject and the nature of the population. The study assumed that the choice of data gathering instruments was appropriate for the task at hand. As this study utilized voluntary participation rather than specific random sampling, extensions of these conclusions to other populations or individuals are understandably weakened.</p><sec><title>Recommendations for general practice</title><p>The training intervention has been incorporated into ongoing training programs. In addition, several strengths of the program can be adopted for use in other training program improvements. The involvement of stakeholders was essential for the development of the training intervention as well as the development and implementation of the research design. The use of a multidisciplinary team for the development of the training intervention elements resulted in high quality training tools that have been shown to be successful with the target audience.</p><p>The attitude evaluation results demonstrated that four areas (I know whom to contact for information, I am monitored, there is oversight, and I feel safe to have a child) should be emphasized in future radiation training intervention implementations. The training elements focused strongly on cognitive knowledge, with the assumption that an increase in knowledge would result in a concomitant improvement of attitudes. It may be possible to develop an additional training element that specifically addresses underlying assumptions and fears. Such an intervention might utilize open discussions or hands-on approaches. The addition of a behavioral psychologist to the multidisciplinary team may improve the outcome.</p></sec><sec><title>Recommendations for future research</title><p>Further research is needed to focus on specifics. While this study evaluated the impact of the entire multi-element training intervention, no conclusions could be drawn for the individual elements of the intervention such as the video, the inservice training sessions, the policies and procedure, or the door signs. Which of these intervention elements had the most impact on increasing cognitive knowledge and/or attitudes is conjectural. The present study was not stratified by age, experience level, gender, hospital unit, or floor, to see if we could identify a nursing sub-population which is more difficult to educate.</p><p>Further research could broaden the study. While this study evaluated the performance and attitude of registered nurses on inpatient floors, a comparison of intervention results for other groups such as registered nurses versus nursing assistants may help to answer the question as to whether education level would affect outcomes. The study should also be extended to other practice settings.</p><p>While the present study utilized a specific set of attitude evaluation questions that concentrated on what the multidisciplinary team believed represented appropriate concerns of oncology nurses, all of the specific fears of the nurses were not requested or learned during this study. An expanded attitude evaluation survey and questionnaire could be beneficial in identifying real fears with respect to radiation. The intervention elements themselves could then be modified to include other methods designed to specifically address identified fears and/or misconceptions.</p></sec></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>LD conceived of the study, collected the bulk of the data, and drafted the manuscript. JK, CH, and JS participated in the design and coordination of the study. JS and JK revised the nursing procedures. All authors were members of the multidisciplinary team that designed the training intervention elements and test instrument. All authors also read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1472-6920/6/32/prepub"/></p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p><bold>Memorial Sloan-Kettering Cancer Center registered nurse radiation safety questionnaire</bold>. This file contains the complete cognitive evaluation and attitude assessment questionnaire. The file is included as a standard pdf document.</p></caption><media xlink:href="1472-6920-6-32-S1.pdf" mimetype="application" mime-subtype="pdf"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Continuous admission to primary school and mental health problems
|
<sec><title>Background</title><p>Younger children in a school class have higher rates of mental health problems if admission to primary school occurs once a year. This study examines whether this relative age effect also occurs if children are admitted to school continuously throughout the year.</p></sec><sec sec-type="methods"><title>Methods</title><p>We assessed mental health problems based on parent-reports (using the Child Behavior Checklist, CBCL) and on professional assessments, among two Dutch national samples of in total 12,221 children aged 5–15 years (response rate: 86.9%).</p></sec><sec><title>Results</title><p>At ages 5–6, we found a higher occurrence of mental health problems in relatively young children, both for mean CBCL scores (p = 0.017) and for problems assessed by child health professionals (p < 0.0001). At ages 7–15, differences by relative age did not reach statistical significance.</p></sec><sec><title>Conclusion</title><p>Continuous admission to primary school does not prevent mental health problems among young children, but may do so at older ages. Its potential for the prevention of mental problems deserves further study.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Reijneveld</surname><given-names>Sijmen A</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Wiefferink</surname><given-names>Carin H</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Brugman</surname><given-names>Emily</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Verhulst</surname><given-names>Frank C</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Verloove-Vanhorick</surname><given-names>S Pauline</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Paulussen</surname><given-names>Theo GW</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>Studies have shown that the youngest children in a school year may be disadvantaged by the educational system [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. Drabman and co-workers showed that in Ohio and Mississippi, USA, the youngest children in a class were referred for academic and behavioral problems more often than their older peers, in a clinical sample aged < 10 years[<xref ref-type="bibr" rid="B1">1</xref>]. Goodman and co-workers showed that in the systems in England, Wales, and Scotland, younger children in a school year are at a greater risk of psychiatric disorder than older children, in a large (n = 10,438) community sample aged 5–15 years[<xref ref-type="bibr" rid="B2">2</xref>]. Lien and co-workers found that in Oslo, Norway, more peer problems and psychological stress levels, and lower average school grades occurred among the youngest children in a school year, in a sample of 6,752 adolescents aged 15–16 years[<xref ref-type="bibr" rid="B3">3</xref>]. Finally, Thompson and co-workers showed that suicide under the age of 20 years in Alberta, Canada, occurred more frequently among those who were younger than their class-mates[<xref ref-type="bibr" rid="B4">4</xref>]. An hypothesized explanation for these adverse outcomes among the youngest in a class is that a reduced self-esteem because of relatively lower academic and athletic pursuits is the underlying factor[<xref ref-type="bibr" rid="B5">5</xref>].</p><p>In the school systems of the countries concerned, all children of a specific age are admitted to school at one moment in the year. As a result of this system, a group of children enters the school at the same moment. In that group the youngest children are almost one year younger than the oldest ones. The youngest can be expected to perform worse in that group at the moment of entry, because they are in an earlier developmental stage, both physically and psychologically. Because of that, Goodman and co-workers proposed to introduce a system that sensitizes teachers to the age position of individual children, thereby reducing the likelihood of unrealistic expectations being placed on younger children[<xref ref-type="bibr" rid="B2">2</xref>].</p><p>In the Netherlands, children are admitted to primary school throughout the year, on the first day of the month following their 4<sup>th </sup>birthday. Thus, children enter primary school one by one, which can be expected to sensitize teachers for their relatively young age and to allow them to pay attention specifically to these youngest children. As such, this system realizes the solution proposed by Goodman and co-workers to prevent mental health problems among children, i.e. '.. to sensitise teachers to the age position of individual children within the class, ..' (page 475)[<xref ref-type="bibr" rid="B2">2</xref>].</p><p>However, in the Dutch system all children enter grade 3 at one specific moment, i.e. after the summer holiday if they are 6 on October 1 of the school year concerned. Children thus stay in grades 1 and 2 for a varying period. Based on this system, children born in October will stay 34 months in grades 1 and 2 (from 1 November till about 1 September – in the Netherlands classes start after the summer holidays someday around September 1), but children born in September only 23 months. All children thus have a similar calendar age at entry, but the latter will be younger at the moment of leaving grade 2, implying that they will never be the oldest in a grade 1/2 class. Retention and acceleration are allowed but no data are available on the proportion of children for whom this goes. This could imply that in grades 1 and 2 fewer problems related to children's relative age occur. Whether the prevalence of mental health problems increases after age 6, remains to be seen. If not, a long-lasting reduction in mental health problems may be reached by admitting children to school continuously instead of once a year.</p><p>The aim of this study is to examine whether such a continuous admission to primary school is indeed associated with fewer differences in mental health problems by relative age within a class at ages 5 and 6 (grade 2), and, if yes, whether this effect persists after the age of 6 years.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Participants</title><p>We used data on 12,221 children from two sources, both within the framework of the routine preventive health assessments that are provided regularly to all Dutch children. The first source was a cross-sectional national study in 1997 and 1998 on children aged 5–15 (response rate 90.1%; n = 4,480), representative for the Dutch population[<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. The second was a randomized controlled trial (RCT) in 2001 and 2002 on the improvement of the early detection of psychosocial problems by child health physicians and nurses (child health professionals, CHP) by a training program, among a national sample (exclusive of the big cities) of children aged 5–6 (response rate 85.2%; n = 7,737)[<xref ref-type="bibr" rid="B8">8</xref>]. The latter data source comprised all children that participated in that trial, i.e. the 6,375 children that were analyzed and the 1,477 that were excluded from the primary analysis because of being under treatment or of non-Dutch ethnicity [<xref ref-type="bibr" rid="B8">8</xref>]. of 15 of this total group, data on month of birth were missing.</p><p>In both studies we obtained participants by means of a two-step procedure. First, Child Health Services were asked to participate. And second, the participating CHS were asked to provide data on a specified number of children. In the cross-sectional sample, this second step concerned a team of child health professionals that were asked to provide data on a sample of 75 children for each age group. In the RCT each participating child health professional was asked to provide data on 50 children for each measurement period, 150 in total. Both studies had been approved by the local Institutional Review Board, including verbal informed consent by parents.</p></sec><sec><title>Measurements</title><p>In both studies parents and CHPs filled-out similar questionnaires with the same wording of the items that have been included in this study. Parents first completed the Child Behavior Checklist (CBCL), a well-validated questionnaire on behavioral and emotional problems over the preceding six months[<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. It contains 120 problem items on the basis of which a Total Problems score can be computed. Children were allocated to a normal range or a clinical (elevated) range, using the 90th percentiles of the Dutch normative sample for the validated Dutch version[<xref ref-type="bibr" rid="B10">10</xref>]. Next, the CHPs took a routine history and physically assessed each child, and then completed the following question: "Does the child have a psychosocial problem at this moment?" (yes or no). If a problem was identified, the CHP was asked to rate its severity as mild, moderate, or severe. All participating CHPs had been trained by the research team on the recording and classification of these problems.</p></sec><sec><title>Analysis</title><p>We analyzed whether the occurrence of mental health problems, i.e. mean CBCL Total Problems scores and prevalence of psychosocial problems rated by the CHP as moderate or severe[<xref ref-type="bibr" rid="B8">8</xref>], differed by relative age (old = October 1 – January 31, middle = February 1 – May 31, young = June 1 – September 30), by age group (5–6, 7–12 and 13–15 years, i.e. grades 1–2, and grades 3–8 of Dutch primary school; and the first grades of Dutch secondary school, respectively). These analyses were similar to those of Goodman et al. [<xref ref-type="bibr" rid="B2">2</xref>], except that these used a psychiatric diagnosis whereas we employed assessment by a CHP. We repeated this with adjustment for differences in background of children in the three relative-age groups, which yielded similar p-values in all cases (not shown).</p></sec></sec><sec><title>Results</title><p>Table <xref ref-type="table" rid="T1">1</xref> presents the information on the background of the children in the two samples. We found no differences in background characteristics by relative-age group (range of p-values: 0.062 to 0.850, chi-square tests). Table <xref ref-type="table" rid="T2">2</xref> shows that at ages 5–6 years (grades 1 and 2), differences by relative age in mental health problems existed both for parent-reports on the CBCL (p = 0.017) and for assessments by CHPs (p < 0.0001). At ages 5–6 years, differences by relative age in CBCL Total Problems scores are larger in the RCT data than in the cross-sectional data whereas regarding CHP assessed problems they were slightly larger in the cross-sectional data. For the cross-sectional data, differences at ages 5–6 years were not statistically significant, due to smaller numbers of children involved in that study. At older ages, differences by relative age diminished and statistical significance decreased further.</p></sec><sec><title>Discussion</title><p>Our results show that shortly after school entrance the association between the relative age of children in a class and mental health problems was similar in a system in which children were admitted to primary school continuously throughout the year and in a system in which admission occurred once a year[<xref ref-type="bibr" rid="B2">2</xref>]. After the age of 6 years, continuous admission seems to have some favorable effects on these relative age effects, though.</p><sec><title>Explanations</title><p>The existence of relative age effects in our study may be explained as a lack of support for the contribution of a lower self-esteem of relatively young children to the development of mental health problems among them. The one by one entrance of children in a school class might allow teachers to take into account their young relative age, but this seems to have no favorable effects at that time. An alternative explanation, an earlier physical and psychological developmental stage seems thus to be more likely, because of the independence of this relative age effect from the kind of system of school entry in various countries. We can not exclude some favorable effects of a continuous admission, though, since in our study differences by relative age diminished at older ages whereas they persisted in a system of admission once a year[<xref ref-type="bibr" rid="B2">2</xref>]. Finally, it should be realized that children born in October (immediately after the cut-off for admission to grade 3 once a year, October 1) or in one of the following months, stay in grade 1/2 for a longer period than those born in September or the preceding months. The former will thus be the oldest in a grade 1/2 class at some moment, even though all children enter this class at the same age of 4 years. This may affect the self-esteem of the latter group to some extent, and might explain some of the still visible relative age affects.</p></sec><sec><title>Limitations</title><p>Our results are unlikely to be biased. Data came from two national studies with both very high response rates [<xref ref-type="bibr" rid="B6">6</xref>-<xref ref-type="bibr" rid="B8">8</xref>]. Moreover, we used both well-validated questionnaires and professional judgments[<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>], which limits the likelihood of information bias. Finally, we had a very large sample size, slightly larger than that of Goodman et al. [<xref ref-type="bibr" rid="B2">2</xref>], and (much) larger than the other studies that showed a relative age effect regarding mental health problems[<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>].</p><p>However, the two datasets that we studied have been collected for different purposes, one essentially being a prevalence study[<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>], the other being an RCT on the effect of training CHPs to improve diagnostic quality[<xref ref-type="bibr" rid="B8">8</xref>]. This is unlikely to have biased our results though, for a number of reasons. Both studies used a very similar methodology, were performed by teams of researchers from the same institute with the same coordinator of the data collection in both studies (SAR), and covered the same population (i.e. the entire Netherlands) though weightings by regions differed somewhat.</p><p>Furthermore, the cross-sectional study had a smaller sample size than the RCT which may explain some of the lack of statistical significance for differences by relative age in the former study. However, using only data from the cross-sectional study's absolute differences between the youngest and the oldest in a class also decrease in the higher grades of primary school.</p><p>Moreover, country-specific factors may have influenced results, especially the decrease of relative age effects at older ages in our study. It could be that the Dutch educational system in higher grades is effective in preventing problems at those ages. This especially holds for selective acceleration and retention at the entry of grade 3. This process is likely to be facilitated by the opportunity to observe children for two years in a mixed class, followed by a transition of a fixed cohort to the next year, a process in which retention or acceleration is not that much accentuated as it is in 'ordinary' classes.</p><p>Finally, we did not obtain a formal psychiatric diagnosis. Therefore, our results need confirmation from studies on the use of the two admission systems within one country, preferably using an experimental design to be able to control for other potential confounders like the type of coaching at school or the degree to which children have attended preschool classes, and with psychiatric assessments.</p></sec><sec><title>Implications</title><p>Our study shows the prevalence of mental health problems at ages 5–6 years to be 36% higher for the relatively youngest tertile of children compared to the relatively oldest one. A similar effect exists in other countries, even at older ages. For instance, in the UK the prevalence rate of any psychiatric diagnosis was about 20% higher in the youngest tertile compared to the oldest tertile[<xref ref-type="bibr" rid="B2">2</xref>], and in Alberta the number of suicides in the youngest half was about 25% higher than in the oldest half[<xref ref-type="bibr" rid="B4">4</xref>]. As it concerns all children in a country, any intervention that decreases the size of this relative age effect may have a relatively large effect. The development of interventions specifically aiming at these relatively young children is thus urgently needed. Moreover, a system of continuous admission deserves further study as our results provide some indications that it has favorable effects at older ages. This may be due to selective retention and acceleration at the transition to grade 3, which is probably facilitated by the existence of mixed grade 1/2 classes with continuous entry. Such a selective transition to grade 3 deserves additional study, especially if it is embedded in a system of continuous admission to grades 1/2.</p></sec></sec><sec><title>Conclusion</title><p>Continuous admission to primary school does not prevent mental health problems among young children, but may do so at older ages. Its potential for the prevention of mental problems deserves further study.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>SAR had the original idea for the project, wrote the study protocol and coordinated this study and the research part of both studies that provided the data for this secondary analysis. All authors discussed the protocol and formulated the final design. EB and CHW supervised the data collection of the cross-sectional study and the RCT, respectively. SAR did the statistical analyses, which were discussed by all authors. SAR wrote the final manuscript, which was discussed, edited and revised by all authors. All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/145/prepub"/></p></sec>
|
The acceptability and impact of a randomised controlled trial of welfare rights advice accessed via primary health care: qualitative study
|
<sec><title>Background</title><p>Qualitative research is increasingly used alongside randomised controlled trials (RCTs) to study a range of factors including participants' experiences of a trial. The need for a sound evidence base within public health will increase the need for RCTs of non-clinical interventions. Welfare rights advice has been proposed as an intervention with potential to reduce health inequalities. This qualitative study, nested within an RCT of the impact of welfare rights advice, examined the acceptability of the intervention, the acceptability of the research process and the perceived impact of the intervention.</p></sec><sec sec-type="methods"><title>Methods</title><p>25 men and women aged 60 years or over were recruited from four general practices in Newcastle upon Tyne (UK), a sub-sample of those who consented to be contacted (n = 96) during the RCT baseline interview. Semi-structured interviews were undertaken and analysed using the Framework Method.</p></sec><sec><title>Results</title><p>Participants viewed the trial positively although, despite agreeing that the information leaflet was clear, some had agreed to participate without being fully aware of what was involved. Some participants were unaware of the implications of randomisation. Most thought it fair, but a few concerns were raised about the control condition. The intervention was acceptable and made participants feel confident about applying for benefit entitlements. 14 out of 25 participants received some financial award; median weekly income gain was £57 (€84, $101). The perceived impact of additional finances was considerable and included: increased affordability of necessities and occasional expenses; increased capacity to deal with emergencies; and a reduction in stress related to financial worries. Overall, perceived independence and ability to participate in society increased. Most participants perceived benefits to their mental well-being, but no-one reported an improvement in physical health. The RCT showed little or no effect on a wide range of outcome measures.</p></sec><sec><title>Conclusion</title><p>Participation in the trial and the intervention was acceptable to participants. Welfare rights advice targeted at people aged 60 years or over and accessed via primary care had a positive impact on quality of life and resulted in increased social participation. The divergence of qualitative and quantitative findings suggests that both methods make important contributions to the evaluation of complex social interventions.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Moffatt</surname><given-names>Suzanne</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Mackintosh</surname><given-names>Joan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>White</surname><given-names>Martin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Howel</surname><given-names>Denise</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Sandell</surname><given-names>Adam</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>Qualitative research is increasingly used alongside randomised controlled trials (RCTs) to study factors such as recruitment[<xref ref-type="bibr" rid="B1">1</xref>], participants' experiences[<xref ref-type="bibr" rid="B2">2</xref>], lay understandings of randomisation[<xref ref-type="bibr" rid="B3">3</xref>] and informed consent[<xref ref-type="bibr" rid="B4">4</xref>], and to illuminate reasons why the intervention may or may not have worked[<xref ref-type="bibr" rid="B5">5</xref>]. Much of this work has focused on participant experiences in clinical trials, but the need for a sound evidence base for public health[<xref ref-type="bibr" rid="B6">6</xref>] will increase the need for RCTs of non-clinical interventions. While evidence of outcomes can be derived from RCTs, 'evidence of the process by which those outcomes were achieved, the quality of implementation of the intervention, and the context in which it occurred is likely to come from qualitative data'[<xref ref-type="bibr" rid="B7">7</xref>] (p528).</p><p>Welfare rights advice can lead to significant financial and non-financial gains [<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref>], and it has been proposed as an intervention with significant potential to reduce inequalities in health[<xref ref-type="bibr" rid="B11">11</xref>]. Previous qualitative research has indicated that welfare rights advice accessed via primary care is viewed positively and is perceived to address both social and health needs[<xref ref-type="bibr" rid="B12">12</xref>]. A number of studies of welfare rights advice accessed via primary care indicate the usefulness of such services and their perceived acceptability among recipients, primary care and welfare rights staff [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B17">17</xref>]. However, no robust qualitative work has examined the acceptability and health impact of this social intervention within the context of an RCT among those aged 60 years or over.</p><p>In this qualitative study, undertaken within the context of a pilot RCT reported in an accompanying paper[<xref ref-type="bibr" rid="B18">18</xref>], we aimed to assess: the acceptability of a domiciliary welfare rights advice service accessed via primary care; the acceptability of the research process; and the perceived impact of the intervention, in order to inform future studies.</p></sec><sec sec-type="methods"><title>Methods</title><p>The study took place in parallel with a single blind pilot randomised controlled trial with allocation of individuals to intervention (receipt of welfare rights advice immediately) or control condition (receipt of intervention after a six month delay), the methods of which are described in detail in the accompanying paper[<xref ref-type="bibr" rid="B18">18</xref>]. The intervention comprised a domiciliary assessment of current welfare status and benefits entitlement, active assistance in making claims where appropriate, and follow-up for unresolved claims as required.</p><p>Participants were selected using general practice databases and the selection process is summarised in Figure <xref ref-type="fig" rid="F1">1</xref>. A random sample of patients aged 60 years or over from each of four participating practices in Newcastle upon Tyne (UK) was invited to take part in the trial. The invitation comprised a covering letter signed by the GP together with a leaflet (see table <xref ref-type="table" rid="T1">1</xref>) explaining the intervention and the RCT. Those who consented to participate were interviewed with a structured baseline assessment, at which time they were asked if they would be willing to take part in a further qualitative study for which written informed consent was obtained. The sampling frame for the qualitative study was formed by those (n = 96) who consented to be approached for the qualitative study during the RCT baseline interview. All participants were then randomised into intervention and control groups. The intervention (welfare rights assessment interview) took place approximately two weeks after the baseline assessment for the intervention group and approximately six months after the baseline assessment for the control group. The qualitative study sample comprised respondents from intervention and control groups, purposively selected to include those eligible for the following resources: financial only; non-financial only; both financial and non financial; and none. Sampling continued until no new themes emerged from the data[<xref ref-type="bibr" rid="B19">19</xref>].</p><p>Appointments were arranged by telephone and semi-structured interviews were undertaken by SM between April and December 2003 in participants' homes. The average time between the welfare rights consultation and research interview was ten months for the intervention group and five months for the control group because the control group received their welfare rights assessment six months after the intervention group. The topic guide covered perceptions of the impact of material and/or financial benefits on physical/mental health, health related behaviours, social benefits and the acceptability of the intervention and research process. Interviews were audio-recorded and transcribed in full.</p><p>We undertook thematic analysis following the Framework method,[<xref ref-type="bibr" rid="B20">20</xref>] with constant comparison[<xref ref-type="bibr" rid="B21">21</xref>] and deviant case analysis[<xref ref-type="bibr" rid="B22">22</xref>] to enhance internal validity. Resulting typologies were derived and higher level descriptive and explanatory categories developed.</p><sec><title>Ethical approval</title><p>The protocol was approved by the Newcastle upon Tyne joint universities and NHS research ethics committee.</p></sec></sec><sec><title>Results</title><sec><title>Participants and benefits received</title><p>Twenty-five semi-structured, in-depth interviews were undertaken (14 intervention, 11 control). Ten participants were interviewed with partners who made active contributions. Fourteen participants received some financial award. The median weekly income gain was £57 (€84, $101) (range £10 (€15, $18) to £100 (€148, $178)) representing a 4–55% increase in household income. Eighteen participants were in receipt of benefits, either as a result of the current intervention or because of claims made prior to the study. Although 96 of the total sample (126) agreed to participate, data saturation was reached after interviewing 25 respondents.</p></sec><sec><title>Acceptability and understanding of the research process</title><p>Participation in the RCT required considerable commitment. All participants were interviewed four times over 24 months with the same structured interview schedule and the qualitative sub-sample participated in a further in-depth interview. Participation rates for the RCT remained high (87%) after two years (five deaths; one moved, 11 declined follow-up).</p><p>The research process was a positive experience overall (table <xref ref-type="table" rid="T2">2</xref>). Factors which contributed to this included: enjoying the company of others; learning through participation; the domiciliary nature of the intervention and research; trust in the institutions involved (NHS and University); and building a relationship with the researchers. No one was concerned about the time commitment, divulging personal information or breaches of confidentiality.</p><p>During the qualitative interview, participants re-read and commented on the participant information leaflet. There was unanimous agreement that it was clear, straightforward and also reflected their experiences of being in the study. However, it became apparent that a number of people had agreed to take part in the study without being fully aware of what it entailed. Some participants disclosed that they had not realised there was the potential for gaining additional resources, despite this being clearly stated during invitation and consent. The main reasons given for taking part were: altruism – helping others and highly regarded institutions like the NHS; importance of research for generating new knowledge; and, finding out about welfare benefit entitlements (table <xref ref-type="table" rid="T3">3</xref>). Their GP's signature on the invitation letter was another reason given for participation, as it conferred a degree of legitimacy on the research. Participants also felt they were reciprocating for the care they had received from their GPs, whose involvement in the research, they felt, signified a caring attitude.</p><p>Although the randomisation procedure was fully explained, some participants were not aware of its implications and a number were unaware to which group they were randomised (table <xref ref-type="table" rid="T4">4</xref>). However, when the procedure was explained, most participants, whether in the intervention or control group, thought that randomisation was fair. Two participants did voice reservations about the control condition on the basis that this resulted in a denial of income.</p></sec><sec><title>Views on the intervention</title><p>The intervention was regarded positively by participants, irrespective of whether they gained personally (table <xref ref-type="table" rid="T5">5</xref>). The reasons for this were linked to the fact that this particular welfare rights service did not operate in a conventional way. The assessment was offered rather than sought and any claims arising were as a result of a knowledgeable professional actively assisting respondents to make claims. This effectively sanctioned claims as well as lessening the chances of rejection. Providing the service in the respondent's home meant that it was convenient, safe and particularly accessible for those with sensory or mobility problems. The most common description of the service was 'helpful', but many participants spoke at length about having the chance to talk, and having someone take an interest, suggesting that the welfare consultation itself had a therapeutic effect. Among those who did not qualify for benefits, there was no evidence that their hopes had been raised and dashed by no subsequent claim being made.</p><p>The process of the intervention required the disclosure of a considerable amount of information about personal finances. No-one objected to this in the context of the welfare rights assessment, although one participant did not wish to disclose actual amounts. Most participants saw this as a necessary part of the process, and many were already required to disclose finances for other matters, such as qualifying for council tax rebates. Participants highlighted the fact that the service made them feel relaxed, at ease and confident about applying for benefits.</p></sec><sec><title>Impact of the intervention</title><p>Fourteen participants received additional financial resources. The perceived impact of extra money was considerable and it was used on a wide range of items. From participants' accounts, four linked categories were identified (table <xref ref-type="table" rid="T6">6</xref>). Firstly, increased affordability of <italic>necessities</italic>, without which maintaining independence and participating in daily life was difficult. This included accessing transport, maintaining social networks and social activities, buying better quality food, paying bills, preventing debt and affording paid help for household activities. Secondly, <italic>occasional expenses </italic>such as clothes, household equipment, furniture and holidays were more affordable. Thirdly, extra income was used as a means of dealing with potential <italic>emergencies </italic>and to increase <italic>savings</italic>. Fourthly, all participants described the easing of financial worries as bringing '<italic>peace of mind'</italic>. Figure <xref ref-type="fig" rid="F2">2</xref> summarises the ways in which these categories may be linked.</p></sec></sec><sec><title>Discussion</title><sec><title>Main findings and interpretation</title><p>Participation in this RCT was regarded as a positive experience by most interviewees; the intervention was highly regarded and had wide-ranging impacts, particularly for those who gained new benefits.</p><p>Participants' perspectives in RCTs have generally been measured by using attitude questionnaires with a view to improving participation rates[<xref ref-type="bibr" rid="B3">3</xref>]. Although limited by the questionnaire format, such studies show that the main reasons given for anticipated participation in trials were altruism[<xref ref-type="bibr" rid="B23">23</xref>]and personal benefit[<xref ref-type="bibr" rid="B24">24</xref>], and major disadvantages identified were time involved and problems with travel[<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B26">26</xref>]. Altruism, generating new knowledge through research and personal benefit were the main reasons given for participation in this study. Because of its domiciliary design, the study did not require participants to travel, something that many, particularly those with limited mobility or sensory impairments, appreciated and which may have increased recruitment and reduced attrition. Our study also suggests that the involvement of a trusted health professional in the recruitment can increase participation rates.</p><p>Although every effort was made to ensure that consent was fully informed, it became clear that a number of people did not understand the nature of the intervention or were not fully aware that they were being randomised. This occurred despite ethical committee approved information leaflets that were regarded as clear and a good reflection of participants' experiences in the trial. This has also been found in clinical research[<xref ref-type="bibr" rid="B4">4</xref>]. With the exception of two participants, no-one was unhappy about randomisation to the control condition which involved waiting six months for welfare rights advice. Participants rationalised this in three main ways: firstly that it was a fair way to 'ration' a service that was otherwise not usually available; secondly, that people must face the consequences of their informed consent; and, thirdly, a degree of fatalism and/or ignorance about group allocation. Participants in this study appeared to have a more relaxed attitude to randomisation than has been reported in clinical trials[<xref ref-type="bibr" rid="B3">3</xref>]. This could be explained by differences in the nature and context of clinical and social interventions, the former concerning treatment for an existing condition with anticipated health benefits; the latter, as in this case, concerning access to a potentially valuable service, but with unknown health effects. This study demonstrates that it was acceptable to randomise the intervention with a delay of six months to the control group. However, the pilot RCT findings suggest that this is unlikely to be adequate time for any measurable health benefits to become evident [<xref ref-type="bibr" rid="B18">18</xref>]. A longer delay between the study and control group's receipt of the intervention raises ethical issues that require further consideration [<xref ref-type="bibr" rid="B27">27</xref>].</p></sec><sec><title>Strengths and limitations</title><p>The qualitative data were derived from a sub-sample of the RCT participants and it is possible that only those willing to be interviewed further held such positive views. Comparison between the quantitative and qualitative samples on a number of social and economic variables indicated similarities according to age, proportion of those living alone/as a couple, council tax band and long term limiting illness, suggesting that both samples were not qualitatively different [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B28">28</xref>].</p></sec></sec><sec><title>Conclusion</title><p>The effects of the intervention were wide-ranging and positively regarded by participants. However, this was not mirrored by the outcome measures used in the pilot RCT. Despite measuring a wide range of physical, psychological and social outcomes, the pilot RCT found little or no differences between intervention and control groups, or between those who did and did not receive additional resources suggesting that the intervention had no impact on these outcome measures [<xref ref-type="bibr" rid="B18">18</xref>,<xref ref-type="bibr" rid="B29">29</xref>]. It is not uncommon for qualitative and quantitative studies to produce divergent findings [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B30">30</xref>,<xref ref-type="bibr" rid="B31">31</xref>] and it is likely that each method, with its different epistemological underpinnings, captured different aspects of phenomena under investigation. The qualitative approach enabled participants to give an account of the various ways in which the intervention impacted on their lives, such as increased independence and improved quality of life, which were not explicitly measured in the pilot RCT and are challenging to capture quantitatively[<xref ref-type="bibr" rid="B29">29</xref>]. This qualitative study has thus not only provided valuable information about outcome measures for a future trial, but helped with the overall interpretation of the pilot RCT. The study also suggests that inclusion of a qualitative component will help to illuminate the process and outcome of a future trial.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that we have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>SM and MW with DH and AS initiated the RCT and obtained funding. SM designed the qualitative study, undertook the interviews, analysis and initial interpretation. JM assisted with devising the coding framework. SM, JM, MW, DH and AS jointly undertook final interpretation of the data. SM wrote the first draft of this paper, JM, MW, DH and AS commented on subsequent drafts. All authors approved the final version. All authors are guarantors and accept full responsibility for the conduct of the study and the contents of this paper.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/163/prepub"/></p></sec>
|
Vitamin G: effects of green space on health, well-being, and social safety
|
<sec><title>Background</title><p>Looking out on and being in the green elements of the landscape around us seem to affect health, well-being and feelings of social safety. This article discusses the design of a research program on the effects of green space in the living environment on health, well-being and social safety.</p></sec><sec><title>Methods/design</title><p>The program consists of three projects at three different scales: at a macro scale using data on the Netherlands as a whole, at an intermediate scale looking into the specific effect of green space in the urban environment, and at micro scale investigating the effects of allotment gardens. The projects are observational studies, combining existing data on land use and health interview survey data, and collecting new data through questionnaires and interviews. Multilevel analysis and GIS techniques will be used to analyze the data.</p></sec><sec><title>Discussion</title><p>Previous (experimental) research in environmental psychology has shown that a natural environment has a positive effect on well-being through restoration of stress and attentional fatigue. Descriptive epidemiological research has shown a positive relationship between the amount of green space in the living environment and physical and mental health and longevity.</p><p>The program has three aims. First, to document the relationship between the amount and type of green space in people's living environment and their health, well-being, and feelings of safety. Second, to investigate the mechanisms behind this relationship. Mechanisms relate to exposure (leading to stress reduction and attention restoration), healthy behavior and social integration, and selection. Third, to translate the results into policy on the crossroads of spatial planning, public health, and safety. Strong points of our program are: we study several interrelated dependent variables, in different ordinary settings (as opposed to experimental or extreme settings), focusing on different target groups, using appropriate multilevel methods.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Groenewegen</surname><given-names>Peter P</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>van den Berg</surname><given-names>Agnes E</given-names></name><xref ref-type="aff" rid="I3">3</xref><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>de Vries</surname><given-names>Sjerp</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Verheij</surname><given-names>Robert A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>The briefest summary of our program is in its title: Vitamin G, where G stands for the green space around us. Notions of beneficial effects of nearby green space have persisted throughout history [<xref ref-type="bibr" rid="B1">1</xref>]. Research on this topic, mainly experimental research, has focused on demonstrating the relationship between exposure to green environments and well-being [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B3">3</xref>]. Most of the evidence on health benefits comes from laboratory experiments that exposed participants to photographic simulations of various types of natural environments [<xref ref-type="bibr" rid="B4">4</xref>], or controlled field studies that compared residents with a view of urban greenery to residents without such view [<xref ref-type="bibr" rid="B5">5</xref>]. This research has demonstrated that mere exposure to views of nature can improve people's health and well-being by providing restoration from stress and mental fatigue. Moreover, this research has shown that views of nature can improve feelings of neighborhood safety and even lead to decreases in aggression and crime rates [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. To maximize effects, scientists have selected extreme settings, such as urban areas with nearly no green space at all, and concentrated on stress reduction and attention restoration as the most noticeable outcomes. Theoretical developments have followed this focus, and the dominant theories in the field [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B2">2</xref>] consider stress reduction and restoration as a central causal mechanism. Although this focus on extreme settings and restorative effects has highlighted the importance of green space to well-being, it potentially obscures the scope and underlying mechanisms of these effects. Very little is known about the positive effects of green space on well-being through mechanisms of increased and prolonged physical activity, and improved social cohesion [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Not only causal mechanisms might explain the effects of green space; research in naturalistic settings also has to take into account the possibility of direct or indirect selection [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Vitamin G aims to establish the relationship between the amount and type of green space in people's living environment and their health, well-being, and feelings of safety, to study the mechanisms behind this relationship, and to specify the implications for policy making. Our program differs from previous studies in two respects.</p><p>First of all, field studies will be conducted instead of experimental studies; this is especially important for applied purposes, because it provides a better indication of the relative size of the effects in real-life settings. Field studies also have higher ecological validity, a more direct social relevance, and a focus on long-term rather than on short-term effects (almost inevitable in the case of laboratory studies). Secondly, the focus is on ordinary settings instead of 'extreme' settings in which people are especially stressed or frustrated (hospitals, prisons), or live in extremely poor or barren circumstances (as in some of the previous studies: [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>]). This increases the generalizability of the effects and the relevance to the European and especially the Dutch situation.</p><p>In this article we describe the research questions and background of the program as a whole and the design and methodology of the three separate projects that form the program. Vitamin G started on 1 January 2005 and will run for four years.</p><sec><title>The program vitamin G</title><p>The general problem formulation of our research program is: what is the direction and strength of the relationship between the amount of green space in people's living environment and their health, well-being and perceived safety, how can this relationship be explained and how can the results be made useful for policy intervention? This general question will be answered in three projects at three different scales: at a macro scale using data on the Netherlands as a whole, at an intermediate scale looking into the specific effect of green space in the urban environment, and at micro scale investigating the effects of allotment gardens. The specific problem formulations for each of these three projects are:</p><p>1. How strong is the relationship between the amount of green space in people's living environment and their perceived health and well-being, and feelings of safety and is this relationship stronger for specific population segments and/or types of green space? How can this relationship be explained?</p><p>2. Do urban neighborhoods that differ in the amount and type of green space in the vicinity, also differ in the health, well-being and perceived safety of the people living in these neighborhoods? Have urban neighborhoods that went through a large change in the amount of green space, changed in these respects? If so, which aspects of the green space seem to be the most influential ones?</p><p>3. Is having an allotment garden related to health, well-being and perceived safety in urban dwellers and how can this relationship be explained?</p><p>Our approach to answer these questions is based on analyzing the multilevel relationships between environment and people [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. People live in a shared environment that influences their well-being in a general sense, even partly by virtue of the fact that it is a shared environment (as in the case of social integration). These relationships have been schematized in figure <xref ref-type="fig" rid="F1">1</xref>.</p><p><italic>Exposure </italic>to green space consists of direct physical exposure and the psychological processes through which exposure influences health and well-being. These psychological processes will be further developed, using theories about stress and restoration [<xref ref-type="bibr" rid="B4">4</xref>]. Restorative effects can be achieved by merely looking at nature or natural elements, indicating that the aesthetic experience of nature may play a role in this mechanism. Besides providing relief from stress, an aesthetically attractive living environment may also improve well-being by enhancing satisfaction, attachment, and a sense of responsibility. Related to stress reduction, (American) evidence suggests that exposure to natural environments may reduce feelings of anger, frustration and aggression (e.g., [<xref ref-type="bibr" rid="B6">6</xref>]). In turn, this may enhance feelings of social safety, and even reduce actual rates of aggressive behavior and criminal activity [<xref ref-type="bibr" rid="B7">7</xref>]. Physical exposure to cleaner air may play a role also. Traffic density seems to be the most important source of polluted air in the direct vicinity, while the overall level of air pollution is rather constant in The Netherlands [<xref ref-type="bibr" rid="B18">18</xref>].</p><p>The <italic>behavioral mechanism </italic>will be developed, using sociological theories about life style, combining structural aspects (socio economic status) and opportunities (availability, social integration) and choices people make (behavior) [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Natural environments are perceived as more attractive than built environments. Because of this, green areas may stimulate residents to undertake healthy physical activities such as walking or cycling or to choose these activities as a mode of transport, and to spend more time in them [<xref ref-type="bibr" rid="B21">21</xref>]. Attractive green areas in the neighborhood may serve as a focal point of tacit coordination for positive informal social interaction, strengthening social ties and thereby social cohesion [<xref ref-type="bibr" rid="B15">15</xref>]. Social cohesion by itself is thought to have a positive effect on well-being and feelings of safety. It is important to distinguish the effect of green space from that of population density or urbanicity, which has an established relationship with (mental) health [<xref ref-type="bibr" rid="B22">22</xref>].</p><p>Apart from these causal mechanisms, part of the effect may be the result of <italic>selection</italic>. Selective migration to or retention in particular living environments might explain part of the relationship between green space in people's living environment and their health. Direct selection occurs, when people's well-being influences their chances of living in a favorable environment; indirect selection, when people with certain characteristics, such as a high income, that are related to well-being can afford to live in a favorable environment [<xref ref-type="bibr" rid="B23">23</xref>]. Migration flows in general are related to such socio-demographic characteristics as age, income and education [<xref ref-type="bibr" rid="B24">24</xref>]. Consequently, indirect selection might play a role in explaining relationships between the amount of green space in people's living environment and health and well-being. It is therefore important to take into account and control for the possibility of selection.</p><p>The three questions formulated above, will be answered in three projects.</p><p>The first question will be addressed in a macroscopic project, establishing the strength of the relationships and testing hypotheses about the mechanisms that explain the association between green space and health, well-being, and feelings of safety. Recent epidemiological research by our team has shown a relationship between a green living environment and perceived health indicators in a large population sample [<xref ref-type="bibr" rid="B25">25</xref>]. This was the first study in the general population, showing that this relationship was not exclusive to extreme and controlled settings. People living in greener areas tend to perceive their physical and mental health status as better than their counterparts living in less green areas (controlling for socio-economic and demographic spatial clustering). Whether such a positive relationship will also be found when looking at other health indicators is not known. The same applies to people's feelings of safety, which could even be negatively influenced by the presence of green space in one's living environment (because the lack of social control may turn urban green spaces into 'hot spot' for criminal activities).</p><p>The second project, addressing the second question, zooms in on the urban environment. Green space is scarcer in urban areas and access to it might be more skewed. Several studies have demonstrated positive relationships between the presence of greenery in urban neighborhoods and residents' health, well-being, and social safety [<xref ref-type="bibr" rid="B25">25</xref>,<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B26">26</xref>]. These relationships have been explained mostly through the mechanism of stress reduction. Indeed, there is some evidence that exposure to local greenery in an urban context may reduce stress and mental fatigue. For example, Honeyman [<xref ref-type="bibr" rid="B27">27</xref>] found that stressed participants who viewed images of vegetated urban scenes showed the highest levels of stress reduction, even higher than those viewing countryside, while those viewing barren urban scenes exhibited an increase in stress levels. However, besides stress reduction, there may be other mechanisms underlying beneficial effects of local greenery, in particular increased physical activity and improved social cohesion.</p><p>The third project, addressing the third question, focuses on the micro geographic scale. Being also located in an urban environment, it studies people with and without an allotment garden. There is a long history of the use of gardens to improve psychological well-being and physical health [<xref ref-type="bibr" rid="B28">28</xref>]. However, few studies have systematically investigated the health benefits of gardens in general, and allotment gardens in particular. Allotment gardens originated at the turn of the 20<sup>th </sup>century and have known revivals during and after the two world wars to increase supplies of fresh foods [<xref ref-type="bibr" rid="B29">29</xref>]. Today, food production is only one of the many functions of allotment gardens. These gardens are now generally assumed to contribute to a wide array of public health and livability issues [<xref ref-type="bibr" rid="B30">30</xref>]. Beneficial effects of allotment gardens have been attributed to various factors, including enhanced physical activities, reduced levels of stress and mental fatigue, and a better social and cultural integration [<xref ref-type="bibr" rid="B31">31</xref>,<xref ref-type="bibr" rid="B32">32</xref>]. Several studies have investigated physical activities associated with gardening [<xref ref-type="bibr" rid="B33">33</xref>,<xref ref-type="bibr" rid="B34">34</xref>]. In one study among elderly men in The Netherlands, participants spent a greater amount of time per week doing gardening than doing other activities such as walking or cycling [<xref ref-type="bibr" rid="B35">35</xref>]. Gardening activities have typically been related to specific health benefits such as reduced cholesterol levels [<xref ref-type="bibr" rid="B36">36</xref>]. But there is some evidence that activities on allotment gardens may also contribute to health and well-being in a more general way [<xref ref-type="bibr" rid="B37">37</xref>]. When they are asked to describe their reasons for participating in an allotment garden, people often refer to the stress reducing effects of gardening [<xref ref-type="bibr" rid="B37">37</xref>]. It has been suggested that in addition to promoting physical activity and reducing stress, allotment gardens may also help to establish a sense of social and cultural integration among gardeners [<xref ref-type="bibr" rid="B38">38</xref>]. Especially for older people, allotment gardens may provide a supportive environment that combats social isolation and contributes to the development of their social networks [<xref ref-type="bibr" rid="B37">37</xref>].</p></sec></sec><sec><title>Methods/Design</title><p>Methods and design will be discussed for each of the three projects that comprise the program separately.</p><sec><title>Vitamin G1: natural environments – healthy environments. exploring the mechanisms</title><p>Starting point for the analyses is the positive relationship between green space and people's health that was found by De Vries et al. [<xref ref-type="bibr" rid="B25">25</xref>]. In the first step of our analyses we will attempt to replicate the analyses of De Vries et al. using larger, more recent, and more comprehensive datasets that are better tuned to each other. The second step entails the theoretical analysis of the mechanisms responsible for the relationship between people's living environment and their health and well-being. This will result in a number of hypotheses, relating to specific segments of the population, specific types of green space and specific health outcomes, which will subsequently be tested empirically, using (and combining) existing data sets.</p><p>Table <xref ref-type="table" rid="T1">1</xref> lists the datasets involved in the analyses. The analyses on health and well-being will be conducted on two datasets that were collected in 2001. The first dataset contains information on perceived general health of about 300.000 people (single question indicator [<xref ref-type="bibr" rid="B39">39</xref>]) and can be linked to diagnose-coded contacts with general practice during one year. This large number of subjects guarantees a sufficient power to differentiate between relatively small subgroups in the population. The second dataset contains a much larger set of indicators of health and well-being (acute complaints, chronic illness, mental health, disabilities), health behavior, and socio-economic and demographic variables but on a smaller number of people (N = 13.000). These two datasets do not contain information on feelings of safety. For this particular part of the project we will use data from a population survey on safety and crime (the so-called Politiemonitor [<xref ref-type="bibr" rid="B40">40</xref>], N = 90.000). The three individual level datasets can be geographically linked to the fourth dataset, containing information on land use in each 25 by 25 meter gridcell in the Netherlands.</p><p>GIS techniques will be used to link the individual level data to the land use data and construct the core independent variables (for example the total amount of green space in a 1 and 3 km radius around one's home). Because spatially clustered data (individuals in their environments) are involved, multilevel research techniques will be used to take into account the hierarchical structure of the data when estimating parameters [<xref ref-type="bibr" rid="B41">41</xref>-<xref ref-type="bibr" rid="B43">43</xref>].</p></sec><sec><title>Vitamin G2: effects of greenery in urban neighborhoods on health, well-being and social safety</title><p>We will apply a twofold methodology: a longitudinal study based on existing neighborhood data and a cross-sectional study based on primary data collection. For the longitudinal study, we will compare data of municipal health services (GGD's) collected prior to and after a considerable change in the local green structure of residential areas. The selection of neighborhoods for the longitudinal study will strongly depend upon the availability of GGD-data on health for neighborhoods that experienced a substantial change in their local greenery situation, prior and after this change. The longitudinal analysis will be rather coarse, since only figures at the level of a residential area as a whole will be available. Furthermore, besides the local greenery situation other characteristics may have changed over time.</p><p>The cross-sectional study is more detailed in nature. About eight large Dutch cities will be selected, having a comparable level of urbanity. Within each of these cities about ten neighborhoods will be selected with (as much as possible) a homogeneous and similar population, to diminish the possibility of strong selection effects and to enable comparison of people with similar social-economic characteristics across local greenery situations. The neighborhoods will have to differ on the set of local greenery characteristics that are considered relevant. Based on a review of the literature the most important aspects of the local greenery for each of the proposed mechanisms will be identified (e.g. amount, structure, type, design, maintenance). The inventory of local urban greenery will include site visits and contacts with municipalities of the cities at hand (green department, health department, recreation department, police). GIS-analysis will be used to quantify local greenery characteristics.</p><p>Data will be collected by means of a postal self-administered questionnaire (see table <xref ref-type="table" rid="T2">2</xref>).</p><p>The way questions are posed will be coordinated with those in the other two projects (same phrasing etc.). The same validated measurement scales will be used regarding health as in Vitamin G1. To aid in the identification of green areas, a detailed map will be included in the questionnaire.</p><p>Given that the data concern observational units at different levels (individual, neighborhood), the data will be analyzed using multilevel techniques [<xref ref-type="bibr" rid="B41">41</xref>-<xref ref-type="bibr" rid="B43">43</xref>]. About 80 observational units at the second level (that of the neighborhood) are required to estimate effects of neighborhood characteristics and cross-level interactions [<xref ref-type="bibr" rid="B44">44</xref>]. Within each neighborhood about 100 addresses will be randomly selected to participate. Given an expected response rate of 30%, this should result in 30 filled-in questionnaires per neighborhood.</p></sec><sec><title>Vitamin G3: health benefits of allotment gardens</title><p>This project focuses on individual residents of deprived urban neighborhoods who spend considerable amounts of time in allotment gardens. The allotment gardeners will be questioned with respect to the same health-related perceptions and behaviors that will be studied in the other two projects. In addition, data will be collected on relevant background variables, such as housing condition and leisure activities. Data will be collected using a mixed methodology of semi-structured face-to-face interviews and the completion of standard weekly diaries over a prolonged period of time (see table <xref ref-type="table" rid="T3">3</xref>). Diaries have the advantage of being able to measure exposure time, activities, and mental states in more detail than questionnaires at one point in time. Compared to the other two projects, these data collection methods offer the advantage of gaining detailed insight into the emotional, physical, and spiritual experiences of the gardeners and the factors influencing these experiences.</p><p>The selection of appropriate control groups is of critical importance in this project. Ideally, the control group should be comparable to the allotment gardeners in every respect, except for the time spent in an allotment garden. Because this is difficult to realize in practice, we will use two different control groups, each with its own advantages and disadvantages. The first control group consists of close neighbors of the allotment gardeners. Because most deprived neighborhoods in The Netherlands consist of similar row houses or apartments, the members of this control group are likely to be similar to the allotment gardeners with respect to their housing circumstances and other variables. However, self-selection may constitute a problem with this control group, since individuals are more likely to rent allotments when they are in good physical condition. To control for self-selection, we will also compare the allotment gardeners with future allotment gardeners who are on a waiting list for the same gardening complex.</p><p>Because of the time-consuming nature of the data collection, this project will include fewer respondents than the other two projects. A feasible design would include 80 gardeners from four different complexes, systematically varying in size (small vs. large) and gardening philosophy (productive vs. recreational). Each gardener will be matched on age, gender, ethnic background and major health risks (e.g., smoking, drinking) to either a neighbor, or a person from the waiting list. This design results in a total sample of 160 respondents, divided in three groups (80 allotment gardeners, 40 neighbors, and 40 future gardeners on a waiting list). Assuming that effect sizes will be medium (0,5), this design provides a power of over 90% at an alpha of 0.05 for detecting differences between the gardeners and the control groups if the two control groups can be combined, and a power of 72% at an alpha of 0.05 for detecting differences between the gardeners and each of the two control groups separately.</p><p>An interviewer at home will interview each respondent personally, using a combination of closed and open-ended questions. The formulation of questions will be coordinated with those in the other two projects as much as possible. In as far as possible, validated measurement scales will be used. In addition, gardeners will also be asked to keep a weekly diary for a month. The diary will ask structured questions concerning the gardeners' health and well-being, time spent and activities at the allotment over the previous week, with additional unstructured space in which the respondents are encouraged to discuss events over the week that may have influenced their health, and their thoughts and feelings concerning these events.</p></sec></sec><sec><title>Discussion</title><p>Urban green space is under strong pressure [<xref ref-type="bibr" rid="B25">25</xref>]. Due to increasing urbanization, combined with a spatial planning policy of densification, more people face the prospect of living in less green residential environments. Especially people from low economic strata, without resources to move to greener areas outside the cities, will be affected. This may lead to environmental injustice with regard to the distribution of (access) to public green space. Until now, the possible effects of these developments on public health and well-being have not been explicitly incorporated in Dutch policy making (see also the advice of the Council for Rural Areas [<xref ref-type="bibr" rid="B46">46</xref>]. Policy makers tend to view green space more as a luxury good than as a basic necessity, and appear to overlook the potentially important effects of green space on health, well-being, and safety. It is vital that these findings become implemented in urban planning and design. At present, however, there is not sufficient knowledge to translate findings into guidelines for urban planning and design. In particular, little is known about the strength of the relationships, possible social differences, and the spatial conditions that promote beneficial effects of nearby nature.</p><p>The research program Vitamin G aims to fill up these knowledge gaps. Compared to existing studies strong points are:</p><p>- We study several interrelated aspects of human well-being that until now have been studied separately: self perceived health, physical complaints, mental health, and perceived safety.</p><p>- We include a large number of different settings so that aspects of well-being can be linked to physical characteristics (amount and type of green space needed, visual quality, lay-out, management).</p><p>- Field studies will be conducted instead of experimental studies; this is especially important for applied purposes, because it provides a better indication of the relative size of the effects in real-life settings.</p><p>- The focus is on ordinary settings instead of 'extreme' settings in which people are especially stressed or frustrated (hospitals, prisons), or live in extremely poor or barren circumstances.</p><p>- A focus on different target groups within society increases the policy relevance.</p><p>- Distinguishing between individuals and environments makes it possible to analyze these two levels with appropriate statistical models (multilevel analysis).</p><p>- The use of similar dependent measures in the three projects and the macroscopic to microscopic approach enables the comparison and integration of the outcomes of the different projects.</p><p>The program also has its weak points.</p><p>The studies within the program are mainly cross-sectional. As a consequence, selection effects cannot be excluded, although they can be made less probable by appropriate statistical controls. However, longitudinal studies take much more time and are much more expensive. Especially in Vitamin G1 we use datasets that have been collected for other purposes. This means that some variables have not been measured or have been measured in a less appropriate way. An example of a variable that has not been measured in the existing datasets is differential exposure to greenery. However, in the other two projects where primary data will be collected, this will be measured. Especially Vitamin G2 will provide information on the effects of differential exposure.</p><p>The program focuses on the beneficial effects of being in a green environment. However, undeniably there are also potential negative effects. Experimental research has shown that nature, in particular wild and uncontrolled nature, is a powerful and probably genetically determined source of fear and anxiety [<xref ref-type="bibr" rid="B47">47</xref>,<xref ref-type="bibr" rid="B48">48</xref>]. Moreover, urban green space may provoke (feelings of) social unsafety, and tick bites may make people ill. Partly we will take these issues into account by asking respondents what is (un)attractive in green space and how they feel about urban green space. However, because negative health impacts of nature have received a relatively large amount of attention as compared to beneficial effects, we deem it more important to focus on the latter effects.</p><p>The program aims to contribute to spatial and health policy-making. Urban green space is under strong pressure [<xref ref-type="bibr" rid="B45">45</xref>]. Due to increasing urbanization, combined with a spatial planning policy of densification, more people face the prospect of living in less green residential environments. Especially groups with a low economic status, who do not have the resources to move to greener areas outside the cities, will be affected by these developments. This may lead to environmental injustice with regard to the distribution of (access) to public green space. Until now, the possible effects of the increasing urbanization and environmental injustice on public health and well-being have not been explicitly incorporated in Dutch policy making. Dutch policy makers tend to view green space more as a luxury good than as a basic necessity, and appear to overlook the potentially important effects of green space on health, well-being, and safety. It seems vital that these findings become implemented in urban planning and design. At present, however, there is not sufficiently known about these effects to translate these findings into guidelines for urban planning and design. In particular, little is known about the strength of the relationships, possible group differences, and the spatial conditions that promote beneficial effects of nearby nature.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>All authors participated in the design of the program. PG was the principal applicant of the program as whole and drafted the manuscript. RV is responsible for Vitamin G1; SdV for Vitamin G2 and AvdB for Vitamin G3. All authors read the drafts of the manuscript and approved the final version.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/149/prepub"/></p></sec>
|
Urban-rural disparities in smoking behaviour in Germany
|
<sec><title>Background</title><p>It is currently not clear whether individuals living in metropolitan areas differ from individuals living in rural and urban areas with respect to smoking behaviours. Therefore, we sought to explore the relation between residential area and smoking behaviours in Germany.</p></sec><sec sec-type="methods"><title>Methods</title><p>We used a nationwide German census representative for the general population of Germany. A number of 181,324 subjects aged 10 years or older were included. Information on the average daily usage of cigarettes that have or had been smoked formerly or currently was available in subjects who have ever smoked. A daily consumption of more than 20 cigarettes was considered heavy smoking. Logistic regression analyses were performed sex-stratified and adjusted for relevant confounders.</p></sec><sec><title>Results</title><p>Analyses revealed inhabitants of metropolitan areas to be more likely current smokers than inhabitants of rural areas (odds ratio 1.56, 95%-confidence interval 1.51; 1.62). Among current and former smokers those who lived in urban communities had also increased odds for being heavy smokers than those who lived in rural communities.</p></sec><sec><title>Conclusion</title><p>We conclude that living in an urban and particularly living in a metropolitan area is a determinant of both smoking and severity of current smoking. Tobacco control programs should recognize the difference in living conditions between rural and urban areas.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Völzke</surname><given-names>Henry</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Neuhauser</surname><given-names>Hanne</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Moebus</surname><given-names>Susanne</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Baumert</surname><given-names>Jens</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Berger</surname><given-names>Klaus</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Stang</surname><given-names>Andreas</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Ellert</surname><given-names>Ute</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Werner</surname><given-names>André</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Döring</surname><given-names>Angela</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>Life in urban areas might be more stressful than life in rural areas. Populations that experience higher levels of stressful events have higher proportions of current smokers who also smoke more heavily than populations with respective lower levels [<xref ref-type="bibr" rid="B1">1</xref>]. This health behaviour model of stress in which populations under stress engage in behaviour which is highly detrimental to health has repeatedly been demonstrated in the context of low income and social status which may lead to an increased risk of smoking [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B5">5</xref>]. Communities can produce stress in individuals but can also provide the coping resources that help modify these stressors. Therefore, it is <italic>a priori </italic>not clear whether a higher or even a lower prevalence of smoking can be assumed in individuals living in urban areas compared to those living in rural areas. In particular the association between living in metropolitan regions and smoking has yet received sufficient attention. Knowledge on such differences would be of great relevance for smoking prevention programs, which have been demonstrated to be highly effective [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>Previous studies [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B14">14</xref>] on this topic were predominantly performed in North America [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B13">13</xref>] and yielded conflicting results. Some studies [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B14">14</xref>] revealed subjects from urban areas to be more commonly current smokers than subjects from rural areas. These findings, however, have not always been confirmed by others [<xref ref-type="bibr" rid="B12">12</xref>] and further studies [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B13">13</xref>] revealed even the opposite, namely that individuals from rural areas to be at a higher risk of being smokers than individuals living in urban areas. Limitations of previous studies include inconsistent definitions of urban and rural areas, relatively small study populations, and reduced representativeness through investigations of selected populations thereby compromising the comparability as well as generalizability of those findings.</p><p>Moreover, there is a considerable lack of information regarding smoking habits in metropolitan compared to rural and urban regions. The mortality attributable to tobacco smoking in German Federal States is particularly high in states with a high proportion of metropolitan areas such as Berlin and Hamburg [<xref ref-type="bibr" rid="B15">15</xref>]. Likewise, a higher incidence of lung cancer has been demonstrated in metropolitan compared to urban or rural German regions [<xref ref-type="bibr" rid="B16">16</xref>]. Although several factors including general air pollution might explain these findings, it might also indicate a higher proportion of smokers in metropolitan regions.</p><p>The present study was designed to investigate the relation between residential area and smoking behaviours in Germany using data of the nationwide German Microcensus.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study population</title><p>For this study, we used data of the Microcensus 1999 that represents a household sample representative for residents in Germany across the whole age range [<xref ref-type="bibr" rid="B17">17</xref>]. The sample was selected using population registries in which every resident's address, age and sex is included by law. For the survey, 1% of the households in Germany were randomly sampled. Selected subjects were legally obligated to participate in the Microcensus. Data were collected in April 1999. A number of 506,897 individuals took part in the survey, corresponding to a response proportion of 97%. A health-related questionnaire was applied to 50% of the sample which was also randomly selected. Questions on smoking behaviour were addressed to 210,268 participants aged 10 years or older. No information on current smoking behaviour was available from 28,944 individuals leaving a study population of 181,324 subjects (item-specific response proportion 86.2%) who were available for analyses. Informed written consent was obtained from all participants. The study was approved by the Ethics Committee and public data protection agencies.</p></sec><sec><title>Measurements</title><p>Data on smoking behaviours were collected by personal interviews. With regard to current and former smoking, participants were questioned for smoking with respect to all tobacco products. Information on the average daily usage of cigarettes that have or had been smoked formerly or currently was available in subjects who have ever smoked. A daily consumption of more than 20 cigarettes was considered heavy smoking. We further selected potential confounders for the association between residential area and smoking behaviours. Education was categorized into three levels (<10 years, 10 years, >10 years) according to the German three-level schooling system. The current marital status comprised four categories (never married, married, divorced, widowed). The net income per capita was divided into four categories (<1000 DM, 1000 – <2500 DM, 2500 – <4000 DM, ≥ 4000 DM, 100 DM = 51.13 Euro). Subjects had their residency in East Germany if they lived in a region that belonged to the former German Democratic Republic including Berlin. This variable was considered as potential confounder because there were considerable changes in smoking behaviour among East Germans following the re-unification of Germany [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>] and at the same time descriptive statistics revealed an association between residency in East Germany and urbanicity in our data:</p></sec><sec><title>Statistical analyses</title><p>The study population was divided into subjects living in rural (<20,000 inhabitants), urban areas whereby the latter category was divided in two subcategories comprising small cities (20,000 – 500,000 inhabitants) and metropolitan areas (>500,000 inhabitants). The Microcensus originally covered five categories for defining community sizes: <5000, 5000–<20,000, 20,000 – <100,000, 100,000 – 500,000 and >500,000 inhabitants). The choice of the classification used in our study was based on the complete availability of the three aforementioned levels for all of the 16 German Federal States. Comparisons between rural and urban communities were made using logistic regression analysis (nominal data) or analysis of variance (ANOVA) (interval data) as appropriate. Multivariable logistic regression analyses were performed adjusted for age (decades), sex, marital status (4 categories), school education (3 categories), income (4 categories) and residency in East Germany. Odds ratios (OR) were calculated, values are given with 95%-confidence intervals (CI). A value of p < 0.05 was considered statistically significant. All statistical analyses were performed with SPSS software, version 11.5 (SPSS GmbH Software, Munich, Germany).</p></sec></sec><sec><title>Results</title><p>Selected characteristics of subjects who completed the health-related questionnaire of the Microcensus are given in table <xref ref-type="table" rid="T1">1</xref>. Subjects living in urban communities were less often of male, better educated and more often single or widowed compared to subjects living in rural communities. Furthermore, there was a lower proportion of subjects with the lowest income and a higher proportion of East Germans among individuals from metropolitan areas compared to individuals from rural areas.</p><p>With regard to smoking habits, there was a higher proportion of current smokers and a lower proportion of never-smokers in urban communities compared to rural ones (Table <xref ref-type="table" rid="T2">2</xref>). Albeit in part statistically significant, the starting age and the type of tobacco products used differed only marginally with respect to residential area. There was a higher proportion of subjects who smoked more than 20 cigarettes daily in urban compared to rural communities. All the differences between urban and rural communities were most pronounced for the comparison between metropolitan and rural areas (Table <xref ref-type="table" rid="T2">2</xref>).</p><p>The association between urbanicity and current smoking remained after adjustment for potential confounders (Table <xref ref-type="table" rid="T3">3</xref>). Age-stratified analyses yielded that inhabitants of metropolitan areas aged 40 – 79 years were particularly more likely to be smoker compared to individuals living in rural areas (Figure <xref ref-type="fig" rid="F1">1a</xref>). Other factors significantly related to current smoking in the total population were male sex, younger age (<50 years), marital status other than married, low school education, and a monthly income between 1000 and 4000 DM (Table <xref ref-type="table" rid="T3">3</xref>). High school education was inversely associated with current smoking, whereas residency in East Germany was not related to current smoking. Sex-stratified analyses revealed a more pronounced relation between residential area and smoking in the female subpopulation (Table <xref ref-type="table" rid="T3">3</xref>, Figure <xref ref-type="fig" rid="F1">1b</xref> and <xref ref-type="fig" rid="F1">1c</xref>). Among women, the differences between rural and metropolitan communities with respect to current smoking were present over all ages but particularly evident for subjects aged 50 years or older (Figure <xref ref-type="fig" rid="F1">1c</xref>). In men, a higher monthly income was inversely associated with current smoking, but there was a direct association in women (Table <xref ref-type="table" rid="T3">3</xref>). Modifications of the logistic regression model (usage of age as continuous variable, alternative categorizations of income, usage of the specific state instead of East or West Germany) did not considerably alter the estimates for the association of interest.</p><p>Logistic regression analyses that were performed with the same set of confounders using ever-smoking as the dependent variable also revealed an association between the size of the population and the endpoint. Compared to subjects living in rural communities, both subjects with residency in small cities and metropolitan communities exhibited higher odds for being ever-smokers (OR 1.18, 95%-CI 1.15; 1.21 and OR 1.53, 95%-CI 1.48; 1.59, respectively).</p><p>Among current and former smokers those who lived in urban communities also had increased odds for being heavy smokers in multivariable analyses. Compared to inhabitants of rural communities the OR for heavy smoking in individuals from small cities and metropolitan communities were 1.18 (95%-CI 1.16; 1.21) and 1.56 (95%-CI 1.51; 1.62), respectively. Compared to subjects from rural areas, women from metropolitan areas had higher odds for ever smoking (OR 1.79, 95%-CI 1.70; 1.87) than men (OR 1.36, 95%-CI 1.30; 1.43).</p></sec><sec><title>Discussion</title><p>In the present study we applied data of the Microcensus 1999 to analyze the associations between residential area and smoking. While there was a relation between urbanicity and smoking in general, residency in metropolitan areas with more than 500,000 inhabitants was particularly related to higher odds of smoking compared to residency in rural areas. Furthermore, subjects living in urban areas were more often heavy smokers compared to those living in rural areas. Again, this relation was particularly evident for the comparison between rural and metropolitan areas. All investigated relations between urbanicity and smoking behaviours were present in both sexes but stronger among women than men.</p><p>Stress might be one explanation for urban-rural differences. Particularly in metropolitan areas this stress may be related to work as well as social factors. Work-related conditions may include longer commuting to work, a greater fear of becoming unemployed and a higher pressure to perform in larger companies. Social factors may include less social relations and a higher effort to find and to foster social contacts. These factors may affect women more than men [<xref ref-type="bibr" rid="B18">18</xref>] thereby explaining sex-related differences found in our study. This is in general concordance with results from two other studies [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. The first [<xref ref-type="bibr" rid="B8">8</xref>] demonstrated that 57% of urban women were current smokers compared to 43% or rural women, whereas in men no such differences were found. The latter study [<xref ref-type="bibr" rid="B14">14</xref>] revealed that women living in cities with more than 500,000 inhabitants smoke by a factor of 50% more commonly than women from other communities [<xref ref-type="bibr" rid="B14">14</xref>].</p><p>Stress, however, cannot be the only explanation for the associations found in our study. Thus, suicide rates are higher in rural than in urban regions, particularly among younger individuals [<xref ref-type="bibr" rid="B19">19</xref>]. Internal migration might further explain the relation between urbanicity and smoking. Speculatively, more smokers than non-smokers might move from rural to urban communities than vice versa. Even a more general explanation is possible: individuals who move from rural to urban communities might be more susceptible for risk behaviours than individuals who move from urban to rural communities or individuals who do not move at all. Moreover, the general susceptibility for risk behaviours might represent the underlying cause for such migration activities. Unfortunately, data on migration were not available in our study. Therefore, we were not able to analyse the highly interesting change of exposure over time in the present context.</p><p>A specific problem in Germany is the relative absence of effective activities in smoking prevention. In contrast to many other countries there are few restrictions regarding smoking and cigarette advertisement as well as availability of cigarette machines in public places. Since inhabitants of larger German cities are more exposed to these circumstances, the findings described in the present paper appear to be plausible and may be representative for countries with little anti-smoking activities. The specific political environment as well as stress and migration, however, might not fully explain the relations reported herein. Therefore, studies are needed to further explore mechanisms underlying the association between urbanicity and smoking and to investigate the change of this association under an environment of improving tobacco control.</p><p>The present results are in concordance with results of other studies [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B14">14</xref>] that also found higher prevalence of smoking in metropolitan compared to rural regions. Conflicting results [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B13">13</xref>] may have resulted from methodological reasons and cultural disparities. All of these studies [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B13">13</xref>] were conducted in Northern America where anti-smoking activities are much better established than in Germany [<xref ref-type="bibr" rid="B20">20</xref>]. The awareness of smoking-related health hazards might be higher among inhabitants of urban compared to rural communities under such specific political environment. Furthermore, one of the former studies [<xref ref-type="bibr" rid="B12">12</xref>] investigated selected populations such as Native Americans. Cultural and social aspects might hence further explain the discrepancies between this study [<xref ref-type="bibr" rid="B12">12</xref>] and ours. In addition, all other studies [<xref ref-type="bibr" rid="B9">9</xref>-<xref ref-type="bibr" rid="B13">13</xref>] recruited only adolescents or students whereas our analyses covered the population aged 10 years and older.</p><p>There are some limitations of the present study. First, the comparability of studies on the relation between urbanicity and smoking including is limited since no standardized and generally accepted definitions of rural, urban and metropolitan areas are available. Thus, the definition of rural populations varied from ≤ 5000 [<xref ref-type="bibr" rid="B14">14</xref>] to <50,000 [<xref ref-type="bibr" rid="B10">10</xref>] inhabitants per community and the definition of metropolitan populations from >50,000 [<xref ref-type="bibr" rid="B11">11</xref>] to >500,000 [<xref ref-type="bibr" rid="B14">14</xref>] inhabitants per community. Other studies [<xref ref-type="bibr" rid="B12">12</xref>] used the population density to classify urbanicity. The choice of the classification used in our study was based on complete availability of data in the three levels <20,000, 20,000 – 500,000 and >500,000 inhabitants for all 16 German Federal States. Second, <bold>w</bold>ith the present data we were unable to analyze potential differences between communities within the metropolitan areas. In urban areas diversity is found within the same community, whereas diversity is more pronounced across rural communities [<xref ref-type="bibr" rid="B11">11</xref>]. Since such diversities may be important for the design of tobacco control programs [<xref ref-type="bibr" rid="B5">5</xref>], further analyses of regionally conducted studies should address this issue. Third, this study shares with others the limitations inherent to cross-sectional data. Because of a lack of time sequence, the relations reported here, while robust, should not be interpreted as causal.</p><p>Some strengths of the present study merit consideration. These strengths include the large size of the study population used in the analysis. In particular the high precision of risk estimates reflects the latter point. Moreover, due to the very high response proportion in the Microcensus 1999, the present data are characterized by a high grade of representativeness.</p></sec><sec><title>Conclusion</title><p>We conclude that living in an urban and particularly living in a metropolitan area is a determinant of both smoking and severity of current smoking. Given the high prevalence of smoking in Germany, general efforts should be made to intensify anti-smoking policy. Tobacco control programs should recognize the difference in living conditions between rural and urban areas.</p></sec><sec><title>Abbreviations</title><p>OR Odds ratio,</p><p>CI Confidence interval</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>HV has made substantial contributions to the conception of the analyses, performed the analyses, drafted the manuscript and has given final approval of the version to be published. HN has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. SM has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. JB has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. KB has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. AS has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. UE has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. AW has made substantial contributions to the interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published. AD has made substantial contributions to the analysis and interpretation of data, revised the manuscript for important intellectual content and has given final approval of the version to be published.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/146/prepub"/></p></sec>
|
Randomised controlled trial of welfare rights advice accessed via primary health care: pilot study [ISRCTN61522618]
|
<sec><title>Background</title><p>Little research has directly evaluated the impact of increasing financial or material resources on health. One way of assessing this lies with assisting people to obtain full welfare benefit entitlements. In 2000–1, 2.3 million pensioners were living in poverty in the UK and estimates suggest that around one million do not claim the financial support to which they are entitled. The effectiveness of welfare rights advice services delivered via primary health care to promote health and reduce health inequalities is unknown.</p></sec><sec sec-type="methods"><title>Methods</title><p>The main objectives of this study were to assess the feasibility and acceptability of a randomised controlled trial of welfare rights advice in a community setting and identify appropriate health and social outcome measures in order to plan a definitive trial.</p><p>This was a single blind, community-based, pilot randomised controlled trial. 126 men and women aged 60 years and over, recruited from 4 general practices in Newcastle upon Tyne, UK, participated. The intervention comprised a structured welfare rights assessment followed by active assistance with welfare benefit claims over the following 24 months. The control group received the intervention after a six month delay. A range of socio-economic, health, behavioural and psycho-social outcomes were measured.</p></sec><sec><title>Results</title><p>126 out of 400 people invited agreed to participate and 109 were followed up at 24 months. Both the intervention and research procedures were feasible and acceptable to participants and professionals involved. 68 (58%) of all participants received a welfare benefit award (31 financial, 16 non-financial and 21 both). Median time to receipt of benefits from initial assessment was 14 (range 1 to 78) weeks and median financial award was £55 (€81, $98) per household per week. There was little evidence of health-related differences between groups or over time, which could be due to limitations of the study design.</p></sec><sec><title>Conclusion</title><p>Modification of the study design, including selection of study participants, timing of interventions and length of follow up are recommended for a definitive trial. More appropriate health and psycho-social outcome measures relevant to the elderly population should be sought, particularly focussing on those issues highlighted in the accompanying qualitative study.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Mackintosh</surname><given-names>Joan</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" corresp="yes" contrib-type="author"><name><surname>White</surname><given-names>Martin</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Howel</surname><given-names>Denise</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Chadwick</surname><given-names>Tom</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Moffatt</surname><given-names>Suzanne</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Deverill</surname><given-names>Mark</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Sandell</surname><given-names>Adam</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Public Health
|
<sec><title>Background</title><p>Major inequalities in health result from differences in socio-economic position between individuals, families and population groups[<xref ref-type="bibr" rid="B1">1</xref>]. Whilst the link between resources and health is well established[<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B4">4</xref>], there has been little conclusive research evaluating the impact of increasing resources on health[<xref ref-type="bibr" rid="B5">5</xref>]. In the UK large amounts of welfare benefits are unclaimed[<xref ref-type="bibr" rid="B6">6</xref>], particularly amongst vulnerable groups such as older people. It has been estimated that only around 40–60% of those eligible actually claim the health-related benefits to which they are entitled[<xref ref-type="bibr" rid="B7">7</xref>]. Appropriate targeting and active assistance with benefit claims can result in substantial increases in financial and non-financial resources (e.g. parking permits, household aids and adaptations) for eligible non-claimants[<xref ref-type="bibr" rid="B8">8</xref>-<xref ref-type="bibr" rid="B10">10</xref>]. In the UK welfare rights advice is offered through local government social services departments, Citizens Advice Bureaux (one of the UK's largest voluntary organisations, providing free advice and information on money, legal or other problems) or primary care[<xref ref-type="bibr" rid="B11">11</xref>], with clients accessing the services either through self-referral, referral from another agency or both. However, these services are not available to everyone.</p><p>Evaluating the health impact of complex social interventions poses ethical and methodological challenges. In particular, the ethical acceptability of withholding welfare advice or benefits from control group participants has been questioned[<xref ref-type="bibr" rid="B12">12</xref>]. Pilot studies play an important role in health research and are a necessary prerequisite for definitive randomised controlled trials[<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. Evaluations of the health effects of social interventions are essential to identify effective ways to reduce health inequalities[<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B12">12</xref>].</p><p>We report the results of a pilot randomised controlled trial of welfare rights advice in a primary care setting, whilst Moffatt <italic>et al</italic>, in an accompanying paper[<xref ref-type="bibr" rid="B15">15</xref>], report the findings of an embedded qualitative study. We aimed to evaluate the methods, including a range of potential outcome measures, estimate effect sizes and sample size for a definitive trial.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Study design</title><p>We conducted a single blind RCT with individuals randomly allocated to intervention (receipt of immediate welfare rights assessment, advice and active assistance with claims) or control condition (receipt of the intervention after a six month delay).</p></sec><sec><title>Recruitment</title><sec><title>General practices</title><p>Four practices working from five premises in Newcastle upon Tyne participated. Three premises were in the top ten per cent of the most deprived wards in England and two were in the top one per cent, measured using the Index of Multiple Deprivation[<xref ref-type="bibr" rid="B16">16</xref>].</p></sec><sec><title>Participants</title><p>A random sample of 400 patients aged 60 years or over was generated using participating practices' computer systems, and invited to participate. Patients likely to have received full welfare assessments, such as the permanently hospitalised or those in residential or nursing care homes, were excluded. The sample size was pragmatic and aimed to enable us to estimate sample size for a future definitive trial.</p></sec></sec><sec><title>Intervention condition</title><p>Within 3 weeks of the baseline assessment, a welfare rights officer from Newcastle City Council Social Services undertook a structured assessment of current welfare status and benefits entitlement, including a full assessment of household income and expenditure. Participants were then offered active assistance with making benefit and other welfare claims where appropriate over the following months. Advice was offered either at home or at the GP surgery, but all participants preferred a domiciliary service.</p></sec><sec><title>Control condition</title><p>Participants in the control group were given an appointment for a follow up interview with the researcher six months after their baseline assessment. Following this second interview, the control group were offered an appointment with the welfare rights officer, who offered a full welfare benefits assessment and active assistance with claims as appropriate, following the same procedure as the intervention group.</p></sec><sec><title>Randomisation and blinding</title><p>Following the baseline assessment, participants were randomly allocated to the intervention or control group by the project secretary using a sequential allocation table independently generated from random number tables prior to recruitment. The researcher conducting the assessments was blind to the randomisation group.</p></sec><sec><title>Baseline assessment and outcome measures</title><p>Written informed consent was obtained at baseline assessment. The following outcomes were collected in a structured, face-to-face interview at baseline, 6, 12 and 24 months:</p><sec><title>Health</title><p>Short Form 36 (SF36)[<xref ref-type="bibr" rid="B17">17</xref>], Hospital Anxiety and Depression Scale (HADS)[<xref ref-type="bibr" rid="B18">18</xref>], activity limiting long term illness[<xref ref-type="bibr" rid="B19">19</xref>], symptoms inventory[<xref ref-type="bibr" rid="B20">20</xref>], the Pittsburgh Sleep Quality Index[<xref ref-type="bibr" rid="B21">21</xref>], and self-reported height and weight (used to calculate Body Mass Index (weight in kg/height in m<sup>2</sup>)).</p></sec><sec><title>Health related behaviours</title><p>The Dietary Inventory for Nutrition Education (DINE),[<xref ref-type="bibr" rid="B22">22</xref>] the Physical Activity Scale for the Elderly (PASE)[<xref ref-type="bibr" rid="B23">23</xref>], current smoking status and weekly alcohol consumption[<xref ref-type="bibr" rid="B24">24</xref>].</p></sec><sec><title>Psycho-social factors</title><p>Social Support Questionnaire[<xref ref-type="bibr" rid="B25">25</xref>], the Self-Esteem Inventory[<xref ref-type="bibr" rid="B26">26</xref>], the Personal Mastery Scale[<xref ref-type="bibr" rid="B27">27</xref>] and the Life Events Inventory[<xref ref-type="bibr" rid="B20">20</xref>].</p></sec><sec><title>Socio-economic status</title><p>Affordability (financial vulnerability)[<xref ref-type="bibr" rid="B28">28</xref>] and Standard of Living Index[<xref ref-type="bibr" rid="B29">29</xref>].</p><p>In addition data were collected at baseline on age, sex, ethnicity and educational attainment. The salary, travel and administrative costs of providing welfare rights advice were estimated.</p><p>Participants were contacted by telephone for repeat welfare assessments at 6, 12 and 24 (intervention) or 6 and 18 months (control) from the initial welfare assessment. When an application for benefits was made, the Welfare Rights Officer asked participants to notify her when a decision letter arrived, so that an accurate record could be kept of benefits received. Participants who had not notified her were followed up with telephone calls.</p><p>Intervention cost per case was estimated by recording the number of minutes spent on each participant's case by the Welfare Rights Officer, including advice sessions, telephone calls and letter writing, and recording the number of miles travelled. Travel was costed at £0.404 (€0.595, $0.716) per mile, and staff time at £0.934 (€1.374, $1.656) per minute. Household income and major expenditure (housing costs, council tax, loan repayments and energy costs) were used to estimate household disposable income.</p></sec></sec><sec><title>Statistical methods</title><p>Initial analyses were conducted on an 'intention-to-treat' basis, without regard to whether participants had received benefits as a result of the intervention. Explanatory analyses then compared outcomes in the intervention at different time points.</p><p>The differences between mean changes from baseline in each outcome measure for the intervention and control groups are reported along with 95% confidence intervals for those differences. Bootstrap methods were used when the distributions were markedly non-normal. ANOVA was used to compare outcomes at different time points.</p></sec><sec><title>Ethical approval</title><p>The protocol for the study was approved by Newcastle upon Tyne joint universities and NHS research ethics committee.</p></sec></sec><sec><title>Results</title><sec><title>Participant flow and follow-ups</title><p>Out of the 400 people invited to participate, 199 (49.8%) did not respond, and 58 (14.5%) declined to participate. Of the latter, 30% gave no reason, 32% said they were not interested or felt they would be ineligible for welfare benefits and 39% cited being too old, sick or frail. Between August and December 2002, 126 people were recruited and randomised (figure <xref ref-type="fig" rid="F1">1</xref>). Characteristics of the intervention and control groups were similar at baseline (table <xref ref-type="table" rid="T1">1</xref>). SF-36 scores showed poorer physical health than expected for their ages, although mental health scores were within normal range[<xref ref-type="bibr" rid="B17">17</xref>]. There was little drop out in subsequent assessments (figure <xref ref-type="fig" rid="F1">1</xref>).</p></sec><sec><title>Material and financial outcomes</title><p>Table <xref ref-type="table" rid="T2">2</xref> shows the distribution of initial disposable income and any awards made to participants, along with the period they had to wait for financial benefits. Sixty eight (58%) of all participants received a welfare benefit award (31 financial, 16 non-financial and 21 both). Non-financial awards included the disabled parking permit ("blue badge"), aids and adaptations around the home, "Staywarm"- a national energy scheme, and the Community Care Alarm scheme. Benefit eligibility was greater in general practices with higher levels of deprivation, measured using the Index of Multiple Deprivation[<xref ref-type="bibr" rid="B16">16</xref>] (table <xref ref-type="table" rid="T3">3</xref>).</p><p>Median time to receipt of benefits from initial assessment overall was 14 (range 1 to 78) weeks and median financial award was £55 (€81, $98) per household per week. At 6 months after the initial welfare assessment, eligible participants had been in receipt of benefits for a median of 9 weeks but seven (14%) had not yet received their benefits (table <xref ref-type="table" rid="T2">2</xref>). However, all participants had received their benefits by 12 months (figure <xref ref-type="fig" rid="F2">2</xref>).</p><p>The mean intervention cost per case was £120.18 (€172.34, $209.94). This comprised salary of £115.24 (€165.20, $201.21) and travel of £4.94 (€7.08, $8.62). The mean cost was £161.23 (€231.14, $281.46) for those who received benefits and £63.62 (€91.21, $111.07) for those who did not.</p></sec><sec><title>Health, behavioural and psychosocial outcomes</title><p>There was considerable variability in the change of most outcomes at 6 months (table <xref ref-type="table" rid="T4">4</xref>) but the mean change was close to zero for most scales. The distributions of the changes over 6 months were very similar for both groups. The only significant difference in mean change was for the financial vulnerability score; those in the intervention group felt less vulnerable than those in the control group at six months (difference = -1.6 (95% CI: -2.6 to -0.7)). However, the change on this scale was zero for most participants.</p><p>Since welfare benefits did not always start until late in the initial follow-up period, 6 months may have been too early to detect any health-related improvements. However, the mean outcome measures varied very little across 6, 12 and 24 month time points in the intervention group, apart from sleep quality and social interaction, which both improved between 6 and 12 months and declined between 12 and 24 months (table <xref ref-type="table" rid="T5">5</xref>).</p></sec></sec><sec><title>Discussion</title><sec><title>Main findings</title><p>This is the first pilot RCT to measure a broad range of health and social outcomes in relation to a welfare rights advice intervention accessed via a health care setting. Both the intervention and research procedures proved feasible and acceptable to participants and professionals involved.</p><p>Around 60% of participants were eligible for some benefits and 40% for financial benefits, confirming that offering welfare advice to older people via primary care is an effective way of identifying those entitled to benefits who would otherwise be unlikely to claim[<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B30">30</xref>]. However, there was little evidence of differences in health outcomes between those who did or did not receive the advice at 6 months, or within the intervention group over time. These could reflect a genuine lack of effect, or limitations in the study design; this was a pilot study and not powered to detect small differences. Previous studies examining the impact of welfare rights advice on health have been observational and therefore provide only weak evidence[<xref ref-type="bibr" rid="B10">10</xref>,<xref ref-type="bibr" rid="B31">31</xref>]. Our qualitative study[<xref ref-type="bibr" rid="B15">15</xref>] suggests that those in receipt of extra benefits were more able to participate in society and experienced greater 'peace of mind', but this was not evident using scales such as Social Interaction[<xref ref-type="bibr" rid="B25">25</xref>] or HAD-Anxiety[<xref ref-type="bibr" rid="B18">18</xref>] in this pilot trial.</p></sec><sec><title>Strengths and weaknesses</title><p>The study was conducted in the UK and the findings should therefore be interpreted with caution in relation to the health and welfare systems of other countries. Nevertheless, most of the implications are methodological in nature and thus widely applicable.</p><p>The methodological and ethical challenges in evaluating an intervention of this type have been discussed before[<xref ref-type="bibr" rid="B12">12</xref>], but the low drop-out rate suggests that participants found our study procedures acceptable, a finding corroborated by the accompanying qualitative study[<xref ref-type="bibr" rid="B15">15</xref>]. This was a single blind RCT and the letter arranging the follow up appointment specifically asked the participant not to tell the researcher whether or not they had seen the welfare adviser. Only one participant told the researcher conducting follow-up interviews (JM) that she had received benefits, but this was an isolated incident and we do not believe that it affected the results. Furthermore, the main analyses were conducted by independent statisticians (DH, TC) who were blind to this information.</p><p>Although the health and social assessment interview covered a wide range of outcome measures, taking between 30 and 90 minutes to complete, participants did not report feeling this was too onerous a task for them.</p><p>Whilst the intervention in this study was the provision of a welfare rights assessment, in 40% of the sample this did not result in any welfare benefits because of ineligibility. This will have diluted any resultant health improvements in the intention to treat analysis. Using practice deprivation scores to target the poorest areas would increase the proportion of participants eligible for welfare benefits in a future trial. Further exploration of individual predictors of eligibility using GP data (e.g. morbidity indicators) in this study was hindered by the small sample size. It may be possible in future work to improve recruitment materials to persuade others likely to gain from a welfare rights assessment to participate. Thirty nine percent of those giving a reason for non-participation said that it was because they were too old, ill or frail. Whilst this is not uncommon in research with the elderly, these are likely to be people who could benefit from a welfare rights assessment. In contrast, the embedded qualitative study[<xref ref-type="bibr" rid="B15">15</xref>] indicated that some participants agreed to take part in this research study out of a sense of altruism, rather than because they thought they might qualify for welfare benefits. All participants reported their ethnic group to be white British. Whilst Newcastle overall has a small ethnic population, one of the study practices has a sizeable proportion of patients from the South Asian community. It is not known how many people from this community were invited to participate and either did not respond or declined. However, the invitation to participate was only available in English and this is likely to have affected recruitment of people for whom English was not their first language.</p><p>The potential health-related impact of additional benefits is hard to quantify. Some concerns were raised about the standard scales used to assess health related outcomes. Interviewer feedback suggested that some of the questions on the scales we used may have been inappropriate for this population; most were not designed specifically for use with older people and, additionally, some may not be sufficiently responsive to small changes in health which would be of interest. A related issue arose in interviews with the 14 (out of the sample of 25) people in the qualitative study[<xref ref-type="bibr" rid="B15">15</xref>] who had received financial benefits. A positive and wide-ranging impact on a number of aspects of quality of life such as 'maintaining independence' or 'peace of mind' was reported. These are all factors relevant to an older population but not specifically measured by the scales used in the RCT.</p><p>The optimum time at which to measure any health benefits was unclear at the trial outset. The six month delay prior to the control group's benefit assessment was thought to be a reasonable compromise between impeding the receipt of benefit entitlements and allowing time for any health improvements to appear. This was justified ethically on the grounds that the intervention is presently rationed and recipients would not have been identified within the welfare rights officer's normal caseload. However, 14% of those eligible for financial benefits did not receive them until after the six month follow up assessment, and the average time between receipt of financial benefits and the next assessment was only two months. Health improvements resulting from increased benefits may not be detectable after such a short time. The time lapse needed for additional resources to have any effect on health is unclear, but there is a methodological argument for delaying the welfare rights advice longer in the control group in a future trial. However, this remains ethically contentious, as it is hard to justify withholding an intervention that is known to be beneficial in financial terms[<xref ref-type="bibr" rid="B12">12</xref>].</p></sec></sec><sec><title>Conclusion</title><p>Our trial design was feasible and acceptable[<xref ref-type="bibr" rid="B15">15</xref>]. The study found a large proportion of participants in the sample was eligible for welfare benefits but not claiming them. However, there was little evidence of differences in health outcome measures between groups or over time. If there was a real effect, possible reasons for a lack of evidence include: the small sample size; inadequate lengths of time for additional welfare benefits to have health and psycho-social effects; many participants did not qualify for any benefits; and the outcome measures used may not have been the most appropriate. These factors combined would have reduced the observed strength of any possible effects.</p><p>Modification of the study design to reduce the dilution effects described above, including selection of study participants, timing of interventions and length of follow up will be necessary for a definitive trial. There is also a need to look at some alternative or additional measures of health outcomes relevant to an older population, particularly those highlighted by the accompanying qualitative study[<xref ref-type="bibr" rid="B15">15</xref>], such as maintenance of independent living : however many of the outcome measures used in this pilot RCT would remain relevant. This study provides an example of how a pilot RCT with an embedded qualitative study identified unforeseen problems that will inform the design of a definitive evaluation.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>MW and SM had the original idea for the study, conducted preliminary work and, with the help of DH, AS, Nick Whitton and Rosemary Bell, developed the proposal for this study and gained funding. Rosemary Bell and Jenny Dover delivered the intervention and AS, JM and SM briefed general practices. JM and RB collected all data. Laura Stokoe and JM prepared the data for analysis. JM, DH, TC and MW, with help from MD, AS and Jean Adams, analysed the RCT data. JM, DH and MW drafted the paper and received critical comments from AS and SM. All authors approved the final version. All authors are guarantors and accept full responsibility for the conduct of the study and contents of this paper.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2458/6/162/prepub"/></p></sec>
|
Non-operative management of posterior tibialis tendon dysfunction: design of a randomized clinical trial [NCT00279630]
|
<sec><title>Background</title><p>Posterior tibialis tendon dysfunction (PTTD) is a common cause of foot pain and dysfunction in adults. Clinical observations strongly suggest that the condition is progressive. There are currently no controlled studies evaluating the effectiveness of exercise, orthoses, or orthoses and exercise on Stage I or IIA PTTD. Our study will explore the effectiveness of an eccentric versus concentric strengthening intervention to results obtained with the use of orthoses alone. Findings from this study will guide the development of more efficacious PTTD intervention programs and contribute to enhanced function and quality of life in persons with posterior tibialis tendon dysfunction.</p></sec><sec><title>Methods/design</title><p>This paper presents the rationale and design for a randomized clinical trial evaluating the effectiveness of a treatment regime for the non-operative management of Stage I or IIA PTTD.</p></sec><sec><title>Discussion</title><p>We have presented the rationale and design for an RCT evaluating the effectiveness of a treatment regimen for the non-operative management of Stage I or IIA PTTD. The results of this trial will be presented as soon as they are available.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Kulig</surname><given-names>Kornelia</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Pomrantz</surname><given-names>Amy B</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Burnfield</surname><given-names>Judith M</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Reischl</surname><given-names>Stephen F</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Mais-Requejo</surname><given-names>Susan</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Thordarson</surname><given-names>David B</given-names></name><xref ref-type="aff" rid="I5">5</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Smith</surname><given-names>Ronald W</given-names></name><xref ref-type="aff" rid="I6">6</xref><email>[email protected]</email></contrib>
|
BMC Musculoskeletal Disorders
|
<sec><title>Background</title><p>Posterior tibialis tendon dysfunction (PTTD) is a common cause of foot pain and dysfunction in adults [<xref ref-type="bibr" rid="B1">1</xref>]. Clinical observations strongly suggest that the condition is progressive. Descriptively, the varying presentations of this condition are divided into four stages. Stage I is characterized by mild swelling and medial ankle pain but no deformity. Stage II is characterized by progressive flattening of the arch, with an abducted midfoot. In stage IIA, the foot is still flexible; however, the tendon is functionally impaired. In Stage IIB, the tendon is incompetent, or even ruptured. Stage III includes all of the signs of stage II; however, the hindfoot deformity becomes fixed. Myerson added Stage IV for patients who progressed to valgus tilt of the talus in the ankle mortise leading to lateral tibiotalar degeneration [<xref ref-type="bibr" rid="B2">2</xref>]. Despite its high prevalence [<xref ref-type="bibr" rid="B3">3</xref>], there are no intervention guidelines for Stage I or II PTTD, and surgical repair is the only definitive treatment for Stage III or IV. Factors associated with PTTD include the following: age-related degeneration, inflammatory arthritides [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>], hypertension, diabetes mellitus, obesity, and less frequently acute traumatic rupture [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>].</p><p>PTTD was previously thought to result from inflammatory processes; however, surgical exploration and tissue analysis has not confirmed the presence of inflammatory cells, such as macrophages. The histopathological findings are characterized by fibroblast hypercellularity, mucinous degeneration, and neovascularization [<xref ref-type="bibr" rid="B7">7</xref>]. It is thought that these changes result in disruption of the linear orientation of the collagen bundles, representing tendon degeneration and poor tissue response to healing [<xref ref-type="bibr" rid="B5">5</xref>]. Ultrasonography demonstrates increased tendon thickening, hypoechoic areas, irregular tendon structure with disrupted fibers, and synovial sheath effusion within a pathological posterior tibialis tendon [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. The pathologies described above are characteristic of tendinosis, not tendinitis [<xref ref-type="bibr" rid="B7">7</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. The consequences of the above findings have significant ramifications in the management of person's with PTTD. The emphasis of an intervention should therefore be on promotion of tissue remodeling and adaptation under optimal loading condition (rest, orthoses, specific exercise), and time (at least three months) [<xref ref-type="bibr" rid="B11">11</xref>]. This process requires effective patient education in regards to length of time to resolution, activity, consistent orthoses wear, exercise, and weight control [<xref ref-type="bibr" rid="B11">11</xref>]. Additionally, it is important that the patient understands that absence of pain does not reflect absence of degeneration.</p><p>Frequently, progressive exercise programs serve as a cornerstone for the non-surgical management of musculoskeletal pathologies. Recent studies demonstrated positive clinical results with eccentric calf muscle training in individuals with painful chronic Achilles tendinosis [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. The participants were instructed in a concentric or eccentric training regimen performed on a daily basis for 12 weeks. Subjects in both groups were instructed to perform the exercises even if tendon pain or discomfort were experienced during exercise. Following completion of the 12-week program, 82% of the participants assigned to the eccentric group were satisfied with the results of the intervention and had resumed their previous activity level (before injury), compared to only 36% of those engaging in concentric training (p < 0.002) [<xref ref-type="bibr" rid="B13">13</xref>].</p><p>The current study in persons with tibialis posterior tendinopathy was motivated by the significant improvements documented for the eccentric training program for persons with Achilles tendinosis. The greater improvements demonstrated in response to an eccentric compared to a concentric strengthening program suggest that exercise interventions which incorporate an eccentric strengthening component may lead to more successful outcomes than concentric strengthening programs in the treatment of tendinosis. Our study will explore the effectiveness of an eccentric versus concentric strengthening intervention while wearing an orthoses to results obtained while using only an orthoses. There are no controlled studies evaluating the effectiveness of exercise, orthoses, or orthoses and exercise on PTTD. Findings from this research will be invaluable to PTTD intervention programs, as these data will provide critical information that will lead to more efficacious treatment interventions to enhance function and quality of life in persons with posterior tibialis tendon dysfunction.</p></sec><sec sec-type="methods"><title>Methods</title><p>The University of Southern California Institutional Review Board granted approval for this study.</p><sec><title>Study population</title><p>Forty-five individuals with a current episode of medial ankle or foot pain who present to the Department of Orthopedics at the University of Southern California, Long Beach Memorial, or one of 17 physical therapy clinics participating in the Clinical Research Network of the Department of Biokinesiology and Physical Therapy at the University of Southern California will be recruited. All subjects will be screened for eligibility according to the inclusion and exclusion criteria by a research investigator. Only subjects presenting with Stage I or IIA PTTD will be enrolled in this study.</p></sec><sec><title>Inclusion criteria</title><p>To be eligible for participation, subjects must meet one of the following criteria:</p><p>• pain at the medial ankle or foot reported for greater than three months duration</p><p>• tenderness localized to the tibialis posterior tendon</p></sec><sec><title>Exclusion criteria</title><p>Participants will be excluded if they have one of the following conditions:</p><p>• fixed foot deformities</p><p>• previous foot surgery</p><p>• presence of any other concurrent foot pathology besides posterior tibial tendon dysfunction</p><p>• inability to walk without assistive device</p><p>• nervous system problems (e.g., stroke, dementia, seizures)</p><p>• cognitive dysfunction (e.g. TBI, CVA)</p><p>• uncontrolled cardiovascular disease</p><p>• evidence of cord compression</p><p>• uncontrolled hypertension</p><p>• infection</p><p>• severe respiratory disease</p><p>• pregnancy</p><p>• current or recent history of low back pain</p><p>• rheumatic joint disease</p><p>• peripheral vascular disease with sensory loss in the foot</p><p>• the subject identifies that she is pregnant</p><p>• any condition that the subject identifies which may limit participation in physical activity</p></sec><sec><title>Outcome measures</title><p>The following outcome measures will be used in this randomized clinical trial:</p><sec><title>Foot Functional Index (FFI)</title><p>The FFI is a self-reported measure of pain, function, and activity level associated with foot dysfunction consisting of 23 items divided into 3 sub-scales. The scores range from 0 to 10, with the higher scores indicating more disability. Both total and sub-scale scores are calculated [<xref ref-type="bibr" rid="B14">14</xref>]. The Foot Function Index (FFI) has been validated and determined to be a reliable instrument for patients with rheumatoid arthritis [<xref ref-type="bibr" rid="B14">14</xref>]. It is also recommended as a reliable measurement scale for use in other foot orthopedic interventions trials [<xref ref-type="bibr" rid="B15">15</xref>] and has been shown to be a reasonable tool for use with low functioning individuals with foot disorders [<xref ref-type="bibr" rid="B16">16</xref>].</p></sec><sec><title>Global Rating Scale (GRS)</title><p>The GRS is a self-reported measure of overall change in subject's condition resulting from a treatment intervention with respect to activity limitations, symptoms, emotions, and quality of life on a 15-point scale [<xref ref-type="bibr" rid="B17">17</xref>]. This scale has been shown to demonstrate minimal differences consistently across domains for both improvement and deterioration [<xref ref-type="bibr" rid="B18">18</xref>].</p></sec><sec><title>Physical Activity Scale (PAS)</title><p>The PAS is a self-reported evaluation of time spent performing daily activities such as sleep, rest, or physical activity (work, leisure time, and exercise activity) on a 24-hour scale. These data will be computed and represented in metabolic equivalents (METs). The PAS has been shown to be a valid self-reporting method for physical activity. It also has high correlation with the physical-activity diary, another self-reporting method of physical activity [<xref ref-type="bibr" rid="B19">19</xref>].</p></sec><sec><title>Short Form 36 Health Survey (SF-36)</title><p>The SF-36 was developed as a multi-purpose measure of health-related quality of life for use on general and specific populations without targeting a specific age, disease, or treatment group [<xref ref-type="bibr" rid="B20">20</xref>]. The SF-36 has proven useful in monitoring general and specific populations, comparing the burden of different diseases, differentiating the health benefits produced by different treatments, and in screening individual patient's health profile. The SF-36 was developed for multiple applications and populations so that the SF-36 scores from different samples could be compared [<xref ref-type="bibr" rid="B20">20</xref>].</p></sec><sec><title>5-Minute walk test</title><p>The 5-minute walk test is a measure of function, endurance, and gait velocity. We developed and field-tested specific verbal instructions to provide consistency in testing. For the 5-Minute Walk Test the instructions read as follows: "This assessment involves walking as far as you comfortably can along this path for 5 minutes. Please let me know when you're ready and I'll give you the signals, "Ready" and "Go" to begin the timing. In order to keep the conditions identical for all the subjects, I will not be encouraging you during the walk, but I will let you know when each full minute has passed. You may stop and rest at any time during the assessment and continue walking when you're able. I'll record the total distance you cover over the 5-minute walk. Do you have any questions at this time?"</p></sec><sec><title>50-Foot walk test</title><p>The Fifty-Foot Walk Test is a measure of gait velocity and function. A specific verbal instruction, similar to that presented in the 5-Minute Walk Test was developed and field-tested for this study.</p></sec><sec><title>Timed Up and Go (TUG)</title><p>The TUG is a test of basic functional mobility for elderly persons. A specific verbal instruction, similar to that presented in the 5-Minute Walk Test was developed and field-tested for this study.</p></sec><sec><title>Visual Analogue Scale for pain (VAS)</title><p>The VAS is a single dimension scale with endpoints marked as "no pain" and "worst pain possible." It is a reliable and valid measure of self-reported pain intensity [<xref ref-type="bibr" rid="B21">21</xref>]. The stability of a VAS score has been reported (r = 0.9) [<xref ref-type="bibr" rid="B22">22</xref>]. The VAS will be used to measure pain intensity after completion of the 5-Minute Walk Test, Fifty-Foot Walk, and Timed Up and Go.</p></sec></sec><sec><title>Evaluation</title><p>The research investigator performing both pre- and post-intervention evaluations will be standardized to the assessments and blinded to the subjects' group assignment.</p></sec><sec><title>Timing of evaluation</title><sec><title>Pre-intervention (PRE)</title><p>A standard orthopedic lower quadrant assessment documenting structural condition, mobility and strength, will be followed by the administration of the <italic>Foot Functional Index, Physical Activity Scale, and SF-36 questionnaires</italic>, and performance of a <italic>5-Minute Walk Test, 50-Foot Walk Test</italic>, as well as the <italic>Timed Up and Go</italic>. The <italic>Visual Analogue Scale </italic>for pain will be administered after each functional test.</p></sec><sec><title>Post-intervention (POST)</title><p>At the conclusion of the intervention, subjects will be re-evaluated following the same testing procedures and using the same tests/questionnaires as during the initial evaluation.</p></sec><sec><title>6-Month follow up (6-MONTH FOLLOW UP)</title><p>Six months after the completion of the intervention, subjects will be asked to complete the <italic>Foot Functional Index, the Physical Activity Scale, SF-36 questionnaire, and the Global Rating Scale</italic>. A pre-addressed, stamped envelope will be included in the correspondence. Patients will also be contacted via phone to ensure a timely response.</p></sec></sec><sec><title>Intervention</title><p>The following interventions will be provided in four clinics located in the greater Los Angeles area:</p><sec><title>Orthoses</title><p>At the time of initial evaluation, all subjects' feet will be assessed and measured for orthoses. The orthoses will be fabricated from a plaster mold obtained with the patient lying prone and the foot in a subtalar joint neutral position, using procedures described by Biomechanical Service (1050 Central Ave. Suite D, Long Beach, CA 92832, 1-800-942-2272). The same company will fabricate the orthoses. The final design of the orthoses will be posted according to the amount of rearfoot pronation when comparing relaxed subtalar joint neutral and relaxed standing, (posting of 50–75% of the difference between these two positions). Additional, in the presence of a bony forefoot deformity, the forefoot will also be posted at 50% of the deformity. The subjects will be asked to comply with the protocol by wearing the orthoses 90% of their waking hours in athletic shoes.</p></sec><sec><title>Stretching</title><p>All participants will perform calf stretches (Figure <xref ref-type="fig" rid="F1">1</xref>). Each subject will be issued a "Slant by OPTP" (OPTP, PO Box Minneapolis, MN 55447-0009) to be used for calf stretching. The slant is a lightweight, portable foam wedge. The subjects will be instructed to place the slant facing away from a wall, within a foot length distance. They will then place the shod foot of the "to be stretched leg" on the slant with the toes pointing up. The subject will be instructed to lean forward until he/she perceives a strong but tolerable stretch in the calf muscles. This maneuver will be repeated 3 times with the knee extended to target the gastrocnemius muscle and 3 times with the knee slightly flexed to more selectively isolate the soleus muscle. This position will be held for 30 seconds. The lumbar spine will be placed in a "neutral" position to reduce the potential risk of strain to the low back region. Additional fine points of the technique, specific to the needs of individual subjects, will be taught to the patient by the intervention therapist. The subjects will receive a pictorial and written description of the stretching technique.</p></sec><sec><title>Progressive resistive exercise</title><p>Previous work has indicated that the tibialis posterior is preferentially recruited during a resisted foot adduction exercise in persons with pes planus [<xref ref-type="bibr" rid="B23">23</xref>] and that this muscle is selectively activated when flat-footed subjects perform the exercise while wearing arch supporting orthoses and shoes [<xref ref-type="bibr" rid="B24">24</xref>]. Therefore, the exercises will consist of isolated loading (plantar flexion and adduction) of the posterior tibialis musculotendinous unit and will be performed with the subjects wearing both orthoses and shoes. The exercises will be performed using a specialized exercise unit, which can be manipulated to progressively load the tendon either concentrically or eccentrically depending on group assignment (Figure <xref ref-type="fig" rid="F2">2</xref>). Subjects will begin resistive exercise when custom orthoses are delivered (1–2 weeks after evaluation).</p></sec></sec><sec><title>Intervention group assignment</title><p>Subjects will be randomly assigned to one of the three groups: 1) orthoses wear and calf stretching; 2) orthoses wear, calf stretching, and a progressively challenging concentric exercise program; 3) orthoses wear, calf stretching, and a progressively challenging eccentric exercise program.</p><sec><title>Orthoses (ORTH)</title><p>Subjects in the orthoses group will wear custom foot orthoses in athletic shoes for 90% of waking hours. They will be instructed in performing calf stretches as described earlier. A series of six, 30-second stretches on the involved leg will be performed twice daily. Subjects will begin to perform calf stretches on the day of initial evaluation.</p></sec><sec><title>Orthoses + concentric exercise program (CONC)</title><p>Subjects in the concentric exercise group will wear custom foot orthoses and perform calf stretches as described for the orthoses group. They will also be instructed in the performance of the concentric exercise program (consisting of plantar flexion and adduction using a specialized exercise unit). The exercise will be performed slowly (5 seconds thought the range of motion). A series of 3 sets of 15 repetitions will be performed twice daily on the involved side. Between sets, rest periods will be 1–2 minutes. The resistance, provided by a constant force extension spring, will be set at 2 pounds for the first week and progressed to tolerance and ability throughout the 10-week concentric intervention program.</p></sec><sec><title>Orthoses + eccentric exercise program (ECC)</title><p>Subjects in the eccentric exercise group will wear custom foot orthoses and perform calf stretches as described for the orthoses group. They will also be instructed in the performance the eccentric exercise regimen using a specialized exercise unit, which requires plantar flexion and resistance to adduction. The exercise will be performed slowly (5 seconds throughout the range of motion). A series of 3 sets of 15 repetitions will be performed twice daily on the involved side. Between sets, rest periods will be 1–2 minutes. The resistance, provided by a constant force extension spring, will be set at 2 pounds for the first week and progressed to tolerance and ability throughout the 10-week eccentric intervention program.</p><p>Subjects will have an Exercise Record Chart where they will record the number of performed stretches and exercises as well as reflections on their performance (ease; if discomfort, then location and intensity). Subjects will meet with an intervention therapist once per week for 10 weeks so that the quality of stretches and motion during the exercise can be assessed and resistance can be added as tolerated or decreased as needed.</p></sec></sec><sec><title>Data analysis</title><p>A 3 × 3 ANOVA with repeated measures will identify differences in the FFI across groups (ORTH, CONC, ECC) and between testing sessions (PRE, POST, 6-MONTH FOLLOW UP). All analyses will be performed using SPSS statistical software (SPSS Inc., Chicago, IL). All significance levels will be set at p < 0.05.</p></sec><sec><title>Sample size</title><p>The selection of 45 subjects for this study (15 subjects per group), is based on clinical observations, attrition rate, and power analysis. Clinical observations have indicated that the inter-subject variability with respect to pain, disability and function is moderate to high. Therefore, all power calculations have taken this increased variability into consideration. A 20% subject attrition rate is anticipated. The possible attrition rate is attributed to length of the rehabilitative program, elimination of symptoms, or increase of symptoms, need for surgery, and prior attrition rate of existing randomized control trials. The <italic>Foot Functional Index </italic>will be used as the primary outcome of this study. The remaining functional, health status, and pain variables will be used as secondary outcome variables. Power calculations were based on an alpha level of 0.05 and beta of 0.8 with respect to the FFI pain score.</p></sec></sec><sec><title>Discussion</title><p>We have presented the rationale and design for an RCT evaluating the effectiveness of a treatment regimen for the non-operative management of Stage I or IIA PTTD. The results of this trial will be presented as soon as they are available.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>JMB, KK, SFR, and SMR were responsible for the design of the study. ABP is the study coordinator and prepared the current manuscript. RWS and DBT contributed to the conceptual framework of the study. All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2474/7/49/prepub"/></p></sec>
|
The shortened disabilities of the arm, shoulder and hand questionnaire (<italic>Quick</italic>DASH): validity and reliability based on responses within the full-length DASH
|
<sec><title>Background</title><p>The 30-item disabilities of the arm, shoulder and hand (DASH) questionnaire is increasingly used in clinical research involving upper extremity musculoskeletal disorders. From the original DASH a shorter version, the 11-item <italic>Quick</italic>DASH, has been developed. Little is known about the discriminant ability of score changes for the <italic>Quick</italic>DASH compared to the DASH. The aim of this study was to assess the performance of the <italic>Quick</italic>DASH and its cross-sectional and longitudinal validity and reliability.</p></sec><sec sec-type="methods"><title>Methods</title><p>The study was based on extracting <italic>Quick</italic>DASH item responses from the responses to the full-length DASH questionnaire completed by 105 patients with a variety of upper extremity disorders before surgery and at follow-up 6 to 21 months after surgery. The DASH and <italic>Quick</italic>DASH scores were compared for the whole population and for different diagnostic groups. For longitudinal construct validity the effect size and standardized response mean were calculated. Analyses with ROC curves were performed to compare the ability of the DASH and <italic>Quick</italic>DASH to discriminate among patients classified according to the magnitude of self-rated improvement. Cross-sectional and test-retest reliability was assessed.</p></sec><sec><title>Results</title><p>The mean DASH score was 34 (SD 22) and the mean <italic>Quick</italic>DASH score was 39 (SD 24) at baseline. For the different diagnostic groups the mean and median <italic>Quick</italic>DASH scores were higher than the corresponding DASH scores. For the whole population, the mean difference between the <italic>Quick</italic>DASH and DASH baseline scores was 4.2 (95% CI 3.2–5.3), follow-up scores was 2.6 (1.7–3.4), and change scores was 1.7 (0.6–2.8).</p><p>The overall effect size and standardized response mean measured with the DASH and the <italic>Quick</italic>DASH were similar. In the ROC analysis of change scores among patients who rated their arm status as somewhat or much better and those who rated it as unchanged the difference in the area under the ROC curve for the DASH and <italic>Quick</italic>DASH was 0.01 (95% CI -0.05–0.07) indicating similar discriminant ability.</p><p>Cross-sectional and test-retest reliability of the DASH and <italic>Quick</italic>DASH were similar.</p></sec><sec><title>Conclusion</title><p>The results indicate that the <italic>Quick</italic>DASH can be used instead of the DASH with similar precision in upper extremity disorders.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Gummesson</surname><given-names>Christina</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Ward</surname><given-names>Michael M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Atroshi</surname><given-names>Isam</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
BMC Musculoskeletal Disorders
|
<sec><title>Background</title><p>Patient-reported outcome measures have become an important part of the assessments used in clinical studies. One of the outcome measures intended for upper extremity disorders is the 30-item disabilities of the arm, shoulder and hand (DASH) questionnaire, which has been assessed regarding reliability, cross-sectional validity and longitudinal validity in a variety of arm disorders [<xref ref-type="bibr" rid="B1">1</xref>-<xref ref-type="bibr" rid="B3">3</xref>]. The use of the DASH has been growing rapidly in clinical trials and other studies of upper extremity disorders and it is now available in several languages [<xref ref-type="bibr" rid="B4">4</xref>].</p><p>From the original DASH questionnaire a shorter version, named the <italic>Quick</italic>DASH, has been developed using what was called a "concept-retention" approach [<xref ref-type="bibr" rid="B5">5</xref>]. The <italic>Quick</italic>DASH consists of 11 items from the original 30-item DASH. The <italic>Quick</italic>DASH may be more appealing to use than the DASH because a shorter questionnaire is associated with less burden on the responder as well as less administrative burden. To date, data regarding the development process and various aspects of reliability and validity have been published only for the English version of the <italic>Quick</italic>DASH [<xref ref-type="bibr" rid="B5">5</xref>]. It is important that translated versions of shortened questionnaires also are subjected to an appropriate validation process. Furthermore, little is known about how the <italic>Quick</italic>DASH scores can be interpreted in comparison to the DASH scores or which version is more favorable with respect to precision of the scoring.</p><p>To determine whether a shortened questionnaire may be used to replace an existing full-length questionnaire, several assessments can be performed to show that the short version should be measuring what the original version is measuring. Different aspects of cross-sectional validity can be compared [<xref ref-type="bibr" rid="B6">6</xref>]. Further, longitudinal construct validity, which concerns the measure's ability to detect a true change in health status and its precision in detecting changes of different magnitudes (also referred to as responsiveness or sensitivity to change) needs to be addressed to determine the clinical usefulness of the short version [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>].</p><p>The purpose of this study was to evaluate the performance of the 11-item <italic>Quick</italic>DASH in comparison to the full-length 30-item DASH regarding different aspects of validity and reliability. The data for the <italic>Quick</italic>DASH were extracted from the full-length DASH.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Design</title><p>This study was designed as a reanalysis of collected data for the 30-item DASH questionnaire, from which scores for both the DASH and the <italic>Quick</italic>DASH were calculated. The data collection process for the assessment of the longitudinal construct validity of the DASH has been described previously [<xref ref-type="bibr" rid="B10">10</xref>]. The study was conducted in agreement with the local ethical guidelines for clinical studies and informed consent was obtained from the participants.</p></sec><sec><title>Questionnaire</title><p>The DASH questionnaire mainly consists of a 30-item disability/symptom scale. The two optional scales of the DASH (sport/music and work) were not part of the study. Each item in the disability/symptom scale has 5 response options. If at least 27 of the 30 items are completed a scale score, ranging from 0 (no disability) to 100 (most severe disability), can be calculated.</p><p>From the full-length DASH the 11 items that constitute the <italic>Quick</italic>DASH were extracted. To calculate a <italic>Quick</italic>DASH score at least 10 of the 11 items must be completed. Similar to the DASH, each item has 5 response options and, from the item scores, scale scores are calculated, ranging from 0 (no disability) to 100 (most severe disability).</p><p>The follow-up questionnaire included an item inquiring about change in the status of the arm as compared to its status before surgery. The item had 5 response options; much better, somewhat better, unchanged, somewhat worse, much worse. This item was accidentally missing in the initially mailed questionnaires and was therefore only completed by the last 83 participants, 82 of whom had <italic>Quick</italic>DASH scores and could be included in the present analysis.</p></sec><sec><title>Setting and participants</title><p>From an orthopedic department 109 of 118 consecutive patients with upper extremity disorders who fulfilled the eligibility criteria (scheduled for elective surgery, 18 years or older, symptom duration of at least 2 months, able to answer questionnaires) responded to the Swedish version of the DASH before surgery and at the follow-up evaluation. The follow-up was done at 6 to 21 (mean 12) months after surgery.</p><p>Of the 109 responders, 105 had responded to at least 10 of the 11 items used in the <italic>Quick</italic>DASH and were included in the analysis. The mean age of the 105 participants was 52 (range 18–83) years; 60 (57%) were women and 45 were men.</p></sec><sec><title>Analysis</title><p>The baseline, follow-up and change scores for the DASH and the <italic>Quick</italic>DASH were calculated for the whole population and for specific diagnostic groups.</p><p>To study the longitudinal construct validity the effect size (mean change score divided by the standard deviation of the baseline scores) and the standardized response mean (mean change score divided by the standard deviation of the change scores) for the DASH and <italic>Quick</italic>DASH were calculated.</p><p>To compare the performance of the DASH and the <italic>Quick</italic>DASH in discriminating among patients who differed in the degree of arm-related disability, receiver operating characteristic (ROC) curves were constructed using change scores (baseline to follow-up) as the test variable and patients' responses to the global item concerning perceived change in arm status after surgery as the dichotomized classifying variable; the difference in the areas under the ROC curves for the two questionnaire versions was calculated [<xref ref-type="bibr" rid="B11">11</xref>,<xref ref-type="bibr" rid="B12">12</xref>]. In the first ROC analysis the DASH and <italic>Quick</italic>DASH were compared with regard to their ability to discriminate the patients who rated their arm status as "much better" or "somewhat better" (combined into one group) from those who rated it as "unchanged". In the second analysis the ability to discriminate the "much better" group from the "somewhat better" group was compared. The difference in the areas under the ROC curves indicates the magnitude of the difference in the discriminant ability of the two measures. The number of patients who had reported worsening was too small to perform an analysis comparing the ability of the 2 measures to detect deterioration.</p><p>To assess reliability the Cronbach alpha coefficient was calculated for the baseline and follow-up item responses. Agreement between the <italic>Quick</italic>DASH and the full-length DASH was assessed with the intraclass correlation coefficient (ICC) using the 2-way mixed and absolute agreement model [<xref ref-type="bibr" rid="B13">13</xref>]. The difference between the DASH scores and the <italic>Quick</italic>DASH scores was assessed with the paired-samples t-test. Because the <italic>Quick</italic>DASH responses were extracted from the full-length DASH some degree of correlation between part of the questionnaire and the whole is expected. To explore the possible effect of this factor we created two hypothetical 11-item short-forms by computer-generated random selection from the 30 items of the full-length DASH. These random 11-item short-forms were analyzed with regard to reliability in a similar fashion as done with the <italic>Quick</italic>DASH.</p><p>Test-retest reliability was studied in a subgroup of 30 patients (14 women) with a mean age of 54 (range 27–79) years, who had completed the full-length DASH on two occasions prior to surgery with a median interval of 5 (range 5–17) days [<xref ref-type="bibr" rid="B14">14</xref>]. The scores for the DASH, <italic>Quick</italic>DASH and the random short-forms from both response times were calculated. The ICC (2-way mixed, absolute agreement) and the paired-samples t-test were used for this analysis.</p></sec></sec><sec><title>Results</title><sec><title>Cross-sectional validity</title><p>The baseline mean DASH score was 34 (SD 22) and the mean <italic>Quick</italic>DASH score was 39 (SD 24) (Table <xref ref-type="table" rid="T1">1</xref>). A best possible score of zero (ceiling) at baseline was recorded for the <italic>Quick</italic>DASH in 3 patients (2.9%) and for the DASH in 1 patient (1%) and a score of less than 10 was found in 19 patients (18%) and 20 patients (19%), respectively (Figure <xref ref-type="fig" rid="F1">1</xref>). At follow-up, 12 patients (14%) had a best possible <italic>Quick</italic>DASH and 10 (9.5%) a best possible DASH score. No patient had a score exceeding 90 at any evaluation except for 1 patient who had a <italic>Quick</italic>DASH score of 93 at follow-up.</p><p>The mean difference between the <italic>Quick</italic>DASH and the DASH scores at baseline was 4.2 (SD 5.4) and the mean difference at follow-up was 2.6 (SD 4.6). The mean difference between the <italic>Quick</italic>DASH and DASH change scores was 1.7 (SD 5.8; 95% CI 0.6–2.8).</p><p>For the different diagnostic groups the mean and median <italic>Quick</italic>DASH scores were higher than the corresponding DASH scores by up to 5 points in most groups (Table <xref ref-type="table" rid="T2">2</xref>). Among patients with shoulder disorders the mean DASH score was 44 (SD 15) and the mean <italic>Quick</italic>DASH score was 49 (SD 18); the difference among patients with CTS was even larger.</p></sec><sec><title>Longitudinal construct validity</title><p>When assessing the magnitude of change from baseline to follow-up the overall effect size and standardized response mean measured with the DASH and the <italic>Quick</italic>DASH were similar (Table <xref ref-type="table" rid="T1">1</xref>). Among the 24 patients with shoulder disorders treated with arthroscopic acromioplasty, the effect size measured with the DASH and <italic>Quick</italic>DASH was 0.79 and 0.74, and the standardized response mean was 0.45 and 0.46, respectively. Among the 19 patients with carpal tunnel syndrome, the effect size of open carpal tunnel release surgery measured with the DASH and <italic>Quick</italic>DASH was 0.66 and 0.89 and the standardized response mean was 0.98 and 1.05, respectively.</p><p>In the ROC analysis of the change scores for the patients who rated their arm status after surgery as better (including "much better" and "somewhat better") and those who rated it as "unchanged", the difference in the area under the ROC curves for the DASH and <italic>Quick</italic>DASH was 0.01 (95% CI -0.05–0.07), indicating no difference in their ability to discriminate between the 2 groups (Table <xref ref-type="table" rid="T3">3</xref>). In the ROC analysis comparing the ability to discriminate the "much better" group from the "somewhat better" group, the difference in the area under the ROC curves for the DASH and the <italic>Quick</italic>DASH was 0.03 (95% CI -0.03–0.09).</p></sec><sec><title>Reliability</title><sec><title><italic>Quick</italic>DASH</title><p>In the assessment of cross-sectional reliability among the 105 responders, the alpha coefficient for the scores exceeded 0.90 and the corrected item-total correlations (ITC) exceeded 0.62, except for 1 item with ITC of 0.42 at baseline (Table <xref ref-type="table" rid="T4">4</xref>). The ICC values for the agreement between the <italic>Quick</italic>DASH and the DASH scores were high, exceeding 0.90 at baseline and follow-up.</p><p>In the analysis of test-retest reliability, the ICC for the <italic>Quick</italic>DASH scores on the 2 response times was high and the mean difference between the <italic>Quick</italic>DASH scores on the first and second response time was almost zero and the 95% confidence interval was within 4 points in each direction.</p></sec><sec><title>Random 11-item forms</title><p>The first short-form included 11 randomly selected items from the full-length DASH (items 1, 3, 5, 8, 10, 11, 12, 19, 20, 24, 27). The mean score was 38 (SD 24) at baseline and 28 (SD 26) at follow-up. For the second short-form (items 3, 7, 11, 12, 13, 15, 22, 25, 26, 27, 28), the mean score was 36 (SD 23) at baseline and 25 (SD 24) at follow-up. The reliability coefficients and agreement with the DASH were high and similar to those for the <italic>Quick</italic>DASH (Table <xref ref-type="table" rid="T4">4</xref>).</p></sec></sec></sec><sec><title>Discussion</title><p>The aim of this study was to compare the performance of the 11-item <italic>Quick</italic>DASH with that of the 30-item DASH, with the <italic>Quick</italic>DASH scores extracted from the responses to the full-length DASH. The results indicate that the DASH can be replaced by the shorter <italic>Quick</italic>DASH. The magnitude of the differences between the DASH and the <italic>Quick</italic>DASH scores found in this study implies that the same questionnaire should be used in longitudinal studies because the score differences between the questionnaires may inflate small random differences and make them reach the level of an important change.</p><p>In all analyses the <italic>Quick</italic>DASH scores were slightly higher than the corresponding DASH scores, which may be an advantage for the <italic>Quick</italic>DASH as this allows for larger improvement to occur, provided that the scores considered as "normal" are equal. Among the different diagnostic groups the <italic>Quick</italic>DASH mean scores were higher; in fact this difference was more pronounced among patients with greater disability, such as those with shoulder disorder, than patients with little disability, such as those with wrist ganglion (Table <xref ref-type="table" rid="T2">2</xref>). This suggested that the <italic>Quick</italic>DASH potentially had better precision in detecting different degrees of disability. To further assess possible differences in the two measures' ability to detect improvement, ROC curves were studied. In all analyses, the confidence intervals for the difference contained null, indicating that no differences were found between the DASH and the <italic>Quick</italic>DASH in their ability to discriminate among groups that differed in the degree of self-rated improvement in arm status after surgery.</p><p>In the study assessing the English-version <italic>Quick</italic>DASH the standardized response mean, calculated for the total population of 171 patients with various disorders, was 0.78 for the DASH and 0.79 for the <italic>Quick</italic>DASH [<xref ref-type="bibr" rid="B5">5</xref>]. In our study the standardized response mean for the DASH and <italic>Quick</italic>DASH also were similar, with values of 0.61 and 0.63, respectively. The mean scores for the DASH and <italic>Quick</italic>DASH in different diagnostic groups were more similar in the study of the English-version <italic>Quick</italic>DASH than in our study. However, limited data was available and the score distributions for the groups were not shown making comparisons difficult.</p><p>In this study, as in the study that reported the development and validation of the English version [<xref ref-type="bibr" rid="B5">5</xref>], the <italic>Quick</italic>DASH scores were computed from the full-length DASH responses. It is not known if patients' responses to the 11 items would have differed if only the <italic>Quick</italic>DASH were administered. In a study of the performance of three SF-36 scales (physical functioning, bodily pain and general health perceptions) no significant differences were found when the scales were administered independently compared to when they were administered within the full 8-scale questionnaire [<xref ref-type="bibr" rid="B15">15</xref>]. However, these were full scales and not selected items as is the case with the <italic>Quick</italic>DASH. The results of the present study, based on <italic>Quick</italic>DASH responses extracted from the full-length DASH, are promising but further assessment of the short version administered to different patient groups would be useful. Because of the small number of patients in certain diagnostic groups as well as the small number of patients with unchanged or worsened self-rated arm status the results involving these groups may need to be interpreted with caution.</p><p>The reliability of the <italic>Quick</italic>DASH was good. However, the 2 randomly constructed 11-item forms also had similarly good reliability and agreement with the DASH. The 2 random short forms showed higher scores than the DASH at baseline and follow-up, which also was found with the <italic>Quick</italic>DASH. Although the differences were statistically significant, their magnitude may not be considered as clinically important. The findings may suggest that the 30-item full DASH may contain redundant items and that fewer items would be sufficient for assessing disability with the same degree of reliability and validity. It might be argued that the random short forms may not cover all relevant domains. However, the results of the DASH or <italic>Quick</italic>DASH are usually not presented as a number of separate components or domains because they are not validated as such. Moreover, the DASH and <italic>Quick</italic>DASH are predominantly composed of activity items that measure physical disability leaving little impact for the non-activity items. Because the item responses were extracted from the responses to the full-length DASH it may not be possible to compare with certainty the individual performance of the <italic>Quick</italic>DASH as compared to other possible short forms of the DASH.</p><p>In this study all participants underwent surgery, an intervention that often results in large score change. The effect size and standardized response mean measured with DASH and <italic>Quick</italic>DASH in populations treated with surgery may be larger than those measured after other interventions. However, the overall effect size in this population was moderate probably because the different diagnostic groups had large variation in the degree of baseline disability with some groups having low scores before treatment allowing only small score improvement. The results support the use of the <italic>Quick</italic>DASH even in the assessment of interventions expected to have smaller effect size.</p><p>The findings of this study are primarily related to the validity and reliability of the Swedish version of the <italic>Quick</italic>DASH (available online [<xref ref-type="bibr" rid="B4">4</xref>]). Although many aspects also may apply to <italic>Quick</italic>DASH versions that are derived from other translated full-length versions with established validity and reliability, other language versions would still require appropriate assessment.</p></sec><sec><title>Conclusion</title><p>The results of this study indicate that the <italic>Quick</italic>DASH can be used instead of the DASH to measure disability/symptom severity with similar precision in a variety of arm disorders.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>CG and IA participated in the design of the study, data collection and analysis, and writing of this manuscript. MW participated in the analysis and writing of this manuscript. All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2474/7/44/prepub"/></p></sec>
|
Comparative genomic mapping of the bovine Fragile Histidine Triad (<italic>FHIT</italic>) tumour suppressor gene: characterization of a 2 Mb BAC contig covering the locus, complete annotation of the gene, analysis of cDNA and of physiological expression profiles
|
<sec><title>Background</title><p>The Fragile Histidine Triad gene (<italic>FHIT</italic>) is an oncosuppressor implicated in many human cancers, including vesical tumors. <italic>FHIT </italic>is frequently hit by deletions caused by fragility at FRA3B, the most active of human common fragile sites, where <italic>FHIT </italic>lays. Vesical tumors affect also cattle, including animals grazing in the wild on bracken fern; compounds released by the fern are known to induce chromosome fragility and may trigger cancer with the interplay of latent Papilloma virus.</p></sec><sec><title>Results</title><p>The bovine <italic>FHIT </italic>was characterized by assembling a contig of 78 BACs. Sequence tags were designed on human exons and introns and used directly to select bovine BACs, or compared with sequence data in the bovine genome database or in the trace archive of the bovine genome sequencing project, and adapted before use. <italic>FHIT </italic>is split in ten exons like in man, with exons 5 to 9 coding for a 149 amino acids protein. VISTA global alignments between bovine genomic contigs retrieved from the bovine genome database and the human <italic>FHIT </italic>region were performed. Conservation was extremely high over a 2 Mb region spanning the whole <italic>FHIT </italic>locus, including the size of introns. Thus, the bovine <italic>FHIT </italic>covers about 1.6 Mb compared to 1.5 Mb in man. Expression was analyzed by RT-PCR and Northern blot, and was found to be ubiquitous. Four cDNA isoforms were isolated and sequenced, that originate from an alternative usage of three variants of exon 4, revealing a size very close to the major human <italic>FHIT </italic>cDNAs.</p></sec><sec><title>Conclusion</title><p>A comparative genomic approach allowed to assemble a contig of 78 BACs and to completely annotate a 1.6 Mb region spanning the bovine <italic>FHIT </italic>gene. The findings confirmed the very high level of conservation between human and bovine genomes and the importance of comparative mapping to speed the annotation process of the recently sequenced bovine genome. The detailed knowledge of the genomic <italic>FHIT </italic>region will allow to study the role of <italic>FHIT </italic>in bovine cancerogenesis, especially of vesical papillomavirus-associated cancers of the urinary bladder, and will be the basis to define the molecular structure of the bovine homologue of FRA3B, the major common fragile site of the human genome.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Uboldi</surname><given-names>Cristina</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Guidi</surname><given-names>Elena</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Roperto</surname><given-names>Sante</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Russo</surname><given-names>Valeria</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Roperto</surname><given-names>Franco</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Di Meo</surname><given-names>Giulia Pia</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A7" contrib-type="author"><name><surname>Iannuzzi</surname><given-names>Leopoldo</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A8" contrib-type="author"><name><surname>Floriot</surname><given-names>Sandrine</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A9" contrib-type="author"><name><surname>Boussaha</surname><given-names>Mekki</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A10" contrib-type="author"><name><surname>Eggen</surname><given-names>André</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A11" corresp="yes" contrib-type="author"><name><surname>Ferretti</surname><given-names>Luca</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Genomics
|
<sec><title>Background</title><p><italic>FHIT </italic>is a tumour suppressor gene that is frequently inactivated by deletions in several human tumours, as it overlaps perhaps the most active of common fragile sites FRA3B, at HSA3p14.2 [[<xref ref-type="bibr" rid="B1">1</xref>], for a review see [<xref ref-type="bibr" rid="B2">2</xref>]]. Infact, <italic>FHIT </italic>is highly responsive to environmental carcinogens, to treatment with aphidicolin – which specifically activates common fragile sites -, and is involved in the development of breast, lung, cervix, stomach, pancreas, head and neck, and kidney tumours, as well as of urothelial tumours, mainly bladder tumours [<xref ref-type="bibr" rid="B2">2</xref>]. The FHIT protein is absent or reduced in most cancers, although transcription of the gene is not necessarily altered, as in many malignancies only one allele is lost [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. The human <italic>FHIT </italic>is huge, spanning about 1.5 Mb at the FRA3B locus. However, it codes for a protein of just 147 amino acids, derived by translation of 10 tiny exons of which only exons 5 to 9 contain the ORF; the transcript is very small too, 1.1 kb [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B5">5</xref>].</p><p>Fragility at FRA3B and the consequent inactivation of <italic>FHIT </italic>are interesting also because of the presence of an integration site for Human Papilloma Virus mapped to intron 4 [<xref ref-type="bibr" rid="B6">6</xref>].</p><p>We are investigating <italic>FHIT </italic>in cattle because tumours of the urinary tract in this species are a good model for human carcinogenesis, and because of the suggested involvement of papillomavirus in tumorigenesis [<xref ref-type="bibr" rid="B7">7</xref>]. Urinary bladder tumours are a common pathology of 4- to 12-year-old cattle and there is a strong pathogenetic relationship between BPV-2 and some carcinogenic principles of bracken fern [<xref ref-type="bibr" rid="B8">8</xref>]. Prolonged ingestion of toxic compounds released by bracken fern, such as ptaquiloside, is responsible for an initial cell transformation of urothelial cells via <italic>RAS </italic>activation; it is believed that papillomavirus infection is associated with tumour progression, whose major clinical symptom is the so called chronic enzootic hematuria (CEH). It is likely that <italic>FHIT </italic>also plays a role in the process [<xref ref-type="bibr" rid="B9">9</xref>], and it is an obvious candidate for deletions affecting the locus that should represent the bovine homologue of FRA3B.</p><p>In a recent study we isolated an exon tag for bovine <italic>FHIT </italic>which allowed us to map the gene by FISH to BTA22q24 and with radiation hybrids to the telomeric region of BTA22 [<xref ref-type="bibr" rid="B10">10</xref>]. Here we used a comparative genomic approach with human <italic>FHIT </italic>to completely annotate the bovine gene. Thus, we describe a 2 Mb 78 BAC contig covering the gene, and present evidence for a very high level of sequence conservation and genomic organization over the whole <italic>FHIT </italic>region between the two species. Finally, we isolated and sequenced four <italic>FHIT </italic>cDNA isoforms, and documented the physiological expression profiles by means of RT-PCR and Northern blot analysis.</p><p>This information will be the starting point to study the role of <italic>FHIT </italic>in papillomavirus-associated tumours of bovine urinary bladder, and more in general will help to interpret the implication of <italic>FHIT </italic>in animal carcinogenesis, including the recently hypothesized role in apoptosis of a phosphorylated form of the FHIT protein [<xref ref-type="bibr" rid="B11">11</xref>].</p></sec><sec><title>Results and discussion</title><sec><title>Assembly of a BAC contig covering <italic>FHIT</italic></title><p>Short sequence tags derived from the <italic>FHIT </italic>human genomic sequence or taken from the trace archive of the bovine genome sequencing project, were used to isolate by PCR clones from an ordered BAC library [<xref ref-type="bibr" rid="B12">12</xref>]. Exon tags were first employed but given the anticipated large size of the genomic region to be covered (1.5 Mb in man), intronic STSs were also used. The STSs that did not amplify comparatively were adapted to the bovine thanks to the homology to non annotated sequence entries available in the bovine genome database (Build 2.1) and bovine genome sequencing trace archive. The whole primer's set is shown in Table 1 [<xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. As many <italic>FHIT </italic>exons are very short, primers were often designed covering the exon-intron boundaries or spanning the exon from within adjacent introns. In most instances the approach was successful, and unexpectedly also intronic tags positioned very far from exons worked in PCR, supporting once more the high degree of sequence conservation between human and cattle.</p><p>By sequential screens of the BAC library with selected primers, a total of 78 BAC clones was retrieved, belonging to three subcontigs. The software Finger Printed Contigs (FPC) was used as explained in Methods to produce the most likely arrangement of the BACs tiling in a single contig. Its structure, shown in Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref> ], was subjected to experimental validation using two approaches, 1) amplification of STS tags on the individual BACs, and, 2) alignment of available sequence from the BAC clones (i.e. mapped STS and BAC-ends) to contigs of sequence retrieved from the bovine genome database.</p><p>The tiling pattern of BACs in the contig as suggested by FPC was consistent with the PCR mapping data for all the exon and intron tags, except for a group of 13 BACs, labeled with an asterisk in Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>]. The inclusion of those BACs in the <italic>FHIT </italic>contig was likely an FPC artefact; indeed, the programme selects the best tiling of a set of clones following a merely statistical evaluation of the similarity in their restriction fingerprints. In our case, however, three lines of evidence allowed to resolve the ambiguity. First, STS-content mapping revealed that none of the 13 BACs contained the expected tags based on the proposed FPC output (i.e. exon 6, 7, 8, intron 8–1 to intron 8–2, see Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>]). Second, BACs bI0643H09 and bI0947B04, boxed in Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>], could be directly linked in the contig, as they were both positive to the intron 8–1 tag. Third, the sequence of the BAC-ends, available for clones bI0195D05, bI0520A05, and bI0545B04, was BLASTed to the bovine genome database and identified a contig assigned to BTA11 [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_988236">NW_988236</ext-link>], which is extraneous to the <italic>FHIT </italic>bovine locus, previously mapped to BTA22 [<xref ref-type="bibr" rid="B10">10</xref>].</p><p>In conclusion, the BAC contig covers a region of about 2 Mb spanning the entire <italic>FHIT </italic>locus. The bovine gene is split into 10 exons like the human, separated by introns that appear also similar in size. The only exception to this highly conserved pattern is exon 3 (58 bp in the bovine, 53 bp in the human), which is completely different with respect to sequence and position in the two species. As shown in Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>], bovine exon 3 is placed between two intron 2 tags (I2–1, I2–2), while the human's exon 3 position is between I2–6 and I3–1 tags. Another difference in the organization of the bovine gene is the presence of two alternative exons 4 (exon 4A and 4B, Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>]) that have no homology in the human <italic>FHIT</italic>. More details on transcripts and alternative splicing are given in a later section.</p></sec><sec><title>VISTA global alignments</title><p>To study further the degree of conservation at the <italic>FHIT </italic>locus between the bovine and the human, all the available contigs covering the region were recovered from the bovine genome database (Build 2.1) using representative tags from the BAC clones of the <italic>FHIT </italic>contig previously described. Seven large contigs were retrieved: [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link>] (314883 bp), containing exon 1; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_001028446">NW_001028446</ext-link>] (64796 bp), containing exon 2 to I2–2 tag; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_984819">NW_984819</ext-link>] (239821 bp), containing I2–3 to I3–5 tag; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] (641699 bp), containing I3–6 to I5–2C tag; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>] (49009 bp), containing I5–2 tag; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_977438">NW_977438</ext-link>] (344934 bp), containing I5–2D to I5–4 tag; [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link>] (1090058 bp), containing exon 6 to exon 10. The contigs were submitted to the mVISTA server along with the human <italic>FHIT </italic>sequence (Build 36.1) in order to analyse the conservation at the locus. However, as most of the bovine contigs represent largely unfinished raw sequence, e.g. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] contained two large stretches (67 and 53 kb each) of unassigned base calls, all N runs longer than 1 kb were removed before submitting the contig files to the mVISTA server. In addition, as there is a maximum limit of 2 Mb for the base sequence to be used as reference, the human <italic>FHIT </italic>genomic sequence from HSA3 contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NC_000003">NC_000003</ext-link>.1] was cut between nucleotides 59500000 and 61500000. The <italic>FHIT </italic>gene coordinates are nt 59710076 to 61212164.</p><p>The graphical output of the mVISTA server analysis is shown in Figure <xref ref-type="fig" rid="F1">1</xref>. The bovine contigs are labeled 1 to 8 according to their expected position with respect to the human reference sequence. The conservation of sequence with the human reference sequence was substantially high, in agreement with the data obtained through the STS mapping of the BAC contig described previously. In addition, as the contribution of exons to the whole alignment is negligible (1 kb vs 2 Mb), the VISTA output clearly suggested that non coding sequence was not only conserved (with an average identity well above 70%), but also arranged very similarly between the two species. Thus, the tiling of bovine contigs followed the expected order and did not suggest the existence of rearrangements at the locus between human and bovine, with possibly one exception, represented by contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] that required special attention. Indeed, the original contig sequence could be aligned only after, 1) removing two large stretches of 53 and 67 kb of Ns (coordinates 214 to 267 kb, and 453 to 520 kb, respectively) from the sequence and, 2) aligning the 5' 84 kb of the modified contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116_A">NW_930116_A</ext-link> in Figure <xref ref-type="fig" rid="F1">1</xref>] independently from the remaining 84–520 kb portion [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116_B">NW_930116_B</ext-link> in Figure <xref ref-type="fig" rid="F1">1</xref>], which had also to be reverse-complemented to successfully align. As shown in Figure <xref ref-type="fig" rid="F1">1</xref>, <ext-link ext-link-type="gen" xlink:href="NW_930116_B">NW_930116_B</ext-link> was proximal to <ext-link ext-link-type="gen" xlink:href="NW_930116_A">NW_930116_A</ext-link>, and contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>] was placed in between them, with a 635 bp overlap at the 5' end with <ext-link ext-link-type="gen" xlink:href="NW_930116_B">NW_930116_B</ext-link> and a 9 kb overlap at the 3' end with <ext-link ext-link-type="gen" xlink:href="NW_930116_A">NW_930116_A</ext-link>. This ordering was also in agreement with the tiling of BACs in the contig of Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>], and with the PCR mapping of exons and introns tags. An explanation for the apparent rearranged structure of the original contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] is that it was formed by merging two sub-contigs (our A and B portions) in the wrong order during the assembly of the whole bovine genome sequence. In this respect, the VISTA alignment clearly suggested that contig [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>] might represent a missing portion in the real arrangement of the correct [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] contig. Our interpretation of the arrangement of the seven retrieved genomic contigs is consistent with experimental data, but shall find a final confirmation only when the complete annotation of the bovine genome will be available.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p><bold>VISTA Global Alignments of the human and bovine <italic>FHIT </italic>loci</bold>. VISTA plot showing sequence similarity in pairwise sequence alignments between the human sequence at the <italic>FHIT </italic>locus [Genbank:<ext-link ext-link-type="gen" xlink:href="NC_000003.1">NC_000003.1</ext-link>] nt 59500000 to 61500000; the <italic>FHIT </italic>gene coordinates are nt 59710076 to 61212164] and bovine genomic contigs: 1. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link>]; 2. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_001018446">NW_001018446</ext-link>]; 3. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_984819">NW_984819</ext-link>]; 4. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116_B</ext-link>] (Reverse-complement of the original [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] from nt 84000 to the 3' end); 5. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>]; 6. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116_A</ext-link>] (the initial 84000 nt of the original [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>]); 7. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_977438">NW_977438</ext-link>]; 8. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link>]. Peaks are shown relative to their position in the reference human sequence (horizontal axis) and percent identity (50–100%) is shown on the vertical axis. Color code is predominantly pink (non coding sequence) as exons (in blue) are virtually invisible due to their very small size (49 to 317 bp).</p></caption><graphic xlink:href="1471-2164-7-123-1"/></fig><p>The exon/intron structure of the bovine <italic>FHIT </italic>gene is summarized in Table 2 [see <xref ref-type="supplementary-material" rid="S3">Additional file 3</xref>]. The gene coordinates were identified using the seven genomic contigs retrieved from GenBank (Bovine Genome Build 2.1) as reference sequence. The size of the introns could not be defined exactly in all cases, as most contigs still contain large blocks of unfinished sequence; however the predicted size of introns 4 and 5, based on this work, is very close to that of the human gene. This is noticeable, especially for intron 5, that is over 560 Kb, and is known to be the major target of deletions that affect the gene structure in human cancers. It is tempting to speculate that <italic>FHIT </italic>intron 5 might have a similar role also in tumours affecting cattle.</p></sec><sec><title><italic>FHIT </italic>cDNA isolation and analysis of expression</title><p>In a previous report, we mapped the <italic>FHIT </italic>gene by FISH and radiation hybrids using a bovine EST that matched human exon 9 [<xref ref-type="bibr" rid="B10">10</xref>]. Here we designed primers on the human <italic>FHIT </italic>cDNA to cross-amplify the bovine cDNA by means of RT-PCR. Thus, oligonucleotides Exon 1F and Exon 10R (Table 1, [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]), were used to amplify nearly the entire cDNA (expected product size 914 bp), on RNA extracted from brain, spleen, kidney, liver, lung and testis. The results are shown in Figure <xref ref-type="fig" rid="F2">2</xref>. As expected, <italic>FHIT </italic>is transcribed ubiquitously, and PCR products were obtained in all tissues, although of different sizes. Thus, four bands were visible within the entire set of samples, and were recovered from the gel and sequenced. A band of 914 bp was present in all tissues except kidney, that showed a band of 878 bp. A third product of 1014 bp was found in brain, spleen, and kidney, but not in lung and liver, and a fourth product of 1029 bp was detected in testis. The existence of four different cDNA isoforms is due to an alternative splicing of exons that belong to the 5' untranslated portion of the gene. This was confirmed by sequencing the isoforms of Figure <xref ref-type="fig" rid="F2">2</xref> and performing additional RT-PCRs with selected combinations of exon-primers on the same panel of tissues, with the addition of bladder. The results are summarized in Figure <xref ref-type="fig" rid="F3">3</xref>, with a schematic representation of the exons arrangement and size of the different isoforms. All the isoforms contain the entire coding portion of the <italic>FHIT </italic>mRNA, so they translate into the same polypeptide of 149 amino acids coded by exons 5 to 9. The difference in the RT-PCR products for kidney, spleen and brain (Figure <xref ref-type="fig" rid="F2">2</xref>) is due to the alternative use of exon 4B (100 bp), yielding a longer cDNA of 1181 bp and a shorter of 1081 bp; the [Genbank:<ext-link ext-link-type="gen" xlink:href="CR849215">CR849215</ext-link>] endometrium cDNA is a bovine non annotated cDNA entry taken from GenBank that has the same structure of the longer cDNA isoform described in the present study for brain, spleen and kidney (i.e. with exon 4B). Testis shows a different exon 4 variant, 4A (115 bp), thus the cDNA in that tissue is 1196 bp long. A fourth cDNA exists and is the shortest of all: it is the only one with exon 3 (58 bp), but is missing exon 4, so its whole length is 1045 bp. This cDNA is one of the two isoforms detected in the kidney.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>RT-PCR analysis of <italic>FHIT </italic>expression</bold>. Total RNA extracted from brain (B), spleen (S), kidney (K), liver (Li), lung (Lu) and testis (T) was reverse-transcribed and then amplified with primers designed to reproduce the entire <italic>FHIT </italic>cDNA, i.e. from exon 1 to 10. Numbers close to bands indicate size in bp. C- is a control reaction with no RNA. RT-PCR products were separated on a 1.5% agarose gel. M are DNA size standards (1 Kbp and 100 bp ladder mixed together).</p></caption><graphic xlink:href="1471-2164-7-123-2"/></fig><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>cDNA isoforms of the <italic>FHIT </italic>gene</bold>. The isoforms of the bovine <italic>FHIT </italic>cDNA identified by RT-PCR analysis in different tissues are listed. Top, schematic of all exons found in the bovine cDNA analysis: non coding (empty boxes), alternatively-used (grey) and coding (black). Size, in bp, is shown for single exons within the box, and at the right for the entire cDNA isoform. Letters (A, B) highlight tissues showing more than one isoform. Endometrium [Genbank:<ext-link ext-link-type="gen" xlink:href="CR849215">CR849215</ext-link>] is a bovine non annotated cDNA entry found in the NCBI database and homologous to <italic>FHIT</italic>. At the bottom is a schematic for the human <italic>FHIT </italic>gene, with four entries of the NCBI database for the known human cDNA isoforms. [Genbank:<ext-link ext-link-type="gen" xlink:href="BM711354">BM711354</ext-link>] and [Genbank:<ext-link ext-link-type="gen" xlink:href="CR745076">CR745076</ext-link>] are 5' RACE products.</p></caption><graphic xlink:href="1471-2164-7-123-3"/></fig><p>For a comparative analysis, human cDNA entries were also retrieved from GenBank. As in the bovine, an alternative use of exons is apparent in the case of the human <italic>FHIT</italic>, as shown in the lower portion of Figure <xref ref-type="fig" rid="F3">3</xref>. Three isoforms are shown that are characterized by alternative splicing of three exons 4 variants, respectively of 49, 72 and 81 bp, none of which has any homology with the bovine alternative exons 4.</p></sec><sec><title>Northern blot analysis</title><p>The expression of <italic>FHIT </italic>was also monitored by Northern blot analysis with a labeled cDNA probe in the tissues used in RT-PCR (testis, spleen, brain, lung, kidney, bladder), and, additionaly, in heart and muscle (Figure <xref ref-type="fig" rid="F4">4</xref>). In all samples a major transcript of about 1.1–1.2 kb was visible, a size very close to the known human <italic>FHIT </italic>mRNA, 1095 bp [Genbank:<ext-link ext-link-type="gen" xlink:href="NM_02012">NM_02012</ext-link>]. As already with RT-PCR, <italic>FHIT </italic>was ubiquitously expressed, with apparently higher levels of mRNA in testicular tissue and spleen. However, although equal amounts of polyA mRNA were used, the evidence coming from the Northern blot is only qualitative and cannot be taken as suggestive of variable expression levels between the investigated tissues.</p><fig position="float" id="F4"><label>Figure 4</label><caption><p><bold>Mapping of <italic>FHIT </italic>transcripts by Northern blot</bold>. Polyadenylated mRNA extracted from testis (T), spleen (S), brain (Br), lung (Lu), kidney (K), bladder (Bl), heart (H) and muscle (M) was run on a 1.2% agarose formaldehyde gel, blotted onto a nylon filter and hybridized with a labeled <italic>FHIT </italic>probe corresponding to the complete cDNA (exon 1 to 10). The position of RNA size standards is shown at the right; arrows point to the main <italic>FHIT </italic>1.1–1.2 kb transcript and to additional transcripts of 0.6 and 0.8 kb that are visible in testis and kidney, respectively.</p></caption><graphic xlink:href="1471-2164-7-123-4"/></fig><p>The resolution of the gel used for the Northern blot did not allow to clearly distinguish the mRNA isoforms as in the RT-PCRs of Figure <xref ref-type="fig" rid="F2">2</xref>. Heart and muscle showed a band of slightly higher molecular weight, but since these tissues were not studied by RT-PCR, it is likely that the major band visible in the Northern blot was a mixture of the cDNA isoforms described before. Only in testis and kidney the probe revealed two additional bands of lower molecular weight, 0.6 and 0.8 kb, respectively. As these mRNA species were not detected with the RT-PCR assay, it is difficult to establish if they were real or rather represented a partial degradation of the <italic>FHIT </italic>mRNA of 1.1–1.2 kb, prior to electrophoresis.</p></sec></sec><sec><title>Conclusion</title><p>In this work we used a comparative genomic approach to fully annotate the bovine <italic>FHIT </italic>gene. A set of 78 BACs was assembled in to a genomic contig that covers 2 Mb of DNA. The bovine <italic>FHIT </italic>turned out to have the same arrangement as in man, and an extraordinary conservation of sequence was found throughout all the locus, in such a way that not only the number, but also the size of introns was virtually identical between the two species. The size of the bovine gene could be inferred from the alignment of the BAC contig tags with the 7 genomic contigs recovered from GenBank (Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>], and Figure <xref ref-type="fig" rid="F1">1</xref>), and was around 1.6 Mb from exon 1 to exon 10. Our data allowed to extend the well known finding of a high degree of synteny between the human and bovine genome down to the level of a single gene, including, at least for <italic>FHIT</italic>, the size and nucleotide sequence of introns as large as 560 kb, like intron 5.</p><p>Our data contribute to the ongoing bovine genome sequencing project with the annotation of a rather large genomic region. The comparative approach that we used allowed us to highlight inconsistencies and errors in the assembly of bovine genomic contigs spanning the <italic>FHIT </italic>gene region. Studies of this kind will be important to speed up and assist the laborious work of finishing the assembly of the whole bovine genome sequence.</p><p><italic>FHIT </italic>is a very important oncosuppressor gene that lays in the middle of the major common fragile sites of the human genome, FRA3B. The role of the FRA3B/<italic>FHIT </italic>locus in the onset and development of a vast category of important cancers in man has been largely documented. A typical feature of <italic>FHIT </italic>involvement is that rearrangement at the locus gives rise to a complex set of homo- and hemizigous deletions that affect gene function [<xref ref-type="bibr" rid="B2">2</xref>]. Among all the tumours showing a loss of FHIT activity, are tumours of the lower urinary tract, especially bladder tumours, one of the most frequent cancers in man. The same situation is likely to exist in cattle tumours, and our complete annotation of the bovine <italic>FHIT </italic>gene will allow to characterize the role of the gene in urothelial tumors, especially of the lower urinary tract. These are a common pathology of 4- to 12-year-old animals grazing in bracken fern (<italic>Pteridium </italic>spp) infested lands [<xref ref-type="bibr" rid="B13">13</xref>] and showing Chronic Enzootic Hematuria [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>].</p><p>Using the set of BACs covering the entire bovine <italic>FHIT </italic>region that we describe in the present work, it will be possible to scan tumour samples in search of markers that are lost due to chromosomal breakage, a finding that would confirm the existence of the bovine homologue of the human fragile site FRA3B.</p><p>Knowledge of the molecular genomic organization of the <italic>FHIT </italic>locus as gathered from our work will be exploited also to study a peculiar feature of the urinary bladder tumors of cattle. Frequently, such tumors are positive to Bovine Papilloma Virus type-2 (BPV-2), and an intriguing synergy has been suggested between the virus and some clastogenic, mutagenic and carcinogenic compounds released by the bracken fern ingested by animals at pasture. The hypothesis has been made that compounds such as ptaquiloside act synergistically with BPV-2 either in a latent or activated form to produce the neoplastic transformation of cells with a mechanism that remains largely unknown. Thanks to the detailed <italic>FHIT </italic>mapping and annotation achieved in this work, insights into the relation between the presence and activation state of BPV and tumors might come from the study of integration of BPV at the <italic>FHIT </italic>locus. As a matter of fact, we have found a short DNA region in the bovine <italic>FHIT </italic>intron 4, that is homologous to a known integration site of Human Papillomavirus type 16 at <italic>FRA3B</italic>/<italic>FHIT </italic>in man (data not shown) [<xref ref-type="bibr" rid="B6">6</xref>].</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Design of sequence tags for the selection of BACs covering the bovine <italic>FHIT </italic>region</title><p>A bovine BAC containing exon 9 of the gene was already available [<xref ref-type="bibr" rid="B10">10</xref>], and was used as a starting point to isolate adjacent clones from a BAC library [<xref ref-type="bibr" rid="B12">12</xref>]. In addition, the human <italic>FHIT </italic>genomic sequence [Build 36.1; Genbank:<ext-link ext-link-type="gen" xlink:href="NC_000003">NC_000003</ext-link>.1], nt 59710076 to 61212124 was used as a source of STSs. Primers were designed in exons, as well as in intronic portions of the gene at an average distance of 50 kb in order to conveniently cover the huge target genomic region with PCR-able STSs of a mean size of 250 bp. The whole set of oligonucleotides used is represented in Table 1 [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. The <italic>FHIT </italic>exons sequences generated have been deposited in GenBank [Genbank:<ext-link ext-link-type="gen" xlink:href="DQ310170-80">DQ310170-80</ext-link>].</p></sec><sec><title>Assembly of a BAC contig</title><p>The screen with the whole set of STSs yielded BAC clones organized in three sub-contigs. The clones were retrieved, the DNA extracted, and <italic>Not</italic>I digests were run on PFGE to estimate the size of inserts. Whithin each sub-contig the end clones were selected in order to look for adjacent overlapping clones. Thus, using the available sequence data for the identified BACs that is posted in GenBank, i.e. the two BAC-ends, primers were designed to amplify the corresponding STSs in the BAC library. Physically assigned STS tags and the output of the Finger Printed Contig (FPC) software tool [<xref ref-type="bibr" rid="B16">16</xref>] were combined to progress in the ordering of clones. In this way additional BACs were identified that allowed to link all sub-contigs together, as shown in Figure [see <xref ref-type="supplementary-material" rid="S2">Additional file 2</xref>], [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>].</p></sec><sec><title>Public database searches and identification of bovine contigs containing <italic>FHIT </italic>sequence</title><p>The available end-sequence of BACs, and the sequence of all the STS generated and mapped to the bovine <italic>FHIT </italic>contig of BAC clones, were used to search the public database of the bovine genome using BLAST [<xref ref-type="bibr" rid="B19">19</xref>]. This yielded 7 contigs: [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link> (314883 bp), <ext-link ext-link-type="gen" xlink:href="NW_001018446">NW_001018446</ext-link> (64796 bp), <ext-link ext-link-type="gen" xlink:href="NW_984819">NW_984819</ext-link> (239821 bp), <ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link> (641669 bp), <ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link> (49009 bp), <ext-link ext-link-type="gen" xlink:href="NW_977438">NW_977438</ext-link> (344934 bp), and <ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link> (1090058 bp)].</p></sec><sec><title>VISTA global alignments</title><p>The mVISTA web server [<xref ref-type="bibr" rid="B20">20</xref>,<xref ref-type="bibr" rid="B21">21</xref>] was used to align genomic sequences from the bovine genome sequencing project with the human sequence of the <italic>FHIT </italic>locus taken as reference. Thus, the HSA3 sequence [Genbank:<ext-link ext-link-type="gen" xlink:href="NC_000003">NC_000003</ext-link>.1, nt 59500000 to 61500000] spanning the entire <italic>FHIT </italic>gene was aligned with the bovine contigs [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link>, <ext-link ext-link-type="gen" xlink:href="NW_001018446">NW_001018446</ext-link>, <ext-link ext-link-type="gen" xlink:href="NW_984819">NW_984819</ext-link>, <ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>, <ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>, <ext-link ext-link-type="gen" xlink:href="NW_977438">NW_977438</ext-link>, and <ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link>]. As many of the bovine sequences are unfinished yet, long runs of unassigned bases were removed prior to analysis; human sequence and its annotation were always used as the reference sequence. Pairwise sequence comparisons were performed with a threshold of 70% identity in a 100 bp window. The VISTA plots showed 50% identity as the minimum value.</p></sec><sec><title>RT-PCR characterization of cDNAs and Northern blot analysis</title><p>Total RNA was isolated from testis, spleen, brain, lung, kidney, bladder, heart, and muscle of an healthy cow using Trizol (Invitrogen). For RT-PCR, 3 micrograms RNA was reverse transcribed with SuperScript II (Invitrogen) at 37°C for 2 hours using an oligo-(dT) primer. The subsequent PCR step was performed with oligonucleotides spanning exon 1 to 10 of the gene (Exon 1F and Exon 10R, Table 1 [see <xref ref-type="supplementary-material" rid="S1">Additional file 1</xref>]. The primers yield a PCR product of variable size around 1 kb due to the existence of four tissue isoforms (see Results).</p><p>For Northern blot analysis, 5 micrograms poly(A) mRNA, purified from total brain, testis, spleen, lung, kidney, bladder, heart, and muscle RNA using a GeneElute mRNA Miniprep Kit (Sigma), was separated on a 1.2% agarose-formaldehyde gel and transferred to HyBond N+ (Amersham). The filter was hybrized with a probe corresponding to the entire <italic>FHIT </italic>cDNA produced as above with primers Exon 1F and Exon 10R (PCR product of 914 bp), and labeled with HexaLabel Plus (MBI Fermentas) in the presence of <sup>32</sup>P α-dCTP.</p><p>The sequence of the four <italic>FHIT </italic>cDNA isoforms has been deposited in GenBank [Genbank:<ext-link ext-link-type="gen" xlink:href="DQ310181-4">DQ310181-4</ext-link>].</p></sec></sec><sec><title>Authors' contributions</title><p>CU and EG developed STS markers, performed the markers assignments to BACs, and finalized the BAC ordering in the final contig. SR, VR and FR organized tissue sample collection and DNA/RNA extraction. GPDM and LI were responsible for the RT-PCR and Northern blot analysis. SF, MB and AE performed the BAC library screening, isolated the BACs and generated with FPC the tiling of BACs in the initial contig. CU and LF performed the sequence analysis and the VISTA global alignments. CU, AE and LF were mainly responsible for conceiving and planning research that was coordinated by LF, who also drafted the first version of the manuscript. All the authors read and approved the final manuscript.</p></sec><sec sec-type="supplementary-material"><title>Supplementary Material</title><supplementary-material content-type="local-data" id="S1"><caption><title>Additional File 1</title><p><bold>Table 1</bold>. A table is presented that describes the sequence and genomic coordinates of all the primers defining the STS tags that were mapped in this work.</p></caption><media xlink:href="1471-2164-7-123-S1.doc" mimetype="application" mime-subtype="msword"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S2"><caption><title>Additional File 2</title><p><bold>Figure. A BAC contig covering the bovine <italic>FHIT </italic>genomic region. </bold>Top: the BAC clones spanning the bovine <italic>FHIT </italic>locus are aligned with the corresponding region of BTA22 into a 2 Mb contig. The identification of BACs is based on Schibler et al. [<xref ref-type="bibr" rid="B16">16</xref>]. The final contig was produced by combining fingerprint data and sequence tags data for BACs of the different subcontigs, with the merging of contigs by means of physically assigned STSs (as shown by vertical lines). Middle: schematic representation of the bovine BTA22 <italic>FHIT </italic>locus region compared to the homologous human HSA3 region shown for reference. The comparison highlights the conservation of individual tags and of the overall locus organization. Exons E1 to E10 are displayed with squares as follows: filled, coding; empty, non coding; grey, alternatively-used. Empty circles are the intronic tags adapted from the human sequence and used in the study. Triangles are three bovine microsatellites. BACs with an asterisk at the right are included by FPC in the contig output but are not linked to the others physically (see text for more details). Bottom: Eight bovine genomic sequence contigs, showing the correspondence with the assembled BAC contig. 1. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link>]; 2. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_001018446">NW_001018446</ext-link>]; 3. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_984819">NW_984819</ext-link>]; 4. and 6. are two independent portions of the original [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_930116">NW_930116</ext-link>] contig separated by contig 5 [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_935020">NW_935020</ext-link>] (for a detailed description of the arrangement of contigs see text and Figure <xref ref-type="fig" rid="F1">1</xref>); 7. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_977438">NW_977438</ext-link>]; 8. [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link>]. The entire contigs are shown, except for [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_978846">NW_978846</ext-link>] and [Genbank:<ext-link ext-link-type="gen" xlink:href="NW_968990">NW_968990</ext-link>] where only the segments that align with the BAC contig are displayed.</p></caption><media xlink:href="1471-2164-7-123-S2.ppt" mimetype="application" mime-subtype="vnd.ms-powerpoint"><caption><p>Click here for file</p></caption></media></supplementary-material><supplementary-material content-type="local-data" id="S3"><caption><title>Additional File 3</title><p><bold>Table 2. Exon/Intron arrangement of the bovine <italic>FHIT </italic>gene</bold>. <italic>FHIT </italic>exons and introns coordinates and size were calculated taking as reference the whole sequence of the bovine genomic contigs retrieved from GenBank (Bovine Genome Build 2.1). *, the contig sequence had to be reverse-complemented to be aligned properly; > the estimated size takes into account the Ns that are present in the unfinished sequence of the corresponding contig, and the Ns that currently separate adjacent contigs; ~, the estimated size may vary according to the number of Ns present in the corresponding contig only; °, The accession numbers for primary sequence data (Ref Seq) within the genomic contigs retrieved from GenBank are shown only for exons, as the number of such sequences for introns would be too high to be included in the table. The complete list of Ref Seq is part of each contig's entry in the GenBank database. The size and sequence of exons was obtained in the present work, see the quoted GenBank accession numbers. Exonic sequence is shown in capital letters. The <italic>FHIT </italic>open reading frame is in bold, with Start and Stop codons underlined.</p></caption><media xlink:href="1471-2164-7-123-S3.XLS" mimetype="text" mime-subtype="plain"><caption><p>Click here for file</p></caption></media></supplementary-material></sec>
|
Effects of dietary calcium on atherosclerosis, aortic calcification, and icterus in rabbits fed a supplemental cholesterol diet
|
<sec><title>Background</title><p>Vascular calcification is implicated in myocardial infarction, instability and rigidity of the aortic wall, and bioprosthetic failures. Although an increase in the calcium (Ca) content in atherogenic diets has been shown to decrease atherosclerosis in rabbits, whether Ca supplementation and deficiency can affect atherosclerosis-related aortic calcification remains unknown.</p></sec><sec><title>Results</title><p>New Zealand White male rabbit littermates were fed an atherogenic diet containing 0.5% cholesterol and 2% peanut oil. The Ca content of the diet, which normally contains 1%, was adjusted to 0.5 or 3%. Segments of thoracic aortas were dissected from rabbits for histological evaluations and Ca and Pi determinations. Rabbits with calcium supplementation were maintained for 4 months, whereas those with calcium deficiency were maintained for 2 1/2 months due to severe icterus beyond this stage. The ratios of intimal to medial areas and calcified to intimal areas were used to semi-quantify lesion accumulation and calcification, respectively. Icterus was estimated from the extent of yellowing of the skin, sclera, and mucous membranes along with gross evidence of hepatic lipidosis and/or biliary obstructions. Statistical analysis of 16 matched littermates shows that Ca supplementation significantly decreased the lesions by 41% (p < 0.05) and markedly inhibited calcification by 62% (p < 0.05). Statistical analysis of 11 matched littermates shows that Ca deficiency significantly increased the lesions by 2.7-fold (p < 0.05) and that the diet caused a small but significant calcification not seen in the sibling groups with normal dietary Ca. Ca supplementation caused a significant 30% decrease in serum cholesterol (p < 0.05). Calcium deficiency increased serum cholesterol by 57% (p < 0.001). Serum cholesterol and LDL-cholesterol levels in Ca deficient rabbits were 2-fold higher than those with high Ca diets. Ca supplementation decreased soluble Ca and Pi content in aortas, suggesting that this effect may underlie the effects of Ca supplementation on calcification. Calcium deficiency increased icterus by 33% (p < 0.05), which may affect hepatic clearance of cholesterol, while calcium supplementation decreased it by 43% (p < 0.001).</p></sec><sec><title>Conclusion</title><p>Ca supplementation to an atherogenic diet inhibits atherosclerosis, aortic calcification, and icterus, whereas a Ca deficient-diet promotes them.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Hsu</surname><given-names>Howard HT</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Culley</surname><given-names>Nathan C</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
Lipids in Health and Disease
|
<sec><title>Background</title><p>Increased Ca content in diets supplemented with cholesterol has been shown to decrease atherosclerosis in rabbits [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. Two epidemiological studies suggested that high levels of Ca in drinking water may decrease atherosclerosis [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. However, whether the increase in Ca intake could also affect aortic calcification through altered levels of Ca × P ion products in aortas was unclear. It is equally possible that Ca supplementation or deficiency could affect the calcification through its primary effect on atherosclerosis. Since vascular calcification is implicated in myocardial infarction [<xref ref-type="bibr" rid="B5">5</xref>], instability and rigidity of the aortic wall [<xref ref-type="bibr" rid="B6">6</xref>], and bioprosthetic failures [<xref ref-type="bibr" rid="B7">7</xref>], it is imperative to study the effect of dietary Ca on aortic calcification.</p><p>During the course of this study, we also found that cholesterol diets often caused increased evidence of icterus from liver impairment, which means the diets may impede the ability of liver to metabolize lipids, thereby influencing serum cholesterol levels. In this report, we sought to determine whether dietary Ca also affects icterus associated with supplemental cholesterol diets. We herein show that Ca supplementation decreases while deficiency increases lesion formation and aortic calcification in rabbit littermates. For the first time, we demonstrate that dietary Ca deficiency increases the chance of developing signs of icterus, while calcium supplementation decreases the chance of developing icterus in rabbits receiving the atherogenic diet.</p></sec><sec><title>Results and discussion</title><sec><title>Effects of calcium supplementation on atherosclerosis and aortic calcification</title><p>We previously demonstrated that cholesterol supplementation induced aortic calcification in the lower zone of lesions (LZP) adjacent to the media progressing along LZP [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>]. After 4 months of the atherogenic diet, gross examinations of the thoracic aortas indicated that the proximal arch was fully covered with lesions. In contrast, about half of the middle sections (2.5 cm from the arch joint) were covered with the lesions. At this stage, about half of the lesions in the proximal sections were filled with mineral deposits identifiable with alizarin red stains (AR), whereas the distal sections were rarely calcified. Thus, the selection of the middle and proximal sections for the quantification of lesions and mineral accumulation, respectively, was preferable for statistical analysis. These effects on lesion formation and calcification were semi-quantified using relative ratios of the intimal to medial areas and the calcified to intimal areas, respectively, as described in the "Experimental Procedure" section. Fig. <xref ref-type="fig" rid="F1">1A & B</xref> show a typical example of the inhibitory effect of calcium supplementation for 4 months on the accumulation of lesions within a sibling group. A rabbit sibling with normal levels of Ca in the atherogenic diet exhibited a 0.77 ratio of the intima to media (Fig. <xref ref-type="fig" rid="F1">1A</xref>) vs. a 0.32 ratio from another sibling with Ca supplementation (Fig. <xref ref-type="fig" rid="F1">1B</xref>). Fig. <xref ref-type="fig" rid="F1">1C & D</xref> show the effect of calcium supplementation on calcification. The ratios of calcified areas to the lesions were 0.47 and 0, respectively, for a sibling with normal Ca diets and the other with calcium supplementation. Statistical analysis of 16 matched littermates shows that Ca supplementation for 4 months significantly decreased the lesions by 41% (p < 0.05) and markedly inhibited calcification by 62% (p < 0.05) (Fig. <xref ref-type="fig" rid="F1">1E</xref>). The control rabbits fed normal chow for 4 months developed neither lesions nor aortic calcification (not shown). The effect of calcium supplementation on lesion formation is consistent with the observation by Yacowitz et al. [<xref ref-type="bibr" rid="B1">1</xref>] and Renaud et al. [<xref ref-type="bibr" rid="B2">2</xref>]. For the first time, we demonstrate that calcium supplementation inhibits aortic calcification. Since aortic calcification visible by AR staining did not occur until the lesions were well developed [<xref ref-type="bibr" rid="B8">8</xref>,<xref ref-type="bibr" rid="B9">9</xref>], it is likely that calcium effects could be a consequence of the primary effect of supplemental Ca on atherosclerosis.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Effects of Ca supplementation for 4 months on the accumulation of plaques and minerals in aortas. Panels A–D show typical examples of the effects of calcium supplementation on accumulation of lesions and calcification in a sibling group. The rationale for the selection of the middle and proximal sections, respectively, for lesion accumulation and calcification was given in detail in the Experimental Procedures section. The cross sections of the middle section of aortas from sibling brothers were stained with H&E (Panels A and B) to show intimal thickening (lesions). The cross sections of the proximal section were stained with AR (Panels C and D) to show mineralization (arrows). A & C are cross sections obtained from rabbits fed the atherogenic diet with normal levels of dietary Ca. B and D are cross sections from rabbits with Ca supplementation. The thickened intima to media ratio was used to semi-quantify the extent of lesion accumulation as described in "Experimental Procedures" section. Likewise, the extent of calcification was expressed as the ratio of calcified area to the area of thickened intima. Intima/media ratios: A, 0.77 vs. B, 0.32. Calcified area/intima ratio: C, 0.47 vs. D, 0.00. Panel E provides the paired Student's <italic>t</italic>-test for the significance of the differences in the ratios between the two respective groups each with 16 siblings.</p></caption><graphic xlink:href="1476-511X-5-16-1"/></fig></sec><sec><title>Effects of dietary calcium deficiency on atherosclerosis and aortic calcification</title><p>Fig. <xref ref-type="fig" rid="F2">2</xref> shows the effect of dietary calcium deficiency for 2 1/2 months on accumulation of lesions and calcification in a sibling group. The Ca deficiency study was not extended beyond this time period since rabbits barely survived due to severe liver impairment and subsequent icterus. The effect was semi-quantified using the ratios of intima thickening/media for lesion formation and calcified area/intima for mineral deposition. A rabbit sibling with normal levels of dietary Ca in the atherogenic diet exhibited neither lesions in the cross section of the middle part of aortas (Fig. <xref ref-type="fig" rid="F2">2A</xref>) nor mineral deposits in the proximal part (Fig. <xref ref-type="fig" rid="F2">2C</xref>). In contrast, a sibling with the calcium deficient diet shows formation of lesions with an intima/media ratio of 0.28 (Fig. <xref ref-type="fig" rid="F2">2B</xref>) and calcified area/intima ratio of 0.12 (Fig. <xref ref-type="fig" rid="F2">2D</xref>). Statistical analysis of 11 matched littermates shows that Ca deficiency significantly increased the lesions by 2.7-fold (p < 0.05) and that the diet caused a small but significant calcification not seen in the sibling groups with normal dietary Ca (Fig. <xref ref-type="fig" rid="F2">2E</xref>). The effect of the Ca deficient diet on atherosclerosis is consistent with the observations of Yacowitz et al. [<xref ref-type="bibr" rid="B1">1</xref>] and Renaud et al. [<xref ref-type="bibr" rid="B2">2</xref>]. For the first time, we demonstrate that calcium deficiency promotes aortic calcification.</p><fig position="float" id="F2"><label>Figure 2</label><caption><p>Effects of the calcium deficient diet for 2 1/2 months on the development of lesions and calcification in sibling rabbits. The cross sections of the middle section of aortas were stained with H&E (Panels A and B) and the proximal section with AR (Panel C and D). Panels A and C: Cross-sections obtained from a sibling fed the atherogenic diet with normal levels of Ca. Panels B and D: The sections from a sibling fed the atherogenic diet with calcium deficiency. The semi-quantification for intimal thickening and calcification was described in Fig. 1 legend. Panel E provides the paired Student's <italic>t</italic>-test for the significance of the differences in these ratios between the two respective groups each with 11 siblings. Note that the graphic bar representing calcification for normal dietary Ca groups was not seen since there was no calcification in this group during 2 1/2 months of cholesterol intervention.</p></caption><graphic xlink:href="1476-511X-5-16-2"/></fig></sec><sec><title>Effects of dietary Ca on serum lipids in rabbits fed high cholesterol diets</title><p>To determine whether dietary Ca may affect levels of various lipid risk factors in the serum and thereby influence atherosclerosis and calcification, we measured the serum lipids from rabbits treated with various Ca diets. Consistent with those findings by Yacowitz et al. [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B10">10</xref>], Ca supplementation caused a significant 30% decrease in serum cholesterol (p < 0.05, Fig. <xref ref-type="fig" rid="F3">3</xref>). Calcium deficiency increased serum cholesterol by 57% (p < 0.001). Except for LDL cholesterol, which was significantly reduced by calcium supplementation and increased by calcium deficiency (data not shown), HDL-cholesterol and triglyceride were not significantly affected by Ca diets. To further support this conclusion, we compared the difference between calcium supplementation and deficiency on these two parameters. As shown in Fig. <xref ref-type="fig" rid="F4">4</xref>, serum cholesterol and LDL-C levels in Ca deficient rabbits were 2-fold higher than those with high Ca diets. Although HDL, HDL/LDL ratios, and triglyceride seemed different between these two groups, the effects were statistically insignificant (p > 0.05). A further statistical analysis revealed the Power levels of 30–40% at α <0.05, indicating that it is unlikely that Ca deficiency affected these parameters.</p><fig position="float" id="F3"><label>Figure 3</label><caption><p>Effects of dietary Ca on serum cholesterol. The total serum cholesterol was measured using a commercial kit (Thermo DMA). Panel A shows the effect of calcium supplementation (3% Ca) vs. the control group (1% Ca) for 4 months. Panel B shows the effect of calcium deficiency (0.5% Ca) for 2 1/2 months. There were 11 matched littermates in each group.</p></caption><graphic xlink:href="1476-511X-5-16-3"/></fig><fig position="float" id="F4"><label>Figure 4</label><caption><p>Effects of Ca deficiency on serum levels of lipid risk factors. Since the serum levels of HDL-cholesterols and triglyceride in calcium deficient group were insignificantly different from the normal Ca group (p > 0.05), a comparison between calcium deficiency and supplementation was used to better see the Ca effects. The LDL- and HDL-cholesterol and triglyceride from diluted serum samples were measured using an automation system in a clinical laboratory. There were 11 matched littermates between high and low calcium groups.</p></caption><graphic xlink:href="1476-511X-5-16-4"/></fig></sec><sec><title>Effect of dietary Ca on icterus development in rabbits fed high cholesterol diets</title><p>During the course of this investigation, we often found cholesterol supplementation caused icterus in the rabbits. Since icterus may be a result of liver impairment which can affect hepatic clearance of lipids, thereby altering serum lipid levels, the effects of Ca on the development of icterus were examined. As shown in Fig. <xref ref-type="fig" rid="F5">5</xref>, calcium deficiency increased icterus by 33% (p < 0.05) while calcium supplementation decreased it by 43% (p < 0.001).</p><fig position="float" id="F5"><label>Figure 5</label><caption><p>Effects of Ca supplementation and deficiency on icterus. Icterus was diagnosed by the extent of yellowing of the skin, sclera, and mucous membranes along with gross evidence of hepatic lipidosis. A scale of 5 was arbitrarily given using the combined signs of icterus to indicate the relative degree of icterus. There were 11 matched littermates among 3 groups with various dietary Ca levels.</p></caption><graphic xlink:href="1476-511X-5-16-5"/></fig></sec><sec><title>Effect of dietary Ca on serum Ca</title><p>The mechanism of dietary Ca on atherosclerosis and aortic calcification is difficult to elucidate since Ca has diverse effects on a variety of cells. It is conceivable that dietary Ca in atherogenic diets could affect the serum Ca levels. The changes in the serum Ca may then affect tissue level of Ca, thereby affecting lipid metabolism or cellular vulnerability to cholesterol. As shown in Fig. <xref ref-type="fig" rid="F6">6</xref>, cholesterol supplementation significantly increased serum Ca levels from a normal level of 3.70 ± 0.57 mM (10) to 4.49 ± 0.69 (10), representing a 20% increase (p < 0.001). Ca levels in sera from rabbits with Ca supplementation and deficiency were 4.92 ± 0.27 (11) and 4.60 ± 0.29 (11), respectively. The differences between these two groups were statistically insignificant (p > 0.05). The effect of Ca diets on serum Pi was statistically insignificant likely due to the large fluctuations in the serum phosphate content in contrast to serum Ca (data not shown). The pathological relevance of increased levels of serum Ca as a result of cholesterol feeding was unclear as was the ineffectiveness of dietary Ca to alter serum Ca levels.</p><fig position="float" id="F6"><label>Figure 6</label><caption><p>Effects of cholesterol and Ca diets on serum Ca. One-Way variance analysis was used to evaluate differences among groups. Cholesterol diet caused a significant higher serum Ca than the control group (p < 0.01). There were no differences among atherogenic diet groups with various levels of Ca.</p></caption><graphic xlink:href="1476-511X-5-16-6"/></fig></sec><sec><title>Effect of dietary Ca on the accumulation of soluble Ca in aortas</title><p>To determine whether Ca diets may alter the accumulation of Ca in aortas, thereby affecting aortic calcification, soluble Ca from aortas was measured. Soluble Ca fractions were obtained by centrifuging the collagenase digests of aorta fragments at 300,000 × g for 30 min to precipitate all subcellular organelles. Cholesterol diets for 4 months caused a significant increase in the concentrations of soluble Ca and Pi (Fig. <xref ref-type="fig" rid="F7">7</xref>). Both mineral and the soluble Ca and Pi were decreased by Ca supplementation (p < 0.05). In rabbits on Ca deficient food for 2 1/2 months, there was a small increase in soluble Ca in aortas (data not shown), consistent with only a 10% of the cross-sections being calcified (Fig. <xref ref-type="fig" rid="F2">2D</xref>).</p><fig position="float" id="F7"><label>Figure 7</label><caption><p>The inhibitory effect of calcium supplementation for 4 months on the dietary cholesterol-induced accumulation of mineral and soluble Ca/phosphate in aortas. The mineral content was expressed as the levels of the Ca (upper panel) and phosphate (lower panel) in 1% acetic acid-soluble fractions of 800 × g precipitates obtained from collagenase digestion of 1-cm aortas [8]. The soluble forms of Ca and phosphate were expressed as the Ca and phosphate content in the supernatant fractions, which were obtained by ultracentrifugation of the collagenase digests at 250,000 × g for 30 min.</p></caption><graphic xlink:href="1476-511X-5-16-7"/></fig><p>In summary, the observations herein reveal that dietary Ca levels in atherogenic diets not only profoundly affect atherosclerosis but also influence aortic calcification and icterus associated with cholesterol supplementation. Although the effects of dietary Ca were remarkable, the precise mechanisms whereby dietary Ca exerts its effects on atherosclerosis and calcification would be difficult to elucidate due to the multiplicity of Ca roles in various signal pathways in target cells and tissues.</p></sec></sec><sec><title>Conclusion</title><p>Ca supplementation to an atherogenic diet inhibits atherosclerosis, aortic calcification, and icterus, whereas a Ca deficient-diet promotes them.</p></sec><sec sec-type="methods"><title>Experimental procedures</title><sec><title>Adjustments of dietary Ca levels in rabbit chows supplemented with cholesterol</title><p>To minimize the variation in the development of atherosclerosis from cholesterol supplementation because of genetic contributions [<xref ref-type="bibr" rid="B11">11</xref>-<xref ref-type="bibr" rid="B13">13</xref>], only siblings of young male 4-month-old New Zealand White rabbits (weighed about 2.5 kg) from identical litters were used. The rabbits were divided into 3 groups according to the levels of Ca in an atherogenic diet containing 0.5% cholesterol and 2% peanut oil. The first group was fed the atherogenic diet containing a normal 1% of dietary Ca by weight. The second group received the same atherogenic diet in which the Ca content was increased to 3 %. The third group was given the atherogenic diet in which the Ca level was reduced to 0.5%, which is the minimal that the manufacturer was able to prepare due to intrinsic Ca ingredients in rabbit chows (Harlan Teklad, Wisconsin). Since approximately 3 to 5 males per litter are available, separate litters at different times were needed for a statistically paired analysis. The content of Ca and cholesterol in the diet was routinely assessed by dissolving pellets in 1% acetic acid and 100% pyridine, respectively.</p></sec><sec><title>Estimation of lesions and mineralization in rabbit thoracic aortas</title><p>Lesions developed fully in the proximal arch of thoracic aortas after 4 months of cholesterol feeding and progressed toward the distal section [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. At this stage, about 50% of the middle sections (2.5 cm from the arch joint) were covered with thickened intima. Thus, we selected the middle section as an index of atherosclerosis to allow optimal statistical analysis for assessing the effects of dietary Ca. To study the effect of dietary Ca on calcification, the proximal sections of the aortas were selected as an index of calcification since calcification occurred mostly in this region. Less than 15% of the mid sections of the aortas was calcified and therefore would be far too small for statistical comparisons. Accordingly, the selection of different locations of the aorta was necessary for statistical analysis of the two processes. To semi-quantify the lesions and mineralization, the cross sections were stained with H&E and alizarin red, respectively, photographed and the images were magnified for photo prints. The weight ratios of the magnified images of the thickened intima to the media from the prints were calculated for lesion quantification. Likewise, for mineralization estimation, the alizarin red stained images were separated from the non-stained area of proximal thickened intima and weighed. The weight ratios of the images of the calcified area to the thickened intima were calculated as an index of calcification.</p></sec><sec><title>Icterus diagnosis</title><p>Icterus was estimated from the extent of yellowing of the skin, sclera, and mucous membranes along with gross evidence of hepatic lipidosis and/or biliary obstructions assessed by the veterinarians at the Animal Resource Support Center. A scale of 0–5 was given to assess the degree of icterus.</p></sec><sec><title>Serum lipids and Ca determinations</title><p>Since both serum and plasma are routinely used in rabbit lipid research, blood samples were collected and allowed to coagulate at room temperature without the use of heparin or EDTA, which otherwise affects Ca determinations. The blood samples were centrifuged at 1800 rpm for 10 min and the sera were collected for lipid analyses. The samples were properly diluted to standard assay ranges. The LDL-C content was directly measured by a standard clinical procedure without the application of Friedewald equation, which requires HDL and triglyceride data inputs to estimate LDL-C. The direct measurement procedure is included in the National Cholesterol Education Program Working Group on Lipoprotein Measurements. The lipid profile was assayed using a Beckman Synchron automation system in a clinical laboratory. The data are means ± S.E. The <italic>p </italic>values from the Student's <italic>t </italic>test denote statistical significance of differences between experimental and control rabbits. Ca and Pi determinations were performed using a commercial Arsenazo III dye kit (Sigma Chemicals Inc.) and a standard phosphate-molybdate complex colorimetric procedure, respectively.</p></sec></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>Howard H.T. Hsu, Ph.D. was responsible for the experimental designs, data interpretations, and execution of experiments. Nathan Culley, D.V. M. played a major role in icterus determinations, welfare and sacrifice of the rabbits, and collections of serum and aortas.</p></sec>
|
Prognostic significance of new onset ascites in patients with pancreatic cancer
|
<sec><title>Background</title><p>The purpose of this study was to determine risk factors for development of malignant ascites and its prognostic significance in patients with pancreatic cancer.</p></sec><sec sec-type="methods"><title>Methods</title><p>A prospective database was queried to identify patients with pancreatic cancer who develop ascites. Stage at presentation, size, and location of primary tumor, treatment received and length of survival after onset of ascites were determined.</p></sec><sec><title>Results</title><p>A total of 15 patients were identified. Of which 4 patients (1 stage II, 3 stage III) underwent pancreaticoduodenectomy and manifested with ascites 2, 3, 24 and 47 months after surgery (tumor size 2.9 ± 1.32 cm). All but one of the remaining 11 patients (tumor size 4.4 ± 3.38 cm) presented with metastatic disease, and all developed malignant ascites 9 months after diagnosis, dying 2 months later. Resected patients lived longer before the onset of ascites, but not after.</p></sec><sec><title>Conclusion</title><p>Once diagnosed, ascites in pancreatic cancer patients heralds imminent death. Limited survival should be considered when determining the aggressiveness of further intervention.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Zervos</surname><given-names>Emmanuel E</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Osborne</surname><given-names>Dana</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Boe</surname><given-names>Brian A</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Luzardo</surname><given-names>German</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Goldin</surname><given-names>Steven B</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Rosemurgy</surname><given-names>Alexander S</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
World Journal of Surgical Oncology
|
<sec><title>Background</title><p>Pancreatic cancer is the 5<sup>th </sup>most common gastrointestinal (GI) malignancy but the third most common cause of death among all GI cancers [<xref ref-type="bibr" rid="B1">1</xref>]. Almost all patients with pancreatic cancer ultimately die of it, usually within 6 months of diagnosis [<xref ref-type="bibr" rid="B2">2</xref>]. The vast majority of these patients are ineligible for surgical therapy at the time of diagnosis due to metastatic spread or local invasion and death usually occurs by inanition rather than gross tumor burden. Aggressive chemotherapeutic regimens are generally highly toxic and prolong life by weeks, rather than months [<xref ref-type="bibr" rid="B3">3</xref>]. Most patients with unresectable pancreatic cancer die of their disease before completing a standard course of therapy.</p><p>Common manifestations of end stage pancreatic cancer include: gastric outlet obstruction due to tumor ingrowth to the duodenum, cachexia, deep venous thrombosis (Trousseau's phenomenon), anasarca and ascites. Ascites manifests in only 20% of pancreatic cancer patients and its cause is multifactorial [<xref ref-type="bibr" rid="B4">4</xref>]. It can occur due to obstruction of diaphragmatic lymphatics, increased production of exudate by the tumor itself and production of osmotically active peptides that alter vascular permeability to favor ascites formation [<xref ref-type="bibr" rid="B5">5</xref>]. In most series, over half of the time, cancer cells can not be identified in a paracentesis sample [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. As such, in this setting, the ascites may not be malignant in the purest sense however; it is undoubtedly due to the cancer even if cancer cells cannot be found in the fluid. Other authors have suggested methods other than cytology to determine the nature of ascites such as ascites/serum albumin gradients or presence of various substances such as telemorase or human gonadotropin-β in the ascites [<xref ref-type="bibr" rid="B7">7</xref>-<xref ref-type="bibr" rid="B9">9</xref>]. We have noted that regardless of whether ascites can be proven to be malignant cytologically, it is usually the final manifestation of a uniformly fatal diagnosis.</p><p>Malignant ascites can be managed for extended periods in patients with other types of cancer such as those of gynecologic, gastric or colonic origin [<xref ref-type="bibr" rid="B10">10</xref>]. We have noted, however, after treating multiple patients with both resected and non-resected pancreatic cancer on various protocols at our institution that regardless of the type of intervention employed, the new onset of ascites in these patients usually heralded imminent demise. The purpose of this study, therefore, was to examine this phenomenon and determine the exact prognostic implications of new onset ascites to better guide subsequent treatment and to counsel patients more effectively. We hypothesized that new onset ascites in patients with pancreatic cancer would be associated with those factors commonly attributed to the final stages of the disease and that death would soon follow its manifestation.</p></sec><sec sec-type="methods"><title>Patients and methods</title><p>A prospective database involving all patients treated on protocols for pancreatic cancer at our institution was begun in 2001. All patients were entered into this database with Institutional Review Board (IRB) approval after giving informed consent. The database chronicles each patient's course from the time of enrollment into a protocol to the time of death, regardless of whether they complete the protocol. Patients who presented with, or developed ascites after having been diagnosed with pancreatic cancer were included in this analysis. Chart review was undertaken to gain basic demographic data including patient age, gender, tumor characteristics such as size, and location. Stage and tumor marker levels were determined at three different time points: presentation, onset of ascites and just prior to death.</p><p>After identifying all patients with ascites, patients were divided into two groups: those having undergone resection and those managed non-operatively. The type of treatment employed for both the primary tumor and the ascites was noted. The incidence of malignant cells in the ascites was noted as was the method by which the ascites was diagnosed. Overall survival, ascites free interval and survival with ascites were determined for the entire cohort. Survival after therapeutic intervention was also noted after separating the patients into those undergoing no treatment, diuretics alone, paracentesis alone, diuretics and paracentesis or peritoneovenous shunting. Double valved Denver<sup>® </sup>(Denver Biomaterials, Denver Colorado) peritoneovenous shunts were employed in all patients undergoing this type of treatment. In cases where a patient's status was unknown, attempts were made to contact the patient or their family. All data is current to March of 2005.</p><p>The database was queried to determine whether any specific predictors of risk of developing malignant ascites could be gleaned. Linear regression was utilized to determine whether tumor size, lymph node status, margin status (if resected), grade, stage at presentation, tumor location, stage at onset of ascites, presence of neural or lymphovascular invasion, or type of therapy (chemotherapy <italic>v</italic>. chemoradiation) would predict the onset of ascites.</p><p>Where appropriate, tumor size and survival characteristics were compared using a two tailed Student's T-test. Significance was accepted with 95% probability.</p></sec><sec><title>Results</title><sec><title>Demographics and basic tumor characteristics</title><p>A total of 120 patients were enrolled in protocols at our institution between 2001 and late 2004. Of these 120 patients, 15 were identified who developed ascites after the diagnosis of pancreatic cancer. All but two were men and average age was consistent with that which is commonly observed in this disease process. Almost all tumors were located in the head of the gland, one patient presenting with intermittent jaundice, an elevated carbohydrate antigen 19-9 (CA 19-9) level and no radiographic evidence of disease was diagnosed with peritoneal carcinomatosis and ascites at the time of celiotomy; his primary tumor was occult. Four patients underwent resection of their tumors and although their tumors were almost 1.5 cm smaller at presentation than those patients not undergoing resection, this difference was not statistically significant. Ascites was most commonly diagnosed by physical examination followed by radiographically and then as an incidental finding at celiotomy. In these patients, ascites manifested between the time of their preoperative computed tomography (CT) scan suggesting resectability and the time of their exploration and none were resected. Demographics for the entire group of patients are depicted in Table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Patient Demographics</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>Number of patients</bold></td></tr></thead><tbody><tr><td align="left"><bold>Age at diagnosis</bold></td><td align="center">60 ± 10.4 years</td></tr><tr><td align="left"><bold>% Male</bold></td><td align="center">87%</td></tr><tr><td align="left"><bold>Location of Tumor</bold></td><td></td></tr><tr><td align="left"> Head</td><td align="center">13</td></tr><tr><td align="left"> Body</td><td align="center">0</td></tr><tr><td align="left"> Tail</td><td align="center">1</td></tr><tr><td align="left"> Unknown</td><td align="center">1</td></tr><tr><td align="left"><bold>Average size of tumor (cm)</bold></td><td></td></tr><tr><td align="left"> Overall</td><td align="center">4.2 ± 3.04 cm</td></tr><tr><td align="left"> Resected</td><td align="center">2.9 ± 1.32 cm</td></tr><tr><td align="left"> Unresectable</td><td align="center">4.4 ± 3.38 cm</td></tr><tr><td align="left"><bold>Presentation</bold></td><td></td></tr><tr><td align="left"> Resected</td><td align="center">4</td></tr><tr><td align="left"> Locally Advanced</td><td align="center">6</td></tr><tr><td align="left"> Metastatic</td><td align="center">5</td></tr><tr><td align="left"><bold>Ascites Diagnosed</bold></td><td></td></tr><tr><td align="left"> At celiotomy</td><td align="center">2</td></tr><tr><td align="left"> Physical exam</td><td align="center">10</td></tr><tr><td align="left"> Radiographically</td><td align="center">3</td></tr></tbody></table></table-wrap></sec><sec><title>Cytology</title><p>All patients underwent paracentesis or intraoperative collection of fluid for cytologic analysis. Only 5 of the 15 patients in this study had malignant cells in their ascites. Figure <xref ref-type="fig" rid="F1">1</xref> is a representative photomicrograph of a positive cytologic preparation in one such patient.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Photomicrograph with cytologically positive specimen in a patient with pancreatic cancer and new onset ascites (hematoxylin and eosin stain, ×1000).</p></caption><graphic xlink:href="1477-7819-4-16-1"/></fig></sec><sec><title>Survival</title><p>Survival for all patients, those undergoing resection and those receiving non-operative management of their cancer is summarized in Table <xref ref-type="table" rid="T2">2</xref>. Overall survival was just less than one year for the group as a whole. Survival in resected patients was near significantly longer (p = 0.06) than non-resected patients. Ascites free survival was significantly longer in patients undergoing resection but once ascites became manifest, survival was no different in patients who had undergone resection than those not undergoing resection. Once ascites developed, patients treated non-operatively lived about 7 weeks longer than those that developed ascites after pancreaticoduodenectomy. Overall survival after ascites developed was slightly over 2 months.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Survival in months</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>Overall</bold></td><td align="center"><bold>Resected</bold></td><td align="center"><bold>Unresected</bold></td><td align="center"><bold>p value</bold></td></tr></thead><tbody><tr><td align="center"><bold>n</bold></td><td align="center">15</td><td align="center">4</td><td align="center">11</td><td></td></tr><tr><td align="center"><bold>From diagnosis</bold></td><td align="center">11.2 ± 12.10</td><td align="center">20.8 ± 20.22</td><td align="center">7.7 ± 5.65</td><td align="center">0.06</td></tr><tr><td align="center"><bold>Ascites free</bold></td><td align="center">9.0 ± 12.06</td><td align="center">20 ± 20.32</td><td align="center">5.1 ± 3.86</td><td align="center">0.02</td></tr><tr><td align="center"><bold>With ascites</bold></td><td align="center">2.2 ± 2.56</td><td align="center">0.8 ± 0.43</td><td align="center">2.6 ± 2.85</td><td align="center">0.24</td></tr></tbody></table></table-wrap></sec><sec><title>Survival based on treatment rendered for ascites</title><p>Patients managed with diuretics alone or peritoneovenous shunts seemed to do better in this group than those receiving no treatment or paracentesis with or without diuretics although none of these intervals were statistically different. Table <xref ref-type="table" rid="T3">3</xref> chronicles outcome based on treatment method.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Survival by Treatment Method</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>None</bold></td><td align="center"><bold>Diuretics</bold></td><td align="center"><bold>Paracentesis</bold></td><td align="center"><bold>Both</bold></td><td align="center"><bold>PV shunt</bold></td></tr></thead><tbody><tr><td align="center"><bold>N</bold></td><td align="center">3</td><td align="center">3</td><td align="center">3</td><td align="center">3</td><td align="center">3</td></tr><tr><td align="center"><bold>Survival (months)</bold></td><td align="center">1.2 ± 1.16</td><td align="center">2.7 ± 2.61</td><td align="center">1.2 ± .80</td><td align="center">1 ± .73</td><td align="center">4.7 ± 4.63</td></tr></tbody></table></table-wrap></sec><sec><title>Stage and tumor markers</title><p>Patients were evenly distributed between stages at the time of presentation but quickly progressed to Stage IV disease at the time their ascites was discovered. This includes those patients who were not automatically staged as metastatic because of positive fluid cytology. Ascites rarely occurred in patients who were not advanced stage.</p><p>Tumor markers were higher at diagnosis than at the onset of ascites reflecting again the advanced nature of the disease at the time it was discovered. At presentation of ascites, tumor markers were lower, but quickly reached their highest point at those levels drawn just before each patient's death. Wide variations in levels of tumor markers at each interval resulted in no statistical difference in CA 19-9 at various time points.</p></sec><sec><title>Multivariate analysis</title><p>Linear regression revealed no characteristics to predict the onset of ascites. Tumor size approached but given the size of the overall database, did not reach statistical significance. Location, lymph node status, histologic grade, and type of or utilization of chemotherapy or chemoradiotherapy neither predicted nor prevented new onset ascites. Presence of unexplained free fluid at the time of celiotomy always predicted eventual onset of ascites, but this was not categorized as "ascites" in the database and, as such, was not predictive.</p></sec></sec><sec><title>Discussion</title><p>Pancreatic cancer remains a diagnostic and therapeutic challenge. No effective medical therapies have emerged, despite rapid advancements in targeted chemotherapeutic agents. Patients with pancreatic cancer continue to die with relatively low tumor burden when compared to other cancer types. Ascites, whether proven to be malignant or not, is a harbinger of the final stages of pancreatic cancer. It occurs relatively infrequently, in part because many patients do not survive long enough to manifest with it. This is the first study that documents a commonly observed clinical phenomenon; the appearance of ascites in pancreatic cancer is associated with minimal remaining life expectancy, regardless of prior intervention for the treatment of the primary tumor and regardless of how the ascites is managed.</p><p>In this study, we have looked at 15 patients obtained from a common database composed of patients enrolled in various adjuvant and palliative protocols for pancreatic cancer at our institution. Most patients had tumors located in the head of their gland, and over two thirds were unresectable. As one would expect, the ascites free interval was longer in the four patients undergoing complete resection of their tumor, but once the tumors recurred, as almost all do, life expectancy with ascites was less than one month. There were no predictive indicators of who would develop ascites based on findings at initial presentation. Patients who were not resected did slightly better in this regard in that they survived slightly longer after the appearance of ascites, and this is probably a manifestation of the survival benefit, albeit minimal, of ongoing cytotoxic chemotherapy. Other authors have shown median survival of approximately the same length once ascites manifests [<xref ref-type="bibr" rid="B11">11</xref>]. Nonetheless, the salutary effects of systemic chemotherapy were not great enough to produce a meaningful difference in survival with ascites between resected patients who had completed their therapy months earlier and non resected patients who were undergoing therapy when the ascites appeared.</p><p>Tumor markers did not correlate with onset of ascites, and this is not surprising given that only one in three patients actually had tumor cells in their ascites fluid; perhaps this is the most important observation that can be gleaned from this study. Cytology does not appear to be important – the mere <italic>presence </italic>of ascites is what really matters. Large tumors shed tumors cells, but smaller tumor secrete vasoactive peptides that change osmotic gradients and cause ascites; this was probably the case in patients with ascites and no carcinomatosis or with radiologically occult primaries. Most patients never came to surgery to document the presence or absence of carcinomatosis; we presume it was present but cannot say for sure. Regardless of whether carcinomatosis could be documented, ascites, in and of itself, was a uniformly poor prognostic sign. This statement is supported by the wealth of data regarding the grave prognostic implications of positive peritoneal washings in patients with pancreatic cancer and radiologically resectable tumors [<xref ref-type="bibr" rid="B12">12</xref>-<xref ref-type="bibr" rid="B15">15</xref>]. At our institution, routine washings of the peritoneum are not obtained. First and foremost, resection is pursued based upon preoperative imaging studies and intraoperative findings since it offers the best survival advantage. As a result of this analysis, we now regard the presence of unexplained intraperitoneal fluid at the time of celiotomy as an absolute contraindication to resection.</p></sec><sec><title>Conclusion</title><p>Based on our clinical impression and now, this study, when we observe it, we seriously weigh the utility of continuing chemotherapy with toxic side effects. Our focus shifts towards palliation and supportive care and, although peritoneovenous shunts do not prolong life, we consider them supportive care in this patient group. Anything that keeps patients out of the hospital and reasonably palliated can only serve to enhance the short time left in the course of their illness.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>EZ was responsible for acquisition of substantial data, drafting of manuscript, and study design. <bold>DO </bold>was responsible for managing database, statistical analysis, data aquisition and study design. <bold>BB </bold>maintained database and assisted in drafting the manuscript. <bold>GL </bold>was responsible for identifying patients and review of patient records. <bold>SG </bold>was involved in coordinating study and design. <bold>AR </bold>was responsible for study conception and design.</p><p>All authors read and approved the final version of the manuscript.</p><table-wrap position="float" id="T4"><label>Table 4</label><caption><p>Stage and tumor markers</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="center"><bold>At diagnosis</bold></td><td align="center"><bold>At ascites</bold></td><td align="center"><bold>At death</bold></td></tr></thead><tbody><tr><td align="center"><bold>AJCC Stage</bold></td><td></td><td></td><td></td></tr><tr><td align="center"><bold>II</bold></td><td align="center">5</td><td align="center">1</td><td align="center">0</td></tr><tr><td align="center"><bold>III</bold></td><td align="center">2</td><td align="center">3</td><td align="center">2</td></tr><tr><td align="center"><bold>IV</bold></td><td align="center">8</td><td align="center">11</td><td align="center">13</td></tr><tr><td align="center"><bold>CA 19-9</bold></td><td align="center"><bold>1532 ± 2751</bold></td><td align="center"><bold>707 ± 761</bold></td><td align="center"><bold>2948 ± 5531</bold></td></tr></tbody></table></table-wrap></sec>
|
The effect of Tai Chi Chuan in reducing falls among elderly people: design of a randomized clinical trial in the Netherlands [ISRCTN98840266]
|
<sec><title>Background</title><p>Falls are a significant public health problem. Thirty to fifty percent of the elderly of 65 years and older fall each year. Falls are the most common type of accident in this age group and can result in fractures and subsequent disabilities, increased fear of falling, social isolation, decreased mobility, and even an increased mortality. Several forms of exercise have been associated with a reduced risk of falling and with a wide range of physiological as well as psychosocial health benefits. Tai Chi Chuan seems to be the most promising form of exercise in the elderly, but the evidence is still controversial.</p><p>In this article the design of a randomized clinical trial is presented. The trial evaluates the effect of Tai Chi Chuan on fall prevention and physical and psychological function in older adults.</p></sec><sec><title>Methods/Design</title><p>270 people of seventy years and older living at home will be identified in the files of the participating general practitioners. People will be asked to participate when meeting the following inclusion criteria: have experienced a fall in the preceding year or suffer from two of the following risk factors: disturbed balance, mobility problems, dizziness, or the use of benzodiazepines or diuretics. People will be randomly allocated to either the Tai Chi Chuan group (13 weeks, twice a week) or the no treatment control group.</p><p>The primary outcome measure is the number of new falls, measured with a diary. The secondary outcome measures are balance, fear of falling, blood pressure, heart rate, lung function parameters, physical activity, functional status, quality of life, mental health, use of walking devices, medication, use of health care services, adjustments to the house, severity of fall incidents and subsequent injuries. Process parameters will be measured to evaluate the Tai Chi Chuan intervention. A cost-effectiveness analysis will be carried out alongside the evaluation of the clinical results. Follow-up measurements will be collected at 3, 6 and 12 months after randomization.</p></sec><sec><title>Discussion</title><p>As far as we know this is the first trial in Europe considering Tai Chi Chuan and fall prevention. This project will answer a pragmatic research question regarding the efficacy of Tai Chi Chuan regarding fall reduction.</p></sec>
|
<contrib id="A1" corresp="yes" contrib-type="author"><name><surname>Zeeuwe</surname><given-names>Petra EM</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Verhagen</surname><given-names>Arianne P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Bierma-Zeinstra</surname><given-names>Sita MA</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>van Rossum</surname><given-names>Erik</given-names></name><xref ref-type="aff" rid="I2">2</xref><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" contrib-type="author"><name><surname>Faber</surname><given-names>Marjan J</given-names></name><xref ref-type="aff" rid="I4">4</xref><email>[email protected]</email></contrib><contrib id="A6" contrib-type="author"><name><surname>Koes</surname><given-names>Bart W</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Geriatrics
|
<sec><title>Background</title><p>Thirty to fifty percent of the elderly of 65 years and older fall each year [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. About twenty percent of them need medical care after a fall and about six percent of the accidents result in a fracture. Moreover falls can result in disabilities, increased fear of falling, social isolation, decreased mobility and even an increased mortality [<xref ref-type="bibr" rid="B3">3</xref>,<xref ref-type="bibr" rid="B4">4</xref>]. The risk to fall is strongly related to previous falls, disturbed balance, dizziness, decreased muscular strength, use of benzodiazepines and diuretics and changes in walking pattern [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Based on these prognostic factors it is feasible to identify elderly who are at risk for falling.</p><p>To prevent fall incidents in the elderly exercise programs may be useful. In the most recent Cochrane review [<xref ref-type="bibr" rid="B2">2</xref>] on the effectiveness of interventions to prevent fall incidents and in the meta-analysis of the FICSIT (Frailty and Injuries: Cooperative Studies of Intervention Techniques) trials [<xref ref-type="bibr" rid="B7">7</xref>] the effectiveness of different exercise programs in healthy elderly populations were examined. In both the review and the meta-analysis exercises seem to have a positive effect to prevent fall incidents, although the evidence is limited. In addition, reviews of Verhagen et al. [<xref ref-type="bibr" rid="B8">8</xref>] and Li et al. [<xref ref-type="bibr" rid="B9">9</xref>] show that the evidence is still weak for the effect of Tai Chi Chuan in preventing falls, improving balance and improving cardiovascular function.</p><p>Tai Chi Chuan is a traditional Chinese exercise that has been practiced for many centuries. It is an integral part of the Traditional Chinese Medicine. It consists of a series of movements (positions) that are performed in a slow and flowing manner. The Tai Chi positions operate on three basic principles. First, the body should be extended and relaxed. Second, the mind must be alert and calm. Last, all body movements require well-coordinated sequencing of segments. The accent is on relaxation and balance control. In the literature many beneficial effects of Tai Chi Chuan are reported, such as improved balance, decreased fall incidents, decreased blood pressure, and increased self-efficacy [<xref ref-type="bibr" rid="B9">9</xref>].</p><p>Although promising, it is not entirely clear yet whether Tai Chi Chuan can indeed prevent fall incidents in the elderly. Only in one trial it is found to reduce the risk of multiple falls by 47,5% [<xref ref-type="bibr" rid="B10">10</xref>]. Since the evidence is still controversial we decided to conduct this study evaluating the effect of Tai Chi Chuan on fall prevention and physical and psychological function in older adults in the Dutch situation.</p></sec><sec><title>Methods/Design</title><sec><title>Study design</title><p>This study is a randomized, partially blinded, clinical trial to assess the effectiveness of 'usual care' and Tai Chi Chuan compared to 'usual care' only on fall prevention in the elderly living at home with an increased risk for falling. 'Usual care' means that people can use or apply for all available services in the area. Figure <xref ref-type="fig" rid="F1">1</xref> presents the flow diagram of the study design. The Medical Ethics Review Committee at Erasmus MC, University Medical Center Rotterdam, approves the study.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Study design.</p></caption><graphic xlink:href="1471-2318-6-6-1"/></fig></sec><sec><title>Study population</title><p>General practitioners (GPs) will be informed by mail in Dordrecht and Zwijndrecht (two towns in the west of the Netherlands) about the background and the goal of our research. Using a response fax they can indicate if they are willing to participate or not. If we do not receive a response within a week or after sending a reminder, we will phone them to get clear if they are willing to participate. An appointment will be made with the participating GPs in order to identify eligible people from their patient databases. Eligible are all people of seventy years and older living at home who are at risk for falling. We will select on the basis of in- and exclusion criteria as shown in table <xref ref-type="table" rid="T1">1</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Inclusion and exclusion criteria</p></caption><table frame="hsides" rules="groups"><tbody><tr><td align="left"><bold><italic>Inclusion criteria</italic></bold></td></tr><tr><td align="left">- 70 years and older</td></tr><tr><td align="left">- living at home</td></tr><tr><td align="left">- at least one self-reported fall accident in the preceding year or at least two of the following self-reported risk factors for falling: disturbed balance, mobility problems, dizziness or the use of benzodiazepines or diuretics</td></tr><tr><td align="left"><bold><italic>Exclusion criteria</italic></bold></td></tr><tr><td align="left">- progressive disorders:</td></tr><tr><td align="left">• neurological disorders (e.g. Parkinson's disease, Alzheimer's disease, multiple sclerosis, epilepsy, grand stroke with rest symptoms, polyneuropathy)</td></tr><tr><td align="left">• metastatic/terminal cancer or undergoing a treatment for cancer</td></tr><tr><td align="left">- unable to undergo the intervention (e.g. confined to bed, restricted to a wheelchair, unable to walk for eight meters without help, unable to reach the location). These items are stated by the GP after inclusion criteria are checked by phone.</td></tr><tr><td align="left">- unable to fill in a Dutch questionnaire</td></tr><tr><td align="left">- not available for measurements or intervention or knowing in advance to miss eighty percent of the lessons</td></tr><tr><td align="left">- received Tai Chi Chuan in the preceding year</td></tr></tbody></table></table-wrap><p>Potential eligible people will receive an information package with detailed information on the study from their GP. Using the included response form they can sign up, ask for additional information or withdraw. In case of withdrawal, people will be kindly requested to write down the reason. In case of willingness to participate they will be asked to sign an informed consent form. A postage free envelope will be included. Reminders will be sent after four weeks.</p></sec><sec><title>Telephone questionnaire</title><p>After returning the response form, potential participants will be interviewed by phone. This serves as the screening instrument for final eligibility (see table <xref ref-type="table" rid="T1">1</xref>) and partly as baseline measurements. Participants are again informed on the randomization procedures, allowing them another opportunity to reconsider participation before study entry. Potential participants will be asked which study group they prefer to be enrolled in, to evaluate whether patient preferences might influence outcome. Finally, participants will be invited to a location near their homes to conduct the remainder of the baseline measurements.</p></sec><sec><title>Baseline measurements</title><p>The baseline measurements consist of the following items. Sociodemographic factors (age, gender, ethnicity, marital status, education, composition of household), physical characteristics (e.g. height, weight, acuity of vision), environmental factors (e.g. pets, adjustments to the house), use of walking devices, medication, and use of health care services (e.g. GP, specialist, physiotherapy, home care/district nurse).</p><p>At baseline the following secondary outcome measures are measured: Balance (Berg Balance Scale), fear of falling (The Falls Efficacy Scale), physical parameters (blood pressure and heart rate at rest, FEV1 (i.e. forced expiratory volume during the first second) and PEF (i.e. peak expiratory flow)), physical activity (Physical Activity for the Elderly), functional status (Groningen Activity Restriction Scale), quality of life (EuroQol), and mental health (SF-36). For further details see 'Secondary outcome measures'.</p></sec><sec><title>Sample size and power</title><p>The effect of Tai Chi in decreasing fall incidents is conservatively estimated on 25 percent [<xref ref-type="bibr" rid="B10">10</xref>]. A power calculation (alpha = 0,05; beta = 0,1; drop out percentage of twenty percent) estimates that 135 participants are needed in each study group. This implies that we intend to enroll 270 participants in total.</p></sec><sec><title>Randomization</title><p>After baseline measurements participants will be assigned to one of the four strata based on gender (male/female) and the prognostic factor fall accidents in the preceding year (yes/no). An independent research assistant will perform randomization into either the control or the intervention group using a randomization list generated by a computer. To avoid unequal group sizes a block randomization will be used using block sizes of four. When two persons live together (e.g. married couples) they will be allocated to the same group, i.e. the group to which the first person contacted is allocated to, in order to avoid contamination.</p></sec><sec><title>Masking</title><p>GPs will be masked because they are not informed which group the patient is allocated to. Tai Chi Chuan teachers and the participants cannot be masked. Both research assistants who perform the measurements and the researcher who performs the data analyses will be masked.</p></sec><sec><title>Control group</title><p>Participants in the control group receive 'usual care' without an additional intervention. 'Usual care' means that they, as before, can use or apply for all available services in the area. Participants in both the control and the intervention group will receive a brochure containing general information on how to prevent falls in and around the house.</p></sec><sec><title>Intervention group</title><p>Participants in the intervention group will receive in addition to the 'usual care' Tai Chi Chuan lessons.</p><p>The Tai Chi Chuan form is based on the one used by Wolf et al. [<xref ref-type="bibr" rid="B11">11</xref>]. This form turned out to be appropriate for the elderly because all components of movement that typically become limited with aging are emphasized. The ten positions are easily comprehensible. It is derived from the Yang style and consists of a form with ten positions: 1. Opening; 2. Grasping the Sparrow's Tail, Left; 3. Grasping the Sparrow's Tail, Right; 4. Cloud Hands; 5. Repulse Monkey; 6. Part Wild Horse's Mane; 7. Brush Knee Twist Step; 8. Lift Kick Left; 9. Lift Kick Right; 10. Closing. The ten positions are linked together in a continuous, smooth-flowing form.</p><p>Four Tai Chi Chuan-teachers are available to give the lessons. All are qualified to give the lessons and are experienced in working with elderly. Together they will develop and finalize the Tai Chi Chuan-protocol.</p><p>In the literature [<xref ref-type="bibr" rid="B8">8</xref>] the recommended frequency of lessons varies from once a day to once a week during a period ranging from 10 weeks to one year. We thought the most optimum and practicable frequency is twice a week for thirteen weeks. Each lesson will last one hour, starting with twenty minutes warming-up and ending with ten minutes cooling down. Hereby Chi Kung exercises will be used. Chi Kung means 'control or mastery of the breath'. Emphasis is placed on relaxation and on contemplation of the breath. In addition to the lessons the participants will be asked to practice the learned Tai Chi Chuan positions at home at least twice a week for about fifteen minutes.</p></sec><sec><title>Outcome measures</title><p>The primary outcome measure is new fall accidents. At the end of the baseline measurements participants will receive a fall calendar. They will be asked to fill in daily for a year if they are fallen, nearly fallen or not fallen. A fall is defined as 'an unintentionally coming to rest on the ground, floor, or other lower level'. A near-fall is defined as 'the person seems to fall, but can prevent the fall by catching or leaning on a person or a thing (e.g. a chair, a drawer or a table)' [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>When a fall or a near-fall is reported people will be asked to give more details (e.g. how it happened, at what time, if they were hurt, if they started to use devices, whether medical care was needed) on an additional standardized form. The forms can be returned in postage free envelopes at the end of each month. If forms are missing or incomplete the research assistant will contact the participant by phone and together they will complete the forms.</p></sec><sec><title>Secondary outcome measures</title><p>- Balance is measured with the Berg Balance Scale. It consists of fourteen items that are scored on a five-points ordinal scale [<xref ref-type="bibr" rid="B13">13</xref>-<xref ref-type="bibr" rid="B15">15</xref>].</p><p>- Fear of falling is measured with the Falls Efficacy Scale [<xref ref-type="bibr" rid="B16">16</xref>]. It is an oral questionnaire and consists of ten items scored on a four points scale, ranging from 'not concerned at all' to 'very concerned'. Participants are asked about how concerned they are that they might fall while performing a series of ten tasks of progressive difficulty in the home environment.</p><p>- Physical parameters: blood pressure and heart rate in rest, FEV1 (i.e. forced expiratory volume during the first second) and PEF (i.e. peak expiratory flow). The latter two are measured by a spirometer.</p><p>- Physical activities are measured with the Physical Activity Scale for the Elderly (PASE) [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>]. It combines the frequency of leisure, household and occupational activities over the previous seven days.</p><p>- Functional status is measured with the Groningen Activity Restriction Scale (GARS), covering activities of daily living (ADL) and instrumental activities of daily living (IADL) [<xref ref-type="bibr" rid="B19">19</xref>]. It consists of eighteen items scored on a four points scale.</p><p>- Quality of life is measured with the EuroQol EQ 5-D. It will be used in the cost-effectiveness analysis [<xref ref-type="bibr" rid="B20">20</xref>-<xref ref-type="bibr" rid="B22">22</xref>].</p><p>- Mental health is measured with the SF-36 (subscales) [<xref ref-type="bibr" rid="B23">23</xref>-<xref ref-type="bibr" rid="B26">26</xref>].</p><p>- Use of walking devices, medication, use of health care services (e.g. GP, specialist, physiotherapy, home care/district nurse), adjustments to the house.</p></sec><sec><title>Follow-up</title><p>As shown in table <xref ref-type="table" rid="T2">2</xref> participants will be followed up during twelve months after baseline measurements. At the end of intervention period (i.e. at three month) and twelve months after the baseline measurement all measurements will be performed. At six months only the questionnaire will be send by mail.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Variables and timing measurements</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Variable</bold></td><td align="left" colspan="4"><bold>Time measured</bold></td></tr></thead><tbody><tr><td></td><td align="left">Baseline</td><td align="left">3 months</td><td align="left">6 months</td><td align="left">12 months</td></tr><tr><td colspan="5"><hr></hr></td></tr><tr><td align="left">Inclusion and exclusion variables</td><td align="left">TQ</td><td></td><td></td><td></td></tr><tr><td align="left">Falls and near-falls</td><td align="left" colspan="4">Continuously every day when fall occur</td></tr><tr><td align="left">Severity of fall incidents</td><td align="left" colspan="4">Continuously every day when fall occur</td></tr><tr><td align="left">Sociodemografic, physical and environmental factors</td><td align="left">WQ</td><td></td><td></td><td></td></tr><tr><td align="left">Balance (Berg Balance Scale)</td><td align="left">M</td><td align="left">M</td><td></td><td align="left">M</td></tr><tr><td align="left">Fear of falling (Falls Efficacy Scale)</td><td align="left">OQ</td><td align="left">OQ</td><td></td><td align="left">OQ</td></tr><tr><td align="left">Physical parameters (blood pressure, heart rate, lung function parameters)</td><td align="left">M</td><td align="left">M</td><td></td><td align="left">M</td></tr><tr><td align="left">Physical activity (Physical Activity Scale for the Elderly)</td><td align="left">WQ</td><td align="left">WQ</td><td align="left">PQ</td><td align="left">WQ</td></tr><tr><td align="left">Functional status (Groningen Activity Restriction Scale)</td><td align="left">WQ</td><td align="left">WQ</td><td align="left">PQ</td><td align="left">WQ</td></tr><tr><td align="left">Quality of life (EuroQol)</td><td align="left">WQ</td><td align="left">WQ</td><td align="left">PQ</td><td align="left">WQ</td></tr><tr><td align="left">Mental health (SF-36)</td><td align="left">WQ</td><td align="left">WQ</td><td align="left">PQ</td><td align="left">WQ</td></tr><tr><td align="left">Use of walking devices, medication, use of health care services, adjustments to the house</td><td align="left">WQ</td><td align="left">WQ</td><td align="left">PQ</td><td align="left">WQ</td></tr><tr><td align="left">Compliance to intervention</td><td></td><td align="left">X</td><td></td><td></td></tr></tbody></table><table-wrap-foot><p>TQ = telephone questionnaire; WQ = written questionnaire; OQ = oral questionnaire; PQ = written questionnaire send by post; D = diary, by self report, concurrent registration; M = measurement (e.g. by a test); X = reported by teacher</p></table-wrap-foot></table-wrap><p>Questionnaires will be checked on missing data and completed with the participants by phone.</p><p>In the invitation letter to visit the location for the follow-up measurement at three and twelve months, participants will be asked not to tell the research assistant whether they are allocated to the intervention or the control group. Moreover a reminder sign will be placed on the location. In this way the research assistant will remain masked.</p><p>Every effort will be made to keep the loss to follow-up as small as possible. When the questionnaire sent after six months is not received within two weeks, a phone call is made in order to remind the participant.</p></sec><sec><title>Process evaluation</title><p>At the end of the intervention period participants and teachers will be asked to fill out a questionnaire to evaluate the Tai Chi Chuan course. This includes the appreciation of the course, the expected benefit, et cetera. The teachers will report their experiences with the protocol, and fill out an attendance list per lesson. Reasons for noncompliance will be written down.</p><p>Dropouts will be phoned to get information on the reason of stopping with the course.</p><p>These results can be used to optimize the content and implementation of the courses in future. Moreover, it can provide information explaining measured effects or explaining why expected effects were not identified. Moreover a comparison between these self-reported values and the measured values (i.e. Berg Balance Scale) can be made.</p></sec><sec><title>Data entry</title><p>Data will be entered in SPSS using a protocol. There will be a check on a random sample of ten percent of the data. Thereafter an overall check will be done using frequency tables to detect unusual values, inconsistencies and missing values.</p></sec><sec><title>Statistical analyses</title><p>To assess if randomization was successful, descriptive statistics of baseline characteristics will be used. In case clinically relevant baseline differences regarding prognosis of future incidence of falls is present, effect estimates will be adjusted using multivariate analysis.</p><p>Dropouts are defined as participants who stop participating during the first three months (i.e. intervention period). Non-compliers are defined as participants who miss more than twenty percent of the lessons.</p><p>The analysis will be performed using the intention-to-treat principle, e.g. all patients are included as randomized irrespective of compliance. Missing values are assigned the last available valid score.</p><p>We will evaluate the effect of the intervention in reducing the number of fall incidents using the Student's t-test and survival techniques. To analyze the secondary outcome measures the Student's t-test, the Mann-Whitney test or the Chi-square test will be used according to the type of variable.</p><p>When drop out is higher than fifteen percent and/or the compliance is lower than seventy percent an additional per-protocol analysis will be carried out. These are restricted to those participants who complied fully with the intervention protocol and outcome measures.</p><p>At the end of the analyses the allocation of participants will be disclosed.</p><p>A cost-effectiveness analysis will be conducted alongside the analysis of the clinical data.</p><p>Direct costs (such as medication, additional therapy, visits to health care providers) as well as indirect costs (such as disability) measured by the EuroQol, will be related to the primary outcome measure (fall incidents).</p></sec><sec><title>Time plan</title><p>The recruitment of the GPs, followed by the search in the patient databases started in September 2003. Thereafter the participants were invited, screened and included. The first intervention group started in February 2004. Follow up will continue during 2006.</p></sec></sec><sec><title>Discussion</title><p>As far as we know this is the first trial in Europe considering Tai Chi Chuan and fall prevention. In the United States some similar trials are done, but circumstances differ from Europe since elderly often live in communities. We have to recruit participants via their GP.</p><p>Most trials on the effect of Tai Chi Chuan are focused on intermediate outcome measures such as balance, muscular strength, whereas we measure fall incidents.</p><p>In this study we use the highest methodological standards. Masking is performed when possible. Moreover randomization will be checked to make clear if baseline characteristics in both groups are equal. If needed adjustments will be made.</p><p>The intervention and control group not only differ regarding the intervention, but also in the amount of attention the individuals receive. As social relations integral to the exercise environment are determinants of subjective well-being in older adults [<xref ref-type="bibr" rid="B27">27</xref>] we considered to make a third group of elderly, receiving another group-based activity to assess the positive effect of social function of the group. But as this project was restricted in time, the funding organization and we decided to work with two groups. This means that this project will answer a pragmatic research question regarding the efficacy of Tai Chi Chuan regarding fall reduction.</p><p>To keep the drop out as low as possible the Tai Chi Chuan-lessons are given at a location close to where the elderly live, the travel costs of the participants will be refunded and as needed we organize transportation.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>APV and BWK developed the design of the randomized clinical trial and are co-author. PZ completed the study design and will perform the research. All authors have read and approved the final manuscript. All results out of this trial will be published using the International Standardised Randomised Controlled Trial Number (ISRCTN).</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2318/6/6/prepub"/></p></sec>
|
Insulin versus oral agents in the management of Cystic Fibrosis Related Diabetes: a case based study
|
<sec><title>Background</title><p>Insulin is the recommend therapeutic agent of choice for the management of Cystic Fibrosis Related Diabetes (CFRD), despite only sub-optimal reductions in glycemic control and increased morbidity and mortality reported by centers using this agent. The newer insulin sensitizing agents demonstrated to have anti-inflammatory mechanisms may provide an alternative management option for CFRD.</p></sec><sec sec-type="methods"><title>Methods</title><p>A prospective case based therapeutic comparison between insulin, sulfonylurea, metformin and thiazolidinedione was observed over one decade with 20 CFRD patients diagnosed using American Diabetes Association guideline standards. Patients entering the study elected treatment based on risk and benefit information provided for treatment options. Patients receiving organ transplant or requiring combination diabetic medications were excluded from the study.</p></sec><sec><title>Results</title><p>No statistical advantage was achieved regarding overall glycemic control for oral agents over insulin. Additional outcome measures including changes in weight, liver function testing and FEV<sub>1 </sub>were not statistically significant.</p></sec><sec><title>Conclusion</title><p>Insulin alone may not be the only therapeutic option in managing CFRD. Oral hypoglycemic agents were equally effective in treating CFRD and may provide an alternative class of agents for patients reluctant in using insulin.</p></sec>
|
<contrib id="A1" equal-contrib="yes" corresp="yes" contrib-type="author"><name><surname>Onady</surname><given-names>Gary M</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" equal-contrib="yes" contrib-type="author"><name><surname>Langdon</surname><given-names>Leora J</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib>
|
BMC Endocrine Disorders
|
<sec><title>Background</title><p>Diabetes mellitus has a 15–30% prevalence in cystic fibrosis (CF) patients [<xref ref-type="bibr" rid="B1">1</xref>]. Only 33% of diabetic CF patients are symptomatic for diabetes [<xref ref-type="bibr" rid="B2">2</xref>]. This subtle presentation of diabetes has resulted in the Cystic Fibrosis Foundation recommending screening guidelines for diabetes [<xref ref-type="bibr" rid="B3">3</xref>]. This guideline further recommends insulin as the therapeutic agent of choice in managing Cystic Fibrosis Related Diabetes (CFRD). Current recommendations for insulin are not always feasible for CFRD patients due to poor compliance and hypoglycemia. Diabetic diets utilized for type 1 and 2 diabetes conflict with the high fat and carbohydrate diet required to compensate for the malabsorption associated with the majority of CF patients. This carbohydrate load further stresses glucose tolerance in CF that is associated with both impaired insulin secretion and insulin resistance [<xref ref-type="bibr" rid="B4">4</xref>]. Increasing insulin resistance is correlated with progressive development of impaired glucose tolerance in CF [<xref ref-type="bibr" rid="B5">5</xref>,<xref ref-type="bibr" rid="B6">6</xref>]. Insulin deficiency progresses with increasing age, but is seldom absolute in CF and ketoacidosis is uncommon [<xref ref-type="bibr" rid="B7">7</xref>]. This disease process may therefore respond to medications used in the management of type 2 diabetes. Investigators testing this therapeutic potential reported the oral hypoglycemic agent, glipizide, effects on insulin secretion in CF diabetics [<xref ref-type="bibr" rid="B8">8</xref>]. Glipizide demonstrated a good response in early stages of disease, but overtly diabetic CF subjects had a poor response to glipizide alone. With the exception of this study, the medical literature has primarily focused on diabetic control with insulin therapy alone.</p><p>Inflammation may further impact insulin resistance in the CF patient. Oral insulin sensitizing agents with anti-inflammatory activity may provide a duel benefit in CFRD and perhaps serve as an alternative agent for managing diabetes. Metformin shares biguanide functional groups in common with protease inhibitors, and metformin has demonstrated protease inhibitor activity with exceptional potency at therapeutic concentrations [<xref ref-type="bibr" rid="B9">9</xref>]. The thiazolidinediones have additionally demonstrated a counter-regulatory protective leukotriene inflammatory effect [<xref ref-type="bibr" rid="B10">10</xref>]. These agents have not been previously evaluated for CFRD, secondary to concern of metabolic acidosis complicated by metformin and the potential of hepatic toxicity with thiazolidinediones [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>This study reports the first clinical outcome for insulin sensitizing agents with potential anti-inflammatory activity in the management of CFRD. The oral agents reviewed above are compared to the insulin treatment group for each clinical outcome. Clinical outcome measures followed changes in glycosylated hemoglobin (HbA<sub>1c</sub>), weight, forced expiratory volulme at 1 second (FEV<sub>1</sub>), and alanine aminotransferase (ALT) for the duration of treatment for each group.</p></sec><sec sec-type="methods"><title>Methods</title><p>Cystic Fibrosis Related Diabetes was diagnosed in 24 patients during a 10 year span from January 1992 through December 2002 at the Dayton Adult Cystic Fibrosis Program. Three CFRD patients receiving lung transplant were excluded from this study secondary to immunosuppressive therapy that included prednisone. A patient on combination diabetic therapy during the study period was also excluded. A total of 20 CFRD patients over this 10 year period remained for this prospective study.</p><p>Patients elected treatment after informed consent was provided for all management options based on the benefit of oral agents approved for use in type 2 diabetes mellitus but with explicit concerns regarding risks of these agents in context of the chronic illness and morbidity associated with CF. The patients were further instructed that the Cystic Fibrosis Foundation recommended the use of insulin as the preferred method in treating CFRD. Both hospital and university institutional review boards approved of this study protocol. Eight patients initially chose insulin, 5 patients sulfonylureas, 4 patients metformin (a biguanide), and 3 patients thiazolidinediones.</p><p>The diagnosis of diabetes was established based on the most current American Diabetes Association guidelines during the decade span of the investigation based on a minimum of two threshold fasting blood glucose and/or random blood glucose values with symptoms. Weights were used over BMI in this adult population as height data remained constant. All weights were obtained standing, without shoes on a balance scale calibrated yearly.</p><p>Patients were initiated on all oral agents starting with the lowest dose recommended based on standard pharmaceutical recommendations. Patients were evaluated at three month intervals at which time HbA1c levels were obtained. All patients with HbA1c >7 had a dose increased to the next recommended incremental level and followed up 3 months later. Once an optimal HbA1c was achieved, follow-up was spaced to 6 month intervals with HbA1c levels drawn and adjusted around the HBA1c >7 parameter. Insulin doses used consisted of split dosing 70/30 NPH/R ranging from 0.3 – 0.8 Units/kg. The oral medication dosing range required to achieve optimal control expressed as a percentage of the highest recommend dose was: sulfonylureas, 0.125 – 0.87; thiazolidinediones, 0.125 to 1.00; and metformin, 0.39 – 1.00. Compliance was followed by timing of prescription renewal with minimal differences identified between treatment groups.</p><sec><title>Statistical analyses</title><p>All continuous data are presented as the mean ± standard deviation (SD). Comparisons between groups on baseline data (age, weight, FEV<sub>1</sub>, and HbA<sub>1c</sub>) and treatment duration were made with one-way analysis of variance (ANOVA), or Welch ANOVA if the variances within groups were not equal. The changes in clinical variables from the start to the end of the treatment period were determined, and then were converted to changes/year to adjust for differences in the duration of treatment. Changes/year was compared between groups with either one-way ANOVA or Welch ANOVA. P values less than 0.05 were considered statistically significant. Chi-squared testing for nominal mortality data from the GraphPad web based statistical program.</p></sec></sec><sec><title>Results</title><p>All patient clinical profiles at the start of the study year, 1992, or at the time of CFRD diagnosis are summarized in Table <xref ref-type="table" rid="T1">1</xref> and their clinical data during the study until death or end of 2002 is graphed in the Figure. Baseline clinical data for patients are grouped by the specific agent used to manage CFRD and is summarized in Table <xref ref-type="table" rid="T2">2</xref>. Data reported in Table <xref ref-type="table" rid="T3">3</xref> reflect the variation over the course of study from the baseline parameters identified in Table <xref ref-type="table" rid="T2">2</xref>.</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Clinical profile of patients followed at start of study</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left"><bold>Pt</bold></td><td align="left"><bold>Age (yrs)</bold></td><td align="left"><bold>Wt (kg)</bold></td><td align="left"><bold>FEV<sub>1 </sub>(%)</bold></td><td align="center"><bold>% HbA<sub>1c </sub>at Dx</bold></td><td align="left"><bold>Rx Course (yrs)</bold></td><td align="left"><bold>Clinical Data</bold></td></tr></thead><tbody><tr><td align="left">I<sub>1</sub></td><td align="left">28</td><td align="left">65.8</td><td align="right">56</td><td align="right">5.5</td><td align="right">4</td><td align="left">PI, Lupus<sup>N</sup>, Deceased</td></tr><tr><td align="left">I<sub>2</sub></td><td align="left">31</td><td align="left">65.3</td><td align="right">44</td><td align="right">14.3</td><td align="right">2</td><td align="left">PI, Deceased</td></tr><tr><td align="left">I<sub>3</sub></td><td align="left">13</td><td align="left">41.9</td><td align="right">57</td><td align="right">9.3</td><td align="right">4</td><td align="left">PS</td></tr><tr><td align="left">I<sub>4</sub></td><td align="left">33</td><td align="left">54.3</td><td align="right">18</td><td align="right">10.0</td><td align="right">2</td><td align="left">PI, Deceased, RI*</td></tr><tr><td align="left">I<sub>5</sub></td><td align="left">14</td><td align="left">49.0</td><td align="right">99</td><td align="right">10.9</td><td align="right">8</td><td align="left">PI</td></tr><tr><td align="left">I<sub>6</sub></td><td align="left">44</td><td align="left">81.5</td><td align="right">30</td><td align="right">9.2</td><td align="right">5</td><td align="left">PI, Cor P, HTN, BPH</td></tr><tr><td align="left">I<sub>7</sub></td><td align="left">24</td><td align="left">53.0</td><td align="right">57</td><td align="right">7.3</td><td align="right">3</td><td align="left">PI, AIHS, RI</td></tr><tr><td align="left">I<sub>8</sub></td><td align="left">21</td><td align="left">70.5</td><td align="right">93</td><td align="right">9.8</td><td align="right">10</td><td align="left">PI</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="left">S<sub>1</sub></td><td align="left">29</td><td align="left">75.5</td><td align="right">71</td><td align="right">7.1</td><td align="right">1</td><td align="left">PI</td></tr><tr><td align="left">S<sub>2</sub></td><td align="left">17</td><td align="left">33.6</td><td align="right">33</td><td align="right">8.1</td><td align="right">2</td><td align="left">PI, Deceased</td></tr><tr><td align="left">S<sub>3</sub></td><td align="left">25</td><td align="left">73.9</td><td align="right">48</td><td align="right">7.8</td><td align="right">4</td><td align="left">PI, Deceased</td></tr><tr><td align="left">S<sub>4</sub></td><td align="left">33</td><td align="left">97.5</td><td align="right">83</td><td align="right">6.3</td><td align="right">1</td><td align="left">PI, Bipolar, HTN</td></tr><tr><td align="left">S<sub>5</sub></td><td align="left">18</td><td align="left">49.7</td><td align="right">30</td><td align="right">6.6</td><td align="right">1</td><td align="left">PI, Deceased</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="left">M<sub>1</sub>I<sub>4</sub></td><td align="left">35</td><td align="left">51.5</td><td align="right">17</td><td align="right">9.7</td><td align="right">1</td><td align="left">PI</td></tr><tr><td align="left">M<sub>2</sub></td><td align="left">24</td><td align="left">71.1</td><td align="right">117</td><td align="right">5.4</td><td align="right">2</td><td align="left">PI</td></tr><tr><td align="left">M<sub>3</sub>I<sub>5</sub></td><td align="left">22</td><td align="left">68.5</td><td align="right">68</td><td align="right">9.6</td><td align="right">2</td><td align="left">PI</td></tr><tr><td align="left">M<sub>4</sub></td><td align="left">19</td><td align="left">70.0</td><td align="right">100</td><td align="right">6.1</td><td align="right">3</td><td align="left">PI</td></tr><tr><td align="left">M<sub>5</sub></td><td align="left">22</td><td align="left">67.7</td><td align="right">88</td><td align="right">13.3</td><td align="right">3</td><td align="left">PI, Hepatic Cirrhosis</td></tr><tr><td align="left">M<sub>6</sub></td><td align="left">15</td><td align="left">41.8</td><td align="right">85</td><td align="right">8.9</td><td align="right">5</td><td align="left">PI</td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td align="left">T<sub>1</sub></td><td align="left">29</td><td align="left">58.0</td><td align="right">47</td><td align="right">13.3</td><td align="right">2</td><td align="left">PI</td></tr><tr><td align="left">T<sub>2</sub></td><td align="left">20</td><td align="left">44.4</td><td align="right">23</td><td align="right">5.1</td><td align="right">1</td><td align="left">PI</td></tr><tr><td align="left">T<sub>3</sub>I<sub>6</sub></td><td align="left">49</td><td align="left">85.5</td><td align="right">29</td><td align="right">9.2</td><td align="right">2</td><td align="left">PI, Cor P, HTN, BPH</td></tr><tr><td align="left">T<sub>4</sub>I<sub>7</sub></td><td align="left">32</td><td align="left">55.7</td><td align="right">65</td><td align="right">7.3</td><td align="right">2</td><td align="left">PI, AIHS, RI</td></tr><tr><td align="left">T<sub>5</sub>S<sub>4</sub></td><td align="left">34</td><td align="left">96.5</td><td align="right">92</td><td align="right">6.3</td><td align="right">1</td><td align="left">PI, Bipolar, HTN</td></tr><tr><td align="left">T<sub>6</sub></td><td align="left">40</td><td align="left">54.0</td><td align="right">56</td><td align="right">9.6</td><td align="right">2</td><td align="left">PI, Lupus, RI ABPA</td></tr></tbody></table><table-wrap-foot><p>Pt, patient (subscripts identify patient number, e.g. M<sub>1</sub>I<sub>4 </sub>was the first patient started on metformin and the fourth patient started on insulin); Rx Course (Yrs), represents the number of years that a patient received consecutive therapy with a specific class of agent; Mon, months; I, insulin; S, sulfonylurea; M, metformin; T, thiazolidinediones, FEV<sub>1</sub>, forced expiratory volume at 1 second; HbA<sub>1c</sub>, percent glycosylated hemoglobin; PI, pancreatic insufficient; PS, pancreatic sufficient; Lupus<sup>N</sup>, lupus with nephritis; RI, renal insufficiency; Cor P, cor pulmonale; HTN, hypertension; BPH, benign prostatic hypertrophy; AIHS, autoimmune hypersplenism; ABPS, allergic bronchopulmonary aspergillosis. *Patient was switched back to insulin when diagnosed with renal insufficiency.</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Baseline clinical data</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>Variable Mean ± SD (Range)</bold></td><td align="center"><bold>Insulin (n = 8)</bold></td><td align="center"><bold>Sulfonylurea (n = 5)</bold></td><td align="center"><bold>Metformin (n = 6)</bold></td><td align="center"><bold>Thiazoli-dinediones (n = 6)</bold></td><td align="center"><bold>P Value</bold></td></tr></thead><tbody><tr><td align="center">Age (Yrs)</td><td align="center">26.0 ± 10.3 (13 – 44)</td><td align="center">24.4 ± 6.9 (17 – 33)</td><td align="center">22.8 ± 6.7 (15 – 35)</td><td align="center">34.0 ± 9.9 (20 – 49)</td><td align="center">0.166</td></tr><tr><td align="center">ALT (U/L)</td><td align="center">65.1 ± 24.1 (25 – 99)</td><td align="center">42.0 ± 26.5 (14 – 73)</td><td align="center">68.3 ± 28.1 (28 – 107</td><td align="center">67.4 ± 8.4 (55 – 76) (n = 5)</td><td align="center">0.250</td></tr><tr><td align="center">FEV<sub>1 </sub>(%)</td><td align="center">56.8 ± 28.0 (18 – 99)</td><td align="center">53.0 ± 23.3 (30 – 83)</td><td align="center">79.2 ± 34.5 (17 – 117)</td><td align="center">52.0 ± 25.2 (3 – 92)</td><td align="center">0.329</td></tr><tr><td align="center">Weight (kg)</td><td align="center">60.2 ± 12.9 (41.9 – 81.5)</td><td align="center">66.0 ± 24.8 (33.6 – 97.5)</td><td align="center">61.8 ± 12.2 (41.8 – 71.1)</td><td align="center">65.7 ± 20.4 (44.4 – 96.5)</td><td align="center">0.911</td></tr><tr><td align="center">HbA<sub>1c</sub></td><td align="center">9.5 ± 2.6 (5.5 – 14.3)</td><td align="center">7.2 ± 0.8 (6.3 – 8.1)</td><td align="center">8.8 ± 2.8 (5.4 – 13.3)</td><td align="center">8.0 ± 3.1 (5.1 – 13.3)</td><td align="center">0.408</td></tr><tr><td align="center">Treatment Duration (Yrs)</td><td align="center">4.8 ± 2.9 (2 – 10)</td><td align="center">1.8 ± 1.3 (1 – 4)</td><td align="center">2.7 ± 1.4 (1 – 5)</td><td align="center">1.8 ± 0.4 (1 – 2)</td><td align="center">0.093</td></tr></tbody></table><table-wrap-foot><p>SD, standard deviation; Yrs, years; ALT, alanine aminotransferase; FEV<sub>1</sub>, forced expiratory volume at 1 second; HbA<sub>1c</sub>, percent glycosylated hemoglobin. P values for Age, ALT, FEV<sub>1</sub>, Weight, and HbA<sub>1c </sub>are from one-way analysis of variance (ANOVA); the P value for Treatment Duration is from Welch ANOVA for unequal group variances.</p></table-wrap-foot></table-wrap><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Clinical data variance over time</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="center"><bold>Variable Mean ± SD (Range)</bold></td><td align="center"><bold>Insulin (n = 8)</bold></td><td align="center"><bold>Sulfonylurea (n = 5)</bold></td><td align="center"><bold>Metformin (n = 6)</bold></td><td align="center"><bold>Thiazoli-dinediones (n = 6)</bold></td><td align="center"><bold>P Value</bold></td></tr></thead><tbody><tr><td align="center">ALT (U/L) Change/Yr</td><td align="center">1.8 ± 3.0 (-2.0 – 7.8)</td><td align="center">4.7 ± 8.6 (-2.5 – 19.0)</td><td align="center">-2.8 ± 22.0 (-41.5 – 26.5)</td><td align="center">-1.1 ± 7.7 (-11.0 – 9.5) (n = 5)</td><td align="center">0.757</td></tr><tr><td align="center">FEV<sub>1 </sub>(%) Change/Yr</td><td align="center">-0.3 ± 3.3 (-3.9 – 6.3)</td><td align="center">1.4 ± 9.6 (-9.0 – 13.0)</td><td align="center">-1.2 ± 3.2 (-7.0 – 1.5)</td><td align="center">5.2 ± 7.8 (-2.5 – 20.0)</td><td align="center">0.422</td></tr><tr><td align="center">Weight (kg) Change/Yr</td><td align="center">1.5 ± 2.2 (-1.2 – 6.3)</td><td align="center">0.5 ± 2.2 (-2.4 – 2.7)</td><td align="center">1.8 ± 3.9 (-3.3 – 8.5)</td><td align="center">4.3 ± 6.0 (0.3 – 16.0)</td><td align="center">0.384</td></tr><tr><td align="center">HbA<sub>1c </sub>Change/Yr</td><td align="center">-0.3 ± 0.2 (-0.5 – 0.1)</td><td align="center">-0.8 ± 0.7 (-1.8 – -0.1)</td><td align="center">-1.1 ± 1.4 (-3.2 – 0.6)</td><td align="center">-0.6 ± 1.3 (-3.2 – 0.5)</td><td align="center">0.283</td></tr></tbody></table><table-wrap-foot><p>SD, standard deviation; Yr, year; ALT, alanine aminotransferase; FEV<sub>1</sub>, forced expiratory volume at 1 second; HbA<sub>1c</sub>, percent glycosylated hemoglobin. P values for ALT and Weight are from one-way analysis of variance (ANOVA); P values for FEV<sub>1 </sub>and HbA<sub>1c </sub>are from Welch ANOVA for unequal group variances.</p></table-wrap-foot></table-wrap><p>Seven individuals (35%) had HbA<sub>1c </sub>levels < 7% at the time of diagnosis. All patients tolerated initial therapy well, and no patient changed medical management due to complications reported for these agents. Four patients chose to discontinue insulin therapy for an oral agent secondary to inadequate glycosylated hemoglobin control of 7.0% while using insulin. Patient M<sub>1</sub>I<sub>4 </sub>weaned off 60 units of insulin per day with the best HbA<sub>1c </sub>control at 9.7% to metformin therapy achieving an averaged HbA<sub>1c </sub>of 6.5%. This individual was the only patient in the metformin group with an FEV<sub>1 </sub>< 60% (refer to Figure), and was insistent on using this agent despite strong advise to select another treatment option. Patient T<sub>5</sub>S<sub>4 </sub>was well controlled on a sulfonylurea, but requested to switch to a thiazolidinedione.</p><p>Mortality rates from our study were highest among patients in the sulfonylurea group (60%); followed by the insulin group (37%); with no deaths observed from the biguanide and thiazolidinedione treatment groups. Death rates between treatment groups were not statistically significant; P = 0.062. Sixty patients had been followed in this adult center during this ten year period with a mortality rate of 23% observed in non-diabetics and 38% in diabetics, of which only a third were female, for a 1.7 increased diabetic mortality risk which is within the range reported by other centers [<xref ref-type="bibr" rid="B12">12</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. To date, only one patient, T<sub>6</sub>, has been identified with diabetic complications. That patient came from the thiazolidinedione treatment group and followed for systemic lupus. A nephropathy developed 18 months after the diagnosis of CFRD was made, with renal biopsy indicating diabetic nephropathy. There have been no further reports of abnormal urine microalbumin measures or retinal examinations indicating microvascular disease in our patients.</p></sec><sec><title>Discussion</title><p>Results demonstrate no statistical advantage of one treatment option over another in achieving overall glycemic control. The number of patients in the various treatment groups was most likely too small to achieve statistical significance. More aggressive overall glycemic control for CFRD may be necessary based on recent reports that HbA<sub>1c </sub>underestimates a true glycemic index in CF patients [<xref ref-type="bibr" rid="B14">14</xref>]. A reduced life span of red blood cells has recently been reported through personal communication in cystic fibrosis patients by researches in Houston, Texas, which may reflect this underestimate of glycemic control. Studies from the CF literature further demonstrate insulin achieves only sub-optimal glycemic control based on HbA<sub>1c </sub>outcomes, which is a significant concern when placed in context with the data demonstrating that HbA<sub>1c </sub>underestimates glycemic control. A retrospective study out of Cleveland, Ohio, evaluated 22 patients on a flexible meal-planning system targeting insulin boluses titrated to each meal to control postprandial blood glucose excursions and report glycosylated hemoglobin reductions from 11.3 to 8.1% [<xref ref-type="bibr" rid="B15">15</xref>]. A prospective study from Paris, France, followed 14 patients early in the diabetic course on insulin therapy finding glycosylated hemoglobin values ranging from 6.6 to 7.8% [<xref ref-type="bibr" rid="B16">16</xref>]. A study conducted in Houston, Texas compared subcutaneous insulin injections versus pump infusion demonstrated improved glycemic control on the insulin pump in lowering HbA<sub>1c </sub>values from 8.6 to 7.3% [<xref ref-type="bibr" rid="B17">17</xref>].</p><p>Morbidity and mortality data reported by centers using insulin as standard management protocol provide concern regarding optimizing clinical outcomes for CFRD. Microvascular disease is reported in 23% of diabetic CF patients [<xref ref-type="bibr" rid="B18">18</xref>] compared to 18% [<xref ref-type="bibr" rid="B19">19</xref>] in the non-CF diabetic population. The median age of survival for CFRD patients is reported at 35.6 years compared to 47 years for non-diabetics with CFRD females having a 7-fold mortality rate [<xref ref-type="bibr" rid="B20">20</xref>]. Significant mortality relative risks of 1.7 [<xref ref-type="bibr" rid="B12">12</xref>] and 2.8 [<xref ref-type="bibr" rid="B13">13</xref>] have been reported in CFRD patients over non-diabetic CF patients at two other centers.</p><p>Our center diabetic outcomes observed no adverse side effects from oral agents during our decade experience of treating CFRD. Caution must still be provided regarding the potential side effects of using the newer oral agents in managing CF diabetics, with the biguanide class presenting the most concerning risk of metabolic acidosis in the CF patient, and therefore should not be offered to the 30% of CF patients that have well documented liver disease by the time they reach adulthood. This study has furthermore identified four patients (20%) treated with renal insufficiency, in which only the one case mentioned above was attributed to diabetic nephropathy. Renal insufficiency is an absolute contraindication to the use of metformin and therefore, renal function must be followed closely in all CF patients as they approach their fourth and fifth decade of life. Likewise, this class of agent should be utilized only in patients with good to moderate pulmonary function. A recent study reported the safety of metformin in 91 randomized patients by intention-to-treat analysis with chronic obstructive pulmonary disease (COPD). Lactic acid values did not differ in the groups on or off metformin and correlated only with serum creatinine and body mass index. Mortality data were identical in the two groups and they conclude that there is no apparent reason that COPD patients should discontinue metformin [<xref ref-type="bibr" rid="B21">21</xref>]. Additionally in a systematic review of 194 randomized control studies of metformin with other diabetic agents, no fatal or nonfatal lactic acidosis events were reported, with Poisson statistics at 95% confidence intervals estimating lactic acidosis incidence at 8.1 and 9.9 cases per 100,000 [<xref ref-type="bibr" rid="B22">22</xref>].</p><p>A remarkable outcome from this center experience is the observed 1.8 kg weight gain for patients taking metformin. Weight loss, is typically reported for the type 2 diabetic, and this outcome would not be tolerated in most cystic fibrosis patients. The largest weight gain observed by a patient in this group was a very respectable 27%. One patient did lose 3% and another lost 7% of weight from this treatment group which explains the large range in the standard deviation of weight for this class. Neither of the patients losing weight demonstrated a loss of pulmonary function, with the patient losing 7% weight actually demonstrating an FEV<sub>1 </sub>gain of 4% per year. The lower pulmonary co-morbidity in the metformin group over other groups may have influenced the higher weight gain observed in this group. While this reduces the matching between groups, statistical significance was not observed over the insulin treatment group. Reasons behind a positive weight gain on metformin may relate to the cystic fibrosis patient having a higher rate of protein breakdown than non-CF patients [<xref ref-type="bibr" rid="B23">23</xref>]. Suppression of proteolysis by protease inhibitor activity of metformin [<xref ref-type="bibr" rid="B9">9</xref>] may both optimize overall nitrogen balance as well as inhibit insulin degradation. There have not been any deaths from our diabetic population maintained on metformin during our decade long experience in treating CFRD. Liver enzymes were least affected in this group as well as in the thiazolidinedione treatment group which also demonstrated a net decrease in liver enzyme changes during the course of therapy, and both were within the range of change observed with the other agents used in managing diabetes.</p><p>This study is limited by the small number of patients treated in a non-randomized, un-blinded fashion under each therapeutic grouping. Additional bias is potentially introduced by including baseline data of patients entering the study diagnosed with CFRD prior to prospective data gathering that was initiated in 1992, at which point all patient data was prospectively reviewed in this cohort study. Pulmonary treatment further evolved throughout the duration of this review in relation to standards of airway clearance utilized and aerosolized maintenance therapies that had become available to these patients.</p></sec><sec><title>Conclusion</title><p>Our CF center has adopted a rational approach to managing CFRD based on this observational experience.</p><p>1) Insulin appears to provide adequate response to patients with significant lung disease (FEV<sub>1 </sub>< 60%), and is likely the agent of choice to initiate management under any inpatient setting.</p><p>2) After initial response to insulin, consideration should be given to add a sulfonylurea or thiazolidinedione for patients with stable liver functions, in attempt to wean off insulin.</p><p>3) In the patient with significant diabetic onset with HbA<sub>1c </sub>> 7 and relatively preserved pulmonary function (FEV<sub>1 </sub>> 60%) with no documented liver disease, metformin can be used for initial management on an outpatient basis. Close monitoring of metabolic profile, renal, and hepatic function at quarterly intervals would be optimal.</p><p>4) The thiazolidinedione class is an alternative agent for outpatient management, particularly for the patient with FEV<sub>1 </sub>close to or approaching 60% predicted. In this situation liver function testing should be monitored quarterly at onset of therapy.</p><p>Our findings suggest that insulin may not be the most beneficial therapeutic agent for the management of CFRD. Oral agents that include the insulin sensitizing agents appear to be safe and as effective as insulin. Larger randomized control trials between insulin and the biguanide or thiazolidinedione class of agents with potential anti-inflammatory activity should be considered as alternative therapy to insulin, especially in light of consistently poor clinical outcomes reported for insulin therapy in the CF patient population.</p></sec><sec><title>Competing interests</title><p>The author(s) declare that they have no competing interests.</p></sec><sec><title>Authors' contributions</title><p>Gary M. Onady was responsible for study design, calculation of statistical outcomes and drafting of the manuscript. Leora J. Langdon was responsible for collection of the majority of patient data and to some of the informational content presented in the manuscript.</p><fig position="float" id="F1"><label>Figure 1</label><caption><p>Clinical response in individual patients.</p></caption><graphic xlink:href="1472-6823-6-4-1"/></fig></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1472-6823/6/4/prepub"/></p></sec>
|
Family doctors' knowledge and self-reported care of type 2 diabetes patients in comparison to the clinical practice guideline: cross-sectional study
|
<sec><title>Background</title><p>It is widely believed that providing doctors with guidelines will lead to more effective clinical practice and better patient care. However, different studies have shown contradictory results in quality improvement as a result of guideline implementation. The aim of this study was to compare family doctors' knowledge and self-reported care of type 2 diabetes patients with recommendation standards of the clinical practice guideline.</p></sec><sec sec-type="methods"><title>Methods</title><p>In April 2003 a survey was conducted among family doctors in Estonia. The structured questionnaire focused on the knowledge and self-reported behavior of doctors regarding the guideline of type 2 diabetes. The demographic and professional data of the respondents was also provided.</p></sec><sec><title>Results</title><p>Of the 354 questionnaires distributed, 163 were returned for a response rate of 46%. Seventy-six percent of the responded doctors stated that they had a copy of the guideline available while 24% reported that they did not. Eighty-three percent of the doctors considered it applicable and 79% reported using it in daily practice. The doctors tended to start treatment with medications and were satisfied with treatment outcomes at higher fasting blood glucose levels than the levels recommended in the guideline. Doctors' self-reported performance of the tests and examinations named in the guideline, which should be performed within a certain time limit, varied from overuse to underuse. Blood pressure, serum creatinine, eye examination and checking patients' ability to manage their diabetes were the best-followed items while glycosylated hemoglobin and weight reduction were the most poorly followed. Doctors' behavior was not related to the fact of whether they had the guideline available, whether they considered it applicable, or whether they actually used it.</p></sec><sec><title>Conclusion</title><p>Doctors' knowledge and self-reported behavior in patient follow-up of type 2 diabetes is very variable and is not related to the reported availability or usage of the guideline. Practice guidelines may be a useful source of information but they should not be overestimated.</p></sec>
|
<contrib id="A1" equal-contrib="yes" corresp="yes" contrib-type="author"><name><surname>Rätsep</surname><given-names>Anneli</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" equal-contrib="yes" contrib-type="author"><name><surname>Kalda</surname><given-names>Ruth</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" equal-contrib="yes" contrib-type="author"><name><surname>Oja</surname><given-names>Ivika</given-names></name><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A4" equal-contrib="yes" contrib-type="author"><name><surname>Lember</surname><given-names>Margus</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib>
|
BMC Family Practice
|
<sec><title>Background</title><p>Clinical practice guidelines (CPG) are systematically developed statements to assist the decisions of the practitioner and patient about appropriate healthcare for specific clinical circumstances [<xref ref-type="bibr" rid="B1">1</xref>]. It is expected that clinical practice guidelines improve healthcare quality, reduce inappropriate variations between providers and predispose dissemination of the evidence-based medicine concept in daily practice. Policymakers and payers see guidelines as a tool for making healthcare more consistent and efficient. However, there is no certain evidence that guidelines may change practice behavior. Different studies have shown contradictory results in quality improvement as a result of guideline implementation. Some studies describe improvement in disease management after guideline dissemination [<xref ref-type="bibr" rid="B2">2</xref>-<xref ref-type="bibr" rid="B4">4</xref>] but evidence to the contrary exists that none of the guideline intervention strategies led to improvements in patient quality of life, quality of diabetes care or performance activity and adherence to the guideline [<xref ref-type="bibr" rid="B5">5</xref>-<xref ref-type="bibr" rid="B7">7</xref>]. The potential barriers to physicians' adherence to guidelines can be divided into three themes: physician knowledge, attitudes and behavior. Meanwhile the sequence of the "knowledge-attitude-behavior" model is important in modifying physicians' practice patterns [<xref ref-type="bibr" rid="B8">8</xref>].</p><p>In recent decades care of patients with diabetes type 2 (DM2) has shifted from specialist care to primary care [<xref ref-type="bibr" rid="B9">9</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. The same trend has taken place in Estonia where in the 1990s previous highly specialized primary medical care was changed into a primary care-oriented and family doctor-based system [<xref ref-type="bibr" rid="B11">11</xref>]. To improve the quality of care, the Estonian Society of Family Doctors started to develop national practice guidelines in collaboration with specialist societies in 1994. The type 2 diabetes guidelines for family doctors (FD) are some of the latest, developed in 2000 by a multidisciplinary team, led by FDs and based on the International Diabetes Federation Europe DM2 guideline [<xref ref-type="bibr" rid="B12">12</xref>]. The guidelines was introduced to FDs at an educational seminar and disseminated by mail for all FD Society members.</p><p>Estonian family doctors' awareness of practice guidelines has not been assessed before. Hence our aim was to compare FDs' knowledge and self-reported care of type 2 diabetes patients with the recommendations of the clinical practice guideline.</p></sec><sec sec-type="methods"><title>Methods</title><p>A questionnaire-based survey was conducted in 2003. Every second doctor (n = 354) from the list of the Estonian Society of Family Doctors received a questionnaire by mail. A second mailing with a reminder letter and an additional questionnaire were sent to those who had not responded three weeks after the initial mailing. The questionnaire had been compiled by a research team and had been piloted before using it in the study.</p><p>The questionnaire covered the following items:</p><sec><title>Background characteristics</title><p>Independent variables included the year of graduation from medical school, the year of specialization as a family doctor, practice type and location, practice size and the number of diabetes patients.</p></sec><sec><title>Availability of the guideline</title><p>To the questions about the DM2 guideline availability, its use in daily practice and its estimated applicability, yes/no responses were required.</p></sec><sec><title>Specialist accessibility</title><p>The doctors were asked about the possibility to consult an endocrinologist, the distance to the nearest endocrinologist and the opportunity to consult an endocrinologist by telephone.</p></sec><sec><title>Following the guideline</title><p>The doctors were asked to report the level of blood glucose at which they usually start treatment with medications if lifestyle changes have been ineffective, and the level at which they are content with treatment outcome. In the DM2 guideline, HbAc1 is suggested for assessment of glucose control and corresponding target levels of capillary plasma glucose levels are provided. In the current study the doctors were asked to provide the respective fasting capillary plasma glucose levels, as this analysis was more widely used by the doctors at the time of study performance. The suggested level for starting treatment with medication is >6.5% for HbAc1 and >6.1 mmol/L (109.8 mg/dl) for capillary plasma glucose.</p><p>In the next section the family doctors were asked about the frequency at which they perform the following tests and examinations prescribed in the guideline: checking symptoms/complications, checking the patients' ability to manage their diabetes, smoking habit, blood pressure, weight/BMI, foot exam, eye exam, HbAc1, lipids (LDL, HDL, and TG), urinary protein, urinary albumin, and serum creatinine. The response options were "once a month," "at least once every three months," "at least annually," "rarely," and "I do not consider it necessary."</p><p>In the Estonian type 2 diabetes guideline, checking the patients' ability to manage their diabetes and the blood pressure measurement is suggested to be performed at every visit. In case of the latter, the doctors' responses "every visit" and "at least once every three months"in the questionnaire were deemed appropriate according to the guideline. HbAc1 is recommended to be checked every three months and other items annually.</p><p>Score for adherence to the guideline was calculated for each physician depending on how many guideline recommendations of 12 test and examinations have been timely performed according to their self-assessment report. More frequent performance was assessed as non-adherence as unneeded use of resources.</p></sec><sec><title>Statistical analysis</title><p>The obtained responses were entered in a database and were analysed by SPSS 10.0 (Statistical Package for the Social Sciences) for Windows. Statistical analysis included the chi-square test for the categorical variables and analysis of variance for the mean continuous variables. All calculated p-values were two-tailed. The P-values higher than 0.05 were considered insignificant.</p></sec></sec><sec><title>Results</title><p>Of the 354 doctors who received the questionnaire 46% (n = 163) responded. There were no significant differences between respondents in the first and second mailings. The background characteristics of the respondents in our sample correspond to that of the members of the Estonian Family Medicine Society, except for working area. Representation of doctors from cities was lower among respondents (19% vs. 37%) but no difference was found in any comparison between urban and rural doctors.</p><sec><title>Background characteristics and availability of the guideline</title><p>The mean size of the patient list was 1830 ± 407 and the average (± SD) working experience 22 ± 7.0 years. Regarding their previous specialty, the majority of the respondents had been district doctors for adults. Fifty-three percent of the doctors worked in solo practices and the rest worked in group practices (Table <xref ref-type="table" rid="T1">1</xref>).</p><table-wrap position="float" id="T1"><label>Table 1</label><caption><p>Background characteristics of the primary care providers according to the previous specialty, type of practice and working area</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Criterion</td><td align="left">n</td><td align="left">%</td></tr></thead><tbody><tr><td align="left">Status before specialization as a family physician</td><td></td><td></td></tr><tr><td align="left"> District doctor for adults</td><td align="left">109</td><td align="left">67</td></tr><tr><td align="left"> District pediatrician</td><td align="left">36</td><td align="left">22</td></tr><tr><td align="left"> Other specialties</td><td align="left">9</td><td align="left">6</td></tr><tr><td align="left"> Family physician through residency</td><td align="left">8</td><td align="left">5</td></tr><tr><td></td><td></td><td></td></tr><tr><td align="left">Type of practice</td><td></td><td></td></tr><tr><td align="left"> Solo</td><td align="left">85</td><td align="left">53</td></tr><tr><td align="left"> Group</td><td align="left">76</td><td align="left">47</td></tr><tr><td></td><td></td><td></td></tr><tr><td align="left">Working area</td><td></td><td></td></tr><tr><td align="left"> Urban</td><td align="left">67</td><td align="left">42</td></tr><tr><td align="left"> Rural</td><td align="left">63</td><td align="left">39</td></tr><tr><td align="left"> City</td><td align="left">30</td><td align="left">19</td></tr></tbody></table></table-wrap><p>The median number of diabetes patients in the list was 35.</p><p>Seventy-six percent of the respondents stated that the guideline was available while 24% reported that it was not. Eighty-three percent of the doctors considered it applicable and 79% reported using it in daily practice. The availability and use of the guideline were not related to working area, practice type and size, previous status before specialization as an FD, waiting time or distance to an endocrinologist.</p></sec><sec><title>Treatment decision and treatment goals</title><p>On average, the doctors tended to start treatment with medications at higher fasting blood glucose (FBG) levels than the levels recommended in the guideline (Table <xref ref-type="table" rid="T2">2</xref>). More than half of the doctors made a decision to start treatment with medications on FBG above 7 mmol/l, while a few made this decision at FBG values below 6.1 mmol/l (Figure <xref ref-type="fig" rid="F1">1</xref>). The decision about when to start treatment with medications was not related to whether the doctors had the guideline available, whether they considered it applicable, or whether they actually used it. There were no differences in the treatment behavior depending on the number of patients with diabetes in the doctors' list.</p><table-wrap position="float" id="T2"><label>Table 2</label><caption><p>Fasting blood glucose values (mmol/l) at which a decision to start treatment with medications was taken and the values at which doctors were content with treatment outcome</p></caption><table frame="hsides" rules="groups"><thead><tr><td></td><td align="left">Mean (SD)</td><td align="left">Minimum</td><td align="left">Maximum</td><td align="left">Standard in DM guideline</td></tr></thead><tbody><tr><td align="left">Decision to start treatment with medication</td><td align="left">7.2 (1.3)</td><td align="left">5.5</td><td align="left">15.0</td><td align="left">6.1</td></tr><tr><td align="left">Satisfaction with treatment outcome (glycemic control)</td><td align="left">6.8 (1.4)</td><td align="left">5.0</td><td align="left">14.0</td><td align="left">5.5</td></tr></tbody></table></table-wrap><fig position="float" id="F1"><label>Figure 1</label><caption><p>Distribution of doctors according to their decision to start treatment with medications depending on fasting blood glucose level.</p></caption><graphic xlink:href="1471-2296-7-36-1"/></fig></sec><sec><title>Following the guideline</title><p>The DM2 guideline includes 12 tests and examinations which should be performed during the year. According to the self-reported performance of the tests and examinations, it varied from overuse to underuse (Figures <xref ref-type="fig" rid="F2">2</xref>, <xref ref-type="fig" rid="F3">3</xref>).</p><fig position="float" id="F2"><label>Figure 2</label><caption><p><bold>Proportions of the doctors and their variation in performing the tests recommended in the guideline</bold>. Recommended frequency in the CPG: A – annually, B – every 3 months</p></caption><graphic xlink:href="1471-2296-7-36-2"/></fig><fig position="float" id="F3"><label>Figure 3</label><caption><p><bold>Proportions of doctors and their variations in performing the exams recommended in the guideline</bold>. Recommended frequency in the CPG: A – annually, B – every 3 months, C – every visit</p></caption><graphic xlink:href="1471-2296-7-36-3"/></fig><p>Blood pressure, serum creatinine, eye examination and checking patients' ability to manage their diabetes were the best followed tests and examinations while glycosylated hemoglobin and weight reduction were the most poorly followed. Checking symptoms and complications as well as checking urinary protein and albumin were performed more often than recommended in the guideline.</p><p>According to the score of adherence none of the family doctors performed all the required tests and examinations on time (Table <xref ref-type="table" rid="T3">3</xref>). At least half of the tests and examinations listed in the guideline were performed on time by 52% of the family doctors. Maximum score 12, was not reached by any of the doctors. The respondents' behavior in performing the tests and examinations did not depend on whether they had the guideline available or whether they had used it. There also was no difference in the doctors' behavior regarding the reported adherence to the guideline in terms of background characteristics and specialist accessibility.</p><table-wrap position="float" id="T3"><label>Table 3</label><caption><p>Distribution of the doctors according to the adherence to the DM2 guideline (adherence score)</p></caption><table frame="hsides" rules="groups"><thead><tr><td align="left">Adherence score</td><td align="left">Number of doctors</td><td align="left">Cumulative %</td></tr></thead><tbody><tr><td align="left">12</td><td align="left">0</td><td align="left">0</td></tr><tr><td align="left">11</td><td align="left">0</td><td align="left">0</td></tr><tr><td align="left">10</td><td align="left">1</td><td align="left">1</td></tr><tr><td align="left">9</td><td align="left">6</td><td align="left">5</td></tr><tr><td align="left">8</td><td align="left">19</td><td align="left">26</td></tr><tr><td align="left">7</td><td align="left">27</td><td align="left">32</td></tr><tr><td align="left">6</td><td align="left">33</td><td align="left">52</td></tr><tr><td align="left">5</td><td align="left">29</td><td align="left">70</td></tr><tr><td align="left">4</td><td align="left">26</td><td align="left">86</td></tr><tr><td align="left">3</td><td align="left">12</td><td align="left">93</td></tr><tr><td align="left">2</td><td align="left">6</td><td align="left">97</td></tr><tr><td align="left">1</td><td align="left">1</td><td align="left">98</td></tr><tr><td align="left">None</td><td align="left">3</td><td align="left">100</td></tr><tr><td align="left">TOTAL</td><td align="left">163</td><td align="left">100</td></tr></tbody></table></table-wrap></sec></sec><sec><title>Discussion</title><p>The present study assessed whether FDs were aware of and familiar with the recommendations in the DM2 guideline and whether this was related to FDs' previous specialty, background characteristics or specialist accessibility. This is the first time CPG application has been studied in Estonia.</p><p>CPGs have been developed and used in the US for more than three decades and about 1500 guidelines are available for the American Medical Association. This is a large number of guidelines to expect to be followed. A recent trend in Estonia is to facilitate the development of different guidelines. At present, about 20 guidelines have been compiled for family doctors. However, as the number of guidelines is increasing, their usage as well as the actual awareness of them may decline and expected improvement in the quality of care is questionable.</p><p>According to the present study, almost two-thirds of the family doctors have the DM2 guideline at their disposal, and the majority of the doctors consider them applicable and use it in daily practice. This is comparable to the results of other studies demonstrating a positive attitude towards CPGs. Eighty-three percent of Israeli family physicians believed the guideline could be implemented and 75% attained help in the management of patients with DM [<xref ref-type="bibr" rid="B13">13</xref>]. Those findings are significantly different from several other studies where guidelines are available for only about one-fourth or less of family physicians, while even fewer still report using them [<xref ref-type="bibr" rid="B14">14</xref>,<xref ref-type="bibr" rid="B15">15</xref>]. The reason for this may be that doctors are sometimes doubtful about CPGs, and they consider them less useful than other sources of medical information, as they are developed for reducing healthcare costs and may not be applicable for individual patients and for local settings [<xref ref-type="bibr" rid="B16">16</xref>,<xref ref-type="bibr" rid="B17">17</xref>]. The usage of the guideline and the knowledge about it may also depend on how CPGs are distributed and if special educational activities are undertaken. However, the results of this study did not reveal any difference in the provision of diabetes care between the doctors who had CPGs available and used it in comparison with those who did not.</p><p>Regarding general factors that might influence usage of CPGs, some studies have shown that younger doctors considered CPGs more useful than older doctors [<xref ref-type="bibr" rid="B18">18</xref>] but there was no difference in our study.</p><sec><title>Decision of treatment</title><p>The results of the current study demonstrate that FDs mostly start treatment at higher FBG levels and the number of patients whom they considered to be compensated is low. The doctors who had guidelines and used them reported acting in the same way as those who did not. This shows that even if FDs had the relevant knowledge, many of them were reluctant to use it. On the other hand, the knowledge that they use may not have been derived from CPGs but from other sources.</p><p>The overwhelming majority of the FDs in our study tended to start treatment and were content with treatment outcome at higher levels of FBG than recommended in the guideline. It has been similarly reported that Italian physicians are content with treatment outcome at quite high FBG levels [<xref ref-type="bibr" rid="B19">19</xref>]. Other studies support the idea that doctors are not fully aware of the recommended criteria of CPGs for intensive blood glucose treatment [<xref ref-type="bibr" rid="B19">19</xref>,<xref ref-type="bibr" rid="B20">20</xref>]. Characteristics such as practice location, practice type, list size and length of experience in our study did not predict the glycemic control of type 2 diabetic patients, which coincides with the finding of another study [<xref ref-type="bibr" rid="B21">21</xref>].</p></sec><sec><title>Following the guideline</title><p>Despite the fact that the majority of the doctors reported using the guideline, their knowledge of the tests and examinations recommended in the guideline were very variable. Blood pressure measurement was followed best, which is consistent with the findings from another study [<xref ref-type="bibr" rid="B22">22</xref>]. Of the laboratory tests, the performance of creatinine showed the best concordance with the guideline. In a similar American study, the performance of creatinine, proteinuria and HbAc1 tests was higher than 90% and had increased compared with the early nineties [<xref ref-type="bibr" rid="B22">22</xref>].</p><p>There were significant differences between the performances of the tests, from underuse to overuse, which might indicate the preferences of individual doctors. A factor leading to underuse of laboratory tests might be a lack of resources, which depends on the healthcare system and the financial system [<xref ref-type="bibr" rid="B11">11</xref>].</p><p>The results of this study showed that regardless of whether the doctors had the guideline available or not, the process of diabetes care remained very variable. It can be presumed that guidelines are not the only source for acquiring knowledge and information.</p><p>Following the guideline might be influenced by the guideline's dissemination and implementation strategy. It is evident that a part of Estonian FDs do not have a copy of the guideline available thus the process of translating diabetes guidelines into practice has occurred by diffusion [<xref ref-type="bibr" rid="B23">23</xref>] and partial dissemination [<xref ref-type="bibr" rid="B23">23</xref>]. Nevertheless there is imperfect evidence to support decisions about which guideline dissemination and implementation strategies are likely to be efficient under different circumstances [<xref ref-type="bibr" rid="B24">24</xref>].</p><p>However, guidelines cannot address all uncertainties in current clinical practice and should only be seen as one strategy among others that can help improve the quality of care that patients receive.</p></sec><sec><title>Limitations of the study</title><p>Despite the low response rate, the results can still be generalized as the final structure of the respondents and non-respondents did not differ from each other. It is quite possible that self-reported data overestimates the real behavior of the doctors. Hereby according to our data there may be even more cause for concern for provided care. Even with these limitations the present study provides valuable information about the knowledge and behavior of FDs, who most frequently provide care for DM2 patients in Estonia.</p></sec></sec><sec><title>Conclusion</title><p>Guidelines are widely available, and are perceived as a useful and helpful source by most practitioners. Nevertheless, the way FDs take care of patients with diabetes varies remarkably. Updated and evidence-based guidelines can be useful as an educational tool but still a knowledge gap and variable behavior in clinical performance exist.</p></sec><sec><title>Authors' contributions</title><p>ML, RK and AR participated in the design of the survey. RK, AR and IO carried out the study. ML, RK and AR contributed to the statistical analyses of the work. AR drafted the manuscript. AR, RK, IO and ML participated in the critical revisions of the manuscript. All authors read and approved the final manuscript.</p></sec><sec><title>Pre-publication history</title><p>The pre-publication history for this paper can be accessed here:</p><p><ext-link ext-link-type="uri" xlink:href="http://www.biomedcentral.com/1471-2296/7/36/prepub"/></p></sec>
|
ProFace: a server for the analysis of the physicochemical features of protein-protein interfaces
|
<sec><title>Background</title><p>Molecular recognition is all pervasive in biology. Protein molecules are involved in enzyme regulation, immune response, signal transduction, oligomer assembly, etc. Delineation of physical and chemical features of the interface formed by protein-protein association would allow us to better understand protein interaction networks on one hand, and to design molecules that can engage a given interface and thereby control protein function on the other hand.</p></sec><sec><title>Results</title><p>ProFace is a suite of programs that uses a file, containing atomic coordinates of a multi-chain molecule, as input and analyzes the interface between any two or more subunits. The interface residues are shown segregated into spatial patches (if such a clustering is possible based on an input threshold distance) and/or core and rim regions. A number of physicochemical parameters defining the interface is tabulated. Among the different output files, one contains the list of interacting residues across the interface. Results can be used to infer if a particular interface belongs to a homodimeric molecule.</p></sec><sec><title>Conclusion</title><p>A web-server, ProFace (available at <ext-link ext-link-type="uri" xlink:href="http://www.boseinst.ernet.in/resources/bioinfo/stag.html"/>) has been developed for dissecting protein-protein interfaces and deriving various physicochemical parameters.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Saha</surname><given-names>Rudra P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Bahadur</surname><given-names>Ranjit P</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Pal</surname><given-names>Arumay</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Mandal</surname><given-names>Saptarshi</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Chakrabarti</surname><given-names>Pinak</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Structural Biology
|
<sec><title>Background</title><p>Most proteins function by interacting with other molecules; the binding sites have evolved for achieving specific interactions and avoiding undesirable associations that would be deleterious to the normal functioning of the cell. Thus the interfaces between two protein subunits provide context for understanding the principles of molecular recognition. A large volume of structural data on protein interactions, either complexes between independent polypeptide chains, or oligomeric assembly of subunits, is available in the Protein Data Bank (PDB) [<xref ref-type="bibr" rid="B1">1</xref>], which has been used to generate diverse datasets of protein-protein interfaces [<xref ref-type="bibr" rid="B2">2</xref>]. The physical and chemical features of the interfaces have been analyzed [<xref ref-type="bibr" rid="B3">3</xref>-<xref ref-type="bibr" rid="B8">8</xref>] and softwares/websites, such as Protein-Protein Interaction Server [<xref ref-type="bibr" rid="B6">6</xref>], MolSurfer [<xref ref-type="bibr" rid="B9">9</xref>], SPIN-PP [<xref ref-type="bibr" rid="B10">10</xref>], etc. are available for their calculations. Nevertheless, our understanding of the biomolecular interactions is not adequate enough, for example, to infer unambiguously the arrangement of the subunits in an oligomeric protein from crystallographic studies [<xref ref-type="bibr" rid="B11">11</xref>], or to ascertain a high success rate for the prediction of models of protein-protein complexes through docking methods [<xref ref-type="bibr" rid="B12">12</xref>].</p><p>Recently, protein-protein interfaces have been dissected from new perspectives [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. It has been shown that many large interfaces are not contiguous, but built of spatially demarcated surface patches. Such segregation into patches is also indicative of the location and distribution of water molecules held in the interface [<xref ref-type="bibr" rid="B15">15</xref>]. Additionally, one can also divide the interface into core and rim regions using the difference of solvent accessibilities of residues and the chemical properties of each region are quite distinct. Interestingly, this division also mirrors the degree of conservation of interface residues in a family of homologous proteins [<xref ref-type="bibr" rid="B16">16</xref>], and this represents an important signature of protein interaction sites. Various other physicochemical parameters have also been developed [<xref ref-type="bibr" rid="B17">17</xref>,<xref ref-type="bibr" rid="B18">18</xref>], which in combination, can distinguish the true oligomeric state (dimer, in particular) from the lattice contacts observed in protein crystals. In this article we describe a web-server, ProFace that dissects a given <underline>pro</underline>tein-protein inter<underline>face</underline> and obtains various parameters to characterize it.</p></sec><sec><title>Implementation and results</title><sec><title>Input file and parameters</title><p>All the protein chains should be contained in the input file in the PDB format and the user must indicate which chains (a maximum of three allowed) constitute each of the two components forming the interface between them. Also, one has to specify the way to display the dissected interface, i.e., to show the residues belonging to core and rim and/or in spatial patches. For clustering into patches the threshold distance has to be supplied. This distance should typically be half the maximum distance between any two interface atoms on a given protein chain – the latter distance is listed along with the other parameters in the output. Ideally, the number of patches should be the same on both the components and if this is not the case the threshold value may have to be slightly changed (increase to reduce the number of patches and vice-versa) to achieve this. The suggested values are 15 Å for protein-protein complexes [<xref ref-type="bibr" rid="B13">13</xref>] and 22 Å for homodimers [<xref ref-type="bibr" rid="B14">14</xref>], as these gave patches that were visually meaningful in the vast majority of the cases.</p></sec><sec><title>Output files and parameters</title><p>There are five types of output: a) plot of interface residues with secondary structural information; b) statistics of interface parameters; c) coordinates of interface atoms and the PDB files in which the interface residues are tagged; d) list of residue contacts across interface; and e) the view of the interface atoms.</p><sec><title>Plot of interface residues with secondary structural information</title><p>The secondary structural elements (α-helix and β-strand) are computed using the program DSSP [<xref ref-type="bibr" rid="B19">19</xref>] and shown below the residue names (one-letter code) along the sequence for the individual chains. The sequence information is based on residues for which coordinates are available (and not on the basis of SEQRES records). There are three options to show the interface residues: (i) to simply show the interface residues (in red color); (ii) to show them dissected into core/rim regions (red/blue color); and to show them dissected in two different ways – spatial patches (in different colors) and core/rim regions (upper/lower case). An example of option (iii) is displayed in Figure <xref ref-type="fig" rid="F1">1</xref>.</p></sec><sec><title>Statistics of interface parameters</title><p>A typical example of output parameters is shown in Table <xref ref-type="table" rid="T1">1</xref>. The interface area is the sum of the solvent accessible surface areas (ASA) of the two components less that of the pair. ASA is calculated using program NACCESS [<xref ref-type="bibr" rid="B20">20</xref>]. All protein atoms or residues contributing more than 0.1 Å<sup>2 </sup>to the interface area are counted as interface atoms or residues, whose numbers are tabulated. Non-polar interface area is the area contributed by non-polar interface atoms (i.e., all atoms excluding O, N and S). Interface area/surface area is the ratio of the interface area to the rest area of the protein surface in the two components. Fraction of non-polar atoms is based not on the area contributed, but on the number of atoms. Fraction of fully buried atoms is the ratio of interface atoms that are completely buried in the complex (with ASA = 0) to the total number of interface atoms (which also include atoms that do not have zero ASA in the complex). Residue propensity score and local density are defined in Bahadur et al. [<xref ref-type="bibr" rid="B17">17</xref>]. Residues with at least one fully buried interface atom are designated as core residues, while rim residues do not contain any interface atom that is fully-buried. Once a residue is identified as core, all its constituent atoms are assumed to be in core (irrespective of the atom being fully or partially buried) and the interface area contributed by the atoms of the residue is part of the core region. Statistics also include atoms/residues/areas divided into core and rim regions (Table <xref ref-type="table" rid="T2">2</xref>). Also the number of patches in individual chains and their respective sizes are tabulated (Table <xref ref-type="table" rid="T3">3</xref>).</p></sec><sec><title>Output files</title><p>The 4-digit code used to name the output files are randomly generated and does not have any correspondence to the input file name. The coordinates are stored in two types of files (with extensions .pdb and .int) and there are two files (corresponding to individual components) of each type. In the .pdb file the interface residues are distinguished from the remaining atoms in the structure on the basis of the content in the two columns – occupancy factor and B-factor. The non-interface residues have a value of 0.00 in these columns. For the interface residues, a) the occupancy is replaced by -5.00 (if it is a core residue) or 5.00 (if it is a rim residue); b) the B-factor column is replaced by a value 1.00 through 9.00, depending on the patch to which the residue belongs. In the .int file, only the interface atoms are kept, with the occupancy and the B-factor column modified as above (and an additional information on patches is also provided by appending labels a, b, c, ... to the keyword ATOM to correspond to patch numbers 1, 2, 3,...). Moreover, there are two additional columns, in which the ASAs of the constituent atoms in the individual component and in the complex are provided. One can use this information to calculate the interface area contributed by individual residues and, for example, correlate with the thermodynamic data on the free energy of binding [<xref ref-type="bibr" rid="B16">16</xref>].</p><p>Another output file (with extension .cont) provides the list of residue contacts across the interface. For an interface residue in the first component the list shows the interface residues from the other component which are within a distance of 4.5 Å. If a pair of residues in contact have the same residue name and number, this is indicated by the symbol '<< ---' at the end of the line. This interaction has been designated as self-contact and indicates that the interface may have been formed by components/chains related by a 2-fold symmetry [<xref ref-type="bibr" rid="B18">18</xref>]. An example of the presence of self-contacting residues in a homodimer structure is presented in Figure <xref ref-type="fig" rid="F2">2</xref>. Some of the parameters in Table <xref ref-type="table" rid="T1">1</xref>, along with the information on self-contacting residues may be used to ascertain if a 2-fold related contact observed in a crystal structure truly represents a biological homodimer.</p></sec><sec><title>View of the interface atoms</title><p>This can be done using either RasMol [<xref ref-type="bibr" rid="B21">21</xref>] or CHIME [<xref ref-type="bibr" rid="B22">22</xref>], depending on whichever program has been configured by the user on the machine. Clicking on the RasMol link will first enable the user to download the PDB file (with interface atoms), which can then be viewed by either program. Clicking on the CHIME link loads the PDB file directly in CHIME. As the B-factor column of the PDB file has been replaced by number codes indicating the patch to which the atoms belong, the interface atoms can be colored on the basis of patches using RasMol. Also, the PDB file generated by the program can be used in GRASP [<xref ref-type="bibr" rid="B23">23</xref>] to color the molecular surface according to the criterion of patch or core/rim region.</p></sec></sec></sec><sec><title>Conclusion</title><p>ProFace can be used to dissect a protein-protein interface, deriving physicochemical parameters. The output can be used to display the interface with standard softwares and understand the biological significance of the interaction.</p></sec><sec><title>Availability and requirements</title><p>• Project name: ProFace</p><p>• Project home page: <ext-link ext-link-type="uri" xlink:href="http://www.boseinst.ernet.in/resources/bioinfo/stag.html"/></p><p>• Operating system(s): Platform independent</p><p>• Programming language: Java, C++</p><p>• Other requirements: JRE 1.4.2.04 or higher, Chime plug-in 2.6 or higher; all of them are available for download at the above web address</p><p>• License: Free</p><p>• Any restrictions to use by non-academics: None</p></sec><sec><title>Authors' contributions</title><p>RPS and RPB wrote the source codes, participated in developing the server. AP and SM participated in developing the server. PC conceived the study, and participated in its design, analysis, and coordination. RPS, RPB, AP, SM and PC all contributed to writing the final manuscript and interpretation of data.</p></sec>
|
Does a preceding hand wash and drying time after surgical hand disinfection influence the efficacy of a propanol-based hand rub?
|
<sec><title>Background</title><p>Recently, a propanol-based hand rub has been described to exceed the efficacy requirements of the European standard EN 12791 in only 1.5 min significantly. But the effect of a 1 min preceding hand wash and the effect of one additional minute for evaporation of the alcohol after its application on the efficacy after a 1.5 min application time has never been studied.</p></sec><sec sec-type="methods"><title>Methods</title><p>We have investigated a propanol-based hand rub (Sterillium<sup>®</sup>, based on 45% propan-2-ol, 30% propan-1-ol and 0.2% mecetronium etilsulfate) in three variations: with (A) and without (B) a 1 min hand wash before the disinfection of 1.5 min with immediate sampling after the disinfection; and (C) without a hand wash before the disinfection but with sampling 1 min after the disinfection. The efficacy of the three variations was compared to the reference treatment of EN 12791. All experiments were performed in a Latin-square design with 20 volunteers. Pre- and post-values (immediate and after 3 hr) were obtained according to EN 12791.</p></sec><sec><title>Results</title><p>The 3 min reference disinfection reduced resident hand bacteria on average by 1.8 log<sub>10 </sub>steps (immediate effect) and 1.4 log<sub>10</sub>-steps (sustained effect) respectively. Method A was equally effective as the reference (immediate efficacy: 1.5 log<sub>10</sub>-steps; sustained efficacy: 1.6 log<sub>10</sub>-steps). Method B seemed to be more effective (immediate efficacy: 2.3 log<sub>10</sub>-steps; sustained efficacy: 1.7 log<sub>10</sub>-steps). Method C revealed the best efficacy (immediate efficacy: 2.3 log<sub>10</sub>-steps; sustained efficacy: 2.0 log<sub>10</sub>-steps). A comparison of all three treatment variations and the reference treatment revealed a significant difference for the immediate efficacy (p = 0.026; Friedman test), but not for the sustained efficacy (p = 0.430). A post-hoc-test for the immediate efficacy indicated a significant difference between methods A and C (p < 0.05; Wilcoxon-Wilcox test). Hence, none of the treatment variations was significantly less effective than the reference treatment.</p></sec><sec><title>Conclusion</title><p>An application of the propanol-based hand rub for 1.5 min after 1 min hand wash fulfills the efficacy requirements of EN 12791. The efficacy can be improved to some extent by omitting the preceding hand wash and by awaiting the evaporation of the alcohol which is clinical practice anyway. The preceding hand wash has the most negative effect on the immediate effect. Based on our data hands should not be routinely washed before the disinfection period unless there is a good reason for it such as visible soiling.</p></sec>
|
<contrib id="A1" contrib-type="author"><name><surname>Hübner</surname><given-names>Nils-Olaf</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A2" contrib-type="author"><name><surname>Kampf</surname><given-names>Günter</given-names></name><xref ref-type="aff" rid="I1">1</xref><xref ref-type="aff" rid="I2">2</xref><email>[email protected]</email></contrib><contrib id="A3" contrib-type="author"><name><surname>Kamp</surname><given-names>Philipp</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib><contrib id="A4" contrib-type="author"><name><surname>Kohlmann</surname><given-names>Thomas</given-names></name><xref ref-type="aff" rid="I3">3</xref><email>[email protected]</email></contrib><contrib id="A5" corresp="yes" contrib-type="author"><name><surname>Kramer</surname><given-names>Axel</given-names></name><xref ref-type="aff" rid="I1">1</xref><email>[email protected]</email></contrib>
|
BMC Microbiology
|
<sec><title>Background</title><p>Alcohol-based hand rubs are recommended and widely used for surgical hand disinfection in many countries [<xref ref-type="bibr" rid="B1">1</xref>,<xref ref-type="bibr" rid="B2">2</xref>]. The main aim of the pre-operative treatment of hands is the reduction of resident skin bacteria to a minimum in order to reduce the risk of surgical site infections in case of perforated surgical gloves [<xref ref-type="bibr" rid="B3">3</xref>]. Alcohols are considered to have better antimicrobial efficacy and dermal tolerance which are two clear advantages in comparison to antimicrobial soaps [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B5">5</xref>]. The application time for alcohol-based surgical hand rubs has become shorter over the last decades, and there was increasing evidence that with well-formulated hand rubs an equivalent efficacy on the resident hand flora can be achieved at an application time of 1.5 min [<xref ref-type="bibr" rid="B6">6</xref>,<xref ref-type="bibr" rid="B7">7</xref>]. At the same time evidence was cumulating to suggest that a 1 min hand wash which is recommended and often performed before the disinfection period may reduce the efficacy of the alcohol to some extent [<xref ref-type="bibr" rid="B8">8</xref>]. But this effect has so far only been studied for a 3 min application time of the hand rubs. The influence of a preceding hand wash as well as of drying by evaporation of the hand rub on the efficacy of disinfection with an application time of 1.5 min is unknown. Therefore, these influences were studied by determination of the efficacy of a hand rub according to EN 12791 and compared with the 3 min reference treatment [<xref ref-type="bibr" rid="B9">9</xref>]. The test design of EN 12791 ensures that the treated hands are sampled immediately after the specific application time even if they are still wet with the hand rub. In clinical practice, however, it is recommended for healthcare workers to put on surgical gloves only onto dry hands due to a better skin tolerance and a lower risk for impairment of glove integrity [<xref ref-type="bibr" rid="B2">2</xref>,<xref ref-type="bibr" rid="B10">10</xref>]. Therefore, the test design of EN 12791 does not reflect the clinical practice of surgical hand disinfection in that particular point.</p></sec><sec sec-type="methods"><title>Methods</title><sec><title>Products and application</title><p>The following preparations were used: propan-1-ol (60%, v/v) as reference alcohol of EN 12791, and Sterillium based on 45% (w/w) propan-2-ol, 30% (w/w) propan-1-ol, and 0.2% (w/w) mecetronium etilsulfate (Bode Chemie GmbH & Co. KG, Hamburg, Germany). Sterillium was applied to the hands for 1.5 min as described recently [<xref ref-type="bibr" rid="B6">6</xref>]. The reference alcohol was applied to the hands for 3 min according according to EN 12791 [<xref ref-type="bibr" rid="B9">9</xref>].</p></sec><sec><title>Wash phase</title><p>The volunteers washed their hands with a non-medicated soap (sapo kalinus) as described by EN 12791. Thereafter hands were rinsed with running tap water and dried with a sterile paper towel.</p></sec><sec><title>Determination of the pre-values and post-values</title><p>Sampling, cultivation and calculation of values were done according to EN 12791. Volunteers rubbed the phalanges of one hand (randomly selected) for one minute in TSB (pre-value) or TSB with addition of 3% Tween 80, 3% saponin, 0.1% histidine, and 0.1% cysteine as neutralizers (immediate effect after disinfection or drying), because these neutralizers are shown to be effective [<xref ref-type="bibr" rid="B11">11</xref>]. The other hand was gloved for 3 hours for the assessment of the sustained effect. After taking off the glove sampling was done in the same way as for the immediate effect. For each volunteer the logarithmic reduction factor (RF) was obtained as the difference between the log<sub>10 </sub>pre-value and log<sub>10 </sub>post values.</p></sec><sec><title>Disinfection phase</title><p>Each of the 20 volunteers was treated with the reference product for 3 min as well as with Sterillium for 1.5 minutes. Between each product application, a rest period of at least one week elapsed in order to allow the reconstitution of normal skin flora.</p></sec><sec><title>Design</title><p>We have investigated surgical hand disinfection with Sterillium in three variations: with (A) and without (B) a 1 min hand wash before the disinfection of 1.5 min with immediate sampling after the disinfection; and (C) without a hand wash before the disinfection but with sampling 1 min after the disinfection. All three applications were tested in a Latin-square design against the reference treatment.</p></sec><sec><title>Statistics</title><p>A product is considered effective for surgical disinfection if the mean of the RF of both the immediate and sustained effect is not significantly lower than the corresponding mean RF of the reference treatment [<xref ref-type="bibr" rid="B9">9</xref>]. Differences in the immediate and sustained effects between the reference treatment and the three variations of product application were investigated by non-parametric Friedman tests, followed by the Wilcoxon-Wilcox procedure for pairwise comparisons. A p-value < 0.05 was chosen to indicate a significant difference.</p></sec></sec><sec><title>Results</title><p>The 3 min reference disinfection reduced resident hand bacteria on average by 1.8 log<sub>10 </sub>steps (immediate effect) and 1.4 log<sub>10</sub>-steps (sustained effect; Table <xref ref-type="table" rid="T1">1</xref>) An application of the hand rub after a 1 min hand wash was equally effective (immediate efficacy: 1.5 log<sub>10</sub>-steps; sustained efficacy: 1.6 log<sub>10</sub>-steps). An application of 1.5 min of the hand rub without a 1 min hand wash was slightly more effective (immediate efficacy: 2.3 log<sub>10</sub>-steps; sustained efficacy: 1.7 log<sub>10</sub>-steps). Application of the hand rub for 1.5 min followed by 1 min drying time revealed the best efficacy (immediate efficacy: 2.3 log<sub>10</sub>-steps; sustained efficacy: 2.0 log<sub>10</sub>-steps). A comparison of all three treatment variations and the reference treatment revealed a significant difference for the immediate efficacy (p = 0.026; Friedman test), but no significant difference for the sustained efficacy (p = 0.430). A post-hoc-test of pairwise differences for the immediate efficacy revealed only that method C was significantly more effective than method A (p < 0.05; Wilcoxon-Wilcox test). None of the pairwise comparisons of methods A to C with the reference reached statistical significance.</p></sec><sec><title>Discussion</title><p>For the first time it was demonstrated that the efficacy requirements of EN 12791 can be fulfilled with a propanol-based hand rub applied for 1.5 minute time with three different application procedures. The lowest efficacy was observed with a preceding 1 min hand wash, the highest efficacy was found without a preceding hand wash and with an additional drying time of 1 min after the disinfection period.</p><p>It has been shown before that a 1 min hand wash followed by drying hands with a paper towel has a significant effect on skin hydration which significantly increases for up to 9 min [<xref ref-type="bibr" rid="B8">8</xref>]. At the same time it was found that the efficacy of different types of alcohols is impaired by a hand wash to some extent [<xref ref-type="bibr" rid="B8">8</xref>]. The value of a hand wash before the disinfection period has been questioned recently in a draft European recommendation on surgical hand disinfection for two reasons [<xref ref-type="bibr" rid="B12">12</xref>]: First, a 1 min hand wash does not really reduce the number of skin bacteria to a relevant extent [<xref ref-type="bibr" rid="B4">4</xref>,<xref ref-type="bibr" rid="B13">13</xref>]. Second, it has been suggested in some studies that the efficacy of the alcohol-based hand rub is even reduced if hands are washed before disinfection [<xref ref-type="bibr" rid="B13">13</xref>,<xref ref-type="bibr" rid="B14">14</xref>]. The difference in efficacy of the same treatment was even more remarkable with and without a 2 min hand wash [<xref ref-type="bibr" rid="B15">15</xref>]. We can now add similar findings with an application time of 1.5 minutes.</p><p>Practical implications have to be adressed, too. The whole procedure of surgical hand disinfection is usually performed in the surgical theater block. If we assume that the surgeon enters the theater block with clean hands then there is no need to wash them before the disinfection period. If we assume that the surgeon enters the theater block with dirty hands then there is something wrong on another level. If we assume that hands have become dirty in the surgical theater block they should be washed immediately after soiling [<xref ref-type="bibr" rid="B1">1</xref>]. In that respect there is no reason to include a 1 min hand wash into the standard procedure of pre-operative treatments of hands especially since recent evidence points out some clear disadvantages of the hand wash.</p><p>The best efficacy was found without a hand wash and with additional drying time after the application of the hand rub. The immediate effect was only marginally better (0.03 log<sub>10</sub>-steps), the sustained effect improved somewhat more (0.3 log<sub>10</sub>-steps). It is certainly correct that the efficacy of a hand rub which is applied for a specific time such as 1.5 min should be assessed exactly after 1.5 min. At the same time the application by the healthcare workers will be different in the hospital as they have to await the evaporation of any residual alcohol from their hands before gloving. But the effect of the additional drying time has to our knowledge never been investigated with an alcohol-based hand rub although it has a clear practical relevance. We observed a trend towards a better overall efficacy if hands are allowed to dry for 1 min after the disinfection period. This finding was expected and can be explained by the additional and variable contact time of the alcohol with the skin. That is why the healthcare worker can expect that the "real efficacy" of an alcohol-based hand rub for surgical hand disinfection is to some extent better than measured according to EN 12791.</p></sec><sec><title>Conclusion</title><p>An application of the propanol-based hand rub for 1.5 min after 1 min hand wash fulfills the efficacy requirements of EN 12791. The efficacy can be improved to some extent by omitting the preceding hand wash and by awaiting the evaporation of the alcohol which is clinical practice anyway. The preceding hand wash has the most negative effect on the immediate effect. Based on our data hands should not be routinely washed before the disinfection period unless there is a good reason for it such as visible soil.</p></sec><sec><title>Competing interests</title><p>The second author is paid employee of Bode Chemie GmbH & Co. KG, Hamburg, Germany.</p></sec><sec><title>Authors' contributions</title><p>GK and AK designed the study, NOH, GK, PK and TK analysed the data, NOH, GK and AK wrote the manuscript. All authors read and approved the final manuscript.</p></sec>
|