Internal exposure to pollutants and body size in Flemish adolescents and adults: Associations and dose-response relationships

Article (PDF Available)inEnvironment international 36(4):330-7 · February 2010with40 Reads
DOI: 10.1016/j.envint.2010.01.005 · Source: PubMed
Abstract
Flanders is densely populated with much industry and intensive farming. Body size of 14- to 15-year old adolescents and of adults aged 50-65 was studied in relation to internal exposure to pollutants. 1679 adolescents (887 boys and 792 girls), 775 men and 808 women were selected as a random sample of the population. Concentrations of pollutants in blood or urine were measured in accordance with quality control/quality assurance procedures. Self-assessment questionnaires provided information on personal and life-style factors. Height and weight of subjects were measured. Confounding factors and significant covariates were taken into account. For boys and girls, height and body mass index (BMI) showed a negative association with urinary concentration of cadmium and BMI also with serum concentration of hexachlorobenzene (HCB) and with the sum of serum concentrations of polychlorinated biphenyls (PCBs) 138, 153, and 180 (marker PCBs), whereas BMI showed a positive association with serum concentration of PCB 118. For boys, height showed a negative association with urinary concentration of 1-hydroxypyrene (1-OHP) and positive associations with serum concentrations of HCB and PCB 118. For adults no significant associations between internal exposure and height were observed. For men, BMI showed negative associations with urinary cadmium concentration and with serum levels of marker PCBs and positive associations with serum levels of HCB, p,p'-dichlorodiphenyldichloroethylene (p,p'-DDE), PCB 118 and the dioxin fraction of dioxin-like activity. For women, BMI showed a negative association with urinary cadmium concentration, with blood lead concentration and with the concentration of marker PCBs in serum, and a positive association with serum concentrations of HCB, p,p'-DDE and PCB 118. Associations between biological effects and internal exposures were, in terms of the regression coefficient, often stronger at exposures below the median. Environmental exposures to pollutants resulting in "normal" levels of internal exposure were associated with quite substantial differences in body mass index.
Internal exposure to pollutants and body size in Flemish adolescents and adults:
Associations and doseresponse relationships
Willem Dhooge
a,1
, Elly Den Hond
b,1
, Gudrun Koppen
b
, Liesbeth Bruckers
c
, Vera Nelen
d
,
Els Van De Mieroop
d
, Maaike Bilau
e
, Kim Croes
f
, Willy Baeyens
f
,
Greet Schoeters
b,g
, Nicolas Van Larebeke
h,
a
Department of Endocrinology, Ghent University Hospital, Ghent, Belgium
b
Environmental Toxicology, Flemish Institute of Technological Research, Mol, Belgium
c
University of Hasselt, Diepenbeek, Belgium
d
Provincial Institute of Hygiene, Antwerp, Belgium
e
Department of Public Health, Ghent University, Ghent, Belgium
f
Analytical and Environmental Chemistry (ANCH), Vrije Universiteit Brussel, Brussels, Belgium
g
University of Antwerp, Antwerp, Belgium
h
Study Centre for Carcinogenesis and Primary Prevention of Cancer, Department of Radiotherapy and Experimental Cancerology, Ghent University, Ghent, Belgium
abstractarticle info
Article history:
Received 23 October 2009
Accepted 29 January 2010
Available online 24 February 2010
Keywords:
Lead
Cadmium
Organochlorines
Polychlorobiphenyls
PCB 118
Dioxin-like activity
Hexachlorobenzene
p,p-Dichlorodiphenyldichloroethylene
Body mass index
Height
Health effects
Low dose
Flanders is densely populated with much industry and intensive farming. Body size of 14- to 15-year old
adolescents and of adults aged 5065 was studied in relation to internal exposure to pollutants. 1679 adolescents
(887 boys and 792 girls), 775 men and 808 women were selected as a random sample of the population.
Concentrations of pollutants in blood or urine were measured in accordance with quality control/quality
assurance procedures. Self-assessment questionnaires provided information on personal and life-style factors.
Height and weight of subjects were measured. Confounding factors and signicant covariates were taken into
account. For boys and girls, height and body mass index (BMI) showed a negative association with urinary
concentration of cadmium and BMI also with serum concentration of hexachlorobenzene (HCB) and with the
sum of serum concentrations of polychlorinated biphenyls (PCBs) 138, 153, and 180 (marker PCBs), whereas BMI
showed a positive association with serum concentration of PCB 118. For boys, height showed a negative
association with urinary concentration of 1-hydroxypyrene (1-OHP) and positive associations with serum
concentrations of HCB and PCB 118. For adults no signicant associations between internal exposure and height
were observed. For men, BMI showed negative associations with urinary cadmium concentration and with serum
levels of marker PCBs and positive associations with serum levels of HCB, p,p-dichlorodiphenyldichloroethylene
(p,p-DDE), PCB 118 and the dioxin fraction of dioxin-like activity. For women, BMI showed a negative association
with urinary cadmium concentration, with blood lead concentration and with the concentration of marker PCBs
in serum, and a positive association with serum concentrations of HCB, p,p-DDE and PCB 118. Associations
between biological effects and internal exposures were, in terms of the regression coefcient, often stronger at
exposures below the median. Environmental exposures to pollutants resulting in normal levels of internal
exposure were associated with quite substantial differences in body mass index.
© 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Flanders is one of the most populated areas in Europe, with a dense
network of trafc roads, industrial activities and intensive farming
close to habitation. The ve-year (20012006) biomonitoring pro-
gram on neonates, adolescents and adults (5065 years) by the
Flemish Centre for Environment and Health aimed at measuring
internal exposure to pollutants in areas differing in pollution pressure
and assessing whether place of residence or observed differences in
internal concentrations of pollutants were associated with biological
and health effects. All public information on the project can be found
on the website www.milieu-en-gezondheid.be.
In this study, we report on body size of 50- to 65-year old adults and
14- to 15-year old adolescents in relation to internal exposure to
environmental pollutants suspected to affect hormonal equilibrium.
PCBs are known to have estrogenic, anti-estrogenic and anti-androgenic
activities (Bonefeld-Jorgensen et al., 2001); p,p-DDE was reported to
Environment International 36 (2010) 330337
The study was approved by the medicalethical committee of the University of
Antwerp on the 4th of July 2002.
Corresponding author. Ninoofse steenweg 379, B1500 Halle, Belgium. Tel.: + 32 2
380 14 10, + 32 475 44 99 55 (mobile); fax: +32 2 381 16 45.
E-mail address: nicolas.vanlarebeke@ugent.be (N. Van Larebeke).
1
These authors have contributed equally.
0160-4120/$ see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envint.2010.01.005
Contents lists available at ScienceDirect
Environment International
journal homepage: www.elsevier.com/locate/envint
have anti-androgenic properties (Kelce et al., 1995); HCB was reported
to affect oestradiol levels in animals (Alvarez et al., 2000; Foster et al.,
1995) and to interact with hormone receptors (Li et al., 2008); cadmium
was observed to be able to interact with both estrogen and androgen
receptors (Stoica et al., 2000; Martin et al., 2002); lead was reported to
have xeno-estrogenic activity (Martin et al., 2003) and to affect pubertal
development in girls (Selevan et al., 2003); polycyclic aromatic
hydrocarbons were reported to affect development (Choi et al., 2006)
and display AhR as well as estrogen receptor-mediated activity
(Hilscherova et al., 2000). We wanted to test the hypothesis that low
differences in levels of internal exposure (such as these occurring in the
general population in Flanders) to endocrine disrupting substances
result in differences in body size parameters. In addition, will this effect
be larger at the lower end of the range of measured internal exposures
than at the higher end of this range? This is what could happen if the
above mentioned pollutants interact with receptors, as their dose
response curves might be expected to follow MichaelisMenten kinetics
(Sheehan et al., 1999; Castano and Flores-Saaib, 2008).
2. Materials and methods
2.1. Selection and recruitment of participants
2.1.1. Adolescents
A Stratied Clustered Multi-Stage Design was used to select 1600
participants as a random sample of the adolescents residing in the
study areas, comprising 22% of the Flemish territory, 20% of the
Flemish population and 20% of the Flemish municipalities as described
in detail in Schroijen et al. (2008). The study areas were chosen to
represent different types of environmental pressure occurring in
Flanders. Sampling took place in three steps: rst by study area,
second by entities for access to participants (i.e. the schools), and
third by selection of the participants in accordance with the inclusion
criteria. The adolescents were enrolled via 42 schools located in the
nine selected regions, and sampled between October 2003 and July
2004. For the areas around waste incinerators it was not possible to
enroll adolescents through schools, because each separate a rea
around a particular incinerator was small and comprised only a few
streets. Therefore, adolescents living near an incinerator received a
home addressed letter for participation. Inclusion criteria were: being
born in 1988 or 1989, studying in the third year of secondary
education, living for at least ve years in the same study area, and
giving informed consent (both adolescent and parents). Of all pupils
who received an invitation, 28.4% did not respond, because they did
not fulll the inclusion criteria or because they were not interested,
and of those who did respond, 14.7% refused to participate. So 61.07%
of the pupils contacted wanted to participate. Among the pupils who
wanted to participate, 2.3% were excluded by the researchers because
they did not reside in the area since 5 years, and 1.9% because of
incomplete questionnaires or insufcient blood or urine. So, nally
58.5% of the pupils contacted participated in the study. The
recruitment resulted in a total of 1679 adolescents.
2.1.2. Adults
A Stratied Clustered Multi-Stage Design was used to select 775
men and 808 women (n = 1583) aged 50 to 65 as a random sample of
the population of the areas under study as described by De Coster et al.
(2008). Sampling took place in three steps: rst by study area,
secondly by sub-municipality entities for access to participants, and
thirdly by selection of the participants in accordance with the
inclusion criteria.
All participants signed an informed consent form and had the right
to withdraw from the study at any time. The study design was
approved by the medical
ethical committee of the University of
Antwerp on July 4th, 2002.
2.2. Blood and urine collection
Length and body weight of the participants were measured by a
study nurse. Each participant donated a urine sample of about 200 mL
and a blood sample of 40 mL for subsequent analysis. Serum samples
were prepared by immediate centrifugation of the coagulated blood.
Urine, whole blood and serum samples were fractionated immedi-
ately and stored at 20 °C until analysis.
2.3. Measurement of biomarkers of exposure
Lead and cadmium concentrations in whole blood were deter-
mined after an acid digestion pre-treatment destroying the organic
matrix and a ten times dilution, followed by high resolution-
inductively coupled plasma-mass spectrometry detection (ICP-MS)
as described by Schroijen et al. (2008). Detection limits for cadmium
and lead in the whole blood were 0.09 and 2.0 µg/L respectively for
digested blood samples diluted 10 times. Isotope Cd114 was used to
quantify the amount of cadmium in urine using ICP-MS. Urinary
cadmium levels were expressed in µg/g creatinine. Urine samples
were diluted in nitric acid (0.7%). Rhodium was used as an internal
standard. The detection limit for urinary cadmium was 0.002 µg/L. The
creatinine content in urine was determined by spectrophotometry.
Polychlorobiphenyl (PCB) 118, PCB 138, PCB 153, PCB 180, hexa-
chlorobenzene (HCB) and p,p-dichlorodiphenyldichloroethylene (p,p-
DDE) were measured in serum using gas-chromatography-electron
capture detection (GC-ECD) as described by Schroijen et al. (2008).The
detection limit of all chlorinated compounds in serum was 0.02 µg/L.
Blood fat was calculated on the basis of serum cholesterol and serum
triglycerides (Covaci et al., 2006). Levels of chlorinated compounds were
expressed in ng/g lipid.
CALUX analyses of the dioxin fraction of dioxin-like activity in
blood plasma were performed (only in the study on adults) as
described by Van Wouwe et al. (2004) and Schroijen et al. (2006).
Briey, 5 mL of blood plasma was extracted with acetone and n-
hexane and dried on a Celite/Na
2
SO
4
column. The extract was then
transferred on an acid silica column in series with an activated carbon
column (XCARB column). After elution of the sample with n-hexane,
the acid silica column was discarded and the XCARB column was then
differentially eluted to yield 3 fractions:
1. a mixture of n-hexane/acetone allows the elution of some toxic or
interfering compounds
2. the PCB fraction is eluted with a mixture of n-hexane/toluene/ethyl
acetate
3. the fraction with polychlorodibenzodioxins (PCDDs) and poly-
chlorodibenzofurans (PCDFs) is collected with 20 mL of toluene.
After this clean-up, fractions 1 and 2 were discarded and only the
dioxin fraction was used for the bio-analysis. The solution containing
the dioxins was then evaporated and exposed to the mouse hepatoma
H1L6.1 cell line developed by Xenobiotic Detection System, Inc. After
an exposure time of 20 h, cells were lysed and measurements were
made with a luminometer. TEQ-values were calculated after compar-
ison of the obtained signals to a 2,3,7,8-tetrachlorodibenzodioxin
calibration curve.
The determination in urine of 1-hydroxypyrene (1-OHP), a
metabolite of pyrene, was performed with high performance liquid
chromatography (HPLC) as described by Schroijen et al. (2008). The
detection limit was 0.030 µg/L.
t,t-Muconic acid (t,t-MA), a metabolite of benzene, was deter-
mined in urine by means of ion chromatography using SPE-SAX
columns as described by Schroijen et al. (2008). The detection limit
was 0.0086 mg/L. Levels of 1-OHP and t,t-MA were expressed in µg/g
creatinine and mg/g creatinine respectively.
All laboratories involved in the analyses of biomarkers applied
standard agreed quality control/quality assurance procedures.
331W. Dhooge et al. / Environment Internation al 36 (2010) 330337
2.4. Measurement of hormone levels
Commercial immunoassays were used to determine serum levels
of total testosterone (T), luteinizing hormone (LH), follicle stimulat-
ing hormone (FSH), sex hormone binding globulin (SHBG;) and total
17β-oestradiol (E2) as described by Croes et al. (2009). The free
fractions of T (fT) and E2 (fE2) were calculated from T and SHBG in
serum, as described by Croes et al. (2009). The intra- and interassay
coefcients of variation for all assays were less than 12%. For each
individual, the aromatase index the ratio of T on E2 (T/E2) was
calculated as pmol/pmol. Data on hormone levels were, in this
publication, only used as covariates in some analyses concerning
body size. A full report on our observations concerning associations of
hormone levels with internal exposure to pollutants is published
(Dhooge et al., 2009 ).
2.5. Questionnaires
Information on personal and life-style factors and on health status
was obtained by self-assessment questionnaires lled in by the
participants or, in the case of the adolescents, by the adolescents and
by their parents. Information provided by the parents included data
on educat ion, weight, height and health status of the parents.
Information provided by the participating adults and adolescents
included data on health status, contraception, smoking behavior and
consumption of alcohol and for the adults also education. For both
adolescents and adults highest education in the family was used as
parameter in statistical analysis. The participants also completed a
semi-quantitative food frequency questionnaire concerning daily
consumption of fruit and vegetables during the last year, and, as
described by Bilau et al. (2008), a semi-quantitative food frequency
questionnaire to assess the daily consumption of fat-containing food
items during the last year. In addition, the consumption of locally
produced food was recorded. On the basis of these questionnaires a
number of parameters were calculated, including, for adolescents,
consumption in g/day of fresh fruit and vegetables, cereals, sh, dairy
products, meat and daily intake of animal fat. For adults also the total
consumption of fat in g/day was calculated.
2.6. Data on sexual development
Data on growth and sexual development for 767 boys and 636 girls
were obtained from the 25 Centers for School Health Examination,
where all pupils are examined once every two years by school doctors.
All pupils were examined during the school year 20032004. Sexual
development was routinely measured through the international score
of Marshall and Tanner (1969, 1970), on a scale from 1 (start of
puberty) to 5 (adult stage). Genital (stages G1 to G5) and pubic hair
(P1 to P5) development were assessed in boys; breast (B1 to B5) and
pubic hair (P1 to P5) development in girls. Shortly before the
biomonitoring study was conducted, all school doctors had received
a re-training for the assessment of pubertal development since new
Flemish growth curves were developed in the period 20002004 (ref:
http://www.vub.ac.be/groeicurven/english.html). Data on sexual de-
velopment were, in this publication, only used as covariates in some
analyses concerning body size. A full report on our observations
concerning associations of sexual development with internal expo-
sure to pollutants is submitted (Den Hond et al., submitted).
2.7. Data treatment
Database management and statistical analyses, for male adults,
female adults, male adolescents and female adolescents separately,
were performed with SAS for Windows, version 9.1.3 and Statistica,
version 7.1. Data that were not normally distributed were subjected to
natural logarithmic transformation for use as dependent variables in
Ancova, multiple regression or logistic regression. Geometric means
or medians (10th and 90th percentile) are reported. The sum of
marker PCBs (PCB 138 + PCB 153+PCB 180) was used to reect
exposure to PCBs in general, whereas the serum concentration of PCB
118, considered to reect exposure to dioxin-like PCBs (Park et al.,
2007), was used separately.
Raw data were adjusted by regression analyses for some pre-
speci
ed, literature based confounders and also for covariates
showing signicant association with effect parameters. For adoles-
cents, confounders of data on height and BMI were age, height of
father and height of mother. For adults age was considered a
confounder for BMI. Adjustment for smoking was performed for
adolescents using the parameter daily smoking or not and for adults
by the number of cigarettes smoked per year. Adjustment for sexual
maturation, performed in some analyses, was done through classify-
ing the boys as having reached stage P3 and the girls as having
reached stage P4. Additionally, parameters related to food intake and
other parameters derived from the questionnaires were included as
covariates in multiple linear regression when they showed in simple
regression signicant asso ciation (p b 0.1) with the dependent
variable under study.
To evaluate the association of biological effect parameters with
parameters of internal exposure after adjustment, if relevant, for other
concurrent internal exposures, a series of forward stepwise multiple
regressions was performed in which we set the F value to enter at 1.2
and the F value to stay at 1.1, with the biological effect parameter as
dependent variable and as independent variables all predetermined
confounders and all parameters related to exposure, food intake, life
style or personal characteristics that showed (in simple regressions)
signicant (pb 0.1) association with the biological effect parameter
used as dependent variable. Confounders were forced in the resulting
model. For every analysis reported in the tables the list of confounders
and covariates included in the model is mentioned.
To assess the strength of the association (in terms of the regression
coefcient) in function of the intensity of internal exposure, two
additional sets of multiple regressions, including confounding factors,
were performed: one including only subjects with exposures above
the median for a particular pollutant, and one including only subjects
with exposures e qual to or below the median. To facilitate
interpretation of this comparison, the parameter for the particular
pollutant under study was used without natural logarithmic trans-
formation. A Piecewise Linear Regression model was used to assess
statistical signicance of differences in these additional multiple
regressions. To assess the strength of association in function of the
intensity of internal exposure after correction for other signicant
covariates or exposures, such additional sets of multiple regressions
were also performed including signicant covariates in addition to
confounding factors and nally also including all covariates and
exposure variables present in the models resulting from stepwise
regressions.
Analysis of residuals was performed after each multiple regression.
For calculation of binomial probability the probability of success on
a single trial was set at 0.5.
3. Results
3.1. Characteristics of participants, nutritional and other life-style factors
Of the boys (n =887), 14.5% drank alcoholic beverages at least weekly, 8.2%
smoked daily and 46.6% lived in a family in which at least one parent had received
higher education. Respective data for girls (n=792) were 13.8%, 7.7% and 53.1%. Oral
contraception was used by 9.3% of the girls. Other characteristics of participants are
summarized in Table 1a.
Of the adult men, (n=775), 22.1% were present smokers and 37.6% lived in a
family in which at least one partner had received higher education. Of the adult women
(n= 808), 14.4% were present smokers and 41.1% lived in a family in which at least one
partner had received higher education. Other characteristics of participants are
summarized in Table 1b.
332 W. Dhooge et al. / Environment Internation al 36 (2010) 330337
3.2. Internal exposure to pollutants
Data on internal exposure (median values and 10th and 90th percentiles) to
pollutants for boys and girls separately are given in Table 2. For adolescents, exposure
data and also differences in function of area of residence are discussed in detail by
Schroijen et al. (2008). Levels of lead and organochlorine chemicals were substantially
higher in boys than in girls. In boys, the following signicant (pb 0.05) positive
correlations with a Pearson correlation coefcient (r) of at least 0.3 were observed:
serum concentration of marker PCBs with serum concentrations of PCB 118 (r =0.52)
and HCB (r= 0.45); serum concentration of HCB with serum concentration of PCB 118
(r= 0.30). For girls, the following signicant positive correlations with a Pearson
correlation coefcient of at least 0.3 were observed: serum concentration of marker
PCBs with serum concentrations of PCB 118 (r = 0.65) and HCB (r = 0.44); serum
concentration of HCB with serum concentration of PCB 118 (r = 0.43).
Data on internal exposure (median values and 10th and 90th percentiles) to
pollutants for adults are given in Table 3. In men the following signicant (p b 0.05)
positive correlations with a Pearson correlation coefcient of at least 0.3 were
observed: serum concentration of marker PCBs with serum concentrations of PCB 118
(r= 0.69), HCB (r = 0.43) and p,p-DDE (r = 0.39); serum concentration of HCB with
serum concentrations of p,p-DDE (r = 0.40) and PCB 118 (r =0.44); urinary
concentration cadmium with blood concentration of cadmium (r=0.47). In women
the following signicant (pb 0.05) positive correlations with a Pearson correlation
coefcient of at least 0.3 were observed: serum concentration of marker PCBs with
serum concentrations of PCB 118 (r = 0.74), HCB (r= 0.43) and p,p-DDE (r =0.50);
serum concentrations of HCB with serum concentrations of p,p-DDE (r =0.42), and
PCB 118 (r =0.49); serum concentration of p,p-DDE with serum concentration of PCB
118 (
r= 0.48); urinary cadmium concentration with blood cadmium concentration
(r= 0.44).
3.3. Adolescents: body size in relation to internal exposure to pollutants
After adjustment for confounders and signicant covariates through a series of
multiple regressions, each comprising a body size parameter as dependent variable and
one parameter of internal exposure among the independent variables, signicant
associations were found between height or BMI and several parameters of internal
exposure. To evaluate the association of height and BMI with parameters of internal
exposure after additi onal adjustment, if relevant, for other concurrent internal
exposures, a series of forward stepwise multiple regressions was performed as
described under Materials and methods. Table 4 show s associations that were
signicant (pb 0.05) in multiple regressions including only one parameter of internal
exposure and that still showed the same trend in the stepwise multiple regressions in
which all signicant parameters of internal expos ure were included. In addition the
more complex data concerning BMI and PCB 118 are included.
Negative associations with height were observed for urinary cadmium (both boys
and girls) and 1-OHP (only boys), and these negative associations persisted after
additional correction for urinary creatinine. For both boys and girls the associations
between levels of urinary cadmium with height were only slightly weakened by
adjustment for sexual maturation and remained signicant.
Levels of HCB and PCB 118 showed a positive association with height in boys. The
association with PCB 118 was no longer statistically signicant after additional
adjustment for other parameters of internal exposure, although a positive trend was
still observed. As co-linearity between concentrations of PCB 118 and HCB (Pearson
correlation coefcient = 0.30) might have affected the association between PCB 118
and length after this additional adjustment, we performed a stratied analysis testing
the association of PCB 118 concentration with height after adjustment for covariates
and other exposures in 13 strata of HCB concentration. In 9 of these 13 strata a positive
association was observed between the concentration of PCB 118 and height (binomial
probability = 0.087), and in 2 of these strata the positive association was signicant
(pb 0.05) whereas in none of the strata the negative association was signicant. When
testosterone level was included as covariate, the positive association between HCB and
height disappeared almost completely, whereas the positive association between PCB
118 and height was also weakened but persisted to a greater extent (data not shown).
Whereas the association of HCB level with height in boys lost signicance after
adjustment for sexual maturation, the association of PCB 118 level with height was not
weakened at all by adjustment for sexual maturation (data not shown).
Levels of urinary cadmium, marker PCBs, HCB and p,p-DDE showed negative
associations with BMI for both boys and girls. These negative associations persisted
after additional correction for sexual maturation (data not shown).
The negative association of PCB 118 with BMI observed in simple regression as well
for girls as for boys might be due to the fact that those adolescents having higher PCB
118 levels often also had higher marker PCB levels (correlation coefcient 0.52 for boys
and 0.65 for girls) and marker PCBs showed a strong negative correlation with BMI.
After additional adjustment for other exposures a positive association was observed
between PCB 118 concentrations and BMI. To conrm this nding we performed a
stratied analysis testing the association between PCB 118 concentration and BMI after
adjustment for covariates and other exposures in 13 strata of marker PCB concentra-
tions for boys and 14 strata of marker PCB concentrations for girls. In 12 of these 13
strata for boys a positive association was observed between the concentration of PCB
118 and BMI (binomial probability= 0.0016), and in 2 of these strata the positive
association was signicant (p b 0.05). In 13 of these 14 strata for girls a positive
association was observed between the concentration of PCB 118 and BMI (binomial
probability = 0.0009), and in 2 of these strata the positive association was signicant
(pb 0.05).
We also studied the strength of association (in terms of the regression coefcient)
in function of the intensity of internal exposure. Comparison of the regression
Table 1a
Adolescents: characteristics and food intake by sex.
Boys (n= 887) Girls (n= 792)
Age (year) 14.8 (14.315.7) 14.9 (14.315.6)
Height(cm) 171 (159180) 164 (157172)
BMI (kg/m
2
) 19.8 (17.323.9) 20.4 (17.324.8)
Food and nutrient intake:
Fresh fruit and vegetables (g/day) 238 (76549) 270 (89591)
Cereals (g/day) 168 (76313) 138 (60242)
Fish (g/day) 14.6 (2.540.6) 14.1 (1.435.6)
Dairy products (g/day) 262 (67593) 185 (57464)
Meat (g/day) 111 (44196) 90 (31171)
Animal fat (g/day) 30.3 (16.658.6) 23.6 (12.646.0)
Data are medians (10th percentile90th percentile).
Table 1b
Adults: characteristics and food intake by sex.
Men (n= 775) Women (n= 808)
Age (year) 58.5 (51.963.4) 57.1 (51.462.5)
Height(cm) 175.0 (166180) 162 (154170)
BMI (kg/m
2
) 27.0 (23.131.9) 25.7 (21.432.8)
Alcohol intake (glasses/week) 10 (427) 7 (216)
Smokers: cigarettes/day 20 (530) 15 (427)
Total fat consumption g/day 58.0 (31.5101.2) 45.5 (24.177.4)
Animal fat consumption g/day 24.6 (12.249.7) 19.2 (9.538.9)
Green vegetables g/day 72 (29134) 72 (34143)
Other vegetables g/day 87 (27218) 101 (39250)
Fresh fruit g/day 104 (12301) 127 (22334)
Cereals g/day 160 (64284) 124 (52205)
Fish g/day 25.6 (7.663.1) 21.4 (6.653.4)
Meat g/day 96 (39176) 75 (27143)
Data are medians (10th percentile90th percentile).
Table 2
Internal exposure to pollutants in boys and girls.
Boys (n= 887) Girls (n= 792)
Blood cadmium (µg/L) 0.387 (0.0451.270) 0.386 (0.0451.243)
Blood lead (µg/L) 25.0 (12.051.2) 18.1 (8.838.1)
Urinary cadmium (µg/g creatinine) 0.199 (0.1150.352) 0.192 (0.1120.376)
Hexachlorobenzene (HCB) in serum
(ng/g fat)
22.8 (15.234.5) 18.3 (12.326.6)
p,p-DDE in serum (ng/g fat) 103.6 (46.8403.9) 84.0 (39.3247.1)
Sum of marke r PCBs in serum (ng/g fat)
a
79.8 (42.7141.3) 53.1 (30.398.5)
PCB 118 in serum (ng/g fat) 7.28 (2.8513.58) 6.00 (2.3711.59)
Urinary 1-OHP (µg/g creatinine) 0.088 (0.0190.463) 0.079 (0.0180.285)
Urinary t,t-MA (mg/g creatinine) 0.084 (0.0130.243) 0.090 (0.0160.285)
Data are medians (10th percentile90th percentile).
a
Sum of marker PCBs: sum of PCB congeners 138, 153 and 180.
Table 3
Internal exposure to pollutants in men and women.
Men (n= 775) Women(n =808)
Blood cadmium (µg/L) 0.44 (0.131.11) 0.51 (0.191.29)
Blood lead (µg/L) 40.7 (22.177.8) 37.6 (18.674.4)
Urinary cadmium (µg/g creatinine) 0.50 (0.301.10) 0.70 (0.401.50)
Hexachlorobenzene (HCB) in serum
(ng/g fat)
44.2 (23.790.9) 71.9 (36.2143.3)
p,p-DDE in serum (ng/g fat) 443 (1231398) 556 (1671818)
Sum of marker PCBs in serum (ng/g fat)
a
354 (222552) 335 (203521)
PCB 118 in serum (ng/g fat) 22.8 (10.446.8) 28.8 (12.952.9)
Urinary 1-OHP (µg/g creatinine) 0.135 (0.0240.588) 0.147 (0.0410.730)
Urinary t,t-MA (mg/g creatinine) 0.315 (0.1160.957) 0.554 (0.1671.794)
Data are medians (10th percentile90th percentile).
a
Sum of marker PCBs: sum of PCB congeners 138, 153 and 180.
333W. Dhooge et al. / Environment Internation al 36 (2010) 330337
coefcients obtained from additional multiple regressions including confounding
factors performed as described under Materials and methods showed that 10 out of 11
associations of body size parameters with internal exposure (Table 5) tended to be
stronger at exposures below the median value than at exposures above the median and
this difference was statistically signicant (p b 0.05) for 7 of these associations.
Additional multiple regressions including, besides confounding factors, also either
signicant covariates or all covariates and exposures present in the models resulting
from stepwise regressions, produced similar results: 10 out of 11 associations of
hormone levels with internal exposure tended to be stronger at exposures below the
median value (data not shown). These differences persisted after additional correction
for sexual maturation (data not shown). The association between PCB 118 serum
concentration and BMI after adjustment for the confounding effect of marker PCBs was
also stronger at exposures below the median (data not shown), corroborating the trend
for most other associations described here.
3.4. Adults: body size in relation to internal exposure to pollutants
After adjustment for confounders and signicant covariates through a series of
multiple regressions, each comprising a body size parameter as dependent variable and
one parameter of internal exposure among the independent variables, signicant
associations were found between BMI and several parameters of internal exposure
(Table 6). For adults, no signicant associations were found between height and internal
exposure to the measured pollutants. To evaluate the association of BMI with parameters
of internal exposure after additional adjustment, if relevant, for other concurrent internal
exposures, a series of forward stepwise multiple regressions was performed as described
under Materials and methods (Table 6). For men, signicant positive associations were
observed between BMI and serum concentrations of HCB, p,p-DDE, PCB 118 and dioxin-
likeactivity which persistedafter additional adjustment for concurrent internal exposures.
Asignicant negative association was found between BMI and serum c oncentration of
marker PCBs after adjustment for concurrent internal exposures and between BMI
and urinary cadmium concentration, but with reg ard to u rinary cad mium stat istical
signicance was lost after additiona l adjustme nt for c onc urr ent interna l exposures.
For women, BMI showed signicant positive associations with serum concentrations
of HCB, p,p-DDE and PCB 118 and signicant negative associations with serum
marker PCB concentration and with urinary cadmium concentration, all of which
persisted after additi onal adjust ment for concu rrent internal ex posures. I n additi on, a
signicant negative association was found between BMI and blood lead concentra-
tion, but statistical signicance was lost after additional adjustment for concurrent
internal expo sures.
We also studied the strength of association (in terms of the regression coefcient)
in function of the intensity of internal exposure. Comparison of the regression
coefcients obtained from additional multiple regressions including confounding
factors performed as described under Materials and methods showed that 11 out of 12
associations of body size parameters with internal exposure (Table 7) tended to be
stronger at exposures below the median value than at exposures above the median and
this difference was statistically signicant (p b 0.05) for 7 of these associations.
4. Discussion
Our cross sectional biomonitoring study has certain limitations in
terms of demonstrating causal relationships. Also, the biological
parameters under consideration could not only be inuenced by
exposures and other factors present at the time of the study, but also
by such factors acting many years before. Notwithstanding these
limitations, several potentially interesting associations were ob-
served. Our understanding of the mechanisms underlying these
associations is however incomplete.
Urinary cadmium concentration showed a negative association
with height in boys, a negative association with height and BMI in girls
and a negative association with BMI for both adult men and women.
In boys, the negative association with BMI almost disappeared and
lost statistical signicance after adjustment for concurrent exposures.
That negative associations with height persisted after additional
correction for urinary creatinine, suggests that these associations
were not simply due to higher creatinine levels in adolescents of
greater height. That we found, in the boys participating in this study,
also a negative association between urinary cadmium levels and sex
hormone levels (Dhooge et al., 2009) suggests that effects on height
might result from effects on hormone levels. However, that negative
associations with height observed for adolescents persisted after
additional correction for sexual matura tion suggests that these
associations were not solely due to effects on hormone levels and
sexual maturation. Cadmium showed, in vitro, toxic effects on the
Table 4
Adolescents: association of height and BMI with internal exposure.
Height or BMI Exposure marker Boys. Multiple regression with all confounders and
signicant covariates
Boys. Forward stepwise multiple regression starting with all
confounders, signicant covariates and signicant internal exposures
a
Change (95%CI) in height or BMI for a
doubling of exposure
p Change (95%CI) in height or BMI for a doubling of
exposure
p
Height (cm)
b
Urinary cadmium (µg/g creatinine)
c
1.09 ( 0.41/1.76) 0.0016 1.18 ( 0.48/1.87) 0.0009
Height (cm)
b
HCB (ng/g fat)
c
+1.26 (+0.17/+2.35) 0.024 + 1.30 (+0.20/+2.39) 0.020
Height (cm)
b
Urinary 1-OHP (µg/g creatinine)
c
0.33 ( 0.03/0.64) 0.033 0.32 ( 0.01/0.63) 0.042
Height (cm)
b
PCB 118 (ng/g fat)
c
+0.73 (+0.12/+1.34) 0.019 + 0.32 ( 0.37/+1.01) 0.37
BMI
d
Urinary cadmium (µg/g creatinine)
c
0.42 ( 0.13/0.71) 0.0040 0.18 (+0.08/0.44) 0.18
BMI
d
HCB (ng/g fat)
c
2.27 ( 1.83/2.71) b 0.000001 0.69 ( 0.20/1.18) 0.0059
BMI
d
Marker PCBs (ng/g fat)
c
2.20 ( 1.94/2.46) b 0.000001 2.45 ( 2.07/2.82) b 0.000001
BMI
d
PCB 118 (ng/g fat)
c
0.76 ( 0.51/1.02) b 0.000001 +0.56 (+0.27/+0.85) 0.00015
Girls. Multiple regression with all confounders and
signicant covariates
Girls. Forward stepwise multiple regression starting with all
confounders, signicant covariates and signicant internal exposures
a
Height (cm)
e
Urinary cadmium (µg/g creatinine)
c
0.61 ( 0.14/1.08) 0.012 0.53 ( 0.05/1.01) 0.013
BMI
f
Urinary cadmium (µg/g creatinine)
c
0.47 ( 0.18/0.76) 0.0016 0.63 ( 0.34/0.92) 0.000028
BMI
f
HCB (ng/g fat)
c
1.64 ( 1.13/2.15) b 0.000001 0.64 ( 0.03/1.25) 0.041
BMI
f
Marker PCBs (ng/g fat)
c
1.73 ( 1.43/2.03) b 0.000001 2.01 ( 1.62/2.41) b 0.000001
BMI
f
PCB 118 (ng/g fat)
c
0.38 ( 0.10/0.65) 0.0081 +0.74 (+0.41/+1.07) 0.000014
a
The resulti ng models comprised, as independent variables for associations with height of boys: urinary concentration of cadmium and 1-hydroxypyrene, serum concentration of
HCB, age, height of mother, height of father and alcohol consumption; for associations with BMI of boys: urinary concentration of cadmium, serum concentration of marker PCBs, PCB
118, p,p-DDE and HCB, age, height of mother, height of father, smoking, parental education and alcohol consumption; for associations with height of girls: urinary concentration of
cadmium and 1-hydroxypyrene, age, height of mother, height of father and smoking; for associations with BMI of girls: urinary concentration of cadmium, serum concentration of
marker PCBs, HCB and PCB 118, age, height of mother, height of father and smoking.
b
Confounders and signicant covariates comprised height of mother, height of father, age, smoking and use of alcohol; signicant internal exposures comprised serum
concentrations of marker PCBs, HCB and PCB 118, and urinary concentrations of cadmium and 1-OHP.
c
ln transformed.
d
Confounders and signicant covariates comprised height of mother, height of father, age, parental education, smoking and use of alcohol; signicant internal exposures
comprised serum concentrations of marker PCBs, HCB, p,p-DDE and PCB 118, and urinary concentration of cadmium.
e
Confounders and signicant covariates comprised height of mother, height of father, age, smoking and parental education; signicant internal exposures comprised urinary
concentrations of cadmium and 1-OHP.
f
Confounders and signicant covariates comprised height of mother, height of father, age, smoking and parental education; signicant internal exposures comprised serum
concentrations of marker PCBs, HCB, p,p-DDE and PCB 118, blood concentration of lead and urinary concentration of cadmium.
334 W. Dhooge et al. / Environment Internation al 36 (2010) 330337
testis (Steinberger and Klinefelter, 1993), was found to inhibit the
expression of steroidogenic acute regulatory protein (StAR) (Gun-
narsson et al., 2004) and to cause a decrease in growth hormone levels
in rats (Lafuente et al., 2001) and to cause, in vitro in human cells, a
decrease in the number of estrogen receptors (Garcia-Morales et al.,
1994). Cadmium can also activate the estrogen receptor alpha (Stoica
et al., 2000) and, in view of the stimulation of lipid mobilization and
lipolysis known to be induced by sex hormones (Hackney et al., 2000),
this xeno-estrogenic activity might contribute to the negative
association with BMI observed for girls, men and women, whereas
the fact that cadmium is associated with lower testosterone levels in
boys might explain the absence, after adjustment for concurrent
exposures, of a signicant negative association in boys. That all
observed effects were more pronounced at exposure values below the
median than at higher exposures, is certainly compatible with a
receptor-mediated mechanism (see below).
HCB serum levels showed a positive association with height in
boys. This increase in height might result from the higher sex
hormone levels and the inhibition of aromatase which we found in
association with higher HCB serum levels in the boys participating in
this study (Dhooge et al., 2009). Inhibition of aromatase has been used
to stimulate growth in short boys (Hero et al., 2005).
Urinary 1-OHP concentration showed a negative association with
height in boys. That the negative association with height persisted
after additional correction for urinary creatinine suggests that this
association was not simply due to higher creatinine levels in
adolescents of greater height. Our ndings are consistent with the
negative association reported for prenatal exposure to polycyclic
aromatic hydrocarbons and growth (Choi et al., 2006; Tang et al.,
2006) and with the reported effect of polycyclic aromatic hydro-
carbons on cells of the skeletal system, adversely affecting growth
(Holz et al., 2007).
That, in contrast with the observations for adolescents, we found
no signicant associations between internal exposure to pollutants
and height in adults aged 5065, suggests that the concentrations of
pollutants measured in adults at the time of the study showed not
enough correlation with the internal concentrations that prevailed
when these subjects were growing as children or adolescents.
The negative association observed between BMI and blood lead
concentration for adult women aged 5065 might be due in part to the
xeno-oestrogenic activity of lead described by Martin et al. (2003),
Table 5
Adolescents: strength of association in function of intensity of internal exposure.
Body size
parameter
Exposure marker Multiple regression with inclusion of confounders
a
Regression coefcient (standard error of the mean)
Boys with exposure
equal to or below
median
Boys with
exposure above
median
p
b
Height (cm) Urina ry cadmium
(µg/g creatinine)
9.73 (9.64) 1.19 (1.14) 0.023
Height (cm) HCB (ng/g fat) 0.063 (0.114) +0.081 (0.050) 0.51
Height (cm) Urinary 1-OHP
(µg/g creatinine)
27.8 (14.7) 1.7 (0.9) 0.19
Height (cm) PCB 118 (ng/g fat) +0.437 (0.210) +0.062 (0.051) 0.19
BMI Urinary cadmium
(µg/g creatinine)
7.05 (4.70) 0.96 (0.42) 0.087
BMI HCB (ng/g fat) 0.356 (0.051) +0.009 (0.015) b 0.000001
BMI Marker PCBs
(ng/g fat)
0.135 (0.009) 0.005 (0.002) b 0.000001
Girls with exposure
equal to or below
median
Girls with
exposure above
median
p
Height (cm) Urina ry cadmium
(µg/g creatinine)
1.28 (7.52) +0.05 (0.50) 0.0066
BMI Urinary cadmium
(µg/g creatinine)
13.04 (4.94) 0.06 (0.30) 0.00097
BMI HCB (ng/g fat) 0.348 (0.070) 0.062 (0.027) 0.00016
BMI Marker PCBs
(ng/g fat)
0.139 (0.016) 0.004 (0.003) b 0.000001
a
As indicated under Materials and methods.
b
Calculated as described under Materials and methods.
Table 6
Adults: association of BMI with internal exposure.
Exposure marker Male adults. Multiple regression with confounders
and signicant covariates
a
Male adults. Forward stepwise multiple regression starting with
all confounders, signicant covariates and signicant internal
exposures
b
Change (95%CI) in BMI for a doubling of exposure p Chan ge (95%CI) in BMI for a doubling of exposure p
Urinary cadmium (µg/g creatinine)
c
0.384 ( 0.021/0.746) 0.038 0.210 (+0.105/0.524) 0.19
HCB (ng/g fat)
c
+1.586 (+1.267/+1.906) b 0.000001 +1.642 (+1.260/+2.024) b 0.000001
p,p-DDE (ng/g fat)
c
+0.559 (+0.387/+0.731) b 0.000001 +0.609 (+0.404/+0.815) b 0.000001
PCB 118 (ng/g fat)
c
+0.789 (+0.494/+1.083) b 0.000001 +0.824 (0.463/+1.185) 0.000009
Marker PCBs (ng/g fat)
c
0.198 (+0.255/0.651) 0.39 3.201 ( 2.61/3.79) b 0.000001
Dioxin fraction of dioxin-like activity TEQ (pg/g fat)
c
+0.369 (+0.142/+0.596) 0.0015 +0.273 (+0.071/+0.476) 0.0083
Female adults. Multiple regressi on with confounders
and signicant covariates
d
Female adults. Forward stepwise multiple regression starting
with all confounders, signicant covariates and signicant
internal exposures
e
Urinary cadmium (µg/g creatinine)
c
0.545 ( 0.123/0.967) 0.011 0.367 ( 0.012/0.721) 0.043
HCB (ng/g fat)
c
+1.81 (+1.38/+2.24) b 0.000001 +2.73 (+ 2.29/+3.17) b 0.000001
p,p-DDE (ng/g fat)
c
+0.304 (+0.081/+0.528) 0.0077 +0.674 (+0.445/+0.902) b 0.000001
Marker PCBs (ng/g fat)
c
2.11 ( 1.54/2.69) b 0.000001 6.18 ( 5.48/6.87) b 0.000001
PCB 118 (ng/g fat)
c
+0.45 (+0.06/+0.85) 0.025 + 1.28 (+0.83/+1.72) b 0.000001
Lead (µg/L)
c
0.75 ( 0.31/1.18) 0.00083 0.21 (+0.15/0.56) 0.25
a
Confounders and signicant covariates comprised age, smoking, alcohol consumption, highest education in the family, consumption of vegetables and consumption of meat.
b
Confounders and signicant covariates comprised age, smoking, alcohol consumption, highest education in the family, consumption of vegetables and consumption of meat.
Signicant internal exposures comprised serum concentrations of marker PCBs, PCB 118, HCB and p,p-DDE, dioxin-like activity of the serum dioxin fraction (TEQ) and urinary
concentrations of cadmium and of hydroxypyrene. The resulting model comprised the following parameters: serum concentrations of marker PCBs, PCB 118, HCB and p,p-DDE,
dioxin-like activity of the serum dioxin fraction (TEQ), urinary concentration of cadmium, consumption of alcohol, meat and vegetables and highest education in the family.
c
ln transformed.
d
Confounders and signicant covariates comprised age, alcohol consumption, highest education in the family, consumption of vegetables and consumption of meat. Signicant
internal exposures comprised serum concentrations of marker PCBs, PCB 118, HCB and p,p-DDE, blood concentration of lead and urinary concentration of cadmium.
e
Confounders and signicant covariates comprised age, alcohol consumption, highest education in the family, consumption of vegetables and consumption of meat. Signicant
internal exposures comprised serum concentrations of marker PCBs, PCB 118, HCB and p,p-DDE, blood concentration of lead and urinary concentration of cadmium. The resulting
model comprised the following parameters: serum concentrations of marker PCBs, PCB 118, HCB and p,p-DDE, urinary concentration of cadmium, consumption of alcohol, meat and
vegetables; and highest education in the family.
335W. Dhooge et al. / Environment Internation al 36 (2010) 330337
but a more general toxic effect of lead (Doumouchtsis et al., 2009)
might also play a role.
We have only a limited understanding of how the organochlorines
measured in our study might affect BMI. That different mechanisms
predominate in adults compared to adolescents seems likely. The quite
strong negative associations in adolescents between organochlorines
and BMI observed previously by Nawrot et al. (2002) and, for
organochlorines other than PCB 118, also in this study, are contrasting
with the positive associations with BMI found for HCB, PCB 118 and p,p-
DDE in both adult men and women and, for men, also for dioxin-like
activity. The negative associations observed in adolescents can probably
be explained partly by a transient dilution effect in adipose tissue during
growth as proposed by Nawrot et al. (2002). However, a dilution effect
cannot explain the positive associations observed in adolescents
between PCB 118 and BMI after adjustment for other exposures, nor
why the association between BMI and organochlorine pollutants was in
all cases stronger at exposures below the median than at exposures
above the median. That we observed, per unit of exposure, more
pronounced effects at exposures below the median could be consistent
with the hypothesis that the effects of organochlorines on body size are
hormone- or receptor dependent, as receptor binding substances can be
expected to show MichaelisMenten kinetics and thus relatively
stronger effects at low doses (Castano and Flores-Saaib, 2008). In adults
a higher consumption of animal fat over a long period of time might lead
to higher serum levels of HCB, p,p-DDE, PCB 118 and dioxin-like activity
in persons with a higher BMI. The effects of the endocrine disrupting
pollutants depend probably on the one hand on changes induced in the
production or metabolism of endogenous hormones, resulting in
changed levels in the target tissues, and, on the other, on the direct
interaction of the pollutants with the hormonal receptors. Differing
interactions with aryl hydrocarbon (AhR) and hormonal receptors and
levels of endogenous hormones might be responsible, at least in part, for
the differences observed between the different pollutants in their
association with BMI and for some of the differences between adults and
adolescents as to associations with BMI.
Serum level of PCB 118, a marker for total dioxin-like PCB con-
centration in human serum (Park et al., 2007), showed a positive
association with BMI as well in boys as in girls and in both adult men
and women, and a trend towards a positive relationship with length in
boys. The positive association with height in boys was not affected by
correction for sexual maturation and persisted partly after correction
for levels of testosterone, suggesting that this association was not
solely due to effects on hormone levels and/or sexual maturation. Our
ndings concerning PCB 118 and BMI are consistent with the increase
in BMI reported in several studies to be associated with increasing
internal exposure to dioxins (Collins et al., 2007). Possibly PCB 118
and other dioxin-like PCBs binding on the AHR receptor have a growth
stimulating effect (Vondracek et al., 2005). The dioxin-like PCB 77
induced adipocyte differentiation and promoted obesity and athero-
sclerosis in mice (Arsenescu et al., 2008). Dioxins can show anti-
oestrogenic activity (Kietz et al., 2004), and such an effect might
contribute to a positive association with BMI.
PCBs have xeno-estrogenic activity, mainly through their hydroxyl-
ated metabolites (Waller et al., 1995; Arulmozhiraja et al., 2005), and
can also have estrogenic effects by inhibiting the metabolism of
endogenous oestradiol (Kester et al., 2002). In the boys participating
in this study (Dhooge et al., 2009), serum concentrations of marker PCBs
showed a positive association with serum levels of endogenous sex
hormones. The oestrogenic effects of PCBs might, through stimulation of
lipid mobilization and lipolysis known to be caused by sex hormones
(Hackney et al., 2000), explain the strong negative associations with BMI
for both male and female adolescents, and also for female adults. For
male adults too a negative association was observed between PCB levels
and BMI, that was however only signicant after correction for
concurrent exposures.
In adolescents, the observed association of HCB with higher sex
hormone levels (Dhooge et al., 2009
) might be the dominant effect in
adolescents and might cause the quite strong negative association
between HCB serum level and BMI, whereas the anti-androgenic
characteristics of p,p-DDE (Kelce et al., 1995; Li et al., 2008) and the
anti-androgenic and anti-oestrogenic characteristics of HCB (Li et al.,
2008) might be dominant mechanisms in adults, causing the (for HCB
strong) positive association with BMI. That HCB might in addition
induce a slightly hypothyroidal status (Foster et al., 1993; Meeker et al.,
2007) might contribute to a positive association withBMIas a signicant
negative association between serum HCB concentration and free
triiodothyronine was indeed observed in the adolescents participating
in this study (unpublished results).
5. Conclusion
Environmental exposures to pollutants resulting in normal levels
of internal exposure were associated with small differences in height
of adolescents and, particularly concerning organochlorines, with
quite substantial differences in body mass index of both adolescents
and adults. Associations between biological effects and internal
exposures were, in terms of the regression coefcient, often stronger
at exposures below the median and can probably, at least in part, be
explained by endocrine disrupting effects.
Acknowledgements
We thank Marc Elskens and Tim Nawrot (statistics) and Hans Keune
and Ilse Loots (communication) for advice. We gratefully acknowledge
the collaboration of the 42 Flemish schools, the (psycho-medical)
centres for guidance of pupils, all participating adolescents and their
parents.
Table 7
Adults: strength of association of BMI with internal exposure in function of intensity of
internal exposure.
Exposure marker Multiple regression with inclusion of confounders
a
Regression coefcient (standard error of the mean)
Male adults with
exposure below
median
Male adults with
exposure above
median
p
b
Urinary cadmium
(µg/g creatinine)
-1.45 -0.38 0.35
HCB (ng/g fat) 0.097 0.026 0.000068
p,p-DDE (ng/g fat) 0.00442 0.00003 0.000001
PCB 118 (ng/g fat) 0.117 0.006 0.00033
Marker PCBs (ng/g fat)
c
0.0162 0.0039 0.0031
Dioxin fraction of dioxin-like
activity TEQ (pg/g fat)
0.0548 0.0055 0.0088
Female adults with
exposure below
median
Female adults with
exposure above
median
p
Urinary cadmium
(µg/g creatinine)
0.54 0.60 0.99
HCB (ng/g fat) 0.045 0.029 0.13
p,p-DDE (ng/g fat) 0.00075 0.00008 0.0071
Marker PCBs (ng/g fat) 0.0114 0.0075 0.29
PCB 118 (ng/g fat)
d
0.098 0.055 0.065
Lead (µg/L) 0.084 0.011 0.024
a
As indicated under Materials and methods. Concerning marker PCBs for the men
and PCB 118 for the women signicant concurrent exposures were also included in the
multiple regressions, because of their important impact (see Table 6)onthe
corresponding associations.
b
Calculated as described under Materials and methods.
c
Concurrent exposures included in the multiple regressions were serum concentra-
tions of PCB 118, HCB and p,p-DDE, the dioxin fraction of dioxin-like activity in blood
plasma and urinary concentrations of cadmium and 1-OHP
.
d
Concurrent exposures included in the multiple regressions were serum concentra-
tions of marker PCBs, HCB and p,p-DDE, blood concentration of lead and urinary
concentration of cadmium.
336 W. Dhooge et al. / Environment Internation al 36 (2010) 330337
The study was commissioned, nanced and steered by the
Ministry of the Flemish Community (Department of Science, Depart-
ment of Public Health and Department of Environment), without any
responsibility for the scientic content.
References
Alvarez L, Randi A, Alvarez P, Piroli G, Chamson-Reig A, Lux-Lantos V, et al. Reproductive
effects of hexachlorobenzene in female rats. J Appl Toxicol 2000;20:817.
Arsenescu V, Arsenescu RI, King V, Swanson H, Cassis LA. Polychlorinated biphenyl-77
induces adipocyte differentiation and proinammatory adipokines and promotes
obesity and atherosclerosis. Environ Health Perspect 2008;116:7618.
Arulmozhiraja S, Shiraishi F, Okumura T, Iida M, Takigami H, Edmonds JS, et al. Structural
requirements for the interaction of 91 hydroxylated polychlorinated biphenyls with
estrogen and thyroid hormone receptors. Toxicol Sci 2005;84:4962.
Bilau M, Matthys C, Baeyens W, Bruckers L, De BG, Den HE, et al. Dietary exposure to
dioxin-like compounds in three age groups: results from the Flemish environment
and health study. Chemosphere 2008;70:58492.
Bonefeld-Jorgensen EC, Andersen HR, Rasmussen TH, Vinggaard AM. Effect of highly
bioaccumulated polychlorinated biphenyl congeners on estrogen and androgen
receptor activity. Toxicology 2001;158:14153.
Castano E, Flores-Saaib RD. A mathematical approach for the transactivation of
hERalpha. Philos Transact A Math Phys Eng Sci 2008;366:225363.
Choi H, Jedry chowski W, Spengler J, Camann DE, Whyatt RM, Rauh V, et al. International
studies of prenatal exposure to polycyclic aromatic hydrocarbons and fetal growth.
Environ Health Perspect 2006;114:174450.
Collins JJ, Bodner K, Burns CJ, Budinsky RA, Lamparski LL, Wilken M, et al. Body mass
index and serum chlorinated dibenzo-p-dioxin and dibenzofuran levels. Chemo-
sphere 2007;66:107985.
Covaci A, Voorspoels S, Thomsen C, van BB, Neels H. Evaluation of total lipids using
enzymatic methods for the normalization of persistent organic pollutant levels in
serum. Sci Total Environ 2006;366:3616.
Croes K, Baeyens W, Bruckers L, Den HE, Koppen G, Nelen V, Van De ME, Keune H,
Dhooge W, Schoeters G, Van Larebeke N. Hormone levels and sexual development
in Flemish adolescents residing in areas differing in pollution pressure. Int J Hyg
Environ Health 2009;212:61225.
De Coster S, Koppen G, Bracke M, Schroijen C, Den HE, Nelen V, et al. Pollutant effects on
genotoxic parameters and tumor-associated protein levels in adults: a cross
sectional study. Environ Health 2008;7:26.
Dhooge W., Den Hond E., Koppen G., Bruckers L., Nelen V., Van De Mieroop E., Bilau M.,
Croes K., Baeyens W., Schoeters G., Van Larebeke N. Internal exposure to pollutants
and sex hormone levels in Flemish male adolescents in a cross-sectional study:
associations and doseresponse relationships. Journal of Exposure Science and
Environmental Epidemiology. (advance online publication, 16 December 2009).
doi:10.1038/jes.2009.63.
Doumouchtsis KK, Doumouchtsis SK, Doumouchtsis EK, Perrea DN. The effect of lead
intoxication on endocrine functions. J Endocrinol Invest 2009;32:17583.
Foster WG, Pentick JA, McMahon A, Lecavalier PR. Body distribution and endocrine toxicity
of hexachlorobenzene (HCB) in the female rat. J Appl Toxicol 1993;13:7983.
Foster WG, McMahon A, Younglai EV, Jarrell JF, Lecavalier P. Alterations in circulating
ovarian steroids in hexachlorobenzene-exposed monkeys. Reprod Toxicol 1995;9:
5418.
Garcia-Morales P, Saceda M, Kenney N, Kim N, Salomon DS, Gottardis MM, et al. Effect of
cadmium on estrogen receptor levels and estrogen-induced responses in human
breast cancer cells. J Biol Chem 1994;269:16896901.
Gunnarsson D, Svensson M, Selstam G, Nordberg G. Pronounced induction of testicular
PGF(2 alpha) and suppression of testosterone by cadmium-prevention by zinc.
Toxicology 2004;200:4958.
Hackney AC, Muoio D, Meyer WR. The effect of sex steroid hormones on substrate
oxidation during prolonged submaximal exercise in women. Jpn J Physiol 2000;50:
48994.
Hero M, Norjavaara E, Dunkel L. Inhibition of estrogen biosynthesis with a potent
aromatase inhibitor increases predicted adult height in boys with idiopathic short
stature: a randomized controlled trial. J Clin Endocrinol Metab 2005;90:6396402.
Hilscherova K, Dusek L, Kannan K, Giesy JP, Holoubek I. Evaluation of cytotoxicity,
dioxin-like activity and estrogenicity of complex environmental mixtures. Cent Eur
J Public Health 2000;8 Suppl:289.
Holz JD, Sheu TJ, Drissi H, Matsuzawa M, Zuscik MJ, Puzas JE. Environmental agents
affect skeletal growth and development. Birth Defects Res C Embryo Today
2007;81:4150.
Kelce WR, Stone CR, Laws SC, Gray LE, Kemppainen JA, Wilson EM. Persistent DDT
metabolite p, p-DDE is a potent androgen receptor antagonist. Nature 1995;375:5815.
Kester MH, Bulduk S, van TH, Tibboel D, Meinl W, Glatt H, et al. Potent inhibition of
estrogen sulfotransferase by hydroxylated metabolites of polyhalogenated aro-
matic hydrocarbons reveals alternative mechanism for estrogenic activity of
endocrine disrupters. J Clin Endocrinol Metab 2002;87:114250.
Kietz S, Thomsen JS, Matthews J, Pettersson K, Strom A, Gustafsson JA. The Ah receptor
inhibits estrogen-induced estrogen receptor beta in breast cancer cells. Biochem
Biophys Res Commun 2004;320:7682.
Lafuente A, Marquez N, Pazo D, Esquino AI. Cadmium effects on dopamine turnover
and plasma levels of prolactin, GH and ACTH. J Physiol Biochem 2001;57:2316.
Li J, Li N, Ma M, Giesy JP, Wang Z. In vitro proling of the endocrine disrupting potency
of organochlorine pesticides. Toxicol Lett 2008;183:6571.
Marshall WA, Tanner JM. Variations in pattern of pubertal changes in girls. Arch Dis
Child 1969;44:291303.
Marshall WA, Tanner JM. Variations in the pattern of pubertal changes in boys. Arch Dis
Child 1970;45:1323.
Martin MB, Voeller HJ, Gelmann EP, Lu J, Stoica EG, Hebert EJ, et al. Role of cadmium in the
regulation of AR gene expression and activity. Endocrinology 2002;143:26375.
Martin MB, Reiter R, Pham T, Avellanet YR, Camara J, Lahm M, et al. Estrogen-like
activity of metals in MCF-7 breast cancer cells. Endocrinology 2003;144:242536.
Meeker JD, Altshul L, Hauser R. Serum PCBs, p, p-DDE and HCB predict thyroid
hormone levels in men. Environ Res 2007;104:296304.
Nawrot TS, Staessen JA, Den Hond EM, Koppen G, Schoeters G, Fagard R, et al. Host and
environmental determinants of polychlorinated aromatic hydrocarbons in serum
of adolescents. Environ Health Perspect 2002;110:5839.
Park H, Lee SJ, Kang JH, Chang YS. Congener-specic approach to human PCB
concentrations by serum analysis. Chemosphere 2007;68:1699706.
Schroijen C, Van Wouwe W, Sanctorum H, Goeyens L, Baeyens W. CALUX analysis in
small amounts of human serum. Organohalogen Compd 2006;68:25114.
Schroijen C, Baeyens W, Schoeters G, Den HE, Koppen G, Bruckers L, et al. Internal
exposure to pollutants measured in blood and urine of Flemish adolescents in
function of area of residence. Chemosphere 2008;71:131725.
Selevan SG, Rice DC, Hogan KA, Euling SY, Pfahles-Hutchens A, Bethel J. Blood lead
concentration and delayed puberty in girls. N Engl J Med 2003;348:1527 36.
Sheehan DM, Willingham E, Gaylor D, Bergeron JM, Crews D. No threshold dose for
estradiol-induced sex reversal of turtle embryos: how little is too much? Environ
Health Perspect 1999;107:1559.
Steinberger A, Klinefelter G. Sensitivity of Sertoli and Leydig cells to xenobiotics in in
vitro models. Reprod Toxicol 1993;7(Suppl 1):2337.
Stoica A, Katzenellenbogen BS, Martin MB. Activation of estrogen receptor-alpha by the
heavy metal cadmium. Mol Endocrinol 2000;14:54553.
Tang D, Li TY, Liu JJ, Chen YH, Qu L, Perera F. PAH-DNA adducts in cord blood and fetal
and child development in a Chinese cohort. Environ Health Perspect 2006;114:
1297300.
Van Wouwe N, Eppe G, Xhrouet C, Windal I, Vanderperren H, Debacker N, et al.
Validation of the CALUX bioassay for PCDD/F analyses in human blood plasma and
comparison with GC-HRMS. Talanta 2004;63:115767.
Vondracek J, Machala M, Bryja V, Chramostova K, Krcmar P, Dietrich C, et al. Aryl
hydrocarbon receptor-activating polychlorinated biphenyls and their hydroxylated
metabolites induce cell proliferation in contact-inhibited rat liver epithelial cells.
Toxicol Sci 2005;83:5363.
Waller CL, Minor DL, McKinney JD. Using three-dimensional quantitative structure
activity relationships to examine estrogen receptor binding afnities of poly-
chlorinated hydroxybiphenyls. Environ Health Perspect 1995;103:7027.
337W. Dhooge et al. / Environment Internation al 36 (2010) 330337
    • "Although their use was banned in most countries several decades ago, DDTs and PCBs are still found at considerable levels in human adipose tissue due to their earlier widespread use and persistency [1] [2] [3]. Studies have reported an association between obesity and plasma levels of certain PCBs and pesticides [4] [5] [6] [7], suggesting a possible relationship between POP exposure and the current obesity epidemic as well as type 2 diabetes [7] [8] [9] [10]. However, a causal relationship between POP exposure and obesity development has not yet been demonstrated, and conflicting data have been reported [11] [12] [13] [14]. "
    [Show abstract] [Hide abstract] ABSTRACT: Accumulation of persistent organic pollutants (POPs) has been linked to adipose tissue expansion. As different nutrients modulate adipose tissue development, we investigated the influence of dietary composition on POP accumulation, obesity development and related disorders. Lifespan was determined in mice fed fish-oil-based high fat diets during a long-term feeding trial and accumulation of POPs was measured after 3, 6 and 18months of feeding. Further, we performed dose-response experiments using four abundant POPs found in marine sources, PCB-153, PCB-138, PCB-118 and pp'-DDE as single congeners or as mixtures in combination with different diets: one low fat diet and two high fat diets with different protein:sucrose ratios. We measured accumulation of POPs in adipose tissue and liver and determined obesity development, glucose tolerance, insulin sensitivity and hepatic expression of genes involved in metabolism of xenobiotics. Compared with mice fed diets with a low protein:sucrose ratio, mice fed diets with a high protein:sucrose ratio had significantly lower total burden of POPs in adipose tissue, were protected from obesity development and exhibited enhanced hepatic expression of genes involved in metabolism and elimination of xenobiotics. Exposure to POPs, either as single compounds or mixtures, had no effect on obesity development, glucose tolerance or insulin sensitivity. In conclusion, this study demonstrates that the dietary composition of macronutrients profoundly modulates POP accumulation in adipose tissues adding an additional parameter to be included in future studies. Our results indicate that alterations in macronutrient composition might be an additional route for reducing total body burden of POPs.
    Full-text · Article · Oct 2015
    • "The Flemish population is therefore possibly exposed to a wide variety of pollutants. In the first Flemish Environment and Health Study (FLEHS I), conducted between 2002 and 2006, mean values were established for a wide range of biomarkers, including cadmium (Cd) and lead (Pb), and the collected data were related to health and area of living (Bilau et al., 2008; Dhooge et al., 2010; Keune et al., 2008; Koppen et al., 2009; Schroijen et al., 2008 ). This work presents the results of trace elements (essential ones such as Mn, Cu and Cr and toxic metals or metalloids) of the second study conducted by the Flemish Centre for Environment and Health (FLEHS II) between 2007 and 2011. "
    [Show abstract] [Hide abstract] ABSTRACT: The Flemish Centre for Environment and Health started with human biomonitoring in 2002 (FLEHS I: 2002-2006). The main goal of the second human biomonitoring cycle (FLEHS II: 2007-2011), was to determine mean values for a large number of pollutants in a representative sample of the general Flemish population. Values for Cd and Pb were updated, and a group of previously undetermined metals and metalloids (As, Mn, Cu and Tl) were included in some of the age groups. In this human biomonitoring program, three different age groups of the general Flemish population were monitored: 255 newborns and their mothers, 210 adolescents aged 14-15, and 204 adults between 20 and 40 years old. Trace elements were determined in cord blood and maternal blood of the mothers, in blood and urine of adolescents and in urine of adults. Determinants of life-style and personal factors were taken into account. The levels of trace elements in cord blood and maternal blood were for most elements at the lower end of the range found in literature. For Pb, As and Tl, a strong correlation (respectively r = 0.43, 0.55 and 0.33; p < 0.05) was found between levels in cord blood (respectively 8.6, 0.54 and 0.017 μg/L) and maternal blood (11.1, 0.64 and 0.028 μg/L), indicating that they are transported via the placenta from mother to fetus. The levels found in the adolescents and adults were compared with results from international biomonitoring studies, and were found to be in the same ranges. With the exception of Pb, all trace elements increased with increasing age group population. Finally, the results also showed that the levels of Cd and Pb in blood for this campaign (e.g. for Pb 8.6 and 14.8 μg/L in neonates and adolescents respectively) were lower compared to the first campaign (e.g. for Pb 14.7 and 21.7 μg/L in neonates and adolescents respectively), indicating a decrease over time. However, differences in sampling strategies might partially explain this observed trend.
    Full-text · Article · Nov 2014
    • "A similar pattern was observed with our data, with PCB concentrations decreasing with increasing BMI (Table 2a). Accumulation in body fat deposits in those subjects with a higher BMI is one of the likely causes for this negative association (Dhooge et al., 2010). One of our critical concerns is to determine how much human serum PCB levels have decreased since the ban. "
    [Show abstract] [Hide abstract] ABSTRACT: This manuscript presents the levels of six indicator polychlorinated biphenyl (PCB) congeners (IUPAC nos. 28, 52, 101, 138, 153 and 180) in the serum of 1880 individuals from a representative sample of the Spanish working population recruited between March 2009 and July 2010. Three out of the six PCBs studied (180, 153 and 138) were quantified in more than 99% of participants. PCB 180 was the highest contributor, followed by PCBs 153 and 138, with relative abundances of 42.6%, 33.2% and 24.2%, respectively. In contrast, PCBs 28 and 52 were detected in only 1% of samples, whereas PCB 101 was detectable in 6% of samples. The geometric mean (GM) for ΣPCBs138/153/180 was 135.4 ng/g lipid (95% CI: 121.3–151.2 ng/g lipid) and the 95th percentile was 482.2 ng/g lipid. Men had higher PCB blood concentrations than women (GMs 138.9 and 129.9 ng/g lipid respectively). As expected, serum PCB levels increased with age and frequency of fish consumption, particularly in those participants younger than 30 years of age. The highest levels we found were for participants from the Basque Country, whereas the lowest concentrations were found for those from the Canary Islands. The Spanish population studied herein had similar levels to those found previously in Greece and southern Italy, lower levels than those in France and central Europe, and higher PCB levels than those in the USA, Canada and New Zealand. This paper provides the first baseline information regarding PCB exposure in the Spanish adult population on a national scale. The results will allow us to establish reference levels, follow temporal trends and identify high-exposure groups, as well as monitor implementation of the Stockholm Convention in Spain.
    Full-text · Article · Sep 2014
Show more