Daily intake of bisphenol A and triclosan and their association with anthropometric data, thyroid hormones and weight loss in overweight and obese individuals

Environment International 76 (2015) 98–105

Contents lists available at ScienceDirect

Environment International

j ourna l homepage: www.e lsev ie r .com/ locate /env int

Full length article

Daily intake of bisphenol A and triclosan and their association withanthropometric data, thyroid hormones and weight loss in overweightand obese individuals

Tinne Geens a,1, Alin C. Dirtu a,1, Eveline Dirinck b, Govindan Malarvannan a, Luc Van Gaal b,Philippe G. Jorens c, Adrian Covaci a,⁎a Toxicological Centre, University of Antwerp, 2610 Wilrijk, Antwerp, Belgiumb Department of Endocrinology, Diabetology and Metabolic Disease, University of Antwerp and Antwerp University Hospital, 2630 Edegem, Antwerp, Belgiumc Department of Clinical Pharmacology, University of Antwerp and Antwerp University Hospital, 2630 Edegem, Antwerp, Belgium

⁎ Corresponding author at: Toxicological Center, Universi1, 2610 Wilrijk, Belgium.

E-mail address: [email protected] (A. Cov1 Equal contribution as the first author.

http://dx.doi.org/10.1016/j.envint.2014.12.0030160-4120/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Article history:Received 24 July 2014Accepted 8 December 2014Available online xxxx

Keywords:Bisphenol ATriclosanUrineObesityWeight lossBelgium

Bisphenol A (BPA) and triclosan (TCS) were determined in urine of Belgian overweight and obese (n = 151) andlean (n = 43) individuals. After the first urine collection (0 M), obese patients started a diet program or haveundergone bariatric surgery. Hereafter, three additional urine samples from obese patients were collected after3 (3 M), 6 (6 M) and 12 (12 M) months. Both compounds were detected in N99% of the samples. BPA had medianconcentrations of 1.7 and 1.2 ng/mL in obese and lean groups, respectively, while TCS had median concentrationsof 1.5 and 0.9 ng/mL in the obese and lean groups, respectively. The obese group had higher urinary concentrations(ng/mL) of BPA (p b 0.5),while no significant differenceswere found for TCS between the obese and lean groups. Notime trends between the different collection moments were observed. The BPA concentrations in the obese groupwerenegatively associatedwith age,while no gender difference or relationshipwithbodymass indexwas observed.For TCS, no relationshipswith gender, BMI, or agewere found. The temporal variability of BPA and TCSwas assessedwith calculation of the intraclass correlation coefficient, Spearman rank correlation coefficients, and surrogatecategory analysis. We observed evidence that single spot urine samples might be predictive of exposure over alonger period of time. Dietary intakes of BPA and TCS did not differ significantly among the time points consideredafter obese individuals started losing weight (6 and 12 months). Multiple linear regression analyses after adjustingfor age andweight loss revealednegative associations betweenurinary TCS and serumFT4 in the0M and3M femaleobese individuals and positive associations between urinary BPA and serum TSH in the lean group.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Due to exponential use of synthetic chemicals combinedwith delayedactions taken for their regulation, advances in exposure sciences com-binedwith improved analyticalmethodologies evidenced thewidespreadhuman and animal exposure to such chemicals (Meeker, 2012). Apartfrom these, some compounds, such as bisphenol A (BPA) or triclosan(TCS), are capable of disrupting endocrine functions through mimickingor blocking endogenous hormones (Fenichel et al., 2013; Koeppe et al.,2013; Schug et al., 2011). Alterations in thyroid and steroid hormonefunction and activation of peroxisome proliferator-activated receptorsmight occur due to exposure to such endocrine disrupting chemicals(EDCs) influencing the adipocyte differentiation a'nd energy storage(Tang-Péronard et al., 2011; Thayer et al., 2012). Additionally, the

ty of Antwerp, Universiteitsplein

aci).

presence of such synthetic chemicals in humans has been associatedwith elevated triglycerides and cholesterol, impaired fasting glucose anddiabetes (Meeker, 2012; Tang-Péronard et al., 2011), factors that are allrelated to the body's natural weight controlmechanisms potentially lead-ing to obesity. Exposure to EDCs, such as BPA and phthalates, has alreadybeen linked to obesity in animal studies (Miyawaki et al., 2007; R.R.Newbold et al., 2007). Such compounds are widely used as plasticizersand stabilizers in the manufacture of consumer products includingchildren's toys and food-packaging materials (Koch and Calafat, 2009).As a consequence, human exposure to EDCs might take place throughmultiple pathways, but ingestion of contaminated food or water, inhala-tion of polluted air or dermal exposure is the most relevant (CDC, 2009).

In parallel with the increase in the production and use of several EDCs,a clear increase of the child and adult obesity was documented, especiallyfor the developed countries (Baillie-Hamilton, 2002). Since commoncauses, such as over-eating, inactivity and genetic pre-disposition, donot fully explain the current obesity epidemic (Tang-Péronard et al.,2011; Thayer et al., 2012), it was hypothesized that human exposureto EDCs might play an important role in obesity development. As a

99T. Geens et al. / Environment International 76 (2015) 98–105

consequence, several studies tried to link human exposure to specificEDCs, such as BPA, to obesity indicators, including glucose metabolism,type 2 diabetes or cardiovascular diseases (Lakind et al., 2014; Shankaret al., 2012; Thayer et al., 2012; Vom Saal et al., 2012).

Since EDCs might interfere with the programming of endocrine-signaling pathways established during vulnerable periods of life (R.Newbold et al., 2007), particular attentionwas given to associations be-tween BPA exposure in early life stages (fetal and/or childhood) andobesity indicators (Harley et al., 2013; Li et al., 2013; Valvi et al., 2013;Wells et al., 2014). Therefore, it was shown that prenatal BPA exposurewas positively associatedwith increasedwaist circumference (WC) andbody mass index (BMI) for 4 years old children, although not in infancy(Valvi et al., 2013). However, another study reported positive associa-tions only between increased postnatal BPA exposure at 9 years old chil-dren and obesity indicators (BMI, WC, fat mass), while increasing BPAconcentrations in mothers during pregnancy were associated with in-creased BMI, body fat, and overweight/obesity among their daughtersat 9 years of age (Harley et al., 2013). Significant associations betweenincreasing urinary BPA levels and another obesity indicator, waist toheight ratio (WHR), among a nationally representative sample of chil-dren (more than 2800 participants) were recently reported for USA(Wells et al., 2014).

Associations between BPA exposure, obesity, and related outcomeswere studied also for adults (Carwile and Michels, 2011; Gallowayet al., 2010; Shankar et al., 2012; Wang et al., 2012). Some studiesreported that higher BPA exposure, measured through urinary BPAconcentrations, is associated with general and central obesity in theadult population, but reverse causation is of concern due to the cross-sectional nature of such studies. Nearly all research on BPA and obesityindicators used a cross-sectional design and relied mostly on a singlemeasure of BPA exposure, while this may result in serious exposuremisclassification (Lakind et al., 2014). However, for all obesity out-comes, results across studies were inconsistent. Even studies that usedthe same data, but different statistical methods, produced different re-sults (Lakind et al., 2014).

A link between human exposure to BPA and TCS and obesity indica-tors was mostly investigated for BPA, while studies on TCS tried to evi-dence its influence on the ED system and especially on the thyroidhormones (Koeppe et al., 2013). Previous studies suggested that BPAand TCS might accumulate in adipose tissue, although to a lesser extentwhen compared to neutral persistent organic pollutants, and mightlater be released into the human fluids generating additional exposure(Christensen et al., 2012). Therefore, studies evaluating their levels inpersons undergoing weight loss programs became important. Due totheir fast metabolism, evaluation of the human exposure to suchchemicals is commonly assessed through the urinary measurement offree excreted forms or conjugated metabolites.

In the light of the above discussion, the aims of the currentmanuscriptwere to: 1) compare the BPA and TCS urinary levels in obese and leanindividuals sampled in Belgium; 2) evaluate the dynamics of urinaryBPA and TCS levels in obese individuals during one year of weight loss;3) estimate the BPA and TCS daily intake (DI) and investigate intakedifferences during weight loss based on accepted intake scenarios andexcretion fractions reported for these compounds in humans; 4) relateBPA and TCS urinary levels to anthropometric data and serum thyroidhormones; and 5) evaluate variations in exposure sources depending onthe treatment method applied for weight loss in the obese individuals.

2. Materials and methods

2.1. Sample collection and target analytes

The Endorup trial, conducted at theweightmanagement clinic of theDepartment of Endocrinology, Diabetology and Metabolism of theAntwerp University Hospital, is a prospective study to unravel the hy-pothesis of endocrine disruption by organic contaminants in obesity

(registered at clinicaltrials.gov number NCT01778868). Previously, wehave reported data on persistent organic pollutants from serum and ad-ipose tissue (Dirinck et al., 2014; Dirtu et al., 2013a; Malarvannan et al.,2013), as well as non-persistent chemicals such as phthalate metabo-lites from urine samples collected from obese and lean participantsincluded in this study (Dirtu et al., 2013b). In this manuscript, we pro-vide data on the urinary levels of BPA and TCS, as well as estimationsof the daily intake of these compounds in a cohort of 151 overweightand obese men and women who were prospectively selected from pa-tients visiting the weight management clinic between November 2009and February 2012 for their first visit. All participants were older than18 years of age. Anthropometric measures were taken in the morningwith patients in a fasting state and undressed. Height was measuredto the nearest 0.5 cm and body weight was measured with a digitalscale (with a precision of 0.2 kg). The subject characteristics are de-scribed in Table 1. Urine samples were taken before starting weightloss program (0 M, N = 151) and after three (3 M, N = 95), six (6 M,N = 53) and twelve months (12 M, N = 39). For comparison purposes,a group of 43 lean men and women, matched by age and sex, was re-cruited from hospital staff and volunteers during the same period. Themedian age of the lean population (43 y) was not significantly differentfrom the obese population (41 y) (p N 0.37). There was also no differ-ence in gender composition of the two groups (p N 0.45). The subjectcharacteristics are described in Table 1.

The 24-h urine collections were available (except for the 3 M obesepatients and lean population which could not be kept in hospital for24 h). These were obtained as the following: the 24-h urine collectionstarted with the second void on the day prior to the in-hospital investi-gation and ended with the first void on the day of the in-hospital inves-tigation. For safety reasons, the study participants were not allowed tocollect the urine in glass containers, while BPA-free plastic containerswere used for urine collection. After collection, the 24-h urine wasstored in glass containers. For the 3 M obese and controls, we haveused the second void of the day; this was the day of the in-hospital in-vestigations for 3 M patients. This urine was collected in glass con-tainers, under the supervision of a trained nurse and was further usedfor the BPA and TCS measurements.

On the day of the in-hospital investigations, patients came to thehospital in a fasting state. Beingmonitored in hospital for 24 h, our pop-ulation was proven to have a stable kidney function and therefore thecreatinine clearance is expected to be stable from one day to another.Urinary creatinine concentration was measured using an enzymaticmethod to correct for the urine dilution (Cocker et al., 2011). However,as stated above, urinary creatinine data were available only for baseline(0 M), and for 6 and 12months (6 M and 12M) obese and therefore wecould normalize BPA and TCS concentrations to creatinine levels onlyfor these time points (Table 2).

Serum Thyroid Stimulating Hormone (TSH) and free Thyroxine(fT4) were measured for all participants to the study including thetime of their enrolling (0 M), but also the later time points (3 M, 6 M,12 M) and also for the lean individuals. The analysis method of eachhormone involved a chemiluminescence LOCI method (Siemens,DimensionVista) at theAntwerpUniversityHospital laboratory. Venousblood sampleswere obtained in the fasting state between 08.00 AMand10.00 AM into sterile BD Vacutainer tubes. This study was approved bythe Ethical Committee of the Antwerp University Hospital (BelgianRegistry number B30020097009) and all participants provided theirwritten informed consent.

2.2. Sample preparation and analysis

Full description of the method optimization, development and vali-dation for the analysis of the total BPA and TCS in urine was previouslydescribed (Geens et al., 2009). In brief, a volume of 3 mL of urine wasspiked with 2.5 ng of 13C-labeled BPA and TCS (Cambridge Isotope Lab-oratories, Andover, MA, USA), 50 μL of β-glucuronidase/sulfatase (from

Table 1Characteristics of the sampled population (median togetherwith range interval) at different time points ofweight loss in overweight and obese individuals (0M to 12M) togetherwith thelean group from Belgium.

No. of participants Men/women Age (y) Weight (kg) BMI (kg/m2) Creatinine (mg/day) TSH (mIU/L) fT4 (pmol/L)

0 M (baseline overweight and obese) Median 151 46/105 41 109.8 38.5 1441 1.505 13.4Min 18 73 26.2 680 0.01 9.3Max 84 197.4 62.3 3565 14.63 20

3 M (after 3 months) Median 95 31/64 40 98.6 34.2 – 1.470 12.9Min 18 64.8 24.2 – 0.03 9.9Max 84 143.2 48.1 – 10.37 17.7

6 M (after 6 months) Median 53 15/37 44 91.8 34.3 1265 1.145 12.9Min 18 63.8 23.6 625 0.01 10.1Max 70 134.8 48.9 2400 4.21 20.7

12 M (after 12 months) Median 39 11/28 46 92.0 31.7 1226 1.165 12.45Min 21 57 20.3 667 0.02 9.7Max 60 135 48.1 2030 3.08 17.9

Lean Median 43 13/30 43 61.4 21.9 – 1.745 12.9Min 19 42.8 17.5 – 0.46 10.3Max 59 87.4 25.3 – 8.1 18.7

Differences between 0 M overweight and obese individuals and lean group (p) p N 0.05 p b 0.01 p b 0.01 – p N 0.05 p N 0.05

100 T. Geens et al. / Environment International 76 (2015) 98–105

Helix Pomatia, Merck, Darmstadt, Germany) and 750 μL of 1 M Na-acetate (pH = 4.5). After vortexing, the mixture was held at 40 °C for30 min for analyte deconjugation. Later, 200 μL of formic acid (Merck,Darmstadt, Germany) was added followed by 15 min sonication. OasisHLB columns (60mg/3mL,Waters,Milford, MA, USA)were used for ex-traction and analyteswere elutedwith amixture of 5mLMeOH–dichlo-romethane (1:1, v/v). After evaporation of the extract, 50 μL of 2MKOH,50 μL pentafluorobenzoylchloride (PFB-Cl, 99% purity, Sigma-Aldrich—

used here as 5% in hexane, v/v), 1 mL MilliQ water and 2 mL hexanewere added before applying an extractive derivatization. The mixturewas vortexed for 2 min and, after phase-separation, the hexane layerwas transferred. After the addition of 2 mL fresh hexane, the later oper-ation was repeated for a quantitative transfer of the analytes. The com-bined hexane layers were evaporated to dryness into a N2 stream, andthe analytes were reconstituted in 100 μL of iso-octane and furtherinjected into a GC–MS system (Agilent 6890 GC coupled to a 5973mass spectrometer, operated in ECNI mode). Details concerning the

Table 2Overview of the BPA and TCS concentrations (median together with 10th–90th and 25th–75thcollected from obese individuals entering a weight loss program (0 M) and after 3 (3 M), 6 (population fromBelgium.Data for bariatric surgery and diet and physical exercises following obbold* shows significant differences (p b 0.05).

BPA (ng/mL)

All samples Bariatric surgery Diet

0 M (baseline overweight and obese) Median 1.7 – –

10th 0.7 – –

90th 5.1 – –

25th 1.1 – –

75th 2.8 – –

3 M (after 3 months) Median 2.0 2.0 2.010th 0.7 1.2 0.690th 5.3 5.4 4.125th 1.4 1.4 1.275th 3.2 3.2 3.3

6 M (after 6 months) Median 1.7 1.7 1.810th 0.8 0.8 0.890th 3.7 3.3 4.425th 1.0 1.0 1.275th 2.8 2.5 3.3

12 M (after 12 months) Median 1.5 1.6 1.510th 0.9 0.9 1.090th 4.9 4.4 5.125th 1.1 1.1 1.375th 2.9 2.1 3.3

Lean Median 1.2 – –

10th 0.6 – –

90th 2.9 – –

25th 0.9 – –

75th 1.9 – –

targeted analytes together with additional information concerningtheir analysis (e.g. internal standards used for their quantification, ionsused for the analyte quantification/qualification, instrumental tech-nique) are given in Table SI.1. Additional information on the instrumen-tal conditions employed for the analysis of BPA and TCS is given inTable SI.2. All solvents used in analyses (n-hexane, MeOH, dichloro-methane, iso-octane) were of gradient grade for liquid chromatographyand they were obtained from Merck.

2.3. Quality assurance/quality control (QA/QC)

The internal QA/QCwas performed by regular analyses of proceduralblanks and blind sample duplicates, analysis of pooled urine samples,and random injection of standards, spiked samples and solvent blanks.The external quality control scheme was also assessed through partici-pation to inter-laboratory comparison exercises, as follows: AMAP(2010) for total BPA in urine (obtained values were deviating with

percentile intervals, expressed in ng/mL and μg/g creatinine) measured in urine samples6 M) and 12 months (12 M) after enrolling the program together with results for a leanese population are also presented separately for comparisonpurposes; data evidencedwith

TCS (ng/mL) BPA (μg/g creatinine) TCS (μg/g creatinine)

All samples Bariatric surgery Diet

1.5 – – 1.7 1.40.3 – – 0.5 0.3

73 – – 4.7 520.6 – – 0.9 0.63.7 – – 2.9 4.31.3 2.4* 1.0* – –

0.3 0.4 0.3 – –

110 135.2 64.8 – –

0.7 0.8 0.6 – –

6.2 10.7 2.9 – –

1.9 1.5 2.2 1.9 2.30.3 0.3 0.3 0.8 0.3

205 5.0 291 4.4 2120.7 0.45 0.7 1.0 0.67.9 3.49 12.4 3.1 11.81.9 2.7 1.1 2.1 3.80.3 0.3 0.4 0.8 0.4

250 362 52 4.7 2800.7 0.8 0.7 1.2 0.6

15.2 121 3.7 3.0 17.90.9 – – – –

0.3 – – – –

52 – – – –

0.6 – – – –

3.1 – – – –

101T. Geens et al. / Environment International 76 (2015) 98–105

less than 20% from the consensus values), several inter-laboratory com-parison exercises within the framework of the European DEMOCOPHESfor total BPA in urine (all resultswerewithin±1× SDof themean valueof the participating reference laboratories), and inter-laboratoryexercises GERMAN EQUAS 2011 and 2013 (for the analysis of totalBPA in urine) when the obtained values werewithin the accepted refer-ence interval (G-EQUAS, 2011, 2013).

2.4. Daily intake calculations

Since excretion of ingested BPA into urine is essentially complete in24h (Völkel et al., 2002; Völkel et al., 2005), we assumed that in order toestimate the total daily BPA and TCS intake (DIBPA,TCS expressed inμg/kg bw/24 h), the following formula will apply:

DIBPA;TCS

μgkg

bw

24 h

0B@

1CA ¼ UCBPA;TCS

μgL

� �� UV Lð ÞBW kgð Þ

where UCBPA,TCS (μg/L) represents the urinary concentrations of totalBPA or TCS (free and conjugated after hydrolysis of the conjugates),UV is the measured volume (L) of the 24 h urine collection and BW(kg) represents the body weight of the sampled individual. Each indi-vidual stayed in the hospital for 24 h at each sampling time point. How-ever, due to logistics issues, obese individuals for 3 M and also the leangroup could not be kept in the hospital for 24 h and therefore the DIdata for these categories were not calculated in our study.

2.5. Statistical analysis

All statistical analyses were performed using XLStat-Pro version2013.3.03 (Addinsoft, 2012). Levels below themethod LOQwere assigneda value of DF × LOQ, with DF being the proportion of measurements withlevels above the LOQ or the detection frequency (Voorspoels et al., 2002).Data were not normally distributed (Shapiro–Wilk test, p N 0.05) andtherefore they were log-transformed (y = log(x + 1)). For non-normally distributed data after log-transformation, differences betweenmultiple groups of samples (different age categories, sampling timepoints) were assessed using the Kruskal–Wallis test, while the Mann–Whitney U test was used for differences in concentrations betweengroups (gender, treatment method influence or different time points).Correlations were carried out using parametric Pearson correlation onthe log-transformed data after removal of outliers (Grubbs' test). The sig-nificance level was set at α= 0.05 throughout the study.

The intraclass correlation coefficient (ICC) was defined as the per-cent of total variance explained by between-person variance (Rosner,2000). The ICC was calculated from a one-way random-effects ANOVAmodel estimated independently for each analyte. An ICC of at least0.40 is considered as an indication of fair to good reproducibility andan ICC of 0.75 or greater is considered as excellent (Rosner, 2000).

The surrogate category analysis was conducted for each analyteusing data from 21 obese individuals who had urine samples collectedat all of the following time points: baseline (0 M); 3 M, 6 M and 12 Mmonths after baseline. The “surrogate sample” was defined as the setof 21 samples collected at one of the four collection time points, thatis, baseline. The 12-month mean concentration (mean of the four re-peated measures over the 12-month period) was calculated to repre-sent the average analyte exposure for this time period. We calculatedtertile cut-off points using the analyte concentration distribution ofthe 21 baseline samples (one per person). These cut-off points wereused to classify each obese individual into low (0–33rd percentiles),medium (34th–67th percentiles) or high (N67th percentile) concentra-tion groups. We then computed three tertile-specific 12-month grandmeans by averaging the 12-month mean concentrations of the individ-uals who were assigned to each tertile (N = 7). All calculations were

performed on the log-transformed data (log(x + 1)) and were thenback transformed to obtain mean concentrations.

3. Results and discussion

An overview of the median concentrations (expressed in ng/mL ofurine) together with 10th–90th and 25th–75th percentile intervals,expressed in ng/mL and μg/g creatinine for BPA and TCS measured inobese individuals at each sampled time point (0 M to 12M) and in con-trol samples are presented in Table 2.

3.1. Urinary levels of BPA and TCS

Both BPA and TCS could be detected in almost all samples withDF N99%. Additionally, a high variability in the levels of BPA and TCSwas seen within the sampled groups of obese and lean individuals, withTCS concentrations ranging between 2 and 3 orders of magnitude. Com-bined with the high DFs, this suggests a wide range of exposure sourcesto BPA and TCS. Yet, BPA concentrations covered a much smaller rangecompared to TCS levels. While BPA concentrations were in the ng/mLrange, concentrations up to the μg/mL range could be observed for TCS.However, the largest fraction (40%) of TCS and BPA concentrations wasfound between 0 and 1 ng/mL and between 1 and 2 ng/mL, respectively.As a consequence, comparisons between data at different time pointswere difficult to generate significant differences. Significant differences(p b 0.05) were obtained only for BPA concentrations (ng/mL) creatinineunadjusted between samples collected from 0 M obese compared to leansamples, between 0 M and 3 M obese and also between 12 M obese andlean samples (significance data for all tested comparisons are given inTable SI.3). Furthermore, no clear differences could be evidenced for TCSin any of the comparisons performed for both unadjusted and creatinineadjusted levels (Table SI.3).

Previously, it was shown that urinary BPA levels are significantlydecreasing during fasting, thus evidencing the importance of the dietto the human BPA intake (Christensen et al., 2012). A possible explana-tion for the higher BPA levels in the obese population at 0 M comparedto the lean group may be thus the difference in food intake, both interms of the amount of food and the composition of the diet (e.g. contri-bution of canned food) (Christensen et al., 2012; Geens et al, 2010).AlthoughCarwile andMichels (2011) suggested that the difference in cal-ories intake between normalweight, overweight, and obese individuals isminimal, the composition of the diet would only explain our results be-tween the obese and the lean group. Additionally, if the amount of foodintake would be responsible for the difference observed between obeseand lean individuals in the present study, a decline in the BPA concentra-tions would be expected between 0 M and 3 M obese individuals, and es-pecially in the group of obese following weight loss treatment throughcontrolled diet. Although non-significant, an increase in BPA concentra-tion was yet observed between the 0 M and 3 M obese individuals, bothfor the diet controlled as for the surgery groups (Table 2). Previously,we reported that high consumption of canned food or beverages mightsignificantly contribute to daily intake of BPA due to the presence ofBPA-based epoxy resins as coating agents of canned food (Geens et al,2010). It is thus more likely that differences in diet composition betweeneach BMI categories, specifically the proportion of the diet derived fromcanned food and beverages, will account for differences in the urinaryBPA levels. Another hypothesis suggested that small reservoirs of BPAfrom past exposures are released from storage (e.g. adipose tissues) andexcreted during weight loss (Christensen et al., 2012). Although a signif-icant weight loss was recorded for the obese individuals over the 12months of our study, no clear trend could be evidenced neither for theBPA or TCS urinary levels. Supplementary data on the consumption ofcanned food by the populations included in our studywould be necessaryin order to make further conclusions.

Since its exposure routes are not necessarily food related, the TCSbehavior observed in our study for the time trends and the differences

102 T. Geens et al. / Environment International 76 (2015) 98–105

between obese and lean individuals may be easier to explain by linkingto TCS sources, such as cosmetics, personal care and cleaning products(NICNAS, 2009). Although the use of cosmetics might expect to differamong exposed populations depending on the gender, in our study nosignificant gender differences could be observed for the urinary TCSlevels (p N 0.05). The use of such consumer products seems to lead tonon-specific differences in the urinary TCS levels between the variousgroups of individuals included in our study.

The median urinary BPA concentrations (with and without creati-nine normalization) measured in our study are comparable to thosefound in other previous studies, although most of the literature reportson samples collected from general population and not necessarily fromobese individuals (Koch et al., 2012). When referring to studies whichhave investigated the relationships between BPA concentrations and obe-sity indicators, the reported results seem to be inconsistent (Carwile andMichels, 2011; Galloway et al., 2010; Lakind et al., 2014; Shankar et al.,2012; Wang et al., 2012). No association was reported between urinaryBPA and overweight or obesity in a cross-sectional analysis of Italianadults [Galloway et al., 2010] or an analysis of 2003–2004 NHANES(National Health and Nutrition Examination Survey) biomonitoringdata of the US general population (Lang et al., 2008). However, a cross-sectional analysis of the 2003–2006 NHANES data revealed that, relativeto participants with a urinary concentration in the lowest BPA quartile,participants in the upper three BPA quartiles had similar prevalenceodds of obesity, with odds ranging from 1.60 to 1.85 (Silver et al., 2011).In addition, participants in the upper three BPA quartiles had non-significantly increased prevalence odds of overweight relative toparticipants in the lowest BPA quartile. The authorswarned that reversecausation is of concern due to the cross-sectional nature of the study(Carwile and Michels, 2011). Applying similar statistical approach,positive associations between increasing levels of urinary BPA andBMI and WC, independent of potential confounding factors includ-ing, smoking, alcohol consumption, and serum cholesterol levels,were observed when evaluating data of the 2003–2008 NHANES(Shankar et al., 2012).

Although we could not find clear differences for TCS urinary levelsbetween 0 M obese and lean individuals who participated in ourstudy, the results of the present work might be important to be addedto the international database on the human exposure to multiplechemicals and its possible relation to obesity. Since other studies haveshown that through including the different sets of samples within thesame population, the output could be different (Lakind et al., 2014);such comparisons might clarify the needs which have to be consideredon how to appropriately adjust the analyses for key confounding factors(eg. dietary intake) (Magliano and Lyons, 2013).

3.2. Daily intake estimations

There were no significant differences of the estimated DIs for BPAand TCS (ng/kg bodyweight/day, Table SI.4) at the different time pointsfor obese individuals. All comparison tests between samples wereperformed considering the estimated DI values expressed both perkg body weight/day, but also per total body weight/day in order toalso evaluate the total DI values of BPA and TCS independently of the in-dividualweight (Tables SI.5 and SI.6). Although significantweight losseswere recorded in time, it was expected that a decrease in the diet wouldalso lead to a decrease in the intake of BPA. After differentiating theobese individuals based on theweight loss method (bariatric surgeryor diet-controlled), no significant change in the DI for BPA after thisintervention could be evidenced. Since TCS intake is not necessarilyinfluenced by alterations in diet, no changes between estimatedDI for TCS between 0 M and 3 M obese individuals were expectedwhen the losing weight method was considered as a comparisonvariable. Indeed, results showed that no significant differenceswere obtained between considered samples: diet group p N 0.5 andsurgery group p N 0.4.

By comparing the estimated DIs for BPA (ng/kg body weight/day)obtained in this study for 0 M, 6 M and 12 M obese individuals withpreviously available data for Belgium or other European countries(Geens et al., 2011, 2012), it was noticed that our estimations forobese individuals (20–25 ng/kg body weight/day) are at the lowerliterature reported limit intervals estimated for general lean population(15–60 ng/kg body weight/day). However, previously reported datawere derived from urinary BPA levels and considering an estimatedurinary output of 1500 mL/24 h and a body weight of 60 kg, while ourdata are based on real excreted urinary volume and individual weightsfor each participant to the study.

3.3. Variations in the BPA and TCS concentrations in overweight and obeseindividuals during weight loss

The temporal variability was assessed based only on the obese pa-tients who donated a urine sample on all four time points. To assessthe temporal variability of BPA and TCS, we used a method describedby Hauser et al. (2004) and Teitelbaum et al. (2008). Three statisticalmethods were used: intraclass correlation coefficient (ICC), Spearmancorrelation coefficients between concentrations measured at differenttime points, and surrogate category analysis to determine how wellthe tertile categories based on a single measurement represented a12-month average concentration.

The results of the ICC and Spearman rank correlation coefficient aresummarized in Table SI.7. The ICC is defined as the percent of total var-iance explained by between-person variance (Rosner, 2000). With anICC of respectively 0.594 and 0.476 for BPA and TCS, the ICC measureof reproducibility was fair to good for both compounds. As shown inTable SI.7, the rank of BPA concentrations, expressed by the Spearmancorrelation coefficient, was only significantly correlated for 3 M and6 M samples apart from each other (Spearman coefficient r = 0.225,p=0.029 for 3M and r=0.349, p=0.01 for 6M). The samples collect-ed with an interval of 12 months were not significantly correlated(Spearman coefficient r = 0.121, p = 0.463). No clear trend for theSpearman correlation coefficient for BPA could be found during collec-tion interval (Table SI.7). TCS was correlated at all the collection inter-vals (p b 0.001) with Spearman correlation coefficients increasingwith an increasing collection interval (Table SI.7).

The surrogate category analysis showed consistent results for everysurrogate tertile (Table SI.8). For each of the four surrogate BPA and TCSsample tertiles (0 M, 3 M, 6 M and 12 M), a monotonic increasing meanconcentration was observed. This monotonic increase in the surrogatecategory analysis indicates consistent ranking of a 12-month BPA andTCS mean concentration by a single sample collected at various timeswithin the 12-month interval. It provides some assurance that we canreliably rank individuals relative to each other based on analyte levelsin a single spot urine sample (Mervish et al., 2012).

With an ICC of 0.572 for BPA, the within-person variance was small-er than the between-person variance. The within-person variance com-bines both biological and methodological sources. Due to the goodperformance of the analytical method (see Supplementary InformationandQuality assurance/quality control QA/QC section), the analytical im-precisionwill be b 10% and thuswill be lower than the typical biologicalvariance. Biological sources of within-person day-to-day variability areexpected, because these compounds have short biological half-livesand are quickly excreted (Mervish et al., 2012; Teitelbaum et al.,2008). The composition of the food (canned or non-canned) for BPAor the use of TCS-containing personal care products can largely influ-ence the day-to-day variability.

Human exposure to these chemicals is generally assessed in a singlespot urine sample; however, the concentrations of BPA- and TCS-conjugates represent a person's recent exposure. Food intake (determi-nant for BPA) and use of personal care products (determinant for TCS)can change substantially from day-to-day. Because of their short half-lives, BPA and TCS are eliminated in the urine within a few hours after

103T. Geens et al. / Environment International 76 (2015) 98–105

exposure, and therefore, a single urine sample may not be representa-tive for assessing the long-term exposure. An approach to minimizethe amount ofwithin-person variance and the bias caused by the collec-tion of single spot urine samples would be to use a mean concentrationbased on several urine samples obtained from each person, or to collecturine samples that integrate over longer time periods (e.g. 24 h)(Mervish et al., 2012). However, this approach is not always feasible.Despite thewithin-person variability in urinary BPA and TCS concentra-tions, the present study provided evidence that a single sample mightbe used as predictor of long-term exposure (at least 3 months for BPAand 12 months for TCS) and can be used to classify individuals intotertiles in epidemiological studies (Mahalingaiah et al., 2008).

3.4. Associations between urinary BPA and TCS levels, age and anthropo-metric data

No gender difference was observed in our study for the obese popu-lation at 0 M (Table SI.9), both for BPA (p = 0.431) as for TCS (p =0.453). A significant difference (p=0.004) could be observed in urinaryBPA concentrations between different age categories (Kruskal–Wallismultiple comparison test). Median concentration decreased from3.21 ng/mL in the 15–25 y group to 1.37 ng/mL in the N55 y group(Table SI.9). The difference was only significant between the 15–25 yand each of the other age group categories, but no significant differencescould be recorded between any of the age groups 26–35 y, 36–45 y,46–55 y and N55 y (Mann–Whitney test). No significant differencesbetween the age categories could be detected for TCS, p = 0.122(Table SI.9). Additionally, the Pearson correlation coefficients betweenBPA or TCS and age and BMI for the obese population were calculatedfor the four different time points considered in our study. Except forthe obese individuals at 3 months (3 M) after they start losing weight,BPA was negatively correlated with age at the four time points, with ap-value ranging between b0.001 and 0.047 (Table SI.10). TCS was notcorrelated with age, except for the 3 M obese individuals (Table SI.10).No significant correlations with age could be found for both BPA andTCS levels for the lean group. For both obese and lean populations, BMIwas not correlatedwith BPA and TCS levels (Table SI.10). Different studieshave reported the negative correlation between BPA concentrations andage. In the US-NHANES 2003–2004 study (N = 2517), children (6–11 y) had higher concentrations than adolescents (12–19 y) (p b 0.001),who in turn had higher concentrations than adults (20–59 y) (p =0.003). Also in the same study, the category N60 y had the lowest urinarylevels of BPA (Calafat et al., 2008). A negative correlation between age andBPA levels (p b 0.001)was also observed by Lakind andNaiman (2011) inthe US-NHANES 2005–2006 study (N = 2548).

Among the obese patients sampled for this study, a subgroup of 66individuals underwent bariatric surgery later on. According to Belgianlaw, patients are eligible for bariatric surgery if BMI N 40 kg/m2 orBMI N 35 kg/m2 with any of the following comorbidities: diabetesmellitus, obstructive sleep apnea, or therapy resistant arterial hyperten-sion. The other patients entered a conservative weight loss programwith dietary and lifestyle counseling. Therefore, we tested also for thedifferences recorded for the BPA and TCS levels between the two groupssampled (Table SI.11). However, although a significant decrease couldbe recorded for the BMI for each of the tested group, for both analytesmeasured, no significant differences (p N 0.05) could be recorded be-tween the two groups of samples after 3, 6 and 12 months they startedthe weight loss program (Table SI.11).

3.5. Associations between urinary BPA and TCS levels and thyroid hormones

It has been suggested that exposure to non-persistent chemicals,such as phthalatesmay affect thyroid signaling through differentmech-anisms (Boas et al., 2006; Koeppe et al., 2013; Meeker and Ferguson,2011; Zoeller, 2007). Therefore, we also tested for potential relation-ships between urinary BPA and TCS levels and serum thyroid hormones

(THs), such as free thyroxine (FT4) and thyroid-stimulating hormone(TSH) for obese and lean individuals. There were no statistically signif-icant differences (p N 0.05) for both TSH and fT4 between 0 M patientsand lean population included in our study.

Linear regression analyses between these parameters andWC wereperformed with age, weight loss (for 0 M, 3 M and 6 M obese individ-uals) and gender included in the model. The adjusted regression coeffi-cients (β) obtained for BPA and TCS togetherwith the significance of thecontaminant level contribution to the whole model are presented inTable 3. Only few significant (p b 0.05) relationships could be foundoverall: negative associations for TCS in relation to FT4 levels for 0 Mand 3 M female obese individuals and positive associations for BPAin relation to TSH for the lean group. Previous literature suggestedthat TCS might have multiples modes of action, but there is someevidence that it may inhibit sulfotransferase (SULT) and uridinediphospho-glucuronosyl-transferases enzyme activity associatedwith Phase-II thyroid metabolism (Schuur et al., 1998; Wang et al,2004). Since sulfation for T3 and glucuronidation for T4 seems to be therate-limiting step in biliary excretion (Klaassen andWatkins, 2010), pos-itive associations previously reported for urinary TCS in relation withserum T3 levels in an adolescent population (Koeppe et al., 2013) indi-cated the SULT inhibition as the major mode of action for TCS. Althoughour data was generated from a rather small sample size, we evidencedsome significant negative relationships between urinary TCS levels andserum FT4 in female obese individuals before they start losing weightand 3 months later, while this relation was not significant for the leangroup (Table 3). Therefore, a potential influence of the TCS exposureto the Phase-II thyroid metabolism might be suggested by our findings:TCS might alter the activity of the uridine diphospho-glucuronosyl-transferases enzymes and therefore influencing the T4 glucuronidationprocess.

There is limited literature focused on the BPA relationship with thethyroid hormone levels so far. An inverse relationship between urinaryBPA concentrations and TSH was evidenced in a cross-sectional studyof 167 men from an infertility clinic (Meeker et al., 2010) and in theUS-NHANES 2007–2008 study on 1346 adults (Meeker and Ferguson,2011). Our results give support for a positive relationship between uri-nary BPA and serum TSH levels, but only for the whole lean populationand not for obese individuals included in our study. However, whenstratifying the analysis by considering the gender of the sampled controlpopulation, the significance of ourfindings decreased consistently keep-ing the relationship to the border of significance only for the females(p = 0.082).

4. Conclusions

In our study on estimating potential differences in the human expo-sure to non-persistent chemicals like BPA or TCS, we could not evidencesignificant differences between obese and lean individuals, nor betweenobese individuals at different time points after they followed a weightloss program. This observationwas noticed independently of theweightloss method followed: bariatric surgery or conservative weight lossprogram with dietary and lifestyle counseling. This would suggestthat sources of human exposure to such chemicals did not differ foreach case considered. These findings were also supported from theestimations of the DI of BPA and TCS, which did not differ significant-ly among the time points considered after obese individuals startedlosing weight (6 and 12 months). After testing for potential relation-ships between urinary BPA and TCS levels and serum thyroid hormones,only few significantfigureswere obtained: negative associations for TCSin relation to FT4 levels for 0 M and 3 M female obese individuals andpositive associations for BPA in relation to TSH for the lean group.Therefore it might be suggested that TCS exposure in the case of obeseindividuals might have an influence on the activity of the uridinediphospho-glucuronosyl-transferases enzymes and therefore influenc-ing the T4 glucuronidation process.

Table 3Regression analyses of associations between urinary BPA and TCS (ng/mL) and freethyroxine (FT4), thyroid-stimulating hormone (TSH) and waist circumference (WC) inoverweight and obese individuals sampled in Belgium compared to lean individuals.Age, weight loss (for 3 M and 6 M obese individuals) and gender influence were includedin themodel. Significance figures (p values) only of the urinary BPA and TCS levels influenceon the overall model are highlighted: p b 0.05 (green) and 0.05 b p b 0.1 (orange).

Outcome

0MAll Males Females

β p β p β pTSH

BPA –0.11 (–0.29 to 0.07) 0.218 –0.03 (–0.36 to 0.31) 0.882 –0.13 (–0.35 to 0.08) 0.220TCS 0.01 (–0.17 to 0.17) 0.990 –0.15 (–0.46 to 0.16) 0.326 0.08 (–0.12 to 0.29) 0.428

FT4BPA 0.02 (–0.15 to 0.20) 0.803 0.02 (–0.31 to 0.35) 0.902 0.02 (–0.19 to 0.24) 0.840TCS –0.18 (–0.34 to –0.01) 0.034* –0.11 (–0.42 to 0.20) 0.477 –0.22 (–0.42 to –0.02) 0.032*

WCBPA 0.04 (–0.14 to 0.21) 0.677* 0.10 (–0.23 to 0.44) 0.532 –0.04 (–0.25 to 0.18) 0.721TCS 0.01 (–0.16 to 0.17) 0.928* –0.01 (–0.32 to 0.31) 0.980 –0.01 (–0.21 to 0.20) 0.934

3MAll Males Females

β p β p β pTSH

BPA 0.10 (–0.14 to 0.33) 0.414 0.07 (–0.37 to 0.52) 0.738 0.18 (–0.11 to 0.46) 0.215TCS –0.01 (–0.24 to 0.22) 0.061** –0.25 (–0.69 to 0.19) 0.249 0.06 (–0.22 to 0.34) 0.653

FT4BPA 0.03 (–0.20 to 0.26) 0.805 0.07 (–0.38 to 0.53) 0.743 0.04 (–0.23 to 0.31) 0.778*TCS –0.01 (–0.03 to 0.01) 0.061** 0.06 (–0.40 to 0.53) 0.787 –0.30 (–0.55 to –0.05) 0.022*

6MAll Males Females

β p β p β pTSH

BPA –0.01 (–0.30to 28) 0.958 0.21 (–0.67 to 1.08) 0.607 0.10 (–0.26 to 0.45) 0.580*TCS 0.10 (–0.18to 0.37) 0.491* –0.22 (–1.01 to 0.56) 0.536 0.19 (–0.13 to 0.52) 0.235*

FT4BPA –0.05 (–0.38 to 0.27) 0.742 0.13 (–0.69 to 0.95) 0.727 –0.04 (–0.43 to 0.35) 0.842TCS –0.09 (–0.41 to 0.22) 0.548 –0.05 (–0.78 to 0.69) 0.893 –0.16 (–0.52 to 0.21) 0.381

LEANTSH

BPA 0.35 (0.05 to 0.66) 0.026** 0.29 (–0.50 to 1.08) 0.429 0.34 (–0.05 to 0.725) 0.082TCS 0.08 (–0.25 to 0.42) 0.618 0.28 (–0.44 to 1.01) 0.398 0.03 (–0.39 to 0.45) 0.888

FT4BPA –0.19 (–0.52 to 0.13) 0.235 –0.19 (–0.98 to 0.61) 0.605 –0.24 (–0.64 to 0.16) 0.224TCS –0.07 (–0.40 to 0.26) 0.674 0.02 (–0.72 to 0.77) 0.943 –0.12 (–0.54 to 0.29) 0.545

WCBPA –0.27 (–0.57 to 0.04) 0.084* 0.12 (–0.53 to 0.76) 0.697 –0.22 (–0.58 to 0.14) 0.224*TCS 0.14(–0.17 to 0.46) 0.366** –0.24 (–0.82 to 0.33) 0.361** 0.33 (–0.03 to 0.68) 0.068*

β — Adjusted regression coefficients corresponding to BPA/TCS influence on the thyroidhormone levels; bold values correspond to significant (p b 0.05) and borderline non-significant (0.05 b p b 0.1) figures of the overall linearmodel: *— significance of the overalllinear model (all parameters included) is p b 0.05; ** — significance of the overall model(all parameters included) is 0.05 b p b 0.1.

104 T. Geens et al. / Environment International 76 (2015) 98–105

Acknowledgments

The research performed in this study was funded by the Universityof Antwerp through a GOA project (Endocrine disrupting environ-mental chemicals: From accumulation to their role in the global“neuro-endocrine” epidemic of obesity and its metabolic consequences;FA020000/2/3565). ACD was financially supported through a postdoc-toral fellowship from the Research Scientific Foundation-Flanders(FWO), Belgium. GM thanks the University of Antwerp for a post-doctoral fellowship. Mrs. Rie Braspenning and other nurses and nutri-tionists are also greatly acknowledged for helping with sampling andhandling.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.envint.2014.12.003.

References

Baillie-Hamilton, P., 2002. Chemical toxins: a hypothesis to explain the global obesityepidemic. J. Altern. Complement. Med. 8, 185–192.

Boas, M., Feldt-Rasmussen, U., Skakkebaek, N.E., Main, K.M., 2006. Environmentalchemicals and thyroid function. Eur. J. Endocrinol. 154, 599–611.

Calafat, A.M., Ye, X., Wong, L.Y., Reidy, J.A., Needham, L.L., 2008. Exposure of the U.S.population to bisphenol A and 4-tertiary-octylphenol: 2003–2004. Environ. HealthPerspect. 116, 39–44.

Carwile, J.L., Michels, K.B., 2011. Urinary bisphenol A and obesity: NHANES 2003–2006.Environ. Res. 111, 825–830.

Centers for Disease Control and Prevention (CDC), 2009. Fourth National Report onHuman Exposure to Environmental Chemicals. URL, http://www.cdc.gov/exposurereport/pdf/FourthReport.pdf (accessed June 2014).

Christensen, K.L.Y., Lorber, M., Koslitz, S., Brüning, T., Koch, H.M., 2012. The contribution ofdiet to total bisphenol A body burden in humans: results of a 48 hour fasting study.Environ. Int. 50, 7–14.

Cocker, J., Mason, H.J., Warren, N.D., Cotton, R.J., 2011. Creatinine adjustment of biologicalmonitoring results. Occup. Med. 61, 349–353.

Dirinck, E.L., Dirtu, A.C., Govindan, M., Covaci, A., Van Gaal, L.F., Jorens, P.G., 2014. Expo-sure to persistent organic pollutants: relationship with abnormal glucosemetabolismand visceral adiposity. Diabetes Care 37, 1951–1958.

Dirtu, A.C., Dirinck, E., Malarvannan, G., Neels, H., Van Gaal, L., Jorens, P.G., Covaci, A.,2013a. Dynamics of organohalogenated contaminants in human serum from obeseindividuals during one year of weight loss treatment. Environ. Sci. Technol. 47,12441–12449.

Dirtu, A.C., Geens, T., Dirinck, E., Malarvannan, G., Neels, H., Van Gaal, L., Jorens, P.G., Covaci,A., 2013b. Phthalate metabolites in obese individuals undergoing weight loss: urinarylevels and estimation of the phthalates daily intake. Environ. Int. 59, 344–353.

Fenichel, P., Chevalier, N., Brucker-Davis, F., 2013. Bisphenol A: an endocrine andmetabol-ic disruptor. Ann. Endocrinol. 74, 211–220.

Galloway, T., Cipelli, R., Guralnik, J., Ferrucci, L., Bandinelli, S., Corsi, A.M., Money, C.,McCormack, P., Melzer, D., 2010. Daily bisphenol A excretion and associations withsex hormone concentrations: results from the InCHIANTI adult population study. En-viron. Health Perspect. 118, 1603–1608.

Geens, T., Neels, H., Covaci, A., 2009. Sensitive and selective method for the determinationof bisphenol-A and triclosan in serum and urine as pentafluorobenzoate-derivativesusing GC-ECNI/MS. J. Chromatogr. B 877, 4042–4046.

Geens, T., Apelbaum, T.Z., Goeyens, L., Neels, H., Covaci, A., 2010. Intake of bisphenol Afrom canned beverages and foods on the Belgian market. Food Addit. Contam. A 27,1627–1637.

Geens, T., Goeyens, L., Covaci, A., 2011. Are potential sources for human exposure tobisphenol-A overlooked? Int. J. Hyg. Environ. Health 214, 339–347.

Geens, T., Aerts, D., Berthot, C., Bourguignon, J.P., Goeyens, L., Lecomte, P., Maghuin-Rogister, G., Pironnet, A.M., Pussemier, L., Scippo, M.L., Van Loco, J., Covaci, A., 2012.A review of dietary and non-dietary exposure to bisphenol-A. Food Chem. Toxicol.50, 3725–3740.

Harley, K.G., Schall, R.A., Chevrier, J., Tyler, K., Aguirre, H., Bradman, A., Holland, N.T.,Lustig, R.H., Calafat, A.M., Eskenazi, B., 2013. Prenatal and postnatal bisphenol A expo-sure and body mass index in childhood in the CHAMACOS cohort. Environ. HealthPerspect. 121, 514–520.

Hauser, R., Meeker, J.D., Park, S., Silva, M.J., Calafat, A.M., 2004. Temporal variability of uri-nary phthalate metabolite levels in men of reproductive age. Environ. HealthPerspect. 112, 1734–1740.

Klaassen, C.D., Watkins, J.B., 2010. Casarett & Doull's Essentials of Toxicology. 2nd ed.McGraw-Hill Companies, Inc., New York.

Koch, H.M., Calafat, A.M., 2009. Human body burdens of chemicals used in plastic manu-facture. Philos. Trans. R. Soc. Lond. B Biol. Sci. 364, 2063–2078.

Koch, H.M., Kolossa-Gehring, M., Schroter-Kermani, C., Angerer, J., Bruning, T., 2012.Bisphenol A in 24 h urine and plasma samples of the German Environmental Speci-men Bank from 1995 to 2009: a retrospective exposure evaluation. J. Expo. Sci. Envi-ron. Epidemiol. 22, 610–616.

Koeppe, E.S., Ferguson, K.K., Colacino, J.A., Meeker, J.D., 2013. Relationship between uri-nary triclosan and paraben concentrations and serum thyroid measures in NHANES2007–2008. Sci. Total Environ. 445–446, 299–305.

Lakind, J.S., Naiman, D.Q., 2011. Daily intake of bisphenol A and potential sources of expo-sure: 2005–2006 National Health and Nutrition Examination Survey. J. Expo. Sci. En-viron. Epidemiol. 21, 272–279.

Lakind, J.S., Goodman, M., Mattison, D.R., 2014. Bisphenol A and indicators of obesity, glu-cose metabolism/type 2 diabetes and cardiovascular disease: a systematic review ofepidemiologic research. Crit. Rev. Toxicol. 44, 121–150.

Lang, I.A., Galloway, T.S., Scarlett, A., Henley, W.E., Depledge, M., Wallace, R.B., Melzer, D.,2008. Association of urinary bisphenol A concentration with medical disorders andlaboratory abnormalities in adults. JAMA 300, 1303–1310.

Li, D.K., Miao, M., Zhou, Z., Wu, C., Shi, H., Liu, X., Wang, S., Yuan, W., 2013. Urinebisphenol-A level in relation to obesity and overweight in school-age children. PLoSOne 12, e65399.

Magliano, D.J., Lyons, J.G., 2013. Bisphenol A and diabetes, insulin resistance, cardiovascu-lar disease and obesity: controversy in a (plastic) cup? J. Clin. Endocrinol. Metab. 98,502–504.

Mahalingaiah, S., Meeker, J.D., Pearson, K.R., Calafat, A.M., Ye, X., Petrozza, J., Hauser, R.,2008. Temporal variability and predictors of urinary bisphenol A concentrations inmen and women. Environ. Health Perspect. 116, 173–178.

Malarvannan, G., Dirinck, E., Dirtu, A.C., Pereira-Fernandes, A., Neels, H., Jorens, P.G., VanGaal, L., Blust, R., Covaci, A., 2013. Distribution of persistent organic pollutants intwo different fat compartments from obese individuals. Environ. Int. 55, 33–42.

Meeker, J.D., 2012. Exposure to environmental endocrine disruptors and child develop-ment. Arch. Pediatr. Adolesc. Med. 166, 952–958.

Meeker, J.D., Ferguson, K.K., 2011. Relationship between urinary phthalate and bisphenolA concentrations and serum thyroidmeasures in U.S. adults and adolescents from theNational Health and Nutrition Examination Survey (NHANES) 2007–2008. Environ.Health Perspect. 119, 1396–1402.

Meeker, J.D., Calafat, A.M., Hauser, R., 2010. Urinary bisphenol A concentrations in relationto serum thyroid and reproductive hormone levels in men from an infertility clinic.Environ. Sci. Technol. 44, 1458–1463.

105T. Geens et al. / Environment International 76 (2015) 98–105

Mervish, N., Blount, B., Valentin-Blasini, L., Brenner, B., Galvez, M.P., Wolff, M.S.,Teitelmaum, S.L., 2012. Temporal variability in urinary concentrations of perchlorate,nitrate, thiocyanate and iodide among children. J. Expo. Sci. Environ. Epidemiol. 22,212–218.

Miyawaki, J., Sakayama, K., Kato, H., Yamamoto, H., Masuno, H., 2007. Perinatal and post-natal exposure to bisphenol A increases adipose tissue mass and serum cholesterollevel in mice. J. Atheroscler. Thromb. 14, 245–252.

Newbold, R.R., Padilla-Banks, E., Snyder, R.J., Phillips, T.M., Jefferson, W.N., 2007a. Devel-opmental exposure to endocrine disruptors and the obesity epidemic. Reprod.Toxicol. 23, 290–296.

Newbold, R., Padilla-Banks, E., Snyder, R., Jefferson, W., 2007b. Perinatal exposure toenvironmental estrogens and the development of obesity. Mol. Nutr. Food Res. 51,912–917.

NICNAS, 2009. National Industrial Chemicals Notification and Assessment Scheme. Prior-ity Existing Chemical Assessment Report No. 30 — Triclosan.

Rosner, B., 2000. Fundamentals of Biostatistics. 5th edn. Duxbury, Pacific Grove, CA.Schug, T., Janesick, A., Blumberg, B., Heindel, J., 2011. Endocrine disrupting chemicals and

disease susceptibility. J. Steroid Biochem. Mol. Biol. 127, 204–211.Schuur, A.G., Legger, F.F., van Meeteren, M.E., Moonen, M.J., vanLeeuwen-Bol, I., Bergman,

A., et al., 1998. In vitro inhibition of thyroid hormone sulfation by hydroxylated me-tabolites of halogenated aromatic hydrocarbons. Chem. Res. Toxicol. 11, 1075–1081.

Shankar, A., Teppala, S., Sabanayagam, C., 2012. Urinary bisphenol a levels and measuresof obesity: results from the national health and nutrition examination survey 2003–2008. ISRN Endocrinol. 2012, 965243.

Silver, M.K., O'Neill, M.S., Sowers, M.R., Park, S.K., 2011. Urinary bisphenol A and type-2diabetes in U.S. adults: data from NHANES 2003–2008. PLoS One 6, e26868.

Tang-Péronard, J.L., Andersen, H.R., Jensen, T.K., Heitmann, B.L., 2011. Endocrine-disrupting chemicals and obesity development in humans: a review. Obes. Rev.12, 622–636.

Teitelbaum, S.L., Britton, J.A., Calafat, A.M., Ye, X., Silva, M.J., Reidy, J.A., Galvez, M.P.,Brenner, B.L., Wolff, M.S., 2008. Temporal variability in urinary concentrations ofphthalate metabolites, phytoestrogens and phenols among minority children in theUnited States. Environ. Res. 106, 257–269.

Thayer, K.A., Heindel, J.J., Bucher, J.R., Gallo, M.A., 2012. Role of environmental chemicalsin diabetes and obesity: a National Toxicology Program workshop review. Environ.Health Perspect. 120, 779–789.

The Arctic Monitoring and Assessment Program (AMAP) Ring Test, 2010. InstituteNational de Sauté Publique Quebéc, Centre de Toxicologie, Report 2010.

The German External Quality Assessment Scheme (G-EQUAS) organized by Institute andOut-Patient Clinic for Occupational, Social and Environmental Medicine of the Uni-versity Erlangen-Nuremberg, for Analyses in Biological Materials. Reports from2011 and 2013, http://www.g-equas.de/.

Valvi, D., Casas, M., Mendez, M.A., Ballesteros-Gómez, A., Luque, N., Rubio, S., Sunyer, J.,Vrijheid, M., 2013. Prenatal bisphenol A urine concentrations and early rapid growthand overweight risk in the offspring. Epidemiology 24, 791–799.

Völkel, W., Colnot, T., Csanady, G.A., Filser, J.G., Dekant, W., 2002. Metabolism and kineticsof bisphenol A in humans at low doses following oral administration. Chem. Res.Toxicol. 15, 1281–1287.

Völkel, W., Bittner, N., Dekant, W., 2005. Quantitation of bisphenol A and bisphenol A glu-curonide in biological samples by high performance liquid chromatography–tandemmass spectrometry. Drug Metab. Dispos. 33, 1748–1757.

Vom Saal, F.S., Nagel, S.C., Coe, B.L., Angle, B.M., Taylor, J.A., 2012. The estrogenic endocrinedisrupting chemical bisphenol A (BPA) and obesity. Mol. Cell. Endocrinol. 354, 74–84.

Voorspoels, S., Covaci, A., Maervoet, J., Schepens, P., 2002. Relationship between age andlevels of organochlorine contaminants in human serum of a Belgian population.Bull. Environ. Contam. Toxicol. 69, 22–29.

Wang, L.Q., Falany, C.N., James, M.O., 2004. Triclosan as a substrate and inhibitor of 3′-phosphoadenosine 5′-phosphosulfate-sulfotransferase and UDP-glucuronosyl trans-ferase in human liver fractions. Drug Metab. Dispos. 32, 1162–1169.

Wang, T., Li, M., Chen, B., Xu, M., Xu, Y., Huang, Y., Lu, J., Chen, Y., Wang,W., Li, X., Liu, Y., Bi,Y., Lai, S., Ning, G., 2012. Urinary bisphenol A (BPA) concentration associates withobesity and insulin resistance. J. Clin. Endocrinol. Metab. 97, 223–227.

Wells, E.M., Jackson, L.W., Koontz, M.B., 2014. Association between bisphenol A andwaist-to-height ratio among children: National Health and Nutrition Examination Survey,2003–2010. Ann. Epidemiol. 24, 165–167.

Zoeller, R.T., 2007. Environmental chemicals impacting the thyroid: targets and con-sequences. Thyroid 17, 811–817.