Serum concentrations of Organochlorine pesticides and growth among Russian boys. (2024)

BACKGROUND: Limited human data suggest an association oforganochiorine pesticides (OCPs) with adverse effects on children'sgrowth.

OBJECTIVE: We evaluated the associations of OCPs withlongitudinally assessed growth among peripubertal boys from a Russiancohort with high environmental OCP levels.

METHODS: A cohort of 499 boys enrolled in the RussianChildren's Study between 2003 and 2005 at 8--9 years of age werefollowed prospectively for 4 years. At study entry, 350 boys had serumOCPs measured. Physical examinations were conducted at entry andannually. The longitudinal associations of serum OCPs with annualmeasurements of body mass index (BMI), height, and height velocity wereexamined by multivariate mixed-effects regression models for repeatedsures, controlling for potential confounders.

RESULTS: Among the 350 boys with OCP measurements, median serumhexachlorobenzene (HCB), [beta]-hexachlorocyclohexane ([beta]HCH), andp,p'-dichlorodiphenyldichioroethylene (p,p '-DDE)concentrations were 159 ng/g lipid, 168 ng/g lipid, and 287 ng/g lipid,respectively. Age-adjusted BMI and height z-scores generally fell withinthe normal range per World Health Organization standards at entry andduring follow-up. However, in adjusted models, boys with higher serumHCB, [beta]HCH, and p,p'-DDE had significantly lower mean [95%confidence interval (CI)] BMI z-scores, by --0.84 (--1.23, --0.46),--1.32 (--1.70, --0.95), and --1.37 (--1.75, --0.98), respectively, forthe highest versus lowest quintile. In addition, the highest quintile ofp,p'-DDE was associated with a significantly lower mean (95% CI)height z-score, by --0.69 (--1.00, --0.39) than that of the lowestquintile.

CoNcLusIoNs: Serum OCP concentrations measured at 8--9 years of agewere associ.ated with reduced growth, particularly reduced BMI, duringthe peripubertal period, which may affect attainment of optimal adultbody mass and height.

KEY WORDS: BMI, children, DDE, environment, epidemiology, HCB,height, hexachiorocyclohexane, organochlorine pesticides. Environ HealthPerspect 120:303--308 (2012). http://dx.doi.org/1O.12891ehp.1 103743[Online 7 October 2011]

Dichlorodiphenyltrichloroethane (DDT), its metabolitep,p'.-dichlorodiphenyltrichlorocthylene (p,p'-DDE),hexachiorobenzenc (HCB), and [beta]-hexachlorocyclohexane ([beta]HCH)are organochlorine pesticides (OCPs) that are ubiquitous envIronmentalpollutants, despite being banned (HCB and [beta]HCH) or greatlyrestricted (DDT) (United Nations Environmental Programme 2009). Theselipophilic compounds accumulate in the food chain and have half-lives ofyears to decades (Longnecker 2005; Wolff et al. 2000). In humans, themost common route of exposure is diet (Darnerud et al. 2006; Marti-Cidet al. 2008). OCPs readily cross the placenta (Carrizo et aI. 2006) andconcentrate in breast milk (Karmaus et al. 2001; Wolff et at 2005),thereby leading to infant exposures. Although body burdens of OCPs havedecreased over time (Colles et al. 2008; Link et al. 2005), childrencontinue to be exposed.

Children may be especially vulnerable to the effects ofendocrine-disrupting OCPs (Eskenazi et al. 2009; Landrigan et al. 2004).Prenatal exposure to p,p'-DDE and HCB has been associated withreduced birth weight and length, independent of gestational age (Eggesboet al. 2009; Ribas-Fito et al. 2002; Siddiqui et al, 2003; Weisskopf etal. 2005; Wolff et aI. 2007), although estimated associations withpostnatal growth have been inconsistent (Eskenazi et al. 2009; Mendez etal. 2011; Smink et al. 2008). Despite concern regarding the effects ofDDT and its merabolites on children's health, including growth(Eskenazi et al. 2009), expanded DDT use is advocated for malariacontrol (Griffin et al. 2010).

We investigated childhood exposures to OCPs in a cohort of boys inChapaevsk, Russia, a town highly contaminated with HCB, [beta]HCH,dioxins, and polychlorinated biphenyls (PCBs) from local chemical plants(Ecological Analytical Center 2007; Revich et al. 1999). We report dataon the associations of serum OCPs at study entry with serial measures ofgrowth during 4 years of follow-up.

Methods

Study population. The Russian Children's Study is aprospective cohort study of 499 boys in Chapaevsk, Russia, described indetail elsewhere (Burns et al. 2009). The boys, identified using thetownwide health insurance information system, were enrolled at 8 or 9years of age from 2003 to 2005. Exclusion criteria included beinginstitutionalized or having severe cerebral palsy. OCPs were notmeasured for the first 144 boys recruited into the study, and five boyswith severe chronic illnesses were excluded from the present analysis,leaving 350 boys with OCPs measured. The retention rate was 86% after 4years. The study was approved by the human studies institutional reviewhoards of the Chapaevsk Medical Association, Harvard School of PublicHealth, Brigham and Women's Hospital, and University ofMassachusetts Medical School. The parents or guardians signed informedconsent forms, and the boys signed assent forms.

At study entry, the boys had a physical examination and blood draw.The mother or guardian completed a nurse--ad ministered health andlifestyle questionnaire (Hauser et al. 2005; Lee et al. 2003) thatincluded birth history, famil and child's medical history,occupational and residential history, household income, and parentaleducation. Birth weight and gestational age were obtained from mcdicalrecords. A validated Russian Institute of Nutrition semiquantitativefood frequency questionnaire was used to ascertain the child'sdietary intake (Martinchik et al. 1998; Rockert et al. 1997).

Physical examination. At study entry and annual follow--up visits,a standardized an thro-pometric examination was performed by a singlestudy investigator (O.S.) per written protocol and without knowledge ofthe boys' pesticide levels. Height was measured to the nearest 0.1cm using a stadiometer. Weight was measured to the nearest 1 00 g with abalance scale. Age-adjusted z-scores were calculated for height and bodymass index (BMI; kilograms per square meter) using the World HealthOrganization (WHO) standards (WHO 2011), and for weight using theCenters for Disease Control and Prevention (CDC) standards (CDC 2009)because WHo standards are unavailable For this age group. Annual heightvelocity (HV) was calculated for five 1-year intervals (ages 8--13years) by computing the difference in height between visits, with eachboy contributing up to Four measurements.

Blood sample analyses. Sera from enrollment blood samples werestored at --35[degrees]C until shipment on dry ce to the CDC (Atlanta,GA, USA) for organochlorine analysis. The samples, including methodblank and quality control samples, were spiked with 'C1,-1ahcledpesticides, extracted by a C18 solid-phase extraction (SPE) Followed bya multicolumn amomated cleanup and enrichnent procedure using eitherlarge-volume (Turner et al. 1 997) or smali--volume (Sjodin et al. 2004)SPE and analyzed using high- resolution mass spectrometrv in selectiveion monitoring (Barr et al. 2003). Sera were analyzed For dioxin-likecorn pounds [D LCs (polychlorinated dibenzo-p-d ioxins, dibenzofurans,coplanar PCBs)1 and PG.13s, and whole blood was analyzed for blood leadlevels (BLLs) as described previously (Burns et al. 2009; Hauser et al.2008; Williams et aI. 2010). Lipid-adjusted total 2005 toxic equivalents(TEQs; a measure of toxicity for DLCs) was calculated (Burns et al.2009).

Statistical analysis. We evaluated the aSsociatiOnS of serum OCPconcentrations measured at 8--9 years of age with the boys'age-adjusted BMI and height z-scores and HV across five study visits(entry and up to four annual Follow-up visits) through May 2008.Individual serum OCPs (HCB, [beta] HCH, and p,p'--DDE) were dividedinto quintiles, with the lowest quintile as the reference group. We usedmixed-effects regression models for repeated measures with anunstructured covariance to examine the assoCiations of OCP quill tileswith growth measures. We evaluated univariare associations based onprior literature, fitted a full multivariate model including allcovariates with p [+ or -] 0.20, and then reduced it to a core modelincluding covariates with p < 0.10 and those required a priori forbiological interpretability of other covariates. 'Ihis core modelwas used for all statistical analyses and included boy's age, birthweight, and gesrational age categories (< 37, 37--42, > 42 weeks);household income categories (< $US 175, $ 175--250, > $250 permonth); total calories; percent calories from carbohydrate, Fat, andprotein; and high (> 5 [micro]g/dL) versus low BLL. In our analyses,statsticaI significance for main effects and interactions was set at =0.05. Tests for trend over OCP levels were performed by modelingquintiles of exposure as a continuous variable. In sensitivity analyses,we adjusted for parental height and weight because these data wereavailable for only 67% (n = 236) of fathers and 94% (n = 329) ofmothers. Our primary models did not adjust for pubertal stage becauseOCPs may affect pubertal stage and thus be on the causal pathway betweenOCP exposures and growth. However, we conducted sensitivity analysesadjusting for pubertal stage based on Tanner genitalia staging (Tannerand Whitehouse 1976) (stage 4--5, 2--3, or 1) to confirm OCPsassociations with growth. We also performed sensitivity analysesadjusting for quintiles of serum total TEQs, DLCs, and PCBs(non-dioxin-like and mono-ortho) because of our prior findings ofassociations between these compounds and longitudinal growth measures(Burns et al. 2011). We assessed whether the associations between OCPsand growth were modified by ages using an interaction term of continuousage times an indicator of OCP concentrations above the median. Inaddition, we fitted a joint model including all OCPs simultaneously toevaluate their independent contributions in the context of multipleexposures.

Results

Study population and serum OGF trations. Among the 350 boys, ratesof prematurity (< 37 weeks) and low I)irth weight (<2,500 g were8.9% and 5.1%, respectively, similar to U.S. rates (National VitalStatistics Reports 2007). The percentages of calories from dietaryprotein, fat, and carbohydrate (data not shown) were within age-specificnu*tritionally appropriate ranges (Food and Nutrition Board 2005). Birth,maternal, and household characteristics are presented in Table 1.

Table 1. Descriptive characteristics of the participantsof the Russian Children's Study (n=350)Characteristic(a) MeasureGrowth measurements at study entry8-year-old boys(n=297)Height (cm) 128.3 [+ or -] 59Height z-score(b) 0.15 [+ or -] 1.02Weight (kg) 26.4 [+ or -] 53Weiqht z-score(b) 0.04 [+ or -] 1.25BMI (kg/[m.sup.2]) 15.9 [+ or -] 2.2BMI z-score(b) 0.12 [+ or -] 1.28pubertal onset (genitalia 65 (22)stage[greater than or equal to]2)9-year-old boys (n= 53)Height (cm) 132.7 [+ or -] 5.3Height z-score(b)Weight (kg) 27.9 [+ or -] 5.7Height z-score(b) -0.27 [+ or -] 1.15BMI (kg/[m.sup.2]) 15.8 [+ or -] 26BMI z-score(b) 0.42 [+ or -] 1.06pubertal onset (genitalia 17 (32)stage[greater than or equal to]2)Birth and neonat history (all boys,n = 350)Birth weiqht (kg) 34 [+ or -] 0.5Gestationa age (weeks) 39.O [+ or -] 1.8Duration of breast-feeding (weeks) 26.1 [+ or -] 33.9Breast-fed 297 (85)Maternal characteristics andexposures during pregnancy (allboys, n=350)Moter's age at sons birth < 25 222(64)yearsMother overweight at study entry 132 (40)Maternal alcohol consumtion 48 (14)Maternal tobacco smoking 25 (7)Any household tobacco smoking 165 (48)Maternal reported activities andexposures (all boys, n= 350)Ever employed at chemical plat 20 (6)Herbicide/pesticide occupattonal 5 (1)exposureLocal gardening 210 (60)Herbicide/pesticide personal 309 (88)BLL (pq/dL) at study entry (all 3.7 [+ or -]boys, n = 350) 2.5Househod caactersiics 1a boys. ii =35OParental education (maximum)Secondary education or less 29 (8)Junior college/teechnical training 198 (57)University graduate 121 (35)Household income reported at studyentry (US$/month)<175 107 (31)250 88 (25)>250 154 (44)Both parents living in the home at 227 (65)study entryData are mean [+ or -] SD. or n(%) unless stated otherwise.(a) Missing information: birth weight (n=1), gestational age (n=2),breast-fed (n=5), mother's age (n=3), mother over-weight (n=21),mother drank during pregnancy (n=11), mother smoked during pregnacy(n=8), any household smoking (n=6), mother employed at chemicalplant (n=10), pesticide occupational exposure (n=12), local gardening(n=6), pesticide/herbicide personal use (n=9), parental education(n=2), household income (n=1). (b) WHO age-adjusted z-score (WHO) 2011)

We compared boys with serum OCP measurements and those in thecohort without OCP measurements (data riot shown) and found nosignificant differences in height, weight, BMI z-scores at study entry,or birth or family characteristics, except that more boys with OCPmeasurements had higher household income (69% high income vs. 54%, p =0.001). Boys with OCP measurements had significantly lower total TEQs,DI, Cs, PCBs, and BLLs than did boys without oCP measurements. Mothersof boys with OCP measurements reported more personal pesticide usc (91%vs. 77%, p < 0.00 1)

Median HCB and p,p'-DDE concentrations (Table 2) were 11 and2.5 times higher, respectively, than the upper 95% confidence intervals(CIs) for the median values reported for 12- to 19-year-old U.S.adolescents in the 2003--2004 National Health and Nutrition ExaminationSurvey. The median 3HCH concentration in the same U.S. adolescent groupwas below the limit of detection and was not in the detectable rangeuntil the 95th percentile; thus, the Russian boys' median [beta]HCHlevel was almost 12 times higher than the upper 95% confidence limit ofthe 95th percentile fOr U.S. teens (Patterson et al. 2009), The medians(25th, 75th percentiles) for serum total TEQs, DLCs, and PCBconcentrations were, respectively, 19.4 pg/g lipid (12.5, 29.2), 323pg/g lipid (261, 416), and 208 ng/g lipid (151, 329). The median (25th,75th percentiles) for BLL was 3.0 (2.0, 4.0) [mu]g/dL. The OCPs werepositively correlated with each other and the other organochlorinecompounds1 with the highest Spearman relation (0.70) between [beta]HCHand both total TEQs and PCBs. fHCH was also moderately correlated withp,p'-DDE (0.58) and HCB (0.54). BLL showed weaker correlations withthe OCPs (0.10--0.24).

Table 2. Distribution of measured OCPs (ng/g lipid) (a) among8-year-old boys enrolled in the Russian Childern's Study (n=350) percentileOCP n 10th 25th 50th (median) 75th 90thHCB 350 80 107 159 247 365[bate] HCH 350 82 114 168 272 421p,p'-DDE 350 122 189 287 492 866(a) None below the limit of detection

Growth measures at entry and during follow-up. At study entry, mostof the boys' height, weight, and BMI (Table 1) were within thenormal range according to child growth standards from WHO (de Onis etal. 2007) and CDC (McDowell et al. 2009). However, 18% of the boys wereoverweight (> 1 SD above the mean) (de Onis et al. 2007), 6% wereunderweight (defined as > 2 SD below the mean) (de Onis et al. 2007),and 2% of the boys' heights were > 2 SD below the mean. Duringfollow-up, the mean height, weight, and BMI z-scores remained relativelyunchanged (Figure 1). The mean ([+ or -] SD) F-IVs at 8--13 years of agewere, respectively, 5.3 [+ or -] 0.8, 5.5 [+ or -]0.8, 4.9 [+ or -] 1.0,5.8 [+ or -] 1.8, and 6.7 [+ or -] 2.2 cm/year.

Boys' characteristics, household income, and BLLS aspredictors of growth measures. In multivariate models, for each 1-kgincrease in birth weight, boys had significantly higher estimatedz-scores (95% CI) for BMI and height of 0.65 (0.35, 0.94) and 0.63(0.41, 0.85), respectively, over 4 years of follow-up. Boys from thelowest household income category had significantly lower estimated BMIz-score (95% CI) of --0.38 (--0.69, --0.07). Preterm birth wasassociated with significantly greater estimated height z-score (95% CI)over 4 years of follow-up of 0.59 (0.19, 0.98). High BLL was associatedwith significantly lower height z-score (95% CI) of --0.44 (--0.67,--0.21). However, BLL was not significantly associated with a change inBMI z-score.

Multivariate associations of serum OCPs with growth measures. Inboth univariate [see Supplemental Material, Table 1(http://dx.doi.org/10.1289/ehp.1 103743)] and multivariate models, boyswith higher serum HCB, f3 HCH, and p,p'-DDE concentrations hadlower mean BMI z-scores over 4 years of follow-up (Table 3, Figure 2A).At 12 years of age, the adjusted mean (95% CI) BMIs were 16.2 (15.5,16.9) and 18.9 (18.0, 20.0) kg/rn2 for the highest and lowestp,p'-DDE quintiles, respectively. The pattern for both HCB and[bate]HCH showed a linearly decreasing trend over quintiles until thehighest quintile, where there was either a plateau ([bate]HCH) or areversal (HCB) (Table 3). Sensitivity analyses including quintiles ofTEQs, DLCs, or PCBs in the model had minimal impact on the OCPassociations (data not shown), although including PCBs attenuated theOCPs' associations with BMI z-scores (Figure 2A). In the modelincluding all three OCPs, each individual esti mare was attenuated,although significant associations of [bate]HCH and p,p'-DDE withBMI z-scores remained (see Supplemental Material, Table 2).

Table 3. Association of OCP with measures of growth overyears of follow-up in boys from the Russian Children's Study's(a) (m=350) HCB (b) [beta]HCH (c)Growth Estimate p-Value Estimate p-Valuemeasure/quintile (95 CI) (95%)of exposureAnnual whoage-adjusted BMIz-scores (n=350)Quintile 1 Reference Reference(lowest)Quintile 2 -0.36 0.06 -0.61 0.001 (-0.73, (-O98. 0.02) -0.24)Quintile 3 -0.70 < 0.001 -1.09 < 0.001 (-1.07, (-1.46, -0.32) -0.76)Quintile 4 -1.30 < 0.001 -1.33 < 0.001 (-1.68, (1.70, -0.91) -0.97)Trend test -0.84 < 0.001 -1.32 < 0.001 (-1.23, (-1.70, -0.16) -0.95)) < 0.001 < 0.001AnnUal WHO Reference Referenceage-adjustedheight z-scores(n = 345)Quintile 1(lowest)Quintile 2 -0.25 0.09 -0.24 0.11 (-0.55. (-0.55. 0.04) 0.061)Quintile 3 -0.04 0.81 -0.21 0.18 (-0.33, (-0.54, 0.26) 0.08)Quintile 4 -0.33 0.03 -0.41 0.006 (-0.63, (-0.72, -0.03) -0.12)Quintile 5 -0.19 0.22 -0.28 0.08(highest) (-0.49, (-0.59, 0.11) 0.02) 0.18 0.03Annual HV (ri =329)Quintile 1 Reference Reference(lowest)QuntiIe 2 0.13 0.20 -0.06 0.57 (-0.07. (-0.26, 0.33) 0.14)Quintile 3 0.06 0.60 0.05 0.64 (-0.14, (-0.16, 0.25) 0.26)Quintile 4 -0.09 0.37 -0.16 0.13 (-0.29, -0.36, 0.11) 0.04) -0.05 0.66 -0.03 0.81 (-0.25, (-0.23, 0.16) 0.18) 0.19 0.47 p.p-DDE (d)Growth Estimate p-Valuemeasure/quintile (95% CI)of exposureAnnual whoage-adjusted BMIz-scores (n=350)Quintile 1 Reference(lowest)Quintile 2 -0.75 < 0.00 (-1.12, 1 -0.38)Quintile 3 -1.19 < 0.00 (-1.56. 1 -0.82)Quintile 4 -1.10 < 0.00 (-1.48, 1 -0.72)Trend test < 0.00 1AnnUal WHOage-adjustedheight z-scores(n = 345)Quintile 1(lowest)Quintile 2 -0.25 0.09 (-0.53, 0.05)Quintile 3 -0.24 0.20 (-053. 0.05)Quintile 4 -0.52 < 0.00 (-0.81, 1 -0.22)Quintile 5 -0.69 < 0.00(highest) (-1.00, 1 -0.39) < 0.00 1Annual HV (ri =329)Quintile 1 Reference 0.13(lowest)QuntiIe 2 -0.15 0.64 (-0.34, 0.06)Quintile 3 -0.05 0.64 (-0.25. O.16)Quintile 4 -0.24 0.02 (-0.45. -O.04)(a)Mixed-effects repeated measures rec'-ession model adjustedfor age, birth weight, gestational age, household income, totalcalories consumed, percent calories from carbohydrate, protein,and fat, arid BLL.(b) HCB quintiles (Q1--Q5, ng/g lipid): Qi, 31--98, 0.2, 99--135;Q3, 136--184; Q4, 185--282; Q5, 283--2,660.(c)[bate]HCH quintiles (0.1--0.5, ng/g lipid): 0.1, 39--104; Q2,105--144; Q3, 145--196; Q4, 197--302; 0.5,303--2,860. (d)PP'. DDEquintiles (0.1--Q5, ng/g lipid): 0.1,48--172; 0.2, 173--246;0.3,247--354; 0.4,355--549; Q5, 550--9,370. eRedliced number becauseat least two consecutive measures are required for calculation ofchange in height (e.g., HV).

Similar to the univariatc associations [see Supplemental Material,Table 1 (http://dx.doi.org/10.1289/ehp.1103743)], higher serum p,p'-DDE concentrations were associated with iowcr height z-scoresover 4 years of follow--up, with a monoronic trend (Table 3). At 12years of age, the adjusted mean (95% CI) heights were 146.8 (144.6,149.1) and 151.5 (149.3, 153.7) cm for the highest and lowest p,p'--DDE quintiles respectively. Adjustment for total TEQs and DLCsdid not affect these associations (data flOt shown); however, adjustingfor PCBs attenuated the associations between p,p -DDE quin tiles withheight z-scores (Figure 2B). Height showed a significant decreasingtrend with higher quintiles of [bate]HCH, similar to the univariateassociations (see Supplemental Material1 Table 1), but this trend becamenonsignificant after adjustment for other OCPs. Although in theunivariate model higher serum HCB was associated with lower heightz-scores (see Supplemental Material, Table 1), there was no consistentassocianon in the multivariate model between HCB quintiles with heightz-scores (Table 3). In the multiple OCP model, there was no change inthe association between p,p'-DDE quinriles with height z-scores(see Supplemental Material, Table 2).

After adjustment for covariates, boys with the highest quintile ofp,p'-DDE had a mean (95% CI) HV that was significantly decreased by--0.22 (--0.43, --0.01) cm/year over 4 years compared with the lowestquinti1e with a nonlinear dose response (Table 3). After furtheradjustment for either pubertal stage or PCB quintiles, p,p'-DDE wasno longer significandy associated with HV (data not shown). Neitherserum HCB nor [bate]HCH concentrations were associated with HV over 4years of follow-up (Table 3).

In sensitivity analyses that included parental height and BMI andboy's pubertal status, the associations were statisticallysignificant a in the expected positive directions; however, the observedassociations of OCPs with growth were not affected (data not shown);thus, we did not include these parental and pubertal status measures inthe Final models. We did not find evidence that age modified OCPs'associatiOnS with growth (data not shown).

Discussion

In the present study, Russian boys with higher peripuhertal serumOCPs had lower age-adjusted BMI z-scores OVer 4 years of follow-up.These associations persisted alter adjustment for serum DLCs, PCBs, andBLL At 1 2 years of age, the difference in estimated BMI was 2.7 kg/rn2lower among boys in the highest than in the lowest p,p'-DDEquintile. In addition, higher serum p,p'-DDE was associated withlower height z-scores over 4 years, independent of serum DLCs, PCBs, andlead. At 2 years of age, the difference in estimated height was 4.7 cmlower for boys in the highest than in the lowestp,p'-DDE quintile.

During pubertal maturation, especially among boys interference withthyroid hormones, insulin--like growth factor 1 (IGF-l), andtestosterone signaling may affect linear growth and weight gain. Inanimal studies, DDE was shown to act as an antiandrogen (Kelce andWilson 1997). In a study among Spanish children, DDE was associated withlower serum ICF--1 (Zumbado et al. 20 1 0). In several human studies,DDE, [bate]HCH, and HCB were associated with thyroid hormone disruption(Pearce and Braverman 2009).

Most prior epidemiologic studies have not found an association ofserum DDT or DI)E with height (Cupul-Uicah et al. 2010; Dhooge et al.2010; Gladen et al. 2004; jusko et al. 2006; Karmaus et al. 2009; Pan etal. 2010; Verb Verulst et al. 2009). All of' these studies, withone exception (Dhooge et al. 2010), measured a different window ofexposure. that is, prenatal or lactational exposure to DDT or DDE. Although studies using historical (1959--1967) (C laden et al. 2004; Juskoet al. 2006) or agricultural (Cupul-Uicab et al. 201 0) exposures hadhigher DDE exposures than ours, others were conducted after DDT wasbanned, so exposure levels were lower than our cohort. Mans' ofthese studies followed their cohorts only through the prepubertal periodand did not extend Followup through the adrenarchal hormone--mediatedincrease in weight or the sex steroid--induccd pubertal growth spurt (Cupul--Uicab et al. 2010; Jusko et al. 2006; Pan et al. 2010; Verhulstet al. 2009), which may partially contribute to their null Findings.However, two studies Found associations of DDE with reduced height. Aprospective U.S. study (n = 1,712) observed an association of DDEconcentrations with shorter height through 7 years of age based onstored prenatal serum samples collected during a period of DDT usc(1959--1966) (Ribas-Fito Ct al. 2006). A German study (n = 343) usingestimated early childhood serum DDE concentrations and both prospectiveand retrospective growth data found higher serum DDE associated withshorter height in 8-year-old girls (Karmaus et al. 2002).

In multivariate anaiysis, we found higher serum [3HCHconcentrations associated with lower height z-scorcs over 4 years offollow-up. However, after adjusting for all serum OCPs, the associationwas no longer significant. To our knowledge, only one previous small (n= 12) study of hospitalized children in the Aral Sea region examined,and did not find, a relationship of serum I3HCH with childhood lineargrowth (Mazhitova et ai. 1998).

We observed significant negative associations or all OCPs withlower BMI z-scores over 4 years of follow-up. Two cross-sectionalstudies found associations of childhood serum DDE (Mazhitova et al.1998) and HCB (a Flemish cohort1 n = 1,679; Dhooge et al. 2010) withlower BMI. However, most of the published literature concerns studies ofprenatal orllacrational OCP exposure, compared with our peripuhertalmeasurements, and found either positive associations between prenatalOCP concentrations and childhood 13M1 (Gladen et al. 2000; Mendez et aI.2011; Smink et aI. 2008; Verhulst et al. 2009) and adult weight (Karmauset at. 2009) or null associations between prenatal (Cupul--Uicah et al.20 1 0; Gladen et al. 2004; Jusko et al. 2006) or lactational (Gladen etal. 2000; Pan et al. 2010) OCP concentrations and BMI. Populations instudies that reported null associations of prenatal DDE with BMI hadhigher levels than our pcripubertal DDE concentrations. In thosestudies, their lowest prenatal DDE categories were comparable to thehighest quintile in our cohort. However, most studies that reportedpositive associations between prenatal OCPs and BMI had lower OCPconcentrations than our cohort, comparable to our lower quintiles ofexposure. Interestingly, Mendez et al. (20 11) reported positiveassociations between BMI and DDE concentrations up to 750 ng/g lipid butobserved a decrease in BMI when DDE COncentrations exceeded 750 ng/glipid, comparable to the median in our highest DDE quinrile. In ourcohort, we did not find any evidence of a nonlinear dose--responserelationship between DDE and BMI z-scores. 'These contradictoryresults across studies suggest that range and timing of exposure may beimportant factors for OCPs' association with BMI.

An ahernative explanation fOF asSociations of higher serum OCPconcentrations with lower BMJ Z-SCOrCS is that serum OCPs may he lowerin those with a larger body mass because of a greater volume ofdistribution and sequestration in adipose tissue. A dilutional effect ofincreased growth (Wolff et al. 2005) on serum OCP concentrations wouldindicate reverse causation whereby' BMJ led to a decrease in OCPconcentrations, rather than a causative effect of OCPs on BMI. Thereverse causation hypothesis may be consistent with the strongerassociations observed between OCPs and BMI z-scores than between OCP andheight z-scores. However, these differences may also reflect differencesin biological effects of the individual OCPs on BMI and height, ratherthan dilittionaL effects. If prenatal and early IifC exposures, throughplacental transfer and breast-feeding, were the primary source of OCPexposure, then dilutional effects would be a more plausible explanationfor the associations we observed. However, in our cohort there isongoing OCP exposure from local soil and foods (Ecological AnalyticalCenter 2007). It was not possible to estimate the contributions fromthese sources to the boys' serum concentrations. n the presentanalysis, we cannot determine whether associations between serum OCPsand BMI z-scores were attributable to differences in body composition,such as reduced lean muscle mass versus body fat because BMI is a crudeapproximation of body fat. Prenatal exposure to dioxins and furans wasnot associated with childhood weight among the Yu-Chcng cohort but wassignificantly associated with reduced lean muscle mass (Guo et al.1994). Therefore, alterations in the ratio of lean muscle mass and bodyfat may result from exposure to some organochionne compounds. In futureanalyses, we will examine whether OCPs are associated with alterationsin body composition using longitudinal skin fold and hioclectricimpedance data.

Limitations of our study were that we did not have measures ofprenatal OCPs and may have missed a critical window for OCPs'effect on growth, although findings on prenatal exposures have beeninconsistent (Eskenazi et al. 2009; Mendcz er aI. 2011; Smink et al.2008). Moreover, our childhood measures of serum OCPs may he a surrogatefor prenatal exposure because childhood levels of lipophilic compoundsoften track closely with prenatal levels (Patandin et al. 1999),especially in a primarily breast-fed population such as ours. Also,there was likely continuing environmental exposure to these compounds inthis community; thus, the boys' serum concentrations reflect bothpre- and postnatal exposures.

Our results from the Russian Children's Study provide evidencethat OCPs affect children's growth during the critical peripubertalperiod. These compounds, especially p,p'-DDE, were associated withlower height and BMI z-scores, even after adjustment for otherenvironmental exposures and known predictors. Childhood exposure toOCPs, especially DDE, is still a public health concern because of theirenvironmental persistence and the continued use of DDT in somecountries. OCPs may affect children's growth by affecting hormonesassociated with growth (Kelce and Wilson 997; Li et al 2008; Pearce andBraverman 2009; Zumbado et al. 2010) and body composition (Guo et al.1994; Smink et al 2008). our future research in Chapaevsk will examinewhether these compounds are associated with alterations in the ratio offat to muscle mass and growth-associated hormones.

REFERENCES

Barr JR. Maggio VL Barr DB, Turner WE, Sjodin A, Sandau CU, et al.2003. New high-resolution mass spectrometric approach for themeasurement of polychiorinated biphenyls and organochiorine pesticidesin human serum. J Chromatogr B 794(1):137--148.

Burns JS, Williams PL, Sergeyev 0, Korrick S. Lee MM, Revich B, etal. 2009. Predictors of serum dioxins and PCBs among peripubertalRussian boys. Environ Health Perspect 117:1593--1599.

Burns JS, Williams PL, Sergeyev 0, Korrick 5, Lee MM, Revich B, etal. 2011. Serum dioxins and polychiorinated biphenyls are associatedwith growth among Russian boys. Pediatrics 127(1):e59--e68,

Carrizo U, Grimalt JO, Ribas-Fito N, Sunyer J, Torrent M. 2006.Physical-chemical and maternal determinants of the accumulation oforganochiorine compounds in four-year-old children. Environ Sci Technol40(5):1420--1426,

CDC (Centers for Disease Control and Prevention). 2009. GrowthCharts. Available: http://www.cdc.gov/growthcharts/zscore. htm [accessed22 June 2011].

Colles A, Koppen G, Hanot V. Nelen V. Dewolt MC, Noel E, et al.2008. Fourth WHO-coordinated survey of human milk for persistent organicpollutants (POPs): Belgian results. Chemosphere 73(6):907--914.

Cupul-Uicab LA, Hernandez-Avila M, Terrazas-Medina EA, LongneckerMP. 2010. Prenatal exposure to the major DDT metabolitel,1-dichloro-2,2-bis(p-chlorophenyl)ethyl ene (DDE) and growth in boysfrom Mexico. Environ Aes 1 i0(6):595--603,

Darnerud P0, Aturna S, Aune M, Berselius R, Glynn A, Grawe KP, etal. 2006. Dietary intake estimations of organohalogen contaminants(dioxins, PCB, PBDE and chlorinated pesticides, e.g. 0D11 based onSwedish market basket data. Food Chem Toxicol 44(9):1597--1606,

de Onis M, Onango AW, Borghi E, Siyam A, Nishida C, Siekmann J.2007, Development of a WHO growth reference for school-aged children andadolescents. Bull World L Heafth Organ 85(9):660--667.

Dhooge W, Den Hond E, Koppen G, Bruckers 1, Nelen V, Van De MieroopE, et al. 2010. Internal exposure to pollutants and body size in Flemishadolescents and adults: associations and dose-response relationships.Environ mt 36:330-337.

Ecological Analytical Center. 2007. Complex Assessment of theEnvironment for Dioxins and Other Pollutants in Chapaevsk, Samara Region[in Russiani. Final Report. Moscow:Ecoloqical Analytical Center.

Eggesbo M, Stigum H, Longnecker MP, Polder A, Aldrin M, Basso 0, etal. 2009. Levels of hexachlorobenzene (HCB) in breast milk in relationto birth weight in a Norwegian cohort. Environ Res 109(5):559-566.

Eskenazi B, Chevrier J, Rosas LG, Anderson HA, Bornman MS, BouwnmanH, et al. 2009. The Pine River Statement: human health consequences ofDOT use. Environ Health Perspect 117: 1359-1367.

Food and Nutrition Board, Institute of Medicine. 2005.Macronutrients and healthful diets. In: Dietary Reference Intakes forEnergy, Carbohydrate, Fiber, Fatty Acids, Cholesterol, Protein, andAmino Acids. Washington, DC:National Academy Press, 769-879.

Gladen BC, Kiebanoff MA, Hediger ML, Katz SH, Barr DB, Davis MO, etal. 2004. Prenatal DOT exposure in relation to anthropometric andpubertal measures in adolescent males. Environ Health Perspect 112:1761--1767.

Gladen BC, Ragan NB, Rogan WJ. 2000. Pubertal growth anddevelopment and prenatal and lactational exposure to polychlorinatedbiphenyls and dichiorodiphenyl dichioroethene. J Pediatr136(4):490--496.

Griffin JT, Hollingsworth TD, Okell LC, Churcher TS, White M,Hinsley W, et al. 2010. Reducing Plasmodium falciparum malariatransmission in Africa: a model-based evaluation of interventionstrategies. PLoS Med 7(8):e1000324; doi:10.i371/ Journal.pmed.1000324[Online 10 August 20101.

Guo YL, Lin CJ, Yao WJ, Ryan JJ, Hsu CC. 1994. Musculoskeletal changes in children prenatally exposed to polychlorinated biphenyls andrelated compounds (Yu-Cheng children). J Toxicol Environ Health41(1):83--93.

Hauser R, Sergeyev 0, Korrick St Lee MM, Revich B, Gitin E, et al.2008. Association of blood lead levels with onset of puberty in Russianboys. Environ Health Perspect 116:976--980,

Hauser R, Williams P, Altshul L, Korrick S, Peeples L, Patterson DGJr, et al. 2005. Predictors of serum dioxin levels among adolescent boysin Chapaevsk, Russia: a cross-sectional pilot study. Environ Health4(1):8; dai:10.1186/1476-069X-4-8 [Online 26 May 2005].

Jusko TA, Koepsell TO, Baker RJ, Greenfield TA, Wiliman EJ, CharlesMJ, et al. 2006. Maternal DOT exposures in relation to letal and 5-yeargrowth. Epidemiology 17(6):692--700.

Karmaus W, Asakevich St lndurkhya A, Witten J, Kruse H. 2002.Childhood growth and exposure to dichiorodiphenyl dichloroethene andpolychlorinated biphenyls. J Pediatr 140(1 ):33--39.

Karmaus W, DeKoning EP, Kruse H, Witten J, Osius N. 2001. Earlychildhood determinants of organochiorine concentrations in school-agedchildren. Pediatr Res 50(3):331- 336.

Karmaus W, Osuch JR, Eneli I, Mudd LM, Zhang J, Mikucki U, et al.2009. Maternal levels of dichlorodiphenyl-dichioroethylene (DDE) mayincrease weight and body mass index in adult female offspring. OccupEnviron Med 66(3):143--149.

Kelce WR, Wilson EM. 1997. Environmental antiandrogens:developmental effects, molecular mechanisms, and clinical implications.J Mol Med 75(3):198--207.

Landrigan PJ, Kimmel CA, Correa A, Eskenazi B. 2004.Children's health and the environment: public health issues andchallenges for risk assessment. Environ Health Perspect 112:257--265.

Lee MM, Sergeyev 0, Williams P, Korrick S, Zeilert V, Revich B, etal, 2001 Physical growth and sexual maturation of boys in Chapaevsk,Russia. J Pediatr Endocrinol Metab 16(2)169--178.

Li J, Li N, Ma M, Giesy JP, Wang Z. 2008. In vitro profiling of theendocrine disrupting potency of organochiorine pesticides. Toxicol Lett183(1--3):65--7l.

Link B, Gabrio T, Zoeliner I, Piechotowski I, Paepke 0, Herrmann T,et al. 2005. Biomonitoring of persistent organochlorine pesticides,PCDD/PCDFs and dioxin-like PCBs in blood of children from south westGermany (Baden-Wuer-ttembergl from 1993 to 2003. Chemosphere58(911185--1201.

Longnecker MP. 2005. Invited Commentary: why OUT matters - now, AmJEpiderniol 162(8):726--728.

Marti-Cid R, Liobet JM, Castell V, Domingo JL. 2008. Human dietaryexposure to hexachlorobenzene in Catalonia, Spain. J FoodProt7l(10):2148--2152.

Martinchik AN, Baturin AK, Baeva VS, Feoktistova Al, PiatnjtskajaIN, Azizbekjan GA, et al. 1998. Development of a method of studyingactual nutrition according to analysis of the frequency of consumptionof food products: creation of a questionnaire and general evaluation ofthe reliability of the method [in Russiani. Vopr Pitan (3):8--13.

Mazhitova Z, Jensen S, Ritzen M, Zetterstrom R. 1998. Chlorinatedcontaminants, growth and thyroid function in schoolchildren from theAral Sea region in Kazakhstan. Acta Paediatr 87(9):991--995.

McDowell MA, Fryar CD, Ogden CL 2009. Anthropometric reference datafor children and adults: United States, 1988--1994. Vital Health Stat11(249):1--68.

Mendez MA, Garcia-Esteban R, Guxens M, Vrijheid M, Kogevinas M,Goni F, et al. 2011. Prenatal organochlorine compound exposure, rapidweight gain and overweight in infancy. Environ Health Perspect119:272--278.

National Vital Statistics Reports. 2007. Births: Final Data for2005. Hyattsville, MD:National Center for Health Statistics.

Pan IJ, Daniels JL, Herring Aft, Rogan WJ, Siega-Riz AM, GoldmanBD, et al. 2010. Lactational exposure to polychlorinated biphenyls,dichiorodiphenyltrichloroethane, and dichlorodiphenyldichloroethyleneand infant growth: an analysis of the Pregnancy, Infection, andNutrition Babies Study. Paediatr Perinal Epidemiol 24(3):262--271.

Patandin S, Dagnelie PC, Mulder PG. Op de CouP E, van der Veen JE,Weisgias-Kuperus N, et al. 1999. Dietary exposure to polychiorinatedbiphenyls and dioxins from infancy until adulthood: a comparison betweenbreast-feeding, toddler, and long-term exposure. Environ Health Perspect107:45--51.

Patterson DO Jr, Wang LV, Turner WE, Caudill SP, Dipietro ES,

McClure PC, et al. 2009. Levels in the U.S. population of thosepersistent organic pollutants (2003--2004) included in the StockholmConvention or in other long range transboundary air pollutionagreements. Environ Sci Technol 43(4):121 1--1218.

Pearce EN, Braverman LE. 2009, Environmental pollutants and thethyroid. Best Pract Res Clin Endocrinol Metab 23(6):801--813.

Revich B, Brodsky E, Sotskov Y. 1999. Dioxin in environmental,blood, breast milk, cow milk in Chapaevsk town. Organohalogen Compounds44:229--232.

Ribas-Fito N, Gladen BC, Brock JW, Kiebanoff MA, Longnecker MP.2006. Prenatal exposure to 1,1-dichioro- 2,2-his(p-chlorophenyl)ethylene (p,p'-DDE) in relation to child growth.lntJ Epidemiol 35(4):853--858.

Ribas-Fito N, Sala M, Cardo E, Mazon C, IJe Muga ME, Verdu A, etal. 2002. Association of hexachlorobenzene and other organochiorinecompounds with anthropometric measures at birth. Pediatr Res52(2):163--167.

Rockett HR, Breitenbach M, Frazier AL, Witschi J, Wolf AM, FieldAE, et al. 1997. Validation of a youth/adolescent food frequencyquestionnaire. Prey Med 26(6):808--816.

Siddiqui MK, Srivastava S, Srivastava SP, Mehrotra PK, Mathur N.Tandon I. 2003. Persistent chlorinated pesticides and intra-uterinefoetal growth retardation: a possible association. mt Arch Dc cupEnviron Health 76(1):75--80,

Sjodin A, Jones RS, Lapeza CR, Focant JF, McGahee EE Ill, PattersonDO Jr. 2004. Semiautomated high-throughput extraction and cleanup methodfor the measurement of polybrominated diphenyl ethers, polybrominatedbiphenyls, and polychiorinated biphenyls in human serum. Anal Chem76(7):1921--1927.

Smink A, Ribas-Fito N, Garcia R, Torrent M, Mendez MA, Grimalt JO,et al. 2008. Exposure to hexachlorobenzene during pregnancy increasesthe risk of overweight in children aged 6 years. Acta Paediatr 97(10):1465--1469. Tanner JM, Whitehouse RH. 1976. Clinical longitudinalstandards for height, weight, height velocity, weight velocity, andstages of puberty. Arch Dis Child 51: 170-179.

Turner W, DiPietro E, Lapeza C, Green V, Gill J, Patterson DGJ.1997. A fast universal automated cleanup system for the isotope-dilutionhigh-resolution mass spectrometric analysis of PCDOs, PCDFs, coplanarPCBs, PCB congeners, and persistent pesticides from the same serumsample. Organohalogen Compounds 31:26--31.

United Nations Environmental Programme. 2009. Stockholm Conventionon Persistent Organic Pollutants. Available:http://chm.pops.jnt/Convention/tabjd/54/Default.aspx [accessed 29 June20111.

Verhuist SL, Nelen I, Hond ED, Koppen G, Beunckens C, Vael C, etal. 2009. Intrauterine exposure to environmental pollutants and bodymass index during the first 3 years of life. Environ Health Perspect117:122-126.

Weisskopf MG, Anderson HA, Hanrahan LP, Kanarek MS, Falk CM,Steenport OM, et al. 2005. Maternal exposure to Great Lakes sport-caughtfish and dichlorodiphenyl dichloroethylene, but not polychiorinatedbiphenyis, is associated with reduced birth weight. Environ Res97(2):149-162.

WHO (World Health Organization). 2011. The WHO Child GrowthStandards. Available: http://www.who.int/childgrowth/en/ [accessed 22June 2011].

Williams PL, Sergeyev 0, Lee MM, Korrick SA, Burns JS, Humblet 0,et al. 2010. Blood lead levels and delayed onset of puberty in alongitudinal study of Russian boys. Pediatrics 125(5):e1088-e1096

Wolff MS. Britton JA, Teitelbaum SL, Eng S, Deych E, Ireland K, etal, 2005. Improving organochiorine biomarker models for cancer research.Cancer Epidemiol Biomarkers Prey 1 4(9):2224-2236.

Wolff MS, Engel S. Berkowitz G, Teitelbaum S, Siskind J, Barr OB,et al. 2007. Prenatal pesticide and PCB exposures and birth outcomes.Pediatr Res 61(2):243-250.

Wolff MS, Zeleniuch-Jacquotte A, Dubin N, Toniolo P. 2000. Risk ofbreast cancer and organochlorine exposure. Cancer Epidermol BiomarkersPrey 9(3):271-277.

Zumbado M, Luzardo OP. Lara PC, Alvarez-Leon EE, Losada A,Apolinaria ft et al. 2010. Insulin-like growth factor-I (IGE-I) serumconcentrations in healthy children and adolescents: relationship tolevel of contamination by DOT-derivative pesticides. Growth Horm IGF Res20(1):63-67.

Address correspondence to J.S. Burns, Environmental andOccupational Medicine and Epidemiology Program, 665 Huntington Ave.,Building I, Room 1404E, Boston, MA 02115 USA. Tdcphone: (617) 432-1829.Fax: (617) 432-3441. E-mail: jburns@ hsph.harvard .edu

Supplemental Material is available online(http://dx.doi.org/10.1289/ehp. 1103743).

This work was funded by the U.S. Environmental Protection Agency(R82943701) and the National Institute of Environmental Health Sciences(ES014370, E5000002, and ESO17I 27). M.M.L. is a member of the UMassDiabetes and Endocrine Research Center (DK32520).

The opinions expressed in this article are those of the authors anddo not necessarily reflect the official opinion of the Centers forDisease Control and Prevention.

L.A. is cmployed by Environmental Health and Engineering, inc.(Needham, MA, USA). D.G.P. is employed by Axys Analytical Solutions(Sidney, BC, Canada), EnviroSolutions Consulting, Inc. (Auburn, GA,USA), Exponent, Inc. (Maynard, MA, USA), and Fluid Management Systems(Boston, MA, USA). The other authors declare they have no actual orpotential competing financial interests.

Received 28 March 2011; accepted 7 October 2011.

Jane S. Burns, (1) Paige L. Williams, (2) Oleg Sergeyev, (3), (4)Susan A. Korrick, (1), (5) Mary M. Lee, (6) Boris Revich, (7) LarisaAltshul, (8), (9) Julie T. Del Prato, (1) Olivier Humblet, (10) DonaldG. Patterson Jr., (11), (12), (13), (14), (15) Wayman E. Turner, (11)Mikhail Starovoytov, (16) and Russ Hauser (1)

(1) Environmental and Occupational Medicine and EpidemiologyProgram, Department of Environmental Health, and (2) Department ofBiostatistics, Harvard School of Public Health, Boston, Massachusetts,USA; (3) Department of Physical Education and Health, Samara StateMedical University, Samara, Russia; (4)Chapaevsk Medical Association,Chapaevsk, Russia; (5)Channing Laboratory, Department of Medicine,Brigham and Women's Hospital, Harvard Medical School, Boston,Massachusetts, USA; (6) Pediatric Endocrine Division, Departments ofPediatrics and Cell Biology, University of Massachusetts Medical School,Worcester, Massachusetts, USA; (7) Department of Environmental Health,Institute for Forecasting, Russian Academy of Sciences, Moscow, Russia;(8) Environmental Health and Engineering, Inc., Needham, Massachusetts,USA; (9) Exposure, Epidemiology, and Risk Program, Department ofEnvironmental Health, Harvard School of Public Health, Boston,Massachusetts, USA; (10) Division of Immunology and Allergy, Departmentof Pediatrics, Stanford University, Stanford, California, USA; 11Centersfor Disease Control and Prevention, Atlanta, Georgia, USA; (12)EnviroSolutions Consulting, Inc., Auburn, Georgia, USA; (13) AxysAnalytical Solutions, Sidney, British Columbia, Canada; (14) FluidManagement Systems, Boston, Massachusetts, USA; (15) Exponent, Inc.,Maynard, Massachusetts, USA; 16Russian Institute of Nutrition, Moscow,Russia