Dioxin Exposure and Age of Pubertal Onset among Russian Boys

Background: Animal data demonstrate associations of dioxin, furan, and polychlorinated biphenyl (PCB) exposures with altered male gonadal maturation. It is unclear whether these associations apply to human populations. Objectives: We investigated the association of dioxins, furans, PCBs, and corresponding toxic equivalent (TEQ) concentrations with pubertal onset among boys in a dioxin-contaminated region. Methods: Between 2003 and 2005, 499 boys 8–9 years of age were enrolled in a longitudinal study in Chapaevsk, Russia. Pubertal onset [stage 2 or higher for genitalia (G2+) or testicular volume (TV) > 3 mL] was assessed annually between ages 8 and 12 years. Serum levels at enrollment were analyzed by the Centers for Disease Control and Prevention, Atlanta, Georgia, USA. We used Cox proportional hazards models to assess age at pubertal onset as a function of exposure adjusted for potential confounders. We conducted sensitivity analyses excluding boys with pubertal onset at enrollment. Results: The median (range) total serum TEQ concentration was 21 (4–175) pg/g lipid, approximately three times higher than values in European children. At enrollment, boys were generally healthy and normal weight (mean body mass index, 15.9 kg/m2), with 30% having entered puberty by G2+ and 14% by TV criteria. Higher dioxin TEQs were associated with later pubertal onset by TV (hazard ratio = 0.68, 95% confidence interval, 0.49–0.95 for the highest compared with the lowest quartile). Similar associations were observed for 2,3,7,8-tetrachlorodibenzo-p-dioxin and dioxin concentrations for TV but not G2+. Results were robust to sensitivity analyses. Conclusions: Findings support an association of higher peripubertal serum dioxin TEQs and concentrations with later male pubertal onset reflected in delayed testicular maturation.


Introduction
The transition from pre-puberty to sexual maturity entails rapid physical, hormonal, and behavioral development. Alterations in the timing of pubertal onset or pace of its progression can adversely impact not only physical and sexual maturation but also social, cognitive and behavioral development as well as adult health (Graber et al. 2004;Michaud et al. 2006). For example, earlier puberty incurs risk for metabolic syndrome and obesity in later life and delayed puberty is associated with decreased bone mineral density in adults raising concerns about increased fracture risk (Biro et al. 2003;Finkelstein et al. 1996;Van Lenthe et al. 1996).
In recent decades, suggestive evidence of earlier onset of breast development and age at menarche has been observed in girls but data in boys are limited (Biro et al. 2010;Euling et al. 2008;Herman-Giddens et al. 1997;Herman-Giddens et al. 2001;Sorensen et al. 2010).
Explanations for this possible trend include changes in diet and activity, and environmental exposures. Chemicals that can disrupt gonadal steroidogenesis and neuroendocrine pathways, such as organochlorine pollutants including dioxins, furans, and polychlorinated biphenyls (PCBs), are of particular concern (Jacobson-Dickman and Lee 2009; Schoeters et al. 2008).
Despite efforts to limit dioxin emissions, and longstanding bans on PCB manufacture and use, human exposure is ongoing, primarily through diet. For example, fish and dairy are potential exposure sources due to organochlorines' lipophilic properties, long half lives, on the order of years to decades, and propensity to bioconcentrate (Schecter et al., 2001).
Gestational or lactational dioxin exposures in animals have been consistently associated with delayed male pubertal onset (Bell et al. 2007;Hamm et al. 2003;Theobald et al. 2003).
However, the few human epidemiologic studies of organochlorine exposures and puberty in boys did not examine onset and had inconsistent findings. Where direct exposure biomarkers were used, dioxins, furans or PCBs were associated with no differences in (Den Hond et al. 2002;Gladen et al. 2000;Leijs et al. 2008) or earlier (Den Hond et al. 2010) male genital maturation among older adolescents but inferences were often limited by small sample sizes (Den Hond et al. 2002;Leijs et al. 2008).
Chapaevsk, Russia is an industrial town contaminated with dioxins consequent to past production of chemical warfare agents and recent production of chlorinated chemicals. Waste from these industries resulted in pervasive dioxin contamination of soils, water, and local food (Revich et al. 1999;Sergeyev et al. 2007). Community concerns regarding potential health hazards from this contamination led to a study of dioxins and male pubertal development.

Study Population
The Chapaevsk study is an ongoing prospective study of 499 generally healthy boys (Hauser et al. 2008 kg/m 2 ) was calculated from measured height and weight. Pubertal maturation was graded from 1 to 5 by visual inspection according to established criteria (Tanner and Whitehouse 1976).
Testicular volume (TV) was measured using an orchidometer. Pubertal onset was defined as stages 2 or higher for genitalia (G2+) or testicular volume (TV) > 3 ml for either testis.

Questionnaire Assessment:
At enrollment, each mother/guardian completed a nurse-administered questionnaire ascertaining the child's birth and medical history, breastfeeding status, physical activity; family demographics, income, residential history; and parental reproductive and medical history, occupation, education, smoking and alcohol consumption. Birth weight and gestational age were obtained from medical record review. Diet was ascertained using a food frequency questionnaire modified from a validated Russian Institute of Nutrition instrument (Burns et al. 2009;Martinchik et al. 1998).

Organochlorine Exposure Assessment:
Fasting blood samples were collected prior to baseline examination and the serum fraction was stored at -35 o C until shipment for analysis at the National Center for Environmental Health at the Centers for Disease Control and Prevention, Atlanta, GA. Analytes included 7 polychlorinated dibenzo-p-dioxins (PCDDs or dioxins), 10 polychlorinated dibenzofurans (PCDFs or furans), 4 co-planar polychlorinated biphenyls (co-PCBs), 6 mono-ortho substituted PCBs, and 31 other PCBs (non-dioxin-like PCBs) described in Burns et al. (2009).
For dioxin-like analytes, sera, method blanks, and quality control samples (aliquots of pooled bovine sera) were spiked with a mixture of 13 C 12 -labeled PCDDs/PCDFs and co-PCBs as internal standards and serum analytes were isolated by solid phase extraction (SPE) followed by a multicolumn automated cleanup and enrichment procedure (Turner et al. 1997). Analytes were separated on a DB-5 MS capillary column and quantified using selected-ion-monitoring (SIM) high resolution (10,000 resolving power) mass spectrometry (HRGC-ID/HRMS) (Patterson et al. 1987). Quantification by isotope dilution mass spectrometry used calibration standards containing 13 C 12 -labeled, and unlabeled analytes.
A similar approach was used for mono-ortho and non-dioxin-like PCBs (Barr et al. 2003). Samples were spiked with 13 C 12 -labeled PCBs, extracted by either large (Turner et al. 1997) or small (Sjordin et al. 2004) volume SPE, and analyzed using high resolution GC/MS in SIM (Barr et al. 2003).
For all analytes, quality control sample coefficients of variation (CVs) combining between and within run reproducibility were generally below 15%. All concentrations were expressed on a per lipid basis with serum total cholesterol and triglycerides measured enzymatically, and total lipids calculated using the Phillips equation (Phillips et al. 1989).
Congener concentrations below the limit of detection (LOD) were assigned the LOD divided by the square root of 2.
Standard Cox proportional hazards models were used to assess time to pubertal onset as a function of exposure adjusted for potential confounders. Age of pubertal onset was assigned to the midpoint between age at the previous visit and age at the visit at which onset was noted. For boys in puberty at study enrollment (n=141 by G2+; n=66 by TV>3 ml), age at onset was defined as 6 months prior to age at enrollment. Observations were censored at the last visit for boys not yet in puberty.
Sensitivity analyses were performed using both interval censored likelihood-based models and repeated measures generalized estimating equation (GEE) models. The interval censored approach does not assign a specific time of onset, but instead assumes that pubertal onset occurred in the interval between study visits. This approach was used to estimate overall mean age of pubertal onset, assuming a normally distributed age at onset, and mean age at pubertal onset for each exposure quartile, adjusted for confounders. The GEE approach was used to fit a logistic regression model for pubertal onset at each visit as a function of age at visit, with adjustment for potential confounders and correlation among multiple visits via an autoregressive structure. GEEs were also used to evaluate the impact of clustering within household for twins (4 pairs) and siblings (3 pairs). To account for possible examiner and laboratory drift over time, uncertainty regarding age of pubertal onset, and the potential for reverse causation (due to dilution of dioxin concentrations in larger, more mature boys), additional sensitivity analyses were performed excluding boys with pubertal onset at study entry and adjusting for year of organochlorine analysis.
Covariates considered in models included potential determinants of pubertal onset: child's age at exam, birthweight, gestational age, breastfeeding; nutrition, height, weight, and BMI at enrollment; household income; maternal age at birth and parity; prenatal smoking (active and second hand) and alcohol intake; parental education; and blood lead (Williams et al. 2010).
A core model was developed by first assessing the univariate relation of covariates to each pubertal onset measure and retaining those with a p-value <0.20. Covariates meeting this criterion were included in a 'full' model and then backward selection (likelihood ratio test) was used to iteratively exclude the least important covariates (retain p<0.15). Covariates were retained if they were significant for at least one pubertal onset measure, or if they resulted in a 10% or greater change in exposure effect estimates when added, one at a time, into our final model. Because height and BMI at enrollment may be proxies for pubertal onset or on the causal pathway relating dioxins with onset, sensitivity analyses were performed excluding these covariates from the final model. As mother's age at menarche was missing for 8% of participants, this covariate was added to the final models in sensitivity analyses.
The association of pubertal onset with each of the nine different exposure measures was assessed, one exposure at-a-time. These nine measures were moderately to strongly correlated (Spearman r = 0.44-0.90) so secondary analyses were performed to assess the independent relation of dioxin-like versus non-dioxin-like exposures with pubertal onset. Specifically, final models for the relation of each dioxin-like measure (total TEQ, TCDD, as well as dioxin, furan, and co-PCB TEQ and concentration measures) with pubertal onset were re-run with noncoplanar PCB concentrations (ΣPCBs) added to the models.

Demographic and exposure characteristics:
The boys were racially homogeneous (all Caucasian) and most were full term with normal birth weight and enrollment BMI (Table 1). Of the 489 boys, 473 had dioxin/furan measures and 468 had PCB measures. The median (range) total serum TEQs at 8-9 years was 21.1 (4.0 to 174.7) pg/g lipid, about three times higher than levels among similar aged European children (Table 2) Link et al. 2005).

Pubertal onset characteristics:
Most (85%) boys had four annual examinations between ages 8-11 or 9-12 with 6%, 4%, and 5%, respectively, examined three times, twice or once. At study entry (both 8-and 9-yearolds), 30% had entered puberty by G2+ and 14% by TV criteria. By age 12, most had entered puberty (92% by G2+ and 83% by TV criteria). Overall, the estimated mean (95% CI) age of onset by G2+ and TV, based on interval censored models, was 9.4 (9.2, 9.6) and 10.5 (10.3, 10.7) years, respectively. In multivariable models (results not shown), pubertal onset was significantly earlier with higher birth weight (both measures), lower gestational age (G2), higher percentage of dietary fat (TV), and greater height or BMI at the initial study visit (both measures). Conversely, pubertal onset was significantly later with maternal alcohol consumption during pregnancy (TV), low household income (TV), older maternal age at menarche (both measures), and high blood lead (both measures).

Association of dioxins, furans, and PCBs with pubertal onset:
In multivariable Cox proportional hazards models, pubertal onset was later with increasing dioxin exposure for TV but not G2+ (Table 3; Figure 1). For example, for TV > 3ml, the hazard ratio (HR) was 0.69, 95% CI: 0.48, 0.98 for the highest compared with the lowest quartile of serum TCDD; similar associations were observed for PCDD TEQs (Table 3; Figure   1). There was suggestive evidence of later pubertal onset (TV) with increasing PCDF or co-PCB concentrations but not with PCDF TEQs, co-PCB TEQs, or ΣPCB levels (Table 3; Figure 1).
Similar findings were observed in adjusted interval censored models; for example, TCDD was associated with approximately 5½ months' later onset (TV) for the highest compared with lowest quartile of serum levels (95% CI: -0.6, 11.9; p trend = 0.07) (Supplemental Material, Table 1).
In sensitivity analyses, the observed associations of dioxins with later pubertal onset by TV criteria were similar in GEE models and, in Cox models, were essentially unchanged after inclusion of maternal age at menarche, excluding boys who were pubertal at enrollment, excluding twins and siblings, or adjusting for year of serum analyses. However, adjustment for enrollment height and BMI attenuated the findings. For example, for onset by TV, the 4 th (vs. ΣPCB levels were not associated with pubertal onset in the primary study analyses (Table   3), but secondary analyses, although mostly non-significant, supported a tendency for earlier pubertal onset by both TV and G2+ criteria with increasing ΣPCBs in models adjusted for dioxin-like measures (Supplemental Material,  (Bell et al. 2007;Hamm et al. 2003;Theobald et al. 2003).
In contrast, findings have been inconsistent among previous epidemiologic studies that examined late pubertal milestones rather than pubertal onset (potentially a more sensitive endpoint) and have been limited either by small sample size, cross-sectional design, self-reported pubertal staging, or lack of an exposure biomarker. For example, 80 Belgian teenage boys living near an incinerator (a presumed dioxin exposure source) had later sexual maturity, including smaller TV, compared with boys in an unpolluted town but TV did not correlate with indirect (CALUX assay) measures of dioxin exposure (Den Hond et al. 2002). Conversely, in a large (n=887) cross-sectional study of 14-to 15-year-old Belgian boys, serum organochlorine levels (PCBs, p,p'-DDE, and hexachlorobenzene) were associated with earlier genital development  Table   3). Among 14-to 18-year-old Dutch boys, pubertal maturation, including TV, was not associated with perinatal or concurrent dioxin levels but there were only 15 boys in this study and, as with the Belgian studies, assessments focused on late stages of puberty ).
Lastly, among 244 (primarily 12-to 14-year-old) North Carolina boys, age of self-reported pubertal stages was not associated with measures of prenatal PCB or DDE exposure (Gladen et al. 2000). Although larger than most other studies, the North Carolina assessments did not include measures of dioxin, TV, or pubertal onset.
Several potential limitations impact interpretation of study findings. First, although both TV and G2+ maturation reflect hypothalamic-pituitary-gonadal (HPG) activation, associations were observed for TV but not genital staging (Table 3) with suggestive but mostly nonsignificant G2+ delays observed in association with dioxin-like measures only after adjustment for ΣPCBs. Assessment of TV by palpation and comparison with a standardized orchidometer is considered a more precise measure of gondal development and pubertal status than genital staging (Biro et al. 1995;Euling et al. 2008) with minimal intraobserver variability (Carlson et al., 2000) which could account for its greater sensitivity. In addition, testicular growth reflects both luteinizing hormone (LH) and follicle stimulating hormone (FSH) stimulation and paracrine androgen actions whereas penile and scrotal maturation (G2) are primarily influenced by circulating androgens (Macleod et al. 2010;Raivio et al. 2007). In animal models, early life dioxin exposure inhibits androgen biosynthesis and disrupts the HPG axis (Clements et al. 2009;Cooke et al. 1998;Fukuzawa et al. 2004;Kakeyam et al. 2008). If dioxins disproportionately impair gonadotropin secretion relative to androgen biosynthesis, TV could be affected more than G2.
Although the boys' mean age of pubertal onset by TV (10.5 years) is consistent with other studies (Herman-Giddens et al. 2001;Susman et al. 2010), later onset has been observed in Danish boys (Sorensen et al. 2010) and Chapaevsk boys' mean age of onset by G2 (9.4 years) is younger than reported elsewhere. Whether the apparent earlier G2 reflects a true difference between Chapaevsk boys and other populations is unclear as most data on male pubertal development is cross-sectional and/or collected at older ages (Biro et al. 1995;Sun et al. 2005;Susman et al. 2010) making comparisons difficult. But, given that study exams were all performed by a single physician, internal comparisons among study boys should be valid and unbiased.
Additional study limitations include the observation that a number of Chapaevsk boys had entered puberty prior to study enrollment (e.g., 12% of 8-year-olds and 18% of 9-year-olds by TV); it is unclear whether exposure measures obtained after pubertal onset reflect relevant exposure risk. However, the relationship of dioxins with TV persisted after exclusion of boys in puberty at enrollment. Although lower BMI and socioeconomic status were associated with both higher serum dioxins and later pubertal onset in this cohort (Burns et al. 2009) and thus may confound results, the relationship of dioxins with TV persisted after adjustment for height, BMI, and measures of socioeconomic status (Table 3).
Lastly, if dioxin-associated alterations in male pubertal development are aryl hydrocarbon receptor (AhR) mediated toxicities (Theobald et al. 2003), then the apparent stronger association of TV with PCDD TEQs compared to PCDF or co-PCB TEQs is difficult to explain. It is possible that non-AhR-mediated mechanisms are relevant (Butler et al. 2004) and congeners contributing to such mechanisms correlate better with some TEQ measures than others. Also, sources of environmental contamination with PCDDs, PCDFs or PCBs may vary, as the latter are a manufactured product rather than a byproduct of chemical production or incineration. Thus, confounding by differing unmeasured co-occurring exposures may explain apparent differential TEQ effects. The results of secondary analyses suggestive of earlier, rather than later, puberty in association with non-co-planar PCB exposures, are consistent with the possibility that non-AhR-related mechanisms may be important to organochlorine-associated alterations in male pubertal onset. Although Taiwanese boys with substantial prenatal furan and PCB exposure, had no apparent exposure-associated differences in pubertal stage (Hsu et al. 2005), concurrent PCB levels have been associated with earlier, not later, male genital development in other populations (Den Hond et al. 2010).

Conclusions
Although recent emphasis has been placed on environmental risk factors for earlier breast development and menarche in girls, environmental contaminants may also delay puberty (Selevan et al. 2003;Wu et al. 2003) and impact puberty in boys. In the present study, serum dioxins measured at age 8 or 9 years were associated with later male pubertal onset (by TV criteria). Although this was not indicative of clinically delayed onset, modest changes in the mean value of a health indicator, such as pubertal onset, can signal substantial changes in the prevalence of clinically evident disease within a population (Korrick and Bellinger 2007).  ΣPCBs e 331 ± 312 164 250 394 4248 ______________________________________________________________________________ a The median limit of detection (LOD) for TCDD was 0.60 pg/g lipid; 123 (26%) of TCDD values were less than this LOD (Burns et al. 2009).