Hypospadias, a congenital malformation in which the urethral opening is on the ventral side of the penis, is one of the most common congenital malformations, affecting four to six males per 1,000 male births [Paulozzi et al., 1997; Paulozzi, 1999; Dolk et al., 2004]. Normal urethral closure, which occurs during the 7^th–15^th weeks of pregnancy (i.e., weeks after the last menstrual period) [Sadler, 2004], depends on fetal conversion of testosterone to dihydrotestosterone, binding of dihydrotestosterone to the androgen receptor, and proper subsequent androgen receptor signaling. Given the importance of endocrine function to urogenital development, it has been proposed that environmental contaminants that disrupt endocrine function may contribute to hypospadias etiology.

Hypospadias can be induced experimentally by administration of several anti-androgenic pesticides, including vinclozolin, prochloraz and procymidone [Gray, Jr. et al., 2004; Noriega et al., 2005; Ostby et al., 1999]. A few human studies have attempted to assess the association between environmental contaminants and hypospadias. Most of these studies had limited exposure assessment (e.g., self-reported occupation), time windows of exposure that were not specific to urethral development, and/or case information that was passively reported {Pierik, 2004 1504/id; Weidner, 1998 951/id; Kristensen, 1997 481/id; Carbone, 2006 2126/id; Olshan, 1991 2127/id; Zhu, 2006 2128/id; Ormond, 2009 2474/id}. Three studies have measured maternal levels of DDE, which has known estrogenic and anti-androgenic properties, in populations with relatively high exposure [Bhatia et al., 2005; Longnecker et al., 2002; Flores-Luevano et al., 2003]. The odds ratios comparing high versus low values in these studies were 1.2 (95% confidence interval (CI) 0.6, 2.4) [Longnecker et al., 2002], 1.2 (0.5, 3.0) [Bhatia et al., 2005], and 0.5 (0.2, 1.6)[Flores-Luevano et al., 2003]. One study examined maternal serum PCBs and found that levels of individual congeners were not associated with hypospadias. Other studies observed that maternal serum PCB levels during pregnancy {McGlynn, 2009 2473/id} and historical maternal exposure to polybrominated biphenyl {Small, 2009 2475/id} were not associated with hypospadias.

Each woman’s serum specimen was linked with delivery outcome information to determine whether her fetus had a structural malformation, which was determined via medical record review by the California Birth Defects Monitoring Program [Shaw et al., 2009].

Among the more than 100,000 pregnancy specimens collected for testing in 2003 and that resulted in a liveborn delivery in 2003, we identified 63 cases with hypospadias. Among the 29 cases for which two serum specimens were available, we selected all 16 for whom the degree of severity was not specified and the four that were coronal, to give a total of 20 cases. As controls, we randomly selected 28 specimens that were collected at the same lab during the same time period and corresponded to non-malformed, liveborn male infants delivered in 2003. Thus, this was a nested case-control study. The selection of samples was limited to singletons. All samples were obtained with approval from the California Health and Welfare Agency Committee for the Protection of Human Subjects.

Specimens for the 48 subjects were sent on cold packs to the Centers for Disease Control and Prevention (CDC) for measurement in 2006. The serum samples were extracted by solid phase extraction (SPE) and after lipid removal analyzed by GC-IDHRMS (isotope dilution high-resolution mass spectrometry). The final analytic results were reported as fresh weight concentration (pg/mL serum) and lipid-adjusted concentration (ng/g lipid) [Sjodin et al., 2004b]. The measured analytes included nine polybrominated flame retardants (PBDEs), 30 polychlorinated biphenyls (PCBs) and nine persistent pesticides.

The limit of detection (LOD) was sample-specific, based on the sample volume. Values that were below the LOD were imputed based on a truncated multivariate normal distribution. This approach allowed us to incorporate information on correlations between related analytes and to restrict imputed values to being below the individual’s LOD for each particular analyte. This type of imputation approach using truncated normal distributions has been applied previously in numerous settings [Lubin et al., 2004; Uh et al., 2008].

Mean time elapsed between blood collection and AFP analysis was 2.9 days (SD 0.9) among cases and 3.5 (SD 1.2) among controls. Mean time elapsed between AFP analysis and storage at −70°C was 13.0 days (SD 2.0) among cases and 12.9 (SD 1.4) among controls. Among the controls, correlations of the studied analytes with time elapsed between sample collection and AFP testing were negative (ranging from −0.37 to −0.06), except for DDT and DDE (0.09 and 0.08, respectively) (data not shown). The confidence intervals for the coefficients all included one. Correlations of the analytes with time elapsed between AFP testing and storage at −70° C were more variable (ranging from −0.36 to 0.22), and all confidence intervals included one.

In comparisons of analyte levels among the cases and controls, means were greater for cases than controls for 11 of the 17 reported analytes (4 of 5 PBDEs, 7 of 9 PCBs, and 0 of 3 other persistent pesticides), but none of the differences were statistically significant (Table 2). Medians were greater for cases than controls for 9 of the 17 analytes. Eleven of the 17 odds ratios exceeded one (the same analytes that had greater means), but none of the confidence intervals excluded one (Table 3). After adjustment for foreign-born Hispanic race-ethnicity and time elapsed between sample collection and AFP analysis, confidence intervals still excluded one, but only four of the odds ratios exceeded one.

Hypospadias is considered part of the testicular dysgenesis syndrome, which comprises several reproductive tract abnormalities – hypospadias, cryptorchidism, testicular germ cell cancer and impaired spermatogenesis – that may share a common etiology resulting from impairment in endocrine function during fetal development [Sharpe and Skakkebaek, 2008]. Accordingly, it has been proposed that these outcomes may be associated with many of the studied contaminants, given that PCBs and the other studied persistent pesticides have known effects on androgen-dependent reproductive health outcomes [Gray, Jr. et al., 2004; Cooke et al., 2001], and evidence for similar effects of PBDEs is accumulating [Legler and Brouwer, 2003; Stoker et al., 2005]. One recent study suggested increased risk of cryptorchidism was associated with higher maternal breast milk levels of PBDEs [Main et al., 2007]. In general, however, results of studies examining potentially endocrine-disrupting chemicals and male reproductive health outcomes are equivocal [Vidaeff and Sever, 2005].

Few studies have examined the association of hypospadias with any of the analytes considered in the current analysis. Two studies examined DDE and PCBs and included third trimester maternal serum samples from women enrolled from 1959–1965, a time when levels were much higher than today {McGlynn, 2009 2473/id}{Longnecker, 2002 960/id}. One study examined DDE and DDT and included serum samples that were primarily obtained soon after delivery, for deliveries from 1959–1967 {Bhatia, 2005 1346/id}. Another study of DDE, DDT and beta-hexachlorobenzene included serum samples obtained on average more than five years after delivery, among women in Mexico {Flores-Luevano, 2003 2334/id}. Maternal exposure to polybrominated biphenyl (congener 153) in the 1970s was measured many years before pregnancy by a study that included male offspring with genitourinary conditions, including five cases with hypospadias cases {Small, 2009 2475/id}. None of the studies found strong associations with hypospadias, although an interesting finding by McGlynn et al. was a potential association of hypospadias with the sum of PCBs {McGlynn, 2009 2473/id}. Thus, the literature includes studies that are not particularly comparable with regard to time period or geography, and none collected specimens at the time of urogenital development. Also, none of them had information regarding the location of the urethral meatus.

Although the current study is limited by its small sample size, it is strengthened by its inclusion of specimens from early pregnancy and its timeliness. Other strengths include the prospective study design; utilization of laboratory techniques that enabled detection of low levels of multiple analytes in a relatively small sample volume; and the multiple imputation approach to estimation of analyte levels below the limit of detection. Limitations included the small number of subjects, low sample volume, and unspecified location of the urethral meatus in most of the selected cases. In particular, a new study with 20 cases and 28 controls would have 80% power using a type 1 error rate of 5% to detect a 0.82 SD difference in exposure levels between cases and controls. Budgetary restrictions and sample availability restricted our ability to examine more subjects. Although the number of subjects was small – and thus the study was exploratory – it is noteworthy that some expected differences were observed between cases and controls, including fewer of the mothers of cases being Hispanic, more of the mothers of cases having higher education and being nulliparous, and more of the cases being low birthweight, relative to the controls [Carmichael et al., 2003; Carmichael et al., 2007]. Given the dearth of comparable data in the literature, a small-scale study seemed like a prudent first step toward understanding exposure levels in the study population and their association with hypospadias. Low sample volume resulted in a reliance on imputed data, which could lead to spurious results. The sample collection parameters also limited what compounds we could measure; for example, other potentially endocrine-disrupting contaminants, such as phthalates, bisphenol-A and other more volatile compounds, were not measurable. In addition, given the limited sample size and lack of association with individual compounds, we did not explore mixtures of compounds based on their known functions (e.g., examining analytes with known estrogenic effects as a single group), although this may be an important area of future inquiry [Sharpe and Skakkebaek, 2008; Bloom et al., 2007; Rider et al., 2009]. We also were unable to examine genetic variability in biotransformation of halogenated organics and end-organ susceptibility, which would be an important consideration for future studies. Sample collection during mid-pregnancy was a strength (e.g., versus at delivery or post-partum), but nevertheless it occurred at the end of or shortly after completion of the urethral closure process. Given the persistence of the studied compounds, however, the levels are likely to be a good reflection of at least the few preceding weeks.

A benefit of this study is that it establishes the feasibility of measuring the selected analytes under the given sample collection and handling conditions. Although sample volume was limited, it was adequate to allow detection of reasonable levels of many of the analytes in most samples. A potential concern was contamination of the samples with ambient PBDEs (e.g., from dust). However, the median ratio of PBDE-99 to PBDE-47 was 0.31 among cases and 0.32 among controls, which is consistent with previous studies and suggests that substantial contamination did not occur during sample processing (exogenous PBDEs tend to have a higher ratio) [Bradman et al., 2007; Sjodin et al., 1998; Sjodin et al., 2005]. Also consistent with other studies, the correlations of PBDEs with PCBs were low (e.g., among controls, the correlation of PBDE-47 with PCB-153 was 0.05) and support the idea that exposure pathways differ for the two classes of compounds [Bradman et al., 2007; Schecter et al., 2006].