|Home | About | Journals | Submit | Contact Us | Français|
Exposure to endocrine disrupting chemicals, such as polychlorinated biphenyls (PCBs), may alter hormonal balance and thereby, increase risk of testicular germ cell tumors (TGCT). To study the relationship of PCBs to TGCT, pre-diagnostic serum samples from 736 cases and 913 controls in the Servicemen’s Testicular Tumor Environmental and Endocrine Determinants study were analyzed. Adjusted odds ratios (OR) and 95% confidence intervals (95%CI) were estimated using logistic regression. PCB levels were examined in association with all TGCT and, separately, with each histologic type (seminoma, nonseminoma). Risks associated with seven functional groupings of PCBs, as well as sum of PCBs, were also examined. There were significantly decreased risks of TGCT in association with eight PCBs (PCB-118, PCB-138, PCB-153, PCB-156, PCB-163, PCB-170, PCB-180, PCB-187) and no association with the remaining three (PCB-99, PCB-101, PCB-183). The same eight congeners were significantly associated with decreased risk of nonseminoma, while five (PCB-138, PCB-153, PCB-156, PCB-163, PCB-170) were associated with decreased risk of seminoma. All functional groupings of PCBs were also associated with decreased risk of TGCT and of nonseminoma, while 3 of the 7 functional groups were associated with decreased risk of seminoma. Sum of PCBs was significantly associated with decreased risk of TGCT (ptrend=0.0008), nonseminoma (ptrend=0.001) and seminoma (ptrend=0.03). Overall, these data do not support the hypothesis that PCB exposure increases the risk of TGCT.
The incidence of testicular germ cell tumors (TGCT) has been increasing in the U.S. and in other developed countries for at least five decades (1). Reasons for the increase are not clear as the etiology of TGCT is poorly understood. The only well-described risk factors are cryptorchism, prior history of TGCT, family history of TGCT and increased adult stature (2). The association of TGCT with a congenital anomaly regulated by androgen levels, as well as the similarity between testicular carcinoma in situ and primordial germ cells, has suggested that TGCT risk may be determined very early in life and may be affected by hormone levels (3-7). The ability of exogenous exposures, such as synthetic hormonal drugs, phytoestrogens and persistent organochlorine compounds to mimic the effects of endogenous hormones has led to speculation that these exposures might also be related to the development of TGCT and the other male reproductive disorders (cryptorchism, hypospadias, impaired spermatogenesis) collectively known as the testicular dysgenesis syndrome (8, 9).
Recently, we reported that p,p’-dichlorodiphenyldichloroethylene (p,p’-DDE), the most persistent metabolite of DDT, was significantly associated with increased risk of TGCT (10). Like p,p’-DDE, other organochlorine compounds share the ability to act as endocrine disruptors. While p,p’-DDE has been demonstrated to have anti-androgenic properties (11), various PCB congeners are known to have estrogenic, anti-estrogenic, androgenic and antiandrogenic properties (12). The effect of PCBs on TGCT risk, could then be similar to that of p,p’-DDE, or could be dissimilar depending on the mix of PCB congeners present. One prior TCGT case-control study, conducted in Sweden, found no association between TGCT and total PCB level (13). The study, however, did find that mothers of the case men had significantly higher total PCB levels than did mothers of the control men (13). To pursue this lead and to examine the relationship between TGCT and specific PCBs, a study was conducted among members of the U.S. Armed Forces.
The source population for the current study is more than 4 million U.S. military personnel whose blood samples are stored at −30°C in the Department of Defense Serum Repository (DoDSR) in Silver Spring, Maryland (14). In the late 1980s, the DoDSR began storing surplus sera from HIV testing of military personnel. The Repository now includes more than 40 million samples from active-duty and reserve personnel, with approximately 2.3 million samples added each year. Blood samples are drawn at the time of service entry and, on average, are collected every 2 years thereafter. The earliest blood sample available from each individual was selected for the current study. The blood samples included in the current study were donated between 1987 and 2002.
Case identification was accomplished by linking the DoDSR computerized database, using a person-specific ID, to the Defense Medical Surveillance System (DMSS) in order to determine which of the serum donors had developed testicular cancer subsequent to donation. The histology of all reported testicular cancers was determined by examination of the original pathology reports or by review (6.5%) of the pathology slides. As the study was focused on testicular germ cell tumors, the histologies were limited to classic seminoma or nonseminoma (embryonal carcinoma, yolk sac carcinoma, choriocarcinoma, teratoma, mixed germ cell tumor). The database linkage identified nine hundred sixty-one case men who appeared to meet the study criteria. Of these men, 76 could not be traced, 27 had died, 3 were deployed and 2 were deemed ineligible, leaving 853 possible participants. Of these men, 22 were in the process of being contacted when the study closed. Thus, of the 831 men contacted, 754 (91%) agreed to participate. In the instances where the potential case participant was deceased (n=27), the study attempted to obtain proxy information from the man’s mother. Thirteen proxy questionnaires were completed. The TGCT cases were diagnosed between 1988 and 2003.
Men with a sample in the DoDSR who had not subsequently developed testicular cancer were eligible to participate as controls. The study was designed as a pair-matched case-control study, though additional controls were initially identified due to the transient nature of the military population. From the list of all possible controls for each case, four individuals who matched on birth date (within 1 year), race/ethnicity (white, black, other) and date of available serum sample (within 30 days) were chosen at random as the control set. Among the controls, 2579 were evaluated for inclusion. Of these men, 385 could not be traced, 18 had died, 64 were deployed, 2 were deemed ineligible and 928 could not be contacted within 30 days. Of the remaining 1182 men, 32 were in the process of being contacted when the study closed. Thus, of the 1150 men contacted, 928 (81%) agreed to participate. Among the 754 cases and 928 controls enrolled, 720 were matched case-control pairs. The volume of serum in the samples of eighteen cases and fifteen controls was insufficient to conduct PCB analysis, thus, the final analysis set included 736 cases and 913 controls.
Study participation included completion of the study questionnaire, donation of a buccal cell mouthwash sample, permission to use a DoDSR serum specimen and written informed consent to participate. In addition, each participant was asked for permission to contact his mother in order to enroll her in the study. For thirty days, every attempt was made to enroll the first control man in the set, via tracing attempts, multiple letters and telephone calls. If the potential control could not be contacted, the enrollment process began anew with the next man in the set. All participants were enrolled between April 2002 and January 2005. The study was approved by Institutional Review Boards of the National Cancer Institute, Bethesda, MD and the Walter Reed Army Institute for Research, Silver Spring, MD.
Each participant was administered a computer-assisted telephone interview composed of nine modules. Cases were asked questions in reference to a date one year prior to their TGCT diagnosis (referred to as the ‘reference date’). Control participants were assigned the same reference date as their matched case. For the current analysis, participants were asked to report whether they had a personal history of cryptorchism, a family history of testicular cancer in first and second relatives and their height and weight as of the reference date.
Levels of 15 PCB congeners were analyzed by the Human Toxicology Laboratory of the Institut National de Sante Publique du Quebec. The PCBs examined, by IUPAC number, were: PCB-28, PCB-52, PCB-99, PCB-101, PCB-105, PCB-118, PCB-128, PCB-138, PCB-153, PCB-156, PCB-163, PCB-170, PCB-180, PCB-183, and PCB-187. The plasma samples, enriched with isotopically labeled internal standards, were denatured using formic acid. Analytes were then automatically extracted from the matrix by solid phase extraction. Extracts were automatically cleaned on a florisil column and analyzed by gas chromatography-mass spectrometry (GC-MS). Detection of ions generated from negative chemical isolation was accomplished in the selected ion monitoring mode. Evaluation of concentrations was done by considering recoveries of the labeled internal standards. Linear calibrations extended up to 10 μg/L for most analytes. Higher concentrations were determined after appropriate dilutions. The mean detection limits approximated to 0.005 μg/L. All control samples were run in the same batch, on the same day, as their matched case sample. Average within-day variability ranged from 2% to 5%. Average recoveries were 80%. The PCB limits of detection, median levels and coefficients of variation, which include both within- and between-batch variability determined from replicate QC samples, are shown in the Appendix table. Accuracy of results was validated through successful participation in two external quality assessment schemes, the German External Quality Assurance Scheme (Erlangen University) and the AMAP Ring Test (Quebec).
In order to adjust all measurements for lipid levels, samples were analyzed for triglycerides, free and total cholesterol and phospholipids. The measurements were made with enzyme bio-assays using kits produced by Randox Laboratories (Antrim, U.K.). Measurements of PCBs and lipid levels were obtained for 739 cases and 915 controls.
In addition to analyzing each congener separately, the relationships of TGCT to previously suggested groupings of congeners were examined. The groupings employed were as follows: Wolff groupings (15): Wolff group 1B (PCB-101, PCB-187), Wolff group 2A (PCB-118, PCB-156), Wolff group 2B (PCB-138, PCB-170), Wolff group 3 (PCB-99, PCB-153, PCB-180, PCB-183). Phenobarbitol inducers (16) (PCB-99, PCB-101, PCB-153, PCB-163, PCB-180, PCB-183). Mixed function oxidase inducers (16): (PCB-118, PCB-138, PCB-156, PCB-170). UDP-GT, CYP1A, CYP2B inducers (17): (PCB-99, PCB-101, PCB-118, PCB-153, PCB-156, PCB-180, PCB-183, PCB-187). The PCBs in each of the groupings were calculated by summing over the individual results.
To conduct logistic regression analyses, lipid-adjusted PCB levels were categorized into quartiles based on the levels in controls. PCB levels that fell below or were equal to the limit of detection were imputed as the midpoint of the limit and were included in the first quartile for regression analysis. Odds ratios (ORs) and 95% confidence intervals (95%CIs) were calculated to estimate the association of each PCB with risk of TGCT, and separately with risk of seminoma and nonseminoma. One case was excluded from histology-specific analyses because tumor histology was unavailable.
Given the matched case-control design, risk estimates adjusting for confounders were first generated using conditional logistic regression, restricting the analysis to only the matched sets. Modeling using unconditional logistic regression was subsequently performed utilizing the data from all participants. As the latter involved breaking the match, risk estimates derived from the unconditional logistic regression models were adjusted for the three matching factors: age at reference date, race/ethnicity, and date of serum sample collection. Both logistic regression models were adjusted for cryptorchism, family history of testicular cancer, age at serum draw, adult stature and body mass index (BMI) as BMI is associated with plasma PCB levels (18). As previous research in the STEED population had identified p,p’-DDE as a risk factor, serum p,p’-DDE level was also included in the models (10). Tests for trend in risk were computed using scored variables for PCB levels, based on the median levels of each quartile, to evaluate possible dose-response relationships. As results using conditional and unconditional logistic regression were similar, only those using the latter approach are presented.
All tests were two-sided, with p<0.05 defined as the level of statistical significance. Statistical analyses were conducted using SAS Release 9.1 (SAS Institute Inc., Cary, NC).
As shown in Table 1, 913 controls and 736 cases (313 seminoma, 422 nonseminoma, 1 histology unavailable) were included in the analysis. Cases and controls were matched on age and ethnicity, resulting in no significant differences in the distributions of these variables. The seminoma cases were somewhat older than all the controls, while the non-seminoma cases were somewhat younger. The cases, particularly the nonseminoma cases, were significantly more likely than the controls to have a history of cryptorchism (OR=3.3, 95%CI=1.8–5.9, p<0.0001). The cases were also significantly more likely to have a family history of testicular cancer (OR=2.9, 95%CI=1.5–5.5, p=0.0009), to be taller (p=0.0009) and to have higher serum levels of p,p’-DDE (p=0.005) than the controls. No difference (p=0.86) in body mass index was evident, however.
Prior to examining the relationship between PCBs and TGCT, two preliminary analyses were conducted: an examination of the length of the interval between serum collection and diagnosis, and a comparison of mean total-lipid levels among cases and controls. The mean and median intervals between serum collection and diagnosis/reference date were 3.9 years and 3.3 years, respectively. A comparison of the intervals between the cases and controls found no significant difference (p=0.89). Similarly, a comparison of the mean lipid levels between the cases and controls found no significant difference (p=0.11) (data not shown). As persons were exposed to mixtures of PCBs, rather than to single PCB congeners, the correlations between the congeners and the groupings are shown in Table 2. As anticipated, the PCB groupings were highly correlated with each other and with the more highly chlorinated congeners.
Of the fifteen congeners analyzed, four (PCB-28, PCB-52, PCB-105, PCB-128) were excluded from data analysis as fewer than 35% of the study samples had values above the limit of detection (Appendix). The results of the adjusted analyses of the other PCBs are displayed in Table 3. There were significant inverse associations between TGCT risk and eight of the eleven PCBs analyzed: PCB-118 (Q4 OR=0.55, 95%CI=0.40–0.76, ptrend=0.0007), PCB-138 Q4 OR=0.46, 95%CI=0.32–0.66, ptrend=0.0001), PCB-153 (Q4 OR=0.45, 95%CI=0.31–0.66, ptrend=0.0003), PCB-156 (Q4 OR=0.57, 95%CI=0.40–0.81, ptrend=0.002), PCB-163 (Q4 OR=0.59, 95%CI=0.42–0.83, ptrend=0.001), PCB-170 (Q4 OR=0.56, 95%CI=0.39–0.80, ptrend=0.002), PCB-180 (Q4 OR=0.56, 95%CI=0.38–0.82, ptrend=0.003) and PCB-187 (Q4 OR=0.004, 95%CI=0.42–0.86, ptrend=0.004). When the analysis was restricted to the nonseminoma cases, the same eight congeners were significantly inversely associated with risk. Among the seminoma cases, five of the congeners were statistically significantly inversely associated with risk, though the gradient of the trends tended to be less steep than the gradient of the nonseminoma relationships.
The analyses of PCBs by functional groups are shown in Table 4. Total TGCT was significantly inversely associated with all PCB groupings examined: Wolff group 1B (ptrend=0.02), Wolff group 2A (ptrend<0.0001), Wolff group 2B (ptrend=0.0002), Wolff group 3 (ptrend=0.002), the inducers of UDP-GT, CYP1A and CYP2B (ptrend=0.0005), the phenobarbital inducers (ptrend=0.0003), the mixed function oxidase inducers (ptrend<0.0001), and the sum of PCBs (ptrend=0.0004). Similarly, nonseminoma was significantly inversely associated with all groupings, while seminoma was associated with all groups except Wolff group 1B (ptrend=0.33).
As DDE had previously been shown to be related to increased risk of TGCT in the STEED population, an examination of sum of PCBs, stratified by DDE was undertaken. As shown in Table 5, none of the trends in total PCB exposure were significant in the strata of individuals with low DDE levels. Among the individuals with high DDE levels, however, PCBs levels were significantly inversely related to risk of all TGCT (ptrend=0.03) and to seminoma (ptrend=0.04). The interaction analyses, however, did not find, however, that the relationship between PCBs and TGCT differed significantly by DDE stratum.
To determine whether the congener-specific or the PCB grouping analyses were affected by the inclusion of persons whose interval between serum collection and diagnosis was less than 1 year, subgroup analyses that eliminated the data from those individuals were performed. The subgroup analyses resulted in no differences in results (data not shown).
The incidence of TGCT has been increasing in the U.S. since prior to World War II (19). Though few risk factors have been identified, a number of studies have reported that there is a pronounced birth-cohort effect on risk, suggesting that changes in exogenous exposures may be related to the trend (20). One exogenous exposure, endocrine disrupting chemicals, including PCBs, has been the subject of much speculation as animal data suggested that they might be related to a variety of male reproductive disorders in humans (8, 20). The results of the current study, however, that PCBs are inversely related to risk of TGCT, do not support the hypothesis. In that the inverse association appeared more pronounced, although nonsignificantly so, among persons with high serum levels of p,p’-DDE, it is possible that the effect of PCBs will depend on the mix of other endocrine disruptors also present. Even at lower serum p,p’-DDE levels, however, there was no indication that PCBs served to increase risk.
PCBs are a group of related compounds composed of two carbon-linked benzene rings to which are attached between one and ten chlorine atoms. Between 1929 and 1977, PCBs were manufactured and used in the U.S. as insulators and coolants in electrical equipment, and in the production of numerous household products. Concerns about possible long-term health effects first surfaced in the 1960s when PCBs were reported to be prevalent in wildlife and persistent in the environment (21). Subsequent animal studies reported that PCB exposure resulted in a number of outcomes, including neurobehavioral changes, hypothyroidism, reproductive disorders and tumors. Among humans, mass PCB food-poisonings in Japan (1968) and Taiwan (1979) resulted in chloracne, menstrual irregularities, altered immune responsiveness and general fatigue. In 1977, the sole manufacturer of PCBs in the U.S., the Monsanto Company, ceased production two years prior to the formal ban by U.S. Environmental Protection Agency. Based on the animal data and on occupational studies in humans, IARC has classified PCBs as being a probable carcinogen in humans (22).
In humans, PCBs are stored in adipose tissue and levels tend to increase with age (12). Thus, it is difficult to determine in the present study when the study participants were exposed. PCBs, however, can cross the placental barrier and are present in breast milk, thus some of the exposure may have occurred in utero and/or via breast feeding. If significant exposure occurred in utero, PCBs may also affect the risk of the male reproductive congenital anomalies that are associated with TGCT. To date, very few studies of either cryptorchism or hypospadias have examined PCB levels, however. Two studies of cryptorchism and PCBs have been reported and neither has found a relationship (23, 24). In contrast, a study of the prevalence of hypospadias in Greenland found an unexpectedly low rate despite high levels of PCBs in the population (25), suggesting that PCBs might be inversely associated with risk of hypospadias.
Among the testicular dysgenesis syndrome disorders (9) that become evident in adulthood (impaired spermatogenesis and TGCT), more studies have examined the relationship of PCBs with the former than the latter. In general, the results of the PCB-fertility studies are somewhat equivocal. Several studies have found statistically significant associations with impaired sperm parameters (26-28), while others have found associations only in subsets of their populations (29-31), or not all (32). In contrast, several studies have reported direct associations between PCB levels and fertility (33, 34). The summary of the international INUENDO study of PCBs and fertility in four populations, however, concluded that a representative congener, PCB-153, did not appear to affect fertility or to have direct hormone-like activity (35).
Only one prior study of TGCT and PCBs has been reported to date (13). A case-control study of Swedish men found no association of TGCT with sum of PCBs, estrogenic PCBs, or enzyme-inducing PCBs. An examination of PCB levels in the mothers of the men, however, found that mothers of the cases had significantly higher levels of 21 of the 37 congeners tested. When the PCBs were examined by functional group, the analysis found that the case mothers had significantly higher levels of sum of PCBs and enzyme-inducing PCBs. Interpretation of the mothers’ results is rather difficult, however, as the mothers’ blood samples were obtained approximately 30 years after their sons were born. Body burdens of PCBs in women are affected by weight changes, child bearing and lactation over time, so it is unclear to what extent the mothers’ PCB levels were representative of their levels during pregnancy with their sons. It is uncertain why the sons’ results differed from the results of the current study, but the discrepancy may be related to the small size of the Swedish study (n=58 cases, 61 controls), differences in the PCB mixtures used in each country and timing of the collection of blood samples, as population PCB levels have declined since they were banned. The blood samples in the Swedish study were drawn at a later time (1997–2000) than the samples in the current study, suggesting that the PCB levels might be lower in the Swedish study. A comparison of median levels of estrogenic PCBs and enzyme-inducing PCBs, however, only partially supports this conjecture. The median level of estrogenic PCBs among the Swedish control men was 25 ng/g lipid in contrast with the median control level in the current study of 69.7 ng/g lipid. The median level of the enzyme-inducing PCBs, however, was 174 ng/g lipid among Swedish controls and 73.9 ng/g lipid among US controls. It is also possible that the level of risk or protection of PCBs is determined by the mixtures of congeners present in any location, or the presence of other compounds with endocrine disrupting properties. For example, p,p’-DDE was not related to TGCT in the Swedish study, but was related to risk in the STEED population (10). This difference may be explained by the lower general level of p,p’-DDE in Sweden than in the U.S. (36).
Why PCBs would be inversely associated with TGCT remains to be determined. p,p’-DDE, which has demonstrated antiandrogenic properties, was associated with increased risk in the same population. PCBs have a range of estrogenic, antiestrogenic, androgenic and antiandrogenic effects (12). Examining the effect by the Wolff groupings (15), only two (PCB-101, PCB-187) of the PCBs in Wolff Group-1 (potentially estrogenic) were examined in the current study. While PCB-101 had no relationship with risk, PCB-187 was inversely associated with both TGCT and nonseminoma. All of the Wolff Group 2 (potentially anti-estrogenic, dioxin-like) PCBs examined (PCB-118, PCB 156, PCB-138, PCB-170) were significantly inversely related to TGCT risk. Three of the four Wolff Group 3 (phenobarbitol, CYP1A, and CYP2B inducers) PCBs were significantly inversely related to risk (PCB-153, PCB-180, PCB-183) while there was no relationship with PCB-99. Given that there were inverse relations in all the groups, it is not clear what effect is most important in determining risk. In light of the DDE relationship, however, it seems unlikely that an antiandrogenic effect of PCBs would be inversely related to risk. An antiestrogenic effect, however, might be related, especially as the ratio of androgenic to estrogenic exposures has been postulated to be important in TGCT development. Although there was no statistically significant difference in the relationship of PCB levels to TGCT in low and high DDE strata, significant trends only in the high DDE strata may indicate that the presence of other endocrine disruptors is critical to determining the effect of PCBs.
A major advantage of the current study was that pre-diagnostic serum samples were analyzed. Other advantages were that participants were drawn from a well-defined population, the tumors were histologically confirmed and the participants were likely to be representative of a wide spectrum of the underlying population. Study limitations include that some potential participants could not be contacted due to deployment, the analysis adjusted for self-reported body size rather than measured body size and the study could not determine when and how the participants were exposed to PCBs. Inability to contact men due to deployment presents a potential bias in that deployed men might be different in some way that non-deployed men. As most young men in military service are healthy and fit, however, it would seem unlikely that that deployment would confer substantial bias.
In conclusion, the current study suggests that PCBs are inversely associated with the risk of TGCT, particularly with nonseminoma. The results argue for further examination of PCBs and TGCT in other populations as PCBs are detectable in a large proportion of the world’s population. It will be particularly instructive to examine PCB levels in concert with other endocrine disruptor levels in order to determine whether the inverse association detected in the current study is a result of exposure to other environmental chemicals.
The authors wish to thank Emily Steplowski and Leslie Carroll of IMS for their contributions in data management and analysis. The authors are also greatly indebted to the STEED participants, without whom, there would have been no study.
The opinions or assertions contained herein are the private views of the authors, and are not to be construed as official, or as reflecting the views of the U.S. Department of the Army or the U.S. Department of Defense.
This research was funded by the Intramural Research Program of the National Cancer Institute, NIH, DHHS.