A Case-Control Investigation of Immune Function Gene Polymorphisms and Risk of Testicular Germ Cell Tumors

There is reason to suspect that testicular germ cell tumor (TGCT) development may be influenced by cytokines, secreted proteins that modulate tumor immune surveillance activity as well as a variety of processes in the testis. To address this hypothesis, we conducted a case-control analysis (508 cases, 608 controls) of 32 putatively functional single-nucleotide polymorphisms (SNP) in 16 immune function genes among non-Hispanic Caucasian participants in the U.S. Servicemen's Testicular Tumor Environmental and Endocrine Determinants Study. The TGFB1 Ex5−73C>T variant was positively associated with TGCT (CT/TT versus CC: odds ratio, 1.73; 95% confidence interval, 1.01-2.95; P_trend = 0.05); additionally, haplotypes of the assessed TGFB1 SNPs (−509C>T, 327C>T, Ex1−282C>G, and Ex5−73C>T) differed in frequency between cases and controls (all TGCT, P 0.07; seminoma, P 0.04; nonseminoma, P 0.11). We also observed excess frequencies among TGCT cases versus controls of LTA 252G (P_trend = 0.08) and of the TNF variants −1042C (P_trend = 0.06), −1036T (P_trend = 0.07), and −238G (P_trend = 0.09). Analyses of haplotypes for LTA-TNF SNPs (LTA −91C>A, LTA 252A>G, TNF −863C>A, TNF −857C>T, TNF −308G>A, and −238G>A) were similarly suggestive of an association with TGCT (P = 0.06) and nonseminoma (P = 0.04), but not seminoma (P = 0.21). Polymorphisms in other genes were found to be associated only with seminoma (IL2) or nonseminoma (IFNGR2 and IL10). However, none of the associations remained noteworthy after applying the false discovery rate method to control for multiple testing. In conclusion, our findings suggest that polymorphisms in TGFB1 and LTA/TNF, and possibly other immune function genes, may influence susceptibility to TGCT. (Cancer Epidemiol Biomarkers Prev 2007;16(1):77–83)

The etiology of testicular germ cell tumors (TGCT), the most common type of testicular cancer diagnosed among adolescent and young adult males, remains largely unknown; the only well-described risk factors are cryptorchism, Caucasian ethnicity, and family history and personal history of TGCT (1). Epidemiologic findings suggest various factors that may influence TGCT development, including exposure to high maternal estrogen levels in utero, low androgen levels, mechanisms underlying infertility, and viral infection (1). However, these hypotheses remain tentative.

There is evidence that immunologic factors may influence the development of TGCT. Studies of acquired immune deficiency syndrome patients and kidney transplant recipients suggest that immunosuppressed men have an increased risk of seminoma (2-5), a common histologic subtype of TGCT. The elevated risk may be the result of an unchecked viral infection, although no such agents have been identified to date. Another explanation that has been proposed is impaired tumor immune surveillance (4); this hypothesis is supported by evidence that the extent of lymphocyte infiltration in seminomas is associated with a reduced risk of disease recurrence (6).

The study of cytokines may yield insight into the etiology of TGCT. Cytokines are secreted proteins that play a critical role in regulating the immune system; they modulate various activities of target cells, such as cell division, differentiation, and apoptosis. Cytokines may promote or inhibit tumorigenesis through their effects in modulating inflammation and tumor immune surveillance activity (7). Additionally, cytokines expressed in the testis, most notably interleukin (IL)-1, IL-2, IL-4, tumor necrosis factor-α (TNF-α), and transforming growth factor-β (TGF-β), are known to influence a variety of testicular processes (e.g., inflammation, steroidogenesis, spermatogenesis, and testis development; ref. 8) and may affect TGCT development on this basis. Evidence that levels of cytokine gene expression are altered by common genetic polymorphisms (9-11) has led to considerable interest in investigating the etiologic relevance of such variants for a wide variety of diseases.

To explore whether cytokines influence TGCT development, we investigated the relationship between putatively functional single-nucleotide polymorphisms (SNP) in cytokine genes and TGCT risk within a population-based case-control study, the U.S. Servicemen's Testicular Tumor Environmental and Endocrine Determinants Study.

A detailed description of the study methods is provided elsewhere (12). Briefly, the U.S. Servicemen's Testicular Tumor Environmental and Endocrine Determinants Study is a case-control study of TGCT conducted within the population of men who served in the Army and Navy between 1989 and 2002, and have at least one serum sample stored in the Department of Defense Serum Repository (Silver Spring, MD). Using a person-specific ID, specimens in the Department of Defense Serum Repository computerized database were linked to the Defense Medical Surveillance System (13) and to other military medical databases to determine which military personnel had developed medical conditions. Cases and controls were asked to complete a telephone interview, donate a buccal cell sample collected in mouthwash, grant permission to use the Department of Defense Serum Repository serum specimen, and to sign an informed consent document. In addition, each participant was asked for permission to contact his mother to enroll her in the study. The study was approved by Institutional Review Boards of the National Cancer Institute (Rockville, MD) and the Walter Reed Army Institute for Research (Forest Glen, MD).

All men with a sample in the Department of Defense Serum Repository who subsequently developed TGCT while on active duty and who were of ages ≤45 years at diagnosis were eligible to participate as cases. Diagnoses of TGCT were limited to classic seminoma or nonseminoma (embryonal carcinoma, yolk sac carcinoma, choriocarcinoma, teratomas, mixed germ cell tumor). The diagnoses were based on the original pathology reports. In instances where the original reports were unclear (6.5%), the diagnosis was confirmed by one of the coauthors (I.A.S.). Nine hundred sixty-one eligible case participants were initially identified. Of these men, 76 could not be traced, 27 had died, 3 were deployed to a combat zone, 2 were found to be ineligible, 22 did not complete study enrollment before the cutoff date, and 77 refused to participate. Of the remaining 754 men who were enrolled as cases, 590 provided buccal cell samples. Five hundred seventy-seven of these cases were successfully genotyped.

Men with a sample in the Department of Defense Serum Repository who did not subsequently develop TGCT were eligible to participate as controls. The study was designed as a pair-matched case-control study, although additional controls were initially identified due to the transient nature of the military population. Using the computerized Defense Medical Surveillance System database, all available controls were identified for each potential case participant. From the list of possible controls, four individuals who matched each case on age (within 1 year), race (white, black, other), and date of serum collection (within 30 days) were chosen at random as the control set. The first man on the list was designated as the primary control. Every attempt was made to enroll this man for 30 days (average number of attempted contacts = 90). The effort included tracing attempts, multiple letters, and telephone calls. If the man could not be traced, was deployed to a combat zone, deceased, refused to participate, or could not be contacted within a 30-day period, attempts began to enroll the next possible control in the set.

Among the controls, 2,579 were evaluated for inclusion. Three hundred eighty-five men could not be traced, 18 had died, 64 were deployed to a combat zone, 2 were found to be ineligible, 32 did not complete study enrollment before the cutoff date, 928 were lost because they did not respond to attempts to contact them within 30 days, and 222 refused to participate. Of the remaining 928 controls who were successfully enrolled, 712 provided buccal cell samples. Seven hundred eight of these controls were successfully genotyped. Among the 577 cases and 708 controls with genotyping data, there were 547 matched case-control pairs.

Buccal cell specimens were mailed directly to BBI Biotech Research Laboratories, Inc. (Gaithersburg, MD) for DNA extraction using the Gentra AutoPure system (Gentra Systems, Inc., Minneapolis, MN). Genotype analysis was done at the National Cancer Institute Core Genotyping Facility (Gaithersburg, MD). All TaqMan assays (Applied Biosystems, Inc., Foster City, CA) were optimized on the ABI 7900 HT detection system with 100% concordance with sequence analysis of 102 individuals listed on the SNP500Cancer website (14).⁶

We selected 32 SNPs in 16 genes encoding cytokines or cytokine receptors that may influence tumor immune surveillance and/or processes in the testis (Table 1). The SNPs were selected if they possessed a minor allele frequency ≥0.05 and satisfied one of the following criteria: laboratory evidence of functional relevance or a previously reported association with human disease (Supplementary Table S1).

Genotyping success rates ranged between 98% and 100%. Duplicate samples from 99 study subjects were interspersed throughout each batch for all genotyping assays. The concordance rates for QC samples were 96% to 100% for all assays. Two SNPs were found to deviate from Hardy-Weinberg equilibrium among Caucasian controls (IL6 −236C>G, P = 0.01; IL6 −600A>G, P = 0.005); the quality control data were rechecked and the accuracy of these assays was confirmed (100% concordance for each SNP).

The analysis was restricted to non-Hispanic Caucasians (508 cases, 608 controls, and 481 matched case-control pairs; 87% of study subjects). All statistical analyses were conducted using SAS version 8.2 (SAS Institute, Cary, NC). All statistical tests were two sided with an α level of 0.05. The χ² test was used to identify departures from Hardy-Weinberg equilibrium among controls.

Odds ratios (OR) and 95% confidence intervals (CI) were calculated to estimate the relative risk of TGCT in relation to SNP genotype. Data for homozygotes and heterozygotes were combined if the former consisted of fewer than five subjects. For each SNP, tests for trend were conducted by assigning the ordinal values 1, 2, and 3 to homozygous wild-type, heterozygous, and homozygous variant genotypes, respectively, and by modeling these scores as a continuous variable. The risk estimates for TGCT were calculated using unconditional logistic regression, adjusting for age at reference date (case's date of diagnosis) and date of serum sample collection, a study design variable. Both of these variables were modeled as continuous covariates. These unconditional models provided estimates similar to conditional logistic regression models for individually matched pairs (data not shown). Additional adjustment for cryptorchism did not affect the results (data not shown). Separate analyses for seminoma and nonseminoma were also conducted to investigate the existence of subtype-specific effects. To assess heterogeneity across these subtypes, we conducted case-case comparisons for each variant (analyzed assuming the additive model). Analyses of nonseminomas excluding mixed germ cell tumors (N = 21) yielded virtually identical findings.

Haplotype analyses were conducted for genes with multiple polymorphisms assessed. The interrelationships of polymorphisms were assessed via the linkage disequilibrium measures D′ and R² using Haploview,⁷

and haplotypes were reconstructed using the estimation-maximization algorithm in SAS/Genetics (SAS Institute). The effects of individual haplotypes were estimated by fitting an additive model (15), adjusting for age and serum collection date. The overall difference in haplotype frequencies between cases and controls was assessed using the likelihood ratio test.

We assessed the robustness of our findings for individual SNP loci by applying the false discovery rate method (16) to the P values from genotype comparisons made assuming the additive model. The 32 comparisons with all TGCT were assessed as one set, and the 64 subtype-specific comparisons were assessed as a separate set. We considered findings that met a false discovery rate criterion of 20% to be robust.

Table 1 provides a list of the genes and SNPs that were evaluated. A summary of the findings for all of the assessed polymorphisms is provided in Supplementary Table S2. Polymorphisms in TGFB1, LTA/TNF, IL2, IFNGR2, and IL10 were associated with TGCT or one of its histologic subtypes.

The results for TGFB1 variants are presented in Table 2. One of the four SNPs assessed, Ex5−73C>T, was significantly associated with TGCT, with the T allele more frequent among cases than controls (CT/TT versus CC: OR, 1.73; 95% CI, 1.01-2.95; P_trend = 0.05). This association was stronger for seminoma (OR, 2.09; 95% CI, 1.07-4.05) than for nonseminoma (OR, 1.53; 95% CI, 0.81-2.90), although a test of heterogeneity was not significant. We also observed weaker evidence of association with other polymorphisms; the Ex1−282C allele was associated with a reduced risk of seminoma (P_trend = 0.08) whereas a positive association with nonseminoma risk was observed for −509T (P_trend = 0.06). When we analyzed haplotypes of the TGFB1 SNPs (−509C>T, Ex1−327T>C, Ex1−282G>C, and Ex5−73C>T), a global test of haplotype frequencies was statistically significant for seminoma (P = 0.04) and approached statistical significance for all TGCT (P = 0.07) and for nonseminoma (P = 0.11). The risk estimates of individual haplotypes seemed to be largely explained by the observed associations for TGFB1 SNP genotypes.

The findings for variants in LTA and TNF, which lie adjacent to one another on 6q21, are presented in Table 3. Among cases, we observed excess frequencies of LTA 252G (P_trend = 0.08) and of the TNF variants −1042C (P_trend = 0.06), −1036T (P_trend = 0.07), and −238G (P_trend = 0.09). We did not observe any clear differences between seminomas and nonseminomas with regard to these relationships. For most polymorphisms, slightly stronger associations were observed for nonseminomas than for seminomas, although these differences were not statistically significant. We analyzed haplotypes for the six LTA-TNF SNPs (LTA Ex1+49C>A, LTA 252A>G, TNF −1042C>A, TNF −1036C>T, TNF −487G>A, and TNF −238G>A). Global haplotype tests were of borderline statistical significance for all TGCT (P = 0.06) and nonseminoma (P = 0.04), but not seminoma (P = 0.21). Haplotypes positively associated with TGCT or a subtype included A-A-C-T-G-G (all TGCT, P = 0.04; seminoma, P = 0.04) and C-G-C-C-G-G (seminoma, P = 0.04). The C-A-C-C-G-A haplotype was underrepresented among nonseminomas (P = 0.02).

Polymorphisms in IL2, IFNGR2, and IL10 were also found to be associated with seminoma or nonseminoma (Table 4). The IL2 IVS1−100G allele was associated with an increased risk of seminoma (GT: OR, 1.37; 95% CI, 0.98-1.64; GG: OR, 1.95; 95% CI, 1.13-3.36; P_trend = 0.008), whereas IFNGR2 Ex7−128T was inversely associated with nonseminoma risk (CT: OR, 0.89; 95% CI, 0.65-1.23; TT: OR, 0.58; 95% CI, 0.39-0.88; P_trend = 0.01). The IL10 variants −853T and −626A were also associated with decreased risk of nonseminoma (P_trend = 0.04 and 0.07, respectively). Analyses of IL10 haplotypes (−3584T>A, −1116A>G, −853C>T, and −626C>A) were generally null, although the T-A-T-A was associated with reduced nonseminoma risk (P = 0.03).

We also observed that cases were less likely than controls to be heterozygous for the IL6 SNPs −600G>A and −236G>C (P = 0.01 and 0.01, respectively), although the prevalence of homozygous variant carriers did not differ between groups (Supplementary data). However, as we noted earlier, the genotype frequencies of both SNPs deviated from Hardy-Weinberg equilibrium among controls (IL6 −236, P = 0.01; IL6 −600, P = 0.005). To explore the sensitivity of these findings to the observed departures from Hardy-Weinberg equilibrium, we calculated crude ORs using the control genotype frequencies expected under Hardy-Weinberg equilibrium. After adjustment, we observed no association with −600 (OR, 0.83 and 0.83 for AG and GG, respectively) or −236 (OR, 0.86 and 0.86 for CG and CC respectively).

When we applied the false discovery rate method to assess the robustness of our findings, none of the observed associations met a false discovery rate criterion of ≤20%.

A role for the immune system in TGCT development has previously been postulated on the basis of reports of increased incidence rates among immunosuppressed populations and evidence suggesting the possible importance of viral infections in the etiology of this malignancy (3, 4, 17-19). Previous studies of testicular cancer have investigated the etiologic relevance of variation in human leukocyte antigen genes, with mainly inconclusive results (summarized in ref. 20). We present, to our knowledge, the first report describing the relationship between polymorphisms in cytokine genes and TGCT risk. We analyzed 32 SNPs in 16 genes encoding cytokines and cytokine receptors among 508 cases and 608 controls. Our findings, although not robust on adjustment for multiple testing, suggest that genetic variants in TGFB1, LTA/TNF, IL2, IFNGR2, and IL10 may influence susceptibility to TGCT.

In our study, the TGFB1 variant Ex5−73T was associated with an increased risk of TGCT. We also observed weaker, subtype-specific associations, with Ex1−282G and −509T associated with elevated risks of seminoma and nonseminoma, respectively. Comparisons of haplotype (−509C>T, Ex1−327T>C, Ex1−282G>C, and Ex5−73C>T) frequencies between cases and controls approached statistical significance. The Ex5−73C>T polymorphism involves a nonsynonymous substitution at codon 263 of the polar threonine amino acid with apolar isoleucine, within a region of the TGF-β proprotein that is cleaved from the active part at amino acid 278. It has been proposed that the Ex5−73T allele could affect the stability and activation process of TGF-β and possibly lead to decreased levels of this cytokine (21), although no empirical evidence about function has been published. Conversely, the other risk alleles have been associated with increased TGF-β activity. Ex1−282G>C corresponds to a change from arginine to proline at codon 25, with the −282G (25Arg) allele linked to increased TGF-β expression in in vitro stimulated leukocytes (22). The −509T allele has been associated with increased circulating plasma TGF-β concentration in a gene dose–dependent fashion (23).

TGF-β affects a variety of processes in the testis, most notably the inhibition of primordial germ cell development and Leydig cell steroidogenesis (8). This cytokine has been shown to influence the development of many types of cancer, although the nature of this relationship is complex. Whereas TGF-β is a potent inhibitor of cell proliferation and inducer of apoptosis in normal and transformed cell types, it also seems to promote the invasiveness and metastasis of tumors, possibly through the inhibition of antitumor immune responses (24). Expression of TGF-β was detected in the majority of 26 TGCT specimens analyzed in an immunohistochemical study, with both a greater percentage of positive immunostained cells and stronger intensity of staining in tumor than in peritumor nonneoplastic testis tissue (25). These findings suggest an important role of TGF-β in the development of TGCT, although the mechanism of action is unclear. A mutational analysis of TGF-β pathway components among 20 seminomas found identical insertion mutants in a highly conserved residue of SMAD4 in two tumors (26). SMAD4, located on chromosome 18q21, is a critical member of the TGF-β signal transduction pathway, with mutations in this gene found to disrupt growth inhibitory and apoptotic activities (27). The observation that chromosome 18q is frequently lost or deleted in TGCT further supports a role for the TGF-β pathway in the development of these tumors (28-30).

We also found LTA 252G and the TNF variants −1042C, −1036T, and −238G to be weakly associated with increased TGCT risk in our study population. The LTA 252G allele has been linked to increased lymphotoxin-α production by phytohemagglutinin-activated mononuclear cells in vitro (31), whereas TNF −238G has been consistently associated with elevated antigen-stimulated production of TNF-α (32, 33). TNF −863A and −857T were found in two studies to be associated with increased TNF promoter activity (34) and lipopolysaccharide-induced TNF-α production (35), although contradictory findings have also been reported (36, 37). TNF-α and lymphotoxin-α, produced mainly by activated macrophages, are important mediators of local inflammatory processes. TNF-α has also been shown to possess both antitumor and tumor-promoting or proangiogenic effects in in vitro and mouse studies (7). In the testis, TNF-α has been shown to influence Leydig cell steroidogenesis and germ cell survival (8). Elevated levels of this cytokine have also been associated with experimental autoimmune orchitis (38) and male infertility (39, 40), a risk factor for TGCT.

Other polymorphisms were unrelated to overall TGCT risk, but associated with a particular subtype. Seminoma cases were more likely than controls to carry the IL2 IVS1−100G allele. The functional relevance of this variant is unclear; the G allele has been associated with increased promoter activity in a study of transfected Jurkat cells (41) and with decreased IL-2 expression in peripheral lymphocytes (11, 41). IL-2, a T-lymphocyte growth factor, has been shown to be a potent inhibitor of Leydig cell testosterone production (8, 42), and increased seminal plasma concentrations of this cytokine have been associated with male infertility (43). We also found the IFNGR2 Ex7−128T allele and the IL10 −3584T/−1116A/−853T/−626A haplotype to be associated with decreased risks of nonseminoma. The IFNGR2 Ex7−128T allele, located in the 3′-untranslated region, has also been reported to be associated with decreased risk of non-Hodgkin's lymphoma (44). IFN-γ, an important effector of antitumor immune activity, has been reported to be highly expressed in TGCT; the significance of this expression is unclear, although the investigators speculated that it may affect stromal architecture, thereby facilitating metastasis and angiogenesis (45). IL-10 is an anti-inflammatory cytokine that is suspected to facilitate tumor immune evasion through local immunosuppression. The IL10 −1116A/−853T/−626A haplotype has been reported to be associated with low IL-10 production, and the GCC haplotype to be a “high-production” marker (46, 47). It is not clear whether IL-10 is commonly expressed in the testis, although this induction of this cytokine by testicular lymphocytes following lipopolysaccharide stimulation has been shown in a rat model (48).

Strengths of this study include its population-based design and relatively large sample size; the U.S. Servicemen's Testicular Tumor Environmental and Endocrine Determinants Study is one of the largest case-control investigations of testicular cancer etiology conducted to date. However, our analysis also has some limitations. Despite its relatively large size, our study had limited statistical power to detect associations involving weak effects or rare variants, particularly for the analyses of seminoma and nonseminoma. Additionally, given that we assessed many SNPs and examined relationships both overall and by subtype, it is possible that at least some of our findings are due to chance. Indeed, none of the observed associations were found to be robust after applying the false discovery rate method to our findings. It is important for these findings to be replicated in other studies before meaningful inferences about their causal relevance can be drawn.

In conclusion, our results suggest that polymorphisms in TGFB1 and LTA/TNF, and possibly other immune function genes, may influence susceptibility to TGCT. Additional investigations are needed to replicate and extend these findings and to clarify the biological basis for these relationships.