Prospectively Identified Incident Testicular Cancer Risk in a Familial Testicular Cancer Cohort

Testicular germ cell tumors (TGCT) are the most common form of cancer in men of ages 15 to 35 years. Approximately 8,430 new cases, and 380 TGCT deaths are projected for 2015 (1). The Surveillance, Epidemiology, and End Results (SEER) program reported a U.S. testicular cancer incidence of 5.5 per 100,000 white men between 2006 and 2010 (2). Ninety-eight percent of these tumors are thought to arise from arrested primordial germ cells (3). TGCT presents two main histologic types: the more aggressive nonseminomas (peak incidence: ∼25 years of age), and the less aggressive seminomas (peak incidence: ∼35 years of age; refs. 4, 5). TGCT incidence has more than doubled during the last 30 years, most notably among men of European ancestry (6). The basis for this pattern of increasing incidence of a malignancy that strikes men in the prime of their productive lives is not well understood.

TGCT has an estimated heritability that ranks as the third highest among all cancers (7), although it does not fit the classical, high penetrance, monogenic paradigm. Compared with most malignancies—which have familial relative risks between 1.5-to-2.5—retrospective cohort studies with various designs (Table 1) have demonstrated that sons of men with TGCT have a 4- to 6-fold increased risk of TGCT versus the general population, while brothers of affected men have an 8- to 14-fold increased risk (8–20). These risks increase to 37- and 76-fold in dizygotic and monozygotic twins, respectively (21). Although there is a substantial epidemiologic literature aimed at estimating familial risks of TGCT, all prior reports targeted sporadic or unselected TGCT, and used retrospective, cross-sectional, or record linkage designs. There have been no published reports describing prospective TGCT risk among affected and unaffected members of multiple-case families in which follow-up and cancer validation were performed at the individual level.

Despite these strong familial relative risks, the largest genome-wide genetic linkage study performed by the International Testicular Cancer Linkage Consortium did not uncover any major, highly penetrant genes predisposing to TGCT; rather, its data suggested that multiple genes with smaller effect sizes may underlie this familial aggregation (22). The discovery of multiple TGCT-risk variants in recent genome-wide association studies (GWAS) supports the hypothesis that many genetic loci contribute to TGCT risk (6, 23–29). Despite the apparent heritability of TGCT, families with more than two affected members are unusual, unlike other hereditary cancer syndromes in which a single multigenerational pedigree often harbors many affected individuals. TGCT is thought to be a polygenic disorder caused by the combined effects of multiple, common genetic variants, perhaps acting in concert with certain environmental exposures. To date, 19 genomic susceptibility loci and three candidate genes have been identified, implicating biologic pathways involving fertility, spermatogenesis, sex determination, and testicular differentiation (6, 23–30). However, the traditional genetic perspective has been that polygenic disorders should not present as familial clusters, presumably because the penetrance of such variants is low (31). Therefore, familial TGCT (FTGCT) represents an unusual, and potentially informative, exception to this rule.

Cryptorchidism, infertility, positive family history, previous TGCT, and white race are known TGCT clinical risk factors (4, 5, 32, 33). The TGCT relative risk among men with cryptorchidism is 4.8 (95% CI, 4.0–5.7; ref. 32). Male infertility also confers a significantly increased risk of testicular cancer [standardized incidence ratio (SIR), 2.8; 95% CI, 1.5–4.8; ref. 33]. Both our group and British investigators have implicated testicular microlithiasis in TGCT risk (34, 35). There is also evidence that infertility may be increased in males from TGCT families compared with the general population (36).

In an analysis of 985 cases of TGCT from 461 families, we found that the characteristics of FTGCT were largely similar to those observed in sporadic TGCT (37); similarities included: (i) the distribution of seminomas and nonseminomas; (ii) the frequency of bilateral cases and; (iii) a later age-at-diagnosis for seminomas than nonseminomas. In addition, the genomic regions implicated as susceptibility loci by GWAS have been similar for sporadic and FTGCT (24, 25, 28, 29, 38). Differences include a 2- to 3-year earlier age of onset for FTGCT versus TCGT (39). A younger age at tumor diagnosis is observed in many hereditary cancer syndromes, a pattern thought to reflect the role of genetic factors (40). However, despite the cumulative data suggesting an important role of heritable factors in the etiology of FTGCT, no study to date has evaluated whether there is an increased risk of prospectively identified incident testicular cancer in an FTGCT cohort, compared with the general population, a knowledge deficit that produces clinical uncertainty when counseling high-risk family members. Given that two is the most common number of TGCTs in multiple-affected-person families, one might anticipate that such risks would be small, if they could be detected at all. We hypothesized that if there indeed were a genetic component to FTGCT, there should be a substantially increased risk of incident testicular cancer in our prospectively followed FTGCT cohort. This is the first prospective study with long-term follow up that quantifies incident TGCTs in a cohort of FTGCT individuals and bloodline relatives.

Multiple-case families with (i) ≥two confirmed TGCT subjects, (ii) a combination of TGCT and extra-gonadal germ cell tumor (both designated “multiple-affected-person” families), and (iii) families containing only a single individual with bilateral TGCT (designated “sporadic-bilateral-subject” families) were enrolled in the “Multidisciplinary Etiologic Study of Familial Testicular Cancer” (NCI Protocol 02-C-0178; NCT-00039598). In the aggregate, these 3 subsets of families were designated “multiple-case” families, because a subject with sporadic bilateral testicular cancer by definition had two cases of TGCT. Kindreds with a female germ cell tumor patient were excluded from the current analysis. The study protocol explicitly included sporadic-bilateral TGCT subjects (i.e., men with bilateral testicular cancer and a negative family history of TGCT) because bilateral affection of paired organs has long been regarded as one of clinical features, suggesting the presence of an underlying cancer susceptibility disorder. Our original analytic plan was to seek candidate gene germline mutations identified in multiple-affected-person families, within our sporadic-bilateral subjects. It was our a priori hypothesis that at least a subset of sporadic-bilateral TGCT patients would be found to have germline mutations in the same susceptibility gene(s) identified in multiple-affected-person families, that is, that they would have the same genetic cause of their cancer.

Participants completed family, medical, epidemiologic, and psychosocial questionnaires and donated blood samples. All subjects provided written informed consent. Families with two or more affected males or a sporadic bilateral case were eligible for travel to the NIH Clinical Center for a protocol-based etiologic evaluation, including detailed history and physical examination, semen, and laboratory analyses, ultrasound imaging of the testes or ovaries, and ultrasound imaging or computed tomography of the kidneys (41). This study was approved by the NCI Institutional Review Board. Ninety-three percent of all participants reported their racial category as white. Twelve hundred and sixty enrolled individuals from 140 families were included in this study; females and non-bloodline relatives were excluded from the current analysis.

Within each family, we designated the first participant with TGCT to enroll in the study as the index case. Data collected from all participants included gender, vital status, date of birth, and dates of death and/or censorship. Data regarding clinical factors, such as microlithiasis and undescended testis (UDT), were also collected. Also, pathology reports were obtained and reviewed for seven of eight (87.5%) incident cases. The relationships between multiple affected persons in a family were classified as “siblings,” “first cousins,” “father–son,” “uncle–nephew,” or “complex,” a term reserved for families with patterns that did not fit neatly into the one of the other categories. We performed annual follow-up of study participants via mailed questionnaires and telephone contact.

Referent age-adjusted population cancer rates for white males were computed by 5-year age group and 5-year calendar periods using the NCI SEER9 database (1973–2010). The at-risk interval was defined from the family enrollment date (the date on which the first subject from each family signed the study-related informed consent document) to date of cancer diagnosis, death or end of study. Accrued person-years were calculated, and an observed-to-expected SIR for incident TGCT was calculated using SEER*Stat, as previously described (42). All TGCT (n = 224) diagnosed prior to each family's date of study enrollment were excluded from the incident TGCT calculation.

Twelve hundred and sixty men from 140 families with 10,207 person-years of follow-up were included in this study. Eight of the 1,260 subjects developed TGCT during follow-up; six incident cases had no prior testicular cancer history, while two were metachronous TGCTs. Six incident cases occurred in multiple-affected-person families and two incident cases occurred among the relatives of men with sporadic-bilateral TGCT. Table 2 summarizes the demographic and clinical characteristics of the individuals with TGCT prior to enrollment and characteristics of incident TGCTs, including number of individuals affected in the family, presence of microlithiasis, personal history of UDT, familial pattern of affection and TGCT morphology. Prior TGCT cases and incident cases had similar distributions of these variables. Table 3 summarizes the clinical characteristics of study participants with an incident cancer.

Eight TGCTs were observed among the 1,260 familial multiple-case TGCT cohort members during prospective follow-up versus 0.67 cases expected (O/E = 11.9; 95% CI, 5.1–23.4; see Table 4). Analyzing only the 1,036 family members with no personal history of TGCT prior to cohort entry yielded similar results: observed = 6; expected = 0.50; O/E = 12.0; 95% CI, 4.4–26.1. The absolute excess risk of TGCT in this cohort was 7.2 cases per 10,000 (P < 0.001). Table 4 summarizes SIRs stratified by selected characteristics chosen a priori as potential modifiers of TGCT risk. Within the constraints imposed by the small number of events, none of these features identified a subset of study participants as being at markedly greater risk of developing incident TGCT, although the presence of either microlithiasis (O/E = 29.3; 95% CI, 10.7–63.7) or UDT (O/E = 31.1; 95% CI, 8.5–79.7) in the family suggested higher risks. However, the 95%CI associated with these point estimates overlapped with those from the respective “no” categories, indicating that these differences were not statistically significant. Of note, seven of the eight incident TGCT occurred in the 119 families with ≤2 affected individuals (expected = 0.51; O/E = 13.7; 95% CI, 5.5–28.3) versus 1 observed (expected = 0.16; O/E = 6.2; 95% CI, 0.2–35.2) in the 21 families with ≥3 affected. Thus, the handful of heavily loaded families did not drive the occurrence of incident TGCT in this cohort.

Stratifying the data by multiple-affected-person family (n = 82 families; 874 family members; 7,696.9 person-years of observation) versus sporadic-bilateral-subject family (n = 56 families; 373 family members; 2,437.5 person-years of observation) yielded similarly increased SIRs in both the former (observed = 6; expected = 0.52; O/E = 11.6; 95% CI, 4.2–25.1) and the latter (observed = 2; expected = 0.15; O/E = 13.4; 95% CI, 1.6–48.6).

In 2002, the National Cancer Institute's Clinical Genetics Branch initiated an observational, etiologic study of FTGCT (41). During the course of prospective follow-up, 8 persons (6 without, and 2 with, a personal history of TGCT at the time of enrollment) developed TGCT, a nearly 12-fold increase in TGCT risk compared with the number expected from gender-, age- and calendar-time-specific rates from the U.S. white population. These are the first cases of FTGCT to be documented prospectively, and their occurrence permitted us to generate the first quantitative estimates of TGCT risk in the setting of multiple-case families.

Furthermore, stratified analysis revealed that the risk was similarly increased in multiple-affected-person families (O/E = 11.6) and sporadic-bilateral-subject families (O/E = 13.4). Our results confirm that men from both multiple-affected-person TGCT families and sporadic-bilateral subject TGCT families truly do comprise two subsets of the general population that are at substantially increased TGCT risk. Although the number of cancer events in each group is small, and the excess absolute risks are low, each O/E ratio is statistically significantly elevated relative to general population expectation. Nonetheless, our observations in the relatives of men with sporadic-bilateral TGCT warrant replication, a task that may be approachable using the Scandinavian population-based registry system.

Our results are somewhat surprising given that a combination of low penetrance genes is thought to underlie the etiology of familial testicular cancer, and that about 75% of families contain only two cases, because it is generally believed that polygenic susceptibility does not produce familial aggregations of disease (31). To the best of our knowledge, ours is the only existing longitudinal cohort study targeting men from extended multiple-case TGCT families that could be used to address this fundamental question. In particular, the prospective occurrence of incident TGCT in the relatives of men from sporadic-bilateral-subject families further supports the broader notion that there is a genetic component to this pattern of affection. This unexpected result is consistent with the recognition that men who are homozygous for KITLG TGCT-associated risk alleles have a TGCT odds ratio that is greater than 6 (25, 26), the strongest SNP/cancer association yet reported. FTGCT may be the first well-documented example of a disease presentation that will become more common now that our ability to identify polygenic disorders has become more tractable. Potential mechanisms for this phenomenon include (i) the existence of intermediate-risk variants, like KITLG; (ii) the presence of common, low-penetrance variants acting as modifiers of the risks associated with as yet undiscovered rare, high-penetrance variants; and (iii) common variants proving to be highly active functionally.

We attempted to determine whether specific clinical features might permit identification of a subset of family members that was at particular risk of developing incident TGCT. Within the constraints imposed by the small number of prospective cancer events, none of the characteristics we examined (Table 4) were significantly correlated with cancer risk above and beyond the level seen in the entire set of family members. The SIRs associated with a family history of either microlithiasis (O/E = 29.3) or undescended testes (O/E = 31.1) trended toward greater risks, but these differences were not statistically significant. We are continuing to enroll and follow additional FTGCT kindred, and hope to eventually achieve sufficient statistical power to answer these questions definitively. We should note that our prior report linking microlithiasis to the risk of FTGCT included many of the same families analyzed here (35); thus, these findings do not comprise independent confirmation of that provocative observation, which does merit corroboration in the context of elucidating the pathogenesis of testicular cancer. The microlithiasis association question is one of the major foci of our ongoing research.

This is the first study to demonstrate quantitatively that the incidence of testicular cancer is substantially increased relative to the general population in a cohort of multiple-case families, including both multiple-affected-person and sporadic-bilateral-subject kindreds. Although there is a substantial epidemiologic literature aimed at estimating familial risks of TGCT, all prior reports targeted sporadic/unselected TGCT, and used retrospective, cross-sectional or record linkage designs (Table 1). In contrast, our study was family-based, prospective, excluded prevalent cases from the risk assessment, had clinical details on a significant fraction of study participants, included a relatively large number of individuals at risk, had central pathology review of TGCT cases performed (87.5% of incident cases), and was based on an average follow-up of more than 8 years. Nonetheless, the number of cancer events was small, limiting our ability to more precisely define subsets of family members that might be at particularly high risk. In addition, individual level information relative to testicular microlithiasis was available only for the 132 individuals who had undergone testicular ultrasound, either as part of our study or during the course of their routine clinical care. This restricted our stratified SIR analysis of microlithiasis to families rather than individuals. Critical risk factor information, such as history of undescended testicle, was based largely on unconfirmed subject self-report. Medical record documentation of UDT was exceedingly difficult to obtain. The study was not designed to disentangle the relative contributions of genetic, developmental and environmental factors to the etiology of TGCT. Rather, its primary focus was on susceptibility gene discovery, toward which end our annotated DNA samples have been contributed to multiple analyses that have shaped our current understanding of TGCT genomics (4, 6, 22, 23, 29, 37, 38, 41, 43, 44).

Given the rarity of testicular cancer and its favorable prognosis even at advanced stages, the United States Preventive Services Task Force (USPSTF) has recommended against testicular cancer screening, concluding that the limited benefits do not outweigh the potential screening-related harms (45, 46). We concur that there is no proven testicular cancer screening strategy available for clinical use, and further believe that the relative rarity of TGCT coupled with its high curability rate make it unlikely that such a strategy will be developed and formally validated. The USPSTF concluded that these characteristics make it unlikely that screening asymptomatic men from the general population will produce additional benefits above and beyond clinical detection (45, 46). However, for the first time, our study has demonstrated prospectively that men from multiple-affected-person and sporadic-bilateral-subject TGCT families are at substantially elevated risk relative to the general population.

What can one do with this information in the absence of a clinically validated risk-reducing strategy? This conundrum is becoming increasingly prevalent, as our ability to identify persons at increased genetic cancer risk is outstripping the development of evidence-based cancer site-specific screening and risk-reducing capabilities. First, the results are of etiologic importance in that we have documented high TGCT risk in a genetic context where the presence of a rare, highly penetrant, single gene Mendelian trait seems very unlikely (22). If the currently accepted polygenic model of TGCT heritability is correct, our findings suggest that substantial cancer risks can result nonetheless.

Second, even in the absence of proven benefit, best clinical judgment would seem to support advising members of multiple-affected-person and sporadic-bilateral-subject families to perform testicular self-examination on a regular basis, and to bring new abnormalities (testicular mass; persistent testicular pain) to the attention of their health care providers promptly. Outside the research setting, we do not advise periodic, routine testicular ultrasound examination for high-risk individuals. We reserve such imaging for the evaluation and management of testicular cancer signs or symptoms, an approach that is practical given the very high rates of cure among patients with advanced-stage TGCT. Nonetheless, there is real potential to avoid the acute and chronic toxicities (e.g., coronary artery disease, neurotoxicity, nephrotoxicity, ototoxicity, pulmonary fibrosis and treatment-related second cancers; ref. 47) related to 3 to 4 cycles of platinum-based chemotherapy if TGCT can be detected at a sufficiently early stage to permit management with surgery and surveillance rather than chemotherapy and/or radiation. And treatment delay has been associated with reduced TGCT survival (48, 49). Thus, a family history of bilateral TGCT or ≥2 TGCT cases might be considered clinically actionable, despite the absence of an effective screening program.

Finally, modeling exercises have suggested that combining data from GWAS risk loci and strong clinical risk factors (e.g., family history, UDT, infertility) might permit the development of risk stratification models that could identify specific subsets of men with even more dramatic elevations in risk, upon whom more aggressive education and surveillance activities might be appropriately focused (50, 51), especially if it could be demonstrated that the GWAS risk SNPs were not also associated with the clinical risk factors, a question for which limited data are contradictory (52, 53). Thus, for example, men aged 30–34 in our study who were in the top 1% of genetic risk and who also had a personal history of cryptorchidism were estimated to be at a 50-fold increase in TGCT risk relative to average population risk, assuming that the TGCT risk SNPs were not also associated with undescended testicle risk (50). We are continuing to develop this line of research in the hope that clinically actionable levels of risk can be defined.

This study presents the first prospectively collected data on incident testicular cancer in a multiple-case familial testicular cancer cohort, providing strong evidence that TGCT incidence is substantially higher in this group than in the general population. These findings support the notion that the combined effect of common, low-penetrance mutations can confer a significant risk of cancer, and provide a rationale for developing more sophisticated risk stratification strategies that might unambiguously identify subsets of men that warrant enhanced education and TGCT surveillance.

No potential conflicts of interest were disclosed.

Conception and design: M.H. Greene

Development of methodology: M.H. Greene

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.T. Loud, K. Nichols, M.H. Greene

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Pathak, C.D. Adams, D.R. Stewart, M.H. Greene

Writing, review, and/or revision of the manuscript: A. Pathak, C.D. Adams, J.T. Loud, D.R. Stewart, M.H. Greene

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.T. Loud, K. Nichols, M.H. Greene

Study supervision: J.T. Loud, M.H. Greene

The authors gratefully acknowledge the participation of our patients, without whom this work could not have been done.

All authors were supported by the NCI Intramural Research Program, Division of Cancer Epidemiology and Genetics, Human Genetics Program, under NIH Project Z01-CP010144-16, Clinical Genetic Studies of Familial and Hereditary Cancer Syndromes.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.