Prediagnostic Hormone Levels and Risk of Testicular Germ Cell Tumors: A Nested Case–Control Study in the Janus Serum Bank

Testicular germ cell tumors (TGCT) are the most commonly occurring cancers among men between the ages of 15 and 40 years (1). Over 90% of all TGCTs occur in this age group and almost all are histologically classified as either seminomas or nonseminomas (2). Nonseminomas, which account for approximately 45% of cases, have a peak incidence at age 25 years, while seminomas account for approximately 55% of cases and have a peak incidence at age 35 years (1). In almost all countries, incidence rates among men of European ancestry are notably higher than rates among men of other backgrounds (2).

Major risk factors for TGCT are not well described. However, TGCTs are thought to be part of a syndrome of male reproductive disorders known as the testicular dysgenesis syndrome (TDS; refs. 3, 4). The TDS disorders, which in addition to TGCT, include cryptorchidism, hypospadias, and impaired spermatogenesis/subfertility, can co-occur in the same individual and are thought to have a common fetal origin (3). The TDS hypothesis suggests that a primary, in utero, insult results in damage to one or more of the testicular stromal cells, the Leydig and Sertoli cells, thereby adversely affecting the development of the germ cells. Both Leydig and Sertoli cells, which receive signals from gonadotropins produced by the anterior pituitary gland, are necessary for proper maturation of germ cells (5). Luteinizing hormone (LH) stimulates the Leydig cells to produce testosterone, while follicle-stimulating hormone (FSH) acts on Sertoli cells to initiate spermatogenesis. Malfunctioning Leydig cells could underproduce testosterone, while malfunctioning Sertoli cells could fail to nurture the developing gonocytes. Among men with impaired spermatogenesis/subfertility, while gonadotropins levels are known to be aberrant, androgen levels have been reported to be low by some studies, but not by others (6, 7). Whether hormone levels vary between men who will develop TGCT and men who will not, however, has not been demonstrated. While studies have examined hormone levels among men after a first TGCT (8–10), identifying sources of prediagnostic sera from men at risk of TGCT has proven to be more challenging. To investigate the hypothesis that higher levels of serum gonadotropins and lower levels of androgens are related to the development of TGCTs, we examined the relationship between gonadotropin and sex steroid hormone levels and risk of TGCT in Norway, the country with the highest incidence of TGCT in the world (2).

The Janus Serum Bank at the Cancer Registry of Norway is a population-based biobank established in 1973 to provide premorbid serum for cancer research (11). Specimens were collected from two sources: county health examinations in selected counties of Norway and Red Cross blood donations. The majority of the blood samples were obtained from approximately 264,000 health examination donors recruited from Finnmark, Oslo, Sogn og Fjordane, and Oppland who were enrolled between 1974 and 1977 (n = 87,647), and other areas in Norway who were enrolled between 1985 and 1991 (n = 176,881). At the time of the health examinations, participants filled out a questionnaire that collected demographic and health-related information. Participants’ age at entry was between 20 and 49 years, with the majority of persons being between 35 and 49 years of age. In addition to the participants enrolled at health examinations, approximately 29,000 Red Cross donors, 20 to 65 years of age at entry and primarily from the greater Oslo region, were enrolled between 1972 and 1989. Serum samples from all participants have been stored at −25°C. The Janus Serum Bank maintains the following data on all participants: name, birth date, Norwegian national identity number, gender, and county of residence at the time of serum donation. The participants in the Janus Serum Bank cohort provided broad consent at the time of recruitment that their samples could be used in cancer research. The Regional Committee for Medical and Health Research Ethics in southeastern Norway approved this specific study (reference number 2012/1993) in accordance with the Declaration of Helsinki.

TGCT cases and controls were selected from cohort members with the following characteristics: no prior history of cancer (except nonmelanoma skin cancer) at baseline blood collection; and at least 0.8 mL of stored serum available. TGCT cases diagnosed between enrollment through December 31, 2012 were identified though linkage with the Norwegian Cancer Registry, using the Norwegian national identity number, a unique identifier assigned at birth or registration with the National Population Register. The Norwegian Cancer Registry dates back to 1953 and has been estimated to have 98.8% completeness of registration (12). Men diagnosed with TGCT who had contributed a serum sample prior to their diagnosis were included as cases. Cases were diagnosed with either seminoma (ICD-O3 morphology codes: 9060-9060-9062, 9064) or nonseminoma (morphology codes: 9065-9101).

Two male controls were matched to each case by region, original enrollment (health examinations vs. Red Cross donor), year of blood draw (1-year strata), and age at blood draw (2-year strata). Controls were cancer-free during the blood draw-to-diagnosis window of the respective case. As such, the final analytic dataset consisted of 182 cases and 364 matched controls.

Figure 1 displays a simplified version of the sex steroid hormone synthesis pathway. Concentrations of sex steroid hormones and their metabolites were measured via LC/MS at Laval University (Québec City, Québec, Canada). Sex steroid hormones measured were DHEA, 4-androstenedione (4-dione), Δ5-androstenediol (5-diol), testosterone, DHT, androsterone (ADT), androsterone glucuronide (ADT-G), 5α-androstane-3α, 17β-diol-3 glucuronide (3α-diol-3G), 5α-androstande-3α, 17β-diol-17 glucuronide (3α-diol-17G), estrone, and estradiol. Sex hormone binding globulin (SHBG), FSH, and LH were quantified in serum by direct immunoassay using the ACS180 Automated Chemiluminescence System (Bayer Health Care) at Boston Children's Hospital, Boston, MA. The standards used in the FSH and LH assays were the World Health Organization 2nd International Standard for human FSH (IS 94/632) and LH (IS 80/552). To evaluate the laboratory performance, three replicate samples from a quality control (QC) pool were included in each batch. Percent detected above the lower limit of detection (LLOD), coefficients of variation, and intraclass correlation coefficients (ICC) for the QC replicate samples are summarized in Supplementary Table S1. ICCs were 79.8% or higher for all hormones measured. Matched case–control sets were assayed together in the same batch.

Free testosterone (13) and free estradiol (14) were calculated using published formulae. The ratios of testosterone to estradiol and free testosterone to free estradiol were also calculated with the values expressed as pg/mL.

The frequency and proportion of cases and controls were calculated within all categories of covariates. Geometric means and 95% confidence intervals (95% CI) of the hormone concentrations were calculated by case status and histologic subtypes. ORs and 95% CI for the associations between measured hormone levels and TGCT risk were calculated using conditional logistic regression, adjusting for height, height², and assay batch. Height was included in the analysis as taller height is associated with increased risk of TGCT (15). For men whose height was not recorded, height was set to the median value (179 cm) among the controls. Sex steroid hormone and gonadotropin measures are not normally distributed, thus were log-transformed prior to analysis.

Hormone levels were categorized into tertiles based on the distribution among controls. Hormone values below the LLOD were assigned to the lowest tertile. To compute tests for trend across tertile categories, intracategory medians were modeled as a continuous parameter. Associations were evaluated stratified by histology using the same conditional logistic regression as was used for the whole sample, with the same covariates, but with the sample restricted to either seminoma or nonseminoma. In addition to the main analyses, four sensitivity analyses were performed to assess the robustness of the observed associations. The first sensitivity analysis excluded hormone values below the limits of quantification. The second analysis determined tertiles of hormone levels by using batch-specific percentiles (10 total batches), rather than percentiles determined across all batches. The third and fourth analyses excluded matched sets in which the case was diagnosed in the first 2 years of follow-up (13 matched sets), or in the first 5 years of follow-up (51 matched sets). All tests were two sided, with P value <0.05 defined as statistically significant, without adjustment for multiple comparisons. All analyses were conducted using SAS 9.3 software (SAS Institute).

The data are available from the Janus Serum Bank upon request.

The distribution of selected demographic and health characteristics of the 182 cases and 364 controls are summarized in Table 1. Among the 182 case men, 149 had seminoma and 33 had nonseminoma. The preponderance of seminoma cases is linked to the age of the men at blood collection, as 64.8% of the cases were in the 40–49 years age group. The majority of men (61.5%) donated their samples in the 1980s. Among cases, the samples were, on average, collected 9.4 years prior to cancer diagnosis (range 1.1–34.9 years). The great majority of the men (86.8%) donated blood as part of a health examination.

The geometric mean hormone concentrations and their ranges are shown by case–control status in Table 2. Comparing all TGCT cases with controls, the hormone with the greatest case–control difference was FSH, with mean values of 6.12 mIU/mL in cases and 4.63 mIU/mL in controls. The FSH difference was also evident when comparing seminomas with controls, with a mean of 6.30 mIU/mL in seminomas and 4.70 mIU/mL in controls. In contrast, among nonseminomas, the greatest case–control difference in mean levels was in SHBG, with cases having a mean of 57.3 pg/mL, while controls had a mean of 68.15 mIU/mL.

The Pearson correlation matrix of the hormones in shown in Supplementary Table S2. As anticipated, the highest correlations were seen between testosterone and DHT (ρ = 0.73), and estrone and estradiol (ρ = 0.62). In addition, there were notable correlations between SHBG and both testosterone (ρ = 0.59) and DHT (ρ = 0.62).

The adjusted relationships of the hormones to TGCTs, and to seminoma and nonseminoma separately, are shown in Table 3. Among all TGCTs, compared with the middle tertile, the highest tertile of 3α-diol-3G was associated with increased risk (OR = 2.37, 95% CI = 1.46–3.83, P_trend from the lowest to the highest tertile<0.01), as was the highest tertile of FSH (OR = 2.89, 95% CI = 1.83–4.57, P_trend <0.01). In the analysis of TGCT histology, increased levels of FSH were related to seminoma (OR = 3.55, 95% CI = 2.12–5.95, P_trend < 0.01) but not to nonseminoma (OR = 1.19, 95% CI = 0.38–3.13, P_trend = 0.16). Increased levels of 3α-diol-3G were similarly related to the risks of seminoma (OR = 2.29, 95% CI = 1.35–3.89, P_trend = 0.01) and nonseminoma (OR = 2.71, 95% CI = 0.82–8.92, P_trend = 0.33).

In the sensitivity analysis that excluded case–control pairs in which the case was diagnosed within the first 2 years of follow-up, the results remained very similar to the overall results. Both 3α-diol-3G (OR = 2.10, 95% CI = 1.28–3.44) and FSH (OR = 2.52, 95% CI = 1.59–4.10) remained associated with increased risk. Similarly, in the sensitivity analysis that excluded case–control pairs whose case was diagnosed within the first 5 year of follow-up, both 3α-diol-3G (OR = 2.08, 95% CI = 1.15–3.78) and FSH (OR = 2.57, 95% CI = 1.45–4.52) remained associated with increased risk. In the sensitivity analyses that (i) excluded persons with hormone values outside the limits of quantification, and (ii) determined quartiles of hormone levels by using batch-specific percentiles, the results were very similar to the overall results.

In this study of prediagnostic samples, higher prediagnostic levels of FSH were associated with increased risk of TGCT, specifically, with increased risk of seminoma. Higher 3α-diol-3G levels were associated with increased risk of TGCT, with consistent increases in risk for seminoma and nonseminoma.

The association between FSH levels and TGCT risk is consistent with the hypothesis that damaged Sertoli cells are related to the development of TGCT, as FSH is a pituitary gonadotropin that mediates spermatogenesis by signaling the Sertoli cells (16). In a feedback loop, Sertoli cells secrete inhibin to negatively regulate FSH release (17). Sertoli cell dysfunction can reduce FSH receptor expression (18) and fail to properly produce inhibin to block the feedback inhibition loop (19), resulting in elevated levels of FSH in men with impaired testis. An elevated serum FSH level is generally interpreted as an early indication of impaired testicular function and disturbed spermatogenesis (20).

To our knowledge, no prior study has prospectively examined gonadotropin levels in men who later develop TGCT. Previous studies, however, have examined gonadotropin levels in men with existing disease, usually designed as case-case comparisons. For example, a Norwegian study compared 13 men who developed a second TGCT or had germ cell neoplasia in situ (GCNIS) in the contralateral testicle, with 26 men who had unilateral TGCT (10). The study, which only examined FSH levels, found that men with bilateral disease or GCNIS were more likely to have elevated FSH than men with unilateral disease. In another case-case comparison, 24 Danish men with GCNIS in their contralateral testicle were compared with 30 men with unilateral disease (8). Higher levels of both FSH and LH were found in the men with GCNIS in the contralateral testicle. GCNIS is the precursor lesion of TGCTs of young men, and evidence suggests that GCNIS almost always develop into TGCTs if left untreated (21). In comparison with these case-case analyses, a Danish case–control study compared 63 men with TGCT with 193 population-based controls (9). Their analysis found that the men with TGCT had higher FSH levels, but lower LH levels, than did the control men. The cases also had lower levels of inhibin B, a factor that suppresses FSH released by the anterior pituitary.

The gonadotropin findings of the current study are partly consistent with prior results, although the differences in study design make comparisons difficult to interpret. The most consistent finding across studies is that FSH levels were increased in men with TGCT, or in the case of the current study, men who later develop TGCT. This finding is in line with the hypothesis that dysfunctional Sertoli cells which do not respond appropriately to FSH stimulus, are predispose to the development of TGCT. The lack of appropriate response is suggested by the observation of the Danish case-case study that found not only elevated FSH levels, but also lower inhibin B levels than were seen in control men (9). In contrast to the FSH findings, the LH findings in the current study were not consistent with the prior findings. While the Danish case–control study found that men with TGCT had lower LH levels than control men, the current study found no difference in LH levels. This may suggest that damage to Leydig cells is more pronounced once a TGCT develops, rather than a number of years prior to diagnosis. Alternatively, the differences in the results could be related to differences in the histology distributions in the two studies. While 43% of the TGCTs in the Danish case–control study were nonseminomas, only 18% of the TGCTs in the current study were nonseminomas. Nevertheless, there was little indication in the current study that lower levels of LH could be related the development of nonseminomas.

Androgens are known to play a central role in the development of testis (22), and thus, insufficient androgen levels have long been suspected of being of etiologic importance in TGCT (3). As with gonadotropins, few studies have examined either androgens or estrogens in relation to TGCT risk, however. In the Danish case-case study that compared 24 men with GCNIS in their contralateral testicle to 30 men with unilateral disease, both testosterone and estradiol were measured (8). No differences in either hormone were found. Similarly, the Danish case–control study found no differences in levels of testosterone, free testosterone or estradiol (9). These findings are consistent with the findings of the currently study in that no prediagnostic differences in levels of testosterone, free testosterone or estradiol were detected. In the current study, however, the androgen metabolite, 3α-diol-3G, was associated with increased risk. Evidence suggests that 3α-diol-3G and the other glucuronide metabolites, androsterone glucuronide and 3α-diol-17G, are the obligatory route of inactivation and elimination of androgens (23). Why there would be greater inactivation of androgens in men who later develop TGCT, however, is not clear. However, higher 3α-diol-3G levels in concert with no difference in testosterone or DHT levels, run counter to the hypothesis the TGCT development is related to insufficient androgenicity. The current findings are also somewhat similar to findings of a study of hormone levels at birth and risk of TGCT among adolescents (7). In that study, higher levels of neonatal androstenedione were associated with subsequent risk of TGCT at ages 15 to 19 years.

The current study had several notable strengths, including that it is the only study, to date, to investigate the association between hormone levels and risk of testicular germ cell tumors in prediagnostic samples. In addition, incident TGCT diagnoses were well characterized through linkage to the Norwegian Cancer Registry, which has a 98% registration completeness rate (12). The study participants were followed, on average, for more than 9 years, thus assuring that the hormone levels were not reflective of an underlying TGCT. Importantly, the sex steroid hormone levels were assayed using LC/MS, which is the gold standard for hormone determinations (24). In addition to these strengths, the study also had several limitations. The study population was limited to men in Norway; thus the results may not be generalizable to men of other ancestries. It should be noted, however, that Norwegian men have the highest risk of TGCT of any men in the world. In addition, the study lacked information on family history of TGCT, weight, and time of day at which the blood samples were collected. The study was unable to measure serum levels of inhibin B, which would have provided more information on the FSH feedback loop. In addition, due to the small number of nonseminomas in the study, it was difficult to draw any conclusions specifically about a hormone-nonseminoma relationship. Future studies which include a sufficient number of both seminomas and nonseminomas are thereby warranted. Finally, as the P values were not adjusted for multiple comparisons, the results should be interpreted with appropriate caution.

In conclusion, the current study based in the Norwegian Janus Serum Bank cohort found that higher prediagnostic levels of FSH and 3α-diol-3G were associated with increased risk of TGCT. These results are consistent with the hypothesis that poorly functioning, or a dearth of Sertoli cells, predispose to TGCT development, but do not support the hypothesis that insufficient levels of androgens are related to risk.

No disclosures were reported.

Z. Wu: Writing–original draft. B. Trabert: Writing–review and editing. C. Guillemette: Formal analysis, methodology. P. Caron: Formal analysis, writing–review and editing. G. Bradwin: Formal analysis, writing–review and editing. B.I. Graubard: Formal analysis, supervision, writing–review and editing. E. Weiderpass: Writing–review and editing. G. Ursin: Writing–review and editing. H. Langseth: Resources, writing–review and editing. K.A. McGlynn: Conceptualization, resources, supervision, funding acquisition, writing–review and editing.

This work was supported by the Intramural Research Program of the NIH, the NCI.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.