The incidence of testicular germ cell tumors (TGCT), the most common type of testicular malignancy, has increased during the past 40 years in the United States (US), with the most rapid recent increases observed among men of Hispanic descent (also referred to as Latinos)[1, 2]. It is the most commonly occurring cancer among male adolescents and young adults (AYA), defined by ages 15–19 years and 20–39 years, respectively[3, 4]. Globally, TGCT incidence has increased in almost every country in which it has been studied[5]. TGCT is comprised of two major histologic types: seminomas and nonseminomas. AYA develop both seminomas and nonseminomas, while infants (ages 0–4 years) are almost exclusively diagnosed with nonseminomas (either yolk sac tumors or teratomas)[6]. The increases in incidence rates are greatest among AYA males and men under age 50[7].

TGCT in AYA are characterized by high heritability and polygenic architecture, and it has been suggested that almost half of TGCT are due to inherited genetic factors[8–10]. However, the increasing incidence rates of TGCT are suggestive of environmental factors in TGCT etiology,[11, 12], alone or in combination with genetic factors, although to date, no environmental factors have been identified.

Estrogens and androgens play a central role in the development of the testis,[13] and may have a critical role in the etiology of TGCT. Germ cell neoplasia in situ (GCNIS), the precursor lesion of the seminomatous and non-seminomatous TGCT most common among AYAs, strongly resemble fetal gonocytes, supporting the hypothesis that TGCT is of fetal origin and a late-onset manifestation of the failure of normal fetal differentiation of primordial germ cells to spermatogonia[14]. Infantile TGCT is thought to differ from that of adult and AYA TGCT and be unrelated to GCNIS[15]. Another critical window for the development of TGCT may be the transition from intra-uterine to extra-uterine life, which is marked by a postnatal surge of the newborns’ steroid sex hormones and other regulatory hormones.

While originally hypothesized that TGCT is related to increased exposure to estrogens during development[16], later studies contradicted this theory[17]. More recently, it has been proposed that androgen insufficiency, or an imbalance between androgens and estrogens during testis development, may be relevant[5, 17]. Supporting studies have been retrospective focusing on maternal exposure to exogenous hormones during pregnancy and tumors occurring 20–40 years later in male offspring[5, 18]. However, assessing the potential role of perinatal sex hormones in the neonate has proven challenging, mainly due to difficulties in obtaining perinatal biospecimens, particularly when TGCT onset is typically decades later. Regarding the postnatal period, epidemiologic studies have not indicated that external hormonal exposures are associated with TGCT[5].

The current study is the first to examine the relationship between sex steroid hormone levels at birth and risk of TGCT among infant and adolescent males.

Of the eight sex steroids examined, seven were present in the newborn samples (androstenedione, DHEA, estradiol, estriol, estrone, testosterone, and progesterone). DHT was not detected or below the LOD in all samples and are therefore not included in any tables. The 2.5% – 97.5% range for all hormone measures among controls were similar to those observed from the freshly collected DBS specimens measured in the pilot study, suggesting that major degradation was not observed in blood spots archived for up to 20 years. Similarly, hormone levels were similar across birth years among controls (Supplementary Table 4), demonstrating the relative stability of steroid hormones in DBS specimens up to 20 years old. Descriptive statistics (mean, median, standard deviation, minimum value, maximum value, and range) and the number of samples with measures below the LOD, for each of the 7 measured hormones, are shown in Table 1. The proportion of measures that fell below the LOD ranged from 1.3% (progesterone) to 98.2% (estradiol). Because there were so few subjects with estradiol and estrone measures above the LOD (n=13 and n=39, respectively), they were not included in subsequent analyses. Hormone levels by case-control status, race/ethnicity, and decade of birth are shown in Supplementary Table 4.

The 370 TGCT cases and 344 age and race/ethnicity-matched controls were similar with respect to all socio-demographic and birth characteristics compared (Table 2). As sampled, 25% of cases were diagnosed between ages 0–4 years, and 75% were diagnosed between ages 15–19 years. Eight percent of cases were classified as seminomas, and 92% were non-seminomas, including 35% mixed germ cell tumors, 21% yolk sac tumors, and 11% embryonal carcinomas. All cases were neoplasms of the testis, with 54% arising from descended testes, 1% from undescended testes, and 44% from unspecified descended or undescended testes.

High androstenedione levels were associated with TGCT among boys diagnosed during adolescence (15–19 years) (Q4 vs Q1 OR: 2.33; 95% CI: 1.37–3.97; p-trend<0.01) but not among boys diagnosed in infancy (0–4 years) (Q4 vs Q1 OR: 0.70; 95% CI: 0.28–1.77) (p-value for interaction<0.01) (Table 3). A similar finding was observed for testosterone (15–19 years, Q4 vs Q1 OR: 1.73; 95% CI: 1.00–3.00 vs 0–4 years, Q4 vs Q1 OR: 0.67; 95% CI: 0.26–1.72 (p-value for interaction=0.17). Androstenedione and testosterone exhibit a weak positive linear relationship (r2=0.32). In models mutually adjusting for each hormone (androstenedione and testosterone), the observed association with androstenedione among adolescents remained (Q4 vs Q1 OR: 2.2; 95% CI: 1.25–3.91), while the association with testosterone was attenuated (Q4 vs Q1 OR: 1.28; 95% CI: 0.71–2.33). The associations between high androstenedione and adolescent TGCT appeared to be stronger among non-Hispanic whites (Q4 vs. Q1 OR: 3.29; 95% CI: 1.33–8.12 vs. 1.95; 95% CI: 0.97–3.94 in Hispanics; p-value for interaction = 0.15). Similarly, the association with testosterone was limited to non-Hispanic whites (Q4 vs. Q1 OR: 2.97; 95% CI: 1.21–7.30, compared to 0.82; 95% CI: 0.39–1.74 in Hispanics; p-value for interaction = 0.03) (Table 4). There was no evidence that the relationship between any hormone and adolescent TGCT differed by major histologic subtype (seminoma vs. non-seminoma) or method of delivery (vaginal vs. caesarian birth).

In the first TGCT study with direct measures of sex steroid hormones at birth, we found that elevated androgens (androstenedione and testosterone) were associated with an increased risk of TGCT, while estriol, the predominant estrogen during pregnancy and the only estrogen present in detectable concentrations at birth, was unrelated. The association of androstenedione and testosterone was observed only in TGCT diagnosed in adolescence and not TGCT diagnosed in infancy. The marked difference in risk by age at diagnosis (infants vs. adolescents) was not unexpected; germ cell tumors arising in infants are predominantly teratomas and yolk sac tumors[6], are not preceded by GCNIS, and are thought to be etiologically distinct from those arising in AYA[15]. The association between androgens and TGCT was stronger among non-Hispanic white males, relative to those reporting Hispanic ethnicity. While non-Hispanic white men have higher incidence rates than other ethnic groups within the same geographic region[28], this result does little to further explain the observed recent increase in incidence rates among Hispanic adolescents in the United States[1, 2].

Androgen insufficiency and/or an imbalance between androgens and estrogens during critical windows of testis development have been posited as a factor in the development of TGCT[5]. Direct evidence of these relationships in humans is sparse, however, due to the protracted time between perinatal life and disease onset, and to the rarity of TGCT. Supporting studies have been retrospective, focusing on maternal exposure to exogenous hormones during pregnancy[5, 18].

The results from the current study are based upon hormone levels at birth and reflect the hormone environment during late gestation/extremely early postnatal life; they suggest that biologic mechanisms operating during this period and related to androgens promote germ cell pathogenesis. However, since sex steroid hormone levels vary during and after pregnancy, the measures in our study do not necessarily reflect the hormonal in utero milieu during germ cell differentiation and initiation/promotion of germ cell neoplasia in situ in early pregnancy. Therefore our study may not adequately test the hypothesis that TGCT arises from hormone imbalance during early gestation.

The steady increases in TGCT strongly point to environmental causes, and it has been hypothesized that exposure to endocrine disrupting chemicals (EDCs) may play a fundamental role[4, 29–31]. Sources of exposure to EDCs are diverse and vary widely around the world, and in humans, can disrupt reproductive and sexual development. Fetuses, infants, and children may have greater susceptibility than adults, thus the impact of EDCs during gametogenesis, fetal development, and early life can be particularly important, even though effects may not become apparent until adolescence and adulthood[32]. Male reproductive disorders thought to be related to TGCT, including undescended testes, hypospadias, and poor semen quality, have been induced in rodents after perinatal exposure to EDCs[5, 29, 30], and exposure to EDCs has been associated with shortened anogenital distance, a sensitive marker of androgen action in utero among male newborns[33]. Although it has been speculated that exposure to EDCs during fetal development plays a role in TGCT development[34], the lack of prospectively collected samples from the prenatal period has hindered direct examination of this relationship in humans. [34–36]

Strengths of this study include its inclusion of neonatal pre-diagnostic specimens, unbiased data collection from cases and controls, large sample size, ethnic diversity, and validated laboratory methods. There are some limitations, however, that should be considered. First, given the design of the study, our study did not include TGCT cases that were diagnosed in California but born elsewhere. In addition, it is possible that some of the controls could have moved from California prior to a diagnosis of TGCT, and therefore not linked to the CCR. Given the rarity of TGCT, however, the chance of this is low. More important, there are no data to suggest that such out-migration is associated with hormonal status. Sex steroid hormone measures in this study represent a single point in time; specifically, the window immediately after birth, when the hormonal milieu is a mixture of those being produced by the newborn testis and the residual maternal hormones. As such, these levels do not necessarily reflect levels during other potentially important windows of exposure, such as periods of gonocyte differentiation early in pregnancy and puberty. However, as the first study with direct measures during the neonatal period, the information obtained in this investigation provides the first piece in defining the relationships between these hormones and future risk of TGCT. Due to the low concentrations of some sex steroid hormones at birth and the small amount of biospecimen used for assays, some compounds of interest were below the limit of detection for most or all of the samples, including estradiol, estrone and dihydrotestosterone. As technology and methods for more precise quantification of steroid hormones in dried blood spots are developed, the role of these, and potentially other, compounds can be more clearly defined. Cryptorchidism and personal/familial history of testicular cancer are known risk factors for TGCT, although contributing to a small fraction of testicular cancer[28]. While our record linkage study is the first to measure sex hormones at birth, data for cryptorchidism was not available on California birth record. Future studies measuring sex hormones levels at birth should attempt to collect and account for these characteristics. Finally, the relatively small sample size, particularly for infant TGCT, hindered our ability to examine some associations,

We consider this study to be the first step in a more expansive investigation into the rising incidence of TGCT observed worldwide. Future investigations should include 1) the confirmation of the observed associations with androgens, perhaps in a larger population that includes young men with the highest incidence, 2) environmental in utero and postnatal exposures, including EDCs, and their impact on the developing fetus and future TGCT risk, and 3) an examination of sex steroid hormones at other critical periods of development. In conclusion, this study fills a gap in the current knowledge of early life origins of TGCT, as well as quantifies sex hormone levels at birth among males.