Perinatal factors and the risk of testicular germ cell tumors
Article first published online: 6 MAR 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 122, Issue 11, pages 2600–2606, 1 June 2008
How to Cite
Cook, M. B., Graubard, B. I., Rubertone, M. V., Erickson, R. L. and McGlynn, K. A. (2008), Perinatal factors and the risk of testicular germ cell tumors. Int. J. Cancer, 122: 2600–2606. doi: 10.1002/ijc.23424
- Issue published online: 25 MAR 2008
- Article first published online: 6 MAR 2008
- Manuscript Accepted: 12 DEC 2007
- Manuscript Received: 7 NOV 2007
- Intramural Program of the National Cancer Institute
- National Institutes of Health
- Department of Health and Human Services
- perinatal risk factors;
- testicular cancer
Testicular germ cell tumors (TGCT) are the most common cancer among young men in the United States and Western Europe. Prior evidence suggests that TGCT may arise in perinatal life, although few risk factors have yet been identified. To study the etiology of TGCT, the US Servicemen's Testicular Tumor Environmental and Endocrine Determinants (STEED) case-control study enrolled participants and their mothers between 2002 and 2005. Five hundred twenty-seven mothers of cases and 561 mothers of controls provided information on perinatal variables. Logistic regression was used to calculate odds ratios (OR) and 95% confidence intervals (95%CI) associated with the candidate risk factors. Analyses were conducted for all TGCT together and for each histologic subgroup (seminoma and nonseminoma) separately. Young maternal age (<20 vs. 20–29 years, OR = 1.51, 95%CI: 1.09–2.10), young paternal age (<25 vs. 25–29 years, OR = 1.45, 95%CI: 1.08–1.94), maternal parity (3 vs. 1, OR = 0.63, 95%CI: 0.44–0.90) and breech birth (OR = 1.92, 95%CI: 1.03–3.56) were associated with risk of TGCT. For seminoma, young maternal age (<20 vs. 20–29 years, OR = 1.67, 95%CI: 1.10–2.54), young paternal age (<25 vs. 25–29 years, OR = 1.53, 95%CI: 1.03–2.27), maternal parity (3 vs. 1, OR = 0.58, 95%CI: 0.35–0.96) and low birth weight (<2,500 g vs. 2,500–4,000 g, OR = 1.82, 95%CI: 1.00–3.30) were risk factors. Nonseminoma was associated with breech birth (OR = 2.44, 95%CI: 1.25–4.78) and Cesarean section (OR = 2.10, 95%CI: 1.25–3.54). These results support the hypothesis that TGCT may originate in very early life. © 2008 Wiley-Liss, Inc.
Carcinoma in situ (CIS) is the precursor lesion of testicular germ cell tumors (TGCT) that arise in adolescents and young men.1 CIS, also referred to as intratubular germ cell neoplasia unclassified (IGCNU), is thought to have a premeiotic origin arising from primordial germ cells before or during their migration to the embryonic genital ridge, as evidenced by identical shared mutations in the KIT gene of bilateral TGCT cases.2 Further support for CIS development during early gestation has been provided by comparative studies showing the similarity of these cells to gonocytes and embryonic stem cells3; the similarity of which is considered partly explanatory of the pluripotency demonstrated by TGCTs. CIS is thought to be the initial causal stage of TGCT, the risk for which may then be subsequently modulated by other exposures, the majority of which remain unknown; only cryptorchidism, prior history of TGCT and family history of TGCT have been consistently associated with TGCT risk.4 This working model of TGCT pathogenesis has resulted in a focus upon in utero and early life exposures in attempts to elucidate the etiology of this increasingly incident cancer.5, 6
To date, several studies have investigated perinatal exposures in relation to TGCT risk, although many of these analyses have been of a small number of cases. These studies are likely to have had limited statistical power, especially with regard to any type of histologic stratification. These concerns may be somewhat assuaged in an analysis of the Servicemen's Testicular Tumor Environmental and Endocrine Determinants (STEED) study by provision of a relatively large number of cases of this rare malignancy. This analysis of the STEED study will consider in utero and perinatal factors in relation to the risk of TGCT and its 2 major histologic groups.
Material and methods
The US Servicemen's Testicular Tumor Environmental and Endocrine Determinants (STEED) study methods have been published in detail elsewhere.7 Briefly, between April 2002 and January 2005 participants aged 18–45 years with at least one serum sample stored in the U.S. Department of Defense Serum Repository (DoDSR, Silver Spring, Maryland) were eligible for enrollment. By use of a person-specific identifier, the specimens in the DoDSR computerized database were linked to the Defense Medical Surveillance System (DMSS)8 and to other military medical databases to determine which military personnel had developed medical conditions.
For the STEED study, all men with a sample in the DoDSR who subsequently developed TGCT while on active duty were eligible to participate as cases. Men with a sample in the DoDSR who did not subsequently develop TGCT were eligible to participate as controls. Diagnoses of TGCT were limited to classic seminoma or nonseminoma (embryonal carcinoma, yolk sac carcinoma, choriocarcinoma, teratoma, mixed germ cell tumor), as spermatocytic seminoma occurs primarily among older men and is thought to have an etiology distinct from other TGCTs. The diagnoses were based on the original pathology reports or on review (6.5 percent) of the pathology slides.
The study was designed as a pair-matched, case-control study. Year of birth (within 1 year), race/ethnicity (White, Black, other) and date of available serum sample (within 30 days) were the variables used for matching. The database linkage identified nine hundred sixty-one case men who appeared to meet the study criteria. Further review found that 76 men could not be traced, 27 had died, 3 were known to be deployed to a combat zone and 2 were found 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 agreed to participate, resulting in a participation rate of 91%. In the instances where the potential case participant was deceased, the study attempted to obtain proxy information from the man's mother. Thirteen proxy questionnaires were completed by the mothers of the 27 deceased men. Among the controls, 2,579 were evaluated for inclusion. Of these men, three hundred eighty-five men could not be traced, 18 had died, 64 were known to be deployed to a combat zone and 2 were found to be ineligible. In addition, 928 could not be contacted within 30 days. Of the remaining 1,182 men, 32 were in the process of being contacted when the study closed. Thus, of the 1,150 men contacted, 928 agreed to participate, resulting in a participation rate of 81%. In total, 767 cases and 928 controls were recruited, of which there were 720 matched case-control pairs.
Each participant was asked for permission to contact his mother to enroll her in the study. Permission was given to contact a total of 1,247 mothers, 43 of whom were found to be ineligible, 28 of whom were incompletely enrolled at study completion and 16 of whom could not be located. Of the 1,160 eligible mothers contacted, 72 refused to participate. Overall there were 527 case mothers and 561 control mothers completely enrolled. Participating mothers were interviewed over the phone by a female interviewer. Supervising interviewers listened in, at random, to the interviews, to assure that the interviews were conducted in a similar fashion across all the mothers. The perinatal exposures section included variables such as maternal age at birth, birth weight, birth length and parity. The study was approved by the institutional review boards of the National Cancer Institute, Bethesda, Maryland, and the Walter Reed Army Institute for Research, Silver Spring, Maryland.
Odds ratios (ORs) and 95 percent confidence intervals (95%CI) were calculated to estimate the association of perinatal variables with risk of TGCT. Because of exclusion of cases and controls without a completed mother's questionnaire, unmatched analyses using unconditional logistic regression were performed. As this involved breaking the match, risk estimates derived were first minimally adjusted, taking into account only the 3 matching factors. Further adjustment (in the fully adjusted model) was then made for the known TGCT risk factors: history of cryptorchidism and family history of testicular cancer. Comparable conditional logistic regression models rendered similar results and so only the results of the unconditional logistic regression models are presented. When applicable, tests for linear trend in risk according to the medians of each quantile of a given ordered categorical variable were conducted to evaluate possible dose-response relationships. In addition, p values for the comparison of two likelihood ratios, one derived from retention of the variable of interest in the model and the other from exclusion, are reported for each categorical variable. Analyses were stratified by tumor histology to estimate risks specific to both seminoma and nonseminoma. These 2 histologic groups were also compared against one another in the logistic regression models as to test whether risk estimates were significantly different by histology. As minimally adjusted and fully adjusted results using unconditional logistic regression were similar, only the fully adjusted estimates are presented. Statistical analyses were conducted with STATA 10 software.9 All tests were 2 sided, with p < 0.05 defined as statistically significant.
The characteristics of the study population are shown in Table I. In comparing all TGCTs to controls, age, race/ethnicity and family history of testicular cancer were similar (p ≥ 0.05) while cryptorchidism was significantly more common among the cases (p ≤ 0.01). When stratified by histology, family history of testicular cancer was significantly associated only with seminoma (p = 0.008), while cryptorchidism was significantly associated only with nonseminoma (p < 0.001). However, it should be noted that this analysis is restricted to those mothers of sons who were recruited. These histologic differences are much attenuated in the full STEED study cohort.7
|Variable||Controls (n = 561)||TGCTs1 (n = 527)||Seminoma (n = 210)||Nonseminoma (n = 317)|
|No.||%||No.||%||p value2||No.||%||p value2||No.||%||p value2|
|Family history of testicular cancer3|
The relationships of the perinatal variables to TGCT are shown in Table II. Sons born to younger mothers (<20 years) were at significantly higher risk of TGCT than were sons born to older mothers (OR = 1.51, 95%CI: 1.09–2.10), although the trend and global p values were not significant. Though the association with young maternal age was seen in both TGCT histologic groups, it was statistically significant only among the sons with seminoma. Paternal age showed a similar relationship to risk possibly due to correlation with maternal age (r2 = 0.61). When both parental ages were entered into the model together, neither the younger maternal age group (OR = 1.28, 95%CI: 0.89–1.84) or younger paternal age group (OR = 1.34, 95%CI: 0.98, 1.34) remained statistically significant [data not tabulated]. Because of the collinear relationship of these 2 exposures, both variables were dichotomized at the median for a stratified analysis. Although the risk estimate for younger maternal age (≤23 years) and older paternal age (>26 years) group was statistically significant (OR = 0.63, 95% CI: 0.39–0.99) when compared to the referent age group (maternal ≤23 years, paternal ≤26 years), the magnitude of the estimates for the other age groupings were too similar to single out maternal age as the cause of the observed association (ORmaternal > 23, paternal ≤ 26 = 0.72, 95% CI: 0.47–1.11; ORmaternal > 23, paternal > 26 = 0.81, 95% CI: 0.62–1.05) [data not tabulated].
|Variable||Controls||TGCT1||Seminoma||Nonseminoma||Seminoma vs. Nonseminoma|
|No.||No.||Odds ratio3||95% CI||p value||No.||Odds ratio3||95% CI||p value||No.||Odds ratio3||95% CI||p value||p value|
|Mother's age at birth|
|p for trend||0.21||0.062||0.83||0.10|
|Father's age at birth|
|p for trend||0.13||0.064||0.61||0.15|
|p for trend||0.066||0.021||0.55||0.056|
|p for trend||0.065||0.058||0.43||0.28|
|≥2,500 g and <4,000 g||457||416||1.00||Referent||157||1.00||Referent||259||1.00||Referent||Referent|
|p for trend||0.58||0.91||0.52||0.57|
|Before due date||177||179||1.25||0.91–1.71||0.16||68||1.25||0.82–1.92||0.30||111||1.26||0.87–1.83||0.86|
|On due date||166||135||1.00||Referent||58||1.00||Referent||77||1.00||Referent||Referent|
|After due date||204||192||1.17||0.86–1.59||0.33||76||1.21||0.80 - 1.83||0.38||116||1.15||0.80–1.66||0.83|
|Birth weight and gestational age||0.48|
|≥2,500 g and on or after due date||361||324||1.00||Referent||132||1.00||Referent||192||1.00||Referent||Referent|
|1,500–2,499 g and on or after due date||9||3||0.38||0.10–1.43||0.16||2||0.59||0.12–2.81||0.52||1||0.21||0.03–1.72||0.83|
|≥2,500 g and before due date||151||140||1.04||0.79–1.38||51||0.97||0.66–1.42||89||1.10||0.80–1.52||0.41|
|1,500–2,499 g and before due date||25||37||1.58||0.92–2.69||0.22||17||2.114||1.09–4.12||0.06||20||1.29||0.68–2.43||0.31|
|p for trend||0.26||0.069||0.65||0.19|
|Presentation or delivery|
|Singleton birth vs. multiple birth|
|High blood pressure during pregnancy|
|Extreme nausea during pregnancy|
|Vomiting during pregnancy|
|Excessive bleeding during pregnancy|
A maternal parity of 3 was significantly associated with TCGT (OR = 0.63, 95%CI: 0.44–0.90), though the trend and global p values failed to reach statistical significance. As with maternal age, the effect was statistically significant only among the cases with seminoma (OR = 0.58; 95%CI: 0.35–0.96; p for trend = 0.021). Adjusting for maternal age did not alter the risk estimates attained. When the data were dichotomized into primiparous births versus higher order births, primiparous births were significantly associated with risk of TGCT (OR = 1.27, 95%CI: 1.00–1.63) with the association more clearly evident among men with seminoma (OR = 1.42, 95%CI: 1.02–1.98) than nonseminoma (OR = 1.15, 95%CI: 0.87–1.54) (p valueseminoma vs. nonseminoma = 0.12) [data not tabulated]. The analyses of maternal gravidity also indicated an inverse association.
Birth weight was not associated with an increased risk of TGCT overall. Low birth weight was statistically significant among cases with seminoma (OR = 1.82, 95%CI: 1.00–3.30), although the global p value was not statistically significant. High birth weight appeared to be marginally associated with seminoma, although the estimate did not attain statistical significance (OR = 1.50, 95%CI: 0.94–2.41). Gestational age, modeled as on-due-date, before-due-date, after-due-date, was not significantly related to risk. The ORs for birth weight and gestational age were not altered when maternal age was added to the model [data not shown].
To try to disentangle the variables of birth weight and gestational age, Yerushalmy's classification system10 was used for dichotomization of these variables, albeit using “before due date” and “on or after due date,” rather than 37 weeks, as the cut-point. In full term pregnancies, low birth weight (1,500–2,499 g) appeared to have no effect on TGCT risk, while in early term deliveries there was a suggestion that low birth weight increased TGCT risk (OR = 1.58, 95%CI: 0.92–2.69) and this association was more apparent in the seminoma group (OR = 2.11, 95%CI: 1.09–4.12). This may indicate that birth weight is a more important factor than gestational age in determining TGCT risk, although the numbers of individuals in some cells of this stratified analysis were small and the global p value was not statistically significant.
Breech birth was significantly related to risk of TGCT (OR = 1.92, 95%CI: 1.03–3.56), primarily due to a significant association with nonseminoma (OR = 2.44, 95%CI: 1.25–4.78) (p valueseminoma vs. nonseminoma = 0.07). In addition, Cesarean section was significantly related to nonseminoma (OR = 2.10, 95%CI: 1.25–3.54) (p valueseminoma vs. nonseminoma = 0.01). When maternal parity was added to the Cesarean section model, the risk estimate for TGCT became statistically significant (OR = 1.71, 95%CI: 1.03, 2.86) and the risk of nonseminoma increased further (OR = 2.43, 95%CI: 1.42, 4.18) [data not tabulated], while the estimates for breech birth were not modulated [data not shown]. Maternal age had no effect on these variables, either alone or in conjunction with maternal parity.
High blood pressure during pregnancy was indicative of an increased risk for TGCT, but this association did not reach statistical significance (OR = 1.64, 95%CI: 0.95–2.85). Pre-eclampsia, however, was not associated with risk, though the number of events was not great.
The risk of TGCT was unaltered in relation to son's length at birth, ponderal index, singleton vs. multiple birth status, jaundice, extreme nausea, vomiting or excessive bleeding during pregnancy.
Of the published articles of perinatal exposures in relation to TGCT risk, few have presented extensive analyses and even fewer have stratified by the two main histologic types. This is most likely due to the rarity of this cancer and thus the difficulty associated with recruiting large numbers of cases. The STEED study has been able to recruit a relatively large number of men with TGCT into a case-control study through utilization of military medical databases. Moreover, the mothers of participants were enrolled into the study to provide perinatal data for analysis.
The finding that maternal age is inversely associated with TGCT risk is supported by three large, recently published studies.11–13 Analyses of this variable in relation to testicular cancer risk, however, are far from a state of concurrence. The majority have found no association,14–27 while other studies have found a statistically significant TGCT risk with increasing maternal age.28, 29 Moreover, two analyses which were nonsignificant were still suggestive of an increased risk with increasing maternal age.21, 25 There is no obvious explanation for these antithetical results but it is not inconceivable that an increased TGCT risk is associated with both younger (<20) and older (≥30) maternal age at birth, compared to maternal age in the 20s. Geographic variation in the distribution of maternal age at birth may lend differing levels of power to each analysis, especially if risk modifying effects are apparent only at the extremes of the range. If such were true, it would run contrary to what one would expect from the “estrogen hypothesis,”30 given that the reported levels of these hormones are low for women <20 years, high for 20–24 year-olds and intermediate for those 24years of age and older.31 Alternatively, TGCT risk may only be increased with younger maternal ages, as reported here in the STEED study and by other recent publications.11–13 Although the stratified analysis could not distinguish between young maternal and young paternal age at birth as the important exposure, previous studies have found paternal age not to be associated with TGCT risk,21, 23, 25 while evidence from univariate analyses suggesting an association with paternal age are likely to be a product of its high correlation with maternal age.29
Maternal parity appeared to be inversely associated with TGCT risk, an association which was more clearly evident when primiparous birth and higher parity were compared. This statistically significant association was first reported by Swedlow et al.24 and has subsequently been found by other studies.21, 23, 28, 32 While it should be noted that the majority of previous studies have failed to show such an association,11–13, 15–17, 19, 20, 22, 25–27, 33–36 some of the statistically nonsignificant estimates appear to indicate an increased TGCT risk in sons of primiparous mothers.12, 22, 36 In addition to these observations, previous studies have also indicated that being first born,21, 28 or at least being earlier in the birth order,24 is primarily a risk factor for seminoma which is consistent with this analysis of STEED study data. The causal link between low birth order and TGCT risk has been speculated to be related to endogenous maternal estrogen levels37 but other explanations, such as late exposure to a common infectious agent or a different psychosocial environment, are also possible.
The association between low birth weight and TGCT risk was not statistically significant, although the estimate itself indicated an increased risk which was on the threshold of statistical significance for the seminoma group. Nearly all previous statistically significant estimates have indicated that low birth weight is associated with an increased TGCT risk17, 25, 36, 38 with a further 2 studies finding such a relationship only in seminoma,29, 39 the latter of which was not replicated when the study was updated.13 Of the many statistically nonsignificant estimates for birth weight and TGCT risk,12, 13, 15, 16, 20, 26–28, 34, 35, 40, 41 those with borderline statistically significant estimates also propose low birth weight as a risk factor for TGCT.12, 27, 34, 39 To date, only 2 studies have found a statistically significant increased TGCT risk with high birth weight,13, 22 the former mediated predominantly by seminoma, the latter by nonseminoma. Moller and Skakkebaek25 did find a similar association with seminoma but this was statistically nonsignificant. From these data one may surmise that the overall evidence is in favor of an increased risk with low birth weight, but the findings of a recent meta-analysis do not substantiate such a conclusion, especially given that a subgroup analysis of registry-based data produced an overall OR of 1.01 (95% CI: 0.73, 1.40).42
An intriguing finding was that breech birth and Cesarean section both appeared to be risk factors exclusive to nonseminoma. A previous study of breech birth and TGCT found a similar, albeit statistically nonsignificant, estimate12 to that presented here, while other studies have found no association.17, 25 Depue et al.36 found that Cesarean section to be associated with TGCT with an estimate on the threshold of statistical significance. Other studies have failed to replicate this finding12, 17, 43 and one study even proposed a decreased risk in those delivered by Cesarean section,25 although the confidence interval is inexact due to no Cesarean sections being observed in the TGCT group. This dearth of observations typifies analyses of rare exposures in a rare cancer; many of these studies have low statistical power to analyze breech birth or Cesarean section in relation to TGCT. The evidence presented here and the findings by Coupland et al.12 indicate that breech birth is likely to increase TGCT risk and that this may be confined to nonseminoma, while, from the current epidemiologic evidence, any association between TGCT and Cesarean section remains unclear. The associations of these 2 exposures with TGCT may be mediated through a similar causal pathway, given that Cesarean section is often the preferred mode of delivery for babies in breech position. Unfortunately, reasons for Cesarean section were not known and this prevented analysis of the hypothesis that breech presentation is the underlying cause of the association between Cesarean section and TGCT.
The causes of breech birth are not fully understood and this makes interpretation of an association with TGCT difficult. Breech presentation is known to be associated with cryptorchism,44 which is itself a risk factor for TGCT, although only 2 of the 28 TGCT cases from a breech birth were cryptorchid. Therefore, further elucidation of the etiopathogenesis of breech presentation is warranted to aid our understanding of its association with TGCT.
Advantages of this analysis are that sample sizes are large and the cases and controls were drawn from the same well-defined population (military servicemen). In addition, the male US military population is not limited to a single geographic area or subset of the population, which makes results from this study generalizable to other populations. The study also included only pathologically confirmed TGCT, suggesting that the results are somewhat more precise than studies that have enrolled participants without regard to histology or confirmation of diagnoses. Study limitations include that some potential participants could not be contacted because of deployment and that mothers of participants were asked to remember events that happened years in the past without record validation. The inability to contact men due to deployment presents a potential bias in that deployed men might be different in some way that nondeployed men. As most young men in military service are healthy and fit, however, it would seem unlikely that deployment would confer substantial bias. The reliance on memory of events in the past is common to most case-control studies and is difficult to avoid when researching rare tumors, especially tumors that may have an early life etiology. These concerns may be somewhat abated by the fact that perinatal data were ascertained from the mothers of cases and controls, a source which is known to be a more accurate as compared with the index participants themselves.45
In conclusion, this study finds young parental age at birth, low birth order, low birth weight and breech birth to be statistically significant risk factors for TGCT, all of which are supported by previous research. In addition, some of the associations appeared to be histologically specific, observations of which have also been noted in other studies of TGCTs and are substantiated here by statistically significant p values for comparison of histologic groups (Table II). However, this should not detract from the fact that seminoma and nonseminoma share many risk factors, which is consistent with the idea that they arise from a common precursor, but it should motivate the need to histologically stratify out future analyses. Evidence for many other perinatal variables in relation to TGCT risk remains inconsistent, and so the field must strive to increase statistical power, whether this be through the design of new large studies, pooled analyses by consortia or systematic review and meta-analysis.
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 Department of the Army or the Department of Defense. The authors wish to thank Ms. Emily Steplowski of IMS for her contributions to data management.
- 4The epidemiology of testicular cancer. In: CooneyKA,FoulkesWD, eds. The genetics of male reproductive cancers: prostate and testes. New York: Springer, in press., .
- 9StataCorp. Stata Statistical Software: Release 10. College Station, TX: StataCorp LP, 2007.