Parental Occupational Exposure to Heavy Metals and Welding Fumes and Risk of Testicular Germ Cell Tumors in Offspring: A Registry-Based Case–Control Study Free

Testicular cancer is a relatively rare disease, accounting for about 1% of all male cancers; however, it is the most common cancer type among males aged 15 to 44 years in developed countries (1). Over the last four decades, the incidence of testicular cancer has increased rapidly particularly in affluent countries (2–4) and started to increase in countries shifting toward higher levels of development (4). Descriptive epidemiologic studies have shown temporal and geographic variations in testicular cancer incidence (2–4), suggesting an etiologic role of environmental factors. However, environmental risk factors for testicular cancer have been poorly defined.

The predominant subtype of testicular cancer is testicular germ cell tumor (TGCT), accounting for 98% of testicular cancer cases. It is presumed that the precursor of TGCT, germ cell neoplasia in situ, is derived from primordial germ cells that failed to differentiate in utero (5–7). Two studies assessing parental occupations before childbirth showed that occupations related to metals (8), wood processing (8), and manufacturing (9) are associated with a greater risk of testicular cancer in the offspring. An exposure to heavy metals and/or other toxins that are present in these workplaces may cause various adverse health effects. For instance, cadmium (Cd), lead (Pb), and mercury (Hg) can influence the endocrine system (10), thereby disrupting embryonal programming and gonadal development in utero (6, 11, 12), and chromium (Cr), Cd, iron (Fe), nickel (Ni), Pb, and copper can increase reactive oxygen species (13), which may damage DNA in the male germline and increase health risk in offspring (14).

Existing epidemiologic data suggest an association between parental exposures to heavy metals and risk of hypospadias or cryptorchidism in offspring. A study showed that probable or possible paternal heavy metal exposure significantly increases the risk of hypospadias and cryptorchidism (15). The same study (15) and two other studies (16, 17) suggested that the risk of hypospadias may be elevated with probable or possible maternal heavy metal exposure. While these male urogenital malformations and testicular cancer have been hypothesized to share, to some degree, common etiology (7), no epidemiologic studies have examined the association between parental exposure to heavy metals and testicular cancer in offspring.

The current study aims to assess the associations of maternal and paternal exposures to heavy metals and welding fumes during preconception and/or prenatal periods with the risk of TGCT in offspring using data from a population-based case–control study in the Nordic countries, the NORD-TEST Study.

The general methodology used in the NORD-TEST Study has been described previously (18) and is summarized briefly herein. The NORD-TEST Study is a registry-based case–control study conducted in Finland, Norway, Sweden, and Demark, aiming to investigate various parental occupational exposures in relation to the risk of TGCT in offspring. The current study pooled data from Finland, Norway, and Sweden where the form of collection of occupational information was comparable across the respective registries.

We identified all males newly diagnosed with testicular cancer without a prior primary neoplasm (excluding non-melanoma skin cancer) during the following calendar periods: 1988–2012 for Finland, 1978–2010 for Norway, and 1979–2011 for Sweden via nationwide cancer registries. The following morphologic codes were used to identify testicular cancer cases in each country-specific cancer registry: ICD-7:178; ICD-8 and ICD-9: 186; and ICD-10 and ICD-O-3: C62.

Males without a prior primary neoplasm (excluding non-melanoma skin cancer) at the time of case's diagnosis were randomly selected from each country's central population registry. Four controls were individually matched to each case by country and year of birth.

In the three countries combined, a total of 9,613 cases were identified. Of those, we excluded 196 cases diagnosed at ages under 14 years or over 49 years. We further excluded 321 testicular cancer cases that were not classified as either seminoma or non-seminoma and additional 984 cases with no information on parental occupations in or before the year of case's birth, leaving 8,112 cases for the analysis. For the remaining cases, there were a total of 26,264 matched controls with information on occupation(s) held by father and/or mother in or before the year of control's birth. The final study sample included 314 cases with one matched control, 1,213 cases with two matched controls, 2,816 cases with three matched controls, and 3,769 cases with four matched controls.

The NORD-TEST Study has been approved by the relevant data protection and ethics committees in Finland, Norway, and Sweden, and also by the International Agency for Research on Cancer ethics committee.

Each Nordic country has population registries where every resident is registered with a unique personal identification number. The identification numbers were obtained from cancer registries (cases) or the central population registry (controls, parents, siblings) and were used to link data across registries including census, birth registry, and cancer registry.

Occupational information was collected in population censuses every five years in Sweden (1960–1990 except for the 1965 census) and Finland (1970–1990) and every ten years in Norway (1960–1990). We retrieved job codes indicating parental occupations from the most recent census administered before the year of subject's birth or the census administered in the year of subject's birth. We then applied the Nordic job-exposure matrices (NOCCA-JEM; ref. 19) to determine exposures of Cr, Fe, Ni, Pb, and welding fumes for each parent. The NOCCA-JEM provides country-specific data on the proportion of workers exposed (P) and the mean level of exposure (L) estimated for over 300 occupations, over 20 agents, and four calendar periods covering 1945–1994 (19). For each agent, maternal and paternal occupational exposure indices were calculated as the product of proportion of persons exposed and the mean level of exposure (P × L). The exposure index of Cd was not available in the NOCCA-JEM, but available in the Finnish job-exposure matrix (FINJEM; ref. 20). Therefore, we estimated parental Cd exposure only for Finnish cases and controls.

Furthermore, we collected information on paternal and maternal ages at birth from the birth registry; age at TGCT diagnosis and family history of testicular cancer (father, brother) from the nationwide cancer registries; and personal history of inguinal hernia, cryptorchidism and hypospadias from the birth registry, the deformities or congenital malformation registry, and/or the hospital discharge registry (18).

First, we assessed bivariate associations of TGCT with the following factors: father's history of testicular cancer, brother's history of testicular cancer, personal history of inguinal hernia, hypospadias, and cryptorchidism, age of father at childbirth, and age of mother at childbirth. For this analysis, conditional logistic regression models were fitted to retain the case–control matches and the associations were tested with a Wald test. Furthermore, we estimated pair-wise correlations between binary exposure variables (absence: P × L = 0, presence: P × L > 0).

Next, we fitted logistic regression models conditioned on year and country of birth to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for TGCT risk associated with Cr, Fe, Pb, Ni, and welding fumes (all three countries) and with Cd (Finland only). We analyzed both maternal and paternal heavy metal exposures in two ways. First, three groups for each parental exposure were generated on the basis of the presence or absence of exposure: (i) parental occupations exposed to neither heavy metals (Cr, Fe, Pb, Ni, or Cd) nor welding fumes (reference group); (ii) parental occupations exposed to a specific heavy metal agent with or without other heavy metal(s)/welding fumes; and (iii) parental occupations exposed to heavy metal(s)/welding fumes without the specific heavy metal agent (base model). Second, each parental exposure was categorized on the basis of the exposure index (P × L): (i) parental occupations that are not exposed to a specific heavy metal agent (reference group); (ii) parental occupations with a lower exposure index of the specific agent; and (iii) parental occupations with a higher exposure index of the specific agent. The cutoff between the low and high categories was determined on the basis of the 90th percentile of paternal exposure index among the exposed (Cr 6.00 μg/m³; Fe 1.98 mg/m³; Ni 3.30 μg/m³; Pb 0.30 μmol/L; welding fumes 6.30 mg/m³; Cd 0.30 μg/m³) or the 75th percentile of maternal exposure index among the exposed (Cr 6.00 μg/m³; Fe 0.27 mg/m³; Ni 0.99 μg/m³; Pb 0.30 μmol/L; welding fumes 0.12 mg/m³; Cd 0.02 μg/m³). The 75th percentile cutoff instead of the 90th percentile cutoff was used for maternal exposure to avoid having too few subjects in the high category.

We assessed potential confounding effects by adding the following covariates to the base model one at a time and evaluating the change in the OR estimate: age of father at birth, age of mother at birth, and family history of testicular cancer [father and/or brother(s)]. We considered a 10% or greater change in the estimated OR as presence of confounding effect. In evaluating the potential dose–response relationship of TGCT risk associated with heavy metal exposure, we fit P × L as a continuous variable. We also ran the base model for seminoma and non-seminoma, separately. The potential heterogeneity of effect by TGCT subtype (seminoma, non-seminoma) was then examined by testing whether the regression coefficient associated with each exposure differs by subtype based on Chow test statistic (21).

Finally, we conducted two sensitivity analyses. First, we checked robustness of the results by repeating the analysis using an alternative way of categorizing the parental jobs (Sensitivity analysis 1). The parental exposure was classified on the basis of different combinations of P and L values rather than P × L: (i) P = 0 (nonexposed, reference group), (ii) 0 < P < 50 and 0 < L < median value (among the exposed), (iii) 0 < P < 50 and L ≥ median value, (iv) P ≥ 50 and 0 < L < median value, and (v) P ≥ 50 and L ≥ median value. Second, we repeated the analysis while restricting the study sample to cases and controls with information on parental occupations in the year of or the year before subject's birth, aiming to account for potential information bias due to changes in parental occupations between the time of census and the critical preconception and prenatal periods (Sensitivity analysis 2). When no census is available in the year of or the year before childbirth but the same occupation was indicated in the last census before childbirth and the first census after childbirth, we assumed that the same occupation was held during the preconception/prenatal period.

All statistical tests were two-sided (α = 0.05) and were performed by using SAS 9.4 or STATA 12.1.

The study population consisted of 8,112 cases and 26,264 matched controls. The average age at diagnosis of TGCT was 29.2 years (SD = 7.0; range: 14–49). Of the 8,112 cases, 45% had seminoma and 55% had non-seminoma (Table 1). About 3% of cases and controls' fathers did not have information on their occupations, and about 13% of cases' mothers and 12% of controls' mothers did not have information on their occupations. The most common jobs exposed to heavy metals and/or welding fumes were machine and engine mechanics, turners, and welding and flame cutting among fathers, and electronics and telecommunications and assembly of machine/metalware or electrical/electronic equipment among mothers. The proportions of fathers and mothers who held a job exposed to any heavy metals/welding fumes prior to childbirth were 19.6% and 1.5% for cases and 19.1% and 1.6% for controls, respectively. The conditional logistic regression analysis indicated that neither parent's age at childbirth nor personal history of inguinal hernia was associated with TGCT (all P ≥ 0.53; Table 1), whereas father's (OR = 2.95; 95% CI, 2.11–4.12) or brother's (OR = 5.30; 95% CI, 3.45–8.12) history of testicular cancer and personal history of hypospadias (OR = 2.52; 95% CI, 1.62–3.90) or cryptorchidism (OR = 2.56; 95% CI, 1.94–3.38) were associated with an increased risk of TGCT.

In both fathers and mothers, the most common heavy metal exposure was Pb (17.2% of fathers and 1.4% of mothers). Pairwise correlations between binary exposure variables ranged from 0.58 to 1.00 for paternal exposures and 0.18–1.00 for maternal exposures (Supplementary Table S1). As jobs exposed to welding fumes are also exposed to Fe and vice versa, no distinction was made between these exposures except in the analysis of different exposure levels. For the subsequent analyses, we did not mutually adjust for these exposures due to their high correlations (Cr, Fe, Ni, Pb, and welding fumes) or the absence of substantial change after adjustment for another heavy metal agent (Cd).

We herein present unadjusted results of the conditional logistic regression analysis because the addition of covariates did not substantially alter the OR estimates. The results from the base models showed no statistically significant association between parental exposures to heavy metals/welding fumes and the risk of TGCT in their sons (Table 2). When we examined different levels of exposure index, we found that sons of fathers with a low exposure index of Fe/welding fumes, Ni, or Pb and sons of mothers with a low exposure index of Cr had roughly 8 to 9% higher odds of developing TGCT than those of fathers/mothers with occupations that are not exposed to any of the heavy metals (Table 3). Of these agents, only paternal exposure to Fe/welding fumes was statistically significantly associated with an increased TGCT risk (OR = 1.09; 95% CI, 1.01–1.18). However, this OR was greater than the OR associated with a high exposure index (OR = 0.97; 95% CI, 0.79–1.19). We also observed statistically nonsignificant elevated ORs associated with higher indices of paternal exposure to Cr and maternal exposures to Fe, Ni, Pb, Cd, and welding fumes and no statistically significant dose–response relationship (P_trend ≥ 0.32).

The ORs associated with paternal exposures did not differ significantly by histologic subtype (Table 4). Although the point estimates of ORs associated with maternal exposures to Fe/welding fumes, Ni, and Cd appeared to be somewhat greater for seminoma than for non-seminoma type, and the point estimate of OR associated with maternal Pb exposure appeared to be greater for non-seminoma than seminoma, these differences were not statistically significant (All P ≥ 0.15) and could be due to random variability.

The first sensitivity analysis was conducted only for all paternal exposures to Cr, Fe, Pb, and welding fumes and maternal exposures to Cr, Fe, and Pb as there were too few subjects in two or more categories for other exposures. The patterns of ORs according to different levels of exposure were similar to what we observed in the analysis using exposure indices (Table 5). In this sensitivity analysis, we found significantly elevated ORs associated with the highest category of paternal exposure to Cr (OR = 1.37; 95% CI, 1.05–1.79) and the second highest category of paternal exposure to welding fumes (OR = 1.39; 95% CI, 1.10–1.76).

In the second sensitivity analysis assessing parental occupations held in the year of or the year before childbirth, the ORs associated with all paternal exposures were similar but slightly greater and the ORs associated with maternal exposures to Cr, Ni, and Pb were greater than the ORs observed in the primary analysis (Supplementary Table S2). For instance, the estimated OR for TGCT associated with maternal exposure to Cr changed from 1.03 (95% CI, 0.74–1.43) to 1.22 (95% CI = 0.75–1.99).

To our knowledge, this is the first large-scale epidemiologic study examining parental occupational exposures to heavy metals and welding fumes in relation to the risk of TGCT in offspring. Our results provide little evidence of associations between parental exposure to heavy metals and welding fumes and TGCT risk in offspring with the potential exception of high paternal Cr exposure.

A previous study showed an increased risk of testicular cancer among sons whose fathers were metalworkers or employees of metal products before conception (9). Provided that paternal exposure to heavy metal can cause DNA damage in the male germ line, which is associated with various types of morbidity, including childhood cancer in the offspring (14), it is plausible that sons whose fathers were exposed to heavy metals prior to conception might have an increased risk of TGCT. In the current study, jobs with a low exposure index of Fe/welding fumes was statistically significantly associated with an increased risk of TGCT; however, the ORs associated with a higher exposure index were closer to the null, providing little support for the causal association between paternal heavy metal exposure and the risk of TGCT in offspring.

Animal studies have suggested that prenatal exposure to endocrine disrupting chemicals adversely affects the development of reproductive organs in offspring (22, 23). While evidence in humans remains limited partly due to challenges in measuring prenatal exposures and insufficient statistical power to detect potential associations (24), some epidemiologic studies have indicated an association between prenatal exposure to endocrine disrupting chemicals and offspring's risk of urogenital malformations (23, 25–29). For instance, a case–control study showed that boys with hypospadias and their mothers had higher blood levels of Cd and Pb compared with their respective counterparts (28). It is plausible that prenatal exposure to these heavy metal agents containing endocrine disrupting properties increases the risk of TGCT by disrupting the growth of offspring's testes. However, the current study found no statistically significant association between maternal exposures to heavy metals or welding fumes and TGCT risk in their sons. The OR associated with maternal Cd exposure was somewhat greater than other heavy metals/welding fumes, but the estimate was derived from a relatively small subset of the study sample.

Like other studies where JEM is applied, the current study is subject to exposure misclassification due to inter-individual variations within each occupation. A misclassification of individual exposure status is particularly likely for jobs with low prevalence of exposure. Therefore, a careful consideration of the potential resultant bias is crucial. When we categorized exposures based on different combinations of P and L values, the ORs associated with the highest exposure categories were not statistically significant except for paternal Cr exposure. The data indicated that paternal occupations where both P and L of Cr exposure are high are associated with a roughly 37% greater risk of TGCT compared with paternal occupations without Cr exposure. The OR associated with the second highest category of paternal exposure to welding fumes was also statistically significantly elevated, but the OR associated with the highest category was not. Although the statistically significant finding may be due to inflated type I error, hexavalent Cr is a well-established human carcinogen (30) and disrupts spermatogenesis through oxidative stress in adult monkeys (31). Our finding warrants further studies to examine potential biological mechanisms underlying the association between paternal Cr exposure and TGCT risk in offspring.

Exposure misclassification is also possible when parental occupational information was obtained from a census administered years before the birth of cases and controls. Nearly 68% of parental occupational information was based on censuses administered two to nine years before the year of childbirth. The sensitivity analysis assessing parental occupations held during the year of or the year before childbirth, we found similar or somewhat greater ORs compared with the estimates from the primary analysis, which could be due to the increased precision in the exposure estimate or the greater variability expected from an analysis of a smaller sample.

It is also important to note that 11% of cases and 12% of controls were excluded due to the lack of information on parental occupations. Excluded subjects were more likely to be Finnish, diagnosed at an older age and in earlier years, seminoma type, have younger parents (age < 20 years at childbirth), and are less likely to have a family history of testicular cancer compared with included subjects (data not shown). As these factors did not modulate the association between parental heavy metal exposure and the risk of TGCT in the offspring, we believe that our results are unlikely to have been biased due to the exclusion of subjects without information on parental occupations.

The aim of the current study is to assess the association of TGCT with heavy metal exposure during preconception or pregnancy period; however, we cannot preclude the possibility that a parent's cumulative exposure and/or a subject's exposure after birth plays an etiologic role in TGCT. While we found no association between cases and controls' occupational exposure to heavy metals/welding fumes and their risk of TGCT (data not shown), the associations of TGCT with cumulative parental exposure before childbirth and nonoccupational heavy metal exposures (e.g., tobacco smoke, alcoholic beverages, food, and water) during infancy, childhood, and adolescence remain to be investigated.

Despite these potential limitations, the current study has major strengths. One of the strengths is its large sample size with the high representativeness of cases. As reporting of cancer cases is compulsory in the Nordic countries, the cancer registries cover nearly 100 percent of all cancer cases occurring in respective countries (32). Another strength is the use of high-quality population-based registries (33) where every resident is registered with a unique identification number. It provided us with a unique opportunity to link data between registries for not only cases and controls, but also their parents and siblings. Furthermore, the Nordic JEMs were developed to assess a wide variety of agents occurring in occupational settings in each country (19, 20), allowing us to assess occupational exposures to specific heavy metal agents and welding fumes.

The NORD-TEST Study provides little evidence of association of TGCT with preconception or prenatal exposure to heavy metals and welding fumes, with the potential exception of paternal exposure to Cr. Given the plausible adverse impacts of Cr on DNA in male germline and health in offspring, research on potential biological mechanisms underlying the association with TGCT risk in offspring warrants further attention.

No potential conflicts of interest were disclosed.

Conception and design: C. Le Cornet, M. Feychting, T. Tynes, E. Pukkala, A. Olsson, S.O. Dalton, K.-C. Nordby, N.E. Skakkebæk, B. Fervers, J. Schüz

Development of methodology: C. Le Cornet, T. Tynes, E. Pukkala, J. Hansen, A. Olsson, K.-C. Nordby, B. Fervers, J. Schüz

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Le Cornet, M. Feychting, T. Tynes, E. Pukkala, J. Hansen, S.O. Dalton, S. Uuksulainen, T. Woldbæk, B. Fervers, J. Schüz

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Togawa, M. Feychting, E. Pukkala, B. Fervers, J. Schüz

Writing, review, and/or revision of the manuscript: K. Togawa, C. Le Cornet, M. Feychting, T. Tynes, E. Pukkala, J. Hansen, S.O. Dalton, K.-C. Nordby, S. Uuksulainen, P. Wiebert, N.E. Skakkebæk, B. Fervers, J. Schüz

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Le Cornet, T. Tynes, J. Hansen, P. Wiebert, T. Woldbæk, B. Fervers

Study supervision: B. Fervers, J. Schüz

We would like to thank Marianne Steding-Jessen and Andrea Meersohn from the Danish Cancer Society Research Center, Karin Fremling from the Institute of Environmental Medicine, Karolinska Institutet, and Veronique Luzon from the International Agency for Research on Cancer for their help with the data management. We would also like to thank the NOCCA team who gave us permission to use FINJEM and NOCCAJEM and also provided their expertise in occupational exposures in the Nordic countries. The Family-Cancer Database was created by linking registers maintained at Statistics Sweden and the Swedish Cancer Registry. Data from the Finnish Cancer Registry was extracted with the permission for the research No.THL/1123/5.05.00/2012.

This work was supported by public funding from the Lyric Grant INCa-DGOS-4664 (Institute of Cancer Research, France; received by J. Schüz). K. Togawa was supported by European Commission FP7 Marie Curie Actions–People–Co-funding of regional, national and international programmes (COFUND).

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.