Fertility and offspring sex ratio of men who develop testicular cancer: a record linkage study

Rune Jacobsen 2 3 Erik Bostofte 1 2 Gerda Engholm 2 3 Johnni Hansen 0 2 Niels E.Skakkebaek 2 5 Henrik Mller 2 4 0 Institute of Cancer Epidemiology, The Danish Cancer Society , Strandboulevarden 49, Box 839, DK-2100 1 The Sperm Analysis Laboratory, Health Service Physicians Organisation , Pilestraede, Copenhagen 2 Epidemiology, Institute of Public Health, Faculty of Health Sciences, University of Copenhagen, Panum Institute , Blegdamsvej 3, DK-2200 Copenhagen N , Denmark 3 Centre for Research in Health & Social Statistics , Sejrgade 11, DK-2100 4 Thames Cancer Registry, Guy's, King's and St Thomas' School of Medicine , 42 Weston Street, London SE1 3QD , UK 5 Department of Growth and Reproduction, National University Hospital , Blegdamsvej 9, DK-2100 , Denmark Analysis of associations between testicular cancer, subfertility and offspring sex ratio (proportion of males born among newborns) was performed on 3530 Danish men, born 1945-1980, who developed testicular cancer in the period 1960-1993. As the basis of comparison we used the total population of Danish men born in the period 19451980 (n 1 488 957) and their biological children (n 1 250 989). Men who developed testicular cancer had, prior to the cancer diagnosis, a reduced fertility (standardized fertility rate ratio: 0.93, 95% confidence interval: 0.890.97) and a significantly lower proportion of boys (48.9%, P 0.02) compared with the general population (51.3%). The reduction in fertility was more pronounced in men with non-seminoma but the reduction in offspring sex ratio was independent of histological type. This confirms earlier results from less conclusive studies and indicates that testicular cancer, male subfertility and a female-biased sex ratio among new-born infants are characteristics of male reproduction that are linked by biological mechanisms. - Introduction The increase in incidence of testicular cancer (Coleman et al., 1993; Adami et al., 1994; Forman and Mller, 1994), the decrease in the sex ratio (proportion of males among newborn infants) in many populations (Mller, 1996, 1998) and the possible decrease in semen quality (Carlsen et al., 1992; Swan et al., 1997) lead to the question whether these temporal trends are independent phenomena or, alternatively, are somehow connected to each other (James, 1997; Mller, 1998). Data from a Danish casecontrol study of testicular cancer, based on interviews with 514 cases and 720 controls, suggested a strong association between subfertility and subsequent risk of testicular cancer (Mller and Skakkebk 1999), and produced evidence that men who develop testicular cancer have a lower offspring sex ratio than other men, thus suggesting that testicular cancer, subfertility and low offspring sex ratio are interdependent. However, other studies have found no association between testicular cancer and subfertility (Swerdlow et al., 1989) or between testicular cancer and low offspring sex ratio (Swerdlow et al., 1989; Heimdal et al., 1996). The present paper addresses the hypothesis that there is an association between testicular cancer, subfertility and offspring sex ratio using data on a large complete and unselected cohort of 3530 Danish men who developed testicular cancer. This larger study eliminates the potential problems with interview-based casecontrol studies: information bias due to differential recall or reporting, and selection bias due to non-participation. Materials and methods The population of Danish men born 19451980 who developed testicular cancer in the period 19601993 was identified in the Danish Cancer Registry, which holds information on all cases of cancer in the Danish population (Storm, 1991). The information from the Danish Cancer Registry was linked with data on reproduction from the Fertility Database at Statistics Denmark (Knudsen, 1998). The study population comprised 3530 men who developed testicular cancer and the total population of Danish men born in the period 19451980, regardless of whether they had children or not (n 1 488 957), served as the basis of comparison. The number of children of the men who developed testicular cancer was 3661 and the number of children of men in the comparison group was 1 250 989. The analyses included both live-born and still-born biological children. The men who were married to the mother at the birth of the child were defined as the biological father. In cases where the mother was not married, the man who signed at the birth of the child to be the father was defined as the biological father. When no man had signed, the biological father was identified as the man to whom the biological mother was married or with whom she was living by January 1st in the year of birth of the child. For each man, information was available on date of birth, date of testicular cancer diagnosis, histological type of testicular cancer and date of death. For each child, information was available on sex and date of birth. The analysis was conducted for the testicular cancer group as a whole and separately for the histological groups seminoma and non-seminoma. Fertility rate ratios and offspring sex ratios were calculated for the periods: (i) up to 8 full calendar years before testicular cancer diagnosis, (ii) from 8 years before until 4 years before testicular cancer diagnosis, (iii) from 4 years before testicular cancer until European Society of Human Reproduction and Embryology Histological type of testicular cancer Fertility rate ratio and 95% CIb 0.98 (0.911.05) 0.95 (0.871.05) 0.95 (0.831.09) 0.97 (0.921.02) Fertility rate ratio and 95% CIb 0.86 (0.800.92) 0.84 (0.740.95) 0.88 (0.761.02) 0.87 (0.810.94) Fertility rate ratio and 95% CIb 0.95 (0.901.01) 0.91 (0.840.98) 0.92 (0.831.02) 0.93 (0.890.97) aIncludes 88 cases with unspecified histology. bCompared with paternal fertility rates in the total Danish population, adjusted for paternal age and year of birth and their interaction. CI confidence interval; NS not significant. 2 years before testicular cancer diagnosis. Offspring sex ratios were further calculated from 2 years before until 2 years after testicular cancer diagnosis, and from two years after testicular cancer and onwards. Age and year of birth of the man (in 5 year groups) were included as co-variates in all analyses. Fertility rates were analysed as a function of the covariates using multiplicative Poisson regression models (Breslow and Day, 1987), and fertility rate ratios and 95% confidence intervals (CI) were thereby estimated. The analyses of the proportion of male offspring were similarly carried out by logistic regression analysis (Breslow and Day, 1980). In the analysis of fertility rates, the best fit to the data was obtained by a Poisson regression model that included an interaction term between age and year of birth. This interaction was due to an increase in age-specific fertility with increasing year of birth. Inclusion or exclusion of the interaction term, however, had no material inf (...truncated)