Summary Associations between the incidence of stromal and germ cell ovarian cancer and pregnancies were examined in a follow-up of 1.1 million women aged 20-56 years. Stromal tumours (41 cases) showed no clear associations. Germ cell tumours (71 cases) were related to high-age childbirths and short time since birth.Keywords: stromal tumour; germ cell tumour; parity; population-based study; prospective study (Cramer and Welch, 1983), but few studies have compared the two cancer types. Furthermore, it has been suggested that exposure to oestrogens in utero may be a risk factor for ovarian cancer of germ cell origin Walker et al, 1988). If endogenous female sex hormones play a role in the aetiology of germ cell tumours, the very high levels of such hormones during pregnancy might also influence the risk of germ cell tumours for the mother.The aim of the present study was to investigate the potential relations between the incidence of ovarian cancer of stromal and germ cell origin and the number of full-term pregnancies, age at first and last birth and time since last birth in a large cohort of Norwegian women. We have previously reported on relations between the risk of epithelial ovarian cancer and reproductive factors in the same data set (Albrektsen et al, 1996).
MATERIALS AND METHODSThe present study includes all Norwegian women born in the period 1935 (Serov et al, 1973) as germ cell tumours (35 dysgerminomas, 25 malignant teratomas, nine embryonal carcinomas, two choriocarcinomas) and 41 cases as sex cord/stromal tumours (23 granolusa-stromal cell tumours, eight androblastomas, one Sertoli cell tumour, nine sarcomas). Separate analyses were performed for sex cord/stromal tumours (referred to as stromal tumours) and germ cell tumours.
Statistical analysesPotential relations between the risk of ovarian cancer of stromal and germ cell origin and reproductive factors were examined in a log-linear Poisson regression model of person-years at risk (Breslow and Day, 1987). In this context, a woman was considered to be at risk of developing ovarian cancer from the age of 20 years. Certain analyses included parous women only. Date of each delivery was recorded, and a woman contributed person-years in successive categories of attained age, number of full-term pregnancies, age at last birth and time since last birth. A woman was withdrawn from follow-up at the date of ovarian cancer diagnosis, emigration or death. Women with a diagnosis of ovarian cancer different from the histological classification under consideration (including epithelial ovarian cancer) were withdrawn at date of diagnosis.All analyses were adjusted for attained age in 1-year intervals and birth cohort in 5-year intervals. In the statistical model, 767