Childbearing patterns and ovarian cancer in the Million Women Study: variation by histological subtype
1Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK
A reduced risk of ovarian cancer amongst parous women has been reported by many investigators; reports of associations with breastfeeding have been varied. Histopathological and molecular evidence suggests the different ovarian cancer subtypes have distinct origins, but few epidemiological studies have sufficient cases to explore variation in risk by histological subtype.
We examined associations between ovarian cancer and childbearing patterns in the Million Women Study, a prospective cohort of UK women. Using Cox regression models, we estimated adjusted relative risks (RR) of ovarian cancer in women with different childbearing patterns, using competing risks to explore heterogeneity by histological subtype.
After excluding women with previous cancer, bilateral oophorectomy, and unknown parity, the study population included 1,146,985 women, aged 56 years on average at recruitment. 7570 incident cases of ovarian cancer accrued, after an average of 13.0 years of follow-up. Overall, parous women had a 26% lower risk of ovarian cancer than nulliparous women (RR: 0.74, 95%CI: 0.69-0.79), with significant heterogeneity by histological subtype (p-het<0.00001). Amongst parous women, each additional birth was associated with a 7% reduction in relative risk of ovarian cancer (RR-per-birth: 0.93, 95%CI: 0.90-0.96), with heterogeneity by tumour subtype (p-het = 0.04). We found little evidence of a trend with increasing age at first or last birth. Overall, breastfeeding (ever vs never) was not associated with ovarian cancer (RR: 0.94, 95%CI: 0.89-1.00, p=0.07), and no variation was seen by histological subtype (p-het=0.4). However, there was a reduced risk of ovarian cancer with longer durations of breastfeeding (per 6-month increase in average breastfeeding duration per child - RR: 0.88, 95%CI: 0.83-0.93, p<0.001).
In this largest prospective study of ovarian cancer to date, our results provide new reliable evidence that associations with reproductive factors vary between different subtypes of ovarian cancer, consistent with hypotheses of different aetiologies.