Oral contraceptive use and cutaneous melanoma risk: A French prospective cohort study

Abstract

Introduction Evidence suggests that cutaneous melanoma, the most lethal form of skin cancer, may be influenced by sex hormones. Oral hormones are the leading contraception method in industrialized countries and represent a significant source of exogenous hormone use. Several studies reported a positive association between oral contraceptive (OC) use and melanoma risk. However, findings were conflicting and data from large prospective studies are lacking. Our aim was to explore the associations between OC use and melanoma risk in women. Methods E3 N is a prospective cohort of 98,995 French women aged 40–65 years at inclusion in 1990. Use of oral hormones for contraception was collected through a self-administered questionnaire at baseline, and detailed information on lifetime use of OCs was then recorded every 2–3 years from 1992. We computed hazard ratios (HRs) and 95% Confidence Intervals (CIs) using Cox regression models adjusted for age, birth cohort, pigmentary traits, residential UV exposure in county of birth and at inclusion, and family history of skin cancer. We stratified the results according to melanoma site and histological type using competing-risk models. To examine potential confounding by sun exposure, we used data from E3N-SunExp, a nested case-control study in which a specific questionnaire on lifetime residential and recreational UV exposures was sent to all incident skin cancer cases and 3 controls per case (matched on age, county of birth, education, and length of follow-up in the cohort) in 2008. Analyses in the sub-cohort were performed through logistic regression modelling. Results Between 1992 and 2008, 539 melanoma cases were ascertained among 79,365 women. In age-adjusted models, there was a modest association between ever use of OCs and melanoma risk (HR = 1.18, 95% CI = 0.98–1.42), which was reduced after full adjustment (HR = 1.14, 95% CI = 0.95–1.38). The association was stronger in long-term users (HR = 1.33, 95% CI = 1.00–1.75 for a duration of use of at least 10 years, compared with never use of OCs). Among ever users, there was an inverse association with age at first use (Ptrend = 0.01), consistent with longer durations of OC use in women with a younger age at first use. However, we found no evidence of an association with age at last use or time since last use. When exploring different formulations, the association between ever use of OCs and melanoma risk was restricted to OCs containing high doses of estrogen (≥ 50 μg of ethinyl estradiol) (HR = 1.26, CI = 1.03–1.53). In subtype- and site-specific analyses, ever use of OCs appeared positively associated with all types and sites of melanoma, except the trunk, although no heterogeneity was detected across subtypes (Phomogeneity = 0.54) or sites (Phomogeneity = 0.23). In the E3N-SunExp population, associations between OC use and melanoma risk were not substantially modified after adjustment for residential or recreational UV exposure. However, while we found no association between OC use and sun exposure, we observed that OC use was positively associated with sunscreen use (Ptrend = 0.04 from no protection to increasing SPF) and tanning bed use (odds ratio = 1.14, 95% CI = 1.01–1.29 for ever vs. never use), even after adjusting for age and year of birth. Conclusions Overall, our findings do not support a strong association between OC use and melanoma risk. OC users were more likely to use sunscreen and tanning beds, suggesting a specific profile of OC users regarding their attitude towards tanning. Therefore, we cannot exclude that the observed relation between OC use and melanoma risk, which has been debated in the literature for almost 40 years, could relate to particular sun exposure behaviors in users. Further research, including social and behavioral description of OC users, is thus critical to increase our understanding of the risk profile of women diagnosed with melanoma, and to question the hypothesis of a hormonal influence on melanoma risk.