Plasma metabolomic profiles of oral contraception use associated with breast and ovarian cancers among premenopausal women in Nurse's Health Study II

BACKGROUND: While over 80% of sexually active premenopausal women use oral contraception (OC), the biological mechanisms linking OC to hormone-sensitive cancers are not well understood. OC use lowers ovarian cancer risk while current use modestly increases breast cancer risk, suggesting complex underlying pathways. We sought to identify plasma metabolites associated with OC use in premenopausal women and relationships with ovarian and breast cancer. METHODS: Analysis included 2,072 premenopausal women in Nurses' Health Study II with self-reported OC history. OC use was defined as ≥1 years of previous use, excluding current users. All metabolite values were transformed to probit scores to achieve normality. Associations of prior use ≥1 years versus never and per 5 years use with individual metabolites were evaluated through multivariable linear regression, accounting for the number of effective tests (NEF) based on the number of principal components explaining 90% of the variance. Metabolite groups were identified through metabolite set enrichment analysis (MSEA) using false discovery rate (FDR) for multiple testing. OC use models were adjusted for blood draw variables (age, time, season, fasting status), BMI smoking, diet (Alternative Healthy Eating Index score), alcohol consumption, and physical activity at the time of blood draw. Associations of metabolite groups (FDR<0.20) and metabolites (NEF<0.20) were considered nominally associated with OC use. Associations with incident breast (n=663) and ovarian cancers (n=34) diagnosed at least ≥3 years after blood draw were assessed using conditional logistic regression, using controls matched 1:1 on blood draw variables, adjusting for age, OC duration, parity, and family history of breast or ovarian cancer and using MSEA. RESULTS: Of 2,703 women, 1,681 (81.1%) reported having previously used OC for ≥1 year, with a mean duration of 4.0±3.9 years. Each additional 5 years of OC use was associated with higher C40:6 phosphatidylethanolamines (PE) levels ( beta =0.09; 95% CI: 0.04, 0.14), and both phosphatidylcholines and PEs were negatively associated with OC use and duration but positively for breast and ovarian cancer risk. Increasing OC duration was associated with lower tryptophan ( beta =-0.11; 95% CI: -0.16, -0.05) and higher 1,7-methyluric acid ( beta =-0.11; 95% CI: -0.16, -0.05), which were both lower among breast cancer cases and higher among ovarian cancer cases, but the overall organoheterocyclic compound group was not significantly associated with OC use. CONCLUSION: Our results suggest long-lasting alterations of systemic metabolism may in part explain associations of OC use with breast and ovarian cancers, particularly among metabolites that contribute to lipid metabolism. Analyses are ongoing and will include a deeper investigation into timing and formulation of OC use as well as cancer subtypes.