Evaluating the role of alcohol consumption in breast and ovarian cancer susceptibility using population-based cohort studies and two-sample Mendelian randomization analyses.

Alcohol consumption is correlated positively with risk for breast cancer in observational studies, but observational studies are subject to reverse causation and confounding. The association with epithelial ovarian cancer (EOC) is unclear. We performed both observational Cox-regression and two-sample Mendelian randomization (MR) analyses using data from various European cohort studies (observational) and publicly available cancer consortia (MR). These estimates were compared with World Cancer Research Fund (WCRF) findings. In our observational analyses, the multivariable-adjusted hazard ratios (HR) for a one standard drink/day increase was 1.06 (95% confidence interval; 1.04,1.08) for breast cancer and 1.00 (0.92,1.08) for EOC, both of which were consistent with previous WCRF findings. MR ORs per genetically predicted one standard drink/day increase estimated via 34 SNPs using MR-PRESSO were 1.00 (0.93,1.08) for breast cancer and 0.95 (0.85,1.06) for EOC. Stratification by EOC subtype or estrogen receptor status in breast cancers made no meaningful difference to the results. For breast cancer, the confidence intervals for the genetically derived estimates include the point-estimate from observational studies so are not inconsistent with a small increase in risk. Our data provide additional evidence that alcohol intake is unlikely to have anything other than a very small effect on risk of EOC. This article is protected by copyright. All rights reserved.