Introduction Despite alcohol being an established carcinogen and recent prospective studies showing increased prostate cancer risk among heavy drinkers, alcohol is not an established risk factor for prostate cancer.

We aimed to investigate the causal role of alcohol on prostate cancer through conventional observational epidemiology techniques and Mendelian randomisation, by using genetic variants influencing the propensity to drink or modifying the physiological response to alcohol.

Methods A case-control study was nested in the case identification phase of a large British population-based RCT for treatment of localised prostate cancer (ProtecT). 2400 prostate-specific antigen detected prostate cancer cases of white ethnicity and 12 700 controls matched on age and general practice provided data on alcohol consumption. Eighteen SNPs in alcohol-metabolising genes (ADH's, ALDH2) were genotyped in a sub-sample of cases and controls.

Results There was some evidence of a modest decrease in low Gleason-grade (RR 0.96; 95% CI 0.93 to 0.99) and increase in high-grade (RR 1.04; 95% CI 0.99 to 1.08; p difference=0.004) prostate cancer per 10 alcohol units/week increase in consumption, not explained by current BMI, blood pressure, co-morbidities, or reverse causation.

Results from genetic association analyses including an interaction between an ADH1B functional variant and alcohol consumption suggested that alcohol causally increases prostate cancer risk. However, this study was underpowered to detect a difference between results from instrumental variable analyses and conventional observational epidemiology models.

Conclusion These results support small increases in high-grade prostate cancer risk caused by heavy alcohol drinking. The required independent replication in populations with incident (non prostate-specific antigen-detected) prostate cancer is currently under-way.