Speaker
Description
Although alcohol is recognized as a type 1-carcinogen, prospective studies evaluating the association between alcohol consumption and the risk of pancreatic cancer (PC) have generally faced power limitation to generate consistent evidence among never smokers. Here, we evaluated the association as part of the Diet and Cancer Pooling Project, a large international consortium of prospective studies.
We pooled individual-level data for 2,459,382 participants from 32 cohorts, of whom 10,082 developed incident PC during follow-up (10,953,275 person-years). Alcohol intake at baseline was collected using food frequency questionnaires and expressed in grams of ethanol per day (10g corresponds to ~1 drink). Cox proportional hazards models were used to estimate multivariable hazard ratios (HR) and 95% confidence intervals (CI). Analyses were conducted in men and women combined and stratified by smoking status.
The population comprised 70% drinkers overall (mean intake: 13g/day) and men (38%) drunk on average twice as much as women. In the overall population, a statistically significant positive association between alcohol intake and PC risk was observed in both continuous (total number of cases [nPC]=10,082; for a 10g/day increment: HR=1.03, 95%CI:1.02-1.04) and categories (HR= 1.10, 95%CI: 1.02-1.20 for 30-<60g/day and HR=1.32, 95% CI:1.18-1.47 for ≥60g/day compared with 0.1–<5g/day), with a significant trend across categories (ptrend=<0.001). Among never smokers, there was not statistically significant association either in continuous (nPC=3,835; HR=1.02, 95%CI:0.99-1.05) or in categories (HR=1.25, 95% CI:0.94-1.66 for ≥60g/day compared with 0.1–<5g/day, ptrend=0.19).
Results from a large consortium of prospective studies suggested that alcohol consumption may not be independently associated with PC risk due to heavy confounding by tobacco smoking. Future steps of analysis will entail examination by type of alcoholic beverage, geographical region, and PC histological subtype.
