Abstract
BackgroundHeavy drinking was well associated with an increased risk of hepatocellular carcinoma (HCC), whereas the effect of low-to-moderate drinking on HCC remains under debate.MethodsParticipants from the UK Biobank with detailed information on alcohol use and free of common diseases were included. Daily pure alcohol intake (g/day) was calculated, and the predominant alcoholic beverage type was assigned for each participant. Additive Cox regression model and nonlinear Mendelian randomization (NLMR) analyses were performed to evaluate the association of alcohol intake with HCC.ResultsOf 329,164 participants (52.3% females, mean [SD] age = 56.7 [8.0] years), 201 incident HCC cases were recorded during the median follow-up of 12.6 years. The best-fitted Cox regression model suggested a J-shaped relationship between daily alcohol intake level and HCC risk. However, NLMR analysis did not detect a nonlinear correlation between alcohol use and HCC (nonlinearity P-value: 0.386). The J-shaped correlation pattern was detected only in subjects who mainly drank wine but not in those who mainly drank beer, spirits, or fortified wine. Moderate wine drinking showed a significant alanine transaminase (ALT)- and aspartate aminotransferase-lowering effect compared to that of the nondrinkers. In low-risk populations of HCC including women, people aged < 60 years, subjects with normal ALT levels, and those carrying non-risk genotypes of PNPLA3 rs738409 and TM6SF2 rs58542926, we observed a J-shaped correlation between alcohol use and HCC; however, a positive dose-response correlation was found in their respective counterparts, even in those predominantly drinking wine.ConclusionsLow-to-moderate drinking may be inversely associated with the risk of HCC in low-risk populations, which may be largely driven by wine drinking. However, those in high-risk populations of HCC, such as men and older people, and those with abnormal ALT levels and carry genetic risk variants, should abstain from drinking alcohol. Given the small HCC case number, further validations with larger case numbers are warranted in future works.</p>