Introduction Because studies of the association between alcohol intake and the risk of primary liver cancer use varying cut-off points to classify alcohol intake, it is difficult to precisely quantify this association by meta-analysis of published data. Furthermore, there are limited data for women in prospective studies of the dose-specific relation of alcohol intake and the risk of primary liver cancer.

Methods We analysed original data from four population-based prospective cohort studies encompassing 174 719 participants (89 86 men and 84 856 women). After adjustment for a common set of variables, we used Cox proportional hazards regression to estimate HRs and 95% CIs of primary liver cancer incidence according to alcohol intake. We conducted a meta-analysis of the HRs derived from each study.

Results During 1 964 136 person-years of follow-up, 804 primary liver cancer cases (605 men and 199 women) were identified. In male drinkers, the multivariate-adjusted HRs (95% CI) for alcohol intakes of 0.1–22.9, 23.0–45.9, 46.0–68.9, 69.0–91.9, and ≥92.0 g/day, as compared with occasional drinkers, were 0.88 (0.57 to 1.36), 1.06 (0.70 to 1.62), 1.07 (0.69 to 1.66), 1.76 (1.08 to 2.87), and 1.66 (0.98 to 2.82), respectively (p for trend=0.015). In women, we observed a significantly increased risk among those who drank ≥23.0 g/day, as compared with occasional drinkers (HR: 3.60; 95% CI 1.22 to 10.66).

Conclusion This pooled analysis of data from large prospective studies in Japan indicates that avoidance of (1) heavy alcohol drinking (≥69.0 g alcohol/day) in men and (2) moderate drinking (≥23.0 g alcohol/day) in women may reduce the risk of primary liver cancer.