OBJECTIVETo investigate the association between diabetes and risks of primary liver cancer.

METHODSA Meta-analysis was performed to estimate the pooled relative risk (RR) to evaluate the relationship between diabetes and the risk of primary liver cancer from cohort studies, which were identified by searching in Medline, Chinese CNKI and Wanfang databases from January 1989 to February 2010. A total of 28 publications were found according to this method. Adjusted RRs and their corresponding 95% confidence intervals (95%CI) were calculated by using the fixed-effect and random-effect model in our analysis. We also conducted a number of sub-groups analysis stratified by some important variables, such as source, gender, region and quality of study.

RESULTSA total of 3800 cases of liver cancer and 3 672 248 study subjects from 14 prospective cohorts were included in our analysis. The pooled RR of primary liver cancer was 3.33 (95%CI: 1.82 - 6.10) for persons with diabetes when compared to subjects without diabetes. The results showed a significant association between diabetes and the risk of primary liver cancer based on these cohort studies. Subgroup analysis indicated that the pooled RRs for diabetes were 3.76 (95%CI: 1.69 - 8.38) in the population-based cohorts and 2.41 (1.34 - 4.32) in the hospital-based cohorts. In terms of the sex groups, the pooled RRs for diabetes were 2.32 (95%CI: 1.70 - 3.17) for males and 1.63 (95%CI: 1.08 - 2.47) for females, respectively.

CONCLUSIONAs one of independent risk factors, diabetes was associated with an increased risk of primary liver cancer.