Metformin and Risk of Hepatocellular Carcinoma in Patients With Type 2 Diabetes

Chin-Hsiao Tseng


Liver International. 2018;38(11):2018-2027. 

In This Article


The median follow-up time was 5.3 years in both the unmatched cohort (range: 0.50–5.92 years) and the matched cohort (range: 0.50–5.92 years). In the unmatched original cohort, age and sex differed significantly. The mean age of the never-users was older (64.5 ± 13.5 versus 59.6 ± 12.2 years, P < .0001) and the proportion of men was lower (45.6% versus 47.2%, P < .0001). All other variables, except peripheral arterial disease, sulfonylurea, pioglitazone, alcohol-related diagnoses and Epstein-Barr virus-related diagnoses, differed significantly in the unmatched original cohort. However, in the matched cohort, age and sex were similar and most variables did not differ significantly (except for nephropathy, insulin, sulfonylurea and acarbose; Table 1). The standardized differences in the matched cohort suggested that the 2 groups were well-matched and none of the covariates had a value >10%.

The overall hazard ratios by metformin exposure suggested a significantly reduced risk of HCC in both the unmatched original cohort and the matched cohort. In the tertile analyses, there was a trend towards a decreasing risk of HCC with an increasing cumulative duration of metformin use (Table 2). The P–values suggested a reduced risk in all tertiles of cumulative duration in the unmatched cohort, but a significantly higher risk was observed in the first tertile in the matched cohort (hazard ratio: 1.39, 95% confidence interval: 1.19–1.62).

Sensitivity analyses after excluding patients aged <25 or >75 years did not change the findings remarkably. The overall hazard ratio was 0.45 (0.41–0.49) and the hazard ratios for the first, second and third tertiles of cumulative duration of metformin therapy were 0.84 (0.75–0.93), 0.45 (0.41–0.51) and 0.21 (0.19–0.24) respectively.

In the subgroup analyses of patients with and without liver diseases, a significantly lower risk of HCC associated with metformin use could be seen in all subgroups, and none of the P–interactions were statistically significant (Table 3).

Users of metformin with or without the use of aspirin/statin had a significantly lower risk of HCC, compared to the reference groups. There was a significant interaction between metformin and aspirin or statin (Table 4).