Metformin Use Is Associated with Hepatic Fibrosis Regression in Nonalcoholic Fatty Liver Disease

Abstract

Nonalcoholic fatty liver disease (NAFLD) will soon become the leading etiology of end-stage liver disease, yet there are currently no approved therapies to reduce hepatic fibrosis in NAFLD. Animal studies have shown that metformin can reduce hepatic fibrosis, yet human data is limited, and metformin’s potential as a treatment for NAFLD remains controversial. A commonly cited meta-analysis of 52 patients concluded that metformin does not have histologic benefits in NAFLD, while a pilot clinical trial that found histologic benefits attributed this effect to metformin-induced weight loss. To clarify the effect of metformin on hepatic fibrosis in NAFLD, we evaluated the association between metformin use and fibrosis regression in a cohort of non-cirrhotic NAFLD patients with paired liver biopsies at least one year apart. Using a database of clinical pathology reports, we identified patients who had undergone two clinically indicated liver biopsies at least one year apart, with baseline biopsy demonstrating non-cirrhotic NAFLD, no history of excessive alcohol consumption, and no histological or serological evidence of a competing etiology of chronic liver disease. Among these, we identified 101 patients with either continuous metformin use or no metformin use at either baseline or follow-up time points. The primary outcome was a decrease in fibrosis stage of ≥1 between baseline and follow-up biopsy. We used Cox proportional hazards regression analysis to assess the association between metformin use and fibrosis regression among subjects with non-cirrhotic NAFLD at baseline. Fibrosis regression occurred in 16/30 (53%) metformin users and 28/71 (39%) non-users. The unadjusted hazard ratio (HR) for fibrosis regression among metformin users compared to non-users was 2.21 (95% CI 1.17-4.15). Moreover, after adjustment for established predictors of histologic fibrosis regression, including baseline fibrosis stage, age, diabetes status, body mass index, interval total body weight loss of ≥10%, as well as weight loss surgery, this association remained significant (HR 3.72, 95% CI 1.53-9.06). Among those who regressed, weight loss surgery was less common in metformin users (25%) compared to non-users (43%). The mean annual change in Brunt fibrosis stage between baseline and follow-up biopsy was -0.32 (SE 0.08) in metformin users and -0.11 (SE 0.05) in non-users (p = 0.04). Fibrosis progression was less frequent in metformin users, occurring in 1/30 (3%) metformin users and 18/71 (25%) non-users. However, this association was not statistically significant on Cox regression analysis. Metformin use is associated with an increased frequency of hepatic fibrosis regression in non-cirrhotic NAFLD, even after adjustment for interval weight loss. Our findings support a potential role for metformin in fibrosis regression in non-cirrhotic NAFLD.