Hepatic Events Prevention by Antihyperglycemic Therapies and Intervention Comparisons in Type 2 Diabetes: The HEPATIC-T2DM Network Meta-analysis.

Abstract

Type 2 diabetes mellitus (T2DM) amplifies liver disease burden, yet the comparative hepatic effects of antidiabetic drugs remain poorly defined. To compare associations between antidiabetic drug classes and major adverse liver outcomes (MALOs) in adults with T2DM. PubMed, EMBASE, and Cochrane Central Register of Controlled Trials were searched from December 1946 through 23 August 2025. Studies enrolling adults with T2DM that evaluated associations between antidiabetic drug classes with regard to MALOs were included. Data were extracted on study characteristics, drug exposures, and MALOs. A three-level Bayesian network meta-analysis with study- and database-level random effects was performed. Outcomes are reported as hazard ratios (HRs) and ranked using the surface under the cumulative ranking curve. Forty-six observational studies (N = 7,124,845) were included. Thiazolidinediones were least associated with hepatocellular carcinoma incidence and significantly lower than DPP-4 inhibitors (HR 0.50), GLP-1RAs (HR 0.72), insulin (HR 0.20), and sulfonylureas (HR 0.69). For decompensation (composite), GLP-1RAs were associated with the lowest hazard compared with all other classes (HRs 0.16-0.91; all significant). SGLT2 inhibitors were least associated with cirrhosis (HR 0.66 vs. DPP-4 inhibitors; HR 0.66 vs. GLP-1RAs). GLP-1RAs were least associated with variceal bleeding and hepatic encephalopathy, whereas SGLT2 inhibitors were least associated with liver-related mortality. All included studies were observational, precluding causal inference. Liver-specific risk reduction is not uniform across antihyperglycemic drug classes. Randomized trials are needed to determine whether these associations reflect true drug effects.