Ralph A. DeFronzo

Aim To compare the efficacy of triple therapy (metformin/exenatide/pioglitazone) versus stepwise conventional therapy (metformin → glipizide → glargine insulin) on liver fat content and hepatic fibrosis in newly diagnosed, drug‐naïve patients with type 2 diabetes. Methods Sixty‐eight patients completed the 6‐year follow‐up and had an end‐of‐study (EOS) FibroScan to provide measures of steatosis (controlled attenuation parameter [CAP] in dB/m) and fibrosis (liver stiffness measurement [LSM] in kPa); 59 had magnetic resonance imaging‐proton density fat fraction (MRI‐PDFF) to measure liver fat. Results At EOS, HbA1c was 6.8% and 6.0% in triple and conventional therapy groups, respectively ( P = .0006). Twenty‐seven of 39 subjects (69%) receiving conventional therapy had grade 2/3 steatosis (CAP, FibroScan) versus nine of 29 (31%) in triple therapy ( P = .0003). Ten of 39 (26%) subjects receiving conventional therapy had stage 3/4 fibrosis (LSM) versus two of 29 (7%) in triple therapy ( P = .04). Conventional therapy subjects had more liver fat (MRI‐PDFF) than triple therapy (12.9% vs. 8.8%, P = .03). The severity of steatosis (CAP) ( r = 0.42, P < .001) and fibrosis (LSM) ( r = −0.48, P < .001) correlated inversely with the Matsuda Index of insulin sensitivity, but not with percentage body fat. Aspartate aminotransferase (AST) to Platelet Ratio Index (APRI), non‐alcoholic fatty liver disease fibrosis score (NFS), plasma AST, and alanine aminotransferase (ALT) all decreased significantly with triple therapy, but only the decrease in plasma AST and ALT correlated with the severity of steatosis and fibrosis at EOS. Conclusions At EOS, subjects with type 2 diabetes treated with triple therapy had less hepatic steatosis and fibrosis versus conventional therapy; the severity of hepatic steatosis and fibrosis were both strongly and inversely correlated with insulin resistance; and changes in liver fibrosis scores (APRI, NFS, Fibrosis‐4, and AST/ALT ratio) have limited value in predicting response to therapy.