Response to pioglitazone in non-alcoholic fatty liver disease patients with vs. without type 2 diabetes: A meta-analysis of randomized controlled trials

Background: Pioglitazone is considered a potential therapy for non-alcoholic fatty liver disease (NAFLD). However, different effects of pioglitazone on NAFLD have been demonstrated in diabetic and non-diabetic patients. Herein, a meta-analysis of randomized, placebo-controlled trials was carried out to indirectly compare pioglitazone in NAFLD patients with vs. without type 2 diabetes.

Methods: Randomized controlled trials (RCTs) of pioglitazone vs. placebo involving NAFLD patients with or without type 2 diabetes/prediabetes collected from databases were enrolled into this analysis. Methodological quality was employed to evaluate the domains recommended by the Cochrane Collaboration. The analysis covered the changes in histology (fibrosis, hepatocellular ballooning, inflammation, steatosis), liver enzymes, blood lipids, fasting blood glucose (FBS), homeostasis model assessment-IR (HOMA-IR), weight and body mass index (BMI) before and after treatment, and adverse events.

Results: The review covered seven articles, with 614 patients in total, of which three were non-diabetic RCTs. No difference was found in patients with vs. without type 2 diabetes in histology, liver enzymes, blood lipids, HOMA-IR, weight, BMI, and FBS. Moreover, no significant difference was revealed in adverse effects between NAFLD patients with diabetes and without DM, except the incidence of edema that was found to be higher in the pioglitazone group than in the placebo group in NAFLD patients with diabetes.

Conclusions: Pioglitazone could exert a certain effect on alleviating NAFLD, which was consistent between non-diabetic NAFLD patients and diabetic NAFLD patients in improving histopathology, liver enzymes, and HOMA-IR and reducing blood lipids. Furthermore, there were no adverse effects, except the incidence of edema which is higher in the pioglitazone group in NAFLD patients with diabetes. However, large sample sizes and well-designed RCTs are required to further confirm these conclusions.