Thiazolidinediones and blood lipids in type 2 diabetes.

We evaluated study population characteristics and treatment effects on blood lipids between studies in which either rosiglitazone (RSG) or pioglitazone (PIO) was investigated in patients with type 2 diabetes. We performed a summary analysis of all published double-blind, placebo-controlled studies with RSG (4 and 8 mg/d) and PIO (15, 30, and 45 mg/d). Data were analyzed by the random-effects model. Nineteen trials met our inclusion criteria, yielding 5304 patients, 3236 in studies with RSG and 2068 in studies with PIO. Subjects treated with PIO were more obese and showed more pronounced hyperglycemia and dyslipidemia (increased triglycerides and decreased HDL cholesterol) at baseline than did subjects treated with RSG. By weighted linear-regression analysis, studies with PIO showed greater beneficial effects on triglycerides, total cholesterol, and LDL cholesterol, after adjustment for the respective lipid levels at baseline. RSG 8 mg/d showed greater increases in total cholesterol and LDL cholesterol than did RSG 4 mg/d. PIO 30 mg/d showed greater reductions in triglycerides than did PIO 15 mg/d. Studies conducted with PIO showed more beneficial effects on blood lipids, but also different study population characteristics in comparison with studies conducted with RSG. Differences in both pharmacologic properties between agents and study population characteristics are likely to have influenced the results.