A Meta-Analysis of Six Placebo-Controlled Trials of Thiazolidinedione Therapy for HIV Lipoatrophy

Objective: To determine the impact of thiazolidinediones (TZD) on changes in limb fat mass in HIV-infected individuals, particularly in those not receiving a thymidine analogue. Methods: Individual patient data from placebo-controlled, randomized trials of rosiglitazone (n=5) or pioglitazone (n=1) were combined. Generalized estimating equation (GEE) models were used to estimate the treatment effect on changes in limb fat mass. Results: In the combined dataset of 427 patients, the baseline median age was 45 years, 86% were male, 80% were Caucasian, 63% were receiving stavudine (d4T) or zidovudine (AZT), 66% were on protease inhibitors, and median body mass index was 23 kg/m(2). In a univariate GEE model, TZD was associated with an increase in limb fat mass (coeff = 0.14 kg vs placebo, P=.04). In a multivariable GEE model, patients receiving pioglitazone had significantly higher limb fat mass gains (coeff = 0.35 kg, P <= .01) compared to patients receiving placebo, while patients on rosiglitazone did not (coeff =0.05 kg, P=.48). Interactions between thymidine analogue use and rosiglitazone and pioglitazone were not significant. Conclusions: In this meta-analysis, pioglitazone therapy was more effective than placebo to increase limb fat mass whereas rosiglitazone was not significantly better than placebo. The effectiveness of these drugs did not vary according to whether the patients were receiving thymidine analogues.