We acknowledge that rosiglitazone was associated with modest but statistically significant increases in total cholesterol and LDL cholesterol levels in our study, similar to findings seen in the non-HIV diabetes literature regarding the use of this agent. It is important to recognize, however, that insulin resistance, elevated free fatty acid levels, and hypoadiponectinemia are also significant independent predictors of cardiovascular disease; all of these improved with rosiglitazone therapy in our study sample. Furthermore, evidence shows that peroxisome proliferator–activated receptor- (PPAR-) agonists increase LDL particle size and increase small high-density lipoprotein particles, thereby creating a less atherogenic lipid profile (1). In addition, PPAR- agonists such as pioglitazone may have more favorable effects on lipid levels while improving insulin sensitivity and adipogenesis. We agree that the long-term cardiovascular effects of thiazolidinediones are not known for this population and warrant further investigation.

In contrast to our study, the report by Carr and associates (2) did not demonstrate significant increases in subcutaneous fat after 48 weeks of rosiglitazone therapy compared with placebo in HIV-infected patients with lipoatrophy. There are several important differences between our study and theirs. Hyperinsulinemia, a surrogate marker for insulin resistance, was required in our study and in the study by Gelato and coworkers (3), which also showed increased subcutaneous fat in response to rosiglitazone. The observed increase in subcutaneous fat in our study is consistent with known biological effects of PPAR- agonists in stimulating adipogenesis. Carr and associates (2) showed a 5% mean increase in limb fat with rosiglitazone but a 7% increase with placebo. In contrast, in our study, subcutaneous fat decreased over time in the placebo group but increased in response to rosiglitazone. In Carr and associates' study, negative findings on limb fat may have been related to the spontaneous improvement in limb fat seen in the placebo group. Furthermore, in their study, use of stavudine, a medication associated with progression of lipoatrophy (4), was disproportionate in the rosiglitazone and placebo groups (53% vs. 26%, respectively). In addition to having less severe lipoatrophy, 25% of the patients in our study were women, compared with only 2% in the study by Carr and associates (2). This may also contribute to differences in study results. Although further study is needed, our data indicate a net potential benefit in metabolic variables and body composition in HIV-infected patients with insulin resistance.