Rosiglitazone increases small dense low-density lipoprotein concentration and decreases high-density lipoprotein particle size in HIV-infected patients

After 12 weeks of rosiglitazone treatment, significant increases in total and small dense low-density lipoprotein, and the total: high-density lipoprotein (HDL)-cholesterol ratio were found. The large HDL concentration and HDL particle size decreased significantly with rosiglitazone compared with placebo. These data indicate the production of a more atherogenic lipid profile with rosiglitazone, a consideration when selecting treatment for the growing population of HIV-infected patients with type 2 diabetes and dyslipidemia.

Rosiglitazone is a peroxisome proliferator-activated receptor gamma (PPAR-γ) agonist recognized for its ability to improve glycemic control and insulin sensitivity. Studies of rosiglitazone in HIV-associated lipoatrophy have shown improvements in insulin sensitivity ^[1-6], and in several studies increases in subcutaneous adipose tissue ^[3-5,7]. Undesirable changes in lipid profiles with rosiglitazone have, however, often been documented ^[1-6]. The utility of rosiglitazone for insulin resistance and diabetes in HIV may therefore be limited by its effects on lipids. The negative effect of PPAR-γ agonists on the lipid profile may be ameliorated by specific effects on lipid particle size and changes in atherogenic lipoprotein subclasses. For example, data in type 2 diabetes suggest that some thiazolidinediones increase large low-density lipoprotein (LDL)-cholesterol and LDL particle size ^[8-10].

We previously reported the results of a randomized placebo-controlled trial of rosiglitazone, which showed significant increases in total and LDL-cholesterol in HIV-infected subjects with lipodystrophy ^[3]. Here we present new data on lipoprotein particle size and subclass concentrations, demonstrating a pro-atherogenic effect of rosiglitazone on the lipid profile in HIV to a degree not previously appreciated.

Metabolic and body composition results of the randomized placebo-controlled 12-week trial of rosiglitazone (4 mg/day) for HIV-infected men and women (n = 28) with lipoatrophy and insulin resistance have been reported previously ^[3]. Here we present data on lipid profile and nuclear magnetic resonance spectroscopy of lipoprotein size, particle concentration, and subclass concentration (Liposcience, Raleigh, North Carolina, USA) obtained at baseline and 12 weeks. The methodology employed by Liposcience is described in detail elsewhere ^[11]. Written informed consent was obtained from each subject before the study.

Baseline characteristics were compared between randomization groups using Student's t-test for continuous variables and chi-square statistics for non-continuous variables. Within each group, changes over 12 weeks were calculated by subtracting baseline values from end of study values. The effect of treatment was assessed by performing Student's t-test on the calculated change scores. The non-parametric Wilcoxon rank sum test was used when change scores were non-normally distributed. Additional analyses were performed to adjust for exposure to lipid-lowering medications. All values are presented as mean (SEM), and statistical significance was accepted at the P < 0.05 level. Statistical analyses were performed using SAS JMP software, version 6.0 (SAS Institute, Inc., Cary, North Carolina, USA).

There were no significant differences between groups for age, duration of HIV, duration of antiretroviral therapy, protease inhibitor use, body mass index, CD4 T-cell cell count or the percentage of subjects with HIV viral RNA levels below the limit of detection. Six subjects randomly assigned to placebo, and nine randomly assigned to rosiglitazone were on lipid-lowering medications during the study; one subject on rosiglitazone commenced a lipid-lowering medication during the study.

There were no significant differences between groups in total very low-density lipoprotein (VLDL), LDL or high-density lipoprotein (HDL) concentrations nor the mean lipoprotein particle size at baseline. Furthermore, subclass concentrations of VLDL, LDL, and intermediate-density lipoprotein (IDL) were no different at baseline (Table 1). Total LDL and IDL concentrations increased with rosiglitazone compared with placebo, whereas total HDL and VLDL did not change. Mean VLDL and LDL particle sizes did not change, but the mean HDL particle size decreased in the rosiglitazone compared with the placebo-treated group.

Table 1

Small LDL and very small LDL increased significantly with rosiglitazone compared with placebo, whereas large HDL decreased. There was also a significant difference in the change in the cholesterol: HDL-cholesterol ratio [mean change with rosiglitazone +1.02 (0.41) versus -1.23 (0.40) with placebo, P = 0.0006]. In all cases, the significant differences noted between treatment groups remained significant after the analyses were adjusted for exposure to lipid-lowering therapy.

Lipid abnormalities are common among HIV-infected patients on antiretroviral therapy ^[12,13]. In the present study, rosiglitazone, given to improve lipoatrophy and insulin resistance, worsened the lipid profile in HIV-infected subjects by increasing the concentration of smaller LDL particles, and decreasing the large HDL concentration and HDL mean particle size.

Increases in LDL-cholesterol with the use of rosiglitazone are well documented in studies of diabetes ^[14,15] and in HIV ^[1-6]. In a meta-analysis of the effects of thiazolidinediones on the risk of cardiovascular disease in diabetes, Chiquette et al. ^[14] identified a mean increased LDL-cholesterol of 15 mg/dl (95% confidence interval 13-18 mg/dl). We previously reported a 0.4 mmol/l (15 mg/dl) increase in LDL-cholesterol in the study of rosiglitazone for HIV lipoatrophy ^[3], which is similar to the LDL-cholesterol increases reported in other trials in HIV (range 7-30 mg/dl) ^[2,4,5]. The significance of this modest increase and the impact on overall cardiovascular disease risk, however, is more fully appreciated when lipid particle subpopulations are assessed.

Prospective, population-based studies in the general population have shown that small, dense LDL particles are associated with an increased risk of coronary artery disease ^[16]. Different PPAR-γ agonists may have different effects on LDL particle size. For example, the use of troglitazone increased large LDL and mean LDL particle size after 8 weeks of treatment for type 2 diabetes ^[8]. Reductions in small, dense LDL have been reported with pioglitazone in patients with type 2 diabetes and hypertension ^[10,17]. In a large study of type 2 diabetes and hyperlipidemia ^[18], both rosiglitazone and pioglitazone led to increased LDL particle size, but with a greater effect of pioglitazone. In contrast, in the current study, rosiglitazone was associated with significant increases in small, dense LDL particles compared with placebo, thereby shifting patients to a more atherogenic lipid profile. These data suggest that thiazolidinediones have varying effects on lipids, not only on the overall lipid profile ^[14], but potentially important different effects on lipoprotein size and particle subclasses.

We also identified a significant reduction in HDL size and increases in IDL with rosiglitazone compared with placebo. Stein et al. ^[19] previously reported decreased HDL particle size in HIV-infected patients compared with controls, and this decrease was present irrespective of the use of protease inhibitors. Therefore, decreases in HDL size and the reduction in the total cholesterol: HDL-cholesterol ratio with rosiglitazone marks a further exacerbation of the dyslipidemia present at baseline in this population.

This was a relatively small study and the use of lipid-lowering therapy was not exclusionary; however, both treatment groups had a similar percentage of patients on lipid-lowering therapy. Statistical adjustment for the use of lipid-lowering therapies did not alter the significant findings in the study.

This study indicates a shift towards a more atherogenic lipid profile with the addition of rosiglitazone in patients with HIV, lipoatrophy and insulin resistance. The deleterious effects of rosiglitazone on lipids is one mechanism believed to contribute to the increased risk of myocardial infarction associated with its use observed in a recent meta-analysis of controlled trials ^[20]. Our findings suggest that lipid parameters and the management of dyslipidemia should be taken into consideration when selecting a PPAR-γ agonist to treat insulin resistance or diabetes in HIV.

Sponsorship: This project was funded from the following sources: RO1 DK 59535, K23 DK 20844, MO1 RR 300088, MO1 RR 02635 and a research grant from GlaxoSmithKline.