713    

Background:  Several cross sectional studies have identified determinants of HAART-associated hyperlipidemia. This prospective study in a diverse ART-naïve population provided an opportunity to evaluate potential predictors of hyperlipidemia at 48 weeks, in subjects treated with trizivir (TZV) vs combivir (COM)/nelfinavir (NFV) vs stavudine (d4T)/lamivudine (3TC)/NFV.

Methods:  From the United States, Puerto Rico, Panama, Dominican Republic, and Guatemala, we randomized 254 non-diabetic, ART-naïve subjects with CD4⁺ cell counts (CD4s) >50 cells/mm³ into 3 groups (1:1:1) to receive therapy. Evaluations consisted of fasting lipid profiles, HIV-1 RNA (vRNA), and CD4 counts. Logistic regression analysis was used to assess the relationship between changes in LDL, total cholesterol and triglycerides, and potential co-variates. All significant factors from the univariate models were assessed together in the final model. Statistical significance was defined as p £0.05.

Results:  In this population (50% female, 40% black, 37% hispanic), the groups were comparable at baseline, with overall mean vRNA of 4.43 log₁₀ copies/mL and CD4 of 355 cells/mm³. Univariate models showed that black subjects and those treated with d4T/3TC/NFV were more likely to have an increase in LDL (p = 0.001 and 0.002); higher baseline values for LDL and total cholesterol and Hispanic ethnicity were significantly associated with a decrease (or no change) in LDL at week 48 (p = 0.008, 0.023, and 0.009, respectively). Using stepwise selection, black race, d4T/3TC/NFV treatment, and baseline LDL values remained significant. Age, gender, CD4 count, weight, body mass index, CDC classification, and other baseline laboratory parameters (HDL cholesterol, triglycerides, lactate, or LFT) were not predictive. Blacks and those treated with d4T/3TC/NFV were also more likely to have an increase in total cholesterol in univariate models (p = 0.009 and 0.02, respectively), while baseline CDC class A was associated with increased odds of a decrease or no change (p = 0.027). These factors remained significant in the final model. The only predictor of triglyceride change was baseline triglyceride level. For each unit increase in baseline triglycerides the odds of having a decrease at week 48 were 1.004 (p = 0.037).

Conclusions:  Lipid changes observed after different HAART regimens appear influenced by HIV disease effect and to vary by race and other base;ome variables. Black race and d4T/3TC/NFV treatment were significant predictors of LDL and total cholesterol increase at 48 weeks.

Keywords: Hyperlipidemia; Complications; HAART