572 Pre-treatment Factors that Predict Responses to Potent Antiretroviral Therapy: Findings from AACTG A5001 C.A. Benson*1, A.C. Collier2, R. Bosch3, K. Bennett3, R. Zackin3, A5001 Protocol Team4 1Univ of Colorado Hlth Sci Ctr, Denver; 2Univ of Washington, Seattle; 3Statistical and Data Analysis Ctr, Harvard Sch of Publ Hlth, Boston, MA; and 4Natl Inst of Hlth /NIAID/DAIDS Adult ACTG, Rockville, MD
Background: Defining immunologic, virologic, and host factors that influence response to Antiretroviral Therapy (ART), and their impact on long-term outcomes is key to individualizing therapy for HIV+ pts.
Methods: We analyzed baseline (BL) factors associated with ART response in HIV+, ART-naïve pts in AACTG A5001 (ALLRT). ALLRT is a prospectively planned series of meta- and cross-protocol analyses of pts enrolled in AACTG trials from 1998-2002 that provide randomized ART. ART-naïve and -experienced pts have data collected at entry and Q16 wks to evaluate long-term clinical, virologic, and immunologic outcomes; 2,174 pts are enrolled (mean follow-up 2.6 pt-yrs); 785 ART-naïve pts (47% Caucasian, 17% women, median age 36, BL CD4+ 222 cells/µL, BL viral load [VL] 143,115 c/ml) with available VL (n = 749) and CD4+ counts (N=711) at wk 96 were included in this analysis. BL factors assessed were CD4+, VL, sex, race, age, Hgb, weight, BMI, nadir CD4+, TLC, CD4%, CD8+, CD8%, memory and naïve CD4+. CD4+ and VL were considered as continuous and categorical variables. Kaplan-Meier curves and Cox proportional hazards regression models were used to assess impact of BL factors on time to VL < 50. Probability of VL < 50 and CD4+ change were examined in regression models.
Results: Estimated % of pts achieving VL <50 by wk 96 was 96 [95% CI (95, 98)]. Longer time to VL <50 was associated with higher BL VL. After adjusting for BL VL, older age was significantly associated with shorter time to VL <50 (P=0.025). BL CD4+, sex and race were not additionally predictive. After adding each BL variable to a model with BL VL and age, higher Hgb was associated with shorter time to VL <50 (P=0.038). After adjusting for BL VL, sex, race, and age were each significantly associated with probability of VL <50 at wk 96. After controlling for BL VL and age, sex, race, and Hgb each remained significantly associated with VL <50. The median rise in CD4+ from BL to wk 96 was 237 cells/µL. Lower BL VL, older age, and male sex were each associated with smaller rises in CD4+. In regression models with BL VL, BL CD4+, age, sex, BMI, nadir CD4+, memory CD4+ and CD4%, and naïve CD4% were each significantly associated with change in CD4+; each remained significant in models with both BL VL and age.
Conclusions: BL VL and age were the most important predictors of time to and probability of VL < 50 at wk 96. In stepwise models, greater CD4+ increases were associated with higher BL VL, lower BL CD4+, and higher naïve CD4+%.
