Predictive markers of HIV-related weight loss and determination of differences between populations with weight loss stratified by opportunistic processes

Objective: To describe factors predictive of >10% weight loss among enrolle d participants in clinical trials of the AIDS Clinical Trial Group (ACTG). Design: A retrospective analysis of data from selected ACTG antiretroviral clinical trials completed prior to 1996 (ACTG 116, 117, 155, 175, and 241), which did not include protease inhibitors. Methods: Data were analyzed in Cox proportional hazards models to determine significant predictors for >10% weight loss while on study. Weight loss oc curring within 30 days before or after an opportunistic infection (OI) was defined as "OI-associated." Both univariate and multivariate models were co nsidered; gender-specific models were also analyzed to provide insight into potential gender differences in predictors of weight loss. Results: We found that substantial weight loss is a frequent occurrence amo ng those enrolled in clinical trials of antiretroviral agents; approximatel y 15% of subjects in the studies considered experienced >10% weight loss. C D4 cell count and HIV-1 RNA at week 8, Karnofsky score, and injection drug use status were significant multivariate predictive markers for weight loss associated with an OI; baseline weight, hemoglobin, triglycerides, and gen der were additional predictors for weight loss not associated with an OI. Conclusions: This is the first study to characterize the association betwee n baseline viral load and future weight loss. Baseline and week 8 immunolog ic parameters as well as measures of baseline symptomatology were significa nt predictors of weight loss associated and not associated with an OI.