Background:  Methods for optimizing treatment and improving virologic suppression in HIV-1-infected patients with multi-drug-resistant virus are greatly needed.

Methods:  CPCRA 064 studied HIV⁺ ART-experienced patients with virologic failure and multi-drug-resistant virus, who had switched to a new (optimized) salvage regimen (enrollment, August 2000 to June 2002). Here we characterize the baseline susceptibility to the new regimen chosen at entry in the control (no structured treatment interruption) arm. Using original genotypic data (CPCRA interpretive algorithm) and phenotypic data (PhenoSense) obtained, retrospectively, a value was assigned for each drug in the new regimen (1 if sensitive, 0.5 if intermediate, and 0 if resistant). An incremental phenotypic susceptibility score (iPSS) was defined as the sum of these values according to phenotype; iGSS was defined similarly according to genotype. A dichotomous phenotypic susceptibility score (dPSS) was defined by summing only those drugs to which the virus was fully sensitive by phenotype. To estimate the changes in HIV RNA at months 2 and 4, univariate and multivariate linear regression models were created with dPSS, iPSS, and iGSS as predictors. All models were adjusted for baseline HIV RNA.

Results:  We included 110 patients with the following characteristics:  at baseline, mean CD4 = 182 cells/mm³, nadir CD4 = 79, log₁₀ HIV RNA = 5.0, number of previous ART used = 10.5 and number of ART mutations = 9. Mean number of drugs in the new regimen = 3.9. Mean dPSS, iPSS, and iGSS for the new regimen were 1.28, 1.85, and 1.78, respectively. At months 2 and 4, mean changes in log₁₀ HIV RNA were a decrease of 0.95 and 0.90, respectively. For change in HIV RNA at both time points, dPSS, iPSS, and iGSS were all significant predictors in univariate models. The model with iPSS had the highest R² value. The univariate model with iPSS as a predictor was not significantly improved by adding either iGSS or dPSS to the model. However, the univariate model with iGSS as a predictor was significantly improved by adding iPSS to the model.

Conclusions:  The iPSS of the new salvage regimen was the strongest single predictor of virologic response among the resistance measures analyzed. Adding iPSS to the iGSS model improved the predictive response compared with iGSS alone. These data suggest that for patients with multi-drug-resistant HIV-1 treatment failure and limited effective treatment options, the use of phenotypic susceptibility scoring may provide added benefit in the selection of new drugs, which may result in improved treatment responses.