A comparison of phenotypic, genotypic and clinical/treatment history predictors of virological response to saquinavir/ritonavir salvage therapy in a clinic-based cohort
Originally Published on September 2, 1999
AR Zolopa, K Hertogs, R Shafer, P Dehertogh, V De Vroey, B Efron, S Bloor and B Larder
ABSTRACT:
Objective: We have previously reported in this cohort that baseline genotype was an independent predictor of response to saquinavir/ritonavir salvage therapy. In this study, we compare the predictive capacity of HIV-1 phenotype and genotype to standard clinical evaluation in patients who received saquinavir/ritonavir combination therapy after failing a protease inhibitor-containing antiretroviral therapy.
Methods: A clinical cohort of 54 patients was identified retrospectively from the HIV clinic at Stanford University. All subjects had a HIV-1 genotype by ABI sequencing methods from a baseline line specimen collected within 1 month prior to saquinavir/ritonavir t. Phenotype analysis was performed (blinded) and was categorized as sensitive (IC50 <fourfold difference from wild-type). Intermediate(IC50 >4<10-fold) and resistant (IC50 <10-fold increase). Response to saquinavir/ritonavir was categorized as complete (viral load <100copies/mL), partial (>0.5 log10 reduction) and non-response (<0.5 log10 drop) by 12 weeks. Regression analysis was used to determine the explained variance (R2), positive (+PV) and negative predictive values (_PV) for multivariate models of phenotypic, genotypic and clinical variables.
Results: A phenotypic result was obtained in 51 of the 54 subjects. The best fit model of the clinical and treatment history variables had an explained variance of 41% (R2=0.41). Adding baseline phenotype significantly improved the predictive capacity of the clinical model (R2=0.58, F test=0.05). Resistance to ritonavir was the strongest phenotype variable. Ritonavir phenotype at baseline predicted complete response at week 12 with R2=0.30, +PV 67% and PV 91%. Surprisingly, saquinavir phenotype did not add significantly to a model that included ritonavir phenotype.
Conclusions: HIV-1 phenotype provides prognostic information that is independent of standard clinical evaluation. Additional multivariate models with both phenotype and genotype will be presented
