Abstract
The predictive values of HIV-1 protease genotype and virtual phenotype (vPhenotype) results on the HIV-1 RNA response to lopinavir/ritonavir (LPV/r)-containing salvage regimens were assessed. Data were evaluated from patients with antiretroviral (ARV) resistance testing prior to initiating an LPV/r-containing salvage ARV regimen, from two independent cohorts from Toronto, Ontario and British Columbia, Canada. Multivariate logistic regression models controlling for previous non-nucleoside reverse transcriptase inhibitor use, baseline viral load and AIDS-defining illness were used to assess the impact of different protease genotypic mutations (individual and in combination) and lopinavir vPhenotyping on virological suppression to <50 copies/ml by 12 months. We confirmed that the 11-mutation ‘lopinavir mutation score’ (LMS) was significantly inversely associated with the probability of virological suppression within 12 months [odds ratio (OR)=0.91; P=0.02]. The only single specific protease mutation found to predict virological response in multivariate analyses was 46I (OR=0.55; P=0.02). The most predictive three-mutation combination was 10F/I/R/V, 46I, 82A/F/T (OR=0.18; P=0.0004). We confirmed that a 10-fold increase of lopinavir IC50 is an appropriate clinical cut–off for lopinavir vPhenotype. In univariate analyses, a cut-off of the LMS as low as 3 was significantly associated with a lack of virological suppression (P=0.04). This finding, which is in contrast to those of other studies, may be due to the high degree of ARV experience of our population and lack of active agents in the salvage regimen. Selecting the 11 specific mutations to make the LMS is potentially arbitrary; we determined that when different combinations of 11 protease mutations were chosen randomly from a set of 30, similar associations with virological response were found, probably due to the co-linearity of these mutations.