Background:  Predictors of response to salvage therapy with darunavir (DRV) may identify patients who are likely to benefit most from this drug. We evaluated the relationship between genotypic and virtual DRV inhibitory quotient (gIQ, vIQ) and the virological response to DRV-based salvage therapy.

Methods:  We included 32 patients on DRV/ritonavir 600/100 mg twice daily as part of salvage therapy. Viral load and CD4 cell count were recorded at baseline and at weeks 12 and 24. Viral genotype and virtual phenotype were obtained at baseline. DRV trough concentrations (C_trough) were measured by high performance liquid chromatography (HPLC) at week 4, 12, and 24. DRV gIQ and vIQ were calculated dividing DRV C_trough by the number of DRV-associated mutations or by the DRV fold-change in the virtual phenotype multiplied by the protein binding-corrected DRV IC₅₀ for protease inhibitor (PI) -resistant viruses. Optimized background score was calculated as the number of active drugs according to virtual phenotype. Virological response was defined as viral load <50 copies/mL or a decrease >1 log₁₀ copies/mL at week 24. The relationship between viral response and different predictors was assessed by linear or logistic regression as appropriate. Cut-off values of gIQ and vIQ to classify patients as responders or not were determined by receiver operating characteristic curves.

Results:  Median (IQR) viral load and CD4 count were 4.5 (3.7 to 4.9) log₁₀ copies/mL and 203 (74 to 290) cells/mm³ at baseline. DRV fold-change was 1.6 (0.6 to 3.8), and the number of active drugs in the optimized background score was 1.5 (0.5 to 2.0). At week 24, the median viral load decrease was 2.5 (1.4 to 3.0) log₁₀ copies/mL, 19 patients (59%) had a viral load <50 copies/mL, and the median CD4 count increase was 96 (23 to 169) cells/mm³. Virological response was not related to DRV C_trough or the number of DRV mutations. After adjusting for the optimized background score, DRV gIQ and vIQ were the only independent predictors of the change in viral load at week 12 and 24. Viral load at week 24 decreased >1 log₁₀ copies/mL in a higher proportion of patients when DRV gIQ was ³2500 (95% vs 58%; p = 0.03) or when vIQ was ³1.5 (95% vs 54%; p = 0.04). Patients with DRV vIQ ³1.5 had an 8-fold greater chance of achieving a viral load <50 copies/mL at week 24 (OR 8.6, 95%CI 1.4 to 51.6; p = 0.02). The positive and negative predictive values of DRV vIQ ³1.5 for achieving VL <50 copies/mL at week 24 were 76% and 72%, respectively.

Conclusions:  DRV IQ predict virological response to DRV-based salvage regimens better than the DRV C_trough or DRV mutations alone. A DRV vIQ cut-off value of 1.5 for achieving viral suppression at week 24 is proposed.