Background: Resistance testing is useful in the management of virological failure, although the best method to be used in clinical practice has not been determined. Virtual Phenotype (VirtualP) may have some advantages over phenotype, especially cost.

Methods: This was a multicenter, prospective, randomized, double-blind study of 300 patients on virological failure (HIV-RNA>1000 copies/mL). Patients who had failed previous HAART regimens were randomized 1:1 to P (Antivirogram, Virco) or VirtualP (VircogenII, Virco), and stratified according to previous ART exposure (1-2 vs 3 drug classes) at a central level. An intention to treat and an as treated analysis were performed. Outcomes include VL changes from baseline and percentage <50 copies/mL at 24 weeks.

Results: Analysis of 260 patients who have completed 24 weeks (129 and 131 in the P and VirtualP arms, respectively). Length of exposure to ART was 65, 37, and 14 months for NRTI, PI, and nNRTI, respectively. Most patients (60%) had been exposed to all 3 drug classes. At baseline, mean CD4 count and VL  were 341 and 330 cells/μL and 3.9 and 4.01 log copies/mL for the P and Virtual P groups, respectively. The mean number of active drugs prescribed was 2.96 in the P arm vs 2.85 in the VirtualP arm. At week 24, median decrease in viral load was 0.89 and 1.22 log copies/mL in the P and VirtualP groups, respectively (p= 0.028). In ITT analysis, 46.5% and 56.5 of patients in the P and VirtualP, respectively, had VL<400 copies/mL % (OR 1.75, 95% CI 2.85-1.03, p=0.1). Significant differences were found in the as treated analysis (42.7% vs 56.7%) (OR 1.49, 95% CI 2.43-0.91, p=0.038). In a logistic regression analysis, after adjusting for all potential confounding variables, the probability of virologic failure increased in the P group  (OR 4.02, 95% CI 1.75-9.22, p=0.001).

Conclusions: VirtualP performs as well as or better than P for guiding ART in subjects who have failed one or more antiretroviral regimens.