Abstract

E-mail Abstract Author Add To Itinerary Session Search Abstracts Program

Session 77 Poster Session Resistance Testing in Drug Selection Session Time: 4:30-6:30 pm Room 4E-F

587-T. CREST-- A Randomised Comparison of 2 Resistance Test Platforms: Genotype and Virtual Phenotype C. Workman*1, G. Hales2, P. McKenna3, and for the CREST Study Group 1Ground Zero Med. Ctr., Sydney, Australia; 2NCHECR, Sydney, Australia; and 3 Virco, Mechelen, Belgium Background: Our objective was to compare genotypic and VirtualPhenotypic resistance tests on antiretroviral treatment (ART) prescribing in patients who have failed an ART regimen and to evaluate the concordance between 2 methods of HIV resistance testing. Methods: Patients were randomised 1:1, in a multi-centre study to receive either an HIV genotype (arm A) or an HIV genotype and VirtualPhenotype (arm B). Genotypes were reported using a standardised, rules based, algorithm. Investigators documented at randomisation which ART regimen they planned to prescribe next for a patient based on their best clinical judgment. This was compared to the actual ART regimen prescribed following receipt of the HIV drug-resistance report. An evaluation of the concordance/discordance between the 2 types of resistance report was performed as well as a post hoc analysis of the updated, biological cut-off VirtualPhenotype. Results: 338 patients entered the trial, of whom 330 completed baseline visits and were included in these analyses. Baseline factors were balanced between arm A and B: median CD4+ cell count 290 and 319 cells/mm3, median HIV RNA viral load 16,000 and 17,000 copies/mL. A similar proportion of arm A and arm B patients' planned ART regimen was changed as a consequence of the resistance test results (81% vs 86%, p=0.18). The number of drugs actually prescribed were the same between arms A and B; median 3, p=0.311, as was a scored difference between the planned and prescribed regimens (median three, p=0.376). 64% of investigators in arm A and 61% in arm B cited the resistance results as the most important factor in selecting the patients' new regimen. There was greater resistance reported by the genotypic algorithm than by either version of the VirtualPhenotype report, particularly for NRTIs. Conclusions: Both HIV genotype and VirtualPhenotype results were found to have affected ART selection more than any other factor. The standardised, rules-based, algorithm used to interpret genotypic data consistently reported a greater degree of resistance than the VirtualPhenotype. The clinical sequelae of baseline discrepancies in resistance reporting are not known.

©2002 9th Conference on Retroviruses and Opportunistic Infections