Abstract
HIV-1 genotyping has become a widely accepted tool for monitoring antiretroviral therapy. However, the benefit of genotyping is limited by the need to interpret the mutation pattern in order to obtain a prediction regarding susceptibility to each antiretroviral drug. Several different interpretation systems are currently available, commercially or free of charge; some are in combination with the genotyping test system used. In this study, we compared the results obtained on patient samples, using nine different resistance interpretation systems for HIV-1 genotype, and identified mutation patterns responsible for discordances between these systems. HIV-1 genotypes from 26 patients were obtained using the ViroSeq™ HIV-1 Genotyping System (Applied Biosystems). Nine resistance interpretation systems were used: the ‘virtual phenotype’ systems VirtualPhenotype™ (Virco) and Geno2Pheno (http://cartan.gmd.de/geno2pheno.html), the rule-based resistance algorithms Antiretroviral Drug Resistance Report (Applied Biosystems), Stanford HIV-SEQ program (http://hivdb.stanford.edu/) and the ViroScorer™ system (ABL; including ANRS AC11, Detroit Medical Center, Grupo de Aconselhamento Virológico, CHL, and Rega). Discordance was defined as the same sequence being interpreted as resistant and sensitive to a substance by different systems, with intermediate scores being regarded as neutral. Results for lopinavir were not available from some interpretation systems. None of the 26 patient samples had concordant results for all drugs. The highest degree of concordance was found for the resistance scoring of lamivudine (25/26), followed by nelfinavir (23/26), indinavir, ritonavir, saquinavir (all 22/26), zidovudine, nevirapine (all 21/26), lopinavir, efavirenz (all 18/26), amprenavir, delavirdine (all 17/26), stavudine, abacavir (all 13/26), zalcitabine (7/26) and didanosine (5/26). In most cases, it was only one interpretation system that gave an interpretation different from the others. If this interpretation system was omitted, the overall discordance rate decreased by a statistically significant degree. There exists, therefore, an urgent need for standardization of interpretation algorithms for HIV-1 genotyping.