Correlation between genotypic predictions based on V3 sequences and phenotypic determination of HIV-1 tropism

OBJECTIVE: Replacing phenotypic assays with simple genotypic predictions of HIV-1 coreceptor usage would make the clinical use of CCR5 antagonists easier. DESIGN: Paired genotypic and phenotypic determination of HIV-1 coreceptor usage was performed to assess several genotypic approaches for detecting CXCR4-using and CCR5-using viruses in a clinical setting. METHODS: HIV-1 coreceptor usage was prospectively assessed using plasma samples from 103 patients who were candidates for treatment with a CCR5 antagonist. Direct sequencing of the V3 region and a sensitive recombinant virus phenotypic entry assay were performed in parallel for each patient from the same bulk env PCR product. RESULTS: The 103 patients had a median CD4+ T lymphocyte count of 268 x 10(6)cells/l and nadirs of 98 x 10(6)cells/l. Paired genotypic and phenotypic data were obtained for 98 of the 103 patients. For detecting CXCR4-using viruses, the genotypic rule based on amino-acid residues at positions 11/25 and the overall net charge of V3 was 77% sensitive and 96% specific. The Geno2pheno bioinformatic tool was 88% sensitive and 87% specific. The WebPSSM tool prediction with the SI/NSI matrix was 77% sensitive and 94% specific. The global concordance between genotypic and phenotypic data was 91% with the rule combining the amino-acid residues at positions 11/25 and V3 net charge. CONCLUSION: Genotypic predictions performed well in paired genotypic and phenotypic assessment of HIV-1 coreceptor usage. Multicenter studies analyzing the correlations between the genotypic determination of HIV-1 tropism and clinical response to CCR5 antagonists are needed to validate this approach in clinical practice.