Prediction of HIV type 1 subtype C tropism by genotypic algorithms built from subtype B viruses

Abstract

Background: Genotypic predictions of HIV-1 tropism could simplify CCR5 antagonist usage. However, the genotypic algorithms built from subtype B viruses could be inadequate for non-B subtypes. We therefore performed paired genotypic and phenotypic determination of subtype C tropism.

Methods: We studied 52 patients recruited in Malawi and 21 patients recruited in France. We directly sequenced the V3 env region and performed a recombinant virus phenotypic entry assay in parallel.

Results: The Malawi patients had 29% of CXCR4-using subtype C viruses compared with only 5% in the patients from France. For detecting CXCR4-using subtype C viruses, the genotypic rule combining the amino acids at positions 11/25 and the net charge of V3 was 93.3% sensitive and 96.4% specific. The Geno2pheno tool was 86.7% sensitive and 89.1% specific. The WebPSSM tool with the SI/NSI matrix was 80% sensitive and 98.2% specific in its subtype B version and 93.3% sensitive and 81.8% specific in its subtype C version.

Conclusions: The genotypic determinants of coreceptor usage for HIV-1 subtype C were mainly in V3 and were globally similar to those previously reported for subtype B viruses. The main genotypic algorithms built from subtype B viruses perform well when applied to subtype C viruses.