"Possible" resistance: Comparison of "possible" HIV resistance results derived from a genotyping report versus the IC50 virtual phenotype

Background: HIV drug-resistance testing employs genotyping, phenotyping, or a hybrid method known as virtual phenotyping. Phenotyping is based on the rate that a drug inhibits viral replication. In genotyping, the virus is sequenced and mutations are analyzed permitting the risk of drug resistance to be designated as "high," "possible," or "none." Virtual phenotyping links sequences of phenotyped isolates to genotype-phenotype pairs in a relational database, thus providing quantitative prediction of drug resistance. This 3-month study correlates "possible" genotypic and virtual phenotypic results. Methods: Genotyping and virtual phenotyping utilized the Viroseq and VitualPhenotype systems, respectively. Results: Of the 267 "possible" resistant genotypic results, 47 were resistant by virtual phenotype (17.6%). Protease inhibitors (PI) had the highest percentage of correlated resistant results (22.6%). Conclusion: "Possible" resistant genotypic results obtained for most of the classes of HIV drugs were eventually classified as "sensitive" by virtual phenotyping. A "possible" resistant genotypic result for PI was more closely correlated to resistance by virtual phenotyping.