Software for HIV-1 genotypic drug resistance testing is routinely used to generate clinical drug resistance interpretations. In this study we compare and quantify the differences found in the results obtained with distinct software.
A HIV sequencing data of 45 clinical samples belonging to a cohort of treatment-experienced patients was generated and analysed by using ViroSeq (VS) Genotyping Software v22.214.171.124. All (VS) results were compared to the FDA-registered DPM product and to the RUO ViroScore-HIV system from Advanced Biological Laboratories which include several knowledge databases ie, Stanford HIVdb v7.0.1 (SD) or the virtual-phenotypic-based interpretative system from Geno2Pheno v3.3 (G2P).
Overall, G2P was the algorithm which showed fewer interpretations classified as “Resistant” (8.9%, compared to 9.4% with SD and 9.2% with VS) and VS was the one which showed the highest percentage of “Susceptible” interpretations (86.1%, compared to 75.3% with SD and 78.3% with G2P). For 41 of the samples we were able to retrieve resistance interpretations for 19 drugs with all 3 algorithms, allowing us to compare 779 drug resistance results between algorithms. In 34.1% of the samples, VS reported different resistance interpretations for at least one drug when compared to SD, all of them involving a 1-level lower resistance value [from Resistant (R) to Intermediate (I) or from I to Susceptible (S)]. When considering only the interpretations where SD was in agreement with G2P (714), VS reported 1-level lower resistance values for at least one drug in 12.2% of the samples. Of the 26 different results obtained by VS when compared with SD, Etravirine, Rilpivirine and Ritonavir-boosted Saquinavir jointly account for 53.8% of the cases (19.2%, 15.4% and 11.5%, respectively).
Laboratories performing DR testing should be aware of alternative interpretive systems which could be used to supplement their existing DR reports.