Objective: Drug resistant HIV strains can be identified by genotypic analysis. The prediction of resistance from the HIV pol sequence, however, requires expert interpretation which is currently provided by various interpretation programs. In the present study, the comparability of two of these programs was investigated.
Methods: One hundred and six HIV sequences obtained from patient samples were subjected to interpretation by the Stanford HIV-SEQ program (http://hivdb.stanford.edu) and, in parallel, by virtual phenotype analysis (VircoNET).
Results: The overall concordance between the two assays was high with respect to nonnucleoside reverse transcriptase inhibitors (NNRTIs) and protease inhibitors. Highly discrepant results were obtained from 22 samples (1.5% of all comparisons), and these discrepancies were significantly associated with the interpretation of nucleoside reverse transcriptase inhibitors (NRTIs) (P < 0.01), especially regarding resistance to zalcitabine (ddC), didanosine (ddI) and abacavir (ABC). Nearly all of these discrepant samples showed a sensitive virtual phenotype. Mutations at codons 69 and 74 were associated with a discrepant interpretation for ddC.
Conclusions: The prediction of resistance to certain NRTIs from a given HIV sequence may be contradictory and requires further investigation.