Abstract

The emergence of drug resistance remains a major problem during antiretroviral treatment of patients infected with human immunodeficiency virus type 1 (HIV-1). As phenotypic drug resistance is laborious and expensive to determine, and because numerous specific mutations are known to be correlated with different resistance patterns, genotyping of the reverse transcriptase and protease genes of HIV is fast becoming an integral part of HIV management in industrialized countries. A number of software-based interpretation systems have been developed for the interpretation of the resulting complex nucleotide sequences. These programs either employ rule-based algorithms or are based on a genotype-phenotype database. This paper reviews recent publications that compare different such systems, trying to identify the degree of discordance between different systems and the reasons underlying such discrepant interpretations. The highest discordance rate was observed for nucleoside reverse transcriptase inhibitors (NRTIs) followed by protease inhibitors (PIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs). For the NRTIs, it is the role of nucleoside analogue associated mutations, for the PIs and for the NNRTIs, that of secondary mutations that causes most discrepancies. As the complexity of the mutation pattern is likely to increase further with new drugs becoming available, rule-based genotype interpretation algorithms need to be updated frequently. Whilst not recommending one particular system, the authors believe that the correlation of genotypic with clinical data is probably the best way to develop an optimal algorithm.

Keywords: hiv-1 genotyping, interpretation system, drug resistance associated mutations, rule based algorithm, virtual phenotype