Validation of replicative phenotyping to detect and assign HIV-1 resistance in clinical specimens

The ability of HIV-1 to develop resistance to antiretroviral drugs has limited the overall efficacy of combination therapy to suppress viral replication. Consequently, drug resistance testing has become an integral tool for many HIV specialists, particularly when managing patients who experience failure of antiretroviral therapy. Two principal methodologies are in current use for assessing resistance to antiviral agents: Genotyping and Phenotyping. The relative utility of the two types of resistance tests is a topic of a heated debate. The phenotypic technique, derived directly from patient leukocytes, is more labour-intensive and time-consuming. Nevertheless, it raised much interest with recent developments such as replicative phenotyping, which involve the constitution of recombinant virus genomes containing sequences from patient-derived viral protease and/or reverse transcriptase genes.
One aim of this study was to validate the PhenoTecT system, a “rPhenotyping”, as diagnostic tool applicable for detecting and assigning HIV resistance in clinical specimens. This methodology can also be performed where known limitations of standard Genotyping pose obstacles, e.g. its low sensitivity for detecting viral minority variants or its limits in dissecting mixed virus populations.
The questions to elucidate in this study were the following:
1. Is rPhenotyping as powerful as Genotyping to reveal HIV-1 resistance in routine use? (Chapter I)
2. Can rPhenotyping overcome the limitations of standard Genotyping?
- How do viral mixtures influence a resistance test? (Chapter II)
- Are clinically critical HIV minorities revealed by replicative Phenotyping? (Chapter III)
Statistical analysis of more than 1,000 patient data sets from the InPheno database (PhenoBase®) identified relevant discordances between genotyping (Stanford DB) and phenotyping (PhenoTecT) in the interpretation of resistance profiles. Discordance was defined as determination of reduced susceptibility by one but sensitivity by the other test (Geno-R/Pheno-S), or vice versa (Geno-S/Pheno-R). The set of data used in this study was all diagnostic analyses from 2002 to 2007The ability of HIV-1 to develop resistance to antiretroviral drugs has limited the overall efficacy of combination therapy to suppress viral replication. Consequently, drug resistance testing has become an integral tool for many HIV specialists, particularly when managing patients who experience failure of antiretroviral therapy. Two principal methodologies are in current use for assessing resistance to antiviral agents: Genotyping and Phenotyping. The relative utility of the two types of resistance tests is a topic of a heated debate. The phenotypic technique, derived directly from patient leukocytes, is more labour-intensive and time-consuming. Nevertheless, it raised much interest with recent developments such as replicative phenotyping, which involve the constitution of recombinant virus genomes containing sequences from patient-derived viral protease and/or reverse transcriptase genes.
One aim of this study was to validate the PhenoTecT system, a “rPhenotyping”, as diagnostic tool applicable for detecting and assigning HIV resistance in clinical specimens. This methodology can also be performed where known limitations of standard Genotyping pose obstacles, e.g. its low sensitivity for detecting viral minority variants or its limits in dissecting mixed virus populations.
The questions to elucidate in this study were the following:
1. Is rPhenotyping as powerful as Genotyping to reveal HIV-1 resistance in routine use? (Chapter I)
2. Can rPhenotyping overcome the limitations of standard Genotyping?
- How do viral mixtures influence a resistance test? (Chapter II)
- Are clinically critical HIV minorities revealed by replicative Phenotyping? (Chapter III)
Statistical analysis of more than 1,000 patient data sets from the InPheno database (PhenoBase®) identified relevant discordances b. For a fair comparison each patient’s sample was re-interpreted with the same version of algorithm. As genotypic algorithms are self-optimising systems that improve their performance over time we could observe that some of the discordances previously found between genotyping and phenotyping did not hold up in later versions. Moreover, three independent genotypic prediction systems were used in parallel to assess potential differences for the same patient samples. Significant discordances were observed between interpretations inferred by these different genotyping algorithms. This suggests the urgent requirement to define common rules and possibly even new criteria to be considered and will increase coherence of results and therapy advices within national and international studies.
Nevertheless some intrinsic shortcoming remain that necessitate the use of alternative dissecting methods such as rPhenotyping: Genotyping principally fails to assign new mutations or novel resistance patterns associated with new drugs, and systems are limited in the ability to follow viral speciation during extended treatment periods as this leads to the formation of viral minorities. The evolution of a growing complexity of viral genomes in any given patient requires dissection and assignment of mutations to individual viral genomes. In our study we hypothesised that these limitations could result in discordances between genotypic and phenotypic resistance profiling. To particularly focus on Geno-R/Pheno-S discordances we generated by point mutagenesis infectious, otherwise isogenic HIV-1 clones carrying selected combinations of one to four mutations known to decrease susceptibility to the drug AZT, and we investigated their phenotypic resistance profiles. As expected: analyses of virus mixtures each expressing at least three of the selected mutations was categorised as resistant to AZT by both testing whereas those with a mixture of single mutated virus was interpreted as AZT resistant only by genotypic algorithm whereas PhenoTecT was able to dissect viral mixtures. For Geno-S/Pheno-R samples, we speculated that rPhenotyping is able to detect resistant minority species not scored by genotyping. We prepared defined mixes of wild-type and M184V mutant which were analysed by rPhenotyping or standard genotyping.
Allele-specific and quantitative polymerase chain reaction (PCR) set detection and quantification limits for minor virus populations in vitro and in authentic clinical samples showing geno-/pheno discrepant lamivudine resistance. Data presented in this study demonstrate that the principle of a functional assessment by rPhenotyping can strongly detect: (i) minor HIV populations with a sensitivity level of less than 1%, (ii) viral mixtures by analysing each single virus separately and (iii) viruses presenting mutations with synergistic or antagonistic interactions. This renders it, for these applications, superior to Genotyping. Nevertheless further studies will have to show how this gain in sensitivity will translate into clinical utility and improved disease management (ongoing project with the Swiss HIV cohort and the group of Prof. M.Battegay,USB).
In the future patients should benefit from those analytical improvements by assisting in improved identification of new drug-options. In addition, these observations will form the basis for applications in other viral diseases, notably HBV and HCV, where similar issues are likely to arise with the currently ongoing introduction of potent antiviral drugs.