Optimisation of a multi-subtype multiplex PCR to screen for key drug resistance mutations in HIV

Abstract

BACKGROUND: Sub-Saharan Africa hosts over two thirds of the global HIVinfected population. South Africa (SA), with the highest national incidence of HIV, also has the largest antiretroviral therapy programme globally. In an effort to reach the UNAIDS 90-90-90 targets, SA implemented a universal test-and-treat approach in 2016 and has committed to increasing the number of HIV-infected individuals on ART over the next few years. The first line regimen in SA includes an NRTI backbone of tenofovir and emtricitabine, combined with efavirenz as the NNRTI. Treatment failure as a result of the development of drug resistance mutations is the greatest risk to successful ART. The most prevalent drug resistance mutations in SA, K103N, M184V and K65R, confer broad high level resistance to NRTI and NNRTIs and may affect the efficacy of second-line regimens. HIV drug resistance testing is not recommended at treatment initiation or at first-line regimen failure, due to the high cost and limited laboratory capacity. The HIV Drug Resistance RNV (derived from K65R, K103N, and M184V) Multiplex Assay was designed by the CDC to simultaneously detect the K65R, K103N and M184V mutations in the presence of virological failure. The aim of this study is to optimise the assay in a South African setting, by comparing assay results to the current gold standard, namely nested PCR and sequencing using Sanger sequencing technologies. METHODS: A total of 225 whole plasma specimens collected from patients infected with HIV-1 were selected for study, and include specimens with the K65R, K103N and M184V and wild-type genotypes as well as specimens with both low (<600 copies per ml) and high (≥600 copies/ml) viral load. Nucleic acids were extracted using the MagNA pure LC instrument and template was generated using PCR amplification of the entire protease gene and codons 1-250 of the reverse transcriptase gene. Detection of HIV variants harbouring these mutations was performed by determining the change in the cycle threshold (ΔCt) of the total copy and mutation detection components of the assay. Specimens with discrepancies between conventional genotype and RNV assay were re-sequenced using next generation sequencing.

RESULTS: Of the 225 specimens selected for this study, 47 specimens had a VL < 600 copies/ml, and 178 specimens had a VL ≥ 600 copies/ml. Of 171 specimens with known genotype, 105 specimens had a WT genotype and 66 specimens had one or more mutation present (the remaining 54 specimens were not genotyped due to low VL result). For the total copy component of the assay, which differentiates specimens by VL based on a cut-off of 600 copies/ml, 180 (84%) specimens had results concordant to the previous VL result and 45 (16%) specimens had discordant results. The sensitivity of this component was 91% and specificity 60%. For the mutation detection component of the assay, the 161 specimens that had a high VL using both the RNV assay and standard VL methodologies were screened. Of these 161 specimens, 130 (81%) were concordant with the known genotype, 28 (17%) were discordant and 3 (2%) had no amplification. The sensitivity of this component of the assay was 94%, and specificity 74%. Next generation sequencing detected low prevalence mutations in five specimens that were not present in the Sanger sequence, but were detected in the RNV assay. New thresholds were proposed to improve test statistics: for the total copy component, the recommended threshold was reduced from 25.00 to 21.78 cycles; this increased sensitivity to 95% and specificity to 85%. For the RNV component, a recommended change in cycle threshold value of 10.94 cycles decreased the number of discordant specimens to 25 (15%). This increased the sensitivity of the assay to 84%, and specificity to 86%. CONCLUSIONS: The RNV multiplex assay was successful in amplifying the three key drug resistance mutations in South African Subtype C specimens. The success of the assay thus makes it a viable screening tool to detect the K65R, K103N and M184V drug resistance mutations at first-line therapy failure. However, our recommended cut-offs should be considered by the developers to improve the sensitivity and specificity of the assay