Performance of Celera RUO integrase resistance assay across multiple HIV-1 subtypes.

BACKGROUND: HIV-1 sequence variation is a major obstacle to developing molecular based assays for multiple subtypes. This study sought to independently assess performance characteristics of the ViroSeq™ HIV-1 Integrase RUO Genotyping Kit (Celera, US) for samples of multiple different HIV-1 subtypes.
METHODS: 264 samples were tested in the validation, 106 from integrase inhibitor naïve patients' sent for routine HIV-1 drug resistance testing after failing a 1st- or 2nd-line regimen, and 158 samples from an external virology quality assurance program (VQA). For the latter, 53 unique VQA samples were tested in two to five different laboratories to assess assay reproducibility. For all assays, viral RNA was extracted using the ViroSeq extraction module, reverse transcribed, and amplified in a one-step reaction. Four sequencing primers were used to span codons 1-288 of integrase. The Rega subtyping tool was used for subtype assignment. Integrase polymorphisms and mutations were determined as differences from the HXB2 sequence and by the Stanford database, respectively. Sequences obtained from the different laboratories were aligned and sequence homology determined.
RESULTS: HIV-1 RNA in the 264 samples ranged from 3.15 to 6.74logcopies/ml. Successful amplification was obtained for 97% of samples (n=256). The 8 samples that failed to amplify were subtype D (n=3), subtype C (n=1), CRF01_AE (n=1), subtype A1 (n=2), and an unassigned subtype (n=1). Of the 256 that successfully amplified samples, 203 (79%) were successfully sequenced with bidirectional coverage. Of the 53 unsuccessful samples, 13 (5%) failed sequencing and 40 (16%) did not have full bidirectional sequence, as a result of failure of sequencing primers: Primer A (n=1); Primer B (n=18); Primer C (n=1); Primer D (n=7) or short sequences (n=16). For the 135 VQA samples (30 unique samples) that were assayed by different laboratories, homology of the sequences obtained ranged from 92.1% to 100%. However, Laboratory 2 detected more mixtures (74%) compared to the other four laboratories, whereas Laboratory 1 detected the least number of mixtures (35%), likely due to differences between the labs in the methods of sequence analysis. Mutations associated with integrase resistance were observed in seven of the 106 (7%) clinical samples [one sample: Q148K; E138K; G140A; two samples: T97A and four samples: L74I]. Of the four samples with L74I, 3 were subtype G.
CONCLUSION: Of the total 264 samples tested, 243 (92%) of samples were able to be amplified and sequenced to generate an integrase genotype. Sequencing results were similar between the testing laboratories with the exception of mixture detection. Mutations associated with integrase inhibitor resistance were observed in only 7% of integrase inhibitor naive samples, and some of these mutations are likely to be due to subtype-specific polymorphisms rather than selection by an integrase inhibitor.