Minority drug resistant variants of HIV-1 and response to early combination therapy

Keys, Jessica

Download PDF

Request Version for Screen Reader

Last Modified

March 22, 2019

Creator

Keys, Jessica
- Affiliation: Gillings School of Global Public Health, Department of Epidemiology

Abstract

Initial HIV-1 therapy selection is informed by sequencing of a bulk PCR product to screen for antiretroviral resistance mutations. However, this method does not reliably sample drug resistant variants that occur in <20% of the viral population, and these may re-emerge and impair treatment response once therapy is administered. Alternatively, ultra deep sequencing can detect minority drug resistant variants, but it is difficult to distinguish very low abundance mutations from error. To address deep sequencing error, two regions of the HIV-1 genome spanning reverse transcriptase (RT) codons 34-245 were tagged with a random 8-nucleotide sequence (Primer ID) prior to PCR and sequencing. Primer ID allowed us to use resampled raw sequences sharing the same Primer ID to construct consensus sequences, each representing an original viral template within that sample. We first established a residual error rate for Primer ID using known sequences for both the Roche 454 and Illumina MiSeq deep sequencing platforms. Primer ID reduced 454 and MiSeq errors from 71 to 2.6 and from 24 to 1.2 errors/10,000 nucleotides, respectively. Applying Primer ID corrected 454 deep sequencing to 184 therapy-naïve patients from North Carolina that went on to receive RT inhibitor based combination therapy, we found that 14% of had at least one RT inhibitor mutation, compared to 2.7% using standard bulk sequence analysis. Nearly 10% of 184 patients received regimens that contained fewer than 3 active antiretrovirals, according to the Stanford resistance algorithm. While patients on suboptimal therapy failed faster than patients on fully-active regimens, the effect was driven by resistance detected by standard methods rather than previously undetected minority variants. Overall, the use of Primer ID revealed limited template utilization, limiting the depth of deep sequencing sampling. Primer ID addresses important limitations of deep sequencing and produces less biased estimates of low level resistance mutations in the viral population, which may allow us to more accurately define a threshold at which minority drug resistant variants of HIV-1 begin to compromise treatment response.

Date of publication