Next-Generation Sequencing Reveals a High Frequency of HIV-1 Minority Variants and an Expanded Drug Resistance Profile among Individuals on First-Line ART

Abstract

We assessed the performance and clinical relevance of Illumina MiSeq next-generation sequencing (NGS) for HIV-1 genotyping compared with Sanger sequencing (SS). We analyzed 167 participants, 45 with virologic failure (VL ≥ 1000 copies/mL), i.e., cases, and 122 time-matched participants with virologic suppression (VL < 1000 copies/mL), i.e., controls, 12 months post-ART initiation. Major surveillance drug resistance mutations (SDRMs) detected by SS were all detectable by NGS. Among cases at 12 months, SS identified SDRMs in 32/45 (71.1%) while NGS identified SDRMs among 35/45 (77.8%), increasing the number of cases with SDRMs by 3/45 (6.7%). Participants identified with, and proportions of major SDRMs increased when NGS was used. NGS vs. SS at endpoint revealed for NNRTIs: 36/45 vs. 33/45; Y181C: 26/45 vs. 24/45; K103N: 9/45 vs. 6/45 participants with SDRMs, respectively. At baseline, NGS revealed major SDRMs in 9/45 (20%) cases without SDRMs by SS. Participant MBL/043, among the nine, the following major SDRMs existed: L90M to PIs, K65R and M184V to NRTIs, and Y181C and K103N to NNRTIs. The SDRMs among the nine increased SDRMs to NRTIs, NNRTIs, and PIs. Only 43/122 (25.7%) of participants had pre-treatment minority SDRMs. Also, 24.4% of the cases vs. 26.2 of controls had minority SDRMs (p = 0.802); minority SDRMs were not associated with virologic failure. NGS agreed with SS in HIV-1 genotyping but detected additional major SDRMs and identified more participants harboring major SDRMs, expanding the HIV DRM profile of this cohort. NGS could improve HIV genotyping to guide treatment decisions for enhancing ART efficacy, a cardinal pre-requisite in the pursuit of the UNAIDS 95-95-95 targets.

Keywords: HIV-1 antiretroviral therapy; HIV-1 drug resistance; minority variants; next-generation sequencing.

Figure 1

Study schema summarizing the key aspects of the study design and the major objectives.

Figure 2

Summary of results. Samples from both groups (controls and the cases) were combined for HyDRA analyses, and the results in the last row are subdivided into baseline NGS results on the left (comprising baseline samples of controls and cases) and NGS results for only cases at 12 months. Controls were neither sequenced on the SS nor on the NGS platform at 12 months. Cases were those participants with virologic failure (VL ≥ 1000 copies/mL) at 12 months while controls were participants with virologic suppression (VL < 1000 copies/mL) at both 12 months and at baseline.

Figure 3

Comparison of major SDRM patterns based on SS and NGS genotyping for cases at 12 months: all major SDRMs detectable by SS were also detectable by NGS with the NGS platform detecting additional SDRMs reflected in the overlaps between the bars where they exist.

Figure 4

Major SDRMs detected by both SS and NGS (blue bars) and detected by only NGS (orange bars). * All major SDRMs detected by SS were detectable by NGS.

Figure 5

Major HIV-1 SDRMs detected among controls at baseline previously undetectable by SS.

Figure 6

Minority drug resistance mutations of cases at 12 months post-ART initiation.

Figure 7

Minority drug resistance profiles of controls and cases at baseline.

Figure 8

(A): Baseline HIV-1 drug resistance mutations plotted against endpoint viral load for both controls and cases show that the higher the frequency of baseline minority mutations, the higher the viral load count at the endpoint (Kruskal–Wallis p < 0.005). (B): Baseline HIV-1 viral loads are similar across timepoints for the cases and controls (Kruskal–Wallis, p = 0.231). In (A), The test statistic is adjusted for ties. Multiple comparisons were not performed since there were only two test fields. In (B), the test statistic is adjusted for ties. Multiple comparisons were not performed since the test showed no significant difference. The * (asterisk) are outliers above the fourth quartile and the small circle in (B) is an outlier below the l quartile.

Figure 9

Baseline HIV-1 drug resistance mutations plotted against end point viral load for cases shows that the higher the frequency of minority mutations at baseline, the higher the viral load count at the endpoint (Kruskal–Wallis p = 0.011). The test statistic is adjusted for ties. Multiple regressions were not performed since there were only two test fields. The * (asterisk) represents outliers above the fourth quartile.