Background: This study aimed to clarify whether NGS might be useful for resistance assessment in virologically suppressed highly treatment-experienced (HTE) individuals with multidrug resistance (MDR).

Methods: Ninety-one HTE MDR individuals from the PRESTIGIO registry were analysed. HIV-1 DNA PR/RT/IN and V3 sequences were obtained through NGS on MiSeq platform. Major resistance mutations (MRM) and APOBEC editing estimation (APOBEC mutations [APO-M]; stop codons) were evaluated through HIVdb algorithm. NGS cut-offs at ≥1%, ≥5% and ≥20% were tested. Minority MRM with frequency ranging 1-5% (mV1%) and 5-20% (mV5%) and majority MRM (frequency >20%, mV20%) were compared to historical-GRT (H-GRT). Variants distribution was compared between individuals who experienced virological rebound after NGS-GRT and those who maintained virological control.

Results: At NGS-GRT, individuals had a median (IQR) cART exposure of 23 (21-25) years, had been virologically suppressed since 3 (2-5) years and had a total HIV-DNA of 2,377 (1,274-4,949) copies/106 CD4+ cells. X4 tropism was detected in 61.5% of individuals.
A total of 1,772 MRM were detected. Around half of MRM detected by NGS were already found in H-GRT; on the other hand, NGS-GRT detected a considerable number of additional mutations never observed before (Figure 1). The highest detection rate of historical MRM was obtained by setting NGS at 1% (Figure 2).
NGS set at 1% showed poor reliability, although associated with the highest detection rate of historical MRM. In fact, mV1% (N=337) were frequently detected in samples with stop codons (94.4%) or APO-M (97.4%) providing potential misleading resistance assessment.
Differently, among mV5% (N=370), a substantial proportion of cases was not affected by APOBEC editing and contributed in expanding detection of historical MRM (25.9%) or detecting new MRM (18.6%).
Regarding majority variants, mV20% (N=704) were marginally detected in samples with stop codons (2.9%) or APO-M (5.3%), and mostly contributed to detect (69.4%) historical MRM or detect new MRM (25.4%).
After NGS-GRT, 21 individuals underwent virological rebound with a median (IQR) viremia of 365 (98-7,840) copies/mL. Among them, only the median (IQR) number of mV5% detected exclusively by NGS-GRT was higher (2 [1-3]) compared to those who maintained virological control (1 [0-2], p=0.030, Figure 3). The number of mV5% newly detected by NGS in failing individuals positively correlated with plasma HIV-RNA levels detected at virological rebound (Spearman test, Rho=0.474, P=0.030).

Conclusions: In HTE MDR virologically suppressed individuals, NGS-GRT on HIV-1 DNA allows detection of around 60-70% historical MRM and detects considerable new resistance. Our results confirm that setting NGS at 5% might be a good choice to obtain reliable sequence data. At this setting, an increased number of minority species correlates with loss of virological control and with viremia levels at virological rebound.