Next-generation sequencing (NGS) reveals low-abundance HIV-1 drug resistance mutations among patients experiencing virological failure at the time of therapy switching in Uganda

Introduction

The widespread emergence and transmission of HIV drug resistance (HIV-DR) may jeopardize the success of the currently recommended first-line antiretroviral therapy (ART) regimens in sub-Saharan Africa (SSA) (Phillips et al., 2017). The initial ART regimen generally consists a combination of two nucleoside reverse transcriptase inhibitors (NRTIs) usually lamivudine/abacavir (3TC/ABC) or tenofovir disoproxil fumarate/emtricitabine (TDF/FTC) plus a drug from one of the three classes: a non-nucleoside reverse transcriptase inhibitor (NNRTI); usually efavirenz (EFV) or nevirapine (NVP), an integrase strand transfer inhibitor (INSTI) or a protease inhibitor (PI) (De Clercq, 2013). Additional data has shown to support the use of Dolutegravir/lamivudine (DTG/3TC) for initial treatment of some people with HIV (Cahn et al., 2020).

Recently, there has been reported existence of HIV drug resistance to both classes of first line drugs mentioned above, with reported genotypic variations. Predominantly, the prevalence of pre-treatment NNRTI resistance has been eye-catching in African populations, with notable ranges of 8% in Cameroon, and close to 15% in Uganda (de Waal et al., 2018). In 2018, the WHO recommended use of DTG in first line ART as there was increasing prevalence of pre-treatment drug resistance to most NNRTIs, especially in sub-Saharan Africa (Siedner et al., 2020). This was supported by international data documenting the presence of low-abundance variants, in addition to the known variants, among people on antiretroviral drugs. However, the policy stated that the urgency of the transition would largely depend on the country-specific prevalence of NNRTI resistance (Phillips et al., 2018a). The policy was not based on locally available data and as such calls for validation, particularly in our Uganda population—given that such in depth analyses can avail a baseline guidance on whether a patient living with HIV has a mutated form of the virus that does respond to ART or not (Clutter et al., 2016).

Several mechanisms including phenotypic and genotypic methods have been employed to offer options for HIV drug susceptibility testing. These methods include conventional HIV-1 genotyping techniques as well as Next generation sequencing approaches, that may as well be used to elaborate predominant viral quasispecies in the population (Smyth et al., 2012). The conventional HIV-1 genotyping assays, including Sanger sequencing, have to date been employed as the gold standard for HIV-DR testing although limited to only detecting HIV-1 variants present at frequencies above approximately 20% of the viral quasispecies. Such limitations lead to the collective failure of such approaches to exclusively detect low-levels of HIV-1 variants (with frequencies lower than 20% of viral quasispecies)—even when the viral species harbor HIV-DR mutations (Lee et al., 2020). Low-abundance variants, also known as minority variants, are usually present in small proportions of the virus populations, and have been reported to have a significant association with increased virological failure for first-line nucleoside reverse transcriptase inhibitors (NRTI) and NNRTI based regimens (Mbunkah et al., 2020). This type of drug resistance has been proven to be selected under the appropriate pressure of ART onto the virions during the course of treatment. And with the advent of deep (next-generation) sequencing, unlike the conventional methods, it is very possible to detect drug resistant HIV-1 variants at low frequencies, below 20% of the viral population (Simen et al., 2009). This study therefore aimed at utilizing NGS deep sequencing and correspondent bioinformatics pipelines to detect low-abundance drug resistance HIV-1 variants amongst patients on first line regimen combinations, with reported virological failure in Uganda, at the time of therapy switching.

Methods

Results

Patients’ demographics

A total of 58 samples passed preliminary quality control, and were successfully analyzed in this study. Among the participants, 43 (74.1%) were females and 15 (25.9%) were males with an average age of 32 years. These patients had all been on first line regimens for a period of at least 6 months where 37 (63.8%) of them were on TDF-3TC-EFV. Other ART combinations were TDF-3TC-DTG (10.3%), ABC-3TC-DTG (1.7%), AZT-3TC-NVP (12.1%), AZT-3TC-EFV (1.7%), ABC-3TC-EFV (3.4%) and ABC-3TC-LPV/r (6.9%) as summarized in Figure 1.

Figure 1. Showing the different ART combinations of the patients in the current study at the time of sample collection.

Minority variants of clinical importance

This study also identified some minority variants of clinical importance that were highly prevalent in the analyzed samples but were missed by Sanger sequencing. Figure 2 shows some unusual mutations that have not been reported previously as well as the presence of drug resistance mutation. These included; K14R, K45R, L63P, I64V, I15V, L83I, V77I. Other than the drug resistant mutations, there were also several unusual mutations whose clinical significance is still unknown and they could be nominated for diagnostic targets, drug resistance targets, drug targets and so on.

Figure 2. A graph to display low-abundance variants obtained from analysis of one of the samples (Sample ID HDR 101) using the HIV Stanford database.

The histograms highlighted in grey indicate usual (previously reported) mutations and whereas the ones highlighted in red indicate unusual (not reported before) mutations. A blue dot above the histograms indicates presence of a drug resistance mutation.

Discussion

Affordable, sensitive, and scalable technologies are needed for monitoring antiretroviral treatment (ART) success with the goal of eradicating HIV-1 infection in low resource settings. For instance, in recent years, conventional HIV-1 genotyping assays, including Sanger sequencing have been the standard approaches for HIV- DR testing in sub-Saharan Africa (Manyana et al., 2021). Being the gold standard, Sanger sequencing is still able to detect HIV-DR mutations of clinical relevance as well as drug resistance, even though it is mainly limited by relatively high sequencing costs and low-throughput, henceforth evidently failing to detect the low-levels of HIV-1 variants—with reported frequencies lower than 20% of viral quasispecies—which scenario may well harbor HIV-DR mutations as reported (Lee et al., 2020). These low-abundance variants, also known as minority variants, usually present as minority members of the virus populations, and thus cannot be easily detected. With such challenges of inadequate/inaccurate detection, some studies recommend the usage of high throughput sequencing technologies (Slatko et al., 2018; Manyana et al., 2021). This study therefore aimed at utilizing NGS deep sequencing approaches, and correspondent bioinformatics pipelines, to detect low-abundance drug resistance HIV-1 variants amongst patients on first line regimen combinations, with reported virological failure in Uganda, at the time of therapy switching. The study population included patients on first-line regimen with virological failure where most participants (58.6%) were on TDF, 3TC, and EFV combination; that is, two NRTIs and one NNRTI. This regimen was common for HIV treatment in many countries as reported (Gregson et al., 2016). Sequence data was obtained from the two genes targeted in the routine monitoring of HIV drug resistance testing, namely the reverse transcriptase (RT) and protease (PR) genes. These genes provide information on three classes of drugs; NRTI, NNRTI and PI class. Overall, 782 variants were identified; 757 variants from the NGS data, 90 variants from the Sanger sequencing data whereas 65/782 (8.3%) of the variants were shared between NGS and Sanger datasets. The study further observed that Sanger Sequencing was able to detect 25 variants that were not identified by NGS, whereas NGS revealed 692 variants that were not detected by Sanger sequencing. This elaborates that NGS being a deep sequencing high-throughput detects more variants compared to Sanger sequencing as observed by previous studies (Taylor et al., 2019; Schenkel et al., 2016). The current study further reports 563 (74.4%) minority variants observed within the 757 total variants identified by NGS. This elaborates that NGS is able to detect approximately 74.4% of the minority variants present above a cutoff of 1% of the viral quasispecies. The 563 minority variants were dispersed unproportionally, with some mutations appearing more frequent than others. Such variants were observed in 65.5% (38/58) of the sampled population amongst the study participants—some of which variants included HIV-DR mutations. Several reports have elucidated the presentation of similar minority variants in different populations as reported (Nicot et al., 2012; Capobianchi et al., 2013; Wensing et al., 2019). The current study also highlights the most prevalent minority variants, which included; K65R in 38/58 (65.5%), K14R in 37/58 (63.79%), K45R in 37/58 (63.79%), L63P in 37/58 (63.79%), and I64V in 37/58 (63.79%). The mutation K65R, that also presents the highest prevalence in the current study, is one of the low-abundance drug resistance mutations conferring resistance to at least one class of drugs—and within the NRTI class. However in other studies, this mutation has been categorized as a relatively rare drug resistance conferring variant that causes loss of drug susceptibility of HIV-1 to most NRTIs (tenofovir, didanosine, abacavir and stavudine), with exception to zidovudine (Brenner & Coutsinos, 2009). And essentially, even though there are limited studies that have focused on understanding the incidences of minority variants more so in sub-Saharan Africa, the outstanding prevalence of K65R perfectly relates to findings in a few researches conducted in different geographical regions in HIV high burdened areas. In Israel, there was a reported high incidence of K65R is the analyzed population (Turner et al., 2009); in West Africa, particularly Togo, there was evidence of a high prevalence of K65R (Dagnra et al., 2011); whereas within the Resistance Monitoring study, a multicenter prospective observational cohort of HIV-1-infected adults who received ART in routine circumstances at 13 clinical sites in Kenya, Nigeria, South Africa, Uganda, Zambia, and Zimbabwe—also reported a significant incidence of K65R amongst samples collected from participants (Sigaloff et al., 2011). Other studies conducted in Southern Africa have elucidated significant occurrences of K65R as a key variant amongst HIV-1 positive individuals experiencing virological failure (Skhosana, 2015; Etta et al., 2017). Accordingly, in the current study, it should be noted that K65R remains the most important and significant variant identified. The results from the study findings also elaborate that there is observed co-existence of mutations amongst the analyzed samples, which findings strongly compare with other research that observed co-existence of variants, most notably K65R with S68G. The alteration S68G, has been reported to partially restore the replication defect associated with K65R as reported (Parikh et al., 2006). Additionally, the variants K14R and K70R were observed to be housed, each at a frequency of 60.3% of all the samples analyzed. The variant K14R is a single nucleotide variant that has been reported to confer no resistance to the most important ART drugs (Descamps et al., 2009), whereas K70R has been categorized as a classical Thymidine Analog mutation (TAM) known to cause intermediate-level resistance to AZT, and possibly low-level resistance to d4T and TDF (Quintana & José, 2013; Hachem et al., 2020). The current study also observed several nonpolymorphic mutations from the analyzed data as categorized by the HIV Stanford database. Such variants include; P225H, G190A, M46I, M46L, K70Q, K70N, K70S, K70T, K101E, and L100I—that are categorically believed not to occur in absence of therapy, contrary to the polymorphic mutations that occur frequently in viruses not exposed to selective drug pressure (Shafer & Schapiro, 2008; WHO, 2020). It should also be noted that the mutation P225H is an EFV-selected variant that occasionally occurs in combination with K103N. This results in synergistic reduction of susceptibility to NVP, EFV and DOR (Kouamou et al., 2021). As well, another reported mutation G190A has been associated with high-level resistance to NVP and intermediate resistance to EFV (Yang et al., 2015), whereas K101E is a primary accessory mutation that causes intermediate resistance to NVP and RVP, low-level resistance to EFV, and potentially low-level resistance to ETR as reported (Wu et al., 2015). The mutations M46I and M46L are relatively non-polymorphic Protease inhibitor-selected mutations, associated with reduced susceptibility to PIs with an exception of DRV, as reported (Watera et al., 2021). The mutations K70E and K70G are known to cause low-level resistance to TDF, ABC, 3TC and FTC, and also increase susceptibility to AZT (Siller et al., 2021). Other identified mutations included K70Q, K70N, K70S and K70T, that are rare non-polymorphic NRTI-selected mutations relating to resistance profiles of K70E and K70G described herein above. Earlier studies elaborated that the mutation L100I usually occurred in combination with K103N even though there is evidence that such variants may cause high level resistance to NVP and EFV (Soriano & de Mendoza, 2002). Other than the drug resistant conferring variants, there were also several unusual mutations whose clinical significance is still unknown and they could be nominated and validated as potential diagnostic targets, drug resistance targets, and so on. For instance, the present study has also unveiled a number of minority variants of clinical importance that were highly prevalent in the analyzed samples. These included; K14R, K45R, L63P, I64V, I15V, L83I, V77I among others as summarized in the supporting files HDR-Supp.zip.

Conclusion

Next Generation Sequencing approaches can confidently reveal low-abundance drug resistance variants (minority variants) in populations of HIV-1 resistant individuals that have been reported to have virological failure. NGS, due to its high throughput data is able to detect low abundance variants as low as 1% of the viral quasispecies. This has been a limitation of the conventional Sanger sequencing that is currently considered the gold standard for HIV-DR testing. It is therefore vital that patients experiencing virological failure are tested using NGS as Sanger misses most of the low abundance variants that are of clinical significance. The frequency of drug resistance varies among individuals suggesting that discrete viral populations have different levels of genetic diversity. An estimated 65.5% of the sampled population harbored low-abundance HIV-1 variants, most of which are associated with virological failure, and consequently antiviral drug resistance.

Data availability

Zenodo. Next generation sequencing reveals low-abundance HIV-1 drug resistance mutations among patients experiencing virological failure at the time of therapy switching in Uganda, DOI: https://doi.org/10.5281/zenodo.6326916 (Namaganda et al., 2022).

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

Author contributions

Conceptualization: Maria Magdalene Namaganda, Hakim Sendagire, Gerald Mboowa, Edgar Kigozi, Barbara Castelnuovo, David Patrick Kateete, Moses L Joloba.

Funding Acquisition: Gerald Mboowa, Maria Magdalene Namaganda.

Sample Acquisition: Hakim Sendagire, Susan Nabadda, Isaac Ssewanyana, Chris Okirya.

RNA Extraction: Maria Namaganda, Moses Luutu Nsubuga, Ashaba Fred.

Library preparation and Sequencing: Maria Namaganda, Moses Luutu Nsubuga, Ashaba Fred, Edgar Kigozi.

Bioiformatics Analysis: Maria Magdalene Namaganda, Gerald Mboowa, Stephen kanyerezi, Fredrick Elishama Kakembo, Jupiter Marina Kabahita.

Manuscript writing: All authors have given final approval of the version to be published.