Doravirine-associated resistance mutations in antiretroviral therapy naïve and experienced adults with HIV-1 subtype C infection in Botswana

Highlights • Lower DOR-resistance in ART-naïve versus ART-experienced individuals with VF.• DOR-resistance was higher among individuals failing NNRTI-containing regimens.• DOR use in NNRTI-experienced population should be based on HIV drug resistance data.


Introduction
The use of antiretroviral therapy (ART) has had a significant effect on the HIV-1 epidemic, turning it into a chronic manageable disease [ 1 , 2 ]. The available FDA-approved HIV antiretroviral drugs (ARVs) have increased over time and have improved efficacy and toxicity profiles [ 3 , 4 ]. However, the development of HIV drug resistance mutations (DRMs) threatens the success of combination ART (cART) and further limits future HIV treatment options [ 5 , 6 ]. Recently, a rise in the prevalence of HIV DRMs to first-generation ART non-nucleoside reverse transcriptase inhibitor (NNRTI)-containing regimens has been observed [7][8][9][10] , which has contributed to the use of a dolutegravir (DTG)-based first-line ART regimen. DTG has proven to have high efficacy and a higher genetic barrier of resistance [11] . In June 2016, the Botswana National ART program adopted DTG-based first-line therapy for all people with HIV (PWH) initiating ART [12] . Prior studies have reported the development of HIV variants harbouring DTG resistance, especially in highly treatment-experienced patients with previous raltegravir exposure, and these patients tend to harbour variants with multidrug resistance mutations [ 13 , 14 ]. In addition, higher DTG-associated virologic failure (VF) was observed among NNRTI-experienced individuals harbouring NNRTI-associated resistance [15] . This highlights the need for new ART post-DTG, which lack cross-resistance with the current ARV regimens.
A newly approved third-generation NNRTI, doravirine (DOR), has been shown to be more potent, with distinct resistance patterns compared to the earlier-generation NNRTIs [16] . DOR has a better safety profile and has a long half-life of about 12-21 h, enabling it to be dosed once a day [ 17 , 18 ]. In vitro studies have shown that DOR is efficacious in the presence of most common reverse transcriptase resistance mutations, K103N and Y181C, found in NNRTI-experienced individuals [ 16 , 19 ]. To extend these in vitro studies, it is important to evaluate whether DOR can maintain its potency in a population with a high prevalence of the K103N and Y181C mutations, such as PWH in Botswana. We investigate the prevalence of DOR-associated resistance mutations in ART-naïve and ART-experienced PWH in Botswana, where NNRTIs were part of first-line ART regimens from 2002 to 2016. These data could guide policy on DOR use in the Botswana ART program and other countries where HIV-1 subtype C predominates and have similar ART experiences.

Study population
PWH aged 16-64 years in 30 communities in northern, central and southern parts of Botswana participated in the Botswana Combination Prevention Project (BCPP) between 2013 and 2018 [20] . The BCPP impact evaluation survey enrolled 16 years or older individuals, regardless of HIV or ART status, who were residing in a random sample of approximately 20% of households among 30 communities at study entry (baseline household survey [BHS]) into a longitudinal cohort (annual household survey [AHS]) that was followed for 30 months as described elsewhere [20] . In addition to 15 intervention communities (15 villages that received preventative measures such as three-drug ARV to all people with viral loads of 10 0 0 0 copies/mL and/or CD4 cells ≤500 to track HIV incidence), PWH who were receiving ART in local clinics were recruited to provide a one-time blood sample for HIV sequencing and metadata but were excluded on the longitudinal cohort. For this analysis, samples and data from all PWH in both the AHS cohort and from the clinics were used.

Selection of study participants
In this analysis, we included PWH who were either ART-naïve or ART-experienced, who had HIV-1 viral load (VL) measurement at the first BCPP study visit and had available HIV-1 sequence. HIV-1 VL of participants was quantified using Abbott m20 0 0sp/rt assay (Wiesbaden, Germany) with a range of 40-10 0 0 0 0 0 0 copies/mL [21] . Among participants who were on ART, some were classified as having VF (HIV-1 VL > 400 copies/mL, as per current Botswana's HIV treatment guidelines definition) [22] . Participants with HIV-1 VL ≤400 copies/mL were regarded as virally suppressed ( Fig. 1 ).

Near full-length HIV genotyping
Both HIV-1 proviral DNA sequences and viral RNA sequences were generated by a long-range HIV genotyping protocol described elsewhere [ 21 , 23 ]. The next-generation sequencing (NGS) was performed by the BioPolimers Facility at Harvard Medical School ( https://genome.med.harvard.edu/ ) and through collaboration with PANGEA HIV consortium [21] at the Wellcome Trust Sanger Institute (Cambridge, UK; http:// www.sanger.ac.uk/ ) with highsequencing coverage using Illumina platforms MiSeq and HiSeq [21] .

Analysis of HIV-1 drug resistance mutations
DOR-associated resistance mutations were identified according to the lists of surveillance drug resistance mutations and major drug resistance mutations (DRMs) in the Stanford University HIV Drug Resistance Database algorithm 9.1 ( https://hivdb.stanford. edu/hivdb/by-sequences/ ). The level of DOR resistance was predicted according to the Stanford HIV DRM penalty scores and resistance interpretation (individuals with low-level, intermediatelevel and high-level resistance were considered to have drug resistance). Only intermediate and high-level specific DOR-associated resistance mutations were reported. The list of mutations assessed are shown in Table 1 .
The prevalence of DOR-associated resistance was estimated in the following groupings: (i) overall; (ii) ART-naïve individuals; (iii) virologic failure (VF) on ART; and (iv) viral suppression (HIV-1 RNA  ≤400 copies/mL) on ART. The DOR prevalence was compared between groups (ii) and (iii), as well as between groups (iii) and (iv). The prevalence of specific DOR-associated resistance mutations was estimated within each group and compared among groups. We also assessed the presence of DOR-associated resistance mutations in a subset of participants on NNRTI-based regimens with at least one major efavirenz (EFV)/nevirapine (NVP)-associated resistance mutation (L100I, K101E/P, K103N/S, V106A/M, Y181C/I/V, Y188L/C/H, G190A/S/E and M230L).

Apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like (APOBEC)-induced hypermutations
Guanine-to-adenine transitions (G-to-A) apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like (APOBEC)-induced hypermutations were screened in the viral sequences using the Hypermut program available in Los Alamos National Laboratory HIV Database tools ( http://www.hiv.lanl.gov/ ) [26] . Adjustment for hypermutations was performed [23] using the HIV-1C consensus sequence as a reference. The adjusted hypermutations were accounted for using the cumulative number of mutations across the length of the sequences from the analysed HIV-1 pol gene. The ad-justment for hypermutations was performed before the drug resistance analysis, as a part of quality control. HIV DRMs identified as hypermutations were not included in the prevalence of DOR resistance in this study.

Statistical analysis
Patient demographics between ART-naïve and individuals on ART experiencing VF were compared using Wilcoxon rank-sum test for continuous variables such as plasma log 10 HIV-1 VL and χ 2 test for categorical variables such as sex. The prevalence of mutations was estimated with 95% confidence using the binomial exact method for each group. The prevalence of mutations among ARTnaïve group and individuals experiencing VF on ART was compared using comparison of proportions test. The potential association between the prevalence of HIV-DRM and sex, age, EFV use, NVP use, log 10 HIV-1 VL was assessed by univariate logistic regression (adjusted for clustering by community) in NNRTI-failing individuals with at least one EFV/NVP-associated resistance mutation for detection of DOR-associated resistance mutations. A P value < 0.05 was considered statistically significant. All the statistical analysis was done using STATA version 14 software.

Ethical considerations
The BCPP study was approved by the Institutional Review Board (IRB) at the U.S. Centers for Disease Control and Prevention and the Botswana IRB (HRDC), and it is registered at ClinicalTrials.gov (NCT01965470). All recruited participants provided written informed consent for participation.

Participants characteristics
The median age at enrolment of the participants included in this analysis was 34 years (interquartile range [Q1, Q3: 27.1, 42.0]), and study participants were mostly women (66%). Among 6078  Table 2: females, older participants ( > 29 years) and those on ART were most likely to have HIV-1 VL < 400 copies/mL.

Association of DOR-associated resistance mutations among individuals failing NNRTI-based ART with at least one EFV/NVP-associated resistance mutation
In the univariate logistic regression, among NNRTI failing participants with at least one EFV/NVP-associated resistance mutation, sex, age, EFV use, NVP use and median HIV-1 VL log 10 copies/mL  were not associated with DOR-associated resistance mutations in a small population of 67 participants ( Table 4 ).

Overall prevalence of mutations associated with intermediate and high-level DOR resistance among individuals with viral suppression and virologic failure (VF) on ART
We also compared the prevalence of HIV-DRMs associated with DOR resistance among individuals with viral suppression on ART (HIV-1 VL ≤400 copies/mL) compared to individuals experiencing VF on ART. The overall prevalence of mutations associated with DOR resistance was similar among individuals with viral suppression on ART, 910/4517 (20

Discussion
To the best of our knowledge, this is the first study to use a large data set with a good representation of the Botswana population to assess the prevalence of mutations associated with DOR resistance among PWH. The reported overall prevalence of DORassociated resistance was 12.3% higher than 1.62% predicted in the Polish study [27] ; however, this prevalence was lower compared to 56.4% [28] and 62% [29] reported in other studies. The difference in the prevalence of DOR-associated resistance across studies may be attributed to the use of different genotyping thresholds to detect DOR-associated resistance mutations, different HIV subtypes, different HIV treatment guidelines in the different studies, and different HIV Stanford algorithm versions used in the analysis.
A significantly lower overall prevalence of DOR-associated resistance was reported among ART-naïve compared to individuals experiencing VF on ART ( P < 0.01), confirming the opportunity for the use of DOR-containing regimens among PWH who are ART-naïve in Botswana. However, the prevalence of DOR-associated resistance among ART-naïve indi viduals (11.0%: 139/1261) in this study was statistically higher than the 2.9% prevalence of other NNRTI (first-generation)-associated resistance mutations previously reported in the BCPP study [21] (31/1069, P < 0.01). Some studies reported higher resistance to other NNRTI-containing regimens when compared to DOR. A study from Greece, Italy and France reported a lower prevalence of DOR-associated resistance (1.4%) compared to first-generation NNRTI-associated resistance (EFV and NVP; 4.3%), with no association between HIV-1 subtype and presence of DOR-associated resistance mutations [30] . In the HIV-1C settings, the lower prevalence of DOR-associated resistance mutations was reported at 4.7% compared to 9.4% NVP/EFV-associated resistance mutations among ART-naïve population [31] .
Our study reports the prevalence of 'archived' DOR-associated resistance mutations of 20.1% among individuals with viral suppression on ART, similarly to those with VF on ART. Although the predominant prevalence (62.7%) of DOR-associated resistance mutations was among participants failing NNRTI-based therapy in our findings, this was lower when compared to 84.8% reported in a South African study [31] but higher compared to 42% reported in European studies [ 32 , 33 ]. The prevalence of mutations associated with high-level DOR resistance was low among NNRTI-experienced patients in non-B vs. B HIV-1 subtypes in Italy, with higher prevalence in HIV-1 subtype C; furthermore, the presence of mutation Y188L was associated with HIV-1C subtype [33] . Our results together with other studies support the notion that some DORassociated resistance mutations are HIV-1 subtype-specific, e.g., intermediate DOR-associated resistance mutation V106M. There is evidence that individuals on DOR harbour distinct resistance profiles of HIV-DRMs compared to other NNRTI regimens [ 30 , 34 ]. Furthermore, there is still a need to understand the mechanisms of DOR-associated resistance, especially among individuals who harbour DOR-associated resistance without specific DOR-associated resistance mutations. The presence of DOR resistance in individuals without specific DOR-associated resistance mutations was observed in phenotypic studies [ 19 , 34 ]; this shows that there could be other mutations that play a part in DOR resistance that are yet to be described.
The high-level DOR-associated resistance mutations G190E, F227L and Y318F all showed a similar prevalence among the three groups: ART-naive, individuals with VF, and individuals on ART with viral suppression. The similar prevalence of these mutations across the three groups highlights the importance of gaining a better understanding of their mechanisms, particularly in ART-naïve and ART-suppressed individuals. It was difficult to determine if the presence of these mutations in ART-naïve individuals was due to undisclosed ART use in the BCPP cohort [35] or to transmitted DRM circulating in the Botswana population [9] . A study conducted in 23/24 individuals with baseline first-generation NNRTI-associated resistance mutations reveals that switching to DOR/lamivudine/tenofovir disoproxil fumarate (DOR/3TC/TDF) leads to viral suppression within 48 weeks of follow-up [36] . These findings suggest that some of the firstgeneration NNRTI resistance mutations do not have an effect on DOR activity.
EFV has been associated with an increased risk of developing high-level DOR-associated resistance mutations [33] , but our findings are discordant with these reports. Although we report no association of EFV use with the presence of DOR-associated resistance, the cohort used had a relatively low number of individuals who were failing NNRTI with at least one EFV/NVP-associated resistance mutation experiencing VF; therefore, our study was not powered to reveal the effect of EFV/NVP-associated resistance in the development of DOR-associated resistance mutations. Nonetheless, our results still highlight the importance of genotyping before initiation of a DOR-containing regimen among individuals with previous NNRTI exposure.
Because our study had a small number of participants on DTGbased ART, our findings cannot reveal whether DOR can be used as an alternative drug for DTG-failing participants with multiresistance mutations. As a result, future studies should investigate DOR-associated resistance mutations in DTG-failing participants with multi-drug resistance.
In conclusion, the prevalence of intermediate and high-level DOR-associated resistance mutations was low among ART-naïve and ART-failing individuals, with the ART-failing group showing a statistically higher prevalence compared to the ART-naïve group. Specifically, a higher prevalence was observed among PWH with EFV/NVP-associated resistance experiencing VF. Our results support the use of DOR among ART-naïve patients in Botswana; however, we highly recommend HIV drug resistance testing among NNRTIfailing individuals before DOR-based regimen initiation.

Declaration of Competing Interest
None declared.

Funding
The BCPP Impact Evaluation was funded by the President's Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention (CDC, cooperative agreements U01 GH0 0 0447 and U2G GH0 01911). The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the funding agencies. OTB was supported by the Fogarty International Center (grant no. 5D43TW009610). SG and WTC were partially supported by H3ABioNet. H3ABioNet is supported by the National Institutes of Health Common Fund [U41HG006941]. H3ABioNet is an initiative of the Human Health and Heredity in Africa Consortium (H3Africa) programme of the African Academy of Science (AAS). SM, NK and OTB were supported by the Trials of Excellence in Southern Africa (TESA III), which is part of the EDCTP2 programme supported by the European Union (grant CSA2020NoE-3104 TESAIII CSA2020NoE). SM, SG, WTC and NOM are partly supported through the Sub-Saharan African Network for TB/HIV Research Excellence (SANTHE 2.0) from the Bill and Melinda Gates Foundation (INV-033558). PANGEA-HIV is funded primarily by the Bill and Melinda Gates Foundation (BMFG). S.L. is supported by National Institutes of Health Common Fund [K24 AI131928]. The views expressed in this publication are those of the authors and not necessarily those of AAS, NEPAD Agency, Wellcome Trust or the U.K. government. The funders had no role in the study design, data collection and decision to publish, or in the preparation of the manuscript.

Ethical approval
Not applicable.

Data availability
All relevant data are within the paper and the Supplemental Notes, Figures and Tables. HIV-1 sequences are available on request through the PANGEA consortium ( www.pangea-hiv.org ). BCPP data is available at: https://data.cdc.gov/Global-Health/ Botswana-Combination-Prevention-Project-BCPP-Publi/ qcw5-4m9q .

Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.jgar.2022.08.008 .