Advertisement
Review article| Volume 148, P135-145, August 2022

Acquired HIV drug resistance mutations on first-line antiretroviral therapy in Southern Africa: Systematic review and Bayesian evidence synthesis

Open AccessPublished:February 19, 2022DOI:https://doi.org/10.1016/j.jclinepi.2022.02.005

      Abstract

      Objective

      To estimate the prevalence of NRTI and NNRTI drug resistance mutations in patients failing NNRTI-based ART in Southern Africa.

      Study design

      We conducted a systematic review to identify studies reporting drug resistance mutations among adult people living with HIV (PLWH) who experienced virological failure on first-line NNRTI-based ART in Southern Africa. We used a Bayesian hierarchical meta-regression model to synthesize the evidence on the frequency of eight NRTI- and seven NNRTI-DRMs across different ART regimens, accounting for ART duration and study characteristics.

      Results

      We included 19 study populations, including 2,690 PLWH. Patients failing first-line ART including emtricitabine or lamivudine showed high levels of the M184V/I mutation after 2 years: 75.7% (95% Credibility Interval [CrI] 61.9%-88.9%) if combined with tenofovir, and 72.1% (95% CrI 56.8%-85.9%) with zidovudine. With tenofovir disoproxil fumarate, the prevalence of the K65R mutation was 52.0% (95% CrI 32.5%-76.8%) at 2 years. On efavirenz, K103 was the most prevalent NNRTI resistance mutation (57.2%, 95% CrI 40.9%-80.1%), followed by V106 (46.8%, 95% CrI 31.3%-70.4%).

      Conclusions

      NRTI/NNRTI drug resistance mutations are common in patients failing first-line ART in Southern Africa. These patients might switch to dolutegravir-based regimen with compromised NRTIs, which could impair the long-term efficacy of ART.

      Keywords

      What is new?
      • As many Southern African countries transition to dolutegravir-based antiretroviral therapy, there is a concern that patients who have been treated with NRTIs for a long period may have acquired resistance to NRTI, and then may be in situation of functional monotherapy when switched to dolutegravir-based antiretroviral therapy.
      • We systematically reviewed studies assessing acquired drug resistance to NRTIs in Southern Africa among patients failing first-line ART and developed an innovative Bayesian evidence synthesis model to estimate the occurrence of several drug resistance mutations accounting for different study designs, regimens and treatment durations.
      • We found that drug resistance mutations to NRTIs are common in patients failing first-line ART in Southern Africa, including mutations conferring high-level resistance to NRTIs that will be commonly combined with dolutegravir in the new regimen.
      • We conclude that many patients failing an NNRTI-based ART may switch to a dolutegravir-based ART with compromised NRTIs, which could impair the long-term efficacy of ART in Southern Africa.

      1. Introduction

      In 2014, UNAIDS communicated its goal of ending the AIDS epidemic as a public health threat by 2030 [

      UNAIDS. Ending the AIDS epidemic by 2030.

      ]. The strategy to achieve this objective involved stepping up testing and antiretroviral therapy (ART) in low and middle-income countries. In the last 2 decades, first-line ART consisted of a non-nucleoside reverse-transcriptase inhibitor (NNRTI) combined with two nucleos(t)ide reverse transcriptase inhibitors (NRTI) in most of these countries. For many NNRTIs, a single mutation will lead to high-level drug resistance, for example, the K103N mutation [
      • Bacheler LT
      • Anton ED
      • Kudish P
      • Baker D
      • Bunville J
      • Krakowski K
      • et al.
      Human immunodeficiency virus type 1 mutations selected in patients failing efavirenz combination therapy.
      ]. The combination of this low genetic barrier of NNRTI and their widespread use has lead to a continuous increase of pre-treatment drug resistance (PDR) to NNRTIs [
      • Hauser A
      • Kusejko K
      • Johnson LF
      • Wandeler G
      • Riou J
      • Goldstein F
      • et al.
      Bridging the gap between HIV epidemiology and antiretroviral resistance evolution: Modelling the spread of resistance in South Africa.
      ,
      • Riou J
      • Dupont C
      • Bertagnolio S
      • Gupta RK
      • Kouyos RD
      • Egger M
      • et al.
      Drivers of HIV-1 drug resistance to non-nucleoside reverse-transcriptase inhibitors (NNRTIs) in nine southern African countries: a modelling study.
      ], which has exceeded the WHO threshold of 10% in Southern Africa [

      World Health Organization (WHO). HIV drug resistance report 2017 [Internet]. 2017 [cited 2020 Nov 19]. Available from:https://www.who.int/hiv/pub/drugresistance/hivdr-report-2017/en/.

      ]. Several countries in the region are therefore transitioning from the NNRTI based first-line triple therapy towards an integrase-strand-transfer-inhibitor (InSTI) based first-line regimen. WHO recommends TLD, a fixed-dose combination of tenofovir, lamivudine and dolutegravir (DTG) [
      World Health Organization
      Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection [Internet]. WHO.
      ].
      Among people living with HIV (PLWH) transitioning to DTG-based therapy, those with unsuppressed viral load are likely to have acquired resistance to NNRTIs or NRTIs. As NRTI-resistance reduces the activity of the NRTI backbone of DTG-based regimens, PLWH switching with pre-existing NRTI resistance are at higher risk of DTG-failure and of developing DTG-resistance. A cross-sectional survey in South Africa showed that 83% of patients failing NNRTI-based treatment had the M184V/I mutation, and over half the patients developed K65R, both of which confer resistance to NRTIs [
      • Steegen K
      • Bronze M
      • Papathanasopoulos MA
      • van Zyl G
      • Goedhals D
      • Variava E
      • et al.
      HIV-1 antiretroviral drug resistance patterns in patients failing NNRTI-based treatment: results from a national survey in South Africa.
      ]. The risk of developing drug resistance mutations (DRMs) depends on several factors, including the ART regimen used and its duration [
      • Tang MW
      • Shafer RW.
      HIV-1 antiretroviral resistance: Scientific principles and clinical applications [Internet].
      ].
      In contrast to transmitted drug resistance, only a few studies have reported the prevalence of acquired drug resistance mutations in patients failing first-line ART in Southern Africa [

      World Health Organization (WHO). HIV drug resistance report 2017 [Internet]. 2017 [cited 2020 Nov 19]. Available from:https://www.who.int/hiv/pub/drugresistance/hivdr-report-2017/en/.

      ]. Most studies of acquired drug resistance (ADR) were performed in Europe or North America, where subtype B is most prevalent, while most HIV infections in Southern Africa are caused by subtype C [
      • Wainberg MA.
      HIV-1 subtype distribution and the problem of drug resistance.
      ,
      • Buonaguro L
      • Tornesello ML
      • Buonaguro FM.
      Human immunodeficiency virus type 1 subtype distribution in the worldwide epidemic: pathogenetic and therapeutic implications.
      ]. In-vitro studies suggest that different subtypes might lead to different rates of ADR [
      • Coutsinos D
      • Invernizzi CF
      • Xu H
      • Moisi D
      • Oliveira M
      • Brenner BG
      • et al.
      Template usage is responsible for the preferential acquisition of the k65r reverse transcriptase mutation in subtype c variants of human immunodeficiency virus type 1.
      ,
      • Invernizzi CF
      • Coutsinos D
      • Oliveira M
      • Moisi D
      • Brenner BG
      • Wainberg MA.
      Signature nucleotide polymorphisms at positions 64 and 65 in reverse transcriptase favor the selection of the k65r resistance mutation in HIV-1 Subtype C.
      ,
      • White E
      • Smit E
      • Churchill D
      • Collins S
      • Booth C
      • Tostevin A
      • et al.
      No evidence that HIV-1 subtype c infection compromises the efficacy of tenofovir-containing regimens: cohort study in the United Kingdom.
      ]. The few studies of ADR from Southern Africa and other resource-limited settings are heterogeneous, for example, regarding ART regimens or treatment duration. The heterogeneous nature of these studies makes summarizing and interpreting the evidence regarding ADR difficult. To address this challenge, we performed a systematic review and Bayesian evidence synthesis of studies reporting frequencies of drug resistance mutations in patients failing first-line ART in Southern Africa, a region heavily affected by HIV [

      UNAIDS. AIDSinfo. Number of people living with HIV [Internet]. [cited 2018]. Available from:http://aidsinfo.unaids.org/.

      ].

      2. Methods

      2.1 Literature search

      We searched seven bibliographic databases from inception, including Embase and Medline, using terms for “HIV” AND “antiretroviral therapy” AND “drug-resistance mutations” AND “Southern Africa”. Appendix A provides details on the literature search. We performed the final search on Oct 9 2020, and de-duplicated references in EndNote (version 18.0.0). We registered this review in the International Prospective Register of Systematic Reviews (PROSPERO, No. CRD42017076406) [
      • Booth A
      • Clarke M
      • Dooley G
      • Ghersi D
      • Moher D
      • Petticrew M
      • et al.
      The nuts and bolts of PROSPERO: an international prospective register of systematic reviews.
      ].

      2.2 Inclusion criteria

      We searched for studies from Southern African countries in adult (15 years or older) PLWH on NNRTI-based first-line ART who were at least 3 months on treatment and experienced virological failure. The results of at least 10 genotypic resistance tests covering all major single mutations listed in the Stanford HIV drug resistance database had to be reported [
      • Shafer RW.
      Rationale and uses of a public HIV drug-resistance database.
      ]. First, FG and AH assessed articles based on their title and abstract. Second, AH and either FG or MLR assessed potentially eligible studies based on the full text. We compared the results of full-text screening and reached a consensus on eligibility by discussion.

      2.3 Data extraction

      We extracted details on study populations and settings, study year, the number of patients tested for viral load and with virological failure, and the time spent on ART. We recorded the definition of failure, that is the viral load threshold and the number of measurements required, and the first-line ART regimen. We extracted data on amino-acid substitutions commonly associated with drug resistance and referred to them as drug resistance mutations. We grouped resistance mutations by locus (e.g., M184I/V). We defined the year of study as the year of the analysis, the year enrollment ended, or by the ‘sampling date’ recorded in GenBank [
      • Sayers EW
      • Cavanaugh M
      • Clark K
      • Pruitt KD
      • Schoch CL
      • Sherry ST
      • et al.
      GenBank.
      ]. If studies reported stratified data (e.g., by country or level of urbanization), we extracted the data separately.

      2.4 Bayesian evidence synthesis

      We developed a hierarchical meta-regression model to estimate the prevalence of eight different NRTI mutations: K65N/R, M184I/V, M41L, D67N/G/E, K70E/G/R, L210W, T215F/I/N/S/Y, K219D/E/N/Q/R Fig. 1. and Appendix B detail the model structure. The model includes a random effect to account for study-level heterogeneity and a random effect at the mutation level. Our model accounts for the effect of both treatment duration and antiretroviral regimen on the prevalence of individual resistance mutations. To reflect that resistance is usually acquired during treatment failure, we developed a function that converts time on ART to time on failure (see Section 1.1 of Appendix B). We considered that time to acquiring a mutation was exponentially distributed, thus assuming a risk of developing a DRM is constant over time on failure. We examined the effect of different NRTI drugs on the risk of developing a DRM. We report model estimates of the prevalence of the seven major NRTI DRMs at baseline (treatment start, corresponding to the intercept of the regression model) and after 2 years. We opted for weakly informative prior distributions [
      • Gelman A
      • Simpson D
      • Betancourt M.
      The prior can often only be understood in the context of the likelihood.
      ]. Appendix B reports the model equations and the prior distributions.
      Fig 1
      Fig. 1Bayesian hierarchical model adjusting for the different levels of heterogeneity.
      We considered five NRTI drugs: didanosine (ddI), emtricitabine/lamivudine (FTC/3TC), tenofovir disoproxil fumarate (TDF), stavudine (d4T), and zidovudine (ZDV). As NRTI backbones mainly comprise either ddI or FTC/3TC and TDF, d4T, or ZDV, there were strong correlations between the drugs. We dealt with multicollinearity by selecting the covariates with the strongest effects first. In a sensitivity analysis, we also ran the model with all covariates. We imputed the time on ART in three studies [
      • El-Khatib Z
      • Ekstrom AM
      • Ledwaba J
      • Mohapi L
      • Laher F
      • Karstaedt A
      • et al.
      Viremia and drug resistance among HIV-1 patients on antiretroviral treatment: a cross-sectional study in Soweto, South Africa.
      ,
      • Rupérez M
      • Pou C
      • Maculuve S
      • Cedeño S
      • Luis L
      • Rodríguez J
      • et al.
      Determinants of virological failure and antiretroviral drug resistance in Mozambique.
      ,
      • Wallis CL
      • Mellors JW
      • Venter WDF
      • Sanne I
      • Stevens W.
      Varied patterns of HIV-1 drug resistance on failing first-line antiretroviral therapy in South Africa.
      ] that did not report it, assuming that the missing ART durations followed a gamma distribution with mean and variance calculated from the observed ART durations in the other studies. We also ran the model after discarding these three studies to assess the impact of data imputation (see Appendix B).
      We also estimated the prevalence of any of six thymidine-analogue mutations (TAM), that is M41L, D67E/G/N, K70E/G/R, L210W, T215F/I/N/S/Y, or K219Q/E. As only a few studies reported the prevalence of any TAM, we adopted a method that estimated the correlation between the six TAMs (using results from the nine studies reporting the prevalence of the single TAMs and the prevalence of any TAM), assuming a multivariate Bernoulli distribution. We used this estimate to calculate the prevalence of any TAM (see Appendix B). Finally, we applied the same model to estimate the prevalence of seven NNRTI mutations (K101E/H/P/Q, K103N/R/S, V106A/I/M, V108I, Y181C/I/V, Y188C/H/L, G190A/E/R/S) as for the NRTI DRMs. We report estimates of DRM prevalence after 2 years of efavirenz (EFV) or nevirapine (NVP)-based regimens. The baseline prevalence of NNRTI DRMs was not reported because EFV and NVP select for the same NNRTI DRMs, which prevents the model from providing reliable estimates.
      All Analyses were performed in a Bayesian framework using the rstan package in R (version 4.1.1). Multi-level Bayesian meta-analyses are particularly well-suited when the number of studies is small, as they allow to better incorporate the uncertainty on the between-study variance [

      Williams DR, Rast P, Bürkner P-C. Bayesian meta-analysis with weakly informative prior distributions. 2018 [cited 2021 Dec 4]; Available from:https://psyarxiv.com/7tbrm/.

      ].

      2.5 Risk of bias

      Two of us (AH, ME) independently assessed study designs to gauge the representativeness of the patients for whom HIV genotypes were available. We calculated the percentage among patients with virological failure who had HIV genotypes. We considered studies at high risk of selection bias if (i) genotyping rates were low (<50% of patients with failure were successfully genotyped); (ii) there were relevant differences between the characteristics of genotyped and non-genotyped patient groups or iii) there was no clearly defined source population for patients to be genotyped. In a sensitivity analysis, we removed studies fulfilling one or more of these criteria and reran the analyses to examine the effect these studies had on the results (see Appendix B).

      3. Results

      3.1 Selection and characteristics of studies

      Our initial search produced 8,534 articles; 3,462 were duplicates (Fig. 2). We further excluded 4,836 papers based on title and abstract. We read the full texts of the remaining 236 articles. Of these, 18 studies were eligible, with 19 unique study populations from South Africa (13 studies), Botswana (1), Lesotho (1), Malawi (1), Mozambique (1), Tanzania (1), and Zambia (1) [
      • Steegen K
      • Bronze M
      • Papathanasopoulos MA
      • van Zyl G
      • Goedhals D
      • Variava E
      • et al.
      HIV-1 antiretroviral drug resistance patterns in patients failing NNRTI-based treatment: results from a national survey in South Africa.
      ,
      • El-Khatib Z
      • Ekstrom AM
      • Ledwaba J
      • Mohapi L
      • Laher F
      • Karstaedt A
      • et al.
      Viremia and drug resistance among HIV-1 patients on antiretroviral treatment: a cross-sectional study in Soweto, South Africa.
      ,
      • Rupérez M
      • Pou C
      • Maculuve S
      • Cedeño S
      • Luis L
      • Rodríguez J
      • et al.
      Determinants of virological failure and antiretroviral drug resistance in Mozambique.
      ,
      • Wallis CL
      • Mellors JW
      • Venter WDF
      • Sanne I
      • Stevens W.
      Varied patterns of HIV-1 drug resistance on failing first-line antiretroviral therapy in South Africa.
      ,
      • Marconi VC
      • Sunpath H
      • Lu Z
      • Gordon M
      • Koranteng-Apeagyei K
      • Hampton J
      • et al.
      Prevalence of HIV-1 drug resistance after failure of a first highly active antiretroviral therapy regimen in kwazulu natal, South Africa.
      ,
      • Novitsky V
      • Wester CW
      • DeGruttola V
      • Bussmann H
      • Gaseitsiwe S
      • Thomas A
      • et al.
      The reverse transcriptase 67N 70R 215Y genotype is the predominant tam pathway associated with virologic failure among hiv type 1c-infected adults treated with ZDV/ddI-Containing HAART in Southern Africa.
      ,
      • Rossouw T
      • Nieuwoudt M
      • Manasa J
      • Malherbe G
      • Lessells R
      • Pillay S
      • et al.
      HIV drug resistance levels in adults failing first-line antiretroviral therapy in an urban and a rural setting in South Africa.
      ,
      • Seu L
      • Mulenga LB
      • Siwingwa M
      • Sikazwe I
      • Lambwe N
      • Guffey MB
      • et al.
      Characterization of HIV drug resistance mutations among patients failing first-line antiretroviral therapy from a tertiary referral center in Lusaka, Zambia.
      ,
      • Singh A
      • Sunpath H
      • Green TN
      • Padayachi N
      • Hiramen K
      • Lie Y
      • et al.
      Drug resistance and viral tropism in HIV-1 subtype C-infected patients in kwazulu-natal, South Africa.
      ,
      • Sunpath H
      • Wu B
      • Gordon M
      • Hampton J
      • Johnson B
      • Moosa MYS
      • et al.
      High rate of K65R for antiretroviral therapy-naive patients with subtype C HIV infection failing a tenofovir-containing first-line regimen.
      ,
      • Barth RE
      • Aitken SC
      • Tempelman H
      • Geelen SP
      • van Bussel EM
      • Hoepelman AI
      • et al.
      Accumulation of drug resistance and loss of therapeutic options precede commonly used criteria for treatment failure in HIV-1 subtype-C-infected patients.
      ,
      • Brehm JH
      • Koontz DL
      • Wallis CL
      • Shutt KA
      • Sanne I
      • Wood R
      • et al.
      Frequent emergence of N348I in HIV-1 subtype c reverse transcriptase with failure of initial therapy reduces susceptibility to reverse-transcriptase inhibitors.
      ,
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ,
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ,
      • Hunt GM
      • Dokubo EK
      • Takuva S
      • de Oliveira T
      • Ledwaba J
      • Dube N
      • et al.
      Rates of virological suppression and drug resistance in adult HIV-1-positive patients attending primary healthcare facilities in KwaZulu-Natal, South Africa.
      ,
      • Labhardt ND
      • Bader J
      • Lejone TI
      • Ringera I
      • Hobbins MA
      • Fritz C
      • et al.
      Should viral load thresholds be lowered?.
      ]. Most study populations were from urban settings.
      Fig 2
      Fig. 2PRISMA flow-chart of inclusion of studies and populations in the systematic review.
      Table 1 summarizes the study samples, the numbers of patients included, and their characteristics. Most studies defined virological failure as a viral load >1,000 copies/ml, either as a single measurement or confirmed by a second measure, as recommended by the WHO [
      World Health Organization
      Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection [Internet]. WHO.
      ]. The most commonly used NRTI combinations were FTC/3TC with TDF (45% of patients), d4T (34%), or ZDV (17%). These NRTI drugs were mostly combined with EFV (76%) or NVP (23%). Of 3,915 patients with virological failure, 69% (2,690) had HIV genotype data available. In most study populations (11/19, 58%), the percentage of patients with virological failure with information on the HIV genotype was above 90%. Three large studies are responsible for the gap between the numbers of PLWH with virological failure and the number with genotype data. A national survey in South Africa [
      • Steegen K
      • Bronze M
      • Papathanasopoulos MA
      • van Zyl G
      • Goedhals D
      • Variava E
      • et al.
      HIV-1 antiretroviral drug resistance patterns in patients failing NNRTI-based treatment: results from a national survey in South Africa.
      ] included 1,033 patients but obtained genotypes only for 788 (76.3%), and two other studies in South Africa [
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ,
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ], where only a subsample of the patients failing first-line ART had genotyping of HIV performed. Appendix C provides a detailed description of each study population.
      Table 1Characteristics of the 19 study samples included in the systematic review.
      Screening of virological failure and genotyping
      Total No. of patients with virological failure3915
      No. of patients with genotype data2690 (69%)
      Median (range) number of patients with virological failure102 (31–1033)
      Median (range) number of patients with HIV genotype68 (19–788)
      Median (range) percent patients with HIV genotype among patients with virological failure93% (10–100%)
      Characteristics of patients
      NRTI regimen, proportion of patients
      TDF + FTC/3TC44.7%
       d4T + FTC/3TC34.4%
       ZDV + FTC/3TC17.3%
       Other or missing3.6%
      NNRTI regimen, proportion of patients
       EFV76.4%
       NVP23.1%
       Other or missing0.5%
      Median (range) time on ART (months)29 (5–96)
      Characteristics of studies
      Median (range) study year2012 (2004–2018)
      Definition of virological failure, number of studies (total patients)
       Confirmed >1,000 copies/mL7 (1530)
       Single value >1,000 copies/mL7 (700)
       Value >5000 copies/mL3 (292)
       Single value >400 copies/mL1 (94)
       Confirmed >80 copies/mL1 (74)
      Country, number of studies (total patients)
       Botswana1 (23)
       Lesotho1 (74)
       Malawi1 (237)
       Mozambique1 (61)
       South Africa13 (2206)
       Tanzania1 (21)
       Zambia1 (68)
      Urbanization, number of studies (total patients)
       Urban9 (673)
       Rural6 (1006)
       Both4 (1011)

      3.2 Prevalence of acquired NRTI resistance mutations

      We found a variable prevalence of NRTI DRMs after 2 years of either FTC/3TC and TDF or FTC/3TC and ZDV (Fig. 3A). The prevalence at baseline for NRTI DRMs was low, ranging from 1% to 8.4%. The use of FTC/3TC was associated with high levels of the M184V/I mutation. The prevalence was 75.7% (95% credible interval [CrI] 61.9%–88.9%) after 2 years on FTC/3TC combined with TDF and 72.1% (95% CrI 56.8%–85.9%) on FTC/3TC combined with ZDV. When FTC/3TC was combined with TDF, there was a substantial increase in the K65N/R mutation: from 2.1% (95% CrI 0.7%–3.5%) at baseline to 52% (95% CrI 32.5%–76.8%) after 2 years. The increases over time in the levels of K65N/R and M184V/I on ART are displayed in Fig. 3B and Fig. 3C. In contrast, the prevalence of each of the six TAM mutations after 2 years of FTC/3TC combined with either TDF or ZDV were moderate, ranging from 1.6% to 24.5%. Finally, when the six TAM mutations were combined into a single outcome, the model showed a higher risk of developing any of the six TAMs when FTC/3TC was combined with ZDV rather than TDF: 41.9% (95% CrI 32.1%–56.1%) vs. 24.8% (95% CrI 18.5%–35.5%) after 2 years, suggesting some correlation between the different TAMs (see Section 2.3 of Appendix B).
      Fig 3
      Fig. 3Prevalence of nine NRTI drug resistance mutations by first-line regimen. Panel A: baseline prevalence and prevalence after 2 years of treatment according to NRTI use. Panel B: Prevalence of the K65 mutation over time. Panel C: Prevalence of the M184 mutation over time. Points and vertical lines: median and 95% credibility intervals of baseline prevalence (black), prevalence after 2 years on FTC/3TC + TDF (red) or FTC/3TC + ZDV (blue). Shaded area: 95% credibility interval over time (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

      3.3 Prevalence of acquired NNRTI resistance mutations

      Our model also estimated highly variable prevalences of the seven NNRTI DRMs after 2 years of either EFV- or NVP-based regimens (Fig. 4A). K103N/R/S was the most frequent NNRTI DRM, with a high prevalence after 2 years of EFV-based (57.2%, 95% CrI 40.9%–80.1%) or NVP-based regimens (42.2%, 95% CrI 22.4%–72.3%). Other NNRTI DRMs with prevalence estimates over 20% included V106A/I/M, Y181C/I/V, and G190A/E/R/S. Of note, the model estimated a higher prevalence of the Y181 mutations after the use of NVP (38.7%, 95%CrI 19.3%–64.4%) than with EFV (17.5%, 95%CrI 9.2%–34.6%). The increase over time of ART of K103N/R/S and Y181C/I/V is displayed in Fig. 4B-C, respectively. Appendix B gives estimates of the prevalence of the NRTI/NNRTI DRMs at baseline and after 2 and 3 years of ART.
      Fig 4
      Fig. 4Prevalence of seven NNRTI drug resistance mutations. Panel A: Prevalences after 2 years on either EFV- or NVP-based regimen. Panel B: Prevalence of the K103 mutation over time. Panel C: Prevalence of the Y181 mutation over time. Points and vertical lines: median and 95% credibility intervals of baseline prevalence (black), prevalence after 2 years on EFV (orange) or NVP (green). Shaded area: 95% credibility interval over time. In contrast to , the Panel A does not report baseline prevalence because EFV and NVP select for the same NNRTI DRMs, which prevents the model from providing reliable estimates (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

      3.4 Risk of bias and sensitivity analyses

      We identified five studies at high risk of selection bias [
      • Seu L
      • Mulenga LB
      • Siwingwa M
      • Sikazwe I
      • Lambwe N
      • Guffey MB
      • et al.
      Characterization of HIV drug resistance mutations among patients failing first-line antiretroviral therapy from a tertiary referral center in Lusaka, Zambia.
      ,
      • Barth RE
      • Aitken SC
      • Tempelman H
      • Geelen SP
      • van Bussel EM
      • Hoepelman AI
      • et al.
      Accumulation of drug resistance and loss of therapeutic options precede commonly used criteria for treatment failure in HIV-1 subtype-C-infected patients.
      ,
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ,
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ,
      • Mziray SR
      • Kumburu HH
      • Assey HB
      • Sonda TB
      • Mahande MJ
      • Msuya SE
      • et al.
      Patterns of acquired HIV-1 drug resistance mutations and predictors of virological failure in Moshi, Northern Tanzania.
      ]. Four had proportions of genotyping below 50% [
      • Barth RE
      • Aitken SC
      • Tempelman H
      • Geelen SP
      • van Bussel EM
      • Hoepelman AI
      • et al.
      Accumulation of drug resistance and loss of therapeutic options precede commonly used criteria for treatment failure in HIV-1 subtype-C-infected patients.
      ,
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ,
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ,
      • Mziray SR
      • Kumburu HH
      • Assey HB
      • Sonda TB
      • Mahande MJ
      • Msuya SE
      • et al.
      Patterns of acquired HIV-1 drug resistance mutations and predictors of virological failure in Moshi, Northern Tanzania.
      ]. Three of these, by design, only genotyped a subsample of infections [
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ,
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ,
      • Mziray SR
      • Kumburu HH
      • Assey HB
      • Sonda TB
      • Mahande MJ
      • Msuya SE
      • et al.
      Patterns of acquired HIV-1 drug resistance mutations and predictors of virological failure in Moshi, Northern Tanzania.
      ]. One study [
      • Hoffmann CJ
      • Charalambous S
      • Sim J
      • Ledwaba J
      • Schwikkard G
      • Chaisson RE
      • et al.
      Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
      ] reported that every third patient was systematically sampled for genotyping; however, the number of infections genotyped was substantially lower (10.1% rather than the expected 33.3%, Appendix C). Genotyping in the second study [
      • Hoffmann CJ
      • Ledwaba J
      • Li JF
      • Johnston V
      • Hunt G
      • Fielding KL
      • et al.
      Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
      ] depended on the availability of residual plasma. In the third study [
      • Mziray SR
      • Kumburu HH
      • Assey HB
      • Sonda TB
      • Mahande MJ
      • Msuya SE
      • et al.
      Patterns of acquired HIV-1 drug resistance mutations and predictors of virological failure in Moshi, Northern Tanzania.
      ], only a subsample of the patients, which were said to be representative of the cohort, were selected for sequencing due to financial constraints. Finally, one study [
      • Seu L
      • Mulenga LB
      • Siwingwa M
      • Sikazwe I
      • Lambwe N
      • Guffey MB
      • et al.
      Characterization of HIV drug resistance mutations among patients failing first-line antiretroviral therapy from a tertiary referral center in Lusaka, Zambia.
      ] lacked a clearly defined source population. It was based on referrals of patients with suspected first-line treatment failure at a University Teaching Hospital. Estimating the prevalence of the eight NRTI DRMs and seven NNRTI DRMs excluding the studies at high risk of bias showed that most estimates were very similar to the main analysis but with larger uncertainty (Appendix B). Similarly, removing the studies with missing ART duration (rather than imputing it) or using all the covariates on NRTI use (rather than including a selection step) did not substantially affect the results (see Appendix B and Figures B4 and B5)

      4. Discussion

      This systematic review and Bayesian meta-analysis provide estimates of emerging DRMs in patients failing first-line antiretroviral therapy in Southern Africa. Using a hierarchical structure and adjusting for the use of different drugs, the model estimates the major NRTI/NNRTI mutations while appropriately accounting for between-study heterogeneity. The most frequently acquired mutations among patients with virological failure after 2 years of ART were M184, K65 and K103 (prevalence 52% to 76%). Of note, K65 and M184 confer high-level resistance to FTC/3TC and TDF, the two NRTI backbones that are usually combined in DTG-based regimens. The model also estimated that 28%–44% of patients failing an NNRTI-based regimen had at least one TAM.
      The TenoRes study assessed the levels of NRTI resistance across regions of the world after the failure of NNRTI + FTC/3TC + TDF [
      • Gregson J
      • Tang M
      • Ndembi N
      • Hamers RL
      • Rhee S-Y
      • Marconi VC
      • et al.
      Global epidemiology of drug resistance after failure of WHO recommended first-line regimens for adult HIV-1 infection: a multicentre retrospective cohort study.
      ]. For Southern Africa, it estimated high levels of both M184 and K65 mutations (59% and 56%, respectively), in line with our study. The study found lower levels of M184 and K65 mutations in Europe (34% and 20%) or North America (42% and 22%) [
      • Gregson J
      • Tang M
      • Ndembi N
      • Hamers RL
      • Rhee S-Y
      • Marconi VC
      • et al.
      Global epidemiology of drug resistance after failure of WHO recommended first-line regimens for adult HIV-1 infection: a multicentre retrospective cohort study.
      ]. The authors argued that these differences in NRTI DRM levels might be driven by the higher frequency of viral load monitoring in Europe and North America. Also, in line with our study, the TenoRes study showed that Southern Africa is the only region with a similar prevalence of M184 and K65 mutations. In contrast, a higher prevalence of M184 mutations was observed in all other regions. Subtype C is most prevalent in Southern Africa and in-vitro studies suggest that the K65 mutation might be more likely to emerge in subtype C compared to other subtypes due to a nucleotide template-based mechanism [
      • Coutsinos D
      • Invernizzi CF
      • Xu H
      • Moisi D
      • Oliveira M
      • Brenner BG
      • et al.
      Template usage is responsible for the preferential acquisition of the k65r reverse transcriptase mutation in subtype c variants of human immunodeficiency virus type 1.
      ,
      • Invernizzi CF
      • Coutsinos D
      • Oliveira M
      • Moisi D
      • Brenner BG
      • Wainberg MA.
      Signature nucleotide polymorphisms at positions 64 and 65 in reverse transcriptase favor the selection of the k65r resistance mutation in HIV-1 Subtype C.
      ,
      • White E
      • Smit E
      • Churchill D
      • Collins S
      • Booth C
      • Tostevin A
      • et al.
      No evidence that HIV-1 subtype c infection compromises the efficacy of tenofovir-containing regimens: cohort study in the United Kingdom.
      ]. This may have contributed to the high frequency of K65 in the included studies.
      The high prevalence of the K65N/R and M184I/V mutations means that 39.4% (corresponding to 52.0% times 75.7%) to 52.0% of PLWH failing first-line ART that includes TDF and 3TC might have both, depending on the correlation between the two mutations. In its new guidelines, the South African Department of Health recommends switching patients failing a first-line regimen that includes FTC/3TC and TDF to DTG combined with FTC/3TC and ZDV to have at least one fully active NRTI [
      Republic of South Africa National Department of Health
      2019 ART clinical guidelines for the management of HIV in adults, pregnancy, adolescents, children.
      ]. However, given the toxicity and side effects of ZDV and no fixed-dose combination combining DTG and ZDV, TDF might be preferred over ZDV [
      • Paton NI
      • Musaazi J
      • Kityo C
      • Walimbwa S
      • Hoppe A
      • Balyegisawa A
      • et al.
      Dolutegravir or darunavir in combination with zidovudine or tenofovir to treat HIV.
      ]. In this context, our results show that up to half of these patients will start a DTG-based regimen without a fully active NRTI. The DAWNING trial [
      • Aboud M
      • Brites C
      • Lu H
      • Supparatpinyo K
      • Hercilla L
      • Sievers J
      • et al.
      DTG VS. LPV/R IN SECOND LINE (DAWNING): OUTCOMES BY WHO-RECOMMENDED NRTI BACKBONE | CROI Conference.
      ] showed that DTG-based ART is effective in second-line ART, provided it is combined with at least one fully active NRTI. However, it is uncertain whether a functional DTG monotherapy, that si DTG with no fully active NRTIs, will be effective. Concerns on the efficacy of functional DTG-monotherapy have been raised by studies showing high failure rates with DTG-monotherapy [
      • Wandeler G
      • Buzzi M
      • Anderegg N
      • Sculier D
      • Béguelin C
      • Egger M
      • et al.
      Virologic failure and HIV drug resistance on simplified, dolutegravir-based maintenance therapy: Systematic review and meta-analysis.
      ]. The efficacy of functional DTG-monotherapy might be higher than DTG-monotherapy, as some residual NRTI activity may still exist even in case of resistance. This assumption is supported by the EARNEST and NADIA studies, where the presence of NRTI resistance did not impair virological response to second-line regimen [
      • Paton NI
      • Musaazi J
      • Kityo C
      • Walimbwa S
      • Hoppe A
      • Balyegisawa A
      • et al.
      Dolutegravir or darunavir in combination with zidovudine or tenofovir to treat HIV.
      ,
      • Hakim JG
      • Thompson J
      • Kityo C
      • Hoppe A
      • Kambugu A
      • van Oosterhout JJ
      • et al.
      Lopinavir plus nucleoside reverse-transcriptase inhibitors, lopinavir plus raltegravir, or lopinavir monotherapy for second-line treatment of HIV (EARNEST): 144-week follow-up results from a randomised controlled trial.
      ].
      Our meta-analysis also shows that TAM mutations are present at a moderate level among people failing a NNRTI-based first-line regimen. Interestingly, in several studies, the detection of NRTI resistance and particularly of TAMs before starting second-line ART was associated with better virological suppression, possibly because patients who developed resistance may, on average, have better adherence [
      • Stockdale AJ
      • Saunders MJ
      • Boyd MA
      • Bonnett LJ
      • Johnston V
      • Wandeler G
      • et al.
      Effectiveness of protease inhibitor/nucleos(t)ide reverse transcriptase inhibitor–based second-line antiretroviral therapy for the treatment of human immunodeficiency virus type 1 infection in sub-saharan Africa: a systematic review and meta-analysis.
      ,
      • Boyd MA
      • Moore CL
      • Molina J-M
      • Wood R
      • Madero JS
      • Wolff M
      • et al.
      Baseline HIV-1 resistance, virological outcomes, and emergent resistance in the SECOND-LINE trial: an exploratory analysis.
      ]. The large-scale switch to DTG-based ART should be accompanied by longitudinal, real-world studies of virological failure and drug resistance monitoring.
      We observed a high prevalence of the K103 mutation (57% after 2 years on EFV, 42% on NVP), conferring high resistance to both EFV and NVP. Several Southern African countries recommend adherence support in patients on failing NNRTI-based regimens to achieve viral suppression before switching to DTG. In the case where patients remain unsuppressed after 6 months, these patients should nevertheless switch to DTG. The high level of NNRTI-resistance harbored by these patients questions the efficacy of such a strategy. Indeed, prolonging a failing regimen might increase the risk of accumulation of NRTI DRMs, potentially impairing the efficacy of a future switch to DTG. Among the other NNRTI drug resistance mutations, Y181C/I/V is of particular concern. The estimated prevalence of this mutation was about 17% and 39% after 2 years of EFV and NVP, respectively, reflecting the higher impact of Y181 on NVP, as previously observed [
      • Rhee SY
      • Liu T
      • Ravela J
      • Gonzales MJ
      • Shafer RW.
      Distribution of human immunodeficiency virus type 1 protease and reverse transcriptase mutation patterns in 4,183 persons undergoing genotypic resistance testing.
      ]. This mutation also confers resistance to the newer generation of NNRTIs, such as etravirine and rilpivirine. Finally, the high prevalence of V106 mutation found (46.8% after 2 years on EFV) confirms the higher propensity of subtype-C to acquire this mutation, as observed in in-vitro studies [
      • Brenner B
      • Turner D
      • Oliveira M
      • Moisi D
      • Detorio M
      • Carobene M
      • et al.
      A V106M mutation in HIV-1 clade C viruses exposed to efavirenz confers cross-resistance to non-nucleoside reverse transcriptase inhibitors.
      ]. In-vitro observation could however not explain the lower prevalence of V106 mutation when using NVP (13.4% after 2 years).
      Most previous reviews of HIV drug resistance focused on transmitted drug resistance rather than the resistance that was likely acquired during ART [
      • Ssemwanga D
      • Lihana RW
      • Ugoji C
      • Abimiku A
      • Nkengasong J
      • Dakum P
      • et al.
      Update on HIV-1 acquired and transmitted drug resistance in Africa.
      ], and earlier reviews of ADR on failure rates, CD4-positive lymphocyte counts, and the prevalence of drug resistance mutations overall [
      • Barth RE
      • van der Loeff MFS
      • Schuurman R
      • Hoepelman AI
      • Wensing AM.
      Virological follow-up of adult patients in antiretroviral treatment programmes in sub-Saharan Africa: a systematic review.
      ,
      • Boender TS
      • Sigaloff KCE
      • McMahon JH
      • Kiertiburanakul S
      • Jordan MR
      • Barcarolo J
      • et al.
      Long-term virological outcomes of first-line antiretroviral therapy for HIV-1 in low- and middle-income countries: a systematic review and meta-analysis.
      ], or a subset of mutations [
      • Gregson J
      • Tang M
      • Ndembi N
      • Hamers RL
      • Rhee S-Y
      • Marconi VC
      • et al.
      Global epidemiology of drug resistance after failure of WHO recommended first-line regimens for adult HIV-1 infection: a multicentre retrospective cohort study.
      ,
      WHO
      HIV Drug Resistance Report 2017 [Internet].
      ,
      • Gupta RK
      • Hill A
      • Sawyer AW
      • Cozzi-Lepri A
      • von Wyl V
      • Yerly S
      • et al.
      ]. Our study provides estimates of all relevant ADR mutations according to the drugs used for the whole region of Southern Africa, the region with the highest burden of HIV. To our knowledge, this is the first meta-analysis of this literature that uses a hierarchical structure jointly estimating the respective effects of multiple drugs on the emergence of multiple DRM.
      Our review has several limitations. Many of the included studies were based on patients attending one or few outpatient clinics. We carefully assessed the likely representativeness of the patients undergoing HIV genotyping. Still, even if results reflect the situation in these clinics, they may not be representative of all patients who fail first-line ART in the region. In particular, loss to follow-up might underestimate the proportion of patients with resistance, in the case where loss to follow-up is associated with a higher risk of DRM. Indeed, the between-study heterogeneity was large, and decision-makers should consider the studies most relevant to their settings, as well as the regional data. Many factors may have introduced heterogeneity, including differences between study populations, their levels of adherence, the first-line regimen used, and the time spent on a failing regimen. One main strength of our approach is that the hierarchical structure of the model adjusted for regimen and time spent on a failing regimen, thereby addressing two major sources of heterogeneity, but some residual heterogeneity remains. Of note, in our sensitivity analyses, the likely risk of selection bias did not appear to influence estimates. Finally, we attempted to assess pre-treatment NRTI mutations, but the wide credibility intervals illustrate the difficulty to disentangle pre-treatment from acquired drug NNRTI resistance.
      Although we searched for studies from all over Southern Africa, most of the data included in our analysis were from South Africa. South Africa is one of few countries in the regions that have implemented routine viral load monitoring in patients on ART, and routine viral load monitoring is associated with a reduced probability of drug resistance [
      • Gupta RK
      • Hill A
      • Sawyer AW
      • Cozzi-Lepri A
      • von Wyl V
      • Yerly S
      • et al.
      ]. Therefore, the overrepresentation of South African studies in our meta-analysis might underestimate the frequencies of the different DRMs in the rest of Southern Africa.
      In conclusion, our analysis demonstrates that in Southern Africa, many patients failing first-line ART have DRMs, with important implications for the likely future transmission of drug resistance, the choice of second-line regimen, and the large-scale transition to DTG-based first-line ART. These implications are particularly pertinent to settings where routine viral load and drug resistance testing is not routinely available [
      • Zaniewski E
      • Dao Ostinelli CH
      • Chammartin F
      • Maxwell N
      • Davies M
      • Euvrard J
      • et al.
      Trends in CD4 and viral load testing 2005 to 2018: multi-cohort study of people living with HIV in Southern Africa.
      ].

      Disclaimer

      The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or other funders. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

      Author contributions

      ME, RK, and FG conceived the study, AH and JR developed the Bayesian model. FG and AH searched and collected the literature. FG, AH, and MR performed the screening of publications. AH and JR performed the statistical analysis. ME, FG, and AH analyzed the risk-of-bias. All authors interpreted the results. AH wrote the first draft of the manuscript, which was revised by JR, RK and ME. All authors provided comments on the manuscript.

      Appendix: Supplementary data

      Supplementary materials are available online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. A research replication archive is accessible at https://github.com/anthonyhauser/ADR-meta-analysis.

      Funding

      This study was funded by the National Institute of Health (NIH) under award number U01AI069924. RK is supported by Starting Grant No. 155851 and ME by special project funding (No. 174281) from the Swiss National Science Foundation.

      Acknowledgments

      We are grateful to Doris Kopp and Beatrice Minder for their support in searching and evaluating the literature.

      Appendix. Supplementary materials

      References

      1. UNAIDS. Ending the AIDS epidemic by 2030.

        • Bacheler LT
        • Anton ED
        • Kudish P
        • Baker D
        • Bunville J
        • Krakowski K
        • et al.
        Human immunodeficiency virus type 1 mutations selected in patients failing efavirenz combination therapy.
        Antimicrob Agents Chemother. 2000; 44: 2475-2484
        • Hauser A
        • Kusejko K
        • Johnson LF
        • Wandeler G
        • Riou J
        • Goldstein F
        • et al.
        Bridging the gap between HIV epidemiology and antiretroviral resistance evolution: Modelling the spread of resistance in South Africa.
        PLOS Comput Biol. 2019; 15 (editor) ([Internet][cited 2019 Jul 24] Available from)e1007083
        • Riou J
        • Dupont C
        • Bertagnolio S
        • Gupta RK
        • Kouyos RD
        • Egger M
        • et al.
        Drivers of HIV-1 drug resistance to non-nucleoside reverse-transcriptase inhibitors (NNRTIs) in nine southern African countries: a modelling study.
        BMC Infect Dis. 2021; 21 ([Internet][cited 2021 Dec 20] Available from): 1-10
      2. World Health Organization (WHO). HIV drug resistance report 2017 [Internet]. 2017 [cited 2020 Nov 19]. Available from:https://www.who.int/hiv/pub/drugresistance/hivdr-report-2017/en/.

        • World Health Organization
        Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection [Internet]. WHO.
        World Health Organization, 2016 ([cited 2020 Nov 27]. Available from:)
        • Steegen K
        • Bronze M
        • Papathanasopoulos MA
        • van Zyl G
        • Goedhals D
        • Variava E
        • et al.
        HIV-1 antiretroviral drug resistance patterns in patients failing NNRTI-based treatment: results from a national survey in South Africa.
        J Antimicrob Chemother [Internet]. 2017; 72 ([cited 2018 Feb 15] Available from): 210-219
        • Tang MW
        • Shafer RW.
        HIV-1 antiretroviral resistance: Scientific principles and clinical applications [Internet].
        Drugs. Drugs. 2012; 72 ([cited 2021 May 27]. Available from)
        • Wainberg MA.
        HIV-1 subtype distribution and the problem of drug resistance.
        AIDS. 2004; ([Internet]Available from:): S63-S68
        • Buonaguro L
        • Tornesello ML
        • Buonaguro FM.
        Human immunodeficiency virus type 1 subtype distribution in the worldwide epidemic: pathogenetic and therapeutic implications.
        J Virol. 2007; 81: 10209-10219
        • Coutsinos D
        • Invernizzi CF
        • Xu H
        • Moisi D
        • Oliveira M
        • Brenner BG
        • et al.
        Template usage is responsible for the preferential acquisition of the k65r reverse transcriptase mutation in subtype c variants of human immunodeficiency virus type 1.
        J Virol [Internet]. 2009; 83 ([cited 2020 Nov 27] Available from): 2029-2033
        • Invernizzi CF
        • Coutsinos D
        • Oliveira M
        • Moisi D
        • Brenner BG
        • Wainberg MA.
        Signature nucleotide polymorphisms at positions 64 and 65 in reverse transcriptase favor the selection of the k65r resistance mutation in HIV-1 Subtype C.
        J Infect Dis [Internet]. 2009; 200 ([cited 2020 Nov 27] Available from): 1202-1206
        • White E
        • Smit E
        • Churchill D
        • Collins S
        • Booth C
        • Tostevin A
        • et al.
        No evidence that HIV-1 subtype c infection compromises the efficacy of tenofovir-containing regimens: cohort study in the United Kingdom.
        J Infect Dis [Internet]. 2016; 214 ([cited 2018 Mar 16] Available from): 1302-1308
      3. UNAIDS. AIDSinfo. Number of people living with HIV [Internet]. [cited 2018]. Available from:http://aidsinfo.unaids.org/.

        • Booth A
        • Clarke M
        • Dooley G
        • Ghersi D
        • Moher D
        • Petticrew M
        • et al.
        The nuts and bolts of PROSPERO: an international prospective register of systematic reviews.
        Syst Rev. 2012; 1: 2
        • Shafer RW.
        Rationale and uses of a public HIV drug-resistance database.
        J Infect Dis [Internet]. 2006; 194 ([cited 2018 Feb 15]Available from: https://hivdb.stanford.edu/pages/pdf/Shafer.JID2006.pdf): S51-S58
        • Sayers EW
        • Cavanaugh M
        • Clark K
        • Pruitt KD
        • Schoch CL
        • Sherry ST
        • et al.
        GenBank.
        Nucleic Acids Res [Internet]. 2021; 49 ([cited 2021 Dec 20] Available from): D92-D96
        • Gelman A
        • Simpson D
        • Betancourt M.
        The prior can often only be understood in the context of the likelihood.
        Entropy [Internet]. 2017; 19 ([cited 2021 May 27] Available from): 555
        • El-Khatib Z
        • Ekstrom AM
        • Ledwaba J
        • Mohapi L
        • Laher F
        • Karstaedt A
        • et al.
        Viremia and drug resistance among HIV-1 patients on antiretroviral treatment: a cross-sectional study in Soweto, South Africa.
        Wolters Kluwer Heal | Lippincott Williams Wilkins AIDS Wolters Kluwer Heal [Internet]. 2010; 24 ([cited 2018 Feb 15] Available from): 1679-1687
        • Rupérez M
        • Pou C
        • Maculuve S
        • Cedeño S
        • Luis L
        • Rodríguez J
        • et al.
        Determinants of virological failure and antiretroviral drug resistance in Mozambique.
        J Antimicrob Chemother [Internet]. 2015; 70 ([cited 2018 Feb 15] Available from): 2639-2647
        • Wallis CL
        • Mellors JW
        • Venter WDF
        • Sanne I
        • Stevens W.
        Varied patterns of HIV-1 drug resistance on failing first-line antiretroviral therapy in South Africa.
        JAIDS J Acquir Immune Defic Syndr [Internet]. 2010; 53 ([cited 2020 Sep 14] Available from): 480-484
      4. Williams DR, Rast P, Bürkner P-C. Bayesian meta-analysis with weakly informative prior distributions. 2018 [cited 2021 Dec 4]; Available from:https://psyarxiv.com/7tbrm/.

        • Marconi VC
        • Sunpath H
        • Lu Z
        • Gordon M
        • Koranteng-Apeagyei K
        • Hampton J
        • et al.
        Prevalence of HIV-1 drug resistance after failure of a first highly active antiretroviral therapy regimen in kwazulu natal, South Africa.
        Clin Infect Dis [Internet]. 2008; 46 ([cited 2018 Feb 15] Available from): 1589-1597
        • Novitsky V
        • Wester CW
        • DeGruttola V
        • Bussmann H
        • Gaseitsiwe S
        • Thomas A
        • et al.
        The reverse transcriptase 67N 70R 215Y genotype is the predominant tam pathway associated with virologic failure among hiv type 1c-infected adults treated with ZDV/ddI-Containing HAART in Southern Africa.
        AIDS Res Hum Retroviruses [Internet]. 2007; 23 ([cited 2018 Feb 15] Available from): 868-878
        • Rossouw T
        • Nieuwoudt M
        • Manasa J
        • Malherbe G
        • Lessells R
        • Pillay S
        • et al.
        HIV drug resistance levels in adults failing first-line antiretroviral therapy in an urban and a rural setting in South Africa.
        HIV Med [Internet]. 2017; 18 ([cited 2018 Feb 15] Available from): 104-114
        • Seu L
        • Mulenga LB
        • Siwingwa M
        • Sikazwe I
        • Lambwe N
        • Guffey MB
        • et al.
        Characterization of HIV drug resistance mutations among patients failing first-line antiretroviral therapy from a tertiary referral center in Lusaka, Zambia.
        J Med Virol [Internet]. 2015; 87 ([cited 2018 Feb 15] Available from): 1149-1157
        • Singh A
        • Sunpath H
        • Green TN
        • Padayachi N
        • Hiramen K
        • Lie Y
        • et al.
        Drug resistance and viral tropism in HIV-1 subtype C-infected patients in kwazulu-natal, South Africa.
        JAIDS J Acquir Immune Defic Syndr [Internet]. 2011; 58 ([cited 2018 Feb 15] Available from) (landingpage&an=00126334-201111010-00002): 233-240
        • Sunpath H
        • Wu B
        • Gordon M
        • Hampton J
        • Johnson B
        • Moosa MYS
        • et al.
        High rate of K65R for antiretroviral therapy-naive patients with subtype C HIV infection failing a tenofovir-containing first-line regimen.
        AIDS [Internet]. 2012; 26 ([cited 2018 Feb 15] Available from) (:landingpage&an=00002030-201208240-00011): 1679-1684
        • Barth RE
        • Aitken SC
        • Tempelman H
        • Geelen SP
        • van Bussel EM
        • Hoepelman AI
        • et al.
        Accumulation of drug resistance and loss of therapeutic options precede commonly used criteria for treatment failure in HIV-1 subtype-C-infected patients.
        Antivir Ther [Internet]. 2012; 17 ([cited 2018 Feb 15] Available from): 377-386
        • Brehm JH
        • Koontz DL
        • Wallis CL
        • Shutt KA
        • Sanne I
        • Wood R
        • et al.
        Frequent emergence of N348I in HIV-1 subtype c reverse transcriptase with failure of initial therapy reduces susceptibility to reverse-transcriptase inhibitors.
        Clin Infect Dis [Internet]. 2012; 55 ([cited 2018 Feb 15] Available from): 737-745
        • Hoffmann CJ
        • Charalambous S
        • Sim J
        • Ledwaba J
        • Schwikkard G
        • Chaisson RE
        • et al.
        Viremia, resuppression, and time to resistance in Human Immunodeficiency Virus (HIV) Subtype C during first-line antiretroviral therapy in South Africa.
        Clin Infect Dis [Internet]. 2009; 49 ([cited 2018 Feb 15] Available from): 1928-1935
        • Hoffmann CJ
        • Ledwaba J
        • Li JF
        • Johnston V
        • Hunt G
        • Fielding KL
        • et al.
        Resistance to tenofovir-based regimens during treatment failure of subtype C HIV-1 in South Africa.
        Antivir Ther [Internet]. 2013; 18 ([cited 2018 Feb 15] Available from): 915-920
        • Hunt GM
        • Dokubo EK
        • Takuva S
        • de Oliveira T
        • Ledwaba J
        • Dube N
        • et al.
        Rates of virological suppression and drug resistance in adult HIV-1-positive patients attending primary healthcare facilities in KwaZulu-Natal, South Africa.
        J Antimicrob Chemother [Internet]. 2017; 72 ([cited 2020 Sep 14] Available from): 3141-3148
        • Labhardt ND
        • Bader J
        • Lejone TI
        • Ringera I
        • Hobbins MA
        • Fritz C
        • et al.
        Should viral load thresholds be lowered?.
        Medicine (Baltimore) [Internet]. 2016; 95 ([cited 2018 Feb 15]Available from:) (:landingpage&an=00005792-201607120-00011): e3985
        • Mziray SR
        • Kumburu HH
        • Assey HB
        • Sonda TB
        • Mahande MJ
        • Msuya SE
        • et al.
        Patterns of acquired HIV-1 drug resistance mutations and predictors of virological failure in Moshi, Northern Tanzania.
        PLoS One [Internet]. 2020; 15 (editor) ([cited 2020 Nov 24] Available from https://dx.plos.org/10.1371/journal.pone.0232649)e0232649
        • Gregson J
        • Tang M
        • Ndembi N
        • Hamers RL
        • Rhee S-Y
        • Marconi VC
        • et al.
        Global epidemiology of drug resistance after failure of WHO recommended first-line regimens for adult HIV-1 infection: a multicentre retrospective cohort study.
        Lancet Infect Dis [Internet]. 2016; 16 ([cited 2018 Jan 25] Available from): 565-575
        • Republic of South Africa National Department of Health
        2019 ART clinical guidelines for the management of HIV in adults, pregnancy, adolescents, children.
        Infants and Neonates. 2020;
        • Paton NI
        • Musaazi J
        • Kityo C
        • Walimbwa S
        • Hoppe A
        • Balyegisawa A
        • et al.
        Dolutegravir or darunavir in combination with zidovudine or tenofovir to treat HIV.
        N Engl J Med [Internet]. 2021; 385 ([cited 2021 Dec 4] Available from): 330-341
        • Aboud M
        • Brites C
        • Lu H
        • Supparatpinyo K
        • Hercilla L
        • Sievers J
        • et al.
        DTG VS. LPV/R IN SECOND LINE (DAWNING): OUTCOMES BY WHO-RECOMMENDED NRTI BACKBONE | CROI Conference.
        in: Conference on Retroviruses and Opportunistic Infections [Internet]., Boston, MA2018 ([cited 2018 Apr 27]. Available from)
        • Wandeler G
        • Buzzi M
        • Anderegg N
        • Sculier D
        • Béguelin C
        • Egger M
        • et al.
        Virologic failure and HIV drug resistance on simplified, dolutegravir-based maintenance therapy: Systematic review and meta-analysis.
        F1000Research. 2018; 7: 1359
        • Hakim JG
        • Thompson J
        • Kityo C
        • Hoppe A
        • Kambugu A
        • van Oosterhout JJ
        • et al.
        Lopinavir plus nucleoside reverse-transcriptase inhibitors, lopinavir plus raltegravir, or lopinavir monotherapy for second-line treatment of HIV (EARNEST): 144-week follow-up results from a randomised controlled trial.
        Lancet Infect Dis [Internet]. 2018; 18 ([cited 2020 Aug 12]Available from:): 47-57
        • Stockdale AJ
        • Saunders MJ
        • Boyd MA
        • Bonnett LJ
        • Johnston V
        • Wandeler G
        • et al.
        Effectiveness of protease inhibitor/nucleos(t)ide reverse transcriptase inhibitor–based second-line antiretroviral therapy for the treatment of human immunodeficiency virus type 1 infection in sub-saharan Africa: a systematic review and meta-analysis.
        Clin Infect Dis [Internet]. 2017; ([cited 2018 May 8]; Available from https://academic.oup.com/cid/advance-article/doi/10.1093/cid/cix1108/4767830)
        • Boyd MA
        • Moore CL
        • Molina J-M
        • Wood R
        • Madero JS
        • Wolff M
        • et al.
        Baseline HIV-1 resistance, virological outcomes, and emergent resistance in the SECOND-LINE trial: an exploratory analysis.
        Lancet HIV [Internet]. 2015; 2 ([cited 2018 May 8] Available from): e42-e51
        • Rhee SY
        • Liu T
        • Ravela J
        • Gonzales MJ
        • Shafer RW.
        Distribution of human immunodeficiency virus type 1 protease and reverse transcriptase mutation patterns in 4,183 persons undergoing genotypic resistance testing.
        Antimicrob Agents Chemother [Internet]. 2004; 48 ([cited 2020 Aug 12]Available from: /pmc/articles/PMC478552/?report=abstract): 3122-3126
        • Brenner B
        • Turner D
        • Oliveira M
        • Moisi D
        • Detorio M
        • Carobene M
        • et al.
        A V106M mutation in HIV-1 clade C viruses exposed to efavirenz confers cross-resistance to non-nucleoside reverse transcriptase inhibitors.
        AIDS [Internet]. 2003; 17 ([cited 2021 Dec 4] Available from https://pubmed.ncbi.nlm.nih.gov/12478089/)
        • Ssemwanga D
        • Lihana RW
        • Ugoji C
        • Abimiku A
        • Nkengasong J
        • Dakum P
        • et al.
        Update on HIV-1 acquired and transmitted drug resistance in Africa.
        AIDS Rev [Internet]. 2015; 17 ([cited 2018 Jan 25] Available from): 3-20
        • Barth RE
        • van der Loeff MFS
        • Schuurman R
        • Hoepelman AI
        • Wensing AM.
        Virological follow-up of adult patients in antiretroviral treatment programmes in sub-Saharan Africa: a systematic review.
        Lancet Infect Dis [Internet]. 2010; 10 ([cited 2018 Mar 16] Available from): 155-166
        • Boender TS
        • Sigaloff KCE
        • McMahon JH
        • Kiertiburanakul S
        • Jordan MR
        • Barcarolo J
        • et al.
        Long-term virological outcomes of first-line antiretroviral therapy for HIV-1 in low- and middle-income countries: a systematic review and meta-analysis.
        Clin Infect Dis [Internet]. 2015; 61 ([cited 2018 Mar 16] Available from): 1453-1461
        • WHO
        HIV Drug Resistance Report 2017 [Internet].
        WHO. World Health Organization, 2017 ([cited 2018 Feb 15]Available from:)
        • Gupta RK
        • Hill A
        • Sawyer AW
        • Cozzi-Lepri A
        • von Wyl V
        • Yerly S
        • et al.
        Virological monitoring and resistance to first-line highly active antiretroviral therapy in adults infected with HIV-1 treated under WHO guidelines: a systematic review and meta-analysis. 9. Elsevier, 2009: 409-417 ([Internet]The Lancet Infectious Diseases[cited 2018 Mar 16]Available from:)
        • Zaniewski E
        • Dao Ostinelli CH
        • Chammartin F
        • Maxwell N
        • Davies M
        • Euvrard J
        • et al.
        Trends in CD4 and viral load testing 2005 to 2018: multi-cohort study of people living with HIV in Southern Africa.
        J Int AIDS Soc [Internet]. 2020; 23 ([cited 2020 Nov 23]Available from: https://onlinelibrary.wiley.com/doi/abs/10.1002/jia2.25546)