The University of the Witwatersrand Faculty of Health Sciences Patterns of HIV Resistance in Children Attending an Antiretroviral Clinic in Soweto, South Africa: A Case-Control Study By Zinhle Vilakazi Student Number: 2449467 A Research Report submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in partial fulfillment of the requirements for the degree of Masters of Medicine in Paediatrics November 2023 Type text here Declaration I, Zinhle Vilakazi, declare that this Research Report is my own work. It is being submitted for the Degree of Masters of Medicine in Paediatrics at the University of the Witwatersrand, Johannesburg. It has not been submitted before for any degree or examination at any other University. ________________________________ (Signature of the candidate) ______ day of __ November 2023 in __________________ ii 29 Johannesburg Dedication I dedicate this work to my husband and daughter who have supported and journeyed with me through this degree. iii Presentations and Publications The resistance profile findings of this study were presented at the University of the Witwa- tersrand Paediatric Research Day, 2 December 2021. Title of presentation: “Patterns of HIV resistance in children attending an antiretroviral clinic in Soweto”. The case-control aspect of the study findings were presented at the University of the Witwater- srand Paediatric Research Day, 8 December 2022. Title of presentation: “Factors associated with HIV resistance in children attending an antiretroviral clinic in Soweto, South Africa: a case-control study”. This presentation was awarded best Registrar Research Project prize. iv Abstract Background Exposure to suboptimal serum levels of antiretrovirals (ARVs) places resistance pressure on circulating human immunodeficiency virus (HIV), with consequent emergence of resistance. HIV resistance leads to treatment failure and adverse outcomes. We explored factors associ- ated with the emergence of ARV resistance in children living with HIV (CLWH) attending a treatment clinic in Soweto. Methods We reviewed the clinical and laboratory characteristics, and factors associated with ARV resistance in children aged 0 to 15 years of age that were treated at the clinic from 01 January 2011 through 31 December 2020. The Stanford HIV drug resistance database was used to identify HIV drug resistance mutations and generate resistance profiles. Characteristics of children that underwent drug resistance testing (DRT) were compared to those of children who remained virologically suppressed on fist-line ARVs. Results During the study period, 7,029 children attended the clinic of which 425 (6.0%) underwent DRT (cases) and 953 (13.6%) remained suppressed on first-line ARVs (controls). The resistance dataset included 431 resistance tests that were done in 425 children and adoles- cents that were eligible for the study. Non-nucleoside reverse transcriptase inhibitor (NNRTI) mutations accounted for 50.8% of all mutations, followed by nucleoside reverse transcriptase inhibitor (NRTI) mutations (44.5%) and protease inhibitor (PI) mutations (4.6%). Cases were significantly older at ARV initiation (81.6 vs 45.2 months), had a higher prevalence of ever being diagnosed with tuberculosis (33.2% vs 27.4%), ever being orphaned (57.6% vs 50.6%) and ever experiencing severe acute malnutrition (SAM) (19.8% vs 11.7%). In all v modelling approaches, SAM was consistently associated with ARV resistance (adjusted odds ratios (aOR) ranging from 3.548 (95% confidence interval (CI) 1.979-6.359) to 6.383 (95% CI, 3.811-10.690)). Increasing baseline CD4 percentage was associated with significantly lower adjusted odds of case-status (aOR ranging from 0.971 (95% CI, 0.953-0.989) to 0.951 (95% CI, 0.931-0.972)). Seventeen (5.6%) cases died, compared to two (0.3%) controls; P<0.001. Conclusions Tenuous nutritional status was consistently and significantly associated with the requirement for DRT in this cohort of children and adolescents. Conversely, higher baseline CD4 per- centage was associated with control status. Early ARV initiation, to preserve immunological status, and nutritional support throughout the course of clinic attendance may limit the emergence of drug resistance in CLWH. vi Acknowledgements First and foremost, praises and thanks to the God, the Almighty, for His showers of blessings throughout my research work to complete the research successfully. Immeasurable appreciation and deepest gratitude for the help and support are extended to my supervisors who in one way or another have contributed in making this mmed possible. Dr David P Moore, thank you for giving me the opportunity to this research and providing invaluable guidance throughout.He has taught me the methodology to carry out the research and to present the research works as clearly as possible. His dynamism, vision, patience and motivation have deeply inspired me. Dr Nosisa Sipmabo provided valuable input, insights, and assistance at every stage of the project. Her contributions were critical to the success of this research, and I am deeply grateful for her hard work and dedication. Dr Kim Steegen provided her expertise and insights which were instrumental in shaping the direction and focus of this research and for that I am grateful. I would also like to give special thanks to my husband, daughter and my family as a whole for their continuous support and understanding when undertaking my research and writing my project. Your prayers for me was what sustained me this far. Finally, my thanks go to all the people who have supported me to complete the research work directly or indirectly. vii Table of contents Declaration ii Dedication iii Presentations and Publications iv Abstract v Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi Acknowledgements vii Nomenclature xiv 1 LITERATURE REVIEW 1 1.1 The evolution of ART and PMTCT regimens in South Africa . . . . . . . . . . 2 1.2 Classification of drug resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3 ART class mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.3.1 NNRTI mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.3.2 NRTI mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3.3 Protease inhibitor mutations . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3.4 Integrase strand transfer inhibitor resistance . . . . . . . . . . . . . . . . 8 1.4 Prevalence of and risk factors for ART drug resistance . . . . . . . . . . . . . . 8 1.4.1 Anti-tuberculosis therapy as a risk factor for PI resistance . . . . . . . . 11 1.4.2 Malnutrition as a risk factor for ART resistance . . . . . . . . . . . . . . 11 1.5 Future of ART in children - the threat of resistance . . . . . . . . . . . . . . . . 11 1.6 Justification for this study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 viii 2 MANUSCRIPT IN SUBMISSIBLE FORMAT 13 Abstract 13 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Inclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Sample Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Ethics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Resistance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Characteristics of children that underwent HIV resistance testing . . . . . . . . 19 Parameters at the time of resistance testing . . . . . . . . . . . . . . . . . . . . . 19 Resistance mutations detected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Factors associated with mutations in children and adolescents that underwent HIV resistance testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Parameters at the time of last follow-up in children and adolescents that un- derwent HIV resistance testing . . . . . . . . . . . . . . . . . . . . . . . . 25 Predictors of ART class mutations . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Case-control analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Case-Control Analysis of Factors Associated with HIV Resistance . . . . . . . 25 Multivariable analysis of factors associated with case-status . . . . . . . . . . . 28 Survival analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 ix Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 APPENDIX 1: Supplementary Tables 43 APPENDIX 2: Study Protocol 69 APPENDIX 3: Ethics Approval Certificates 90 x List of Figures 2.1 The ’top 5’ mutations per ART class detected in children and adolescents attending the Harriet Shezi Children’s Clinic, 2011 through 2020 . . . . . . . . 24 2.2 Kaplan-Meier survival curves for cases and controls . . . . . . . . . . . . . . . . 31 2.3 Distributional balance for ART start age and ART start age for NRTI resis- tance mutation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 2.4 Distributional balance for ART start age and ART start age for NNRTI resis- tance mutation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 2.5 Distributional balance for ART start age and ART start age for PI resistance mutation analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 2.6 Distributional balance for ART start age and ART start age in each of the case-control data sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 xi List of Tables 2.1 Characteristics of children that underwent HIV resistance testing, stratified by duration on ART at the time of resistance testing . . . . . . . . . . . . . . . 21 2.2 HIV Clinic Database for Cases (children that underwent resistance testing) and Controls (children that did not have resistance testing done) . . . . . . . . 27 2.3 Multivariable Logistic Regression Outputs: Unmatched and Matched Case- Control Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.4 Characteristics of children that underwent HIV resistance testing, stratified by age at ART initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.5 Characteristics of children that underwent HIV resistance testing, stratified by regimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 2.6 Matching for case-control analysis of risk factors associated with NRTI muta- tions in children and adolescents that underwent ART resistance testing . . . 46 2.7 Multivariable Logistic Regression Outputs comparing cases with and without NRTI resistance mutations: Unmatched and Matched Case-Control Models . . 48 2.8 Matching for case-control analysis of risk factors associated with NNRTI mu- tations in children and adolescents that underwent ART resistance testing . . 50 2.9 Multivariable Logistic Regression Outputs comparing cases with and without NNRTI resistance mutations: Unmatched and Matched Case-Control Models 52 2.10 Matching for case-control analysis of risk factors associated with PI mutations in children and adolescents that underwent ART resistance testing . . . . . . . 54 2.11 Multivariable Logistic Regression Outputs comparing cases with and without PI resistance mutations: Unmatched and Matched Case-Control Models . . . . 56 2.12 Case-Control matching for analyses exploring factors associated with resis- tance testing in children and adolescents attending the ART clinic during the study period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 2.13 Matched case-control data using coarsened exact matching (nearest neighbour matching without replacement within each stratum) . . . . . . . . . . . . . . . . 60 xii 2.14 Univariate outputs for coarsened exact matched case-control data (without replacement) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.15 Matched case-control data using coarsened exact matching (nearest neighbour matching with replacement within each stratum) . . . . . . . . . . . . . . . . . . 62 2.16 Univariate outputs for coarsened exact matched case-control data (with re- placement) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2.17 Matched case-control data using nearest neighbour matching . . . . . . . . . . 64 2.18 Univariate outputs for nearest neighbour matched case-control data . . . . . . 65 2.19 Matched case-control data using full matching . . . . . . . . . . . . . . . . . . . 66 2.20 Univariate outputs for full matching of case-control data . . . . . . . . . . . . . 67 2.21 Characteristics of children with 3 or more PI mutations, compared to those with 1 or 2 PI mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 xiii Nomenclature 3TC lamivudine ADR acquired drug resistance ART antiretroviral therapy CHBAH Chris Hani Baragwanath Academic Hospital CLWH children living with HIV DRMs drug resistance mutations EFV efavirenz FTC emtricitabine HIV human immunodeficiency virus type-1 HSCC Harriet Shezi Childrens Clinic IQR interquartile range LMIC low-middle income country NHLS National Health Laboratory Service NNRTI non-nucleoside reverse transcriptase inhibitors NRTI nucleoside reverse transcriptase inhibitors NVP nevirapine PDR pre-treatment drug resistance PI protease inhibitor PLWH person living with HIV PMTCT prevention of mother-to-child transmission SD standard deviation TDR transmitted drug resistance VL viral load WHO World Health Organization xiv 1 LITERATURE REVIEW Human immunodeficiency virus (HIV) infection remains a major global epidemic. UNICEF estimates that of the 39.0 million people living with HIV (PLWH) worldwide at the end of 2022, 1.5 million of them children between the ages of 0-141,2. In South Africa it is estimated that 7.7 million people were living with HIV in 20211, with almost half a million of them being children between the ages of 0-14 years3. South Africa currently has the largest antiretroviral therapy (ART) treatment programme in the world since its introduction in 2004, with an estimated 5.7 million people receiving ART1. The discovery and roll out of ART has improved the quality of life and life expectancy of PLWH substantially4. As with any other antimicrobials, ART used in unfavourable conditions can result in the development of resistance5. Surveys conducted by the World Health Organization (WHO) in 10 countries in sub-Saharan Africa between 2012 to 2020, found that up to 10% of adults starting ART have drug resistance mutations (DRMs) to the non-nucleoside reverse tran- scriptase inhibitor (NNRTI) drug class, and nearly half of infants newly diagnosed with HIV have NNRTI resistant virus before initiating treatment2. The prevalence of drug-resistant HIV is high in children under 18 months of age and newly diagnosed with HIV2. The WHO defines antiretroviral drug resistance as mutations in the genetic structure of HIV that affects the ability of a specific drug or combination of drugs to block replication of HIV6. The goals of ART are to inhibit viral replication, halt the progression of disease to acquired immunodeficiency syndrome (AIDS), and to allow for partial restoration of the im- mune system7. When HIV replicates in the presence of ART, genomic mutations occur which may alter the viral proteins that are targeted by antiretroviral drugs, which leads to different types of mutations, some of which may lead to treatment failure7. HIV is particularly prone to developing resistance for a number of reasons, including high levels of circulating virus and rapid and error prone reverse transcription of the genome. The process of reverse transcrip- tion is inaccurate, due to the absence of any enzymatic proofreading activity7. This leads to each genome having at least one mutation which results in diverse HIV populations, some 1 of which have a survival advantage especially when conferring resistance to antiretrovirals7. All current antiretrovirals used in the treatment of HIV, including newer classes, are at risk of becoming partly or fully inactive because of the emergence of drug-resistant strains6. 1.1 The evolution of ART and PMTCT regimens in South Africa The development of antiretroviral treatment began in the 1980s, after the initial isolation of HIV, with nucleoside reverse transcription inhibitors (NRTIs) being the first class of an- tiretrovirals used in the treatment of HIV/AIDS8. NRTIs are not without side effects, and have limited durability. Therefore, other classes of antiretroviral drugs such as NNRTIs and proteases inhibitors (PIs), which had different mechanisms of action against HIV, were de- veloped and used in combination therapy in the 1990s for more effective treatment8. The combination of different drug classes proved more effective at suppressing viral replication, and was recommended by the WHO in both adults and children as the pandemic of HIV continued9. A key strategy that was identified in controlling the paediatric HIV pandemic was the prevention of mother to child transmission (PMTCT) through the use of ART9. In South Africa, there was no ART or prophylaxis available in the public health sector prior to 200210. With the introduction of ART in South Africa, the PMTCT programme was introduced. The PMTCT programme aimed to reduce the transmission of HIV from mothers to their infants and young children, either in utero, through the process of delivery of the newborn, or through breastfeeding. Different antiretroviral prophylactic regimens have been implemented and modified over the years depending on the maternal ART status and the exposure risk to the infant11–13. The PMTCT programme has evolved in South Africa from its introduction. In 2002, PMTCT consisted of a single-dose of nevirapine (NVP) administered intrapartum to the mother, and postpartum for the infant. The introduction of Option A, in 2008, saw the addition of zi- dovudine (AZT) monotherapy to pregnant women from 28 weeks and AZT for the infant, for either 7 or 28 days, depending on the duration of AZT in pregnancy of their mother10,12. In 2 the revision of Option A in 2010, the maternal PMTCT regimen consisted of AZT from 14 weeks’ gestation, intrapartum single-dose NVP, 3 hourly AZT during labour, and a postpar- tum single dose of tenofovir (TDF) and emtracitabine (FTC)12. The infant PMTCT regimen was changed to NVP for 6 weeks or longer for infants that were breastfeeding whose mothers were not on lifelong ART10,12. In 2013, Option B was introduced into the PMTCT programme. In option B, all pregnant women with HIV were started on triple therapy ART for the duration of pregnancy, with ART stopped postpartum for those who did not meet the life-long ART eligibility criteria14. Eligibility criteria for lifelong ART were amended to a CD4 count of <350 cells/ţL or WHO clinical stage 3 or 4 disease for pregnant women10,14. Option B was modified in 2015 to option B+, which makes life-long ART accessible to all pregnant women living with HIV, regardless of clinical or immunological staging13,14. Option B+ also saw a change to the infant prophylactic regimen, with NVP being given for 6 weeks or extended to 12 weeks if there was high risk maternal HIV exposure10,14. In 2019, AZT and NVP were given for 6 weeks for all infants that were considered high risk, with NVP extended for a further 6 weeks for breastfed infants15. ART in paediatrics has evolved from NRTI mono-therapy to dual therapy and finally the triple therapy4,16. In the early era of ART, an NNRTI backbone regimen was the recommended ART regimen for children living with HIV (CLWH)4. However, NNRTIs have fallen out of favour due to high levels of resistance to NNRTIs secondary to their use in PMTCT and their low genetic barrier to developing resistance mutations17–19. A meta-analysis found that pre- treatment drug resistance (PDR) prevalence in children in sub-Saharan Africa is very high in PMTCT-exposed and PMTCT-unexposed children, with the prevalence of PDR increasing markedly over time19. The pooled PDR prevalence in African children is 4–15 times higher than that reported in adults2,19. PI-based ART has remained an important component of management of CLWH in the de- veloping world. From 2010, the WHO recommend paediatric first-line ART regimens that 3 include a boosted PI, such as Lopinavir/ritonavir (LPV/r), combined with two NRTIs for children under 3 years of age2. The CHER trial demonstrated that early treatment of infants with boosted PI-based regimens reduces early infant mortality by 76% and HIV progression by 75%20. As of 2019, the paediatric first-line ART regimen has undergone change with the inclusion of integrase inhibitors such as dolutegravir (DTG)4. The South African first line ART regimen now recommends AZT, 3TC and NVP for neonates that are less than 4 weeks and under 3kgs. The infant can then be switched to ABC + 3TC and DTG at 4 weeks and if their weight is above 3kgs. The first line treatment for children that are more than 30 kgs and are above the age 10 years is TDF +3TC +DTG (TLD)21. The PTMCT regimen that an infant receives is dependent upon the maternal viral load which classifies the infant as either high-risk or low-risk. Infants with low-risk exposure receive NVP for 6 weeks. Those with high-risk exposure receive AZT for 6 weeks and NVP for a minimum of 12 weeks21. 1.2 Classification of drug resistance HIV drug resistance is classified into three categories by the WHO: 1) acquired HIV drug resistance (ADR); 2) transmitted HIV drug resistance (TDR); and 3) PDR (pre-treatment HIV drug resistance). ADR (acquired resistance) is the most common type of drug resis- tance and occurs when HIV replicates in the presence of ART, when drug levels are too low to block replication but high enough to exert a positive selection pressure on the virus7,22. TDR (transmitted resistance) is detected among ART naïve persons with no history of an- tiretroviral drug exposure, and occurs through infection with HIV that has drug resistance mutations (DRMs)23. PDR (pre-treatment resistance) refers to resistance that is detected among ART naïve persons at the time of starting ART, or in persons with previous antiretro- viral drug exposure initiating or re-initiating first-line ART23. PDR is clinically important as it is associated with a poor response to first-line therapy and further accumulation of drug resistance mutations19. 4 1.3 ART class mutations A multi-country analysis from sub-Saharan African found that 785 (54%) of 1450 CLWH (median age 18 months) had PDR to 1 or more antiretroviral agents prior to treatment initiation18. The rate at which drug resistance occurs across the different classes of ART is dependent upon the drugs’ barriers to resistance. First generation NNRTIs have a low barrier to resistance and the accumulation of resistance-associated mutations is rapid, often occurring within 3 months of virological failure24. Second-generation NNRTIs, such as etravirine and rilpivirine, have a higher barrier to resistance and confer only partial cross-resistance to nevirapine and efavirenz24,25. PIs have much a higher barrier to resistance and are more robust. A study which investigated the accumulation of HIV-1 drug resistance after continued viro- logical failure on first line ART in adults and children in sub-Saharan Africa, found that at first virological failure, DRMs were detected in 87% of participants, new DRMs accumulated with an average rate of 1.45 (standard deviation (SD) 2.07) DRM per year, with 0.62 (SD 1.11) NNRTI DRMs and 0.84 (SD 1.38) NRTI DRMs per year, respectively26. Predicted antiretroviral susceptibility declined significantly after continued virological failure for all re- verse transcriptase inhibitors (P<0.001 in all analyses)26. Acquired drug resistance patterns were similar in adults and children26. DRMs were observed in 48% (205/430) Nigerian CLWH under 18 months of age, with resis- tance to NNRTI highest (45%), followed by resistance to NRTIs (22%) and PIs (2%)27. The prevalence of mutations in different regions of the globe is dependent upon the exposure of patients to different ART classes28. A systematic review of 30 studies from LMICs in sub-Saharan African, Asia and South America, found that drug-class specific resistance rates were 88% for NNRTIs, 80% for NRTIs, and 54% for PIs, with these resistance rates similar to those reported for European children29. The K103N and M184V mutations, which confer high-level resistance to NVP and EFV and lamivudine (3TC) and emtricitabine (FTC), respectively, are more frequent in South Africa 5 than in other countries. This is likely because of earlier and/or programmatic use of EFV and 3TC or FTC, and possibly by the earlier introduction of Option B+14,30. The prevalent subtype of HIV-1 has a contributory factor to the type of mutations that are seen in different regions. The NNRTI mutation V106M is found most frequently in settings in which HIV-1 subtype C predominates, whereas V106A is more frequently seen in regions with subtype B predominance28. V106A in subtype B is mainly associated with resistance to NVP but not EFV, unlike V106M which confers broad NNRTI class cross-resistance28. The prevalence of the L65A mutation, which also confers resistance to 3TC, is similar regardless of HIV-1 subtype, although this mutation develops more rapidly in populations infected with subtype C28. 1.3.1 NNRTI mutations Due to their low barrier to resistance, NNRTI DRMs frequently occur first in the face of virological failure, as demonstrated in a study of accumulation of DRMs in sub-Saharan adults and children on first-line ART30. In one study, more than one DRM was detected in 87.4% of ART-experienced adult and paediatric patients in sub-Saharan Africa, and among these, 83.2% harboured NNRTI resistance mutations24. In South Africa, the most frequently detected NNRTI mutations include K103N (38.7%), G190A (21.8%), Y181C (20.2%), V106M (8.4%), K101E (8.4%), E138 (7.6%) and V108I (7.6%). The K103N and G190A mutations are linked with NVP exposure and EFV resistance, while Y181C is associated with NNRTI class resistance24. Second generation NNRTIs, such as etravirine and rilpivirine, are also compromised in the presence of NNRTI resistance due to high (20-60%) cross resistance rates5,24,31. These high rates of cross resistance limit the use of second generation NNRTIs in patients failing first-line ART regimes that contained an NNRTI5,24,31. 6 1.3.2 NRTI mutations A survey of five African countries (South Africa, Zimbabwe, Uganda, Swaziland and Mozam- bique) found that NRTI resistance prevalence is considerably lower than NNRTI resistance (8.9% compared to 53.0%)24. NRTI resistance mutations were mostly determined by the stavudine (D4T) and 3TC NRTI backbone which was commonly used at the time24,32. The most frequent NRTI mutations detected were M184V (69.7%), any thymidine analogue muta- tion (9.2%), K65R (5.9%) and K70R (5.0%)24. The prevalence of K65R mutation, associated with resistance to tenofovir (TDF), was low in Swaziland, Uganda and Mozambique while in South Africa and Zimbabwe, position 65 mutations were more common reflecting the greater use of TDF in maternal regimens in those two countries24. M184V mutation is the most frequently encountered mutation to the NRTI class, and compro- mises the susceptibility of FTC and 3TC, however continued treatment with these NRTIs is not contraindicated as M184V impairs viral replication fitness and increases TDF activity24,25. Thymidine analogue mutations (TAMs), which reduce NRTI susceptibility by facilitating primer unblocking, are more prevalent in second-line ART failure25,31. This is likely due the longer duration of time on any failing therapy, whether it be first-line or second-line. Of note is that the Q151M pan-nucleoside mutation is infrequently detected in paediatric patients24. 1.3.3 Protease inhibitor mutations The prevalence of resistance to PIs in infants and young children newly diagnosed with HIV in sub-Saharan countries is low (<3%), probably due to low rates of maternal PI use5. A large, historical cohort of children in Soweto conducted over 10 years, investigating virologic failure on first-line LPV/r-based therapy, reported that only 8/75 (10.7%) children failing LPV/r- based first-line ART had significant PI DRMs, including two with intermediate resistance to darunavir (DRV)33. Among 63/75 (84%) children remaining on LPV/r-based therapy, 32/63 (51%) achieved virologic suppression, despite two of these children having significant LPV mutations33. The findings of that study attested to the robustness of PIs, even in the face of 7 prolonged virological failure. A 2014 study conducted in South Africa of a cohort of 16 children (median age 5 years) who were failing an LPV/r regimen, found a major PI mutation (V82A) in only one child, with the most common pattern for children failing LPV/r being an isolated M184V (i.e. NRTI) mutation34. In Southern Africa, LPV/r and full-dose ritonavir are widely used in paediatric regimens. The most frequently reported mutations from this region were V82A, I54V, and M46I, which are selected by LPV/r28. In Cape Town, the PI mutations M46I, I54L, L/V82A/F and L90M were observed in a group of children failing full-dose first-line ritonavir-based ART24. 1.3.4 Integrase strand transfer inhibitor resistance Since 2018, dolutegravir (DTG) has been recommended by the WHO to be used as first line treatment, with close monitoring for emergence of integrase inhibitor (INSTI) resistance in CLWH failing DTG-based regimens22. South Africa introduced DTG with 2 NNRTIs as the first line regimen for CLWH in 201921. Although INSTI resistance has been rare there have now been multiple case reports of INSTI-associated resistance35. The most frequently detected INSTI resistance mutations included T66A/I/K, E92G/Q, G118R, F121Y, E138A/K/T, G140A/C/S, Y143C/H/R/S, S147G, Q148H/R/K, N155H, S230R and R263K35. 1.4 Prevalence of and risk factors for ART drug resistance The WHO reports that in several low-middle income countries (LMICs), more than 1 in 10 adults starting ART are infected with HIV that is already resistant to efavirenz (EFV) or nevirapine (NVP). In these groups, women were twice as likely as men to carry PDR strains of HIV. This challenges the elimination of mother-to-child transmission of HIV, and impacts on maternal and child health outcomes6. Furthermore, 25% persons initiating first-line ART 8 may have previous exposure to antiretrovirals, either in the form of PMTCT or treatment interruption, and are three times more likely to have PDR to EFV or NVP compared to those with no prior antiretroviral exposure22. In surveys conducted in nine sub-Saharan African countries by the WHO in 2012 through 2018, the prevalence of PDR among children under 18 months with newly diagnosed HIV was high: over half of the infants carried strains of HIV that were resistant to EFV and/or NVP23. Furthermore, a multi-country analysis reported one in two children newly diagnosed with HIV under the age of 18 months as being infected with an HIV strain resistant to NNRTIs, and therefore at high risk for suboptimal virological response to treatment if initiated on an NNRTI-based regimen18. The prevalence of PDR to NRTIs may also exceed 10% in newly diagnosed infants in some countries7. Data suggests that PDR is not the only problem affecting children accessing ART. Viral suppression is often not achieved as a consequence of poor adherence to therapy, which increases the risk of accumulating resistance. A systematic review of resistance data in children from LMICs found that 90% of those failing first-line regimens had at least one detectable resistance mutation29. Children present with more HIV drug resistance than adults, and have a 2-fold higher risk of virological failure (VF) compared to adults after 5 years on ART28. A Meta-analysis of 72 studies that reported on 51,347 CLWH in LMICs, revealed only 73% attained a VL <1000 copies/mL within 12 months of ART initiation26. The probability of virological suppression and long-term treatment success is lower in chil- dren than in adults. Younger children typically have poorer virological responses than do older children, which is associated with high viral loads before treatment and the risk of sub-therapeutic drug concentrations due to limited paediatric formulations, variable phar- macokinetics, and changes in body weight because of growth2,29. An important determinant of poor viral suppression is the continued use of suboptimal ART regimens in children36. Other contributing factors for suboptimal ART use in children include unpalatable formu- lations, high cost of PIs and the scarcity of paediatric-trained clinicians which contributes 9 to the continued, prolonged use of unfavourable regimes36. Established risk factors for viro- logical failure in children include male gender, age, concurrent tuberculosis (TB) treatment, and maternal orphanhood28. Multi-class antiretroviral resistance has also been associated with PMTCT exposure, sub-optimal mono- and dual-therapy ART regimens, lack of child- friendly formulations, defaulting ART and limited availability of approved antiretrovirals for children19. PMTCT exposure is considered a major risk factor for resistance as it has resulted in an increase in HIV drug resistance among infants and children who acquire HIV infection18. A multi-country analysis of five sub-Saharan countries identified that children who had received neonatal prophylaxis independent of maternal prophylaxis were nearly twice as likely to have drug-resistant HIV compared to children who did not receive neonatal prophylaxis, and prevalence of resistance to NNRTIs decreased with increasing age (61.9% in children aged <3 months, but 25.4% in children aged 12–18 months)18. Furthermore, a study performed in Uganda found that HIV drug resistance occurred in 35.7% of children with a history of PMTCT exposure, compared to 15.6% in children with unknown PMTCT history and 7.7% among children who were ART naïve37. Another Ugandan study Of 701 children found that 34% of the patients failed first line ART, with median time to first line ART failure of 26.4 months (interquartile range (IQR), 18.9 – 39.1 months). Factors associated with treatment failure were poor adherence (odds ratio (OR) 10.0, 95% confidence interval (CI): 6.4–16.7), exposure to single dose NVP as part of PMTCT (OR 4.2, 95% CI: 1.8-9.4) and receipt of a NVP-containing regimen (OR 2.2, 95% CI: 1.4-3.6)38. A South African study analysed factors associated with development of antiretroviral resis- tance mutations in 65 young children (median age 16.8 months) failing PI-based ART, and found major PI mutations in 49% of children. Risk factors associated with these mutations were low weight-for-age and height-for-age, longer duration of PI-containing regimens, viro- logical failure, and unsuppressed HIV viral load at 12 months of ART39. Anti-tuberculosis (TB) treatment at the time of ART initiation and use of ritonavir as single PI were also significantly associated with the emergence of PI-resistance39. On multivariate analysis, cu- 10 mulative months on PI-based ART and use of ritonavir as single PI were independently associated with the emergence of PI-resistance39. 1.4.1 Anti-tuberculosis therapy as a risk factor for PI resistance Tuberculosis co-infection is a major risk factor for PI DRMs as a result of the drug-drug interaction between LPV/r and rifampicin which is known to reduce plasma concentrations of LPV/r which may lead to virological failure and development of PI DRMs4. 1.4.2 Malnutrition as a risk factor for ART resistance Malnutrition has been shown to contribute to the development of resistance. In a study from Mozambique, reduced peak weight-for-age was significantly associated with ART resistance40. In the same study, immunosuppression of any form and length of time on ART were associated with the development of antiretroviral resistance40. 1.5 Future of ART in children - the threat of resistance Based on the high prevalence of PDR in children18,19, the WHO currently recommends that all CLWH should have a genotyping resistance test prior to initiation of ART, and if requiring a change to second-line ART to tailor the individual ART regimes22. Genotyping is frequently done in high-income countries but in LMICs such as South Africa, routine resistance testing is rarely performed due to the cost and complexities of resistance genotyping, which limits testing to children that are failing second-line regimens23. Resource limited regions, where the burden of HIV is most prevalent, rely on large scale treatment programmes with standardised empiric ART regimes5; however, surveillance of drug resistance becomes important in resource limited settings to ensure that empiric regimes are appropriate. In South Africa, the current National Department of Health guidelines recommend resistance testing in children when there is virological failure after more than 2 years of being on a PI- or an integrase inhibitor based regimen21. HIV Drug resistance poses a major public health challenge and impacts individual patient 11 care as it reduces the number of available and effective treatment options and threatens the success of both prevention and treatment programs such as the UNAIDS 95-95-95 global treatment targets1,41. Understanding HIV drug resistance in the paediatric population is particularly important because children have higher viral load set points, with more rapid disease progression compared to adults7. 1.6 Justification for this study Understanding HIV resistance mutations in a population impacts on decisions around the choice of ART regimens and ensures sustainability and durability of these regimens. Despite South Africa having the world’s largest ART programme17, there is a paucity of data on the prevalence and types of resistance mutations in children, and a need for more studies in this area. The aim of this study is to describe the resistance mutations of children attending a paediatric HIV clinic in a tertiary hospital. Furthermore, this study aims to determine the risk factors associated with certain mutations and the clinical outcomes of those with mutations compared to those without mutations. 12 2 MANUSCRIPT IN SUBMISSIBLE FORMAT Abstract Background Exposure to suboptimal serum levels of antiretrovirals (ARVs) places resistance pressure on circulating human immunodeficiency virus (HIV), with consequent emergence of resistance. HIV resistance leads to treatment failure and adverse outcomes. We explored factors associ- ated with the emergence of ARV resistance in children living with HIV (CLWH) attending a treatment clinic in Soweto, Johannesburg, South Africa. Methods We reviewed the clinical and laboratory characteristics, and factors associated with ARV resistance in children aged 0 to 15 years of age that were treated at the clinic from 01 January 2011 through 31 December 2020. The Stanford HIV drug resistance database was used to identify HIV drug resistance mutations and generate resistance profiles. Characteristics of children that underwent drug resistance testing (DRT) were compared to those of children who remained virologically suppressed on fist-line ARVs. Results During the study period, 7,029 children attended the clinic of whom 425 (6.0%) underwent DRT (cases) and 953 (13.6%) remained suppressed on first-line ARVs (controls). The resistance dataset included 431 resistance tests that were done in 425 children and adoles- cents that were eligible for the study. Non-nucleoside reverse transcriptase inhibitor (NNRTI) mutations accounted for 50.8% of all mutations, followed by nucleoside reverse transcriptase inhibitor (NRTI) mutations (44.5%) and protease inhibitor (PI) mutations (4.6%). Cases were significantly older at ARV initiation (81.6 vs 45.2 months), had a higher prevalence 13 of ever being diagnosed with tuberculosis (33.2% vs 27.4%), ever being orphaned (57.6% vs 50.6%) and ever experiencing severe acute malnutrition (SAM) (19.8% vs 11.7%). In all modelling approaches, SAM was consistently associated with ARV resistance (adjusted odds ratios (aOR) ranging from 3.548 (95% confidence interval (CI) 1.979-6.359) to 6.383 (95% CI, 3.811-10.690)). Increasing baseline CD4 percentage was associated with significantly lower adjusted odds of case-status (aOR ranging from 0.971 (95% CI, 0.953-0.989) to 0.951 (95% CI, 0.931-0.972)). Seventeen (5.6%) cases died, compared to two (0.3%) controls; P<0.001. Conclusions Tenuous nutritional status was consistently and significantly associated with the requirement for DRT in this cohort of children and adolescents. Conversely, higher baseline CD4 per- centage was associated with control status. Early ARV initiation, to preserve immunological status, and nutritional support throughout the course of clinic attendance may limit the emergence of drug resistance in CLWH. 14 Introduction Human immunodeficiency virus (HIV) infection remains a major global epidemic. South Africa currently has the largest antiretroviral therapy (ART) programme, with 7.7 million people living with HIV, and half a million children living with HIV (CLWH)6,42. ART used in unfavourable conditions can result in the development of drug resistant mutations (DRMs)6. All current antiretroviral drugs, including newer classes, are at risk of becoming partly or fully inactive because of the emergence of drug-resistant strains6. Surveys conducted by the World Health Organization (WHO) found up to 10% of adults starting HIV treatment may have drug resistance to the non-nucleoside reverse transcriptase inhibitor (NNRTI) drug class2. Of even greater concern is that one in two children newly diagnosed with HIV infection under the age of 18 months are infected with a strain that is resistant to NNRTIs, and are therefore at high risk for suboptimal virological response to treatment if initiated on an NNRTI-based regimen18. Children present with more HIV drug resistance than adults, and have a 2-fold higher risk of virological failure (VF) compared to adults after 5 years on ART31. The probability of virological suppression and long-term treatment success is lower in children than in adults31. A critical determinant of poor viral suppression in children is the continued use of suboptimal ART regimens27. Factors contributing to suboptimal ART use in children include unpalatable formulations, high cost of protease inhibitors (PIs) and the scarcity of paediatric-trained clin- icians which contributes to the continued, prolonged use of unfavourable regimens27. A study from Pretoria, South Africa, reported male gender, age, concurrent tuberculosis (TB) treat- ment, and maternal orphanhood as being independently associated with virological failure28. Furthermore, immunosuppression of any form and longer duration of ART are significantly associated with the development of drug resistance19. Multi-class antiretroviral resistance in the paediatric population has also been associated with prevention of mother-to-child transmission (PMTCT) exposure, suboptimal mono- and dual- therapy ART regimens, default off ART, and limited availability of approved antiretrovirals 15 for children5. Furthermore, malnutrition is a risk factor for the development of resistance39. In this study, we aimed to explore the prevalence of antiretroviral drug resistance, patterns of resistance, and risk factors for resistance in a cohort of children and adolescents attending a large public sector ART clinic in Johannesburg, South Africa. Methods We present a retrospective case-control study of paediatric patients under the age of 15 years who were tested for HIV resistance from 01 January 2011 to 31 December 2020. Inclusion criteria Inclusion criteria for cases (children who were tested for HIV resistance) were age <15 years, with any history of ART exposure prior to submission of the resistance test sample. For analysis of potential risk factors associated with ART class mutations among children who underwent HIV resistance testing, children and adolescents that had resistance testing but in whom the ART class mutations were not detected, were assigned to the control group. For analysis of potential risk factors associated with virologic resistance, children and adolescents that were virologically suppressed on first-line ART and therefore never underwent resistance testing were assigned as being the control group. Children and adolescents failing third line ART, and those without HIV resistance testing although they were virologically unsuppressed, were excluded from the analysis. Data sources The Harriet Shezi Children’s Clinic is located at Chris Hani Baragwanath Academic Hospital (CHBAH), a public sector academic hospital in Soweto, South Africa. The clinic was estab- lished in 2000 and has served over 9,000 CLWH (personal communication, Dr Sipambo). From 2013 onwards, the clinic has focused on the outpatient management of complicated CLWH, serving as a referral site for surrounding clinics and hospitals. Prior to 2018, HIV drug resistance testing was performed for all children failing first line (NNRTI- or PI-based), 16 second line, or third line ART regimens. From September 2018 onwards, drug resistance testing was no longer performed on those failing first line NNRTI-based regimens. These changes allowed the clinic to align its HIV resistance testing algorithm with National Department of Health recommendations. Demographic, clinical and laboratory data were extracted from the clinic’s REDCap database43,44. HIV drug resistance test results were obtained, with permission, from the National Health Laboratory Service (NHLS). Definitions Virologic failure was defined as a plasma HIV viral load (VL) >1000 RNA copies/mL based on two consecutive viral load measurements 3 to 6 months apart, with the first VL at least 6 months after ART initiation. Timing of ART initiation was summarised as an ordinal categorical variable, with “early”, “early middle”, “late middle” and “late” periods defined as ART initiation prior to 2004, from 2004 through end-2009, from 2010 through end-2014 and from 2015 through end-2020, respectively. Analysis Exploratory data analysis was conducted to evaluate the clinical and laboratory characteris- tics of cases and controls during the study time period. Continuous variables were evaluated for normality of distribution, and described as mean and standard deviation (SD) for normally distributed variables and median and interquartile range (IQR) for skewed variables. Cate- gorical variables were summarised as proportions. The Student t test was used to compare means between groups, and the Wilcoxon sign rank test was used to compare medians. HIV drug resistance mutations were evaluated using the Stanford University on-line drug resistance database for all sequences submitted to the NHLS during the study time period31,45. Five strategies were used for the case-control analysis, including four approaches to case- control matching using the MatchIt package in R46. The unmatched analysis compared the characteristics of cases and controls using logistic regression, adjusting for age at ART 17 initiation and era in which ART was initiated. In matched analyses, coarsened exact matching with and without replacement within strata, nearest neighbour matching, and full matching were employed, adjusting for age at ART initiation and era in which ART was initiated in all analyses. Once cases were matched to controls, the characteristics of cases and controls in the weighted strata were compared using logistic regression. All logistic regression was conducted as univariate models initially, and variables that were significantly associated with case-status in univariate analyses were combined in the multi- variable model for each case-control analysis. Potential risk factors that were considered in the case-control analyses included sex, age at HIV diagnosis, age at ART initiation, first line ART regimen used (NNRTI- or PI-containing), exposure to PMTCT, baseline CD4 count, baseline HIVVL, previous or current anti-tuberculosis treatment, malnutrition, and orphan status. All analyses were done using R statistical software, version 4.3.147. Sample Size A sample size of at least 360 cases (children with HIV drug resistance testing performed) and 360 controls (children suppressed on first-line ART) was required to detect an odds ratio for risk factors of at least 2.0, assuming high correlation between case and control parameters in univariate analyses ( = 0.60), with 80% power at the 5% significance level. Ethics The study was approved by the Human Research Ethics Committee of the University of the Witwatersrand (clearance number: M210526) and the Ethics Committee of the NHLS (clearance number: PR2119005). 18 Results From 01 January 2011 to 31 December 2020, there were 638 resistance tests conducted on 624 children and adolescents living with HIV that attended Harriet Shezi Children’s Clinic. Twenty-five children and adolescents were excluded from the analysis as age at ART initiation (n=21) and duration of ART therapy at time of the resistance test (n=4) could not be determined. A further 182 children were excluded, as they were over 15 years of age at the time of the resistance test. The resistance dataset included 431 resistance tests that were done in 425 children and adolescents, and analysis included only the first resistance test from each patient (n=425). Resistance Analysis Characteristics of children that underwent HIV resistance testing The median age at ART initiation in children that underwent ART resistance testing was 82.5 months (IQR, 27.7 to 112.8 months; Table 2.1). Half (52.2%) of the children that underwent resistance testing were males. Baseline values for absolute CD4 count, Viral load were taken at initiation of ART. The median absolute CD4 count at baseline was 350 cells/uL (IQR, 123.3 cells/uL to 728.5 cells/uL), and the median HIVVL at baseline was 4.85 log10 RNA copies/mL (IQR 3.83 to 5.56) (Table 2.1). Although most of the baseline characteristics were similar between groups, those on ART for a shorter duration of time were significantly older than those that were on ART for longer at the time of resistance testing (96.4 vs. 72.6 months; Table 2.1). Parameters at the time of resistance testing At the time of resistance testing, the median age was 130.1 months (IQR, 73.8 to 159.5 months). We stratified the children that underwent resistance testing according to the median time on ART prior to a DRT test being performed. The median time on ART prior to DRT was 33.4 months. The children and adolescents that were on a shorter duration of 19 ART (< 33.4 months) at the time of resistance testing being significantly younger (114.4 vs. 143.6 months; Table 2.1). Children who were in the shorter ART duration group also had significantly lower CD4 percentages (18.0% vs. 21.6%) and higher HIVVL (4.51 vs. 4.11 log10 copies/mL) at the time of resistance testing (Table 2.1). At the time of resistance testing, 36.9% (143/425) of the children were on PI-containing ART regimens, with a smaller proportion of those in the shorter ART duration group being on PI-containing ART at the time of the resistance test (30.2% vs. 43.2%; P=0.010) (Table 2.1). 20 Table 2.1: Characteristics of children that underwent HIV resistance testing, stratified by duration on ART at the time of resistance testing Overall ART duration <33.4 months ART duration >=33.4 months P-value n 425 212 213 Baseline Parameters Median age [months, IQR] at ART initiation 82.52 [27.65, 112.80] 96.37 [28.36, 130.80] 72.62 [25.74, 100.59] <0.001 Male (%) 222 (52.2) 111 (52.4) 111 (52.1) 1.000 Median CD4 count [cells/uL, IQR] at baseline 353.50 [123.75, 725.50] 326.00 [115.00, 653.00] 394.00 [159.00, 811.00] 0.099 Median CD4 percentage [%, IQR] at baseline 13.00 [6.80, 20.38] 12.95 [5.84, 20.42] 13.00 [7.04, 20.13] 0.924 Baseline immunologic status (%) 0.335 None 68 (16.2) 28 (13.5) 40 (18.8) Mild 45 (10.7) 26 (12.6) 19 ( 8.9) Advanced 35 ( 8.3) 16 ( 7.7) 19 ( 8.9) Severe 272 (64.8) 137 (66.2) 135 (63.4) Median HIVVL [log10 RNA copies/mL, IQR] at baseline 4.85 [3.84, 5.56] 4.93 [3.77, 5.60] 4.76 [3.86, 5.48] 0.151 Parameters at Resistance Test Median age [months, IQR] at resistance test 130.07 [73.80, 159.45] 114.41 [48.44, 144.30] 143.62 [109.40, 164.96] <0.001 Median ART duration [months, IQR] at resistance test 33.41 [17.60, 58.84] 17.57 [10.89, 25.03] 58.84 [42.81, 79.84] - Mean WFA Z-score (SD) at resistance test -1.05 (1.22) -1.13 (1.36) -0.96 (1.02) 0.574 Mean HFA Z-score (SD) at resistance test -1.92 (1.19) -1.97 (1.27) -1.88 (1.14) 0.619 Median WFH Z-score [IQR] at resistance test 0.42 (1.34) 0.50 (1.62) 0.31 (0.78) 0.576 Median BMI Z-score [IQR] at resistance test -0.20 (1.30) -0.12 (1.57) -0.25 (1.07) 0.509 On PI-based regimen at time of resistance test (%) 143 (36.9) 57 (30.2) 86 (43.2) 0.010 Median CD4 count [cells/uL, IQR] at resistance test 559.50 [321.50, 938.50] 514.00 [217.50, 1006.50] 597.00 [380.00, 878.50] 0.379 Median CD4 percentage [%, IQR] at resistance test 19.49 [12.91, 27.78] 17.99 [10.43, 26.57] 21.60 [15.71, 29.39] 0.010 Mean HIVVL (log10 RNA copies/mL, SD) at resistance test 4.32 (0.84) 4.51 (0.84) 4.11 (0.79) <0.001 Last Follow-up Parameters Median age [months, IQR] at last visit 172.86 [132.30, 210.58] 150.31 [94.21, 180.81] 188.13 [156.66, 216.80] <0.001 Median ART duration [months, IQR] at last visit 86.26 [50.98, 127.27] 45.52 [29.44, 80.05] 109.97 [81.32, 151.72] <0.001 Median CD4 count [cells/uL, IQR] at last follow-up 621.50 [342.25, 922.75] 696.00 [422.00, 1070.50] 532.00 [299.00, 817.00] <0.001 Median CD4 percentage [%, IQR] at follow-up 23.38 [15.05, 30.70] 23.62 [15.03, 30.83] 22.55 [15.11, 30.64] 0.619 Median HIVVL [log10 RNA copies/mL, IQR] at follow-up 2.61 [1.69, 4.09] 2.32 [1.69, 3.88] 3.00 [1.69, 4.25] 0.017 Note: ART = antiretroviral therapy; BMI = body mass index; HFA = height-for-age; HIVVL = HIV viral load in whole blood; IQR = interquartile range; PI = protease inhibitor; SD = standard deviation; WFA = weight-for-age; WFH = weight-for-height.Baseline immunologic status using CDC CD4 percentage immunologic staging. Last follow up paramaters from the last clinic visit of the cohort prior to extraction of the data. P-values are derived from comparison of groups: Chi-square test or Fisher’s exact test, as appropriate, for categorical variables; Student t test for comparison of means; Kruskal-Wallis test for comparison of medians; stratification is by whether the patient was on ART for shorter or longer than the median duration of ART (33.4 months) at the time of the resistance test. 21 Resistance mutations detected In the 431 resistance tests that were done, non-nucleoside reverse transcriptase inhibitor (NNRTI) mutations accounted for 50.8% of all mutations, followed by nucleoside reverse tran- scriptase inhibitor (NRTI) mutations (44.5%) and protease inhibitor (PI) mutations (4.6%). NRTI mutations were detected in 320 (75.3%) of the children who underwent HIV resistance testing. NNRTI mutations and PI mutations were detected in 259 (60.9%) and 19 (4.5%) of the children who underwent HIV resistance testing, respectively. The top mutations according to ART class are presented in Figure 2.1. The most frequent NNRTI mutations detected were K103N (n=154, 20.6%), V106M (n=102, 13.6%), G190A (n=43, 5.7%), P225H (n=40, 5.3%) and V179D (n=35, 4.7%). The top five NNRTI mutations accounted for 374 (49.9%) of all 749 NNRTI mutations. The most frequent NRTI mutations included M184V (n=314, 47.8%), L74V (n=75, 11.4%), Y115F (n=55, 8.4%), K65R (n=22, 3.4%), and any thymidine analogue mutation (n=20, 3%), these five mutations accounted for 486 (74.1%) of all 656 NRTI mutations. There were few PI mutations (n=68) overall, the top five of which were M46I (n= 17, 25%), V82A (n=14, 20.6%), I54V (n= 12,176%), L76V (n=7, 10.3%) and I47IV (n=4, 5.9%). These five mutations accounted for 76.5% of all PI mutations (n=52/68). The detected major mutations were M46I, V82A and I47I/V. The remainder of the top five PI mutations were minor mutations. Factors associated with mutations in children and adolescents that underwent HIV resistance testing Increasing age at ART initiation was significantly associated with NRTI mutations among children and adolescents that underwent HIV resistance testing (adjusted odds ratio (aOR) 1.055; 95% confidence interval (CI), 1.036 to 1.075) in the nearest neighbour matched model which included 105 cases with NRTI mutations and 105 without NRTI mutations (Appendix 22 Table 2.6). Initiation of ART in the late period was associated with a significantly lower adjusted odds of NRTI mutations compared to initiation in the early period (aOR 0.61; 95% CI, 0.036 to 0.103) using a model which used a full matching approach (320 cases compared to 105 controls) (Appendix Table 2.6). Similarly, the development of NNRTI mutations was significantly associated with increasing age at ART initiation (aOR 1.044; 95% CI, 1.033 to 1.055) whereas ART initiation in the late era was protective (aOR 0.056; 95% CI, 0.032 to 0.99) (Appendix Table 2.8). Each month increase in age at ART initiation was associated with a 2.5% (95% CI, 1.1 to 4.1%) reduced adjusted odds of PI resistance mutations in the unmatched model, while ART initiation in the late period (aOR 18.936; 95% CI, 8.328 to 43.057) was associated with PI mutations in the fully matched model (Appendix Table 2.10). Factors such as malnutrition, orphan status and prior tuberculosis therapy were not associated with the development of ART class resistance mutations in these analyses. 23 n=314 n=75 n=55 n=22 n=20 0 10 20 30 40 50 M184V L74V Y115F K65R TAMS NRTI Mutations P er ce nt A n=154 n=102 n=43 n=40 n=35 0 5 10 15 20 K103N V106M G190A P225H V179D NNRTI Mutations P er ce nt B n=17 n=14 n=12 n=7 n=4 0 5 10 15 20 25 M46I V82A I54V L76V I47V PI Mutations P er ce nt C Figure 2.1: The ’top 5’ mutations per ART class detected in children and adolescents attending the Harriet Shezi Children’s Clinic, 2011 through 2020 NRTI = nucleoside reverse transcriptase inhibitor; NNRTI = non-nucleoside reverse transcriptase inhibitor; PI = protease inhibitor; TAMS = thiamine analogue mutations (M41L, L210W, T215Y (Type 1 TAMS); D67N, K70R, T215F, K219Q/E (Type 2 TAMS)). 24 Parameters at the time of last follow-up in children and adolescents that under- went HIV resistance testing At last follow-up clinical assessment , children and adolescents who were on ART for a shorter time period prior to a resistance testing were significantly younger (150.3 vs. 188.1 months), had been on ART for a shorter time period (45.5 vs. 110.0 months), had higher CD4 absolute counts (696 vs. 532 cells/uL) and a better virologic response to change in ART therapy (HIVVL 2.32 vs. 3.00 log10 copies/mL) compared to those that had been on ART for a longer time period at the time of the resistance test (Table 2.1). Predictors of ART class mutations Multiple matching strategies were employed in a multivariable logistic regression to identify risk factors for the different ART class mutations amongst the children that had a resistance test (Table 2.7, table 2.9, table 2.11, Supplementary tables). In the unmatched comparison, initiating ART at an older age (P<0.001) and an unknown PMTCT history (P=0.046) were associated with NNRTI resistance mutations. In the nearest neighbour matching, older ART initiation age was significantly associated with NNRTI and NRTI resistance mutations (P<0.001 for both comparisons). Children that were initiated on ART after 2010 (late middle to late ART era) had increased risk of both NNRTI and NRTI mutations as demonstrated in the full matching comparison. Risk factors for PI mutations were associated with initiating ART at a younger age in the unmatched comparison (P=0.001) and children that initiated on ART in the early middle era (2004-2010) to the late ART era (2015 onwards). Case-control analysis Case-Control Analysis of Factors Associated with HIV Resistance Table 2.2 summarises the characteristics of the cohort of CLWH that were treated during the study time period (7,029 children and adolescents), as well as those of the cases (n=425), 25 controls suppressed on first-line ART (n=953), and CLWH that were not included in the main case-control analysis (n=5,651). Controls were significantly younger at ART initiation (45.2 months; IQR, 13.3 to 91.8 months) compared to cases (81.6 months; IQR, 28.4 to 112.8 months); P<0.001 (Table 2.2). Controls had significantly better immunological status at baseline than did cases, but had similar baseline HIVVL results (Table 2.2). Eighty-one percent of the controls were initiated onto ART in the “early middle” period (2004 to end- 2009), whereas almost half of the cases (46.9%) were initiated onto ART in that time period. A significantly greater proportion of cases compared to controls were ever diagnosed with tuberculosis (33.2% vs. 27.4%; P=0.034), ever orphaned (57.6% vs. 50.6%; p=0.018), or were ever severely malnourished (19.8% vs. 11.7%; 0.001) (Table 2.2). All cases that experienced severe acute malnutrition were malnourished prior to the resistance test. 26 Table 2.2: HIV Clinic Database for Cases (children that underwent resistance testing) and Controls (children that did not have resistance testing done) Overall Cases Controls Suppressed on First-line ART P-value Other Clinic Attendees n 7029 425 953 5651 Male (%) 3540 (50.4) 222 (52.2) 471 (49.4) 0.365 2847 (50.5) Median age at ART initiation (months) [IQR] 48.82 [13.00, 101.17] 81.61 [28.42, 112.78] 45.18 [13.30, 91.81] <0.001 46.97 [12.21, 101.20] Median CD4% at baseline [IQR] 16.38 [9.52, 24.80] 13.77 [7.05, 20.44] 16.63 [10.57, 24.00] <0.001 16.60 [9.56, 25.20] Median CD4 count at baseline [IQR] 537.00 [245.00, 1020.00] 381.00 [162.00, 762.75] 566.00 [294.00, 1019.00] <0.001 544.00 [245.50, 1036.50] Baseline immunologic status (%) 0.039 None 1185 (18.2) 75 (17.7) 176 (18.5) 934 (18.2) Mild 979 (15.0) 43 (10.1) 133 (14.0) 803 (15.6) Advanced 725 (11.1) 42 ( 9.9) 122 (12.8) 561 (10.9) Severe 3620 (55.6) 264 (62.3) 522 (54.8) 2834 (55.2) Log10 HIVVL at baseline [IQR] 4.92 [3.69, 5.67] 4.89 [3.91, 5.58] 4.93 [3.94, 5.70] 0.523 4.92 [3.58, 5.68] ART Initiation Era <0.001 Early (<2004) 84 ( 1.2) 2 ( 0.5) 8 ( 0.8) 74 ( 1.3) Early Middle (2004 to end-2009) 3832 (54.6) 196 (46.9) 772 (81.0) 2864 (50.7) Late Middle (2010 to end-2014) 1991 (28.4) 167 (40.0) 108 (11.3) 1716 (30.4) Late (2015 to end-2020) 1115 (15.9) 53 (12.7) 65 ( 6.8) 997 (17.6) Ever diagnosed with TB (%) 1751 (24.9) 141 (33.2) 261 (27.4) 0.034 1349 (23.9) TB diagnosed pre-resistance test (%) 1733 (24.7) 123 (28.9) 261 (27.4) 0.597 1349 (23.9) Ever orphaned (%) 2695 (38.3) 245 (57.6) 482 (50.6) 0.018 1968 (34.8) Orphaned pre-resistance test (%) 2673 (38.0) 223 (52.5) 482 (50.6) 0.554 1968 (34.8) PMTCT (%) 0.061 Not received 3774 (53.7) 239 (56.2) 562 (59.0) 2973 (52.6) Received 1241 (17.7) 63 (14.8) 169 (17.7) 1009 (17.9) Unknown 2014 (28.7) 123 (28.9) 222 (23.3) 1669 (29.5) Ever SAM (%) 564 (13.4) 82 (19.8) 73 (11.7) 0.001 409 (12.9) SAM pre-resistance test (%) 564 (13.4) 82 (19.8) 73 (11.7) 0.001 409 (12.9) Median duration on ART (months) [IQR] 38.10 [19.06, 68.00] 79.81 [44.86, 119.35] 54.67 [42.71, 71.19] <0.001 31.30 [15.72, 62.40] Died (%) 108 ( 2.5) 17 ( 5.6) 2 ( 0.3) <0.001 89 ( 2.6) Note: ART = antiretroviral therapy; HIVVL = HIV viral load in whole blood; IQR = interquartile range; PMTCT = prevention of mother-to-child transmission; SAM = severe acute malnutrition; TB = tuberculosis. P-values are derived from comparison of Cases (children with HIV resistance testing done) and Controls that were suppressed on First-line ART: Chi-square test or Fisher’s exact test, as appropriate, for categorical variables; Kruskal-Wallis test for comparison of medians. 27 Multivariable analysis of factors associated with case-status Severe acute malnutrition was consistently associated with case-status in each of the multi- variate logistic regression analyses, with aORs ranging from 3.548 (95% CI, 1.979 to 6.359) to 6.383 (95% CI, 3.811 to 10.690) (Table 2.3). Increasing age at ART initiation was signif- icantly associated with HIV resistance testing in the unmatched and fully matched models, with each one month increase in age being associated with a 2.7 to 3.9% increased adjusted odds of case-status (Table 2.3). ART initiation in the late era (2015 onwards) was associated with significantly reduced ad- justed odds of case-status in the unmatched and nearest neighbour matched models (Table 2.3). Baseline immunological status discriminated between case and control status in the coarsened exact matched models, nearest neighbour matched model, and fully matched mod- els with increasing CD4 percent or absolute count associated with control-status (Table 2.3). While all of the models indicated an association between orphan status prior to resistance testing and case-status (aOR ranging from 1.228 to 1.937), this was statistically significant in the nearest neighbour matched model only. Although treatment for tuberculosis prior to resistance testing was associated with case-status in univariate analysis using coarsened exact matching without replacement (OR 1.45; 95% CI, 1.04 to 2.01, Supplementary Table 2.14), it was not associated with case-status in any of the multivariable models (Table 2.3). Receipt of PMTCT was significantly associated with case-status in the fully matched model (aOR 1.746; 95% CI, 1.003 to 3.042) but safeguarded against cases-status in the coarsened exact matched model (aOR 0.535; 95% CI, 0.334 to 0.857) (Table 2.3). 28 Table 2.3: Multivariable Logistic Regression Outputs: Unmatched and Matched Case-Control Models Parameter Adjusted Odds Ratio Lower Est Upper Est P-value Unmatched Intercept 0.073 0.037 0.137 <0.001 ART start age 1.039 1.033 1.045 <0.001 Early Middle (Yes) 6.093 1.580 27.393 0.01 Late Middle (Yes) 0.522 0.169 1.444 0.221 Late (Yes) 0.479 0.284 0.837 0.007 TB pre-resistance test (Yes) 1.121 0.788 1.591 0.524 Orphaned pre-resistance test (Yes) 1.228 0.855 1.766 0.266 PMTCT (Yes) 1.206 0.759 1.911 0.425 PMTCT (Unknown) 1.040 0.698 1.550 0.847 Ever SAM (Yes) 5.519 3.502 8.752 <0.001 Coarsened Exact Matching without replacement Intercept 2.529 1.595 4.009 <0.001 PMTCT (Yes) 0.532 0.332 0.852 0.009 PMTCT (Unknown) 0.661 0.430 1.014 0.058 Baseline CD4% 0.971 0.954 0.989 0.002 TB pre-resistance test (Yes) 0.916 0.618 1.356 0.66 Orphaned pre-resistance test (Yes) 1.547 1.028 2.328 0.037 Ever SAM (Yes) 3.527 1.964 6.335 <0.001 Coarsened Exact Matching with replacement Intercept 3.449 1.135 10.481 0.029 PMTCT (Yes) 0.812 0.484 1.361 0.428 PMTCT (Unknown) 0.892 0.555 1.434 0.637 Baseline CD4% 0.978 0.950 1.007 0.131 Baseline CD4 count 0.999 0.999 1.000 0.003 Baseline log10 HIVVL 0.865 0.735 1.017 0.079 TB pre-resistance test (Yes) 0.667 0.416 1.068 0.092 Orphaned pre-resistance test (Yes) 1.387 0.871 2.206 0.168 Ever SAM (Yes) 4.809 2.846 8.126 <0.001 29 Table 2.3: Multivariable Logistic Regression Outputs: Unmatched and Matched Case-Control Models (continued) Parameter Adjusted Odds Ratio Lower Est Upper Est P-value Nearest Neighbour Matching Intercept 10.878 3.819 30.986 <0.001 Early Middle (Yes) 6.635 2.200 20.012 0.001 Late Middle (Yes) 0.455 0.199 1.042 0.063 Late (Yes) 0.552 0.340 0.896 0.016 PMTCT (Yes) 0.582 0.320 1.057 0.075 PMTCT (Unknown) 0.683 0.438 1.067 0.094 Baseline CD4% 0.952 0.931 0.972 <0.001 Baseline log10 HIVVL 0.766 0.656 0.894 0.001 TB pre-resistance test (Yes) 0.958 0.629 1.460 0.843 Orphaned pre-resistance test (Yes) 1.881 1.254 2.820 0.002 Ever SAM (Yes) 4.237 2.278 7.882 <0.001 Full Matching Intercept 0.212 0.058 0.776 0.019 ART start age 1.030 1.024 1.037 <0.001 Early Middle (Yes) 10.161 2.120 48.711 0.004 Late Middle (Yes) 0.294 0.087 0.996 0.049 Late (Yes) 0.734 0.395 1.365 0.329 PMTCT (Yes) 1.315 0.700 2.473 0.394 PMTCT (Unknown) 1.172 0.741 1.856 0.497 Baseline CD4% 0.969 0.942 0.997 0.033 Baseline CD4 count 1.000 1.000 1.000 0.751 Baseline log10 HIVVL 0.957 0.831 1.101 0.538 TB pre-resistance test (Yes) 1.232 0.836 1.817 0.292 Orphaned pre-resistance test (Yes) 0.952 0.617 1.469 0.825 Ever SAM (Yes) 6.521 3.959 10.741 <0.001 Note: ART = antiretroviral therapy; HIVVL = HIV viral load in whole blood; PMTCT = prevention of mother-to-child transmission; TB = tuberculosis; SAM = severe acute malnutrition. 30 Survival analysis Of 304 cases that underwent resistance testing whose outcome status was known, 17 (5.6%) were known to have died during the study time period. Among controls, 2 (0.3%) of 623 with known outcomes died. Kaplan-Meier estimates of survival indicated significantly better survival outcomes in controls (Figure 2.2). +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++ + p < 0.0001 0.00 0.25 0.50 0.75 1.00 0 50 100 150 200 250 Months on ART O ve ra ll su rv iv al pr ob ab ili ty Study group + +Controls Cases ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++ + 0.00 0.25 0.50 0.75 1.00 0 50 100 150 200 250 Months on ART O ve ra ll su rv iv al pr ob ab ili ty Cases and Controls +A B Figure 2.2: Kaplan-Meier survival curves for cases and controls A = combined survival curve; B = Stratified by study group (cases were children and adolescents that underwent HIV resistance testing, controls were clinic attendees that were virologically suppressed on first line ART). 31 Discussion In this study, we reviewed the prevalence of antiretroviral DRMs and factors that possibly predict the development of ART resistance mutations in children and adolescents under the age of 15 years, attending a large public sector ART clinic in Johannesburg, South Africa. Mutations to NNRTIs accounted for the majority of the mutations, followed by NRTI muta- tions and lastly PI mutations, similar to other studies in the region5,24,29. In low and middle income countries like South Africa, 90% of children failing first line regimens have at least one detectable resistance mutation at time of virological failure29. The rate at which these mutations accumulate was not investigated in our cohort however in a study published in 2016 including Sub-Saharan African children and adults, found DRMs accumulated at an average rate of 1.45 DRM per year24. The M184V mutation was the most frequent mutation detected in our cohort, followed by K103N mutation. The prevalence of mutations in different settings is dependent upon the exposure of patients to different ART classes31. The K103N and M184V mutations are more frequent in South Africa than in other sub-Saharan countries, and likely arise due to earlier and/or programmatic use of efavirenz and lamivudine or emtricitabine, and possibly by the earlier introduction of Option B+ which resulted in life-long ART being accessible to all pregnant women living with HIV, regardless of clinical or immunological staging as a PMTCT strategy14,30. Mutations of NNRTIs accounted for the majority of mutations in the cohort. This is expected as the rate at which drug resistance develops across the different classes of ART is dependent upon the drug’s genetic barriers to resistance. First generation NNRTIs have a low barrier to resistance and the accumulation of resistance-associated mutations is rapid, often occurring within 3 months of virological failure24. The high proportion of children in this study with NNRTI mutations could also represent the pre-treatment drug resistance as a result of ART exposure from PMTCT or the transmission of a resistant virus from the mother22,24 . 32 The top five NNRTI mutations K103N, V106M, G109A, P225H and V179D in our study were similar to those found in other South African cohorts5,24,48. The K103N mutation is associated with reduced susceptibility to NVP and EFV. V106M, a mutation commonly detected in subtype C viruses, also results in resistance to NVP and EFV. V179D reduces the susceptibility of both first generation NNRTI (NVP, EFV) and to second generation NNRTIs such as etravirine (ETR) and rilpivirine by 2-3 fold31. G109A reduces susceptibility to NVP 50-fold, whilst its effect on EFV is less31. The P225H mutation in combination with K103N results in a more than 50-fold reduction in susceptibility to NVP and EFV, and a 5- to 10-fold reduction in doravirine (DOR) susceptibility31. The mutation Y181C, which is associated with resistance to all NNRTIs, was present in our cohort though not as regularly as in other studies in similar settings24. The finding of mutations to second generation NNRTIs in our cohort further supports the current thinking of the limited efficacy of NNRTIs in future regimens. The development and regular use of newer drug classes, such as integrase inhibitors and second generation PIs, is imperative in second and third line regimens in resource limited settings that rely on standardised empiric ART treatment. The NRTI mutations detected in the cohort were similar to what was found in other sub- Saharan studies5,24,49. The M184V mutation, which confers resistance to 3TC and FTC, was the most frequent mutation detected, occurring in 72.9% (341/431) of the DRTs performed. However the presence of M184V is not a contraindication to the use of 3TC/FTC, as the mutation results in impaired viral fitness and increases susceptibility to AZT and TDF7. The mutations L74V and Y115F select for resistance to ABC. The combination of L74V and M184V is a common pattern of mutation in patients on an ABC/3TC regimen which was the recommended first line NRTI paediatric combination since the replacement of d4T with ABC due to the unfavourable side effect profile of d4T5,14. The K65R mutation, associated with TDF and ABC resistance, was found in only 5.1% (22/431) of the DRTs. The low levels of TDF resistance in this cohort, are likely due to TDF 33 being prescribed infrequently in children and the use of TDF being endorsed in children in this cohort who are older than 10 years and weigh more than 35 kilograms15. This finding supports the use of TDF in future regimens as recommended by the current WHO guidelines2. The Type 1 TAMs found in this cohort, individually and in combination, result in high level resistance to AZT and D4T. The Type 2 TAMs, D72N and K70R, also result in high level resistance to mainly AZT and D4T31. The Q151M complex, a pan-nucleotide mutation that confers high level resistance, was not present in the cohort. This is similar to what was found in other paediatric cohorts, unlike adult patients were the Q151M mutation is frequently detected24. This is likely due to Q151M being selected by TDF, which is not used as frequently in children compared to adults. The proportion of children that were on a PI regimen that underwent resistance testing in the study was 31% (139/435). The prevalence of resistance mutations to PIs in this cohort was low at 4.6%, which attests to the robustness of this drug class. In Southern Africa, the most frequently reported PI mutations are V82A, I54V, and M46I, which are selected by lopinavir and ritonavir31 as detected in this study. Factors associated with resistance to NRTIs and NNRTIs were similar, and included older age at initiation of ART, an unknown PMTCT history and initiating on ART beyond 2010. These are likely due to PMTCT being more available and guideline changes to longer duration on NVP/AZT for the infant and widespread availability of ART to pregnant women. Initiating PIs at a younger age was a risk factor for PI mutations and this may be due to developing metabolic pathways, variability in pharmacokinetics, the unpalatable taste of PI’s and issues with the storage of Protease inhibitors39. In a sub-analysis of the 26 children and adolescents with any PI mutation(s), half (53.8%), had 3 to 5 PI mutations detected in their DRTs. Clinical characteristics of children with 1 or 2 PI mutations were similar to those with 3 to 5 PI mutations. The WHO ART guidelines currently recommend dolutegravir (DTG) with 2 NRTIs as first line and as second line in children and adults. DTG in combination with NRTIs is effective, 34 even in patients with extensive NRTI resistance50. There were 953 children that were suppressed on first line that matched to the cases that underwent DRT. CD4 percentage, CD4 count, older age at ART initiation, tuberculosis (TB), orphan status, nutritional status and duration of ART have an impact in the likelihood of having a DRT performed. Children that underwent resistance testing were older at time of initiation of ART compared to the controls. This could be due to older children being initiated on an Efavirenz (NNRTI) regimen, unlike the younger children that would be initiated on the more robust PI based regimen as per guidelines in place at the time. The control cohort had higher CD4 counts and percentages at baseline and this is likely due to the cohort being younger and therefore having higher CD4 absolute counts. The probability of survival diminishes with the longer duration of ART. A diagnosis of TB increased the likelihood of case-status in our analysis, although not sta- tistically significant in the case-control multivariate outputs. High pill burden impairing adherence and the effects of rifampicin on cytochrome P450 activity, which affected the metabolism of PI’s, may be factors associated with the requirement for DRT in children failing ART7. Malnutrition was associated with the development of resistance across all models of compari- son, this is similar to other studies performed in Southern Africa39,40. All of the children with SAM in this cohort had SAM prior to resistance testing. Malnutrition predisposes children to resistance by changing the pharmacokinetics of drugs through altered metabolic function, body composition and intestinal absorption of ART39,40. This highlights the importance of the nutritional status of children living with HIV. Ensuring that children are well nourished with food aid programmes, social grants and the involvement of dietitians/nutritionists in the management of children who are on ART should be considered in preventing resistance. There are few studies that interrogate malnutrition as a risk factor for resistance and more studies should be done to investigate the impact of interventions that improve the nutritional status of children on ART and the risk of resistance development. 35 HIV Drug resistance poses a major public health challenge and impacts individual patient care as it reduces the number of available and effective treatment options and threatens the success of both prevention and treatment programmes such as the UNAIDS 95-95-95 global treatment targets. Routine HIV genotyping is not feasible in resource-limited regions where large scale treatment programmes with standardised empiric ART regimens are used to treat persons living with HIV. Continued surveillance and monitoring of resistance patterns is important to ensure sustainability of ART treatment programmes. Conclusions Surveillance of HIV drug resistance and the predictors of resistance development are impor- tant in resource-limited settings to ensure that empiric regimes are appropriate and sustain- able. The WHO ART guidelines currently recommend dolutegravir (DTG) with 2 NRTIs as first line and as second line in children and adults. DTG in combination with NRTIs is effective, even in patients with extensive NRTI resistance50. Substandard nutritional status was consistently and significantly associated with the require- ment for DRT in this cohort of children and adolescents. Conversely, higher baseline CD4 percentage was associated with control status. Early initiation of ART to preserve immuno- logical status, and nutritional support throughout the course of ART clinic attendance may limit the emergence of drug resistance in CLWH. References 1 UNAIDS. UNAIDS data 2022 Reference. Unaids 2022; : 1–446. 2 World Health Organization. Updated recommendations on HIV prevention, infant diagnosis, antiretroviral initiation and monitoring. 2021. 3 UNAIDS and AIDSinfo. Estimates Adults and children living with Country fact- sheets SouthAfrica | 2020 HIV testing and treatment cascade People living with HIV Coverage of adults and children. Unaids 2021; : 1–6. 36 https://www.unaids.org/en/resources/documents/2023/2022_unaids_data https://www.ncbi.nlm.nih.gov/pubmed/7710379 https://www.ncbi.nlm.nih.gov/pubmed/7710379 https://www.ncbi.nlm.nih.gov/pubmed/7710379 4 World Health Organization. Interim guidelines suppl to the 2016 Consolidated guide- lines. 2018. 5 Nuttall J, Pillay V. Antiretroviral Resistance Patterns in Children with HIV Infection. 2019; 21. DOI:10.1007/s11908-019-0663-z. 6 World Health Organization. HIV Drug Resistance Report 2021. Technical report. 2021. 2021. 7 Rossouw T, Lessells R, de Oliveira T. HIV & TB Drug Resistance and Clinical Man- agement Case Book. Cape Town: South African Medical Research Council, 2013. 8 Vella S, Schwartländer B, Sow SP, Eholie SP, Murphy RL. The history of antiretroviral therapy and of its implementation in resource-limited areas of the world. Aids 2012; 26: 1231–41. 9 World Health Organization. Antiretroviral therapy for HIV infection in infants and children: Towards Universal Access. Recommendations for a public health approach. 2010 Revision. 2010. 10 Mnyani CN, Tait CL, Peters RPH, et al. Implementation of a PMTCT programme in a high HIV prevalence setting in Johannesburg, South Africa: 2002-2015. Southern African Journal of HIV Medicine 2020; 21: 1–7. 11 National Department of Health. Policy and Guidelines for the Implementation of the PMTCT Programme. http://www.doh.gov.za/docs/policy/pmtct-f.html. 2008. http://www.doh.gov.za/docs/policy/pmtct-f.html. 12 National Department of Health. CLINICAL GUIDELINES : PMTCT ( Prevention of Mother-to- Child Transmission ) National Department of Health , South Africa ; Aids 2010. 13 National Department of Health. 2023 ART Clinical Guidelines 2023 for the Manage- ment of HIV in Adults, Pregnancy and Breastfeeding, Adolescents, Children, Infants and Neonates. 2023. 37 https://doi.org/10.1007/s11908-019-0663-z https://www.who.int/publications/i/item/9789240038608 https://www.who.int/publications/i/item/9789240038608 https://researchonline.lshtm.ac.uk/id/eprint/967463/ https://researchonline.lshtm.ac.uk/id/eprint/967463/ https://doi.org/10.1097/QAD.0b013e32835521a3 https://doi.org/10.1097/QAD.0b013e32835521a3 https://doi.org/10.4102/SAJHIVMED.V21I1.1024 https://doi.org/10.4102/SAJHIVMED.V21I1.1024 http://www.doh.gov.za/docs/policy/pmtct-f.html 14 National Department of Health. National Consolidated Guidelines for the Prevention of Mother-To-Child Transmission of HIV (PMTCT) and the Management of HIV in Children, Adolescents and Adults. Department of Health, Republic of South Africa 2015; : 1–128. 15 National Department of Health. Guideline for the Prevention of Mother to Child Transmission of Communicable Infections South African National Department of Health. Department of Health, Republic of South Africa 2019; : 1–44. 16 Levin LJ, Horak JL, Nuttall J. Looking back at paediatric HIV treatment in South Africa. My, how we have grown! Southern African Journal of HIV Medicine 2021; 22: 1–3. 17 Steegen K, Bronze M, Papathanasopoulos MA, et al. Prevalence of antiretroviral drug resistance in patients who are not responding to protease inhibitor-based treatment: Results from the first national survey in South Africa. In: Journal of infectious diseases. Oxford University Press, 2016: 1826–30. 18 Jordan MR, Penazzato M, Cournil A, et al. Human Immunodeficiency Virus (HIV) Drug Resistance in African Infants and Young Children Newly Diagnosed with HIV: A Multicountry Analysis. Clinical Infectious Diseases 2017; 65: 2018–25. 19 Boerma RS, Sigaloff KCE, Akanmu AS, Inzaule S, Hensbroek MBV, Wit TFRD. Alarming increase in pretreatment HIV drug resistance in children living in sub- Saharan Africa : a systematic review and meta-analysis. 2017; : 365–71. 20 Violari A, Cotton MF, Gibb DM, et al. Early Antiretroviral Therapy and Mortality among HIV-Infected Infants. New England Journal of Medicine 2008; 359: 2233–44. 21 National Department of Health. 2019 ART Clinical Guidelines for the Management of HIV in Adults, Pregnancy, Adolescents, Children, Infants and Neonates: Updated March 2020. 2020; : 3–3. 22 World Heaith Organization. Global Action Plan on HIV Drug Resistance 2017-2021: 2018 Progress Report - Hiv Drug Resistance. 2018. 38 https://www.google.com/search?q=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&oq=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&aqs=chrome..69i57j0i512j69i64.444j0j7&source https://www.google.com/search?q=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&oq=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&aqs=chrome..69i57j0i512j69i64.444j0j7&source https://www.google.com/search?q=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&oq=Guideline+for+the+Prevention+of+Mother+to+Child+Transmission+of+Communicable+Infections&aqs=chrome..69i57j0i512j69i64.444j0j7&source https://doi.org/10.4102/SAJHIVMED.V22I1.1283 https://doi.org/10.4102/SAJHIVMED.V22I1.1283 https://doi.org/10.1093/infdis/jiw491 https://doi.org/10.1093/infdis/jiw491 https://doi.org/10.1093/infdis/jiw491 https://doi.org/10.1093/cid/cix698 https://doi.org/10.1093/cid/cix698 https://doi.org/10.1093/cid/cix698 https://doi.org/10.1093/jac/dkw463 https://doi.org/10.1093/jac/dkw463 https://doi.org/10.1056/nejmoa0800971 https://doi.org/10.1056/nejmoa0800971 https://clintonhealthaccess.org/content/uploads/2017/09/2017- https://clintonhealthaccess.org/content/uploads/2017/09/2017- 23 World Health Organization. HIV DRUG RESISTANCE REPORT 2019. 2019 https: //www.who.int/publications/i/item/WHO-CDS-HIV-19.21. 24 Boender TS, Kityo CM, Boerma RS, et al. Accumulation of HIV-1 drug resistance after continued virological failure on first-line ART in adults and children in sub- Saharan Africa. Journal of Antimicrobial Chemotherapy 2016; 71: 2918–27. 25 Orrell C, Levison J, Ciaranello A, et al. Resistance in pediatric patients experiencing virologic failure with first-line and second-line antiretroviral therapy. Pediatr Infect Dis J 2013; 32: 644–7. 26 Boerma RS, Boender TS, Bussink AP, et al. Suboptimal Viral Suppression Rates Among HIV-Infected Children in Low- and Middle-Income Countries: A Meta- analysis. Clin Infect Dis 2016; 63: 1645–54. 27 Boerma RS, Boender TS, Sigaloff KC, et al. High levels of pre-treatment HIV drug resistance and treatment failure in Nigerian children. J Int AIDS Soc 2016; 19: 21140. 28 Makatini ZN, Chb MB, Path FC, et al. High virological failure rates in HIV-1 peri- natally infected children in South Africa : A retrospective cohort study. 2021; 111: 255–9. 29 Sigaloff KCE, Calis JCJ, Geelen SP, Vugt MV, Wit TFRD. HIV-1-resistance- associated mutations after failure of fi rst-line antiretroviral treatment among children in resource-poor regions : a systematic review. The Lancet Infectious Diseases 2011; 11: 769–79. 30 Sánchez PR, Holguín A. Drug resistance in the HIV-1-infected paediatric population worldwide: A systematic review. Journal of Antimicrobial Chemotherapy 2014; 69: 2032–42. 31 Standford University. HIV Drug Resistance Database. 2022. https://hivdb.stanford .edu/. 39 https://www.who.int/publications/i/item/WHO-CDS-HIV-19.21 https://www.who.int/publications/i/item/WHO-CDS-HIV-19.21 https://doi.org/10.1093/jac/dkw218 https://doi.org/10.1093/jac/dkw218 https://doi.org/10.1093/jac/dkw218 https://doi.org/10.1097/INF.0b013e3182829092 https://doi.org/10.1097/INF.0b013e3182829092 https://doi.org/10.1093/cid/ciw645 https://doi.org/10.1093/cid/ciw645 https://doi.org/10.1093/cid/ciw645 https://doi.org/10.7448/ias.19.1.21140%0A21140 https://doi.org/10.7448/ias.19.1.21140%0A21140 https://doi.org/10.1016/S1473-3099(11)70141-4 https://doi.org/10.1016/S1473-3099(11)70141-4 https://doi.org/10.1016/S1473-3099(11)70141-4 https://doi.org/10.1093/jac/dku104 https://doi.org/10.1093/jac/dku104 https://hivdb.stanford.edu/ https://hivdb.stanford.edu/ 32 Kityo C, Sigaloff KCE, Sonia Boender T, et al. HIV Drug Resistance among Children Initiating First-Line Antiretroviral Treatment in Uganda. AIDS Research and Human Retroviruses 2016; 32: 628–35. 33 Meyers T, Sawry S, Wong JY, Moultrie H. Virologic Failure among Children Taking Lopinavir/Ritonavir-Containing First-Line Antiretroviral Therapy in South Africa . Pediatric Infectious Disease Journal 2015; 34: 175–9. 34 Pillay S, Bland RM, Lessells RJ, Manasa J, Oliveira T de, Danaviah S. Drug resistance in children at virological failure in a rural KwaZulu-Natal, South Africa, cohort. AIDS Res Ther 2014; 11: 3. 35 Tzou PL, Rhee SY, Descamps D, et al. Integrase strand transfer inhibitor (INSTI)- resistance mutations for the surveillance of transmitted HIV-1 drug resistance. Journal of Antimicrobial Chemotherapy 2020; 75: 170–82. 36 Hackett S, Teasdale CA, Pals S, et al. Drug Resistance Mutations Among South African Children Living With HIV on WHO-recommended ART Regimens. Clini- cal infectious diseases : an official publication of the Infectious Diseases Society of America 2021; 73: e2217–25. 37 Kityo C, Sigaloff KC, Sonia Boender T, et al. HIV Drug Resistance Among Chil- dren Initiating First-Line Antiretroviral Treatment in Uganda. AIDS Res Hum Retro- viruses 2016; 32: 628–35. 38 Sebunya R, Musiime V, Kitaka SB, Ndeezi G. Incidence and risk factors for first line anti retroviral treatment failure among Ugandan children attending an urban HIV clinic. 2013; : 1–10. 39 Rossouw TM, Feucht UD, Melikian G, et al. Factors associated with the development of drug resistance mutations in HIV-1 infected children failing protease inhibitor-based antiretroviral therapy in South Africa. PLoS ONE 2015; 10: 1–16. 40 https://doi.org/10.1089/aid.2015.0215 https://doi.org/10.1089/aid.2015.0215 https://doi.org/10.1097/INF.0000000000000544 https://doi.org/10.1097/INF.0000000000000544 https://doi.org/10.1186/1742-6405-11-3 https://doi.org/10.1186/1742-6405-11-3 https://doi.org/10.1093/jac/dkz417 https://doi.org/10.1093/jac/dkz417 https://doi.org/10.1093/cid/ciaa1068 https://doi.org/10.1093/cid/ciaa1068 https://doi.org/10.1089/aid.2015.0215 https://doi.org/10.1089/aid.2015.0215 https://doi.org/10.1371/journal.pone.0133452 https://doi.org/10.1371/journal.pone.0133452 https://doi.org/10.1371/journal.pone.0133452 40 Vaz P, Buck WC, Bhatt N, et al. Compromise of Second-Line Antiretroviral Therapy Due to High Rates of Human Immunodeficiency Virus Drug Resistance in Mozambican Treatment-Experienced Children With Virologic Failure. J Pediatric Infect Dis Soc 2020; 9: 6–13. 41 World Health Organization. Paediatric Antiretroviral Drug Optimization ( PADO ) Meeting 4 Meeting objectives Meeting summary. 2018; : 1–18. 42 UNICEF. Regional snapshot : Eastern and Southern Africa Children , HIV and AIDS. 2018; : 2017–8. 43 Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009; 42: 377–81. 44 Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: Building an in- ternational community of software platform partners. J Biomed Inform 2019; 95: 103208. 45 Shafer RW. Rationale and uses of a public HIV drug-resistance database. J Infect Dis 2006; 194 Suppl 1: S51–8. 46 Ho DE, Imai K, King G, Stuart EA. MatchIt: Nonparametric Preprocessing for Para- metric Causal Inference. J Stat Softw 2011; 42: 1–28. 47 R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2023 https://www.R-project.org/. 48 Abbas UL DY Glaubius RL. Drug resistance from preferred antiretroviral regimens for HIV infection in South Africa : A modeling study. 2020; : 1–17. 49 Skhosana L, Steegen K, Bronze M, et al. High prevalence of the k65r mutation in hiv-1 subtype c infected patients failing tenofovir-based first-line regimens in south africa. PLoS ONE 2015; 10. DOI:10.1371/journal.pone.0118145. 41 https://doi.org/10.1093/jpids/piy102 https://doi.org/10.1093/jpids/piy102 https://doi.org/10.1093/jpids/piy102 %7Bhttps://cdn.who.int/media/docs/default-source/hq-hiv-hepatitis-and-stis-library/pado4.pdf?sfvrsn=26d4169c_5%7D %7Bhttps://cdn.who.int/media/docs/default-source/hq-hiv-hepatitis-and-stis-library/pado4.pdf?sfvrsn=26d4169c_5%7D https://data.unicef.org/wp-content/uploads/2018/11/ESAR-Regional-snapshot-2018.pdf https://doi.org/10.1016/j.jbi.2008.08.010 https://doi.org/10.1016/j.jbi.2008.08.010 https://doi.org/10.1016/j.jbi.2008.08.010 https://doi.org/10.1016/j.jbi.2019.103208 https://doi.org/10.1016/j.jbi.2019.103208 https://doi.org/10.1086/505356 https://doi.org/10.18637/jss.v042.i08 https://doi.org/10.18637/jss.v042.i08 https://www.R-project.org/ https://doi.org/10.1371/journal.pone.0218649 https://doi.org/10.1371/journal.pone.0218649 https://doi.org/10.1371/journal.pone.0118145 50 Paton NI, Musaazi J, Kityo C, et al. Dolutegravir or Darunavir in Combination with Zidovudine or Tenofovir to Treat HIV. New England Journal of Medicine 2021; 385: 330–41. 42 https://doi.org/10.1056/nejmoa2101609 https://doi.org/10.1056/nejmoa2101609 APPENDIX 1: Supplementary Tables 43 Table 2.4: Characteristics of children that underwent HIV resistance testing, stratified by age at ART initiation Overall ART start <82.1 months ART start >=82.1 months P-value n 425 212 213 Baseline Parameters Median age [months, IQR] at ART initiation 82.52 [27.65, 112.80] 26.84 [10.35, 57.89] 112.80 [98.68, 132.54] - Male (%) 222 (52.2) 112 (52.8) 110 (51.6) 0.882 Median CD4 count [cells/uL, IQR] at baseline 353.50 [123.75, 725.50] 598.00 [271.50, 1036.50] 205.00 [51.00, 451.00] <0.001 Median CD4 percentage [%, IQR] at baseline 13.00 [6.80, 20.38] 14.55 [8.61, 23.88] 11.00 [3.61, 18.00] <0.001 Baseline immunologic status (%) 0.045 None 68 (16.2) 28 (13.3) 40 (19.0) Mild 45 (10.7) 30 (14.3) 15 ( 7.1) Advanced 35 ( 8.3) 20 ( 9.5) 15 ( 7.1) Severe 272 (64.8) 132 (62.9) 140 (66.7) Median HIVVL [log10 RNA copies/mL, IQR] at baseline 4.85 [3.84, 5.56] 4.99 [3.90, 5.78] 4.71 [3.76, 5.28] 0.003 Parameters at Resistance Test Median age [months, IQR] at resistance test 130.07 [73.80, 159.45] 73.72 [44.65, 123.05] 154.33 [136.42, 168.66] <0.001 Median ART duration [months, IQR] at resistance test 33.41 [17.60, 58.84] 39.15 [21.35, 73.97] 28.36 [15.97, 48.65] <0.001 Mean WFA Z-score (SD) at resistance test -1.05 (1.22) -1.00 (1.16) -1.34 (1.54) 0.424 Mean HFA Z-score (SD) at resistance test -1.92 (1.19) -1.90 (1.12) -1.93 (1.27) 0.870 Median WFH Z-score [IQR] at resistance test 0.33 [-0.13, 1.00] 0.32 [-0.12, 0.98] 0.77 [-0.23, 1.16] 0.895 Median BMI Z-scor