Leveraging Cohort Data and Mathematical Modelling to Predict HIV Drug Resistance Evolution in South Africa

With the advent of antiretroviral therapy (ART), millions of lives have been saved. However, HIV’s rapid adaptation within and between hosts has made HIV drug resistance a persistent challenge since the late 1980. While newer, more potent antiretroviral drugs combined with viral load monitoring have led to a reduction in population level drug resistance in resource-rich settings, acquired and transmitted drug resistance remain a significant threat to the sustainability of HIV treatment, particularly in resource-limited settings where HIV prevalence is high and viral load monitoring less frequent, and with a programmatic approach to treatment management. In response to the exponential rise of pre-treatment resistance to non-nucleoside reverse transcriptase inhibitors (NNRTIs) – previously a cornerstone of WHO-recommended ART regimens used by millions of people – the WHO recommended the integrase strand transfer inhibitor (INSTI) dolutegravir as preferred first-line regimen in 2018. Dolutegravir offers higher potency for viral suppression, fewer side effects and a higher genetic barrier to resistance than NNRTIs which in 2013, shortly after its approval, even prompted speculation about its potential to be “evolution-proof”. Reports of emergent dolutegravir resistance have been scarce; by mid-2023, fewer than 100 cases with sequencing data have been published. Most data originate from resource-rich settings where HIV subtype B predominates. By 2023, dolutegravir-based ART has been adopted as preferred first-line regimen by 116 countries and is used by 25 million people living with HIV. With the widespread use of dolutegravir in settings with highly standardized ART regimen, recycled drugs, limited access to adherence support, infrequent viral load monitoring and higher risks of stock outs, the risk for emergent dolutegravir resistance is increased. In chapter 1, we conducted the first study systematically assessing drug resistance in individuals with viremia on dolutegravir-containing regimens. This was achieved through a cohort collaboration study within the ART Cohort Collaboration (ART-CC) and two additional cohorts from the United Kingdom and South Africa. By assembling the largest dataset on drug resistance in the context of dolutegravir-based ART in real-world settings, we were able to describe the prevalence and patterns of drug resistance mutations and identify risk factors for emergent dolutegravir resistance. These findings have been instrumental in shaping the broader understanding of resistance mechanisms in the context of dolutegravir, as evidenced by their inclusion in the WHO 2024 HIV Drug Resistance Brief Report and discussion at associated stakeholder meetings. In chapter 2, we employed MARISA (Modelling Antiretroviral Resistance In South Africa), a deterministic compartmental model that simulates the HIV epidemic in South Africa. We integrated mutational acquisition pathways for INSTI drug resistance mutations, parametrized using findings from chapter 1. To our knowledge this is the first study modelling dolutegravir resistance projections based on empirical data. Our results indicate that acquired dolutegravir resistance increases rapidly at population level and is strongly dependent on the duration of viraemia while on treatment. While transmitted resistance is currently very rare, our projections show an upward trend which may reach 10% within the next decade; however, these estimates are subject to parameter uncertainty. We additionally evaluated resistance mitigation strategies which are currently being evaluated in clinical trials through counterfactual scenarios. Our results suggest that such strategies may strongly reduce or even halt the emergence of dolutegravir resistance. In chapter 3, we addressed the gap in resistance data from resource-limited settings and non-B subtypes, where the risk for emerging drug resistance is increased due to epidemiological, health care system, and possibly viral factors. Within the framework of IeDEA, a global observational HIV cohort study, we prospectively enrol individuals experiencing virologic failure during dolutegravir-based ART in 26 sites across 16 countries. Enrolment is ongoing, and the final study population is projected to exceed 1’000 individuals. Preliminary results from sub-Saharan sites representing nearly 500 study participants revealed substantially higher rates of dolutegravir resistance compared to resource-rich settings. Moreover, the mutational pathways observed in these populations differed strongly from those described in chapter 1. These findings are essential for informing future guidelines and policies for HIV drug resistance surveillance and ART management globally. In summary, the work presented in this thesis combines epidemiological data, mathematical modelling, and evidence synthesis to predict the likelihood and public health impact of dolutegravir resistance evolution. This work quantifies the current occurrence of resistance, identifies critical determinants of resistance emergence, and provides projections on future incidence and transmission dynamics. Additionally, it outlines actionable public health strategies to mitigate dolutegravir resistance and preserve treatment efficacy. By integrating molecular, clinical, and epidemiological perspectives, this work contributes valuable insights into sustainable ART strategies over the coming decades. These findings also serve as a valuable example on the interaction between evolutionary processes and public health interventions, demonstrating how cutting-edge evolutionary biology research can inform real-world public health applications. $\textit{Full-text embargoed until: 2026-03-06}$