Modelling HIV Drug Resistance in Southern Africa.

Hauser, Anthony Willy (2021). Modelling HIV Drug Resistance in Southern Africa. (Unpublished). (Dissertation, University of Bern, Faculty of Medicine, the Faculty of Science and the Vetsuisse Faculty)

Preview

Text Hauser_PhD_2021.pdf - Accepted Version Available under License Creative Commons: Attribution-Noncommercial-No Derivative Works (CC-BY-NC-ND). Download (4MB) | Preview

Southern Africa is the region most affected by HIV globally. In South Africa, for example, the
prevalence of HIV reaches 17% among adults. In the early 2000s, the roll-out of antiretroviral
therapy (ART), a non-nucleoside reverse transcriptase inhibitor (NNRTI) and two nucleoside
reverse transcriptase inhibitors (NRTIs), had a dramatic impact on decreasing mortality
related to acquired immunodeficiency syndrome (AIDS). However, the recent emergence of
resistance to NNRTI threatens the long-term efficacy of such regimen. As a response, a new
ART first-line regimen is introduced in several countries of Southern Africa, where the NNRTI
drug is replaced by an integrase strand transfer inhibitor (InSTI) drug, called dolutegravir
(DTG). DTG has a high genetic barrier to resistance, is highly effective, well tolerated and
affordable in resource-limited settings. In this thesis, I develop mathematical models aimed
at characterizing different aspects of the dynamics of HIV drug resistance in Southern Africa.
In Chapter 1, I give a brief timeline of the HIV-epidemic in Southern Africa. I then introduce
basic concepts on HIV, ART, and HIV drug resistance. I present the different strategies that
have been implemented in Southern Africa to fight HIV. Finally, I discuss the increasing role
that mathematical models play to gain insight on the HIV-epidemic.
In Chapter 2, I run a systematic review and meta-analysis estimating the prevalence
of NRTI/NNRTI drug resistance mutations among adults failing a first-line NNRTI-based
regimenin Southern Africa. I develop a Bayesianhierarchicalmodelthat synthesizes evidence
from the collected studies. The model estimates high levels of K65 and M184 mutations
after 2 years of regimen including emtricitabine or lamivudine (FTC/3TC) and tenofovir
(TDF), the two NRTI backbones that are now commonly associated in first-line regimen. The
K65 and M184 mutations confer high levels of resistance to FTC/3TC and TDF, respectively.
Therefore, it suggests that between 43% and 55% of people failing a NNRTI-based regimen,
will switch to DTG-based regimen with substantially compromised NRTI backbones, if they
are not optimized. These results show the importance of monitoring DTG-response in this
population, as they have higher risk of DTG-failure, where resistance could develop.
In Chapter 3, I develop a compartmental model, the MARISA model, which captures both
the general HIV-epidemic and the dynamic of NNRTI drug resistance in South Africa.
Data from several sources, including cohort data on thousands of people living with HIV
(PLWH), are used to calibrate the model. The MARISA model also assesses the impact
of counterfactual scenarios reflecting alternative countrywide policies during 2005-2016,
considering either increasing ART coverage, improving management of treatment failure,
broadening ART eligibility, or implementing drug resistance testing before ART initiation. I
identify key driversof theNNRTI resistance epidemic: large-scaleARTroll-out andinsufficient
monitoring of first-line treatment failure. The results also suggest that no simple measure
CHAPTER . ABSTRACT
could have prevented the rise ofNNRTI resistance in the South African context, whereNNRTIs
have been rapidly rolled out.
In Chapter 4, I adapt the MARISA model to assess the impact of different strategies of DTG
introduction on the level of NNRTI resistance in South Africa. I investigate the impact of two
scenarios of the DTG-introduction: 1) DTG as a first-line ART, or 2) DTG replacing NNRTIs for
all patients, including patients on NNRTI-based regimen. Due to safety concerns related to
DTG during pregnancy, the model also considers scenarios whereDTG is prescribed to all men
and in addition to i) women beyond reproductive age, ii) women beyond reproductive age or
using contraception, and iii) allwomen. The simulations showthat, while somestrategies can
stabilize the level of NNRTI resistance, none of the different strategies introducing DTG leads
to its elimination. To halt the increase of NNRTI resistance, DTG should become accessible to
both women and people currently on NNRTI-based therapy. As some women (e.g. women
at risk of pregnancy) will continue to rely on NNRTI-based ART in the future, controlling the
resistance to NNRTI is key to provide them with an effective alternative to DTG.
The Chapter 5 discusses some important public-health questions regarding HIV drug
resistance in sub-Saharan Africa. It stresses the central role of mathematical modelling to
quantify the risk of HIV drug resistance when data is scarcely available. It also discusses the
modelling idea used in Chapters 3 and 4 and shows howmathematical models can bridge the
gap between the wide availability of HIV epidemiological data and the limited knowledge on
HIV drug resistance in the African regions.
In Chapter 6, I summarize the main findings presented in Chapters 2-4 and discuss their
implications. I also present the strengths and weaknesses of the project. Finally, I discuss
the perspective of the potential emergence of resistance to DTG. The Chapter 7 presents an
additional study, in which I was involved but which does not represent the core ofmy thesis.
In this study, a mathematical model reproduces the dynamics of the SARS-CoV-2 epidemics
in several regions of the world and provides estimates of the age-specific mortality related to
SARS-CoV-2.
In this thesis, I use mathematical modelling to capture the emergence of NNRTI resistance in
South Africa. I identify some factors that have driven the development of NNRTI resistance,
such as the long time spent on a failing regimen. Due to its flexibility, the MARISA model is
adapted to investigate future strategies, such as the impact of the DTG-introduction on the
levels of NNRTI resistance. This shows that processes such as the acquisition and the spread
of HIV drug resistance can be reproduced at the population-level using mathematical models
calibrated with clinical resistance data. As South Africa is currently introducing DTG-based
regimen, such modelling approach can be implemented to investigate the future risk of
emergence of DTG resistance. However, even if mathematical models could help to bridge
the gaps between clinical and real-world resource-limited settings, more real-world data
is needed to understand the actual risk of DTG resistance development in the context of a
countrywide implementation of DTG. The meta-analysis in Chapter 2 adds to the body of
evidence, as it highlights the potential threat on the long-term efficacy of DTG posed by the
switch of patients with elevated viral load. Close follow-up and resistancemonitoring of these
patients are therefore key to ensure an early detection of DTG resistance and prevent it from
spreading through the population.

Interest & Impact

Downloads

5 since deposited on 22 Mar 2022
5 in the past 12 months

Citations

5 Citations in Web of Science ®
6 Citations in Scopus

Search

in Google Scholar™

Services

Actions (login required)

Edit item

Actions (login required)

Edit item