📊 statistics

Modelling HIV Drug Resistance Dynamics with Pre- and Post-Treatment Resistance Profiling in Botswana

This study employs mathematical modeling and Bayesian inference to demonstrate that implementing systematic HIV drug resistance profiling in Botswana significantly reduces transmission rates and AIDS progression by optimizing treatment regimens and lowering the effective reproduction number.

Original authors: Refilwe Matlou, Tefa Kaisara, Farai Nyabadza

Published 2026-10-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Refilwe Matlou, Tefa Kaisara, Farai Nyabadza

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the fight against HIV, a virus that has shaped public health for decades, the goal has long been to turn a fatal infection into a manageable chronic condition. This transformation relies on antiretroviral therapy, a daily regimen of medicines that suppresses the virus and allows people to live long, healthy lives. However, the virus is a master of adaptation. It replicates rapidly and makes frequent errors, creating new variations. Sometimes, these variations are naturally resistant to the drugs used to treat them. When a person starts treatment with a virus that is already resistant, or when the virus evolves resistance while on treatment, the first line of defense fails. The patient remains sick, the virus continues to spread, and the healthcare system must scramble to find a more complex, expensive, and often less accessible second or third line of treatment. The challenge for countries with high rates of infection is not just providing medicine, but ensuring the right medicine reaches the right person at the right time.

Botswana, a nation in Southern Africa with one of the highest rates of HIV in the world, has been a global leader in expanding access to treatment. By 2021, the country had achieved remarkable milestones, with nearly all people who knew their status receiving care and the vast majority of those on treatment achieving viral suppression. Yet, even in this success story, the shadow of drug resistance looms. While the country recently shifted its standard first-line treatment to a more robust drug, the threat of resistance remains. A critical question for health officials is whether testing people for drug resistance before they start treatment—rather than waiting to see if the treatment fails—can stop the spread of resistant strains and improve outcomes. To answer this, researchers Refilwe Matlou, Tefa Kaisara, and Farai Nyabadza built a mathematical model to simulate the dynamics of HIV transmission in Botswana, comparing a system where everyone starts on the same first-line drug against a system where patients are tested for resistance first.

The researchers constructed a detailed simulation of the population, dividing people into groups based on their health status and treatment journey. They tracked those who were susceptible to infection, those infected with a virus sensitive to drugs, those infected with a virus already resistant to drugs, and those currently receiving different levels of treatment. They also followed the progression of the disease to its most severe stage. The model allowed them to run two distinct scenarios. In the first scenario, representing the past or a system without pre-treatment testing, every person diagnosed with HIV started on the standard first-line treatment. If the drug did not work, they would eventually switch to a second or third-line regimen, but only after a period of failure. In the second scenario, representing a system with resistance profiling, individuals were tested immediately upon diagnosis. If the test revealed a resistant virus, they were skipped over the first-line treatment entirely and started directly on the more advanced, effective regimen.

The simulations revealed that this proactive approach of testing before treatment significantly lowers the potential for the virus to spread. The researchers found that when people with resistant viruses are identified early and moved directly to the correct, stronger medication, the virus is suppressed much faster. This rapid suppression reduces the amount of virus in their blood, making them far less likely to pass the infection to others. In the model, this shift in strategy lowered the basic reproduction number, a measure of how many new infections one person causes, effectively slowing the epidemic's momentum. The study showed that the benefit comes from two main factors: the immediate identification of resistant cases prevents them from wasting time on ineffective drugs, and the successful treatment of these cases reduces their ability to infect others.

The researchers also explored how the effectiveness of the second-line drugs influences the outcome. Their model indicated that the strategy works best when the advanced treatments are highly effective at suppressing the virus. If the second-line drugs are not potent enough, or if patients struggle to take them consistently, the advantage of the testing strategy diminishes. The simulations suggested that the greatest reduction in transmission occurs when there is a combination of high-quality testing, immediate access to effective second-line drugs, and strong support to ensure patients stick to their medication. The study highlighted that simply having the tests is not enough; the entire system must be ready to act on the results quickly.

Furthermore, the model demonstrated that resistance profiling changes the landscape of treatment. In the simulated future with testing, fewer people would be stuck on failing first-line treatments, and the population receiving advanced therapy would grow more efficiently. This does not mean more people are failing treatment overall; rather, it means the right people are getting the right treatment immediately. The simulations showed a clear reduction in the number of people progressing to the most severe stage of the disease, known as AIDS, and a decrease in the total number of people living with drug-resistant HIV. The researchers concluded that for a country like Botswana, where the burden of HIV is heavy, integrating routine resistance testing into standard care is a vital step. It offers a way to protect the gains made in treatment coverage by ensuring that the medicines being used are actually working, thereby stopping the spread of resistant strains and saving lives.

Technical Summary: Modelling HIV Drug Resistance Dynamics with Pre- and Post-Treatment Resistance Profiling in Botswana

Problem Statement
Despite significant progress in antiretroviral therapy (ART) coverage in Botswana, the emergence and transmission of HIV drug resistance (HIVDR) threaten the sustainability of treatment gains. The study addresses the critical gap in understanding how the transition from reactive resistance management (treating all patients with first-line ART regardless of resistance status) to proactive resistance profiling (testing for resistance before initiating treatment) impacts transmission dynamics. Specifically, the authors aim to quantify the epidemiological benefits of identifying drug-resistant strains in ART-naive individuals and directing them immediately to appropriate second- or third-line regimens, rather than subjecting them to ineffective first-line therapy that fosters further resistance and disease progression.

Methodology
The authors developed two deterministic compartmental mathematical models to simulate HIV transmission dynamics in Botswana:

Pre-Resistance Profiling Model ( $\rho = 0$ ): Represents the standard care approach where all newly infected individuals initiate first-line ART. Treatment failure leads to a switch to second- or third-line regimens only after virological failure is detected.
Post-Resistance Profiling Model ( $0 < \rho \leq 1$ ): Incorporates a resistance profiling parameter ( $\rho$ ), representing the proportion of resistant infections identified prior to treatment initiation. In this scenario, individuals with resistant strains are directed immediately to second- or third-line ART ( $T_2$ ), while those with sensitive strains proceed to first-line ART ( $T_1$ ).

Model Structure:
The population is divided into six compartments: Susceptible ( $S$ ), Drug-Sensitive Infected ( $I_s$ ), Drug-Resistant Infected ( $I_r$ ), First-line ART ( $T_1$ ), Second/Third-line ART ( $T_2$ ), and AIDS ( $A$ ). The force of infection $\lambda(\rho)$ accounts for varying infectivity across compartments, including a reduction factor $\xi$ for individuals on effective advanced regimens.

Analysis and Estimation:

Theoretical Analysis: The authors derived the basic reproduction number ( $R_0$ ) using the next-generation matrix method. They established conditions for the global asymptotic stability of both the disease-free equilibrium (using the Castillo-Chavez approach) and the endemic equilibrium (using Lyapunov function theory).
Parameter Estimation: Model parameters were estimated using Bayesian inference with Hamiltonian Monte Carlo (HMC) sampling via Stan. The model was calibrated against UNAIDS data for adults receiving ART in Botswana from 2010 to 2023.
Sensitivity Analysis: Partial Rank Correlation Coefficients (PRCC) were calculated to identify key drivers of $R_0$ and the effectiveness of resistance profiling.

Key Results

Reduction in Transmission Potential: Theoretical analysis demonstrates that the basic reproduction number under complete resistance profiling ( $R_0^{post}$ ) is strictly less than that of the pre-profiling scenario ( $R_0^{pre}$ ) provided the infectivity reduction parameter $\xi < 1$ . This confirms that directing resistant individuals to effective regimens reduces the overall force of infection.
Parameter Sensitivity: The effective transmission rate ( $\beta$ ) was identified as the most influential driver of $R_0$ . The resistance profiling proportion ( $\rho$ ) showed a strong negative correlation with $R_0$ , indicating that increased profiling coverage significantly lowers transmission potential. Conversely, the infectivity reduction parameter ( $\xi$ ) and the switching rate ( $\epsilon$ ) were positively correlated with $R_0$ ; higher values of $\xi$ imply reduced treatment effectiveness in the $T_2$ class, while the positive correlation for $\epsilon$ suggests that, under the estimated parameter values, faster movement into $T_2$ does not sufficiently reduce infectivity (consistent with the condition $\xi\kappa_3 > \kappa_2$ ).
Simulation Outcomes:
- Treatment Allocation: Increasing $\rho$ shifts the population distribution, decreasing the number of individuals on ineffective first-line therapy ( $T_1$ ) and increasing the proportion on appropriate advanced regimens ( $T_2$ ).
- Disease Burden: Simulations show that resistance profiling leads to a reduction in the total number of drug-resistant infections ( $I_r$ ) and a decrease in the progression to AIDS ( $A$ ).
- Infectivity: The force of infection in the post-profiling scenario ( $\lambda_{post}$ ) decreases linearly as the efficacy of advanced regimens ( $\xi$ ) improves. When $\xi = 1$ (no reduction in infectivity), profiling offers no transmission benefit; as $\xi \to 0$ , the transmission reduction is maximized.

Significance and Claims
The paper claims that systematic resistance profiling, coupled with rapid access to effective second- and third-line regimens, is a necessary strategy to control HIVDR in high-prevalence settings like Botswana. The study provides mathematical evidence that:

Proactive Testing: Identifying resistance before treatment initiation prevents the "waste" of time on ineffective first-line drugs, thereby reducing the duration of infectiousness for resistant strains.
Policy Implications: To maximize the benefits of profiling, health systems must ensure robust laboratory capacity, reliable supply chains for second- and third-line drugs, and adherence support programs. Without these, the parameter $\rho$ cannot be effectively utilized, and the transmission benefits of profiling are negated.
Epidemiological Impact: The transition to resistance-informed care reduces the community-level transmission of resistant HIV and slows the progression of the disease to AIDS, supporting the achievement of UNAIDS 95-95-95 goals.

The authors acknowledge limitations, noting that the model does not account for heterogeneous contact patterns in key populations, pre-exposure prophylaxis (PrEP), or potential delays in resistance testing results, suggesting these as areas for future research.

Technical Summary: Modelling HIV Drug Resistance Dynamics with Pre- and Post-Treatment Resistance Profiling in Botswana

More like this