Abstract
Increasing levels of artemisinin and partner drug resistance threaten malaria control and elimination globally. Triple artemisinin-based combination therapies (TACTs) which combine artemisinin derivatives with two partner drugs are efficacious and well tolerated in clinical trials, including in areas of multidrug-resistant malaria. Whether early TACT adoption could delay the emergence and spread of antimalarial drug resistance is a question of vital importance. Using two independent individual-based models of Plasmodium falciparum epidemiology and evolution, we evaluated whether introduction of either artesunate-mefloquine-piperaquine or artemether-lumefantrine-amodiaquine resulted in lower long-term artemisinin-resistance levels and treatment failure rates compared with continued ACT use. We show that introduction of TACTs could significantly delay the emergence and spread of artemisinin resistance and treatment failure, extending the useful therapeutic life of current antimalarial drugs, and improving the chances of malaria elimination. We conclude that immediate introduction of TACTs should be considered by policy makers in areas of emerging artemisinin resistance.
Introduction
The introduction of artemisinin-based combination therapies (ACTs) into routine clinical use for uncomplicated Plasmodium falciparum malaria has saved millions of lives over the past two decades1. These drugs are the mainstay of antimalarial treatment and remain highly effective in most malaria-endemic regions2. However, resistance to artemisinins has emerged in several malaria-endemic regions, first in the Greater Mekong Subregion (GMS) of Southeast Asia3,4,5, and subsequently in South America6, Papua New Guinea7, and most recently in Eastern Africa8,9. In the GMS, artemisinin resistance was compounded by ACT partner drug resistance, causing ACT treatment failure10,11. Increasing drug resistance threatens global malaria control and elimination efforts. Artemisinin partial resistance, caused by mutations in the pfkelch13 gene12,13, results in slower parasite clearance, increased rates of treatment failure, increased transmissibility, and reduced protection of the partner drugs from resistance emergence and spread. The recent emergence of artemisinin resistance in East Africa is particularly concerning as the previous reduction in the malaria burden has stalled in many parts of the continent. Malaria mortality, which fell substantially between 2000 and 2015, is estimated to have plateaued over the past 8 years2. Worsening drug resistance will result in an increasing death toll, with most of these preventable deaths occurring in African children.
The current high efficacy of antimalarial treatments must be maintained if a lethal reversal in malaria trends is to be avoided. New antimalarial drugs are promised14, but even if their development continues successfully, they may not become available for years and will be more expensive15. Multiple first-line therapies (MFT), which mathematical modeling suggests can slow drug resistance spread16,17, have become policy in more than a dozen endemic countries2. However, there is as yet no field evidence demonstrating the success of MFT at slowing or delaying resistance evolution. A potential alternative solution, based on the same biological principles of combining two different slowly eliminated antimalarial drugs with an artemisinin derivative, are triple artemisinin-based combinations (TACTs), which should provide protection against partner drug resistance and therefore preserve high treatment efficacy, given the extreme rarity of parasites acquiring mutations conferring resistance against both partner drugs over a short period of time. Randomized clinical trials in Asia with dihydroartemisinin-piperaquine-mefloquine (DHA-PPQ-MQ) and artemether-lumefantrine-amodiaquine (ALAQ) have shown that these combinations are well-tolerated, safe, and effective, including in areas with multidrug-resistant falciparum malaria18,19. Dose-optimized TACTs are now being tested in large trials in African and Asian countries (clinicaltrials.gov identifiers NCT03923725 and NCT03939104, respectively). However, the long-term evolutionary benefits of TACT deployment cannot be assessed with clinical trials. Here, we use a consensus mathematical modeling approach to project the potential long-term evolutionary dynamics and clinical treatment outcomes of TACT deployment in different malaria epidemiological settings. The results can inform proactive policies aiming to contain the emergence and spread of antimalarial drug resistance.
Results
The models predicts that replacing ACTs with triple artemisinin-based combination therapies (TACTs) substantially slows the emergence and evolution of artemisinin-resistant alleles and restores the clinical efficacy of first-line therapy (Fig. 1). Notably at model year zero, under a scenario where DHA-PPQ had been used for the previous 15 years, the mutant pfkelch13 580Y alleles were present in both models; at 0.013 (IQR: 0.006–0.043) allele frequency in the PSU model and 0.080 (IQR: 0.065–0.096) allele frequency in the MORU model. Under continued DHA-PPQ use, the allele frequency of pfkelch13 580Y increased to a median frequency of 0.884 (IQR: 0.725–0.968; PSU) or 0.571 (IQR: 0.335–0.769; MORU) after 10 years. In both models, under continued ACT use, ten years was sufficient time for artemisinin-resistant genotypes to become established (Fig. 1 and Supplementary Figs. 2–7). If ASMQ-PPQ was deployed at year zero replacing DHA-PPQ, pfkelch13 580Y alleles were projected to reach frequencies of 0.056 (IQR: 0.014–0.226; PSU) or 0.155 (IQR: 0.112–0.224; MORU) after 10 years. Under ALAQ deployment, 580Y emergence was slower under the PSU model, with predicted median allele frequency of 0.015 after 10 years (IQR: 0.003–0.054), but similar under the MORU model (median = 0.162; IQR: 0.116–0.213). The relative benefit of ALAQ over ASMQ-PPQ was sensitive to the model used and scenario examined (Fig. 2). The risk of an infection carrying the pfkelch13 580Y allele after only 2 years of ALAQ (TACT) deployment was significantly lower than that with continued ACT use, with mean relative risks of 0.49 (95% CI: 0.36-0.66; PSU) and 0.71 (95% CI: 0.55-0.91; MORU). With ASMQ-PPQ (TACT) deployment, the relative risks were 0.49 (95% CI: 0.35–0.68; PSU) and 0.93 (95% CI: 0.73–1.19; MORU). The corresponding relative risks at year 10 were 0.19 (95% CI: 0.16–0.22; PSU) and 0.33 (95% CI: 0.29–0.38; MORU) for ASMQ-PPQ and 0.10 (95% CI: 0.08–0.12; PSU) and 0.31 (95% CI: 0.27–0.36; MORU) for ALAQ.
In 52 out of 54 scenarios examined (27 per model) the 10-year treatment failure rates were lower when TACTs were introduced compared to continued ACT deployment (Fig. 3; nearly all Mann–Whitney p values < 10−4; see Supplementary Figs. 20, 21). In the other two scenarios, both in a low-transmission setting (75% treatment coverage, 0.1% prevalence, baseline AL use, PSU model; and 25% treatment coverage, 0.1% prevalence, baseline DHA-PPQ use, MORU model) the simulations projected median prevalence levels below 0.05% after 10 years with median treatment failure rates of zero both for continued ACT use and switch to TACT. Thus, no scenarios showed any advantages of ACT deployment over TACT deployment–Supplementary Figs. 22–28.
Preventing drug resistance
The short-term benefits of a switch to TACTs (Fig. 1) are likely to be critically dependent on the initial conditions at year zero (the year of TACT introduction) and the expected evolutionary trajectories of drug resistance allele frequencies. Treatment failure and allele frequency dynamics are highly nonlinear and grow exponentially in their early phases (Supplementary Figs. 29–34), suggesting that the timing of TACT introduction (i.e., as early as possible) could be critical to its long- and short-term success.
Following the standard scenarios (Fig. 1), switching to TACTs led to median artemisinin resistance allele frequencies ranging from 0.01 to 0.17 (across the two models and two TACTs) ten years after TACT deployment. If TACT introduction were delayed by three years, the median projected 10-year pfkelch13 580Y allele frequency ranged from 0.08 to 0.22 and shifted to 0.24 to 0.45 if TACT introduction were delayed by 5 years. This behavior is consistent between models across different prevalence settings although the magnitude of the effect over ten years differed between the models (Fig. 4A, B). In very low prevalence settings delays in TACT implementation could preclude elimination (Fig. 5).
It should be noted that while we are presenting 580Y frequency as the signature genetic marker of artemisinin resistance and treatment failure, there are differences in how selection operates across resistance loci. For example, piperaquine resistance and artemisinin resistance are co-selected strongly in a scenario where DHA-PPQ is the recommended ACT (Supplementary Fig. 34). The reason is that the treatment failure rate of DHA-PPQ on the double-resistant genotype is very high. Continued AL deployment, on the other hand, facilitates mdr1 second copy selection (Supplementary Fig 35), which is further enhanced if changing to ASMQ-PPQ but mitigated if changing to ALAQ. AL is the only regimen to select strongly for N86, although this can be reversed after a policy change to TACTs (Supplementary Fig 35). Under high selection pressure scenarios, ASMQ-PPQ consistently selects for a higher frequency of drug resistance-related alleles relative to ALAQ (Supplementary Figs. 34–36). Continued use of ASAQ is predicted to result in high long-term 580Y frequencies but barely noticeable increases in all other resistance-related allele frequencies.
Pre-existing triple resistance mechanisms
The major long-term benefit of combination therapy, that multidrug resistance emerges much later under combination therapy than single-drug resistance does under monotherapy, relies on multidrug-resistant genotypes being absent from the population at the time of deployment. To evaluate the risk posed by pre-existing multidrug resistance, we explored a range of starting ASMQ-PPQ triple-resistant genotype frequencies in a PfPR setting of 0.1%. If the triple-resistant genotype’s frequency was lower than 0.01 at the time of TACT deployment, the models showed no major effect on the emergence of parasites resistant to all 3 drugs (Fig. 6). However, a marked increase in the risk of triple drug-resistant mutant spread was predicted when the initial frequency of these mutants was between 0.01 to 0.05 (Fig. 6A). For example, when triple-mutant frequency was set to 0.04 at year zero, natural selection operated efficiently (i.e., no stochastic disappearance due to genetic drift) to produce median frequencies of 0.35 (IQR: 0.11–0.62; PSU) and 0.25 (IQR: 0.00–0.55; MORU) after 10 years. This selection is enabled by the model-calculated 71% to 73% treatment efficacy of ASMQ-PPQ for infections with the triple-resistant genotype (see PKPD section of the methods)….
