Effects of recombination on multi-drug resistance evolution in Plasmodium falciparum malaria

When multiple beneficial alleles are present in a population but not linked together in any one individual, there is no general evolutionary result that determines whether recombination will speed up or slow down the emergence and evolution of genotypes carrying multiple beneficial alleles. Translated to infectious disease control, this evolutionary uncertainty means that when multiple types of drug resistance are present we do not know whether recombination will act more strongly to (1) bring together single-resistant genotypes into multi-drug resistant (MDR) genotypes, or (2) break apart MDR genotypes into single-resistant genotypes. In this paper, we introduce a new version of an established and validated individual-based malaria transmission model where we have added individual mosquito bites, interrupted feeding by mosquitoes, 25 drug-resistance related loci, and individual recombination events of different Plasmodium falciparum genotypes inside the mosquito. Recombination among P. falciparum genotypes in this model occurs from two sources of variation, multi-clonal infections and interrupted feeding by mosquitoes, and we show that 80% to 97% of MDR recombinant falciparum genotypes are projected to occur from single uninterrupted bites on hosts with multi-clonal infections (for malaria prevalence > 5%). Increases in the model’s interrupted feeding rate slowly increase the number of recombination events occurring from interrupted feeds. A comparison of drug-resistance management strategies with this new model shows that, over a 15-year timeframe, triple artemisinin-combination therapies (ACT) strategies show the largest reductions in treatment failures and the longest delays until artemisinin resistance reaches a critical 1% threshold. Multiple first-line therapies (MFT) are second best under these criteria, and ACT cycling approaches are third best. MFT strategies generate a greater diversity of recombinant genotypes but fewer recombination events generating MDR and slower emergence of these recombinant MDR genotypes.