The last man standing is the most resistant: eliminating artemisinin-resistant malaria in Cambodia

Abstract

Background: Artemisinin combination therapy (ACT) is now the recommended first-line treatment for falciparum malaria throughout the world. Initiatives to eliminate malaria are critically dependent on its efficacy. There is recent worrying evidence that artemisinin resistance has arisen on the Thai-Cambodian border. Urgent containment interventions are planned and about to be executed. Mathematical modeling approaches to intervention design are now integrated into the field of malaria epidemiology and control. The use of such an approach to investigate the likely effectiveness of different containment measures with the ultimate aim of eliminating artemisinin-resistant malaria is described.

Methods: A population dynamic mathematical modeling framework was developed to explore the relative effectiveness of a variety of containment interventions in eliminating artemisinin-resistant malaria in western Cambodia.

Results: The most effective intervention to eliminate artemisinin-resistant malaria was a switch of treatment from artemisinin monotherapy to ACT (mean time to elimination 3.42 years (95% CI 3.32-3.60 years). However, with this approach it is predicted that elimination of artemisinin-resistant malaria using ACT can be achieved only by elimination of all malaria. This is because the various forms of ACT are more effective against infections with artemisinin-sensitive parasites, leaving the more resistant infections as an increasing proportion of the dwindling parasite population.

Conclusion: Containment of artemisinin-resistant malaria can be achieved by elimination of malaria from western Cambodia using ACT. The "last man standing" is the most resistant and thus this strategy must be sustained until elimination is truly achieved.

Figure 1

Schematic diagram of the structure of the mathematical modeling framework. The structure of the model is built up as follows: natural history and pharmacodynamics incorporated as a repeating unit (A) with four compartments – susceptible people, people with liver stage infection, people with noninfectious blood stage infection and people with infectious (blood) stage infection, rates of flow between these compartments (βI/N, δ, γ and σ) and rates of recovery due to each of artesunate and piperaquine treatment (c_Bda, c_Ida, c_Bdb and c_Idb) (Additional file 1- Table S2) is shown. The times to recovery (1/rate) following treatment are then adjusted by a multiplying factor (e_rada or e_rbdb) (0 ≤ e ≤ 1) depending on the degree of resistance to each drug, giving three possible linked variants of unit A (resistant to no drug, artesunate only and piperaquine only) making up a repeating pattern (B). Finally the population dynamics of transmission is shown in (C). This consists of multiple repetitions of (B) with different rates of flow between them at different time points depending on which treatments and interventions are used. For example, for individuals with blood stage infections to begin treatment with ACT, they will move from the 'No drugs' box (1) to the equivalent parts of an 'ACT' box (2) at a rate determined by the time to begin treatment. The dynamics in the 'ACT' box are different from the 'No drugs' box as these individuals will be subjected to faster rates of recovery due to the ACT. Each box is also subject to pharmacokinetic dynamics independent of infection dynamics. This is in the form of waning pharmacodynamic drug effect over time ('loss of...') with sequential loss of DHA and then piperaquine. This results in a percentage of the entire unit moving to a new box 'Piperaquine' (3) which again has different dynamics representing the effect of piperaquine on recovery rates. Interventions shown here are elimation of artemisinin monotherapy and replacement with ACT ('Switch to ACT') and MSAT and MDA with ACT. Each circle represents a population exposed to a particular drug or combination. Key: ACT = dihydroartemisinin/piperaquine combination therapy, Rx = treatment, DHA = dihydroartemisinin. (For more details, please see the Full Model Code in Additional File 2.)

Figure 2

Effect of continuing availability and use of artemisinin monotherapy on the total number of malaria infections (black line), the number of artemisinin-resistant infections (red line) and percentage of infections resistant to artemisinin (red dotted line) over time. Artesunate monotherapy is introduced as treatment in 1975 and a single artemisinin-resistant infection in 1980.

Figure 3

Effect of eliminating artesunate monotherapy and replacement with ACT in 2009 for treatment of symptomatic cases on the total number of malaria infections (black line), the number of artemisinin-resistant infections (red line) and the percentage of infections resistant to artesunate (dotted lines, pink = artesunate red = ACT) (mean-field approximation).

Figure 4

Example of a single stochastic output illustrating the effect of a switch of treatment to ACT in 2009.