Modelling the impact of artemisinin combination therapy and long-acting treatments on malaria transmission intensity

Abstract

Background: Artemisinin derivatives used in recently introduced combination therapies (ACTs) for Plasmodium falciparum malaria significantly lower patient infectiousness and have the potential to reduce population-level transmission of the parasite. With the increased interest in malaria elimination, understanding the impact on transmission of ACT and other antimalarial drugs with different pharmacodynamics becomes a key issue. This study estimates the reduction in transmission that may be achieved by introducing different types of treatment for symptomatic P. falciparum malaria in endemic areas.

Methods and findings: We developed a mathematical model to predict the potential impact on transmission outcomes of introducing ACT as first-line treatment for uncomplicated malaria in six areas of varying transmission intensity in Tanzania. We also estimated the impact that could be achieved by antimalarials with different efficacy, prophylactic time, and gametocytocidal effects. Rates of treatment, asymptomatic infection, and symptomatic infection in the six study areas were estimated using the model together with data from a cross-sectional survey of 5,667 individuals conducted prior to policy change from sulfadoxine-pyrimethamine to ACT. The effects of ACT and other drug types on gametocytaemia and infectiousness to mosquitoes were independently estimated from clinical trial data. Predicted percentage reductions in prevalence of infection and incidence of clinical episodes achieved by ACT were highest in the areas with low initial transmission. A 53% reduction in prevalence of infection was seen if 100% of current treatment was switched to ACT in the area where baseline slide-prevalence of parasitaemia was lowest (3.7%), compared to an 11% reduction in the highest-transmission setting (baseline slide prevalence = 57.1%). Estimated percentage reductions in incidence of clinical episodes were similar. The absolute size of the public health impact, however, was greater in the highest-transmission area, with 54 clinical episodes per 100 persons per year averted compared to five per 100 persons per year in the lowest-transmission area. High coverage was important. Reducing presumptive treatment through improved diagnosis substantially reduced the number of treatment courses required per clinical episode averted in the lower-transmission settings although there was some loss of overall impact on transmission. An efficacious antimalarial regimen with no specific gametocytocidal properties but a long prophylactic time was estimated to be more effective at reducing transmission than a short-acting ACT in the highest-transmission setting.

Conclusions: Our results suggest that ACTs have the potential for transmission reductions approaching those achieved by insecticide-treated nets in lower-transmission settings. ACT partner drugs and nonartemisinin regimens with longer prophylactic times could result in a larger impact in higher-transmission settings, although their long term benefit must be evaluated in relation to the risk of development of parasite resistance.

Figure 1. Overview of Model Structure

The main states and transitions of the transmission cycle and treatment in the human population are shown, including presumptive treatment (dashed lines). Here one age and exposure group and a single type of antimalarial treatment are represented.

Figure 2. Example Model Run over Time for One Age and Exposure Group in One Survey Setting

Initially this shows introduction of infection and baseline endemic equilibrium in the absence of treatment. Nonartemisinins with treatment failure are introduced first and the model is allowed to reach a second equilibrium, matching the prevalence of slide-positive malaria, clinical malaria and self-reported treatment in the Tanzania survey data. ACT or another antimalarial regimen are then introduced, and our main outcomes of interest are the reductions in prevalence and rate of clinical episodes between the second and third equilibriums.

Figure 3. Model Predictions of ACT Impact on Transmission in Six Transmission Settings in Tanzania Compared to the Pre-ACT Scenario with Failing Nonartemisinin Treatment

(A) Relative and absolute reductions in clinical episodes achieved by ACT if treatment rates remained the same and there was a 100% switch to ACT. Also shown is the pre-ACT ratio of treatment to infection rate. (B) Relative reductions in clinical episodes by ACT coverage.

Figure 4. Model-Estimated Impact of Introducing Improved Diagnostic Procedures

Improved diagnostic methods are introduced prior to antimalarial prescription together with a 100% switch to ACT in six transmission settings in Tanzania, compared to no change in current treatment use, (A) on the percentage of clinical episodes prevented and (B) on the efficiency of treatment at reducing transmission.

Figure 5. Model-Estimated Impact of Introducing Antimalarials with 100% Efficacy and Different Pharmacodynamic Properties

Impact is shown on (A) clinical episodes and (B) slide-prevalence of infection in six transmission settings in Tanzania, compared to the pre-ACT scenario with failing nonartemisinin treatment assuming a 100% switch to these treatments. Short-acting nongametocytocidal: prophylactic time = 10 d, no specific gametocytocidal action. Short-acting ACT (as Figure 3A): prophylactic time = 10 d, with the gametocytocidal action of artemisinin. Long-acting nongametocytocidal: prophylactic time = 25 d, no specific gametocytocidal action. Long-acting ACT: prophylactic time = 25 d, with the gametocytocidal action of artemisinin.