Mass campaigns with antimalarial drugs: a modelling comparison of artemether-lumefantrine and DHA-piperaquine with and without primaquine as tools for malaria control and elimination

Abstract

Background: Antimalarial drugs are a powerful tool for malaria control and elimination. Artemisinin-based combination therapies (ACTs) can reduce transmission when widely distributed in a campaign setting. Modelling mass antimalarial campaigns can elucidate how to most effectively deploy drug-based interventions and quantitatively compare the effects of cure, prophylaxis, and transmission-blocking in suppressing parasite prevalence.

Methods: A previously established agent-based model that includes innate and adaptive immunity was used to simulate malaria infections and transmission. Pharmacokinetics of artemether, lumefantrine, dihydroartemisinin, piperaquine, and primaquine were modelled with a double-exponential distribution-elimination model including weight-dependent parameters and age-dependent dosing. Drug killing of asexual parasites and gametocytes was calibrated to clinical data. Mass distribution of ACTs and primaquine was simulated with seasonal mosquito dynamics at a range of transmission intensities.

Results: A single mass campaign with antimalarial drugs is insufficient to permanently reduce malaria prevalence when transmission is high. Current diagnostics are insufficiently sensitive to accurately identify asymptomatic infections, and mass-screen-and-treat campaigns are much less efficacious than mass drug administrations. Improving campaign coverage leads to decreased prevalence one month after the end of the campaign, while increasing compliance lengthens the duration of protection against reinfection. Use of a long-lasting prophylactic as part of a mass drug administration regimen confers the most benefit under conditions of high transmission and moderately high coverage. Addition of primaquine can reduce prevalence but exerts its largest effect when coupled with a long-lasting prophylactic.

Conclusions: Mass administration of antimalarial drugs can be a powerful tool to reduce prevalence for a few months post-campaign. A slow-decaying prophylactic administered with a parasite-clearing drug offers strong protection against reinfection, especially in highly endemic areas. Transmission-blocking drugs have only limited effects unless administered with a prophylactic under very high coverage.

Figure 1

Pharmacokinetics and pharmacodynamics of antimalarial drugs are implemented in an agent-based model of malaria transmission and calibrated to clinical data. (A) A simplified model of drug pharmacokinetics uses a double exponential to model antimalarial drug distribution and elimination. (B) Pharmacokinetics of 5 antimalarials as simulated with the simple PK model in a 50 kg adult. (C) Maximum drug concentration with simple PK including age-based dosing, weight-dependent PK parameters, and WHO/CDC age-weight tables. See Tables 1 and 2 for model PK parameters and age-based dose fractions.

Figure 2

Parasite killing rates of antimalarial drugs are calibrated to clearance time, prevalence, and infectivity. (A) Asexual parasite clearance time is within 3 days post-treatment for AL (red) and DP (blue) in both naive (solid bar) and semi-immune (hashed bar) populations. Black bars: 95% confidence interval. (B) AL (red) and DP (blue) pharmacodynamics are calibrated in a semi-immune population to <3% recrudescence rate (solid lines) at day 42 post-treatment and 50% and 20% reinfection rate (dashed lines) at day 42 post-treatment with a background EIR of 3/month. Shaded areas: 95% confidence interval. (C) Fraction of patients clear of parasites at day 42 post-treatment is dependent on patient age for DP-treated population (blue) but not AL-treated population (red). Young children are more likely to experience recrudescence (solid line) and reinfection (dashed line) than older patients when treated with DP. Shaded areas: 95% confidence interval. (D) Gametocyte clearance time for AL-treated patients of all ages is calibrated to 7–10 days (red lines) post-treatment. Addition of single-dose PQ at 0.75 mg/kg reduces gametocyte clearance time to < 7 days. Shaded areas: 95% confidence interval. (E) Mean fraction of mosquitoes infected on day 2 post-treatment with AL, PQ, AL + PQ, or no drug in a semi-immune population challenged with an infectious bite 25 days prior to treatment. Primaquine was given with a single dose at 0.065, 0.10, 0.25, 0.40, and 0.75 mg/kg. Bars: 95% confidence interval (smaller than dot radius for PQ alone and AL + PQ).

Figure 3

Campaign outcome depends on timing, diagnostic sensitivity, coverage, and compliance. (A) Asexual parasite prevalence (top) and daily EIR (bottom) of a semi-immune population of 1000 people with no intervention (black), multi-round MDA with AL (red), and multi-round MDA with DP (blue). Solid lines: 3-round campaigns. Dashed lines: 2-round campaigns. MDAs were simulated with 100% coverage and 100% compliance. Annual EIR was 50, and infected individuals were those with ≥10 asexual parasites/μL. DP’s long prophylactic tail better protects against reinfection after campaign. (B) Prevalence 1 month after campaign for 3-round and 2-round MDA campaigns with AL or DP. MDAs were simulated with 70% coverage and 100% compliance; annual EIR was 50. Error bars: 95% confidence interval. (C) Prevalence 1 month after 3-round MSAT campaigns with varying sensitivity of MSAT diagnostic. Coverage and compliance were 100%; annual EIR was 50. Shaded areas: 95% confidence interval. (D) Prevalence 1 month after 3-round MDA campaigns with varying coverage and 100% compliance. Annual EIR was 50. Shaded areas: 95% confidence interval. (E) Prevalence 1 month after 3-round MDA campaigns with varying compliance and 100% coverage. All covered individuals take the first dose of AL or DP; subsequent doses are taken with probability equal to the compliance. Annual EIR was 50. Shaded areas: 95% confidence interval.

Figure 4

Efficacy of prophylaxis is greatest under high transmission conditions and moderately high coverage. (A) AL reduces prevalence by curing individuals in an infected population, while additional prevalence reduction observed in DP is due to the long-lasting prophylactic effects of piperaquine. Annual EIR of 50, MDA coverage 90%, compliance 100%, 3 campaign rounds, mean of 100 stochastic realizations. (B) Normalized prevalence 4 months after MDA campaigns in high and low transmission settings. Transmission-blocking refers to prevalence reduction upon addition of PQ. Coverage 90%, compliance 100%, 3 campaign rounds, annual EIR of 50 (left) and 1 (right). (C) Absolute prevalence reduction due to prophylaxis (mean prevalence using AL - mean prevalence using DP) depends on both coverage and EIR.

Figure 5

Primaquine reduces transmission when campaign coverage is very high. (A) At a coverage of 70%, addition of PQ to an AL MDA reduces prevalence 4 months post campaign by <5%. EIR is sampled at constant coverage for 100 stochastic realizations for each EIR. Compliance was 100% and 3 rounds of campaigns were conducted. Mean +/− one standard deviation is shown for each EIR value. A linear regression is shown (dotted line) correlating prevalence after MDA with AL + PQ to prevalence after a 3-round MDA with AL alone. (B) Prevalence reduction due to addition of PQ is highly dependent on coverage. The fractional relative prevalence reduction upon addition of PQ is the slope of the linear regression of prevalence with PQ to prevalence without PQ (panel A) subtracted from 1. Lines: bootstrapped mean fractional relative prevalence reduction at each level of coverage. Shaded areas: 95% confidence interval. (C) Higher coverage with ACTs and addition of primaquine lead to higher probability of local elimination, especially at higher EIR. Fraction of stochastic realizations leading to eradication is shown for 4 bins of annual EIR at each level of coverage, with 100 realizations per EIR per coverage. Compliance was 100% and 3 rounds of campaigns were conducted. The height of each bar is normalized to the total number of realizations in each EIR bin.