COS 183-3 - A population-dynamics approach to evaluate the impact of anti-relapse treatment on epidemic Plasmodium vivax malaria

Friday, August 10, 2012: 8:40 AM
D139, Oregon Convention Center
Manojit Roy, Ecology & Evolutionary Biology, University of Michigan, and Howard Hughes Medical Institute, Ann Arbor, MI, Menno J. Bouma, Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, University of London, London, United Kingdom, Edward L. Ionides, Statistics, University of Michigan, Ann Arbor, MI, Ramesh Dhiman, National Institute of Malaria Research, Delhi, India and Mercedes Pascual, Ecology and Evolutionary Biology, University of Michigan,Howard Hughes Medical Institute, Santa Fe Institute, Ann Arbor, MI
Background/Question/Methods

While Plasmodium falciparum malaria commands much attention because of high human fatality in sub-Saharan Africa, the comparatively “benign” Plasmodium vivax malaria continues to exact a significant toll on 40% of the world population, and is increasingly becoming a major public health concern primarily in south-east Asia. Unlike falciparum malaria, clinical interventions by the interruption of blood-stage transmission is ineffective in P. vivax because of the liver-stage hypnozoites, a dormant phase of the parasite's life cycle that can trigger relapses months after the original infection has subsided. The current “radical cure” for vivax malaria combines blood-stage treatment with primaquine (PQ), a hypnozoitecide in use for over 50 years despite impaired efficacy from drug resistance, poor patient compliance and safety concerns.

Considering the importance of relapses as a hidden reservoir for vivax malaria, especially in epidemic areas with intermittent transmission, the public health implications of using an improved anti-relapse drug are poorly understood due to insufficient tools for quantifying its population-wide benefits. To address this problem we adopt a population dynamics approach as follows. We formulate a dynamical model of P. vivax transmission, and use recently developed likelihood-based inference methods to parametrize it from time series surveillance data from an epidemic region in NW India. We then use the maximum-likelihood model to quantify the fraction of relapses that require effective treatment in order to suppress, and even eliminate, vivax malaria in the study area.

Results/Conclusions

Our model gives maximum likelihood estimates for mean latency and relapse rate of 7.1 months and 31% respectively, which correspond closely to values from clinical evaluation studies in the area. Given the prevailing treatment practices in India as a baseline, the model predicts that a successful treatment of 30-40% of relapses alone would drastically reduce disease burden within a decade, and result in extinction of the parasite species in 25 years. Furthermore, the impact of relapse treatment is inversely related with preceding transmission intensity, which suggests a cost-effective policy of selectively applying treatment following years of low transmission if resources are limited and the development of resistance becomes a concern.

The sensitive dependence of P. vivax on relapses can inform the debate on how to more effectively control the disease. Our approach can provide effective means of evaluating population-wide impacts of relapse treatment in some regions, and should encourage the development of better alternatives to PQ in the light of renewed enthusiasm for global malaria eradication.