Developing age-based dosing regimens for antimalarials

Introduction. Age-based dosing of antimalarials is a widely used alternative to weight-based dosing practices. However, it is rarely taken into account in the drug development phase and standardized methods to devise age-based proxies that optimize the effectiveness and safety of antimalarial drugs do not exist. This has contributed to the variation in existing age-based regimens for antimalarials, at times resulting in poor, but widely-used regimens. Inaccurate dosing poses a threat to the individual (treatment failure, adverse effects) and the population (emergence and spread of resistance). Objectives. To address the lack of standardization and optimization in the development of age- based dosing regimens for antimalarials through the creation of regional weight-for-age growth ,,;,. .... references and the development and testing of a tool that uses these r~rer;JCes to calculate optimized age-based dosing regimens to inform drug developers and policy makers. Methods. To meet the objectives the following three steps were taken. First, a population representative reference data set was compiled. Secondly this data set was used to model regionally representative growth references. Finally these references were applied in a modelling tool to optimise age-based dosing regimens. Together with a team of expert statisticians, I developed a method to model multi-source anthropometric data to generate regional weight-for-age reference curves. I compiled a weight-for-age reference database from malaria-endemic countries using population representative data from health surveys and individual studies. An extension of the new generalized additive model for location, scale and shape (GAMLSS) was developed to generate smoothed country level curves, interpolating missing data from adjacent age categories and neighbouring countries with similar weigh-for- age distributions. These were combined in a finite mixture model weighted by population size or by population at risk of malaria to obtain regional weight-for-age growth references. In combination with drug specific parameters such as the therapeutic dose range and tablet strength, and regimen specific criteria such as the number of age categories and the use of tablet fractions, I determined age cut-offs that would result in regimens with the lowest number of patients receiving drugs outside the therapeutic range. The tool was initially used to support the development of age-based dosing regimens for the new fixed-dose combination of artesunate+mefloquine for Africa, Latin America and the Asia-Pacific regions and for specific countries (i.e. Cambodia and Brazil). I then evaluated the predicted dosing accuracy of all age- based ACT regimens recommended in the 2nd edition of the WHO malaria treatment guidelines; artemether+lumefantrine (AL), dihydroartemisinin+piperaquine (DHA+PPQ) and artesunate (AS) plus amodiaquine (AQ), mefloquine (MQ) or sulfadoxine-pyrimethamine (SP). Alternative age- thresholds were evaluated to further optimize the regimens. Results. Robust region specific weight-for-age references were created using >900.000 measurements from 167 data sources and 67 countries. The models accurately predicted observed regional weight-far-age distributions. Significant inter-regional variation existed in growth patterns (e.g. delayed growth spurt in Asia compared to Africa and Latin America) and attained growth. Dosing accuracies varied greatly by age and drug. ACT regimens with narrow therapeutic ranges (including MQ, AL, PPQ,) likely result in considerable dosing outside WHO recommended dose ranges. Suboptimal drug ratios in fixed-dose drug combinations contribute to low dose accuracies. Conclusions. Our reference and regimen modelling tools for the design of age-based dosing recommendations provide drug developers and policy makers with a powerful decision-support tool to optimize the safety and effectiveness of antimalarials dosed by age. The methods can further be extrapolated to other drugs from the WHO essential medicines list, and can be improved by incorporating pharmacokinetic data and safety data from post marketing surveillance when they become available. This work was supported by grants from the Liverpool School of Tropical Medicine, the British Medical Research Council, the European and Developing Countries Clinical Trials Partnership and the Drugs for Neglected Disease initiative.EThOS - Electronic Theses Online ServiceGBUnited Kingdo