Simpson JA, Zaloumis S, DelIvera AM, Price RN, McCaw JM, Making the most of clinical data: reviewing the role of pharmacokinetic-pharmacodynamic models of anti-malarial drugs, AAPS Journal 16(5): 962-974 (2014) doi: 10.1208/s12248-014-9647-y
Mechanistic within-host models integrating blood anti-malarial drug concentrations with the parasite-time profile provide a valuable decision tool for determining dosing regimens for anti-malarial treatments, as well as a formative component of population-level drug resistance models. We reviewed published anti-malarial pharmacokinetic-pharmacodynamic models to identify the challenges for these complex models where parameter estimation from clinical field data is limited. The inclusion of key pharmacodynamic processes in the mechanistic structure adopted varies considerably. These include the life cycle of the parasite within the red blood cell, the action of the anti-malarial on a specific stage of the life cycle, and the reduction in parasite growth associated with immunity. With regard to estimation of the pharmacodynamic parameters, the majority of studies simply compared descriptive summaries of the simulated outputs to published observations of host and parasite responses from clinical studies. Few studies formally estimated the pharmacodynamic parameters within a rigorous statistical framework using observed individual patient data. We recommend three steps in the development and evaluation of these models. Firstly, exploration through simulation to assess how the different parameters influence the parasite dynamics. Secondly, application of a simulation-estimation approach to determine whether the model parameters can be estimated with reasonable precision based on sampling designs that mimic clinical efficacy studies. Thirdly, fitting the mechanistic model to the clinical data within a Bayesian framework. We propose that authors present the model both schematically and in equation form and give a detailed description of each parameter, including a biological interpretation of the parameter estimates.