Development, Evaluation, and Application of an In Silico Model for Antimalarial Drug Treatment and Failure

Abstract

Pharmacological mechanism-based modeling was refined and used to develop an in silico model of antimalarial drug treatment validated against clinical and field data. We used this approach to investigate key features of antimalarial drug action and effectiveness, with emphasis on the current generation of artemisinin combination therapies. We made the following conclusions. (i) The development of artemisinin tolerance and resistance will, unless checked, have an immediate, large impact on the protection afforded to its partner drug and on the likely clinical efficacy of artemisinin combination therapies. (ii) Long follow-up periods are required in clinical trials to detect all drug failures; the follow-up periods of 28 days recommended by the World Health Organization are likely to miss at least 50% of drug failures, and we confirmed recent suggestions that 63 days would be a more appropriate follow-up period. (iii) Day 7 serum drug concentrations are a significant risk factor of failure, although, paradoxically, receiver operating characteristic curve analysis revealed that their predictive power is relatively poor. (iv) The pharmacokinetic properties of the partner drugs in artemisinincontaining combination therapies are the most important determinants of treatment outcome, particularly the maximum killing rate. We discuss the assumptions made in such modeling approaches and how similar approaches may be refined in future work.

Making the Most of Clinical Data: Reviewing the Role of Pharmacokinetic-Pharmacodynamic Models of Anti-malarial Drugs

“…Figure 1c shows the resulting relationship between k drug and time. Recently, this single ODE model has been employed to evaluate the effectiveness of the ACTs, in particular, as artemisinin resistance spreads (5,7,20).…”

“…2. Continuous-time within-host pharmacokinetic-pharmacodynamic models: a Total parasite burden models (5,7,13,15,18,19,20); b Stage-specific models (11). Note, k death was assumed to be determined by the killing action of the anti-malarial and host immunity; c Stage-specific models (10,17,21).…”

“…Some of these mechanistic PK-PD models have been fitted formally to patient parasite-time profiles to obtain parameter estimates (10,11,14,16,17,21,22), whereas others have simply selected parameter values that generate parasitetime profiles that accord well with the profiles from clinical studies (5)(6)(7)9,12,13,15,(18)(19)(20). Both of these approaches are dependent on empirical data defining the dynamics of the parasite burden.…”

“…Mechanistic PK-PD models for malaria characterize the relationship between the anti-malarial drug concentrations in the blood (the pharmacokinetic (PK) data) and the parasite-time profile (pharmacodynamic (PD) response) within an individual host. Such quantitative models can aid decision making regarding the choice of dosing schemes for current ACTs and new anti-malarials, which can then be selected for evaluation in clinical studies (5)(6)(7). These models can be incorporated in population-level analyses to predict the selection and spread of drug resistance (8).…”

“…These models can be incorporated in population-level analyses to predict the selection and spread of drug resistance (8). A variety of within-host PK-PD models have been described for falciparum malaria, and there is a wide variation in the number of biological parameters describing the parasite life cycle and PD effect of the anti-malarial treatment (5)(6)(7)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). The accuracy of these models in predicting the change in parasite burden following treatment depends on many factors.…”