The selection of sample matrix (e.g. venous blood, capillary blood, plasma, breast milk, placenta, urine) will depend on the drug and the study context and objectives. Capillary blood of fixed volume collected into calibrated capillary tubes and dried on to filter paper is the most feasible sample for use in field studies, and such assays have been validated for chloroquine, quinine, halofantrine, piperaquine, amodiaquine, sulphadoxine/pyrimethamine, atovaquone, mefloquine and proguanil [7, 17–23]. The accuracy of these assays depends on the filter paper used (and its pre-treatment, when needed) and relies on careful training to ensure, for example, an accurate volume of sample, avoiding diluting the sample with interstitial tissue fluids by excessive squeezing of the patient's finger, and preventing contamination by touching the filter paper after touching the study drug. A simple field guide providing simple SOPs is needed to optimize sample collection and storage. Where protein binding is substantial, it is ideal to measure both total and free drug, although this is not possible for whole blood spots dried onto filter paper. Furthermore, the concentration of free drug for drugs with substantial protein binding (i.e. >99%) is very low and will thus in many cases be limited by assay sensitivity. There are currently no validated assays for capillary blood spots on filter paper for pyronaridine, lumefantrine or the artemisinin derivatives.
The selection of sampling times depends on the pharmacokinetic properties, particularly the elimination half-life, of each antimalarial drug, the context of the study and the therapeutic objective of treatment. Although time to reach therapeutic concentrations is of critical importance in treating severe malaria, for the treatment of uncomplicated malaria, an adequate clinical and parasitological response requires the persistence of levels that exceed the minimum inhibitory concentration (a concentration resulting in a parasite multiplication rate of <1/cycle) for at least four 48-hour asexual blood cycles in non-immune patients. The day 7 drug concentration has been shown to be the most important single concentration, in terms of correlation with the area under the concentration time curve and or association with treatment response, for lumefantrine, piperaquine, quinine, mefloquine and sulphadoxine-pyrimethamine [7, 17, 24–27]. As day 7 is a routine visit in therapeutic efficacy studies , measuring drug concentrations at least at this time point is recommended for all medium- to long acting drugs (i.e. drugs with half-lives greater than 12 hrs). This requires drug assays that are sensitive enough to quantify antimalarial drug levels accurately at least until this time point. Treatments with shorter elimination half-lives require more frequent earlier sampling. Further pharmacokinetic studies are needed to define optimal sampling times for the treatment of P. vivax (and other non-falciparum species).
The respective roles of conventional and population pharmacokinetic approaches need to be considered when selecting timing of sample collection. Intensive sampling is needed initially for each antimalarial and in each important target population. Subsequently, population pharmacokinetics can be used to minimize the number of times blood needs to be collected from an individual patient, thereby allowing more patients to be included in the study. Population pharmacokinetic analysis is complex and time consuming, taking between six and 12 months per drug. Extensive capacity building (particularly training in statistical modelling) is needed for population pharmacokinetic modelling to be more widely used, particularly in malaria endemic countries. Occasionally, the population pharmacokinetic approach will not succeed in modelling the pharmacokinetics of a particular antimalarial. Standardization in reporting will also be needed. Given the skew distribution of most pharmacokinetic parameters, these should preferably be reported as medians, inter-quartile ranges and ranges. Although summary measures are most often reported, the extremes of the range are more important in exploring the pharmacokinetic determinants of treatment failure and toxicity. This creates need for accurate characterization of the variability in of pharmacokinetic parameters.