The capability of several population-based approach software to analyze sparse drug plasma concentration data after intravenous bolus injection

Abstract

Monolix, NONMEM, and WinBUGS-PKBUGS are among the available software packages for population-based modeling. The sparse-data of drug plasma concentration versus time (Cp-time) is prevalent in clinically based studies involving patients. It is not ethical in this case, to collect many and large volumes of blood samples. This study was aimed to simulate the capability of Monolix, NONMEM, and WinBUGS-PKBUGS to analyze very sparse Cp-time data after an intravenous bolus drug administration and to estimate the minimum number of Cp-time data required for an adequate analysis. Data of Cp-time were obtained based on simulation using the pharmacokinetic one-compartment open model following an intravenous bolus administration of 50 mg of a hypothetical drug. In this respect, six random values of k (rate constant of elimination) and Vd (volume of distribution) with mean and standard deviation values of 0.3±0.1/h and 30±10 L, respectively, were used to create simulated Cp-time data of 6 subjects. Simulated Cp-time data in each subject were randomly ranked to choose data based on the intended number of samples in each subject. Several sparse Cp-time data scenarios, starting from a minimal state, i.e., with a total of 6 Cp-time data (1 datum per subject) to a rich-data with 48 Cp data-points (8 data per subject), were examined. The goodness-of-fit evaluations, as well as the similarity of individual values of k and Vd to the respective real values (p>0.05), indicate that nonlinear-mixed-effect-model using Monolix, NONMEM, and WinBUGS-PKBUGS can appropriately describe sparse Cp-time data even with only 2 data per subject. This fact is an important finding to support the demand of analytical tool for a limited number of Cp-time data such as obtained in the therapeutic drug monitoring event.