Challenges in modelling the pharmacokinetics of isoniazid in South African tuberculosis patients
Justin J Wilkins (1, 2), G Langdon (1), H McIlleron (1), Goonaseelan Pillai (3), Peter J Smith (1), Ulrika S H Simonsson (2)
(1) Division of Clinical Pharmacology, Department of Medicine, University of Cape Town, South Africa; (2) Division of Pharmacokinetics and Drug Therapy, Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (3) Clinical Modelling & Related Technologies, Clinical Development & Medical Affairs, Novartis Pharmaceuticals AG, Basel, Switzerland
Objectives: The modelling of isoniazid was part of a larger project investigating the interindividual and interoccasional variability in the pharmacokinetics of drugs for the treatment of tuberculosis in South African pulmonary tuberculosis patients. The pharmacokinetics of isoniazid in this group exhibited a number of unusual features, which made model-building interesting. This talk will focus on the specific issues involved in modelling the drug.
Methods: A prospective study was planned and executed in 91 pulmonary tuberculosis patients, using a design involving sparse data on multiple occasions. The resulting isoniazid concentration-time measurements were pooled with rich data from 175 pre-existing pulmonary tuberculosis patients from two previously executed studies. Population pharmacokinetic modelling using NONMEM was carried out to characterize the variability-related characteristics of the drug. During model development, models were selected by graphical analyses of residuals and predictions using NONMEMory, a software utility developed by the first author, as well as by comparing the relative standard errors of parameter estimates. Differences in the objective function value (OFV) were used to discriminate between hierarchical nested models.
Results: The pharmacokinetics of isoniazid in the studied population had a number of unusual characteristics. Firstly, the well-known polymorphism in the acetylation of the drug was clear, as two subpopulations separated by a 2.5-fold difference in clearance were identified. In addition, a subset of patients exhibited sharply reduced and delayed bioavailability in comparison to the rest of the population. Another subpopulation showed significantly different absorption characteristics in comparison with the rest of the group. Enterohepatic circulation has previously been shown in the literature and was evident in the present data. Finally, interoccasion variability, as the primary objective of the project, was investigated for the different parameters on 27 discrete dosing occasions. The final model described the data well, and dealt with the bimodal absorption by using dual absorption compartments assigned differing proportions of total bioavailability. Interoccasion variability on apparent clearance and absorption half-life was included, and acetylator phenotype was addressed through the use of a mixture model. It was not possible to include enterohepatic circulation, since it was not supported by the data.
Conclusions: The modelling process was directed towards addressing the primary purpose of the study, interindividual and interoccasion variability in isoniazid pharmacokinetics. While enterohepatic circulation and the observed malabsorptive syndrome were undeniably interesting and clinically relevant, they were unsupported by the data, and therefore not included in the model. The final model answered the questions posited by the study and described the data well, and was therefore appropriate.