A physiologically based population pharmacokinetic model describing the non-linear disposition and blood distribution of indisulam in Caucasian and Japanese patients
A.S. Zandvliet(1), S. Yasuda(2), W. Copalu(3), J.H.M. Schellens(4,5), J.H. Beijnen(1,5), A.D.R. Huitema(1)
(1)Department of Pharmacy & Pharmacology, Slotervaart Hospital/The Netherlands Cancer Institute, Amsterdam, The Netherlands; (2)Eisai Ltd., Tokyo, Japan; (3)Eisai Ltd., London, UK; (4)Department of Medical Oncology, Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands; Department of Biomedical Analysis, Section of Drug Toxicology, Utrecht University, Utrecht, The Netherlands
Objectives: The pharmacokinetic profile of the investigational anti-cancer agent indisulam is nonlinear. This may partially be explained by its saturable distribution to erythrocytes and the saturable plasma protein binding. The aim of the current study was to develop a mechanistic population pharmacokinetic model for Caucasian and Japanese patients, which adequately describes the pharmacokinetic profile of indisulam in various physiological compartments and which provides a basis for rational PK-PD relationships. The model allows unraveling of various physiological processes and their contribution to the pharmacokinetics of indisulam. A simulation study was performed to examine the role of distribution to erythrocytes and protein binding in indisulam pharmacokinetics.
Methods: Pharmacokinetic sampling was carried out in Caucasian and Japanese patients after administration of indisulam in several phase I studies at a large range of dose levels. The population PK analysis was performed using NONMEM (version V, level 1.1). The first order (FO) method was applied throughout to fit logarithmically transformed data. Concentrations in plasma, plasma ultrafiltrate and in erythrocytes were used to develop a physiological population pharmacokinetic model in which both the distribution of indisulam (in plasma, erythrocytes, interstitial fluid and tissue) and its elimination were integrated. Established physiological knowledge regarding distribution volumes of the different compartments was incorporated in the model a priori. A previously developed Langmuir model for the in vitro saturable binding of indisulam to plasma proteins was verified in vivo. In order to describe the distribution of indisulam to erythrocytes, saturable Langmuir models with or without a non-specific binding component were considered. Linear and saturable models were tested to describe tissue binding and drug elimination. Correlations between individual PK parameter estimates and various demographic, hematological and blood chemistry variables were investigated to identify relevant covariate relationships. The model was evaluated by the performance of sensitivity tests, the jack-knife procedure and likelihood profiling. Simulations were performed to explore the impact of hematocrit and plasma albumin level on the disposition of indisulam.
Results: The backbone of physiological parameters proved to be an adequate starting point while it was amplified with drug-specific parameters. The maximal protein binding capacity in plasma corresponded with individual plasma albumin concentrations and the equilibrium dissociation constant was 0.25 (±0.012) mg/L. Binding to erythrocytes was best described by a two site binding model, comprising one class of saturable binding sites and one class of non-specific binding sites. The population value of the maximal specific erythrocyte binding capacity was 59.0 (± 4.2) mg/L and was similar to typical erythrocyte carbonic anhydrase concentrations. Tissue distribution was represented by a linear and a saturable component and was different for Caucasian and Japanese patients. Free indisulam was cleared from the plasma compartment through a Michaelis Menten pathway (major route of elimination) and a linear first order process. The Michaelis Menten elimination constant (Km) for free indisulam was 1.07 (±0.17) ug/L. Interindividual differences in the maximal elimination rate were partially explained by differences in body surface area. Simulation studies demonstrated that the hematocrit does not have a clinically relevant impact on indisulam disposition. A decreased plasma albumin level, however, considerably influenced total plasma concentrations of indisulam, whereas free concentrations in plasma, concentrations in erythrocytes and the amount distributed to tissue remained relatively unaffected.
Conclusion and discussion: The presented physiological population pharmacokinetic model allowed adequate prediction of the time profiles of indisulam concentrations in Caucasian and Japanese patients in all monitored compartments for a large range of dose levels and several treatment regimens. This project has led to a better understanding of the physiological mechanisms behind the pharmacokinetics of indisulam. This improved insight has facilitated the elucidation of the important impact of the plasma albumin level on indisulam disposition. Simulations have illustrated that total plasma concentrations were highly dependent on the plasma albumin level. Total plasma concentrations may therefore not be a preferable target in pharmacokinetic studies of indisulam. Alternatively, indisulam concentrations in erythrocytes may be well suitable to define PK-PD relationships. This finding may support the development of new rational PK-PD relationships.