Modelling non-linear dose-dependent absorption profiles after oral prolonged release formulations
L. Liefaard (1), D. Huntjens (2), A. Vermeulen (2) & H. Drenth (1)
(1) LAP&P Consultants BV, Leiden, The Netherlands; (2) J&J Pharmaceutical Research and Development, a Division of Janssen Pharmaceutica N.V., Beerse, Belgium
Objectives: After oral administration of prolonged release tablets of a new candidate drug compound, complex absorption profiles were found that showed high inter-individual variability and multiple peaks. In addition, dose dependencies in Cmax but not in AUC were observed after initial exploratory analysis. The objective of the study was to develop a population PK model that was able to describe the complex absorption adequately.
Methods: Data from five phase 1 studies were included in this population PK analysis. 140 patients were dosed (SD) with 50-250 mg of a prolonged release formulation of a new candidate drug compound. 26 subjects took the study drug after a high-fat, high-calorie breakfast (non-fasted conditions), 84 subjects took the study drug under fasted conditions, and 30 subjects took the study drug under fasting and non-fasting conditions in a crossover design. Data for all dose strengths are available under fasted conditions, and for the 250 mg dose only under non-fasted conditions.
The absorption process was explored using deconvolution of the raw data, in which the rate of absorption was estimated for arbitrarily chosen small time frames. For deconvolution, all disposition parameters were fixed to the values obtained from data analysis after administration of an immediate release formulation of the compound and bioavailability was set to 1. Nonmem 6.2 was used for model building.
Results: Deconvolution displayed two peaks in the absorption profiles of which one showed dose dependency. Two absorption routes were used to describe the absorption process. A buffer compartment with first order absorption (ka1) was used to describe the immediate dose-independent absorption peak. A time-dependent first order rate constant (ka2) with saturation was required to describe the slower, dose-dependent absorption peak. Furthermore, ka2 appeared to be doubled for non-fasted subjects. Drug disposition was described by a 1-compartment model with linear elimination from the central compartment.
Conclusions: Deconvolution of the initial raw data is a valuable tool in assessing the components of the absorption process including detection of the dose-dependent process. The data were well described after including a mathematical description of these processes in the proposed PK model.
