Application of a semi-physiological model describing time-varying pharmacokinetics to support optimal clinical study design
J. G. Coen van Hasselt (1), Bruce Green (2), Glynn A. Morrish (3)
(1) Department of Clinical Pharmacology, The Netherlands Cancer Institute, Amsterdam, The Netherlands; (2) Model Answers Pty Ltd, Brisbane, Australia; (3) Goodness of Fit Pty Ltd, Brisbane, Australia
Introduction: Physiological changes during pregnancy can alter drug pharmacokinetics1 (PK). Precise quantification of PK changes during pregnancy is therefore required if drugs are to be administered to expectant mothers, although design of such experiments can be difficult. D-optimal design is one method that can be used to identify sampling schedules that ensure precise parameter estimation. However, prior models that account for time-varying PK during pregnancy are unlikely to exist challenging the use of D-optimality.
Objectives: The objectives of this work were to a) develop a semi-physiological PK model from existing literature, which described the time-varying PK of enoxaparin during pregnancy; b) use this model to determine an informative sampling design for a prospective clinical study quantifying the PK of enoxaparin during pregnancy.
Methods: The physiological variables total body water and creatinine clearance were selected as predictors for pregnancy-related changes in volume of distribution and clearance. The mean change in these physiological variables was linked to a population PK model for enoxaparin in non-pregnant females.
The semi-physiological model was then used to simulate enoxaparin concentrations throughout pregnancy, which were compared concentrations simulated from an empirical model2 that described the population PK of enoxaparin during pregnancy.
The semi-physiological model was then used to develop an optimal sampling design using WinPOPT3, which was evaluated by means of simulation using the previously published empirical model2 and subsequent re-estimation.
Results: The simulated concentration-time profiles from the semi-physiological model were comparable with the profiles of the previously published model2. A D-optimal design was successfully developed using the semi-physiological model. This design resulted in precise parameter estimation using data simulated from the true model. The need for a changing design over pregnancy was not evident, most likely due to the magnitude of change in PK parameters not being large enough.
Conclusions: This work demonstrates how relevant physiological variables published in the literature can be used to make informed predictions on the PK of enoxaparin during pregnancy. Moreover, we have demonstrated that a literature based semi-physiological model can be used to support development of informative sampling designs.
References:
[1]. Anderson GD, Clinical pharmacokinetics (2005); 44:989-1008
[2]. Lebaudy C et al, Clinical Pharmacology and Therapeutics (2008); 84:370-377
[3]. Duffull SB et al, WinPOPT user guide version 1.2 (2008)