Developing an In Vitro – In Vivo Correlation Model Using a Population Approach in NONMEM
Claire Gaynor (1), Marylène Gaignet (1), Marylore Chenel* (1)
(1) Clinical Pharmacokinetics, Institut de Recherches Internationales Servier, France;
Introduction: In vitro drug dissolution-time profiles can be used to predict in vivo plasma drug concentration-time profiles if an appropriate mathematical model can be found to describe the In vitro - In vivo Correlation (IVIVC) defined by the FDA . Such models are beneficial in terms of both time and cost savings as in vitro dissolution fractions can be used as a surrogate for clinical bioequivalence studies during the initial drug development process as well as with scale-up and post approval changes. Traditional methods are suboptimal under certain conditions [2,3]. A modelling approach based on differential equation system (DES) has been proposed [4,5].
Objectives: The aim is to illustrate, using clinical data, the implementation of a population approach to the development of an IVIVC model in NONMEM.
Methods: This study used observed plasma concentration profiles of 12 healthy volunteers following cross-over oral administration of immediate release (IR) and modified release forms as well as 6 in vitro dissolution profiles of molecule S. Three model-building steps were required. First a function, f, describing in vitro dissolution was selected. Secondly, a population PK model was fit to the IR data. Lastly, a model correlating the in vitro and in vivo dissolution-time profiles was developed, taking f and individual PK parameters from the IR model into account. Therefore, in vivo individual plasma concentration profiles were predicted directly in a one-step process from in vitro dissolution data.
The percentage prediction errors (%PE) on observed and predicted Cmax and AUC were calculated to evaluate the predictability of the model.
Results: A Gompertz function was used to describe in vitro dissolution. Individual PK parameters were taken from a 2-compartment IR model (with IIV on clearance, central volume, absorption rate constant and lag time). A cubic polynomial was used to describe the relationship between in vitro and in vivo dissolution. This IVIVC model, with separate in vitro and in vivo residual error models and variability between tablets and subjects, enabled the prediction of individual in vivo plasma concentration profiles. The %PE observed were less than 15%..
Conclusions: This work illustrates the application of a compartmental model-based Pop-PK approach to IVIVC model building, and provides a reliable method to help the initial drug development process.
 Food and Drug Administration (1997) Guidance for Industry: Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In vitro/In vivo Correlations.
 Dunne A., Gaynor C. and Davis J. (2005) Deconvolution Based Approach for Level A In vivo-In vitro Correlation Modelling: Statistical Considerations. Clinical Research and Regulatory Affairs, 22, 1-14.
 Gaynor C., Dunne A. and Davis J. (2008) A Comparison of the Prediction Accuracy of Two IVIVC Modelling Techniques Journal of Pharmaceutical Sciences, 97:8, 3422-3432.
 Buchwald, P. (2003) Direct, Differential-equation-based In-Vitro-In-Vivo Correlation (IVIVC) Method. Journal of Pharmacy and Pharmacology, 55: 495-504.
 Gaynor C., Dunne A., Costello C. and Davis, J. (In press) A Population Approach to In vitro - In vivo Correlation Modelling for Compounds with Nonlinear Kinetics. Journal of Pharmacokinetics and Pharmacodynamics.