A mixture distribution approach to IVIVC modeling of a dual component drug delivery system
Stefaan Rossenu (1), Clare Gaynor (2), An Vermeulen (1), Adrian Dunne (2) and Adriaan Cleton (1)
(1) Johnson and Johnson Pharmaceutical Research & Development, a division of Janssen Pharmaceutica, Beerse, Belgium (2) UCD School of Mathematical Sciences, University College Dublin, Belfield, Dublin 4, Ireland
Objectives: Modification of a non-linear mixed effects modeling approach(1) to establish an in vitro-in vivo correlation (IVIVC) for a dual component drug delivery system and to allow the simultaneous analysis of more than one formulation.
Methods: Four controlled release (CR) formulations of galantamine were used to test this extended IVIVC model. The fraction of each CR dosage unit dissolved at any given time in vitro was described as a mixture of two distributions. A two-compartment pharmacokinetic model with a mixture of two inputs was used to describe the in vivo data. The proportion initially released as a loading dose (25%) was one of these inputs, while the second input corresponded to the remaining part released as CR component (75%). The CR fraction of the dose was modeled using a modified version of the method described by O'Hara et al(1). Three of the four formulations (slow, medium (target) and fast) were used for the development of the IVIVC model, which was used to predict plasma concentration-time profiles for each subject following administration of all four formulations. The AUC and Cmax were calculated for the observed and predicted profiles and used to compute the percentage prediction errors in accordance with the FDA validation guideline(2). Internal predictability was assessed using the three formulations included in the model development, while the fourth was used to similarly evaluate external predictability. The model fitting was performed using NONMEM(3).
Results: The observed and predicted in vitro dissolution and in vivo plasma concentration-time profiles demonstrated a good model fit. The calculated prediction errors for both the internal and external predictability comfortably met the FDA criteria.
Conclusion: A mixture distribution-based model was developed to describe a dual component drug delivery system. This model successfully captured the IR and CR elements of a capsule. The technique was tested on four formulations of galantamine and an IVIVC model was established that meets the FDA criteria for internal and external predictability.
References:
[1] O'Hara T. et al.. J. Pharmacokinetics and Pharmacodynamics 2001;28:277-298.
[2] FDA's Guidance for Industry: Extended release oral dosage forms: development, evaluation, and application of in vitro/in vivo correlations, 1997
[3] Beal SL and Sheiner LB. NONMEM User's Guides, NONMEM Project Group, University of California, San Francisco, 1992.
