Comparison of Alternative Population Modeling Approaches for Implementing a Level A IVIVC and for Assessing the...

Area of Science:

• Pharmacokinetics and Pharmacodynamics
• Drug Development and Formulation
• Biopharmaceutics

Background:

• Establishing in vitro-in vivo correlation (IVIVC) is crucial for drug development.
• Existing methods include deconvolution and convolution analyses.
• Time-scaling is a key factor in complex drug release processes.

Purpose of the Study:

• To implement and compare time-scaled deconvolution and convolution-based approaches for IVIVC assessment.
• To evaluate method performance using population modeling (NONMEM, R).
• To investigate the impact of pharmacokinetic (PK) variability and nonlinearity on IVIVC.

Main Methods:

• Utilized a novel convolution-based model for time-scaled IVIVC.
• Employed Levy's plots for deconvolution-based time-scaling.
• Nonlinear regression determined time-scaling in the convolution-based model.
• Assessed prediction error (%PE) for Cmax, AUC, and partial AUC.

Main Results:

• Both methods met mean %PE criteria, but deconvolution-based approach showed significant impact of inter-individual variability (IIV) on Cmax %PE, potentially failing validation.
• Complex nonlinear PK disposition models resulted in larger Cmax %PE with the deconvolution-based method.
• Convolution-based approach demonstrated more robust performance under challenging PK scenarios.

Conclusions:

• The convolution-based approach is recommended for time-scaled IVIVC assessment due to its better performance with increased IIV and nonlinear PK.
• Deconvolution-based methods may be less reliable for IVIVC validation under complex in vivo drug release conditions.
• Population modeling provides a robust framework for implementing and evaluating IVIVC methods.