A 1-step Bayesian predictive approach for evaluating in vitro in vivo correlation (IVIVC)

Area of Science:

• Pharmacokinetics
• Drug Development
• Biostatistics

Background:

• In vitro-in vivo correlation (IVIVC) methods are crucial for drug formulation changes, often relying on in vivo absorption estimations.
• Current IVIVC models typically use convolution or deconvolution, which can be mathematically complex and unstable.
• These methods estimate in vivo absorption from in vitro dissolution data, introducing potential inaccuracies.

Purpose of the Study:

• To introduce a new Bayesian approach for evaluating IVIVC.
• To eliminate the need for convolution, deconvolution, or approximations in IVIVC calculations.
• To provide a method that explicitly accounts for variability in drug absorption and dissolution measurements.

Main Methods:

• A Bayesian statistical framework was developed for IVIVC analysis.
• The method directly models in vivo plasma concentration as a function of in vitro dissolution percentage.
• It incorporates both between- and within-subject variability without assuming asymptotic normality.

Main Results:

• The proposed Bayesian method successfully evaluated IVIVC for a controlled-release formulation.
• It bypasses the need for complex mathematical operations like convolution and deconvolution.
• Key pharmacokinetic parameters (AUC, Cmax, Tmax) can be derived from dissolution percentage, with explicit uncertainty quantification.

Conclusions:

• This Bayesian approach offers a more stable and direct method for establishing IVIVC.
• It facilitates drug formulation adjustments based solely on in vitro dissolution data, potentially reducing the need for bioequivalence studies.
• The method explicitly addresses sources of variability, leading to more robust IVIVC evaluations.