Novel extended IVIVC combined with DoE to predict pharmacokinetics from formulation compositions

The objective of this study was to develop a novel extended in vitro in vivo correlation (IVIVC) model combined with design of experiment (DoE) that integrates the DoE into IVIVC, which can predict the pharmacokinetics of sustained-release (SR) tablets from their formulation compositions, and vice versa. To develop the extended IVIVC model, ketoprofen was used as a model drug. Nineteen types of ketoprofen SR tablets with different formulation compositions were prepared based on the mixture design and used to derive mathematical relationships between the formulation composition and the in vitro dissolution profiles for DoE. The predictability of the DoE equation was externally validated by using additional seven types of SR formulations with prediction errors (%PE) of less than 11.45%. For the development of IVIVC model, three SR formulations that have fast, medium, and slow drug-releasing rates were selected, and the in vivo pharmacokinetics were assessed in Beagle dogs. The pharmacokinetic properties of ketoprofen SR tablets were described by a population pharmacokinetics (POP-PK) model which incorporated the pH-dependent dissolution of ketoprofen by a time-dependent Hill-type equation. The final POP-PK model could describe the overall in vivo pharmacokinetic profiles and allowed estimation of the in vivo dissolution parameters. The POP-PK model estimated in vivo dissolution parameter, K_{diss, in vivo} were then correlated with the in vitro dissolution parameter, K_{diss, in vitro} by linear regression (R² = 0.9989), establishing IVIVC. Finally, the equation derived from DoE was introduced to the IVIVC model to develop the extended IVIVC, which connects the formulation composition, in vitro dissolution, and in vivo pharmacokinetic profiles. The average %PE of the final extended IVIVC model was 4.24% for C_max and 4.46% and AUC. Finally, the final extended IVIVC was applied to predict the in vivo PK profiles of SR tablets from their formulation compositions as well as to design the optimal formulation to achieve certain target PK profiles. The %PE of the final extended IVIVC model was less than 14.67% for C_max and 12.41% for AUC, satisfying the FDA criteria of conventional IVIVC. The present extended IVIVC model may provide a useful tool towards rationalized design and development of new SR formulations.