Generic physiologically based pharmacokinetic (PBPK) models for rats and humans were developed using in vitro absorption, distribution, metabolism, and excretion (ADME) data. The models accurately predicted plasma concentrations of drugs in independent test sets, performing equally to or better than traditional allometric scaling methods. However, PBPK models have advantages over scaling in accounting for differences in drug chemistry, metabolism, and administration routes between species.

1. David E Leahy Molecular Informatics Group Newcastle University

5. AUC Predictions by PBPK

6. Predicted Plasma Levels A, biperiden; B, acecainide; C, dofetilide; D, budesonide

7. PBPK versus Scaling Training Set Caldwell (n=176) I Ward (n=97) PBPK (n=69) Method I Method II Av. Fold E. 2.16 2.22 2.64 2.48 FE < 2 56% 52% 49% 46% FE < 3 77% 79% 70% 72% FE < 4 88% 86% 79% 75% Test Set Caldwell (n=18) Ward (n=18) PBPK (n=18) Method I Method II Av. Fold E. 2.61 2.53 3.44 2.80 FE < 2 50% 50% 28% 50% FE < 3 56% 56% 44% 61% FE < 4 72% 67% 61% 72% Median P/A 1.74 1.60 2.54 1.11

8. Scaling(Ward Scaling) PBPK Scaling (Caudwell II) Scaling (Cauldwell I)