Development of a Korean population model for physiologically based pharmacokinetic modeling

Abstract

Physiologically-based pharmacokinetic (PBPK) modeling and simulation within SimCYP® Simulator (Certara USA, Inc., Princeton, USA) is available with several virtual population models by ethnicity such as Caucasian, Japanese, and Chinese. These built-in ethnic population models with the population-specific physiologic parameters are useful to evaluate and understand the ethnic impact on the sensitivity of pharmacokinetics. However, a Korean population model is not available yet in SimCYP® simulator for PBPK approach. We aimed 1) to develop a Korean virtual population model which can be incorporated into the SimCYP® Simulator, and 2) to validate the appropriateness of the Korean virtual population model by comparing the simulated pharmacokinetic profiles and those reported previously for several model substrates of major drug metabolizing enzymes and transporters. Because most of the demographic and physiological parameters were the same (2689 out of 2824, 95.2 %) for all ethnic groups, they were mostly employed in the Korean population model. In contrast, ethnic sensitive parameters such as frequency of genetic variants and hepatic parameters including Cytochrome P450 (CYPs), UDP-glucuronosyltransferase (UGT), organic anion-transporting polypeptide (OATP), liver density and volume were modified based on literature, or otherwise some were employed without changes from other ethnic populations. Using the built-in three ethnic population models in the SimCYP® Simulator and the newly-developed Korean population model, pharmacokinetic profiles of the following example substrates were predicted: midazolam (CYP3A4/5), warfarin (CYP2C9), omeprazole (CYP2C19), metoprolol (CYP2D6), lorazepam (UGT), and rosuvastatin (OATP1B1/Breast cancer resistance protein (BCRP)). The predicted pharmacokinetic profiles varied widely by ethnic group with omeprazole having the largest difference. All the simulated values of geometric mean clearances of the six example drugs in developed Korean population model were within 0.5-1.5 fold of observed values (90.9 % of the predictions) except that of rosuvastatin (1.63 fold over-prediction). In conclusion, a Korean population model for PBPK modeling in the SimCYP® Simulator was adequately developed. This study indicated that an adequate Korean model would help understand ethnic sensitivity of a drug, and thus support the evidence for different dosing regimens and designs of bridging studies, and for gene-drug/drug-drug interaction possibilities in clinical drug development.