Presentation Information
[POS-01]Prediction of human PK of drug antibody in a human PBPK model using scaling factors obtained from cynomolgus PK
*TAICHI AKAHOSHI1, Yasunori Komori1, Yoshie Takashima2, Koki Kibe2, Sotaro Naoi3, Kohji Nagano2, Kimio Terao1, Tatsuhiko Tachibana1 (1. Pharmaceutical Science Department, Translational Research Division, Chugai Pharmaceutical Co., Ltd., Yokohama-shi, Japan. (Japan), 2. Research Division, Chugai Pharmaceutical Co. Ltd., 216 Totsuka-cho, Totsuka-ku, Yokohama, Kanagawa, 244-8602, Japan. (Japan), 3. Research Division, Chugai Pharmabody Research, Singapore, Singapore. (Singapore))
Keywords:
PBPK,modeling and simulation,human,antibody
The prediction of plasma and organ interstitial concentrations of drug antibodies is essential for understanding their efficacy and toxicity in humans. PBPK (physiologically-based pharmacokinetic) models incorporate physiological parameters that enable the prediction of organ interstitial concentration, while PBPK-TMDD (target mediated drug disposition) models further account for antigen-mediated drug elimination to predict non-linear PK. To calculate antigen-mediated drug elimination, PBPK-TMDD models utilize antigen expression levels and turnover rates; however, because these parameters are derived from in vitro studies, it is unclear whether they are sufficient for predicting nonlinear PK in humans. In this study, we obtained scaling factors for antigen expression and turnover rate by fitting the parameters to cynomolgus (cyno) PK data. Then, we examined whether these scaling factors were useful for predicting human PK in a human PBPK model.
First, we collected PK parameters for a drug antibody exhibiting non-linear PK. Then, we established a cyno PBPK model and estimated scaling factors for antigen expression and turnover rate by fitting them to cyno PK. The scaling factors were then applied to a human PBPK model to predict human PK (Method 1), and prediction accuracy was calculated based on the average absolute fold error. For comparison, we also predicted human PK using a model without these scaling factors ( Method 2). Overall, Method 1 achieved higher human PK prediction accuracy than Method 2, although both methods yielded reasonable predictions. These results indicate that scaling factors obtained from cyno PK data can improve the prediction of human PK for drug antibodies in a human PBPK model.
First, we collected PK parameters for a drug antibody exhibiting non-linear PK. Then, we established a cyno PBPK model and estimated scaling factors for antigen expression and turnover rate by fitting them to cyno PK. The scaling factors were then applied to a human PBPK model to predict human PK (Method 1), and prediction accuracy was calculated based on the average absolute fold error. For comparison, we also predicted human PK using a model without these scaling factors ( Method 2). Overall, Method 1 achieved higher human PK prediction accuracy than Method 2, although both methods yielded reasonable predictions. These results indicate that scaling factors obtained from cyno PK data can improve the prediction of human PK for drug antibodies in a human PBPK model.