Presentation Information
[SS09-06]Prediction of antibody human pharmacokinetics using a physiologically based pharmacokinetic model coupled with multi-omics data
*Yasunori Komori1, Taichi Akahoshi1, Yoshie Takashima1, Koki Kibe1, Sotaro Naoi1, Kohji Nagano1, Kimio Terao1, Tatsuhiko Tachibana1 (1. Chugai Pharmaceutical Co., Ltd. (Japan))
Keywords:
Antibody drug,physiologically based pharmacokinetic model,Omics measurement,Human prediction
In the development of antibody drugs, accurately predicting their plasma concentration and extrapolating the results to humans are of great importance. For prediction, two types of antibody elimination are generally considered: one is dose-proportional, non-specific elimination and the other is saturable, target protein-mediated elimination. The latter is driven by complex formation between antibody with target protein, followed by degradation of the complex. For both types of elimination, empirical modeling methods that neglect physiological details have been generally used. However, mechanism-based modeling has recently gained momentum due its ability to account for underlying physiological processes.
In this context, researchers have developed physiologically based pharmacokinetics (PBPK) models. These models describe antibody disposition into major organs and their interstitial spaces using physiologically grounded assumptions. They are then calibrated to recapitulate animal/human pharmacokinetic data and can successfully explain dose-proportional nonspecific elimination. In contrast, nonlinear antibody elimination by target proteins has not been well documented due to the lack of target protein information.
In this session, we introduce a PBPK model that incorporates target protein data. As a data source, we utilized protein expression in humans and animals and protein half-life data in human cell lines obtained via proteomic analysis. By incorporating complex formation and subsequent degradation, the model calculates the antibody disposition in each organ, and the plasma concentration of antibody drugs is predicted as a sum of these processes. We evaluated the prediction accuracy by comparing the model’s output with available human pharmacokinetic data for several antibody drugs. For some antibodies, the human plasma concentration was successfully predicted within an acceptable margin of prediction error. After improving the model prediction by scaling model parameters to in vivo PK data using cynomolgus monkey PK datasets, we were able to estimate the concentration trends for the majority of the tested antibodies. The results, discussed in detail at this symposium as well as in our poster (Akahoshi et al.), suggest that such a PBPK modeling approach could be used to facilitate the model-based drug discovery and development of antibody drugs.
In this context, researchers have developed physiologically based pharmacokinetics (PBPK) models. These models describe antibody disposition into major organs and their interstitial spaces using physiologically grounded assumptions. They are then calibrated to recapitulate animal/human pharmacokinetic data and can successfully explain dose-proportional nonspecific elimination. In contrast, nonlinear antibody elimination by target proteins has not been well documented due to the lack of target protein information.
In this session, we introduce a PBPK model that incorporates target protein data. As a data source, we utilized protein expression in humans and animals and protein half-life data in human cell lines obtained via proteomic analysis. By incorporating complex formation and subsequent degradation, the model calculates the antibody disposition in each organ, and the plasma concentration of antibody drugs is predicted as a sum of these processes. We evaluated the prediction accuracy by comparing the model’s output with available human pharmacokinetic data for several antibody drugs. For some antibodies, the human plasma concentration was successfully predicted within an acceptable margin of prediction error. After improving the model prediction by scaling model parameters to in vivo PK data using cynomolgus monkey PK datasets, we were able to estimate the concentration trends for the majority of the tested antibodies. The results, discussed in detail at this symposium as well as in our poster (Akahoshi et al.), suggest that such a PBPK modeling approach could be used to facilitate the model-based drug discovery and development of antibody drugs.