Integrating in vitro testing and physiologically-based pharmacokinetic (PBPK) modelling for chemical liver toxicity assessment—A case study of troglitazone

In vitro to in vivo extrapolation (IVIVE) for next-generation risk assessment (NGRA) of chemicals requires computational modeling and faces unique challenges. Using mitochondria-related toxicity data of troglitazone (TGZ), a prototype drug known for liver toxicity, from HepaRG, HepG2, HC-04, and primary human hepatocytes, we explored inherent uncertainties in IVIVE, including cell models, cellular response endpoints, and dose metrics. A human population physiologically-based pharmacokinetic (PBPK) model for TGZ was developed to predict in vivo doses from in vitro point-of-departure (POD) concentrations. Compared to the 200–800 mg/d dose range of TGZ where liver injury was observed clinically, the predicted POD doses for the mean and top one percentile of the PBPK population were 28–372 and 15–178 mg/d respectively based on C_max dosimetry, and 185–2552 and 83–1010 mg/d respectively based on AUC. In conclusion, although with many uncertainties, integrating in vitro assays and PBPK modeling is promising in informing liver toxicity-inducing TGZ doses.