Chemical-induced cholestasis (CIC) has become a concern in chemical safety risk assessment in pharmaceutical, food, cosmetic, and industrial manufacturing. Currently, known animal and in vitro liver models are unsuitable as high-throughput screening tools due to their high cost, time-consuming, or poor screening accuracy. Herein, a cohort of chemicals validated as cholestatic hepatotoxic in humans, rodents, and in vitro liver models was established for testing. The accuracy and reliability of the detection of CIC in zebrafish larvae were assessed by liver phenotype, bile flow inhibition rate, bile acid distribution, biochemical indices, and RT-qPCR. In addition, the nomogram prediction model was constructed using binomial logistic regression analysis. The model was constructed with three variables: aspartate aminotransferase (AST.FC) level, total bile acid (TBA.FC) level, and fold change in the number of bile acid nodes per unit of bile ducts in the zebrafish liver (NPL.FC), which showed high predictive power (areas under the ROC curve: 0.983). Furthermore, this study demonstrated that zebrafish larvae have some model specificity for CIC risk assessment of estrogen endocrine disruptors and that testing after 10 dpf provides more scientific results. Overall, combining zebrafish larval phenotyping and nomograms is an efficient and powerful tool for CIC risk monitoring of chemicals.
Read full abstract