Integrated in vitro/in silico test battery for the prediction of chemical-induced cholestatic liver toxicity.

Cholestatic liver toxicity induced by chemicals is a multisectoral concern and remains a challenge to predict, partly due to a lack of mechanistic understanding of chemical-induced cholestasis. In the context of modern chemical safety, there has been a demand to step away from animal testing and promote alternative mechanistically-anchored approaches. Based on a previously introduced adverse outcome pathway network and using 7 (non-)cholestatic reference test chemicals, the present study built the initial iteration of a test battery integrating in vitro and in silico approaches. The former investigated the molecular initiating events and key events of transporter changes, bile canalicular dynamics, endoplasmic reticulum stress, mitochondrial impairment and oxidative stress in HepaRG liver cells. The latter consisted of quantitative-structure-activity relationships, alert-based structural activity relationships and molecular docking. Results of the test battery were integrated and quantified using a weight of evidence approach, which allowed to rank the chemicals based on their cholestatic concern. The chemicals perfluorooctanesulfonic acid, cyclosporine A and 1-naphtyl isothiocyanate received the highest scores, while the two non-cholestatic chemicals were correctly identified. As a result, a high specificity of the test battery was observed, in contrast to traditional animal-based approaches. Using multiple endpoints and a weight of evidence approach provides the proof-of-concept of an animal-free mechanism-based test battery for the prediction of the cholestatic potential of chemicals with promising performance. The high specificity supports its potential application in early drug development, reducing the risk of excluding viable drug candidates.