Validation of gene expression profiles from cholestatic hepatotoxicants in vitro against human in vivo cholestasis.

Drug-induced liver injury remains the most common cause of acute liver failure and a frequently indicated reason for withdrawal of drugs. For the purpose of evaluating the relevance of liver cell models for assessing hepatotoxic risks in intact humans, we here aimed to benchmark 'omics-derived mechanistic data from three in vitro models for parenchymal liver function, intended for the investigation of drug-induced cholestasis, against 'omics data from cholestatic patients. Transcriptomic changes in HepG2 cells, primary mouse hepatocytes and primary human hepatocytes exposed to known cholestatic compounds were analyzed using microarrays. Some of the differentially expressed genes in HepG2 cells were also differentially expressed into the same direction in human cholestasis. The overlap between drug-induced transcriptomic responses in primary mouse hepatocytes and primary human hepatocytes appeared limited and no genes overlapping with in vivo cholestasis were found. Thereupon, a pathway for drug-induced cholestasis was used to map the drug-induced transcriptomic modifications involved in bile salt homeostasis. Indications of an adaptive response to prevent and reduce intracellular bile salt accumulation were observed in vivo as well as in the in vitro liver models. Furthermore, drug-specific changes were found, which may be indicative for their cholestatic properties. Furthermore, connectivity mapping was applied in order to investigate the predictive value of the in vitro models for in vivo cholestasis. This analysis resulted in a positive connection score for most compounds, which may indicate that for identifying cholestatic compounds the focus should be on gene expression signatures rather than on differentially expressed genes.