Background and previous systems
Cholestatic and mixed hepatocellular/cholestatic injuries account for up to 50% of all cases of drug-induced liver injury and are thus highly relevant for drug development . Drug-induced cholestasis (DIC) manifests due to impaired bile acid (BA) homeostasis, which results in the intrahepatic accumulation of BAs, causing induction of apoptosis . DIC can result from interference of drugs or their metabolites with the function of the bile salt export pump (BSEP), which constitutes the predominant BA export transporter .
Previous cholestatic models
Preclinical prediction of DIC previously relied exclusively on assessing BSEP activity using membrane vesicles  or hepatocytes in sandwich culture . Yet, this confined perspective neglects other mediators of BA homeostasis that play a role in cholestatic liver injury, including enzymes involved in BA conjugation and sulfation , nuclear receptors  and a variety of BA transporters . Furthermore, symptoms of DIC often only manifest weeks after the patients commenced treatment.
A major limitation of the currently used DIC in vitro models is the inability to maintain hepatic cells in a differentiated state, resulting in loss of expression of BA transporters and conjugating enzymes. In simple 2D monolayer cultures, primary human hepatocytes (PHH) rapidly dedifferentiate within hours, which renders them incompatible with long-term toxicity studies. In sandwich culture, PHH retain functional bile canalicular networks for several days, which is of great value for studies of hepatobiliary transport and DIC. Yet, dedifferentiation in sandwich-cultured PHH is only delayed but not stalled and thus, culturing PHH for the timeframes necessary for DIC to manifest in vivo is not possible .
In the 3D spheroid system the relevant major bile acid transporters are expressed for multiple weeks in culture (Fig. 1a; Hendriks et al., 2016). This feature allows to faithfully identify compounds with cholestatic liability by co-exposing spheroid cultures with BAs and candidate drugs of interest (Fig. 1b).
The cholestatic model could separate cholestatic agents from compounds that have not been implicated in hepatic cholestasis with 86% sensitivity and 100% specificity (Fig. 1c).
- Björnsson et al., Hepatology, 2005
- Yang et al., Journal Pharm Sci, 2013
- Morgan et al., Toxicol Sci, 2010
- Ansede et al., Drug Metab Dispos, 2010
- Alnouti et al., Toxicol Sci, 2009
- Zollner et al., Pharmacol Therapeut, 2010
- Pauli-Magnus et al., Hepatology, 2006
- Rowe et al., Hepatology, 2013