Table 1.

Models for assessing DILI.

Model	Benefits	Limitations	Example references
Conventional cultures/cocultures	(i) High-throughput (ii) Sandwich cultures can maintain cell polarity	(i) Usually lack liver stromal cells (ii) Loss of drug metabolizing capacity occurs within hours	[6, 10–13]
Micropatterned cocultures	(i) Controlled architecture allows for higher functions for 1-2 months (ii) Coculture allows for the study of drug toxicity in a diseased background (iii) Compatible with high content imaging readouts	(i) Usually lack all liver stromal cells (ii) Use nonhuman supporting cells	[14, 17, 18, 20, 27]
Spheroidal cultures/cocultures	(i) Multicellular interactions (ii) Maintenance of major liver functions for 1–3 months	(i) Difficult to control disorganized cell type interactions over time (ii) Necrosis in center of larger spheroids (iii) Size variability (iv) Incompatible with standard high content imaging equipment	[30–38]
Bioprinted cultures/cocultures	(i) Precise control of cell placement (ii) Multicellular interactions	(i) Printing resolution does not allow placement of individual cells (ii) Low-throughput (iii) Heterogeneous distribution of drugs across the bioprinted tissues	[39, 53, 54]
Perfused biochips	(i) Dynamic fluid flow for nutrient and waste exchange (ii) Sustained functionality for at least 1 month (iii) Can be combined with module for real-time toxicity endpoint readouts	(i) Binding of drugs to tubing (ii) Large dead volume requiring higher quantities of novel compounds (iii) Low-throughput (iv) Shear stress may cause lower hepatic functions (v) Can wash away built-up beneficial molecules	[44–52, 55–66]
Precision cut liver slices	(i) Retains native liver architecture and all liver cell types (ii) Can be combined with flow to improve functional lifetime	(i) Low-throughput (ii) Rapid decline in liver functions (iii) Heterogeneous distribution of drug within the slice	[67–72]
Humanized rodents	(i) Human-relevant toxicity profiles in vivo  (ii) Can look at organ-organ interactions with a humanized liver background	(i) Variability in human hepatocyte engraftment efficiency (ii) Low-throughput (iii) Residual murine liver cells can cause confounding results (iv) Interactions with murine organs	[73–85]