Drug-induced liver injury (DILI) is a prominent failure mode in drug development resulting in clinical trial failures and post-approval withdrawal. Improved in vitro models for DILI risk prediction that can model diverse genetics are needed to improve safety and reduce high attrition rates in drug development. In this study, we evaluated the utility of human liver organoids (HLOs) for high-throughput DILI risk prediction and in an organ-on-chip system. The recent clinical failure of inarigivir soproxil due to DILI underscores the need for improved models.

HLOs were adapted for high-throughput drug screening in dispersed-cell 384-well format and a collection of DILI-associated drugs were screened. HLOs were also adapted to a liver-chip system to investigate enhanced in vivo-like function. Both platforms were benchmarked for their ability to predict DILI using combined biochemical assays, microscopy-based morphological profiling, and transcriptomics.

Dispersed HLOs retained DILI predictive capacity of intact HLOs and are amenable to high-throughput screening allowing for measurable IC50 values for cytotoxicity. Distinct morphological differences were observed in cells treated with drugs exerting differing mechanisms of action. HLOs on chips were shown to increase albumin production, CYP450 expression and also release ALT/AST when treated with known DILI drugs. Importantly, HLO liver chips were able to predict hepatotoxicity of tenofovir-inarigivir and showed steatosis and mitochondrial perturbation via phenotypic and transcriptomic analysis.

The high throughput and liver-on-chip system exhibit enhanced in vivo-like function and demonstrate the utility of the platforms in early and late-stage drug development. Tenofovir-inarigivr associated hepatotoxicity was observed and highly correlates with the clinical manifestation of DILI.