A new in vitro rat liver platform capturing biological diversity for drug-induced liver injury assessment and reduction of animal use in drug discovery and development

Drug-induced liver injury (DILI) is a major cause of drug development failure and market withdrawal. Despite the use of animal and human-derived preclinical models, none reliably predict human hepatotoxicity. Conventional animal studies, based on group-level averages, overlook inter-individual variability critical to idiosyncratic DILI. Although several alternatives to animal testing are available, implementation of alternative methodologies into safety evaluations is very slow and to date, no standalone validated alternative models to assess systemic toxicity exist. To address this, we developed a novel serum-educated rat liver spheroid model that captures individual metabolic diversity, allowing DILI to be assessed both in vitro and in vivo within the same animal. Building on this approach, the present study aimed to validate the model’s ability to detect DILI caused by well-characterized hepatotoxic drugs, including diclofenac and bosentan, in rat models. Rats were treated orally with diclofenac or bosentan for 28 days. Blood was collected pre-dosing to generate individualized spheroids containing rat hepatocytes, stellate cells, and macrophages. These spheroids were exposed to each drug across a ten-point concentration range for 3 days and cell viability was quantified using the CellTiter-Glo ATP assay. Clinical chemistry analyses of ALT, AST, ALP, albumin, and bilirubin were performed at in vivo study termination on day 29. In vitro data were analyzed using PredictCan-MIND to derive DILI severity scores and correlate in vitro cytotoxicity with in vivo biomarkers. In vivo , diclofenac caused no significant liver enzyme elevations, although some rats, exclusively females, displayed subtle toxicity revealing sex-dependent susceptibility. Corresponding in vitro spheroids confirmed hepatocellular injury in these individuals. Bosentan produced a mild cholestatic response in vivo without consistent enzyme elevation, while in vitro analysis showed clear hepatotoxicity in 8 of 10 rats. Notably, in vitro DILI severity correlated strongly with in vivo ALP levels, consistent with bosentan’s known cholestatic mechanism. The serum-educated rat liver spheroid model captures inter-individual and sex-related differences in hepatotoxicity and demonstrates translational concordance with in vivo cholestatic markers. This approach improves DILI prediction, aligns with the 3Rs principle, and supports a potential 50%–70% reduction in animal use for preclinical liver toxicity testing.