スケーラブルな肝臓チッププラットフォームによる予測的安全性評価

Non-animal methods, including advanced in-vitro approaches such as liver-on-a-chip models, hold promise for hepatotoxicity testing but remain limited in pharmaceutical and regulatory adoption due to high costs, poor scalability, low reproducibility, and insufficient validation. To address these challenges, we introduce a drug-induced liver injury (DILI) model based on the commercially available and highly adaptable OC-Plex microfluidic platform. Primary human hepatocytes (PHHs) maintained long-term viability and metabolic competence, confirmed by albumin and urea production as well as cytochrome P450 gene expression. The system reliably detected acetaminophen (APAP) - induced hepatotoxicity across six mechanistic readouts-albumin, viability, cytokeratin-18 (CK18), urea, CYP3A4 activity, and mitochondrial function. As an initial proof-of-concept, we tested 17 compounds, including both drugs and cosmetic ingredients. Among 13 compounds with known toxicity liabilities in humans, our model shows a high predictive performance (85.7 % sensitivity, 100 % specificity, and 92.3 % accuracy). Together, these findings demonstrate the feasibility of a cost-effective, scalable, and human-relevant liver-on-a-chip system for predictive hepatotoxicity testing. With future development and validation, this model holds great potential to replace animal testing in chemical safety assessment.