3D human liver spheroids accurately predict drug hepatotoxicity, distinguishing GLP-1R agonist risks.

Drug-induced liver injury (DILI) remains a critical challenge in drug development, frequently causing clinical trial attrition and postmarketing withdrawals. Current preclinical models often exhibit significant translational gaps, failing to accurately predict human hepatotoxicity. This necessitates more physiologically relevant in-vitro systems to better identify potential liver toxicity early. Three-dimensional human liver spheroids offer a promising platform, mimicking in-vivo hepatic architecture and function more closely than traditional 2D cell cultures, thereby improving predictive accuracy for complex mechanisms like mitochondrial dysfunction and cholestasis.

Researchers applied a well-characterized 3D primary human liver spheroid platform, undergoing multicenter technology transfer, to interrogate diverse mechanisms of hepatotoxicity. The spheroids were subjected to repeated-dose exposure to established hepatotoxins, including chlorpromazine and bosentan, to assess concentration-dependent toxicity. Mechanistic resolution involved Seahorse extracellular flux analysis for early mitochondrial dysfunction detection and bile acid coexposure to identify cholestatic liability. The model's performance was benchmarked against recent clinical cases, including Bruton's tyrosine kinase inhibitors (tolebrutinib, evobrutinib), oral GLP-1R agonists (danuglipron, orforglipron), and coexposure of azelaprag-tirzepatide, as well as the IL-17A inhibitor LY3509754 with nonparenchymal liver cells.

The 3D human liver spheroid platform demonstrated stable hepatic functionality, including expression of metabolic enzymes and transporters, albumin secretion, and long-term CYP activity. Repeated-dose exposure to hepatotoxins showed robust, concentration-dependent toxicity with excellent interexperimental (R2 = 0.98) and interdonor (R2 = 0.89) reproducibility. Seahorse extracellular flux analysis enabled sensitive detection of early mitochondrial dysfunction, allowing mechanistic discrimination between mitochondrial and nonmitochondrial toxins. Cholestatic liability was accurately identified using bile acid coexposure, correctly classifying chlorpromazine and bosentan and confirming bile salt export pump downregulation. Importantly, the model recapitulated hepatotoxicity signals observed in recent clinical development, including for Bruton's tyrosine kinase inhibitors (tolebrutinib and evobrutinib), and differentiated oral GLP-1R agonists (danuglipron vs orforglipron).