Extending the role of hiPSC-derived cell models in toxicity risk prediction
Authors: Eunho Kim, Ami Kim, Jieun An, Bomi Kim, Subin Kim, Eueun Kim, Jong-Hwan Lee
Originally Published in: Journal of Pharmacological and Toxicological Methods (August 2026) (Download)
Abstract
Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) offer powerful platforms for applications such as drug screening, disease modeling, and cell therapy development through its ability to faithfully replicate native behavior of human primary cardiomyocyte. In particular, they support cardiotoxicity risk assessment in alignment with regulatory frameworks such as ICH E14/S7B and HESI/CiPA, addressing key safety endpoints in drug discovery. In this study, 2D and 3D models were used to evaluate drug-induced cardiac responses in a wide variety of settings and conditions.
2D hiPSC-CMs were differentiated using proprietary protocols. 3D Engineered Heart Tissue (EHT) was generated through protocol provided by Curi Bio. Functional testing was performed using various platforms including Axion Maestro Pro, Nanion FLEXcyte96, and Curi Bio's Mantarray system. Compounds were prepared in recommended solutions, and corresponding negative controls were prepared accordingly.
Predictive modeling using hiPSC-CMs demonstrated robust stratification of 28 CiPA compounds into high/intermediate and low-risk categories based on FPDcF changes. Application of the FDA's predictive algorithm confirmed statistically significant discrimination between risk groups.
Cardiotoxicity assessment revealed that most induced ≥70% reduction in cell signal after chronic treatment of 12 diverse anticancer compounds. However, compounds such as Idarubicin (anthracyclines), Erlotinib (TKIs), and Cyclophosphamide (DNA crosslinkers) demonstrated minimal signal attenuation, suggesting class-specific and mechanistic variability in toxicity profiles.
In both 2D and 3D EHT models, doxorubicin elicited, concentration- and time-dependent reductions in key contractile parameters, including twitch force/amplitude, contraction velocity, and relaxation velocity. While the kinetics and magnitude of responses differed between models, both reliably captured drug-induced dysfunction.
Overall, our findings highlight the versatility and predictive power of hiPSC-derived cardiomyocyte platforms. Their scalability and physiological relevance support their adoption as key tools for preclinical cardiotoxicity assessment, offering refined alternatives to animal models in the early stages of drug development.