A deep learning platform to assess drug proarrhythmia risk

• Published in: Cell stem cell Vol. 30; no. 1; pp. 86 - 95.e4
• Main Authors: Serrano, Ricardo, Feyen, Dries A.M., Bruyneel, Arne A.N., Hnatiuk, Anna P., Vu, Michelle M., Amatya, Prashila L., Perea-Gil, Isaac, Prado, Maricela, Seeger, Timon, Wu, Joseph C., Karakikes, Ioannis, Mercola, Mark
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 05.01.2023
• Subjects: Action Potentials; Arrhythmias, Cardiac - chemically induced; artificial; cardiomyocytes; CiPA; Deep Learning; drug screening; drug-induced arrhythmia; Humans; induced pluripotent stem cells; Induced Pluripotent Stem Cells - physiology; intelligence; iPSC; Myocytes, Cardiac - physiology; safety pharmacology; Torsades de Pointes - chemically induced; cardiomyocytes; deep learning; artificial; drug screening; drug-induced arrhythmia; iPSC; CiPA; AI; induced pluripotent stem cells; safety pharmacology; intelligence
• ISSN: 1934-5909; 1875-9777
• DOI: 10.1016/j.stem.2022.12.002

Drug safety initiatives have endorsed human iPSC-derived cardiomyocytes (hiPSC-CMs) as an in vitro model for predicting drug-induced cardiac arrhythmia. However, the extent to which human-defined features of in vitro arrhythmia predict actual clinical risk has been much debated. Here, we trained a convolutional neural network classifier (CNN) to learn features of in vitro action potential recordings of hiPSC-CMs that are associated with lethal Torsade de Pointes arrhythmia. The CNN classifier accurately predicted the risk of drug-induced arrhythmia in people. The risk profile of the test drugs was similar across hiPSC-CMs derived from different healthy donors. In contrast, pathogenic mutations that cause arrhythmogenic cardiomyopathies in patients significantly increased the proarrhythmic propensity to certain intermediate and high-risk drugs in the hiPSC-CMs. Thus, deep learning can identify in vitro arrhythmic features that correlate with clinical arrhythmia and discern the influence of patient genetics on the risk of drug-induced arrhythmia. [Display omitted] •Deep learning detects in vitro features that correlate with clinical arrhythmia•Drug safety margin from AI identifies drug risk with AUC = 0.95•Cardiomyopathic genotypes increase sensitivity to drug proarrhythmia risk Serrano et al. used human iPSC-derived cardiomyocytes and deep learning data analysis to establish an in vitro safety margin that predicts clinical proarrhythmic effects of drugs. Their platform shows high accuracy in identifying risky drugs as well as genotypes associated with increased risk of arrhythmia in people.