Toward Standardized Methodologies for Drug-Induced Proarrhythmia Classification: An In Silico Proof of Concept

Background: The Comprehensive in vitro Proarrhythmia Assay integrates experimental electrophysiology with in silico simulations to improve the prediction of drug-induced proarrhythmic risk and to address regulatory limitations associated with traditional QT interval-based safety assessment. In recent years, increasing attention has been directed toward the arrhythmogenic potential of drugs, leading to the development of multiple computational frameworks. However, substantial methodological variability still exists across studies, including differences in electrophysiological models, biomarker selection, and population sampling strategies. Using a model of electrotonically coupled ventricular cardiomyocytes, this study proposes a proof of concept for a standardized in silico framework aimed at predicting drug-induced arrhythmic risk. Methods: A virtual population of ventricular cellular models was generated and calibrated using patient-derived electrophysiological data. Ten drugs representing 3 levels of proarrhythmic risk were evaluated both in isolated-cell simulations and in electrotonically coupled cellular networks. Drug classification was performed using a novel arrhythmic risk score, which integrates 8 established electrophysiological biomarkers. The analysis also evaluated the effects of electrotonic coupling, biomarker reduction, and cellular subsampling on the stability of the risk estimation. Results: The electrotonically coupled cell model reproduced the classifications obtained with the well-established isolated-cell model, demonstrating comparable predictive performance. However, reducing the number of biomarkers from 8 to 2 led to a substantial increase in false-negative classifications. Similarly, excessive reduction of the cellular population produced a non-negligible increase in the variability of the estimated risk score. Conclusion: Electrotonically coupled cellular networks provide a physiologically consistent framework that preserves the risk classifications obtained in isolated-cell simulations. The results indicate that reliable computational assessment of drug-induced proarrhythmic risk requires careful consideration of 3 key methodological elements: the electrophysiological model, the size of the simulated cellular population, and the number of biomarkers used for risk estimation.