Human-based electromechanical modelling and simulation achieved 86% qualitative agreement with ex vivo rabbit experiments using four-channel data, improving to 95% with additional current targets.
Does human-based ventricular electromechanical modelling and simulation accurately predict drug-induced cardiotoxicity compared to ex vivo rabbit Langendorff experiments?
Human-based electromechanical in silico modelling accurately predicts drug-induced changes in contractility and repolarization, supporting its use as a replacement for ex vivo rabbit experiments in preclinical cardiotoxicity screening.
Introduction Effective proarrhythmic and inotropic risk assessment is essential for pharmaceutical development, but current preclinical methods for assessment of cardiac inotropy are flawed and costly, particularly when combined with QTc prolongation studies. Ex vivo rabbit Langendorff isolated heart experiments provide valuable insights into cardiovascular effects and safety, but their high cost, experimental difficulty, and limited applicability to human physiology pose challenges. Human mechanistic in silico modelling and simulation has proven successful in risk assessments of both electrophysiological and cardiac inotropy assessment. Methods This study evaluates the feasibility of replacing ex vivo Langendorff experiments for contractility with human-based ventricular electromechanical modelling and simulations, based on 37 compounds. Results Results show 1) 86% of compounds show qualitative agreement using four channel data (I Kr , I CaL , I Na , I to ), with 73% showing quantitative agreement correlating with higher quality data, 2) sensitivity analysis identified hNCX1 and late hNaV1.5 currents as additional targets, which, when considered alongside the four channel data as input, improved agreement from 86% to 95% (at least qualitatively), 3) incomplete dose-response input data was the key reason for discrepancies between experiment and simulation, while noting only two compounds showed a complete disagreement. Incorporating patient variability through a population of N = 166 human ventricular cell models add further confidence, and highlights increasing inter-subject diversity with increasing concentrations. Conclusion This study supports the adoption of in silico new approach methodologies for accurate prediction of drug cardiotoxicity, and to refine, reduce and replace the use of ex vivo rabbit experiments.
Holmes et al. (Fri,) conducted a other in Drug-induced cardiotoxicity (n=166). Human-based electromechanical modelling and simulation vs. Ex vivo rabbit Langendorff isolated heart experiments was evaluated on Qualitative agreement between in silico predictions and ex vivo experiments. Human-based electromechanical modelling and simulation achieved 86% qualitative agreement with ex vivo rabbit experiments using four-channel data, improving to 95% with additional current targets.