Human in silico modeling guided by transcriptomics predicts drug-induced proarrhythmicity as a function of age

Unexpected cardiotoxicity is a major cause of drug failure during development and can also contribute to market withdrawal. Traditional preclinical models and clinical trials do not fully capture age-related cardiac functional decline, which is associated with increased arrhythmic risk. We aimed to develop human-based in silico electrophysiology models incorporating age-related remodeling to improve the assessment of age-dependent drug-induced cardiotoxicity during drug development.

Biological and chronological transcriptomic age-related human ventricular ionic remodeling was integrated into the O'Hara-Virag-Varro-Rudy model of ventricular electrophysiology at both cellular and tissue levels. Cardiac biological age was represented using two independent gene expression-based indices: SenesAge, based on one gene related to cell senescence, and AppAge, based on nearly 2,000 age-related genes. A panel of 12 reference drugs was used to evaluate age-related drug-induced cardiotoxicity.

In silico aging based on SenesAge and AppAge reproduced age-related electrophysiological phenotypes at the cellular, tissue, and ECG levels more closely than chronological age-based remodeling. Biological age-related remodeling also increased the susceptibility to arrhythmogenic events following drug exposure by more than fivefold with respect to the adult population across the majority of tested conditions, although the effects varied according to the drug and its concentration.

This study provides a novel perspective for incorporating biological age-related remodeling into an established human-based in silico framework for preclinical cardiotoxicity evaluation at the population level. Our findings support biological age-based modeling as a complement to chronological age-based approaches for improving age-specific drug-induced cardiac risk assessment.