Systems Pharmacology of Arrhythmia: Using Mathematical Models to Identify Targets and Improve Toxicity Prediction

Drug-induced proarrhythmia is a persistent problem. Antiarrhythmic drugs as well as noncardiovascular drugs have the ability to provoke deadly arrhythmias. The research presented in this dissertation aims to address two crucial aspects of this problem. In the first study, we address reverse rate dependence, a problematic property of many antiarrhythmic drugs that results in both reduced arrhythmia suppression at fast heart rates and increased arrhythmia risk at slow heart rates. The opposite property, forward rate dependence, would theoretically overcome these parallel problems, yet forward rate dependent antiarrhythmics remain elusive. To determine whether drugs with desirable rate-dependent properties could be rationally designed, we performed comprehensive and systematic analyses of cardiomyocyte models. The analyses identified targets with forward rate dependent properties, and further simulations uncovered the mechanisms underlying these behaviors. In the second study, we worked to improve the prediction of the drug-induced Torsades de Pointes (TdP) arrhythmia, which remains a ubiquitous issue in safety pharmacology. Current assays to evaluate TdP risk are limited by poor specificity, and a mechanistic understanding of these assay limitations is incomplete. To address this, we used dynamical simulations of drug effects on cardiomyocytes combined with statistical analysis and machine-learning to design an improved algorithm to discriminate between arrhythmogenic and safe drugs. This classification algorithm identifies clinically torsadogenic drugs with superior accuracy and is suitable for use in early drug safety testing. Furthermore, our simulations indicate that cardiac ion channels not typically assessed in the drug development process may significantly affect TdP risk, and suggest a critical role for intracellular calcium in the generation of TdP. Overall, the two studies presented in this dissertation succeed through the use unique computational strategies that combine physiological, dynamical models with comprehensive statistical and machine-learning analyses. This approach enables us to systematically assess the proarrhythmic properties of real and hypothetical drugs, and to derive general principles. Through this work, I aim to improve the understanding of drug-induced arrhythmia and to advance the development of safer and more effective therapeutics.