Using Clinical Datasets to Optimize Models of Human Ventricular Electrophysiology: Implications for In Silico Drug Screening

Abstract

Recent advances in Computational Cardiology mean we can now examine the causes, mechanisms and impact of cardiac dysfunction in silico, particularly in regards to risk stratification and treatment of heart rhythm disturbances. As a result, computational cardiology stands at the threshold of clinical utility. For example, in-silico models of human cardiac electrophysiology are being considered by the FDA for prediction of proarrhythmic cardiotoxicity as a core component of the preclinical assessment phase of all new drugs. However, one issue that exists with current models is that they each respond differently to insults such as drug block of ion channels or mutation of cardiac ion channel genes. Clearly this poses a problem in relation to the utility of these models in making quantitative predictions that are physiologically or clinically meaningful. To examine this in detail we tested the ability of three models of the human ventricular action potential, the O’hara-Rudy the Grandi-Bers and the Ten Tusscher models, to reproduce the clinical phenotype of different subtypes of the long QT syndrome. All models, in their original form, produce markedly different and unrealistic predictions of QT prolongation. To address this, we used a global optimization approach to constrain existing in silico models to clinical datasets. After optimization, all models have similar current densities during the action potential, despite differences in kinetic properties of the channels in the different models, and closely reproduced the prolongation of repolarization seen in clinical data. We suggest that models optimized using this approach can be utilized with more confidence in clinical and preclinical applications such as prediction of proarrhythmic risk as part of in silico drug screening, examining pathogenic interactions of electrical dysfunction and structural alteration in the myocardium and assessing the impact of genetic variants in ion channel genes in contributing to heart rhythm disturbances.