In Silico Modeling of Changes to Ventricular Myocyte Action Potential Morphology Caused by Pharmacological Agents

Abstract

Changes in electrocardiogram morphology associated with ventricular arrhythmias are one of the largest causes of the removal or restriction of marketed pharmacological agents and can be difficult to predict during drug development. Since changes to ECG morphology are the consequence of cellular-level alterations in the action potential, fresh insights can be gained by using computational models of cardiac action potentials and simulating drug-induced side effects. Although most pro-arrhythmic drugs block the rapid delayed rectifier current encoded by the gene commonly referred to as HERG, many drugs are non selective. With this in mind we analyzed mathematical models of ventricular action potentials to develop methods to predict non-specific, potentially pro-arrhythmic effects of new pharmacological agents. Through simulations, we generated predictions of changes in action potential duration (APD) caused by different concentrations of hypothetical drugs that specifically blocked with a defined affinity individual ion transport pathways (channels, pumps, or transporters). These relationships were empirically fit to the Hill equation. We then generated predictions of changes in APD caused by hypothetical drugs that blocked multiple pathways with different affinities. Based on linear transformations of the Hill equation fits and minimum least squares, our “reverse engineering” algorithm predicted the pathways that were most likely to be affected by a particular drug based on the APD versus [drug] relationship. Simulations were performed with the ten Tusscher model of the human ventricular myocyte in which hypothetical drugs blocked the rapid delayed rectifier current along with one of six other currents. When the results were fed to the algorithm in a blinded fashion, the program successfully identified the drug targets. The results show promise for the development of methods to determine potential channel candidates when the cause of APD disturbance is unknown.