Drug Proarrhythmic Evaluation in a High Throughput Cardiac New Approach Methodology

Cardiotoxicity is a major cause for drug failure throughout the drug development process, with particular concern for action potential prolongation and arrhythmia. Hence, such liabilities are heavily considered during the early phases of drug design to prevent dangerous compounds from progressing. New approach methodologies (NAMs) that efficiently examine this risk early in the discovery pipeline should help streamline drug development programs. We developed a cardiac NAM, a 384-well open bath platform consisting of cardiac tissue derived from human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes, enabling high-throughput drug screening while maintaining the structural and functional complexity of 3D cardiac micromuscles.

We dramatically increased throughput without compromising physiological relevance provided by the 3D micromuscle structure. Our 384-well open bath high-throughput platform allowed evaluation of multiple compounds at a time, enabling us to study the CiPA (comprehensive in vitro proarrhythmia assay) drug panel for proarrhythmia screening. We obtained phenotypic fingerprints of all 28 compounds (9 low, 11 intermediate, and 8 high arrhythmia risk; https://cipaproject.org) in dose-escalation studies around their respective clinical concentrations. The analysis was augmented with an in silico pipeline that used phenotypic biomarkers to invert data into a mathematical model of cellular currents to infer which ion channels were affected upon drug exposure, and then trained a ML model to predict channel block.

We found accurate detection of arrhythmic potential for most of the compounds, and the in silico model inversions were consistent with published values of compound channel block. All the high risk compounds showed action potential duration (APD) prolongation coupled with either action potential abnormalities, early afterdepolarizations (EADs), or beat cessation. For the intermediate risk group, 9 out of 11 compounds caused APD prolongation alone or in combination with EADs while 2 others showed either beat cessation or beat rate change. Augmentation of APD analysis with detailed biophysical modeling and ML tools provided meaningful insight into the mechanisms involved in APD changes. Overall, our cardiac NAM allowed for fast and relevant screening for mechanistic understanding of APD prolongation and proarrhythmic activity, at massively increased throughput compared to other 3D micromuscle models.