The genetics of pro-arrhythmic adverse drug reactions

A Real-World Data Driven Pharmacovigilance Investigation on Drug-Induced Arrhythmia Using KAERS DB, a Korean Nationwide Adverse Drug Reporting System

This study aims to investigate the prevalence and seriousness of drug-induced arrhythmia and to identify predictors associated with the seriousness of arrhythmia. Drug-induced arrhythmia cases reported to the Korean Adverse Event Reporting System Database (KAERS DB) from January 2012 to December 2021 were investigated. A disproportionality test was performed to detect the association of the etiologic medication classes and types, along with patient demographic information, with the seriousness of drug-induced arrhythmia. Logistic regression was performed to investigate the predictors that increase the risk of serious arrhythmia. The most common etiologic agent for drug-induced arrhythmia was sevoflurane, whereas serious arrhythmia was most prevalent with narcotics. Antibiotics (reporting odds ratio (ROR) 4.125; 95% CI 1.438-11.835), chemotherapy (ROR 6.994; 95% CI 2.239-21.542), and iodinated contrast media (ROR 8.273; 95% CI 3.062-22.352) had a strong association with the seriousness of drug-induced arrhythmia. Among numerous etiologic agents, ioversol (ROR 16.490; 95% CI 3.589-75.772) and lidocaine (ROR 12.347; 95% CI 2.996-50.884) were more likely to be reported with serious arrhythmia. Aging and comorbidity, primarily cancer, are the most contributing predictors associated with serious arrhythmia. Further studies on the clinical significance of patient-specific predictors for the increased risk of serious drug-induced arrhythmia are warranted to promote drug safety.

A deep learning platform to assess drug proarrhythmia risk

Drug safety initiatives have endorsed human iPSC-derived cardiomyocytes (hiPSC-CMs) as an in vitro model for predicting drug-induced cardiac arrhythmia. However, the extent to which human-defined features of in vitro arrhythmia predict actual clinical risk has been much debated. Here, we trained a convolutional neural network classifier (CNN) to learn features of in vitro action potential recordings of hiPSC-CMs that are associated with lethal Torsade de Pointes arrhythmia. The CNN classifier accurately predicted the risk of drug-induced arrhythmia in people. The risk profile of the test drugs was similar across hiPSC-CMs derived from different healthy donors. In contrast, pathogenic mutations that cause arrhythmogenic cardiomyopathies in patients significantly increased the proarrhythmic propensity to certain intermediate and high-risk drugs in the hiPSC-CMs. Thus, deep learning can identify in vitro arrhythmic features that correlate with clinical arrhythmia and discern the influence of patient genetics on the risk of drug-induced arrhythmia.

Modeling susceptibility to drug-induced long QT with a panel of subject-specific induced pluripotent stem cells

A large number of drugs can induce prolongation of cardiac repolarization and life-threatening cardiac arrhythmias. The prediction of this side effect is however challenging as it usually develops in some genetically predisposed individuals with normal cardiac repolarization at baseline. Here, we describe a platform based on a genetically diverse panel of induced pluripotent stem cells (iPSCs) that reproduces susceptibility to develop a cardiotoxic drug response. We generated iPSC-derived cardiomyocytes from patients presenting in vivo with extremely low or high changes in cardiac repolarization in response to a pharmacological challenge with sotalol. In vitro, the responses to sotalol were highly variable but strongly correlated to the inter-individual differences observed in vivo. Transcriptomic profiling identified dysregulation of genes (DLG2, KCNE4, PTRF, HTR2C, CAMKV) involved in downstream regulation of cardiac repolarization machinery as underlying high sensitivity to sotalol. Our findings offer novel insights for the development of iPSC-based screening assays for testing individual drug reactions.