Data science identifies novel drug interactions that prolong the QT interval

Abstract

Drug-induced prolongation of the QT interval on the electrocardiogram (long QT syndrome, LQTS) can lead to a potentially fatal ventricular arrhythmia called torsades de pointes (TdP). 180 drugs with both cardiac and non-cardiac indications have been found to increase risk for TdP, but drug-drug interactions contributing to LQTS (QT-DDIs) remain poorly characterized. Traditional methods for mining observational healthcare data are poorly equipped to detect QT- DDI signals due to low reporting numbers and a lack of direct evidence for LQTS. In this study we present an integrative data science pipeline that addresses these limitations by identifying latent signals for QT-DDIs in the FDAs Adverse Event Reporting System and retrospectively validating these predictions using electrocardiogram data in electronic health records. We present 26 novel QT-DDIs flagged using this method that warrant further investigation.\n\nKey Points- Drug-induced long QT syndrome (LQTS) can lead to potentially fatal arrhythmias (torsades de pointes, TdP). Drug-drug interactions that prolong the QT interval (QT- DDIs) can be clinically significant but remain poorly characterized.\n- Observational health data (such as adverse event spontaneous reporting systems and electronic health records) offer an opportunity to mine for new QT-DDIs, but when used individually these datasets have a number of limitations that prevent identification of true signals.\n- We present an integrative data science approach that combines mining for latent QT- DDI signals in the FDA Adverse Event Reporting System and retrospective analysis of electrocardiogram lab results in electronic health records at Columbia University Medical Center to identify 26 novel QT-DDIs.