Systems Pharmacology of Arrhythmias

Integration of drug and protein interaction data with genetic data enables prediction of adverse drug effects. Arrhythmic Neighborhood By integrating protein interaction data with data about the genetics underlying disease, Berger et al. identified a network associated with a specific disease, long QT syndrome (LQTS). This particular, potentially fatal, cardiac disorder can also be caused by various drugs, both those used to treat cardiovascular disease and those used to treat non–cardiac-related conditions. By combining this LQTS network with other genomic data sets, Berger et al. identified genetic variations likely to influence a person’s susceptibility to LQTS. By combining this LQTS network with data about adverse effects of drugs, they identified drugs that may induce LQTS. This provides an example of the effectiveness of systems pharmacology in linking drug targets and disease genes through protein interaction networks, which is a step toward personalized medicine and safer drug prescribing. Long-QT syndrome (LQTS) is a congenital or drug-induced change in electrical activity of the heart that can lead to fatal arrhythmias. Mutations in 12 genes encoding ion channels and associated proteins are linked with congenital LQTS. With a computational systems biology approach, we found that gene products involved in LQTS formed a distinct functional neighborhood within the human interactome. Other diseases form similarly selective neighborhoods, and comparison of the LQTS neighborhood with other disease-centered neighborhoods suggested a molecular basis for associations between seemingly unrelated diseases that have increased risk of cardiac complications. By combining the LQTS neighborhood with published genome-wide association study data, we identified previously unknown single-nucleotide polymorphisms likely to affect the QT interval. We found that targets of U.S. Food and Drug Administration (FDA)–approved drugs that cause LQTS as an adverse event were enriched in the LQTS neighborhood. With the LQTS neighborhood as a classifier, we predicted drugs likely to have risks for QT effects and we validated these predictions with the FDA’s Adverse Events Reporting System, illustrating how network analysis can enhance the detection of adverse drug effects associated with drugs in clinical use. Thus, the identification of disease-selective neighborhoods within the human interactome can be useful for predicting new gene variants involved in disease, explaining the complexity underlying adverse drug side effects, and predicting adverse event susceptibility for new drugs.