predicTox: An integrated database of clinical risk frequencies and human gene expression signatures for cardiotoxic drugs

Abstract

We have used drug-induced transcriptomic responses and whole-genome sequences in healthy human induced pluripotent stem cells (iPSC)-derived cardiomyocyte lines to identify cellular functional pathways and genomic variants potentially associated with the cardiotoxic effects of tyrosine kinase inhibitors (TKIs) and other cancer drugs. Here, we describe predicTox.org, an interactive website that organizes our data and its integration with knowledge from cell pathways and genomic databases. DrugTox summary cards give results of these analyses and metadata for each drug. Fields include FDA Adverse Event Reporting System (FAERS) cardiotoxicity risk scores, cell pathways, and genomic variants potentially associated with drug-induced cardiotoxicity. At a detailed level, predicTox provides ranked lists of pathway signatures that are common for cardiotoxic TKIs, up- and downregulated pathways associated with cardiotoxicity induced by specific TKIs and TKI-regulated genes mapping to those pathways. predicTox provides downloadable lists of drug-induced differentially expressed genes (DEGs) and pathways, drug-related genomic variants associated with cardiotoxicity, all with their statistical metrics, and mathematical models to simulate drug effects on heart physiology. Building on the results of our algorithm for independent reidentification of the well-known rs2229774 variant for anthracycline-induced cardiotoxicity, we describe how our data can be queried to identify potential variants associated with drug-induced cardiotoxicity by affecting a drugs pharmacodynamics and pharmacokinetics.