[Visualization and analysis of adverse reactions of molecularly targeted anticancer agents using the self-organizing map (SOM)]

Abstract

Molecularly targeted anticancer agents cause a variety of adverse reactions compared with conventional anticancer agents because of their unique mechanisms of action. Sources of drug information such as package inserts (PIs) provide primarily document-based and numerical information. Therefore it is not easy to obtain a complete picture of drugs with similar effects, or to understand differences among drugs. In this study we used the self-organizing map (SOM) technique to visualize the adverse reactions indicated on PIs of 23 molecularly targeted anticancer agents as of March 2013. In both the presence/absence version and the frequency version, SOM was divided into domains according to mechanism of action, antibody drug or low-molecular weight drug, and molecular target. The component planes of the 753 adverse reaction items in the frequency version enabled us to grasp all available information and differences among the drugs. In some component planes in the presence/absence version, an adverse reaction that had not been reported for a drug but had already been reported for its proximally positioned drug(s) as of March 2013, was found to be reported thereafter by the Drug Safety Update (DSU) or the Adverse Event Report Search System "CzeekV," which is based on FDA Adverse Event Reporting System (FAERS). Our results suggest that visualization of the adverse reactions of molecularly targeted anticancer agents by the SOM technique is useful not only to acquire all available information and differences among drugs, but also to predict the appearance of adverse reactions.