Induction Mechanism of Drug-induced Pruritus Using Food and Drug Administration Adverse Events Reporting System Database and Machine Learning

Abstract

We attempted to identify drugs that induce pruritus, using the Food and Drug Administration Adverse Events Reporting System, and analyze biochemical pathways related to the pathogenesis of drug-induced pruritus. The reported odds ratio and P-values for each drug and adverse drug reaction of pruritus were calculated comprehensively. Both statistics were plotted to visualize the pruritus-inducing tendency of the drugs. Additionally, a principal component analysis was also performed using the reported odds ratios, and a hierarchical cluster analysis based on the calculated principal components was performed. On the contrary, the agonist/antagonist activity of pharmaceuticals on the nuclear receptor/stress response pathway was predicted by a machine-learning quantitative structure–activity relationship model to explore biochemical pathways associated with pruritus onset. Systemic antibacterials, ophthalmologicals, and contrast media were identified as the major triggering drugs. In the principal component analysis, the second and third principal components could be interpreted as indicators of the mucocutaneous/dermal system and presence or absence of surgical procedures, respectively. In the hierarchical cluster analysis, systemic antibacterials, immunosuppressants, and antineoplastic agents strongly correlated with the first principal component. A search for relevant biochemical pathways revealed that TSHR, PR, and PXR were significantly correlated. These findings may be applicable to drug selection to avoid pruritus in clinical practice.