When Side Effects Become Therapies: A Mechanistic Taxonomy of Adverse Drug Reactions as Translational Signals for Drug Repositioning

Background: The systematic conversion of adverse drug reactions (ADRs) into new therapeutic indications represents one of the oldest and most productive sources of repositioned drugs; however, it has not been subjected to formal mechanistic analysis. Objectives: This study aims to propose a mechanistic taxonomy of convertible ADRs and a structured workflow for translating documented adverse effects into testable repositioning hypotheses. Methods: In this narrative review, historical cases of ADR-driven repositioning spanning five decades—from sildenafil and minoxidil to semaglutide—were analyzed and classified within a four-dimensional framework organized along the axes of: (1) target identity (on-target versus off-target conversion); (2) dose, route of administration, and reformulation; (3) patient population and clinical context; and (4) pharmacogenomic variability. Results: Each historical case was successfully classified within the proposed taxonomy. A five-step theoretical workflow—comprising signal identification, taxonomic classification, mechanistic verification, feasibility assessment, and preliminary evidence search—was developed to operationalize the framework for prospective hypothesis generation. Computational tools including side-effect similarity networks, network pharmacology, and transcriptomic signature reversal were identified as means to support specific workflow steps. Conclusions: The proposed taxonomy and workflow provide a replicable, mechanism-informed approach to ADR-driven drug repositioning, while also delineating the pharmacological, regulatory, and methodological challenges that separate a repositioning signal from an approved therapeutic indication.