Reverse translation of adverse event reports paves the way for derisking preclinical off-targets

Despite its important role, the Food and Drug Administration Adverse Event Reporting System (FAERS) – the primary source for postmarketing pharmacovigilance - remains susceptible to reporting biases, including poor standardization, admissible drug synonyms and stimulated reporting. We have designed improved parsing methods to illuminate caveats that must be considered when using this and other post-marketing datasets to provide guidance and tools for medical professionals and researchers to enable bias-free analysis of the FAERS database. To decrease the noise in drug-ADR signals, and to reinforce important associations, we mapped over 500,000 drug identifiers used in FAERS to normalized chemical structures of their ingredients and introduced time-resolved analysis of ADR reporting which reveals similarities between drugs and adverse events across therapeutic classes, enabling unbiased classification of adverse events, indications, and drugs with similar clinical profiles as demonstrated with the case of celecoxib and rofecoxib. We also investigated key idiosyncrasies in the data, such as confusion between reported indications of drugs and their ADRs, multiplications, and the correlation of reported adverse effects with public events, regulatory announcements, and with scientific publications. The comparison of the pharmacological, pharmacokinetic and clinical ADR profiles of methylphenidate, aripiprazole and risperidone demonstrates how underlying molecular mechanisms can be an emergent property of co-analysis of ADRs. The precautions and methods of analysis presented here enable investigators to avoid confounding chemistry-based and reporting biases in FAERS, which is increasingly mined in the community, and to illuminate the possibility to conduct comaparative analysis of ADRs in association with their underlying mechanisms