Purpose This study aimed to systematically identify drug‐associated signals for hepatic veno‐occlusive disease/sinusoidal obstruction syndrome (VOD/SOS) using large‐scale real‐world data and to decode the underlying molecular mechanisms of high‐risk agents. Methods We analyzed FDA Adverse Event Reporting System (FAERS) data (2004–2024) using four distinct disproportionality algorithms to strictly detect VOD/SOS safety signals. To validate these signals and elucidate pathogenic mechanisms, we performed an in silico integrative analysis by intersecting the transcriptomic profiles of representative high‐risk drugs (Busulfan and Oxaliplatin) with known VOD/SOS disease targets. Results Among 3726 VOD/SOS reports identified from over 21 million records, 39 drugs met the stringent criteria for a positive safety signal. Alkylating agents (e.g., Busulfan), antimetabolites (e.g., Thioguanine), and antibody–drug conjugates exhibited the strongest statistical associations. Age‐stratified analysis revealed distinct risk profiles, identifying Busulfan and Oxaliplatin as the most frequent suspects in adults and pediatric populations, respectively. Integrative transcriptomic validation uncovered divergent molecular drivers: Busulfan induces toxicity primarily through DNA damage response and apoptotic signaling (involving CDKN1A and GADD45A ), whereas Oxaliplatin is linked to inflammatory responses and tissue development pathways (involving PHLDA3 and MYF5 ). Conclusion This study integrates big data pharmacovigilance mining with transcriptomic validation to provide a comprehensive landscape of drug‐induced VOD/SOS. We identified distinct age‐specific risk profiles and revealed that different high‐risk agents drive hepatic toxicity through divergent molecular mechanisms, offering novel potential biomarkers for precision monitoring. However, safety signals identified for anti‐infective agents should be interpreted with caution, as they likely reflect confounding by indication in high‐risk populations rather than direct hepatotoxicity.
He et al. (Thu,) studied this question.