Disproportionality analysis of off-label intravitreal bevacizumab in the FDA Adverse Event Reporting System database.

Background: Bevacizumab, an anti-vascular endothelial growth factor agent initially approved for cancer treatment, is widely used off-label in retinal vascular diseases; however, this use may carry safety risks, particularly when evidence from controlled clinical trials is limited. In this context, real-world evidence derived from large pharmacovigilance databases plays a critical role in evaluating its safety profile. Objectives: Evaluate the safety profile of off-label intravitreal bevacizumab for the treatment of retinal vascular diseases using the FDA Adverse Event Reporting System (FAERS) database. Design: Disproportionality analysis using data mining of the FAERS database. Methods: Individual case safety reports (ICSRs) of intravitreal bevacizumab from Q1 2015 to Q2 2025 were extracted, and data mining methods (reporting odds ratio, proportional reporting ratio and Bayesian confidence propagation neural network) were applied to identify statistical disproportionality. In addition, a comparative risk analysis (odds ratio) was performed against ranibizumab. Results: Of the 1495 ICSRs (5232 adverse events (AEs)), 114 preferred terms were statistically significant in the disproportionality analyses; 52% were expected, 15% were related to lack of efficacy and 33% were unexpected. Furthermore, the comparative analysis suggests a higher risk of serious cardiovascular, ocular and systemic events with ranibizumab. Conclusion: Most of the disproportionately reported AEs associated with bevacizumab were expected or related to product handling, underscoring the importance of using approved single-dose formulations. While this analysis suggests a lower risk of serious systemic AEs with bevacizumab than with ranibizumab, this finding should be interpreted with caution, as disproportionality analyses based on spontaneous reporting systems cannot confirm causal associations. Further evidence is needed to confirm the observed associations.