Pharmacovigilance and Real-World Evidence of ACTB-Linked Therapeutics in Breast Cancer: A Review

Abstract

This study explores the integration of artificial intelligence (AI) into pharmacovigilance frameworks for β-actin (ACTB)-targeting therapeutics in breast cancer, emphasizing enhanced detection and management of cytoskeleton-related adverse events (AEs). Taxane-induced dose-dependent neuropathy and actin-specific toxicities such as alopecia and edema exhibit variability across breast cancer subtypes, with complex interactions in combination therapies. Mining of large real-world databases (FAERS, WHO-VigiBase) reveals elevated safety signals linked to ACTB disruption, while natural language processing of social media data uncovers patient-reported outcomes correlating with these toxicities. The pharmacovigilance platform exemplifies an AI-augmented approach leveraging prompt-based extraction, machine learning, and real-time signal prioritization to improve causal inference and safety monitoring throughout drug development and post-marketing phases. A phased integration framework is proposed, enabling preclinical off-target risk simulation, clinical trial signal detection, and post-approval risk mitigation, enhanced by iterative prompt engineering tailored to breast cancer molecular subtypes. Retrospective application to paclitaxel data exposed previously underrecognized actin-related cardiotoxicity, illustrating AI’s potential to uncover subtle, clinically meaningful safety concerns. Future directions include federated learning for global pharmacovigilance cooperation and deeper integration with electronic health records to augment real-world evidence. Overall, this study highlights the transformative role of AI-enabled pharmacovigilance platforms in advancing precise, patient-centered safety surveillance for ACTB-modulating cancer therapies.