Cardiac substructure dosimetry in thoracic radiotherapy: moving beyond mean heart dose and toward precision cardiac risk mitigation

Background Radiation-induced heart disease (RIHD) remains a clinically significant consequence of thoracic radiotherapy (RT). Historically, the mean heart dose (MHD) has served as the primary metric for cardiac risk estimation. However, accumulating evidence demonstrates that MHD inadequately captures the heterogeneous dose distribution across functionally distinct cardiac substructures and poorly predicts specific cardiovascular endpoints. Objective This narrative review synthesizes recent evidence (2013–2025) on the transition from whole-heart dosimetry to substructure-level dose–volume analysis in thoracic RT, encompassing dose–response relationships for specific cardiac substructures, advances in artificial intelligence (AI)-driven auto-segmentation, modern cardiac-sparing delivery techniques, biomarker-guided surveillance, pharmacologic cardioprotection, and normal tissue complication probability (NTCP) modeling. Methods A structured literature search was conducted across PubMed, Scopus, and Web of Science databases (January 2013–March 2025) using predefined search terms. Studies reporting cardiac substructure dose–response relationships, auto-segmentation performance, cardiac-sparing techniques, biomarkers, pharmacologic interventions, and NTCP models were included. Results Substructure-specific dose metrics—particularly dose to the left anterior descending artery (LAD), heart base, left ventricle (LV), and cardiac conduction system—predict coronary events, arrhythmias, and overall survival with substantially greater accuracy than MHD. Deep learning auto-segmentation now achieves ≥ 94% clinical acceptability for cardiac chambers, enabling routine substructure-based planning. Deep inspiration breath hold (DIBH) reduces heart and LAD doses by 36–42%, and proton beam therapy (PBT) offers superior cardiac sparing across all tumor sites, though randomized cardiac outcome data remain pending. Statins demonstrate a compelling retrospective cardioprotective signal (up to 66% reduction in major adverse cardiac events), but prospective validation is lacking. The 2022 European Society of Cardiology (ESC) cardio-oncology guidelines and the 2024 American Radium Society (ARS) appropriate use criteria provide the current clinical framework. Conclusion The paradigm shift from MHD to substructure-level cardiac dosimetry represents a transformative advance in thoracic RT. Integration of AI-driven segmentation, advanced delivery techniques, biomarker surveillance, and pharmacologic cardioprotection into routine clinical workflows is essential for precision cardiac risk mitigation in irradiated patients.