Geometric and dosimetric impact of heart and cardiac substructure contouring variability in left-sided breast radiotherapy using VMAT-FFF: A FB versus DIBH analysis.

INTRODUCTION

Accurate heart and cardiac substructures (CS) contouring is essential for cardiac-sparing in left-sided breast radiotherapy. However, inter-observer variation (IOV) persists. While previous studies mainly evaluated IOV in free-breathing (FB) using three-dimensional conformal radiotherapy, deep-inspiration breath-hold (DIBH)'s impact on geometric consistency and dose robustness with modern techniques remains unclear. We aimed to evaluate the heart and 15 CS IOV in paired FB-DIBH scans, and to assess its dosimetric impact using Halcyon™ flattening-filter-free volumetric modulated arc therapy (VMAT-FFF), while examining breathing-related effects on geometric and dosimetric outcomes.

METHODS

Fifty patients underwent FB/DIBH non-contrast computed tomography scans. Two independent observers contoured the heart and CS. Reference VMAT-FFF plans were recalculated on alternate contours using identical geometry/optimization settings. IOV was quantified using seven volumetric/distance-based indices. Reliability of mean/maximum doses (Dmean/Dmax) was evaluated using intraclass correlation coefficients (ICC), and dosimetric impact was quantified as ΔD = |D_alternate-D_reference|. Spearman correlations (r) assessed relationships between geometric indices/ΔD.

RESULTS

Agreement was high for heart (Jaccard: 0.97-FB/DIBH) and most CS but lower for pulmonary veins (Hausdorff: 11.52-mm-FB vs. 10.99-mm-DIBH) and coronary arteries (CAs). DIBH improved consistency for selected CS, including circumflex artery (mean surface distance: 1.33-mm-FB vs. 0.60-mm-DIBH; p < 0.05), while FB showed lower right atrium IOV. Dosimetric reliability was excellent in DIBH (ICCs>0.90 for 93.75 % of structures for Dmean/Dmax vs. 81.25 %/75 % in FB). DIBH significantly reduced CAs ΔDmean, while FB reduced pulmonary artery ΔDmean (p = 0.018). Correlations varied by structure/breathing, from strong (r > 0.5; p < 0.05) to absent.

CONCLUSION

CAs and pulmonary veins remain IOV-sensitive. DIBH enhances dose robustness, particularly for CAs. Geometric indices alone insufficiently predict dosimetric impact, reflecting complex anatomy-breathing interplay.

IMPLICATIONS FOR PRACTICE: Integrated geometric-dosimetric assessment enhances cardiac-sparing consistency, particularly in DIBH, supporting careful review of small or complex CS.