Evaluation of skin dose robustness in breast radiotherapy using VMAT planning technique: a phantom-based surface dosimetry study

OBJECTIVE: Minor anatomical changes from evolving edema during breast radiotherapy can compromise skin dose, particularly for VMAT. This study experimentally evaluated the robustness of conventional, skin-flash, and robust-optimization-based VMAT techniques against simulated contour variations using phantom-based surface dosimetry. 
Approach: Two configurations of an anthropomorphic phantom simulated clinical scenarios: with a breast attachment (large/reconstructed) and without (small/postmastectomy flat chest wall). Four plans were compared: 3D-CRT, conventional VMAT, VMAT with skin-flash, and VMAT with robust optimization. Plan robustness was assessed via TPS simulations and physical OSLD measurements at five surface points. To simulate controlled rigid surface contour change, the couch was shifted anterior-posteriorly by ±2, ±4, and ±6 mm as a simplified surrogate for edema-induced variations. Shifted OSLD measurements were normalized to the 0-mm baseline, and absolute dose deviations were evaluated against the 3D-CRT reference. 
Main results: TPS simulations suggested VMAT with skin-flash offered the greatest stability; however, physical measurements showed the opposite. Conventional VMAT was more variable than 3D-CRT (SD 9.47% vs 3.62% at ±6 mm), and the evaluated VMAT with skin-flash implementation was the least robust (SD 12.60%; worst-case minimum dose 62.49% with 75% of points outside ±5%), despite its favorable TPS prediction. In contrast, VMAT with robust optimization approached the stability of 3D-CRT (SD 5.48%) and showed even smaller deviations than 3D-CRT in the phantom with a breast attachment. 
Significant: These physical measurements demonstrate that, contrary to TPS predictions, the evaluated VMAT with skin-flash implementation does not ensure skin-dose robustness, whereas VMAT with robust optimization substantially improved robustness to a level comparable to that of 3D-CRT, with the greatest benefit observed in the phantom with a breast attachment.