3152 - Evaluation of Robustness against Breast Deformation in Accelerated Partial Breast Irradiation: A Phantom Study

Purpose/Objective(s):

In volumetric modulated arc therapy (VMAT) for whole-breast radiotherapy, breast deformation remains a challenge. However, the extent of dosimetric uncertainties in localized treatments, including accelerated partial breast irradiation (APBI) using VMAT, is unclear. Therefore, we evaluated the robustness of APBI-VMAT plans against breast deformation, focusing on robust optimization (RO) and virtual bolus (VB) methods.

Materials/Methods:

Three CT datasets (CT_-5, CT₀, and CT₊₅) were acquired using a custom breast phantom capable of simulating 5-mm deformations. To ensure contour consistency, three clinical target volumes (CTVs) representing the medial, central, and lateral regions were delineated on CT₀ and propagated to the CT_-5 and CT₊₅ datasets by deformable image registration. As a result, total of nine CTVs were delineated. Planning target volumes were created by adding 5-mm margin to each CTV. Four APBI-VMAT plans were generated using Raystation 10ASP1 (RaySearch Laboratories, Stockholm, Sweden): robust optimization (RO), virtual bolus with densities of 0.4 (VB04) and 1.0 g/cm³ (VB10), and standard optimization (SO) without a robust approach. These plans were designed for a TrueBeam linear accelerator using a 6-MV flattening filter-free photon beam with 30 Gy delivered in five fractions. Dose distributions under 5-mm breast shrinking and swelling were obtained by recalculating these plans on the other datasets. Planning methods were compared based on the relative differences in two representative dose-volume histogram parameters (CTV D_95% and Body D_max) between the original and recalculated dose distributions. To validate the calculated dose distributions, measurements were perfomed using radiochromic film. Measured and calculated dose distributions were compared via global gamma analysis, with acceptance criteria of 3%/2 mm for dose differences/distance to agreement and a 30% threshold.

Results:

For a 5-mm breast deformation, the mean relative differences in CTV D_95% for VB04 and VB10 were within ±1%, whereas those for SO and RO were within ±2%. Regarding Body D_max, RO, VB04, and VB10 maintained equivalent stability compared to SO (1.7% ± 0.4%) under 5-mm shrinkage. However, under 5-mm swelling, Body D_max increased for most methods, with SO, RO, and VB04 reaching 6.3%, 3.7%, and 7.0%, respectively. Notably, VB10 remained consistently robust, maintaining a mean relative difference of 1.6% ± 1.4% even under breast swelling. The gamma analysis results demonstrated excellent agreement between the measured and calculated dose distributions. The gamma pass rates for all measurements exceeded 90%, with an overall mean value of 96.7% ± 2.5%.

Conclusion:

Although CTV coverage was maintained across all methods, VB10 provided the highest protection for Body D_max under both shrinking and swelling. This planning method was the most robust against 5-mm breast deformations in APBI-VMAT.