Dosimetric consequences of image guidance techniques on robust optimized intensity-modulated proton therapy for treatment of breast Cancer

PURPOSE: To investigate the consequences of residual setup error on target dose distribution using various image registration strategies for breast cancer treated with intensity-modulated proton therapy (IMPT). MATERIALS AND METHODS: Among 11 post-lumpectomy patients who received IMPT, 44 dose distributions were computed. For each patient, the original plan (Plan-O) and three verification plans were calculated using different alignments: bony anatomy (VPlan-B), breast tissue (VPlan-T), and skin (VPlan-S). The target coverage were evaluated for each alignment technique. Additionally, 2 subvolumes-BreastNearSkin (1-cm rim of anterior CTV) and BreastNearCW (1-cm rim of posterior CTV)-were created to help localize CTV underdosing. Furthermore, we divided the setup error into the posture error and breast error. Patients with a large posture error and those with good posture setup but a large breast error were identified to evaluate the effect of posture error and breast error. RESULTS: For Plan-O, VPlan-B, VPlan-T, and VPlan-S, respectively, the median (interquartile range) breast CTV D95 was 95.7%(94.7-96.3%), 95.1% (93.9-95.7%), 95.2% (94.8-95.6%), and 95.2% (94.9-95.7%); BreastNearCW D95 was 96.9% (95.6-97.3%), 94.8% (93.5-97.0%), 95.6% (94.8-97.0%), 95.6% (94.8-97.1%); and BreastNearSkin D95 was 94.1% (92.7-94.4%), 93.6% (92.2-94.5%), 93.5% (92.4-94.5%), and 94.4% (92.2-94.5%) of the prescription dose. 4/11 patients had >/=1% decrease in breast CTV D95, 1 of whom developed breast edema while the other 3 all had a > 2(o) posture error. The CTV D95 variation was within 1% for the patients with good posture setup but >2(o) breast error. CONCLUSION: Acceptable target coverage was achieved with all three alignment strategies. Breast tissue and skin alignment maintained the breast target coverage marginally better than bony alignment, with which the posterior CTV along the chest wall is the predominant area affected by under-dosing. For target dose distribution, posture error appears more influential than breast error.