PO96: Treatment Planning Dosimetry Predictive of Treatment Plan Robustness in Permanent Breast Seed Implant

Purpose To identify pre-plan dosimetry metrics that are predictive of delivered clinical target volume coverage for early-stage breast cancer patients treated with the permanent breast seed implant technique. Materials and Methods Permanent breast seed implant (PBSI) is a brachytherapy technique for early-stage breast cancer patients in which stranded Pd-103 seeds are permanently implanted into and around the surgical seroma following lumpectomy. The CTV is defined as the seroma, plus any existing surrounding fibrosis visible on CT. The CTV is expanded 10 to 15 mm to create the PTV. The PTV is cropped to 5 mm from the skin surface and to the chest wall muscle. Pre-planning dose volume histogram (DVH) target coverage indices were collected from clinically treated patients. Pre-plan coverage metrics included the V90%, V100% and D90% for the CTV and the PTV. Using an in-house simulation software, 1000 simulated post-plans were created for each pre-plan, with needles shifted from their planned positions according to a distribution of clinically expected needle uncertainty. An Evaluation Metric was defined to describe the robustness of a given treatment plan. The Evaluation Metric was defined as the CTV V100% that was exceeded by 90% of the simulated post-plans generated from a given pre-plan. Pearson's correlation coefficients (R) were calculated to assess correlations between the DVH metrics from the pre-plans, and the Evaluation Metric (calculated from the simulated post-plans). An alpha level of 1% was used to describe significance. Results A total of 98 patients, treated at two institutions, were included in this study. Median and range values for the pre-plan coverage metrics were collected. These are shown in table 1. For the simulated post-plans, the Evaluation Metric was calculated over the 1000 simulations run for each patient and was found to be median (range) 92.2% (69.1% - 99.6%) across all patients. The correlation coefficients between each treatment plan coverage metric and the Evaluation Metric were 0.25 for CTV V90%, 0.36 for CTV V100%, 0.35 for CTV D90%, 0.30 for PTV V90%, 0.36 for PTV V100%, and 0.47 for PTV D90%. All R values were significant at the 1% level (p less than 0.01) except for CTV V90% (p = 0.013). Conclusions This investigation presents a preliminary assessment of the correlation of pre-planning coverage metrics with simulated post-plan robustness. The strongest correlations were found for the V100% and D90% for the PTV and CTV. This reinforces the importance of planned V100% coverage and motivates further consideration of the D90% in treatment planning. To identify pre-plan dosimetry metrics that are predictive of delivered clinical target volume coverage for early-stage breast cancer patients treated with the permanent breast seed implant technique. Permanent breast seed implant (PBSI) is a brachytherapy technique for early-stage breast cancer patients in which stranded Pd-103 seeds are permanently implanted into and around the surgical seroma following lumpectomy. The CTV is defined as the seroma, plus any existing surrounding fibrosis visible on CT. The CTV is expanded 10 to 15 mm to create the PTV. The PTV is cropped to 5 mm from the skin surface and to the chest wall muscle. Pre-planning dose volume histogram (DVH) target coverage indices were collected from clinically treated patients. Pre-plan coverage metrics included the V90%, V100% and D90% for the CTV and the PTV. Using an in-house simulation software, 1000 simulated post-plans were created for each pre-plan, with needles shifted from their planned positions according to a distribution of clinically expected needle uncertainty. An Evaluation Metric was defined to describe the robustness of a given treatment plan. The Evaluation Metric was defined as the CTV V100% that was exceeded by 90% of the simulated post-plans generated from a given pre-plan. Pearson's correlation coefficients (R) were calculated to assess correlations between the DVH metrics from the pre-plans, and the Evaluation Metric (calculated from the simulated post-plans). An alpha level of 1% was used to describe significance. A total of 98 patients, treated at two institutions, were included in this study. Median and range values for the pre-plan coverage metrics were collected. These are shown in table 1. For the simulated post-plans, the Evaluation Metric was calculated over the 1000 simulations run for each patient and was found to be median (range) 92.2% (69.1% - 99.6%) across all patients. The correlation coefficients between each treatment plan coverage metric and the Evaluation Metric were 0.25 for CTV V90%, 0.36 for CTV V100%, 0.35 for CTV D90%, 0.30 for PTV V90%, 0.36 for PTV V100%, and 0.47 for PTV D90%. All R values were significant at the 1% level (p less than 0.01) except for CTV V90% (p = 0.013). This investigation presents a preliminary assessment of the correlation of pre-planning coverage metrics with simulated post-plan robustness. The strongest correlations were found for the V100% and D90% for the PTV and CTV. This reinforces the importance of planned V100% coverage and motivates further consideration of the D90% in treatment planning.