Comparison of Tumor Bed Delineation Using a Novel Radiopaque Filament Marker vs. Surgical Clips for Targeting Breast Cancer Radiotherapy

Abstract

<h3>Purpose/Objective(s)</h3> The accuracy of tumor bed (TB) delineation is essential for targeting boost dose or partial breast irradiation. Surgical clip markers (CM) are standardly used. However, despite using CM, multiple studies have shown high interobserver variability in TB delineation. We hypothesize that a radiopaque filament marker (FM) woven along the TB at the time of lumpectomy will improve accuracy of TB delineation. <h3>Materials/Methods</h3> An FDA-approved radiopaque filament was intraoperatively used to outline the tumor bed of patients undergoing lumpectomy. Between January 2020 and September 2020, 10 consecutive patients with the FM and 10 consecutive patients with the standard CM were identified for comparison. Five "experts," defined as board-certified radiation oncologists specializing in breast cancer, independently contoured all 20 tumor beds. Dice coefficients were then calculated for each patient by comparing the intersection and aggregate of each expert's contours with that of the other experts individually. An intersection and aggregate of the contours of all 5 experts was also analyzed for each patient. Dice coefficient of 1 represents perfect contour agreement, and 0 represents no agreement. Two-tailed paired samples t-tests were performed to compare dice coefficients between FM and CM cohorts. Furthermore, the center of mass (COM) coordinates were calculated to analyze the distance of the COM of each expert's contour to that of the aggregate of the other experts' contours for each patient. Two-tailed unpaired samples t-tests were performed to compare COM distances between FM and CM cohorts. <h3>Results</h3> For FM, mean dice coefficient and standard deviation for each expert was: 0.61 ± 0.09, 0.60 ± 0.10, 0.60 ± 0.10, 0.60 ± 0.09, and 0.55 ± 0.11, respectively. By comparison, for CM, mean dice coefficient and standard deviation for each expert was: 0.48 ± 0.22, 0.42 ± 0.25, 0.47 ± 0.21, 0.47 ± 0.22, and 0.40 ± 0.22, respectively. Thus, TB delineation agreement was significantly improved with FM (<i>P</i> = 0.0013, <i>P</i> = 0.0001, <i>P</i> = 0.0013, <i>P</i> = 0.0006, and <i>P</i> = 0.0008, respectively). Furthermore, the COM distance of each expert's contour to that of the aggregate of all experts' contours was significantly reduced (<i>P</i> < 0.001) from a mean of 0.96 cm ± 0.89 cm for CM to 0.25 cm ± 0.19 cm for FM. <h3>Conclusion</h3> Radiopaque filament resulted in significantly improved interobserver variability in tumor bed delineation compared with standard surgical clips. This difference was clinically meaningful with a reduction in mean COM contour disagreement from almost 1cm to 2.5mm, and suggests marked superiority of the radiopaque filament in targeting boost dose and/or partial breast irradiation.