Evaluation of Dosimetric Consequences of Seroma Contour Variability in Accelerated Partial Breast Irradiation Using a Constructed Representative Seroma Contour

Purpose: Contouring variability of the seroma can have important implications in the planning and delivery of accelerated partial breast irradiation (APBI). This study aimed to quantify the dosimetric impact of these interobserver and intraobserver contouring variations by construction of a representative seroma contour (RSC). Methods and Materials: Twenty-one patients with a seroma suitable for APBI underwent four computed tomography (CT) scans: one planning CT and three additional CTs on the first, third, and fifth days of treatment. Three radiation oncologists contoured the seroma on each CT scan. For 3 patients, oncologists repeated contouring twice to assess intraobserver variations. Seroma contour variability was quantified by construction of an RSC. In addition, the percent volume overlap (PVO) was calculated. Root-mean-square (RMS) differences in seroma volume, size, and center of mass position compared to those of the RSC were calculated. Treatment fields from the original plan were applied to the repeated CTs by using the same isocenter shifts as the original plan. The dosimetric impact of the contour variations was assessed using V95 (volume receiving at least 95% of the prescribed dose) and equivalent uniform dose (EUD). Results: Interobserver RMS volume differences were, on average, 5.6 times larger than intraobserver differences. The median interobserver RMS seroma volume difference was 1.48 cm3. The median PVO was 51.6%. V95 and EUD of the seroma contours were similar for all patients. The median RMS differences of the seroma V95 and EUD were 0.01% (range, 0%–3.99%) and 0.05 Gy (range, 0–0.98 Gy). Conclusions: Construction of the RSC showed that interobserver variations were most responsible for contour variations of the seroma. Current planning margins provided adequate dose coverage of the seroma despite these contour variations. [Copyright &y& Elsevier]