Session Item

In the Danish Breast Cancer Group (DBCG) Skagen trial 1, 2963 high-risk breast cancer patients receiving loco-regional radiotherapy were randomized to 40 Gy in 15 fractions versus standard 50 Gy in 25 fractions. DICOM data were collected for all patients treated at the Danish centers between 2015 and 2021, allowing for investigation of inter-center variations in contouring practice and doses to organs at risk. While randomization was stratified according to center and systemic therapy thereby adjusting for potential inter-center variation, future analysis on other exposures could be affected by inter-center variations in radiotherapy practice between the centers.  

1937 treatment plans submitted from the seven participating Danish centers to the national DICOM collaboration system (DcmCollab) were available for analysis.  Seven clinical target volumes (CTVs) (CTVp_Breast/CTVp_Chestwall, CTVn_IMN, CTVn_L1, CTVn_L2, CTVn_L3, CTVn_L4) and six organs at risk (OAR) (Heart, A_LADCoronary, Lung_ipsilat, Lung_contralat, Breast_contralat and HumeralHead) were delineated by all centers according to ESTRO guidelines. Full dose volume histograms were calculated for all contoured volumes.

Volumes of CTVs and OARs stratified by center are shown in Figure 1. Good agreements were observed for all CTVs with only the CTVn_IMN showing relevant inter-center variation. For OAR the variation in lung delineations and contralateral breast between centers (data not shown) was low, whereas some variation was observed in both heart and humeral head delineations. Especially humeral head delineations varied in volume likely because several centers included more of the humeral bone than what should be included in the strict anatomical definition of the humeral head.

Since introduction of CT-based treatment planning around 2005-8, DBCG guidelines for both contouring and dose constraints have been agreed upon, and they are continuously evaluated and modified at national meetings and workshops. As a result most volumes were contoured consistently among centers, but surprisingly, systematic differences were present in some volumes. This illustrates the need for further actions in harmonizing delineation practices for these volumes. Future studies investigating outcomes related to dose to volumes such as the CTVn_IMN, heart and humeral head should consider the variation in delineation practices.