ESTRO 2025 - Abstract Book

S1102

Clinical – Head & neck

ESTRO 2025

3833

Mini-Oral HERVIN: variability in target selection and delineation for head and neck cancer radiotherapy in Europe Gaber Plavc 1,2 , Guillaume Beldjoudi 3 , Primož Peterlin 1 , Vincent Grégoire 3 1 Department of radiotherapy, Institute of Oncology Ljubljana, Ljubljana, Slovenia. 2 Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia. 3 Radiotherapy Department, Centre Léon Bérard, Lyon, France Purpose/Objective: We evaluated interobserver variability in clinical target selection and clinical target volume (CTV) delineation among European radiation oncologists (RO) treating head and neck cancer. The aim was to also identify areas of significant discrepancies within CTVs to inform guideline revision and improve contouring consistency. Material/Methods: Participants were invited through national RO societies and the ESTRO. They were provided with a short questionnaire and DICOM CT images of a patient with right-sided oropharyngeal carcinoma, including pre contoured gross tumor volumes (GTVt and GTVn) for CTVs delineation. Their contours were compared to gold standard (GS) CTVs delineated by the lead author of international guidelines using metrics such as the Hausdorff Distance (HaD) and Dice Similarity Coefficient (DSC). Results: Among 354 survey participants, concordance with GS target selection was high for ipsilateral nodal region 1b (91%), 2 (100%), 3 (100%) and 4a (98%), but significantly lower for regions 5 (48%) and 7a (35%). Contralateral concordance was high for regions 2 (97%), 3 (97%) and 4a (88%). However, some participants' CTV included regions not defined in the GS such as ipsilateral regions 4b (44%), 6a (6%), 6b (4%), 7b (27%), and controlateral regions 1b (26%), 4b (35%), 5 (11%), 7a (19%). Analysis of contours from 252 participants across 30 European countries (61% from academic institutions, 43% with >10 years of experience, treating an average of 40 head and neck patients annually) showed an interquartile range (IQR) of 261–325 cc (median 288 cc; GS 250 cc) for total contoured CTV and 99 to 113 cc for high-dose (HD) CTV (median 105 cc, GS 101 cc). The median volume extending beyond the GS was estimated at 11 cc for HD CTV and 67 cc for total CTV volume. Major deviations from the GS were observed in both the controlateral nodal CTVs (median DSC 0.73, median HaD 44 mm) and the ipsilateral nodal CTVs (median DSC 0.77, median HaD 32 mm). Median percentages of GS ipsilateral neck regions not covered by participants’ CTVs were 9% (region 1b), 4% (region 2), 7% (region 3), 17% (region 4), 94% (region 5), and 12% (region 7a). Conclusion: This largest performed study on interobserver variability in head and neck cancer radiotherapy revealed substantial variations in target selection and delineation among European oncologists. These findings emphasize the need for targeted training to standardize pratices. The observed variability also offers opportunities to explore its impact on treatment outcomes through large-scale real-world data studies.

Keywords: head and neck, radiotherapy, observer variation