S659
Clinical – Head & neck
ESTRO 2026
baseline and at least one post-RT MST from 2010-2021 were included. Original RT plans were retrieved and autocontouring applied to delineate submandibular glands (SMG) if absent. Spearman’s rank correlation coefficient was applied to identify the most predictive dose metrics for change in MST at one-year post-RT. Recognized thresholds such as mean dose, V20, V30, V40 and V50 for oral cavity, parotid and SMG were considered per structure (defined as left/right and ipsilateral/contralateral to high dose) and as combined structures for total volume and total % spared. AUC- ROC for the strongest metrics was calculated to quantify predictive performance and identify a threshold for reduction in MST <10 ml at one-year
Conclusion: Deep-learning-based auto-segmentation of LN seems to be a viable option for reducing contouring-time in epHNCP and producing clinically acceptable contours with only minor need for editing. This leads to a reduced interobserver-variability expressed by higher values of DSC and s-DSC. Keywords: autocontouring, planning, radiotherapy
post-RT. Results:
Among 1104 eligible patients, 698 received systemic therapy with RT. Median delivered dose was 70 Gy (range 46-70 Gy). At 1, 6 ,12 and 24 months, the median MST change and number of patients assessed was 18 mm (874 pts), 22 mm (628), 20 mm (481) and 17 mm (210) respectively. The strongest predictive factor of change in MST <10 ml at 1 year was combined SMG volume receiving <40 Gy (AUC = 0.824) (figure 1). The spared volume threshold required to prevent greater loss of function was 3.2 cc <40 Gy (sensitivity 0.79, specificity 0.80) (figure 2).
Digital Poster Highlight 4396
Dosimetric predictors of one-year non-stimulated salivary function following radiotherapy for head and neck cancer Nicole Wisener 1,2 , Iymad Mansour 1 , Nicholas Dietrich 1 , Ali Hosni Abdalaty 1 , Andrew Bayley 1 , Michael Glogauer 3 , Ezra Hahn 1 , Andrew Hope 1 , Shao Hui Huang 1 , John Kim 1 , Nauman Malik 1 , Brian O’Sullivan 1 , Susie Su 4 , John Waldron 1 , Tony Tadic 1 , Erin Watson 3 , Andrew McPartlin 1 1 Radiation Medicine Program, Princess Margaret Cancer Centre, Toronto, Canada. 2 Faculty of Medicine, University of Ottawa, Ottawa, Canada. 3 Dentistry, University Health Network, Toronto, Canada. 4 Biostatistics, University Health Network, Toronto, Canada Purpose/Objective: Chronic reduction in unstimulated saliva production following radiotherapy (RT) for head and neck cancer (HNC) is common and detrimental to quality of life1. Current dosimetric predictors of change such as mean dose and dose-volume metrics are imperfect, informed by relatively small patient series and heterogeneous outcome measures. To address this gap, we analyzed, in a large patient cohort, the association between dosimetric parameters and changes in unstimulated saliva production following RT, assessed via repeat modified Schirmer test (MST). This validated objective measure quantifies unstimulated saliva production via distance absorbed onto a litmus strip placed in the floor of mouth after 3 minutes. At our centre, MST is routinely performed at baseline and then at 1-, 6-, 12- and 24-months post-RT. Material/Methods: All HNC patients receiving definitive (chemo)-RT with
Made with FlippingBook - Share PDF online