S571
Clinical – Head & neck
ESTRO 2026
automatically segmented on the PET/CT images using an in-house deep learning-based algorithm [2]. The resulting delineations were then reviewed and corrected, if necessary, by an experienced nuclear physician. Special attention was given to accurate segmentation of ipsilateral glands, where tumor proximity may confound metabolic signal interpretation. Then, SSF measurements at 12 months were used as a binary endpoint, with a value of 500 mg/min or less being used as a threshold for xerostomia as in the initial clinical study [1]. Finally, PET/CT based measurements at baseline were used to predict xerostomia considering each gland type and laterality separately and in combination. Results: Baseline PET/CT-derived features (SUVMax and volume) were predictive of xerostomia at 12 months, see figure 2. ROC analysis, see figure 1, compared 5 prediction approaches: i) parotid glands only, ii) submandibular glands only, iii) contralateral glands only, iv) ipsilateral glands only and v) all glands combined.While good results are obtained with the contralateral glands (AUC=0.81), as previously reported [3], as well as with the parotid glands alone (AUC=0.81), the best results were achieved when both contralateral and ipsilateral glands were included (AUC=0.91), demonstrating the contribution of ipsilateral glands in the final prediction.Figure 1: ROC analysis of gland based prediction models
with marginal significance observed for OS (P=0.071). Conclusion: DWI-guided DP-VMAT combined with chemotherapy demonstrates superior survival outcomes compared to conventional MRI-based IMRT in locally advanced HNSCC patients, while maintaining comparable toxicity profiles. References: 1.Tan C, Li Y, Jiang C, He L, Xiao S, Fan C, Ye X, Zhao Q, Wu W, Li Y, Liu F. DWI-guided DP-IMRT and conventional MRI-based IMRT in head and neck squamous cell carcinoma: a comparative study. BMC Cancer. 2025 Aug 23;25(1):1364. 2.F DEF, Musio D, Bulzonetti N, Maghella F, Tombolini V. Target volume delineation based on Diffusion-weighted magnetic resonance imaging for locally advanced head and neck Cancer. Anticancer Res. 2016;36(8):4181– 5.3.Houweling AC, Wolf AL, Vogel WV, Hamming-Vrieze O, van Vliet-Vroeginde weij C, van de Kamer JB, van der Heide UA. FDG-PET and diffusion-weighted MRI in head-and-neck cancer patients: implications for dose painting. Radiotherapy Oncology: J Eur Soc Therapeutic Radiol Oncol. 2013;106(2):250–4. Keywords: LA-HNSCC ， DWI-guided Dose Painting Radiotherapy Inclusion of ipsilateral salivary gland PET/CT features improves xerostomia prediction after radiotherapy in head and neck cancer Luis Torres 1 , Santiago F Osorio-Botero 1 , Xavier Palard 1 , Renaud De Crevoisier 1 , Lucía Cubero 2 , Oscar Acosta 1 , Joël Castelli 1 , Mathieu Rubeaux 1 1 Universite de Rennes, CLCC Eugène Marquis, Rennes, France. 2 Bioengineering, Universidad Carlos III de Madrid, Madrid, Spain Purpose/Objective: Xerostomia is a major side-effect associated with radiotherapy in head and neck cancer (HNC) patients and can be clinically assessed through the stimulated salivary flow (SSF) rate. This study evaluates the potential of metabolic 18F-FDG PET salivary glands measurements as a noninvasive alternative for xerostomia assessment and prediction. In particular, we emphasize the added value of incorporating ipsilateral gland information. Material/Methods: PET/CT images and clinical data from 25 stage III-IVB HNC patients enrolled in the prospective trial NCT01874587 were included in the analysis. SSF measurements and PET/CT acquisitions were performed at baseline and 12 months post-treatment, capturing the evolution of xerostomia over time [1]. Parotids and submandibular glands were Digital Poster Highlight 1617
Figure 2: Evolution of gland volume and SUVMax for SSF (mg.min-1) of ipsilateral glands at time of inclusion and after 12 months.
Made with FlippingBook - Share PDF online