S2500
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
Proffered Paper 5191
A PET-PSMA–derived functional salivary gland atlas supports dose sensitivity of the tubarial glands in head and neck cancer radiotherapy Lucía Cubero 1,2 , Oscar Acosta 1 , Javier Pascau 2,3 , Xavier Palard-Novello 1 , Anaïs Barateau 1 , Renaud de Crevoisier 1 , Joël Castelli 1 1 Université Rennes, CLCC Eugène Marquis, Inserm, LTSI - UMR 1099, Rennes, France. 2 Departamento de Bioingeniería, Universidad Carlos III de Madrid, Madrid, Spain. 3 Instituto de Investigación Sanitaria, Gregorio Marañón, Madrid, Spain Purpose/Objective: Xerostomia remains a prevalent late toxicity after HNC RT despite parotid-sparing techniques [1]. While parotid damage is a major contributing factor, recent work by Valstar et al. [2] described the tubarial glands as a potential organ at risk for xerostomia development. This discovery has sparked debate regarding their role and the relevance of their preservation during treatment planning. This study aimed to construct a PET-PSMA–derived functional atlas of salivary glands and compare radiation dose distributions in these functional regions between patients with and without xerostomia. Material/Methods: The functional atlas was built with [ ¹⁸ F]-piflufolastat (PSMA) PET/CT scans from a population of 13 prostate cancer patients registered to a representative patient, selected as the anatomical template. A voxel-wise 65th percentile threshold of SUV across patients was applied to identify regions with consistently high PSMA uptake. To validate its utility in xerostomia studies, 60 oropharyngeal cancer patients from the ARTIX trial [1] were analyzed. Stimulated salivary flow (SSF) was measured 12 months after RT, with xerostomia defined as SSF ≤ 500 mg/min (grade ≥ 2). PSMA voxels with SUV ≥ 3 were used to segment the major salivary glands in the atlas. Planning dose maps of ARTIX patients were deformably registered to the template CT, flipped to align ipsilateral sides, and smoothed with a Gaussian kernel ( σ = 2 mm). For each identified gland, DVHs were extracted and compared between xerostomia and non-xerostomia patients using the Mann–Whitney U-test. Logistic regression models were fitted, and predictive performance was evaluated by bootstrap resampling (n = 1000) with AUC, odds ratio, and 95% CI. Results: The PET-PSMA–derived functional atlas successfully delineated all major salivary glands, including the parotid, submandibular, sublingual, and the previously described tubarial glands (Figure 1).
Mean doses were significantly higher in xerostomia cases for the tubarial, contralateral parotid, and contralateral submandibular glands (p < 0.05). The largest difference was observed in the tubarial glands (Figure 2), with mean doses of 59.4 Gy in xerostomia and 49.6 Gy in non-xerostomia patients.
Conclusion: This study demonstrates the feasibility of constructing a functional salivary atlas from PSMA-PET imaging and
Made with FlippingBook - Share PDF online