ESTRO 2021 Abstract Book

S1523

ESTRO 2021

Conclusion In the era of personalised medicine and tailored treatments, a model that accurately predicts toxicity could represent a useful clinical tool for better patients’ selection. The study shows that the performance of the predictive models is highly dependent on the choice of classifier, features and training/testing sets. Differences between groups can be due to the different characteristics of trials. Further analyses should focus on features and classifiers selection, to improve the performances and generalizability of the models. PO-1797 Relating proton LETd to biological response of salivary glands using PSMA-PET in clinical patients D. Wagenaar 1 , J. Langendijk 1 , V. Mohan 2 , R. Steenbakkers 1 , W. Vogel 3 , S. Both 1 1 University Medical Center Groningen, Radiation Oncology, Groningen, The Netherlands; 2 The Netherlands Cancer Institute (NKI), Radiation Oncology, Amsterdam, The Netherlands; 3 The Netherlands Cancer Institute (NKI), Radiation Oncology and Nuclear Medicine, Amsterdam, The Netherlands Purpose or Objective Variations of the relative biological effectiveness (RBE) of proton therapy with the dose-weighted averaged linear energy transfer (LETd) are predicted from pre-clinical experiments. However, RBE values may substantially vary for different endpoints. Therefore, RBE models need to be validated against clinical endpoints before introduction into clinical treatment planning optimization. An ongoing clinical trial investigating the use of PSMA-PET for head and neck cancer (HNC) patients quantitatively measures the biological response of salivary glands to photon and proton radiotherapy. In the current study, the results of included patients treated with proton therapy are used to investigate a possible relation between biological response and LETd. Materials and Methods Six patients treated with proton therapy were included. These patients received a PET-CT scan using 68Ga (N=1) or 18F (N=5) PSMA before treatment (baseline), during treatment and one month after the last fraction (follow-up). All analyses were performed in RayStation v9AR (v8.99.30.71). Physical dose (Dphys), D*LETd and the follow- up PSMA-PET scan were deformed to the baseline PET-CT using deformable image registration. Parotid gland delineations were adjusted to include voxels which had an uptake of ≥5 g/ml in the baseline PSMA-PET scan. Voxels were smoothed in a 6-mm cube to reduce effects of noise and registration errors. Relative PSMA uptake was defined as the ratio of uptake on the follow-up and baseline PSMA-PET scans. The LETd was calculated from the D*LETd and Dphys. Results A clear relation was observed between physical dose and relative PSMA uptake for five out of six patients (Figure 1). For some patients, PSMA uptake in low-dose voxels was higher in the follow-up scan compared to the baseline scan which may relate to compensation mechanisms within the salivary glands or day-to-day variations in PSMA uptake. For four out of six patients a relation between relative PSMA uptake and LETd could be observed, where the relative PSMA uptake was lower as voxels had higher LET but identical physical dose (Figure 2).