ESTRO 2020 Abstract book

S902 ESTRO 2020

Figure 1. Overall survival: 84% at 12 months and 55% at 36 months. Figure 2. Probability calibration curves. Brier scores: 0.06 (radiomic model 12m), 0.10 (volumetric model 12m), 0.13 (radiomic model 36m), 0.18 (volumetric model 36m). Conclusion Radiomic features, considering all metastases in one patient from baseline imaging and follow-up imaging at 3 months, allow for accurate prediction of survival in metastatic melanoma patients at 12 and 36 months after treatment with immune checkpoint-inhibitors. PO-1572 Use of intravoxel incoherent motion (IVIM) MRI for predicting dysphagia in oropharyngeal carcinoma M. Maddalo 1 , L. Altabella 2 , L. Pegurri 1 , A. Guerini 3 , G. Peretto 3 , A. Alghisi 3 , O. Turla 3 , A. Guaineri 3 , G. Costantino 3 , L. Spiazzi 4 , L. Costa 1 , C. Mozzetti 3 , N. Pasinetti 5 , M. Buglione 3 1 Spedali Civili di Brescia, Department of Radiation Oncology, Brescia, Italy ; 2 Azienda Ospedaliero - Universitaria di Parma, Department of Medical Physics, Parma, Italy ; 3 Unversità degli Studi di Brescia - ASST Spedali Civili di Brescia, Department of Radiation Oncology, Brescia, Italy ; 4 Spedali Civili di Brescia, Department of Medical Physics, Brescia, Italy ; 5 ASST Valcamonica, Depertment of Radiation Oncology, Esine BS, Italy Purpose or Objective The IVIM technique is an advanced MR imaging method that allows to obtain informations on diffusion, perfusion and pseudo-diffusion. This work is a preliminary analysis of the data of a prospective protocol on patients affected by OPC treated with radical radio-chemotherapy that aims to identify an IVIM signal pattern predictive of 1) early response during treatment and early changes to organs at risk 2) loco-regional recurrence of disease 3) acute and late toxicity. Material and Methods In this preliminary analysis, the IVIM MRI parameters (mean, median, skewness, kurtosis and others) of the pharynx constrictor muscles were evaluated before treatment (RM0) and during treatment (RM1). The parameters of the apparent diffusion coefficient (ADC), diffusion (D) and pseudo-diffusion (D*) coefficient maps were correlated with acute and late toxicity data reported both by the physician using the CTCAE scale (dysphagia > G3), and by the patient through patient reported outcomes (PROs), in particular analyzing the responses of the VHNSS- IT questionnaires (Vanderbilt Head and Neck Symptom Survey Italian version) of the subscales related to dysphagia (“nutrition”, “swallow-solid” and “swallow- liquid”). The dysphagia assessment was performed at baseline and at 3, 6 and 12 months of follow-up. Results The RM0 and RM1 data of 19 patients enrolled in the protocol were analyzed (Figure 1).

The analysis showed that the mean and median values of the ADC maps of the constrictor muscles of RM0 and RM1 were correlated with worse dysphagia scores (scores> 7) in VHNSS-IT. The values considered most significant were described using ROC curves (example in Figure 2). The dysphagia reported by the doctor according to the CTCAE scale was instead significant only at baseline and 3 months after the end of therapy. D and D* maps have shown to be less effective in predicting dysphagia. Conclusion The mean and median values of the ADC and D* maps were found to be higher in patients with worse scores of the swallow-solid and nutrition subscale, both in RM0 and this seems to reflect an intrinsic predisposition of the patient to develop major dysphagia, and in RM1 and this could to be correlated to the irradiation of the constrictor muscles. The data provided by the PROs were more accurate in describing dysphagia than the CTCAE toxicity scale. This is justified by the fact that by CTCAE scale assessment only those patients who require enteral or intensive parenteral nutrient supports are considered to have severe dysphagia. In the specific case of our study, only one patient at 6 and 12 months needed this suport. In such a small sample it is therefore difficult to discriminate only those patients with severe toxicity. Such analyzes should therefore be extended to a larger sample. PO-1573 First clinical experience with model-based selection for breast cancer proton therapy K. Verhoeven 1 , G. Vilches-Freixas 1 , K. Limpens 1 , J. Mannens 1 , K. Van der Klugt 1 , M. Velders 1 , F. Visser 1 , S. Peeters 1 , G. Bosmans 1 , L. Boersma 1 1 Maastro, Department of Radiation Oncology- GROW