ESTRO 37 Abstract book

ESTRO 37

S400

Material and Methods We prospectively enrolled 82 cases with ESCC into a cohort study which underwent DWI before CRT . All MR examinations included axial T2WI, T1WI and diffusion- weighted sequences (b=0, b=600 s/mm 2 ). Two groups of tumor features were examined: (1) clinical features (eg, TNM stage, age and gender) and demographics; (2) spatial texture features of apparent diffusion coefficient (ADC), which characterize tumor intensity range, spatial patterns and distribution and associated changes resulting from CRT. A reproducible and no redundant feature set was statistically filtered and validations. A separate radiomics nomogram to predict over survival (OS) was developed using Cox proportional hazards regression model. Results In total, 82 consecutive patients with ESCC, who had all nomogram variables available, were identified. The radiomics signature, which consisted of 9 selected features, was significantly associated with over survival (P <0.001 for both test and retest cohorts). A individualized prediction nomogram based on the radiomics signature(P<0.005), radiation dose (P<0.001), pathology Lesion length (P<0.01), and treatment response (P<0.05) was developed. The radiomics nomogram for OS had a concordance index of 0.72 in the validation set, indicating accurate prediction of OS. Conclusion Our raiomics nomograms predict OS accurately and can serve as a useful prognosis tool for patient counselling and deciding on follow-up strategies. PO-0774 Role of Magnetic Resonance Imaging in radiotherapy target volume definition in Oesophageal Cancer K. Owczarczyk 1 , C. Kelly-Morland 1 , S. McElroy 2 , R. Neji 3 , C. Thomas 4 , M. Siddique 1 , A. Qureshi 5 , G. Cook 1 , V. Goh 1 1 King's College London, Department of Cancer Imaging- Division of Imaging Sciences and Biomedical Engineering, London, United Kingdom 2 Guy's and St Thomas' NHS Foundation Trust, Department of Medical Physics, London, United Kingdom 3 Siemens Healthcare UK, MRI Collaboration Scientist, London, United Kingdom 4 Guy's and St Thomas' NHS Foundation Trust, Department of Radiotherapy Physics, London, United Kingdom 5 Guy's and St Thomas' NHS Foundation Trust, Department of Clinical Oncology, London, United Kingdom Purpose or Objective Multi-modality imaging approach may be more accurate in radiotherapy treatment planning (RTP) in oesophageal cancer. 18F-flourodeoxyglucose positron emission tomography (18F-FDG PET/CT) is established in staging oesophageal cancer. Whilst it has been used increasingly in RTP, 18F-FDG PET/CT alone may underestimate the maximum tumour length (MTL) by up to 38%. The aim of this prospective study was to determine the feasibility of magnetic resonance imaging (MRI) in conjunction with 18F-FDG PET/CT to better estimate gross tumour volume (GTV) and MTL and to determine the interobserver variability (IOV) of the two approaches. Material and Methods Following ethics approval and consent, 34 patients were included in this study. Each subject underwent a staging 18F-FDG PET/CT as well as research 1.5T MRI with anatomical T2w sequences (isotropic SPACE sequence with ECG-triggering and respiratory navigator gating) as well as a respiratory navigated diffusion weighted imaging (DWI) sequence with 3 b values (50, 400, 900s/mm 2 ). The images were fused manually using rigid registration; conformity was assessed using Jaccard Similarity Index (JSI) and Dice Coefficient (DC).

Two sets of MR volumes (GTV) were delineated – on T2-w and on b-900 DWI images by a radiation oncologist and radiologist independently. The MTLs were calculated as the maximum sup-inf extent of the tumour (slice thickness 1.2mm, T2w sequence; 5mm, DWI). MR was compared with PET/CT GTV (automated thresholding at 40% SUVmax) and MTL (manual by nuclear medicine physician) using paired t test, Wilcoxon signed rank test and Spearman's rho. IOV was assessed using Bland-Altman statistics. Results 18 tumours were located in lower oesophagus (LO) and 16 tumours within the gastro-oesophageal junction. Following image fusion, mean JSI was 0.34 (SD 0.16), mean DC was 0.5 (SD 0.15). Using radiologist’s outlining as the point of reference, T2w GTV was larger than PET/CT GTV by a mean of 44.2% (95% CI 21.9 - 63.3; p=0.004, Spearman’s rho 0.75). This was more pronounced in GOJ tumours than LO tumours (48.5% (95%CI 10 - 85, p= 0.02) versus 24% (95%CI -16 - 64, p = 0.18) respectively). DWI GTV was larger than PET/CT GTV by a mean of 6.5% (95% CI -30.9 - 43.9%, p=0.73, Spearman’s rho 0.6). There was no difference between T2w and PET/CT MTL (mean difference -1.7% (95% CI -11.5 - 15.1; p= 0.78, Spearman 0.69) overall, however, when stratified by tumour location, T2w MTL was smaller by 12.7% (95% CI - 27 - 1, p=0.07, Spearman 0.81) in LO tumours but larger by a mean 7.4% (95%CI -16 - 28, p=0.4) in GOJ tumours. Bland-Altman limits of agreement between the 2 observers were -64 to 189%, (bias 62±64) for DWI and - 152.8 to 202%, (bias 23±90) for T2w. Conclusion 18F-FDG PET/CT and MRI complement each other in pre- treatment assessment of primary oesophageal cancer. For RTP, high resolution T2w MRI sequences may be preferable for delineating tumour bulk, particularly in GOJ tumours, however ought to be interpreted with input from a subspecialty radiologist. PO-0775 MRI for GTV delineation in pancreatic cancer: preliminary results of a multi-institutional study. L. Caravatta 1 , F. Cellini 2 , N. Simoni 3 , C. Rosa 1 , R. Niespolo 4 , M. Lupattelli 5 , V. Picardi 6 , G. Macchia 6 , A. Sainato 7 , G. Mantello 8 , F. Dionisi 9 , M.E. Rosetto 10 , C. Guida 11 , R. Basilico 12 , A. De Paoli 13 , G. Boz 13 , V. Fusco 14 , G.C. Mattiucci 2 , A.G. Morganti 15 , D. Genovesi 1 1 “SS Annunziata” Hospital- “G. D’Annunzio” University, Department of Radiotherapy, Chieti, Italy 2 Università Cattolica del S. Cuore, Radiotherapy Department, Roma, Italy 3 Azianda Ospedaliera Universitaria, Radiotherapy Unit, Verona, Italy 4 Azienda Ospedaliera S. Gerardo, Department of Radiation Oncology, Monza, Italy 5 University of Perugia and Perugia General Hospital, Radiation Oncology Section, Perugia, Italy 6 Fondazione di Ricerca e Cura “Giovanni Paolo II”- Università Cattolica del S. Cuore, Radiation Oncology Department, Campobasso, Italy 7 Azienda Ospedaliera Universitaria Pisana, Radiotherapy Unit, Pisa, Italy 8 State Hospital, Radiotherapy, Ancona, Italy 9 Azienda Provinciale per i Servizi Sanitari- APSS, Therapy Unit- Department of Oncology Proton, Trento, Italy 10 Ospedale Belcolle, Department of Radiation Oncology, Viterbo, Italy 11 Azienda Ospedaliera San Giuseppe Moscati, Radiation Therapy, Avellino, Italy 12 “SS Annunziata” Hospital- “G. D’Annunzio” University, Department of Radiology, Chieti, Italy 13 Centro di Riferimento Oncologico-National Cancer Institute, Department of Radiation Oncology, Aviano, Italy