ESTRO 38 Abstract book

S774 ESTRO 38

pathological multivariate model using Cox proportional hazard ratio. Results Out of 41 patients recruited, 33 underwent surgery (27/33 male, median age 65±11, 30 adeno/ 3 SCC, 30 NACT/ 3CRT). 12/33 were good responders, 14/33 showed no response. In univariate analysis, T2 T2w Kurt and T2 T2w Ent were predictive of good response (FDR p= 0.0001 and 0.0002, respectively), whilst T2 T2w MFL, T2 T2w Ent and T2 volume were predictive of no response (FDR p 0.03, 0.04, 0.04, respectively). None of the DWI derived features met the predefined FDR cut-off. T2 T2w Kurt was significantly lower whilst T2 T2w Ent was significantly higher in good responders (p=0.0002, 0.0009, respectively). Spearman ρ between linear Mandard TRG and T2 T2w Kurt and Ent was 0.55 (p=0.002) and -0.62 (p=0.0005), respectively. Dichotomised T2 T2w Kurt (>median vs ≤ median) remained significantly predictive of good response in multivariate model including age, gender, cT (cT2 vs cT3-4), cN (cN0 vs cN+) and histopathology (SCC vs adeno) with an OR 0.04 (0.001- 0.48, p=0.008)

by Kaplan-Meier method and compared using longrank test. Results Ninety-two LAPC patients were enrolled (CG/SG: 46/46 pts; M/F: 52/40; median age: 65.5 years, range: 36.0- 89.0, median follow-up: 17.0 months, range: 3.0-70.0). At univariate analysis no significant differences between the two groups were recorded. Results in terms of OS, LC, DMFS are shown in the table.

Conclusion This case-control study shows that the two treatment approaches achieved comparable outcomes in patients with LAPC. EP-1425 MRI heterogeneity analysis for predicting response to neoadjuvant therapy in oesophageal cancer K. Owczarczyk 1 , C. Kelly-Morland 1 , C. Yip 2 , M. Siddique 1 , N. Maisey 3 , A. Qureshi 3 , J. Gossage 4 , 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 National Cancer Centre Singapore, Department of Oncology, Singapore, Singapore ; 3 Guy's and St.Thomas' Hospital NHS Foundation Trust, Department of Oncology, London, United Kingdom ; 4 Guy's and St.Thomas' Hospital NHS Foundation Trust, Department of Upper GI Surgery, London, United Kingdom Purpose or Objective Response to neoadjuvant treatment has been shown to be a major predeterminant of survival in oesophageal cancer. The ability to predict response prior to oesophagectomy would allow for adaptive treatment approaches with potentially improved outcomes. We hypothesized that heterogeneity analysis of MRI may be predictive of response to neoadjuvant treatment. Material and Methods Prospective IRB approved study of patients with oesophageal cancer undergoing neoadjuvant chemotherapy or chemoradiotherapy (CRT) followed by oesophagectomy. Three MRI scans including a 3D T2w ECG and respiratory-triggered sequence as well as Diffusion Weighted Imaging (DWI) sequence (b=0,100, 900) were carried out at baseline (T0), following first cycle of chemotherapy (or 10# of CRT, T1) and prior to surgery (T2). VOIs were outlined in 3D on T2w sequences & ADC parametric maps, generating 6 VOIs per case (T2w and DWI at each timepoint). Heterogeneity analysis was carried out using in-house software. Highly correlated features (ρ>0.8) were excluded. Seven imaging features were analysed per case – Mean Intensity, Intensity SD, Intensity CoV, Entropy (Ent), Kurtosis (Kurt), Mean Fractal Lacunarity (MFL) and VOI volume (cm 3 ). Primary outcome was Mandard tumour regression grade in the resected specimen (good response TRG1-2 (<10% residual tumour cells) vs no response TRG4-5 (>50%)). Predictive value of imaging features was assessed in univariate analysis using false discovery rate (FDR) corrected cut-off of p<0.05 (Benjamini Hochberg correction for multiple testing). Huber method was used to minimise risk of model overfitting. Wilcoxon non- parametric test was used to compare values between subgroups. Selected features were assessed in a clinico-

Conclusion This prospective data shows that heterogeneity analysis has a role in predicting pathological response to treatment in oesophageal cancer and warrants investigation in a larger patient cohort. EP-1426 Postoperative Radiotherapy In Pancreatic Cancer Patients. A Single Institution Experience M. Caroprese 1 , M. Tirozzi 1 , G. Porri 1 , I.R. Scognamiglio 1 , E. Toska 1 , L. Faraci 1 , A. Perillo 1 , F. Pati 1 , A. Farella 1 , R. Solla 2 , S. Clemente 1 , M. Conson 1 , R. Pacelli 1 1 University “Federico II” School of Medicine, Department of Advanced Biomedical Sciences, Napoli, Italy ; 2 National Research Council, Institute of Biostructures and Bioimages, Napoli, Italy Purpose or Objective Pancreatic cancer is the fourth leading cause of cancer- related deaths worldwide. At present, the only chance for cure and prolonged survival is surgical resection with macroscopic tumor clearance. Even following potential curative resection more than 80% of the patients ultimately die of the disease due to local recurrence and/or distant metastasis. The high rate of local recurrence is predetermined by the microscopic frequently incomplete resections as a result of the anatomical location of the tumor and the growth pattern of pancreatic cancer cells. According to the Royal College of Pathologists (RCP), positive surgical margin status is defined as the microscopic presence of tumor cells within 1 mm of the resection margin. The aim of this study was to determine the impact of margin status on disease free survival (DFS) and overall survival (OS) in patients undergoing adjuvant radiotherapy. Material and Methods Thirty-two consecutive patients treated by radiotherapy for M0 pancreatic adenocarcinoma at the Radiotherapy Department of the University “Federico II” (Naples, Italia)