ESTRO 38 Abstract book

S377 ESTRO 38

Hatt 3 , H. Kaanders 4 , O. Hamming-Vrieze 6 , H. Van Laarhoven 7 , R. Subramiam 8 , S. Huang 9 , B. O'Sullivan 9 , S. Bratman 9 , L. Dubois 10 , R. Miclea 11 , D. Di Perri 12 , X. Geets 12 , D. De Ruysscher 13 , F. Hoebers 14 , P. Lambin 2 1 Maastricht University, The-D Lab: Decision Support for Precision Medicine-, Maastricht, The Netherlands ; 2 Maastricht University, The-D Lab: Decision Support for Precision Medicine, Maastricht, The Netherlands ; 3 Université Brest, 3.Laboratory of Medical information processing, Brest, France ; 4 Radboud University Nijmegen Medical Centre, Department of Radiation Oncology, Nijmegen, The Netherlands ; 5 Academic Medical Centre, 7.Department of Medical Oncology, Amsterdam, The Netherlands ; 6 Antoni van Leeuwenhoek Medical Centre, Department of Radiation Oncology, Amsterdam, The Netherlands ; 7 Academic Medical Centre, Department of Medical Oncology, Amsterdam, The Netherlands ; 8 Johns Hopkins Medical Centre, Division of Nuclear Medicine-, Baltimore, The Netherlands ; 9 Princess Margaret Cancer Center, Department of Radiation Oncology, Toronto, Canada ; 10 Maastricht University, Department of Radiotherapy, Maastricht, The Netherlands ; 11 Maastricht University Medical Centre+, Department of Radiology, Maastricht, The Netherlands ; 12 Université catholique de Louvain, Center of Molecular Imaging, Brussels, Belgium ; 13 Maastro Clinic, 2.Department of Radiation Oncology, Maastricht, The Netherlands ; 14 Maastro Clinic, Department of Radiation Oncology, Maastricht, The Netherlands Purpose or Objective Tumor hypoxia increases resistance to radiotherapy and systemic therapy as well as promoting tumor metastasis. Our primary aim was to develop and validate a CT- and FDG-PET-based hypoxia classification signature based on radiomic features. Material and Methods A total of 131 patients (78 head and neck squamous cell carcinoma, 61 non-small cell lung cancer, 13 pancreatic carcinoma and 13 esophageal carcinoma) with low dose [ 18 F]-HX4-PET/CT and 61 patients with [ 18 F]-FDG-PET from 5 imaging centers were included. One dataset with [ 18 F]- FAZA PET/CT (36 NSCLC) and another dataset (40 HNSCC) with pimonidazole stained tumor biopsies were used as additional external validation datasets. The primary gross tumor volumes (GTV) were manually delineated on CT. In order to create three dichotomized groups between hypoxic and well-oxygenated tumors, [ 18 F]-HX4-derived hypoxic fractions (HF) were thresholded at 10%, 20% and 30%. The HX4 -derived HF was defined as the ratio of the hypoxic region (tumor-to-background ratio > 1.4) to the total GTV. A random forest (RF)-based classifier was trained to stratify patients into hypoxia-positive/ negative cohorts based on radiomic features. Results In the 20% HF threshold group, the area under the receiver operating characteristic curve (AUC) for an RF model combining 5 CT-derived radiomic features achieved 0.79±0.16 in the Boston/ NKI and 0.76±0.18 in the UCL validation sets respectively. In the same threshold group, an RF model combining 10 radiomic features derived from FDG-PET reached an AUC of 0.74±0.23 to classify hypoxia in an external validation dataset (Boston/ Maastro). Conclusion Our hypoxia signatures derived from CT and FDG-PET have the potential to inform interventional trials studying hypoxia-targeting agents by identifying patients with tumors likely to be hypoxic.

D. Yeboa 1 , K. Liao 2 , B.A. Guadagnolo 1 , G. Rao 3 , A. Bishop 4 , C. Chung 4 , J. Li 4 , C.E. Tatsui 3 , L. Rhines 3 , S. Ferguson 3 , A. Paulino 4 , A. Ghia 4 1 MD Anderson Cancer Center, Radiation Oncology- Joint appointment Health Services Department, Houston, USA ; 2 MD Anderson Cancer Center, Health Services Department, Houston, USA ; 3 MD Anderson Cancer Center, Neurosurgery, Houston, USA ; 4 MD Anderson Cancer Center, Radiation Oncology, Houston, USA Purpose or Objective Spinal ependymomas are rare with an incidence of 1 per 100,000. The management and patterns of care for spinal cord ependymomas require further investigation given the paucity of data for high grade disease. Material and Methods A National Cancer Database (NCDB) was searched for eligible patients >18 years old with a diagnosis of spinal ependymoma at a Commission on Cancer facility from 2004 through 2014. Chi-square test, simple logistic regression, and multivariable logistic regressions were used to assess the potential significance of associations between patient characteristics and treatment received. Facility volume characteristics were assessed in relation to survival by evaluating the number of patients diagnosed and the facility volume was organized by relative quartiles. We assessed the average number of patients seen at each facility within a 1-year timeframe and over an equivalent 10-year timeframe. Results Our study of 1,345 patients with high-grade spinal ependymoma and almost one decade of follow-up data uniquely describes practice patterns and long-term overall survival (OS) independent of treatment received for grade II and grade III ependymoma. The majority of grade II patients received surgery alone (81.1%). In contrast, 36.8% of grade III patients received surgery alone. Only 56.3% of grade III spinal ependymomas received adjuvant therapy with either radiation therapy (RT) or combination RT and chemotherapy after surgery. Over 60% percent of patients with grade III disease received initial adjuvant RT, in contrast to only 15.3% of patients with grade II ependymoma (P< 0.001). Patients living ≤20 miles from a RT facility were more likely to receive RT (P<0.001) than those living further away. On multivariable logistic regression, grade (grade III: OR 8.6, P<0.001) and facility distance were significantly associated with receipt of RT (P<0.0001). Among all patients, the 5 and 10-year OS rates were 94.7%/ 85.1% for patients with grade II disease and 58.2%/46.4% for those with grade III disease (P<0.0001). The OS rates were higher at facilities treating at least 1.5 patients per year (i.e., an average of 15 patients over 10 years), which corresponds to the top 81st percentile in facility volume. The 5-year OS rates were 96.7% for patients treated at high-volume facilities, followed by 93.4% at facilities treating 6-14 patients, 92.5% for 2-5 patients, and 86.8% for <2 patients over 10 years In conclusion, our study is the largest series to date reporting observational patterns of care for grade II and III spinal ependymomas. Despite guideline recommendations for adjuvant therapy in all grade III ependymomas regardless of extent of resection, approximately 2 in 5 do not receive immediate adjuvant therapy and our data suggests this may be related to distance lived from a CoC facility. Facilities at which larger numbers of patients with spinal ependymomas are treated may experience better long-term outcomes. PO-0735 Cognitive function after radiation therapy for brain tumours L. Haldbo-Classen 1 , A. Amidi 2 , L. Wu 3 , S. Lukacova 1 , G. Von Oettingen 4 , R. Zachariae 2 , J. Kallehauge 5 , M. Høyer 5 1 Aarhus University Hospital, Department of Oncology, ( P <0.0019). Conclusion

Poster: Clinical track: CNS

PO-0734 Patterns of Care in the Management of WHO Grade II and III Spinal Ependymomas