ESTRO 36 Abstract Book

S364 ESTRO 36 _______________________________________________________________________________________________

validation. The discriminative accuracy of the model is 0.67. The nomogram was able to significantly (p<0.001) distinguish low- from high-probability patients for overall survival, as can be seen in figure 2.

Figure1: Nomogram of the model PRRT of pancreatic NEN. The nomogram is based on a proportional hazards cox regression. A number of points can be looked up for each variable, the total points can be summed up and mapped to the survival scores on the bottom of the nomogram. The predictive value of each variable is proportional to the points score line length associated with the variable

Conclusion Neo-adjuvant CRT followed by resection for primary resectable non-metastatic EC in daily practice results in a 3 yr OS of 49.6% and PFS of 45.6% compared to 58% and 51% within the CROSS-trial. The lower rates can be attributed to less favourable patient- and tumour characteristics in our study population. Toxicity and QOL results will be presented. PO-0696 A predictive nomogram for decision support for patients with pancreatic neuroendocrine tumors A. Jochems 1 , R. Baum 2 , A. Singh 2 , K. Niepsch 2 , H. Kulkarni 2 , P. Lambin 1 1 Maastro Clinic, Radiotherapy, Maastricht, The Netherlands 2 Zentralklinik Bad Berka, Nuclear medicine, Bad Berka, Germany Purpose or Objective Pancreatic neuroendocrine neoplasm (NEN) patients may benefit from peptide receptor radionuclide therapy (PRRT). To facilitate treatment decision support for individual patients, accurate statistical models to predict overall survival (OS) are required. Material and Methods 414 patients with pancreatic NEN treated at censored, country censored, were included in this study. Variables included in the analysis were age, Karnofsky performance status (KPS), grade of tumor, weight loss (at least 5kg) before PRRT (over last 6 months), presence of bone metastases, presence of liver metastases, hepatomegaly and presence of lung metastases. Few missing data were imputed using the predictive mean matching method. Multivariate analyses were based on Cox proportional hazards regression model. Model results were expressed by the c-index (0.5-1.0; random to perfect prediction). High- and low risk strata were identified by taking median prediction percentage of the model over all patients. Results The nomogram is shown in figure 1. Accuracy of the developed model was tested with internal cross

Figure 2: Kaplan Meier curve of model identifying high-risk (solid line) and low risk (dashed line) group. Patients in the high-risk group were identified by the model as above median risk score. Patients in the low-risk group were identified by the model as below median risk score. Conclusion The nomogram is internally validated and able to accurately predict overall survival for pancreatic neuroendocrine neoplasm patients. The model could facilitate decision support in daily clinical practice and can be used for patient counseling and shared decision making, as well as for and generating new hypotheses. PO-0697 Reduced inter- and intra-observer variation in esophageal tumor delineation using fiducial markers. M. Machiels 1 , P. Jin 1 , P. Jelvehgaran 1 , O.J. Gurney- Champion 1,2 , E.D. Geijsen 1 , P.M. Jeene 1 , M.W. Kolff 1 , V. Oppedijk 3 , M.B. Van Herk 4 , T. Alderliesten 1 , M.C.C.M. Hulshof 1 1 Academic Medical Center, Radiation Oncology, Amsterdam, The Netherlands 2 Academic Medical Center, Radiology, Amsterdam, The Netherlands