ESTRO 37 Abstract book

S166

ESTRO 37

parameters) was performed. The best model was reported and covariates of the top 15 performing models were analyzed. NTCP models were validated in dataset 2 after a refit of model coefficients. Results Median follow-up time was 30.8 and 43.3 months in dataset 1 and 2, respectively. Univariate Cox modeling selected MHD (p<0.001), GTV (p<0.001), WHO PS (p=0.006), MLD (p=0.007), age (p=0.03) and current smoker (p=0.07). The best multivariable Cox model included MHD (HR=1.026, p<0.001), GTV (HR=1.002, p<0.001), current smok er (HR=1.39, p=0.03) and WHO PS (HR=1.24, p=0.03). Adding a MHD*GTV interaction did not change these results. Survival curves showed an increased mortality associated with higher MHD starting 6 months post RT (Figure 1). NTCP models for all 4 mortality time endpoints included the GTV covariate (in 100% of best performing models). Additional covariates were age and current smoker at 6 months (AUC=0.73); MHD at 12 months (AUC=0.71); MHD, WHO PS and current smoker at 18 months (AUC=0.71); WHO PS, age and current smoker at 24 months (AUC=0.72). MHD was selected in 40%, 100%, 87% and 47% of best performing models (this was only 27%, 47%, 33% and 33% for MLD) at 6, 12, 18 and 24 months, respectively. The 12 month mortality NTCP model had the highest MHD OR=1.042 (p=0.006) and was selected. The probability of 12 month mortality can be calculated with the formula NTCP=(1+e -S ) -1 with S=- 1.528+0.0408*MHD+0.00570*GTV (Figure 2). In dataset 2, the 12 and 18 month NTCP models had respective AUCs of 0.60 (0.65 when adding WHO PS) and 0.67. MHD OR was 1.050 (p=0.11) at 12 month. Conclusion MHD is a risk factor independent from GTV volume for post RT mortality endpoints later than 6 months and before 18 months. A NTCP model for 12 month mortality could allow patient selection for proton therapy.

PV-0318 External Validation of Radiation-Induced Dyspnea Models on Esophageal Cancer Radiotherapy Patients Z. Shi 1 , L. Wee 1 , K. Foley 2 , E. Spezi 3 , P. Whybra 3 , T. Crosby 4 , J. Pablo de Mey 5 , J. Van Soest 1 , A. Dekker 1 1 Maastricht University, Department of Radiation Oncology MAASTRO Clinic- GROW – School for Oncology and Development Biology, Maastricht, The Netherlands 2 Cardiff University, Division of Cancer & Genetics, Cardiff, United Kingdom 3 Cardiff University, School of Engineering, Cardiff, United Kingdom 4 Velindre Cancer Centre, Department of Clinical Oncology, Cardiff, United Kingdom 5 Maastricht University, Faculty of Health Medicine and Life Sciences, Maastricht, The Netherlands Purpose or Objective Radiation-induced lung disease (RILD), such as dyspnea, is a risk for patients receiving high-dose thoracic irradiation. This study is a TRIPOD (Transparent Reporting of A Multivariable Prediction Model for Individual Prognosis or Diagnosis) Type 4 validation of previously- published lung toxicity models via secondary analysis of esophageal cancer SCOPE1 trial data. We quantify the predictive performance of these two models for predicting dyspnea 6 months after high-dose chemo- radiotherapy for primary esophageal cancer. Material and Methods Lung cancer patients treated at MAASTRO Clinic (The Netherlands) from the period 2002 to 2011 we re used to develop the previous dyspnea risk models. We tested the performance of the earlier models using baseline, treatment and follow-up data on 258 esophageal cancer patients in the UK enrolled into the SCOPE1 multi-centre trial. As some variables were missing randomly and cannot be imputed, 212 patients in SCOPE1 were used for validation of model 1 and 255 patients were used for validation of model 2. The adverse event of interest was dyspnea ≥ Grade 2 (CTCAE v3) within 6 months of the end of radiotherapy. The model parameter Forced Expiratory Volume in 1s (FEV 1 ) was imputed using the WHO performance status. External validation was performed