36 Predicting Long-Term Survival and Time-to-Recurrence After Oesophagectomy in Patients with Oesophageal Cancer

Abstract Aim Long-term survival after oesophagectomy remains poor, with recurrence a feared common outcome. Prediction tools can identify high-risk patients and optimise treatment decisions based on their prognostic factors. This study developed prediction model to predict long-term time-to-recurrence following surgery for oesophageal cancer. Method Patients undergoing curative from the European iNvestigation of SUrveillance After Resection Esophageal Cancer were included. models overall (OS) disease-free (DFS) using Cox proportional hazards (CPH) Random Survival Forest (RSF). Model performance was evaluated discrimination (time-dependent area under curve (tAUC)) calibration (visual comparison predicted observed probabilities). Results included 4719 an OS 47.7% DFS 48.4% at 5 years. Sixteen variables CPH RSF demonstrated good tAUC 78.2% (95% CI 77.4–79.1%) 77.1% 76.1–78.1%) 79.4% 78.5–80.2%) 78.6% 77.5–79.5%) respectively showed agreement between probabilities in all quintiles. 20–80%. Conclusions that statistical accurately oesophagectomy. Identification patient groups risk poor improve outcomes by optimising methods surveillance strategies. Future work evaluating prediction-based against standard decision-making is required understand clinical utility derived use.