Background: Accurate prediction of prognosis in gastro-oesophageal cancer remains challenging. The aim of this study was to develop a robust model for outcome prediction.
Methods: The study included 220 patients with gastric or oesophageal cancer newly diagnosed over a 2-year period. Patients were staged and underwent treatment following discussion at a multidisciplinary team (MDT) meeting. Clinical and investigative variables were collected, including performance and nutritional status, and serum C-reactive protein (CRP) level. Primary endpoints were death within 12 and 24 months.
Results: Overall median survival was 13 months. Advanced clinical stage (P < 0.001), reduced performance score (P < 0.001), weight loss exceeding 2.75 per cent per month (P = 0.026) and serum CRP concentration above 5 mg/l (P = 0.031) were identified as independent prognostic indicators in multivariable analysis. A prognostic score was constructed using these four variables to estimate a probability of death. Applying the model gave an area under the receiver-operator characteristic curve of 0.84 and 0.85 for prediction of death at 12 and 24 months respectively (both P < 0.001).
Conclusion: This model accurately estimated the probability of death within 12 and 24 months. This may aid the MDT decision-making process.
Copyright (c) 2007 British Journal of Surgery Society Ltd.