Background: Tumour node metastasis (TNM) staging predicts survival on the basis of the pathological extent of a tumour. The aim of this study was to develop a prognostic model with improved survival prediction after oesophagectomy.
Methods: Consecutive patients who had potentially curative oesophagectomy for adenocarcinoma of the oesophagus or gastro-oesophageal junction were included. Cox regression analyses were performed to examine the association between risk factors and time to death from oesophageal cancer. The concordance index, calculated after bootstrapping, was used to measure accuracy. A nomogram was designed for use in clinical practice.
Results: Oesophageal cancer-specific survival rates for the 364 included patients who underwent oesophagectomy between 1993 and 2003 were 75.8, 54.9 and 39.2 per cent at 1, 2 and 5 years respectively. A prognostic model using all prognostic variables outperformed TNM staging (concordance index 0.79 versus 0.68 respectively; P < 0.001). A reduced model derived after backward elimination, containing only T stage, lymph node ratio and extracapsular lymph node involvement, also outperformed TNM staging (concordance index 0.77; P < 0.001).
Conclusion: A prognostic model developed to predict disease-specific survival after oesophagectomy was superior to TNM staging. More reliable prognostic information might lead to different approaches to patient follow-up.
Copyright (c) 2007 British Journal of Surgery Society Ltd.