Background: Prognoses vary substantially among patients with advanced gastric cancer following curative surgery. The aim of the current study was to develop and verify the validity of a novel nomogram that predicts the probability of 5-year relapse-free survival (RFS) in patients who underwent curative resection for stage II/III gastric cancer.
Materials and methods: A nomogram to predict 5-year RFS following surgical resection of gastric cancer was constructed based on the data of patients who underwent surgery for primary gastric carcinoma at three institutions in Japan in January 2001-December 2006. Multivariate analysis using a Cox proportional hazards regression model was performed, and the nomogram's predictive accuracy (discrimination) and the agreement between observed outcomes and predictions (calibration) were evaluated by internal validation.
Results: Multivariate analyses revealed that age at operation, depth of tumor, tumor location, lymph node classification, and presence of combined resection were significant prognostic factors for RFS. In the internal validation, discrimination of the developed nomogram for 5-year RFS was superior to that of the American Joint Committee on Cancer TNM classification (concordance indices of 0.80 versus 0.67; P < 0.001). Moreover, calibration appeared to be accurate. Based on these results, we have created free software to more easily predict 5-year RFS.
Conclusion: We developed and validated a nomogram to predict 5-year RFS after curative surgery for stage II/III gastric cancer. This tool will be useful for the assessing a patient's individual recurrence risk when considering additional therapy in clinical practice.
Keywords: Advanced gastric cancer; Cox proportional hazards regression model; Nomogram; Prediction; Recurrence risk.
Copyright © 2016 IJS Publishing Group Ltd. Published by Elsevier Ltd. All rights reserved.