BACKGROUND: Although some predictive factors of long-term survival after a distal gastrectomy for gastric cancer have been reported, only few studies have predicted long-term outcomes based on preoperative parameters. We aimed to evaluate the reliability of perioperative risk calculator for predicting overall survival (OS) after distal gastrectomy in patients with gastric cancer. METHODS: Overall, 337 patients (225 males, 112 females) who had undergone a distal gastrectomy for gastric cancer at the Keio University Hospital, Tokyo, Japan, between January 2009 and December 2013 were enrolled in this study. We investigated the reliability of a risk calculator for the prediction of OS. RESULTS: In multivariate analysis, the risk models for operative mortality and 30-day mortality were identified as predictors of death. Time-dependent receiver operating characteristics (ROC) curve analysis indicated that the estimated area under the curve (AUC) value of the risk model for operative mortality was > 0.870 during the first postoperative 3 years. We set optimal cutoff values of the risk model operative mortality for OS using the Cutoff Finder online tool. The cutoff values of 4.117% were significant risk factors of death. Similar results were observed in the external validation set. CONCLUSIONS: We elucidated the associations among risk calculator values and OS rates of patients with gastric cancer. Time-dependent ROC curve analysis suggested that the AUC value of the risk model for operative mortality was high, indicating that this risk calculator would be useful for not only short-term outcomes, but also long-term outcomes.
BACKGROUND: Although some predictive factors of long-term survival after a distal gastrectomy for gastric cancer have been reported, only few studies have predicted long-term outcomes based on preoperative parameters. We aimed to evaluate the reliability of perioperative risk calculator for predicting overall survival (OS) after distal gastrectomy in patients with gastric cancer. METHODS: Overall, 337 patients (225 males, 112 females) who had undergone a distal gastrectomy for gastric cancer at the Keio University Hospital, Tokyo, Japan, between January 2009 and December 2013 were enrolled in this study. We investigated the reliability of a risk calculator for the prediction of OS. RESULTS: In multivariate analysis, the risk models for operative mortality and 30-day mortality were identified as predictors of death. Time-dependent receiver operating characteristics (ROC) curve analysis indicated that the estimated area under the curve (AUC) value of the risk model for operative mortality was > 0.870 during the first postoperative 3 years. We set optimal cutoff values of the risk model operative mortality for OS using the Cutoff Finder online tool. The cutoff values of 4.117% were significant risk factors of death. Similar results were observed in the external validation set. CONCLUSIONS: We elucidated the associations among risk calculator values and OS rates of patients with gastric cancer. Time-dependent ROC curve analysis suggested that the AUC value of the risk model for operative mortality was high, indicating that this risk calculator would be useful for not only short-term outcomes, but also long-term outcomes.