Development and Validation of a Prognostic Nomogram for Predicting the Overall Survival of Patients with Metastatic Gastric Cancer who Undergo Surgery

Background: The benefits of surgery in patients with metastatic gastric cancer remain unclear. A nomogram may predict overall survival (OS) time of patients with metastatic gastric cancer who undergo surgery. Methods The development set consisted of 1300 patients from the Surveillance, Epidemiology, and End Results Program who were diagnosed with metastatic gastric cancer and underwent surgery in 2004–2015, whereas the validation set consisted of 555 patients diagnosed in 2004–2015. Variables were incorporated into a Cox proportional hazards model to identify independent risk factors for survival. Nomograms predicting 1-, 2-, and 3-year OS probabilities were designed using the development dataset. Receiver operating characteristic (ROC) curves and calibration curves were plotted to determine the accuracy of the nomograms. The net benefit of prediction models was evaluated using decision curve analysis (DCA). Results Age, primary tumor site, T stage, and N stage were risk factors for OS, whereas chemotherapy and radiotherapy were protective factors of OS. A nomogram including all of these variables had areas under the ROC curves (AUC) predicting 1-, 2-, and 3-year OS of 0.684, 0.676, and 0.685, respectively, for the development dataset and 0.689, 0.642, and 0.665, respectively, for the validation dataset. A calibration curve showed that the OS probability predicted by the nomogram was in good agreement with the actual OS probability. The DCA curves suggested that the nomogram provides a superior net benefit. Conclusions This nomogram can help predict the OS of patients with metastatic gastric cancer who underwent surgery. This nomogram can help clinicians make correct clinical decisions to prolong the OS of these patients.