Abstract

The aim of this study was to combine clinicopathologic variables associated with overall survival after gastric resection with D2 lymphadenectomy (D2 gastrectomy) for gastric cancer into a prediction nomogram.We retrospectively analyzed 7,954 patients who underwent D2 gastrectomy for gastric cancer at Seoul National University Hospital (SNUH) in Seoul, Korea. Two thirds of the patients were randomly assigned to the training set (n = 5,300), and one third were assigned to the validation set (n = 2,654). Multivariate analysis by Cox proportional hazards regression was performed using the training set, and the nomogram was constructed. Discrimination and calibration were performed using the SNUH validation set. Additional external validation was performed using the data set (n = 2,500) from Cancer Institute Ariake Hospital (CIAH) in Tokyo, Japan.The multivariate Cox model identified age at diagnosis, sex, location, depth of invasion, number of metastatic lymph nodes, and number of examined lymph nodes as covariates associated with survival. In the SNUH validation set, the nomogram exhibited superior discrimination power compared with the seventh American Joint Committee on Cancer TNM classification (Harrell's C-index, 0.78 v 0.69, respectively; P < .001). Calibration of the nomogram predicted survival corresponding closely with the actual survival. In the CIAH validation set, discrimination was good (C-index, 0.79), and the predicted survival was within a 10% margin of ideal nomogram.We developed a nomogram predicting 5- and 10-year overall survival after D2 gastrectomy for gastric cancer. Validation using the SNUH and CIAH data sets revealed good discrimination and calibration, suggesting good clinical utility. The nomogram improved individualized predictions of survival.