Nomogram for predicting overall survival in metastatic gastric cancer: a population-based study from SEER

Abstract

Background Gastric cancer is a highly malignant tumor, and many patients already have distant metastasis at the time of diagnosis, so the mortality rate is very high. It is necessary to establish an effective prognostic prediction model to guide treatment. Methods The cases diagnosed as metastatic gastric cancer from 2010 to 2018 in Surveillance Epidemiology and End Results (SEER) database were selected, and the clinicopathological characteristics were included in the cox regression model to determine which are the independent factors affecting prognosis. Then all cases were divided into training cohort and verification cohort. A nomogram for predicting overall survival (OS) was established using the training cohort, and the effectiveness was verified with the verification cohort. Result A total of 4208 stage IV gastric cancer patients were enrolled in this study. firstly, they were randomly divided into training cohort and verification cohort according to the proportion of 2:1. The median overall survival was 8 (95% CI 7.6–8.4) months in the training cohort. Age at diagnosis, gender, histology grade, T stage, N stage, surgery for primary site, performance of chemotherapy or radiotherapy, existence of bone, liver or lung metastasis were independent prognostic factors. Then a nomogram was established using the training cohort (2948), and the concordance index (C index) is calculated as 0.696 (95%CI 0.684–0.708). It was also found that the model has good prediction ability by using the calibration chart. ROC curve showed that the model can accurately predict the OS, AUC of 1 year, 3 years and 5 years survival are 0.7, 0.756 and 0.76365 respectively. Finally, the calibration chart and ROC curve were used in the verification cohort to prove these results. Conclusion In this study, based on the data of patients with metastatic gastric cancer in SEER database, a nomogram which can predict overall survival was established, from which several independent prognostic factors were verified. C index, ROC curve and calibration chart were used to verify the effectiveness of the nomogram, which can provide a reference for predicting prognosis and guiding individualized treatment for advanced gastric cancer patients.