Optimizing Clinical Outcomes: A Nomogram for Predicting Gastric Cancer Survival in Elderly Patients Following Gastrectomy

Abstract

Background: Owing to the aging global population, the prevalence of gastric cancer (GC) in the elderly is on the increase. The study aimed to investigate risk variables correlated with cancer-specific survival (CSS) and to construct and validate a nomogram to calculate the probability of CSS after gastrectomy in elderly patients with non-metastatic GC. Methods: Overall, 5462 postgastrectomy patients with geriatric non-metastatic GC who were diagnosed between 2006 and 2015, and 1078 patients diagnosed between 2004 and 2005 from the Surveillance, Epidemiology, and End Results (SEER) database were incorporated in this study. We randomly selected 3846 patients from the 5462 patients to constitute the training cohort to construct the nomogram, and the other 1616 patients were used for internal validation. Additionally, the other 1078 patients were used for external validation. CSS predictors were screened using univariate Cox proportional hazards regression analysis, Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis, Random Survival Forest (RSF), and multivariate Cox proportional hazards regression analysis. A model for predicting CSS was developed using Cox proportional hazards regression analysis, and both static and dynamic web-based nomograms were provided. Results: A total of 10 variables, including gender, age, race, tumor size, primary site, grade, positive lymph nodes (PLNs), examined lymph nodes (ELNs), T stage, and months from diagnosis to treatment (MFDTT), were incorporated in the development of the nomogram. The areas under the curve (AUCs) for the training cohort at 3, 5, and 10 years were 0.780 (95% CI, 0.765–0.795), 0.792 (95% CI, 0.778– 0.807), and 0.803 (95% CI, 0.781–0.826). Following internal validation, these values were 0.809 (95% CI, 0.789–0.830), 0.822 (95% CI, 0.801–0.843), and 0.824 (95% CI, 0.793–0.854). The model-predicted CSS converges with the actual CSS according to the calibration curves. The nomogram showed higher clinical benefit than the TNM staging system, according to the decision curve analysis (DCA) for all three cohorts. The nomogram score was utilized as a basis for grouping high- and low-risk patients, and significant statistical differences (p<0.0001) were demonstrated between the two groups in the Kaplan–Meier (K-M) curves. Conclusions: Our novel nomogram can conveniently and accurately predict CSS in elderly patients with non-metastatic GC after gastrectomy, which can assist clinicians in adopting individualized treatment strategies to improve the prognosis of elderly patients with GC.