Prognostic and Predictive Model of the Log Odds of the Negative Lymph Node/T Stage Ratio in Resectable Gastric Adenocarcinoma Patients.

ABSTRACT

PURPOSE: There are few reports on disease-specific survival (DSS) prediction systems for resected gastric cancer (GC) patients. The aim of this study was to create a nomogram based on the log odds of the negative lymph node/T stage ratio (LONT) for individual risk prediction. METHODS: We applied the Surveillance, Epidemiology, and End Results (SEER) Program database released in 2021 to screen GC patients from 2010 to 2015. Using a competitive risk model, we plotted the cumulative risk curve of variables for gastric cancer-specific death and death from other causes at each time point. According to the minimum BIC, we constructed and assessed a nomogram for the 12-month, 36-month, and 60-month cumulative mortality probabilities assessed by time-dependent ROC curves (time-AUCs), the C-index, Brier scores, decision curve analysis (DCA), and calibration curves. RESULTS: A total of 3895 patients were ultimately included and randomly assigned to two sets the training set (n = 2726, 70%) and the validation set (n = 1169, 30%). The LONT was a remarkable independent predictor of gastric cancer-specific death (high versus low 0.705, 95% CI 0.524-0.95, p = 0.021). The variables selected based on the minimum BIC were as follows location, AJCC, AJCC.T, AJCC.N, radiotherapy, LONT.cat, and chemotherapy. According to the time-AUC, C-index, Brier score, DCA, and calibration curves, the nomogram risk score had excellent survival prediction ability for DSS. CONCLUSIONS: A low LONT was associated with a high cumulative incidence of DSS. A prognostic nomogram model based on the LONT could effectively predict DSS for resectable GC patients.