Development and validation of a tumor size-stratified prognostic nomogram for patients with gastric signet ring cell carcinoma.

Gastric signet ring cell carcinoma (GSRC) is a rare malignancy without a commonly acknowledged prognostic assessment and treatment system. This study aimed to determine the optimal cut-off value of tumor size (TS), and construct a prognostic nomogram in combination with other independent prognostic factors (PFs) to predict 3 year and 5 year overall survival (OS) in GSRC patients. From the Surveillance, Epidemiology, and End Results (SEER) database, this study collected 4744 patients diagnosed with GSRC. These patients were randomized into a training cohort (n = 2320,) and a validation cohort (n = 1142). A restricted cubic spline (RCS) was used to determine the cut-off value for TS, and univariate and multivariate Cox regression analyses were performed in the training cohort to identified significant predictors. A prognostic nomogram was constructed to predict OS at 3 and 5 years. Concordance index (C index), receiver operating characteristics curve (ROC curve), area under curve (AUC), and calibration curve were used to test the predictive accuracy of the model. A non-linear relationship was observed between TS and the risk of OS in GSRC, with TS thresholds at 4.4 cm and 9.6 cm. Survival was significantly lower in GSRC patients with TS > 4.4 cm. Age, marriage, chemotherapy, surgery, TS, SEER stage, regional lymph node status, and total number were independent predictors of OS. The C index in the training cohort was 0.748, and the AUC values for both 3- and 5-year OS were higher than 0.80. Similar results were observed in the validation cohort. In addition, the calibration curves showed good agreement between the predicted 3 year and 5 year OS and the actual OS. TS is a key prognostic factor for patients with GSRC, and patients with a TS of 4.4-9.6 cm and > 9.6 cm may have a poorer prognosis than those with a TS of < 4.4 cm. The TS-stratified nomogram we constructed and validated has favorable accuracy and calibration precision, and may be helpful in predicting the survival rate of patients.
(© 2024. The Author(s).)