A nomogram for the prediction of survival for colorectal signet ring cell carcinoma after surgery: A population-based study.

Abstract

The aim was to construct and verify a nomogram-based assessment of cancer-specific survival (CSS) in patients with colorectal signet ring cell carcinoma after surgery. Patients were collected from Surveillance, Epidemiology, and End Results program between 2004 and 2015. Independent prognostic indicators were determined in the training cohort by Cox regression model. We identified 2217 eligible patients, who were further categorized into the training set (n = 1693) as well as the validation set (n = 524). Multivariate analysis revealed that age at diagnosis, gender, grade, tumor size, T stage, N stage, and M stage were independent predictive indicators. Then, the above 7 predictive factors were incorporated into a nomogram model to assess CSS, which showed good calibration and discrimination capacities in both sets. Both internal and external calibration plot diagrams revealed that the actual results were consistent with the predicted outcomes. The time-independent area under the curves for 3-year and 5-year CSS in the nomogram were larger than American Joint Committee on Cancer and Surveillance, Epidemiology, and End Results summary stage system. Moreover, decision curve analysis indicated the clinical utility of the nomogram. The nomogram demonstrated favorable predictive accuracy of survival in colorectal signet ring cell carcinoma patients after surgery, which should be further confirmed before clinical implementation.