Development and validation of a dynamic nomogram for individualized prediction of survival in patients with colon cancer

Current tools for predicting survival outcomes in colon cancer patients depend predominantly on clinical and pathologic characteristics, but accumulating evidence demonstrates that nutritional status is closely related to patient outcomes. We aimed to establish a new dynamic nomogram for individualized prediction of postoperative overall survival (OS). The clinicopathological and nutritional data of colon cancer patients undergoing radical resection were retrospectively collected and randomly divided into the primary and validation cohorts. Risk factors on OS rates were investigated by Cox analyses and the nomogram was constructed using significant predictors. Among 1,024 patients, 341 deaths were observed after median follow-up of 54 months. Five independent prognostic factors including nutritional status assessments, were incorporated into the nomogram. The C-index regarding 1-, 3-, and 5-year OS were 0.830, 0.859, and 0.757 in the primary cohort and 0.843, 0.870, and 0.773 in the validation cohort, respectively. Calibration curves for the probability of OS exhibited an optimal agreement. The decision curve analyses demonstrated the greater application value of the nomogram than the TNM staging system. Based on the nomogram, these patients could be divided into three scenarios with significant prognostic classification ( P < 0.0001). We developed and validated an easy-to-use dynamic nomogram to predict postoperative OS for patients with colon cancer.