Predicting Distant Metastasis in Young-Onset Colorectal Cancer After Surgery: A Retrospective Study

BACKGROUND: Although overall colorectal cancer (CRC) cases have been declining worldwide, there has been an increase in the incidence of the CRC among individuals younger than 50 years old, which is associated with distant metastasis (DM) and poor prognosis. METHODS: Young-onset CRC patients’ postoperative data were collected from the Surveillance, Epidemiology, and End Results (SEER) database between January 2010 and December 2015. Data from the SEER database were divided into early stage and advanced stage according to whether chemoradiotherapy was recommended in the guidelines. Independent risk factors for DM were explored by using univariate and multivariate logistic regression separately. A predictive model was established and presented as nomogram in the training set of advanced stage. The model was internally verified in testing set and externally validated in a cohort of 145 patients from Zhongnan Hospital of Wuhan University. The accuracy, reliability, and clinical application value were assessed using the receiver operating characteristic curve (ROC), the area under the curve (AUC), calibration curve, and decision curve analysis (DCA), respectively. Different risk subgroups of DM were classified according to the scores of the nomogram in the training set of advanced stage. RESULTS: A total of 5,584 patients were eligible and enrolled in our study in which 1,277 were in early stage and 4,307 in advanced stage. Preoperative CEA positive was found to be an independent predictor of DM in early stage. Multivariate logistic regression analysis showed that tumor size, degree of differentiation, T stage, N stage, preoperative CEA, and whether radiation or chemotherapy performed were independent risk factors for DM (all, p < 0.05) in advanced stage. Great accuracies were achieved in our nomogram with AUC of 0.801 in training set, 0.811 in testing set, and 0.791 in the validation cohort, respectively. Calibration curves and DCA in internal validation and external validation both showed good stability and clinical utility values. CONCLUSIONS: Preoperative CEA positive was a significant predictor of DM for young-onset CRC patients. A novel nomogram containing clinical and pathological features was established for predicting DM of advanced CRC in patients younger than 50 years old. This tool may serve as an early alert for clinicians to DM and make better clinical treatment regimens.