A Novel Prognostic Model and Practical Nomogram for Predicting the Outcomes of Colorectal Cancer: Based on Tumor Biomarkers and Log Odds of Positive Lymph Node Scheme.

BACKGROUND

Emerging evidence shows that serum tumor biomarkers (TBs) and log odds of positive lymph node scheme (LODDS) are closely associated with the prognosis of colorectal cancer (CRC) patients. The aim of our study is to validate the predictive value of TBs and LODDS clinically and to develop a robust prognostic model to predict the overall survival (OS) of patients with CRC.

METHODS

CRC patients who underwent radical resection and with no preoperative chemotherapy were enrolled in the study. The eligible population were randomized into training (70%) and test (30%) cohorts for the comprehensive evaluation of the prognostic model. Clinical implications of serum biomarkers and LODDS were identified by univariate and multivariate Cox proportion regression analysis. The predictive ability and discriminative performance were evaluated by Kaplan-Meier (K-M) curves and receiver operating characteristic (ROC) curves. Clinical applicability of the prognostic model was assessed by decision curve analysis (DCA), and the corresponding nomogram was constructed based on the above factors.

RESULTS

A total of 1,202 eligible CRC patients were incorporated into our study. Multivariable COX analysis demonstrated that CA199 (HR = 1.304), CA125 (HR = 1.429), CEA (HR = 1.307), and LODDS (HR = 1.488) were independent risk factors for OS (all P < 0.0001). K-M curves showed that the high-risk group possessed a shorter OS than the low-risk counterparts. The area under curves (AUCs) of the model for 1-, 3- and 5-year OS were 86.04, 78.70, and 76.66% respectively for the train cohort (80.35, 77.59, and 74.26% for test cohort). Logistic DCA and survival DCA confirmed that the prognostic model displayed more clinical benefits than the conventional AJCC 8 TNM stage and CEA model. The nomograms were built accordingly, and the calibration plot for the probability of survival at 3- or 5-years after surgery showed an optimal agreement between prediction and actual observation.

CONCLUSIONS

Preoperative serum TBs and LODDS have significant clinical implications for CRC patients. A novel prognostic model incorporating common TBs (CA199, CA125, and CEA) and LODDS displayed better predictive performance than both single factor and the TNM classification. A novel nomogram incorporating TBs and LODDS could individually predict OS in patients with CRC.