Development and validation of pathological nomograms for predicting the prognosis of colorectal cancer patients

Abstract

Purpose The prediction of colorectal cancer (CRC) prognosis greatly influences clinical decision-making. The traditional tumor node metastasis (TNM) staging system has limited prognostic accuracy in CRC patients. This study was designed to develop a more robust prognostic tool to aid in prognosis prediction for CRC patients. Methods Novel prognostic nomogram models were developed based on postoperative pathological findings from 2435 patients who underwent curative colorectal tumor resection. In the development cohort, least absolute shrinkage and selection operator (LASSO) regression was used to select parameters for inclusion in the overall survival (OS) and disease-free survival (DFS) nomograms. Receiver operating characteristic (ROC) analysis, calibration plots, and decision curve analysis (DCA) were utilized to compare performance between the models and the traditional AJCC staging. Results Calibration plots indicated that the nomograms developed had good prognostic prediction capability. ROC analysis revealed that the OS-related nomogram predicted 1-, 3-, and 5-year OS with AUCs of 0.786, 0.776, and 0.803, respectively, compared to 0.768, 0.750, and 0.782, respectively, for the TNM staging system. The DFS nomogram predicted 1-, 3-, and 5-year DFS with AUCs of 0.764, 0.777, and 0.789, respectively, in contrast to 0.762, 0.761, and 0.770 for TNM staging. DCA demonstrated that the developed nomograms provided greater net benefits than did the TNM staging system. Conclusion Our developed prognostic model demonstrated enhanced accuracy in predicting CRC prognosis compared to traditional staging methods. Utilizing this model in postoperative survival prediction for CRC patients could facilitate development of more suitable personalized treatment strategies.