Establishment and validation of a postoperative predictive model for patients with colorectal mucinous adenocarcinoma

Abstract

Background: The aim of this study was to develop comprehensive and effective nomograms for predicting overall survival (OS) and cancer-specific survival (CSS) rates in patients with colorectal mucinous adenocarcinoma (CRMA). Methods: A total of 4711 CRMA patients who underwent radical surgery between 2010 and 2018 from the Surveillance, Epidemiology, and End Results (SEER) database were collected and randomized into development (n=3299) and validation (n=1412) cohorts at a ratio of 7:3 for model development and validation. OS and CSS nomograms were developed using the prognostic factors from the development cohort after multivariable Cox regression analysis. The performance of the nomograms was evaluated using Harrell’s concordance index (C-index), calibration diagrams, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA). Results: The study included 4711 patients. Multivariate Cox regression analysis demonstrated that age, tumor size, grade, pT stage, pN stage, M stage, carcinoembryonic antigen, perineural invasion, tumor deposits, regional nodes examined, and chemotherapy were correlated with OS and CSS. Marital status was independently related to OS. In the development and validation cohorts, the C-index of OS was 0.766 and 0.744, respectively, and the C-index of CSS was 0.826 and 0.809, respectively. Calibration curves and ROC curves showed predictive accuracy. DCA showed that the nomograms had excellent potency over the 8th edition of the TNM staging system with higher clinical net benefits. Significant differences in OS and CSS were observed among low-, medium-, and high-risk groups. Conclusions: Nomograms were developed for the first time to predict personalized 1-, 3-, and 5-year OS and CSS in CRMA postoperative patients. External and internal validation confirmed the excellent discrimination and calibration ability of the nomograms. The nomograms can help clinicians design personalized treatment strategies and assist with clinical decisions.