Development and validation of survival prediction nomograms for patients with early-stage rectal cancer: a population-based study.

Abstract

Background: The incidence of colorectal cancer (CRC) has been rising in recent years, with a concurrent increase in early-stage rectal cancer (ESRC) cases. This study aimed to investigate risk factors and developed nomograms to predict overall survival (OS) and cancer-specific survival (CSS) in ESRC patients in order to improve clinical outcomes across diverse patient subgroups.

Methods: Risk factors were investigated in ESRC patients by analyzing data from the Surveillance, Epidemiology, and End Results (SEER) database. We developed and validated nomograms to predict OS and CSS after dividing patients into two risk groups. Then we assessed the potential benefits of various therapies across subgroups after propensity score-matching (PSM).

Results: T stage, tumor grade, age, carcinoembryonic antigen (CEA) levels, tumor size, and surgical options emerged as independent risk factors through univariate and multivariate Cox regression analyses, contributing to the OS nomogram; while for CSS, the identified risk factors were tumor grade, age, elevated CEA levels and surgical options. The Concordance-index of the nomogram surpassed that of the American Joint Committee on Cancer (AJCC) 7^th staging system, with values of 0.69 (C-index, 0.64-0.74) in the training set and 0.65 (C-index, 0.62-0.68) in the testing set. The receiver operating characteristic (ROC) analysis revealed area under the curve (AUC) values of 0.70, 0.70, and 0.67 for 1-, 3-, and 5-year OS in the development cohort, with comparable results in the validation cohort. Calibration plots demonstrated strong alignment between predicted and observed outcomes. Decision curve analysis (DCA) confirmed the nomogram's superior clinical utility relative to the AJCC 7^th staging system, with similar findings for CSS. Kaplan-Meier curves illustrated significant differences in OS and CSS between low- and high-risk groups. Notably, radiation and chemotherapy conferred no benefit, while low-risk patients, especially younger individuals, may benefit from local resection.

Conclusions: This study presents a comprehensive prognostic analysis of patients with ESRC and developed predictive nomograms for OS and CSS. Subgroup analyses highlight the potential benefits of local resection in younger patients with low risk.