Survival analysis and prediction of early-onset colorectal cancer patients post-chemotherapy: an analysis based on the SEER database.

Abstract

Background: The incidence of Early-Onset Colorectal Cancer (EOCRC) has risen markedly in recent years, garnering widespread attention due to its distinctive clinical and biological features. However, systematic research on prognostic risk factors and long-term survival prediction for EOCRC patients undergoing postoperative chemotherapy remains scarce. This study seeks to pinpoint critical prognostic factors for EOCRC patients receiving postoperative chemotherapy and to devise a survival prediction tool employing a Nomogram model.

Methods: Patients diagnosed with EOCRC between 2010 and 2015, who underwent postoperative chemotherapy, were extracted from the SEER (Surveillance, Epidemiology, and End Results) database. Only those meeting the inclusion criteria were included. Univariate and multivariate Cox regression analyses were performed to determine independent risk factors influencing overall survival (OS). A Nomogram model was then developed using significant variables. The model's predictive accuracy and clinical utility were assessed through the concordance index (C-index), calibration curves, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA).

Results: A cohort of 9,205 patients was analyzed, with 6,445 randomly allocated to the training group and 2,760 to the validation group from the SEER database. Independent prognostic factors, including gender, race, marital status, primary tumor location, histological type, TNM stage, CEA levels, bone metastasis, liver metastasis, and lung metastasis, were identified through univariate and multivariate Cox regression analyses. A Nomogram model constructed from these factors yielded a C-index of 0.76 (0.75, 0.77) in the training group and 0.76 (0.75, 0.78) in the validation group, reflecting robust discriminative ability and consistency. The area under the curve (AUC) for predicting 1-year OS was calculated as 0.84 (0.81, 0.86) in the training group and 0.82 (0.78, 0.85) in the validation group. For 3-year OS, AUCs were recorded at 0.83 (0.82, 0.84) and 0.82 (0.80, 0.84), respectively, while for 5-year OS, AUCs reached 0.81 (0.80, 0.82) and 0.82 (0.80, 0.84). Calibration curves demonstrated close alignment between predicted and observed survival rates. Additionally, DCA affirmed the model's clinical decision-making value.

Conclusion: Prognostic risk factors for EOCRC patients receiving postoperative chemotherapy were systematically evaluated in this study, leading to the development of a Nomogram-based survival prediction model. This tool offers a robust scientific foundation for tailoring individualized treatment and guiding follow-up strategies.