Abstract
BACKGROUND: Locally advanced rectosigmoid junction cancer (LARSJC) presents unique challenges in prognosis and treatment due to its anatomical ambiguity between the colon and rectum. Current staging systems may not sufficiently capture the heterogeneity of LARSJC, highlighting a clinical need for more accurate prognostic tools to guide treatment decisions and improve patient outcomes. This study aimed to investigate risk factors and develop nomograms for LARSJC patients to improve clinical outcomes across diverse patient subgroups. METHODS: We developed and validated nomograms to predict overall survival (OS) and cancer-specific survival (CSS) in LARSJC patients using data from the Surveillance, Epidemiology, and End Results (SEER) database (2010-2017). Patients with stage II and III LARSJC were included, excluding those with multiple primary cancers or incomplete data. Potential predictors, including age, sex, tumor stage, grade, size, carcinoembryonic antigen (CEA) levels, and perineural invasion, were assessed. OS and CSS were measured from diagnosis to death or last follow-up. The dataset was randomly split into a 7:3 ratio for training and validation. Predictive accuracy was assessed using C-index, receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). RESULTS: T stage, N stage, poor tumor grade, larger tumor size, age, tumor histology, positive CEA level, tumor deposits, perineural invasion, chemotherapy, and therapy sequence were selected as independent risk factors by univariate and multivariate Cox regression analyses and included in the OS nomogram model. The C-index of the OS nomogram in the training set (0.74; 95% CI: 0.72-0.76) and testing set (0.75; 95% CI: 0.73-0.77) were significantly higher than that of the American Joint Committee on Cancer (AJCC) 7th staging system (0.63; 95% CI: 0.61-0.66). The ROC curve with the calculated area under the curve (AUC) in the development cohort was 0.790, 0.782, and 0.762 for 1-, 3-, and 5-year OS, respectively. The AUC in the validation cohort was 0.752, 0.742, and 0.703 for 1-, 3-, and 5-year OS, respectively. The calibration plots for both cohorts demonstrated good agreement between actual clinical observations and predicted outcomes for 1-, 3-, and 5-year OS. In the training cohort, DCA showed that the nomogram prediction model was more advantageous for clinical application than the AJCC 7th staging system. Kaplan-Meier curves in the low and high groups showed significant differences in OS. Similar results were observed for CSS. CONCLUSIONS: This study provided a comprehensive prognostic analysis of LARSJC patients, developing nomograms that may assist in personalizing treatment plans. Limitations included the retrospective nature of the study and potential missing data in the SEER database. Future researches should focus on validating these models in prospective studies and incorporating additional biomarkers to enhance prognostic accuracy. These findings could guide clinical decision-making, identifying high-risk patients for targeted interventions and improving overall patient management.