Construction and validation of a prognostic model for early-onset rectal adenocarcinoma after surgery: A retrospective cohort study

BACKGROUNDThe incidence of early-onset colorectal cancer has been rising annually, and young patients tend to have a poor prognosis due to higher rates of local recurrence and distant metastasis.Previous studies have evaluated rectal cancer and colon cancer simultaneously; however, there is a lack of research on early-onset rectal adenocarcinoma (EO-READ) specifically. AIMTo identify the clinicopathological characteristics and prognostic factors of patients with locally advanced EO-READ, and to construct and validate a prediction model for cancerspecific survival (CSS). METHODSClinical data and clinicopathological characteristics of EO-READ patients obtained from the Surveillance, Epidemiology, and End Results (SEER) database (2000-2020) and the Third Affiliated Hospital of Anhui Medical University (2020-2025) were retrospectively analyzed.The SEER patients were randomly divided into training and internal validation groups at a ratio of 7:3.The hospital patients were divided into the preliminary external validation group, and this group was included in the validation group together with SEER's internal validation group.Univariate and multivariate Cox analyses were conducted to identify independent prognostic factors for CSS.Based on the independent prognostic variables, a nomogram was constructed and validated to predict CSS for patients with locally advanced EO-READ.A web-based survival calculator was developed and compared with the traditional tumor-node-metastasis (TNM) staging system to determine whether the nomogram had better predictive ability.4 / 42 RESULTSIn total, 7363 patients were enrolled and divided into the training group (5130 patients) and validation group (2233 patients: Internal validation group n = 2199, preliminary external validation group n = 34).There were no statistically significant differences in the clinicopathological characteristics between the groups (all P > 0.05).Multivariate Cox regression analysis showed that T stage, N stage, differentiation grade, perineural invasion, tumor deposits, number of regional nodes, tumor size, and radiotherapy were independent prognostic factors (all P < 0.05).A nomogram was constructed based on the above 8 variables, and a convenient web-based dynamic survival calculator was established.The Concordance-index of the training and validation groups was 0.750 and 0.736, respectively.The area under the receiver operating characteristic curve (AUC) values at 1 year, 3 years, and 5 years were 0.879, 0.935, and 0.834 in the training group, and 0.852, 0.893, and 0.807 in the validation group, respectively.The calibration curve showed that the actual observations were basically consistent with the nomogram predictions.For the TNM staging system, the AUC values at 1 year, 3 years, and 5 years were 0.586, 0.597, and 0.595 in the training group, and 0.617, 0.614, and 0.613, in the validation group, respectively.The Concordance-index values in the training and validation groups regarding the nomogram vs the TNM staging system were 0.750 vs 0.567, and 0.736 vs 0.574, respectively. CONCLUSIONThe web-based dynamic nomogram can predict the probability of CSS in patients with locally advanced EO-READ, which may help clinicians to assess the postoperative risk stratification and follow-up intensity.