Lipid profile-based prognostic nomograms for obstructive colorectal cancer: multi-cohort development and validation

Abstract

Patients with obstructive colorectal cancer (OCRC) have poor prognosis, and conventional prognostic tools such as TNM staging alone are insufficient for accurate assessment; thus, this study aimed to develop and validate nomograms for predicting overall survival (OS) and disease-free survival (DFS) in OCRC patients. A total of 1,650 eligible colorectal cancer patients were enrolled, including 1,520 from Center A (randomly split 6:4 into training and internal validation cohorts) and 130 from Center B as the external validation cohort; clinicopathological and lipid profile data were collected retrospectively, independent prognostic factors were identified via univariate and multivariate Cox regression, and OS/DFS nomograms were constructed and evaluated using ROC curves, C-index, calibration curves, and DCA. The OS nomogram included 8 factors with 1-, 3-, 5-year AUCs of 0.761–0.793 (training), 0.759–0.788 (internal validation), and 0.723–0.856 (external validation), while the DFS nomogram included 6 factors with corresponding AUCs of 0.777–0.822, 0.750–0.841, and 0.775–0.820, and all evaluations confirmed excellent model performance. In conclusion, two nomograms incorporating lipid and clinicopathological factors were developed and validated, which effectively predict OS and DFS in OCRC patients with good discrimination, calibration, and clinical utility, providing new tools for individualized risk prediction.