Development and validation of a prognostic nomogram for predicting overall survival in patients with large retroperitoneal liposarcoma: a population-based cohort study

Objective This study aimed to show the clinicopathological characteristics of large retroperitoneal liposarcoma (RLS) and to developed a customized nomogram model for those patients. Methods A total of 1735 patients diagnosed with RLS were selected from the Surveillance, Epidemiology, and End Results (SEER) database. Among them, 1113 patients with a maximum tumor diameter greater than 150 mm were included for further analysis. Nomogram models were developed based on lasso and multivariate cox regression analyses. The 166 patients collected from the same period at First Medical Center, Chinese People Liberation Army General Hospital (CPLAGH), were used for external validations. The model was further validated through multiple dimensions. Results Larger tumor size in RLS was associated with worse survival outcomes (hazard ratio [HR] = 0.66, 95% confidence interval [CI]: 0.53–0.81, P < 0.05). Lasso and Cox regression analyses consistently identified age, TNM stage, occurrence pattern, histology, and surgery as important prognostic factors for OS. The constructed model demonstrated robust predictive performance, with better time-ROC (Time-Dependent Receiver Operating Characteristic) for 1-year (83.1%), 3-year (83.8%), and 5-year (81.4%) survival in training cohort. The concordance index (C-index) was approximately 0.80 in both the training and validation cohorts, reflecting excellent discriminatory ability of the model. Survival risk stratification analysis revealed significant differences in survival outcomes between the groups (HR = 4.12 [3.31–5.12], P < 0.001 in training cohort). Decision curve analysis (DCA) confirmed that the nomogram provided greater net benefits across a range of threshold probabilities. Conclusion This study identified important prognostic factors for survival in patients with large RLS and developed a reliable nomogram for predicting OS. The model’s strong predictive performance supports its use in personalized treatment strategies, improving prognosis assessment and clinical decision-making for these patients.