The independent prognostic nomogram models for primary and recurrent retroperitoneal liposarcoma: a population-based cohort study

Purpose The aim of this study was to screen and establish independent prognostic models for primary and recurrent retroperitoneal liposarcoma (RLS). Methods A total of 2,429 patients confirmed to have RLS were extracted from the Surveillance, Epidemiology and End Results (SEER) database. The 245 patients collected from the same period at First Medical Center, Chinese People Liberation Army General Hospital (CPLAGH), were used for external validations. Nomogram were built on the basis of clinical practicability, univariate and multivariate Cox analyses. Results After performing a stepwise analysis, the simplified predictive models for primary RLS were primarily based on tumor size (median size, 162 mm [range, 90–230], p < 0.001) and pathological subtypes (WDL vs. DDL, hazard ratio [HR] = 2.11; 95% confidence interval [CI] = 1.71–2.61; p < 0.001), both of which can be readily obtained in outpatient settings. In contrast, TNM stage (HR = 2.18; 95% CI = 1.49–3.20; p < 0.001), an important postoperative prognostic factor, emerged as a significant predictor for recurrent RLS. The area under the time-dependent receiver operating characteristic curve (time-dependent AUC) and the concordance index (C-index) for overall survival (OS) and cancer-specific survival (CSS) models both approached 0.75 in both training and validation cohorts. Moreover, calibration curves and decision curve analysis (DCA) demonstrated that the validated models were not only reliable but also clinically applicable. Conclusion We have developed efficient and independent models for both primary and recurrent RLS. These models will provide invaluable clinical guidance, aiding in prognostication and facilitating personalized therapeutic decision-making.