Development and validation of a nomogram for predicting morbidity in surgically resected primary retroperitoneal sarcoma

BACKGROUND: Surgery is the cornerstone of the treatment for primary retroperitoneal sarcoma (RPS). The purpose of this study was to establish a nomogram predictive model for predicting postoperative morbidity in primary RPS. METHODS: Clinicopathological data of patients who underwent radical resection from 2009 to 2021 were retrospectively analyzed. Risk factor analysis was performed using a logistic regression model, and modeling variables were selected based on Akaike Information Criterion. The nomogram prediction model was built on the basis of a binary logistic regression model and internally validated by calibration curves and concordance index. RESULTS: A total of 319 patients were enrolled, including 162 males (50.8%). 22.9% (n = 73) were over 65 years of age, and 70.2% (n = 224) had tumors larger than 10 cm. The most common histologic subtypes were well-differentiated liposarcoma (38.2%), dedifferentiated liposarcoma (25.1%) and leiomyosarcoma (7.8%). According to the Clavien–Dindo Classification, 96 (31.1%) and 31 (11.6%) patients had grade I–II complications and grade III–V complications, respectively. Age, tumor burden, location, operative time, number of combined organ resections, weighted resected organ score, estimated blood loss and packed RBC transfusion was used to construct the nomogram, and the concordance index of which was 0.795 (95% CI 0.746–0.844). and the calibration curve indicated a high agreement between predicted and actual rates. CONCLUSIONS: Nomogram, a visual predictive tool that integrates multiple clinicopathological factors, can help physicians screen RPS patients at high risk for postoperative complications and provide a basis for early intervention.