Preoperative Assessment for High–Risk Endometrial Cancer by Developing a MRI–and Clinical–Based Radiomics Nomogram: A Multicenter Study in Eastern China

Background: High–risk and low–risk endometrial cancer (EC) differ in the extent of the surgical procedure. Assessment of the risk stratification of EC is essential for planning the appropriate surgery. The aim of this study was to develop an radiomics nomogram for preoperative prediction of high–risk EC. Methods: A total of 718 histopathologically confirmed EC patients from five centers were retrospectively divided into a primary cohort (n = 285, Center A), an internal validation cohort (n = 109, Center A), and an external validation cohort (n = 324, Centers B to E). A binary least absolute shrinkage and selection operator logistic regression analysis was used to select radiomics features from MRIs and multivariate logistic regression analysis was used to select clinical features for high–risk EC. A radiomics nomogram was generated by combining the selected radiomics features (radiomics signatures) with the selected clinical features (clinical signatures). A receiver operating characteristic curve and a decision curve were generated to evaluate the predictive performance and clinical usefulness of radiomics signatures, clinical signatures, and radiomics nomogram. The radiomics nomogram was further used to assess deep myometrial invasion (DMI) and other risk factors of EC. Findings: Of the 358 extracted radiomics features, 17 were ultimately selected from MRIs. The clinical signatures consisted of metabolic syndrome, cancer antigen 125, age, tumor size, and tumor grade following curettage. The radiomics nomogram had highest areas under the curve and achieved the highest clinical net benefit for the decision curve. The radiomics nomogram assessed DMI and other risk factors of EC with high clinical benefit. Interpretation: The radiomics nomogram exhibited good performance in the individual prediction of high–risk EC with good clinical net benefit and good application in assessing DMI and other risk factors of EC. Funding Statement: National Natural Science Foundation of China (No. 81971579). Declaration of Interests: The authors stated: "There are no conflicts of interest." Ethics Approval Statement: This study was approved by the Obstetrics and Gynecology Hospital of Fudan University Institutional Review Board, and informed consent was waived.