Related Experiment Video
Updated: Jul 12, 2026
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
A nomogram for predicting high-risk endometrial cancer based on the transvaginal ultrasonography and
Dongmei Liu1, Yangzheng Xia1, Binyu Zheng1
1Department of Ultrasound, Beijing Shijitan Hospital, Capital Medical University, Beijing, China.
Background:
The high-risk endometrial carcinoma (EC) constitutes a significant threat to the survival of women. However, existing diagnostic modalities exhibit inherent limitations. Contrast-enhanced ultrasonography (CEUS) and transvaginal ultrasonography (TVUS) have demonstrated considerable potential for oncological assessment owing to their diagnostic accuracy and operational simplicity. Therefore, this study aimed to construct a comprehensive diagnostic model for high-risk EC by synergistically integrating TVUS and CEUS parameters.
Methods:
Patients pathologically diagnosed with EC were enrolled and categorized into low-risk and high-risk groups based on pathological risk factors. Demographic information, TVUS, and CEUS examination results were collected. Intergroup comparisons were executed via Chi-square tests, independent samples t-tests, and Mann-Whitney U tests. Univariate logistic regression was employed to identify parameters associated with high-risk EC. Predictors were refined by first excluding collinear variables [variance inflation factors (VIF) ≥5], followed by least absolute shrinkage and selection operator (LASSO) regression-based feature selection. A nomogram model was developed, and its performance assessed using receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and internal validation.
Results:
A total of 128 patients with EC were enrolled in this study, among whom 62 were classified as high-risk cases. Significant differences were observed between high-risk (n=62) and low-risk (n=66) EC patients across multiple clinical characteristics and imaging parameters. Univariate logistic regression analysis revealed 18 variables significantly associated with high-risk EC. Multicollinearity assessment (VIF threshold: 5.0) identified 6 variables with severe collinearity. LASSO regression analysis identified CEUS-derived area under the curve (AUC), enhancement degree (ED), perfusion mode (PM), endometrial-myometrial border (EMB), tumor anterior-posterior diameter (TAP) and menopausal status as independent predictive factors, which were incorporated into the nomogram model. This model achieved an AUC of 0.9213, with substantial clinical net benefit and robust stability.
Conclusions:
This study constructed and validated an efficient nomogram for predicting high-risk EC, by integrating TVUS and CEUS parameters. These findings furnished guidance for the precise identification and timely management of high-risk patients.
