Related Experiment Video
Updated: May 14, 2025
A Biomimetic Model for Liver Cancer to Study Tumor-Stroma Interactions in a 3D Environment with Tunable Bio-Physical Properties
Published on: August 7, 2020
Development of a model combining CEUS LI-RADS and clinical features for predicting glypican-3 expression in
Fen Huang1,2, Jinshu Pang1, Yuquan Wu1
1Department of Medical Ultrasound, The First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi Zhuang Autonomous Region, China.
Objective:
To establish a predictive model incorporating clinical features and contrast-enhanced ultrasound (CEUS) liver Imaging Reporting and Data System (LI-RADS) for predicting glypican-3 (GPC3) expression in hepatocellular carcinoma (HCC).
Methods:
A total of 142 HCC patients between January 2020 to June 2021 in our institution were retrospectively analyzed. All patients underwent CEUS before surgery, and the reference standard was immunohistochemical analysis of surgical specimen. The clinical features, conventional ultrasound features, and CEUS LI-RADS features of patients in the GPC3-positive and GPC3-negative groups were evaluated and compared. The variables screened by multivariable logistic regression were used to develop a model for predicting GPC3 expression and the predictive precision and clinical utility of the model was evaluated using receiver operating characteristic analysis and decision curve analysis.
Results:
Among the 142 HCC patients, 96 (67.6%) were classified as LR-4/5 lesions, 46 (32.4%) were classified as LR-M lesions, 101 (71.1%) were GPC3-positive and 41 (28.9%) were negative. Multivariable logistic regression analysis showed that younger age (OR = 0.947; 95% CI: 0.910-0.985; p = 0.007), alpha-fetoprotein > 400 ng/ml (OR = 5.202; 95% CI: 1.808-14.966; p = 0.002) and LI-RADS M (OR = 2.822; 95% CI: 1.101-7.236; p = 0.031) was independent risk factors for GPC3-positive HCC. The model combining clinical features and LI-RADS categories showed better performance than single variable, with AUC of 0.759 (p < 0.05). The nomogram and decision curves revealed substantial clinical benefit of the prediction model in predicting GPC3 expression.
Conclusion:
The combined model incorporating clinical features and CEUS LI-RADS achieved a satisfactory performance for predicting GPC3 expression in HCC patients.
