An MRI-Based Radiomics Model for Preoperative Prediction of Microvascular Invasion and Outcome in Intrahepatic Cholangiocarcinoma

Abstract

Purpose

Microvascular invasion (MVI) serves as a significant predictor of poor prognosis in intrahepatic cholangiocarcinoma (ICC). This study aims to establish a comprehensive model utilizing MR radiomics for preoperative MVI status stratification and outcome prediction in ICC patients.

Materials and methods

A total of 249 ICC patients were randomly assigned to training and validation cohorts (174:75), along with a time-independent test cohort consisting of 47 ICC patients. Independent clinical and imaging predictors were identified by univariate and multivariate logistic regression analyses. The radiomic model was developed based on robust radiomic features extracted using a logistic regression classifier. The predictive efficacy of the models was evaluated by receiver operating characteristic curves, calibration curves and decision curves. Multivariate Cox analysis identified the independent risk factors for recurrence-free survival and overall survival, Kaplan-Meier curves were plotted, and a nomogram was used to visualize the predictive model.

Results

The imaging model included tumor size and intrahepatic duct dilatation. The radiomics model comprised 25 stable radiomics features. The Imaging-Radiomics (IR) model, which integrates independent predictors and robust radiomics features, demonstrates desirable performance for MVI (AUC_training= 0.890, AUC_validation= 0.885 and AUC_test= 0.815). The calibration curve and decision curve validate the clinical utility. Preoperative MVI prediction based on IR model demonstrated comparable accuracy in MVI stratification and outcome prediction when compared to histological MVI.

Conclusion

The IR model and the nomogram based on IR model-predicted MVI status may serve as potential tools for MVI status stratification and outcome prediction in ICC patients preoperatively.