Decoding Tumor Heterogeneity with Imaging Biomarkers Predicts Microvascular Invasion and Postoperative Risk...

Area of Science:

• Hepatocellular Carcinoma Research
• Medical Imaging Analysis
• Oncology

Background:

• Microvascular invasion (MVI) is a key factor in hepatocellular carcinoma (HCC) recurrence and poor prognosis.
• Current biopsy methods for MVI detection are invasive and prone to errors.
• There is a need for noninvasive methods to predict MVI and stratify HCC recurrence risk.

Purpose of the Study:

• To develop and validate a noninvasive preoperative model for MVI prediction in HCC.
• To integrate multiphase MRI-derived habitat/intratumoral heterogeneity (ITH) features with clinical variables.
• To stratify postoperative recurrence risk in HCC patients.

Main Methods:

• Retrospective analysis of 978 HCC patients from four centers.
• Development and validation of seven models, including a trimodal decision model.
• Performance assessment using AUC, sensitivity, specificity, calibration, and decision curve analysis.
• Prognostic value evaluation for recurrence-free survival (RFS) using C-index and ROC analysis.

Main Results:

• The decision model achieved high AUCs (0.912 training, 0.875 internal, 0.882 external validation).
• The model significantly outperformed clinical and unimodal models for MVI prediction.
• It demonstrated superior recurrence-risk stratification compared to clinicopathologic models (higher C-index and AUCs).
• Performance was consistent across subgroups (HBV status, histologic grade).

Conclusions:

• The developed preoperative decision model shows robust performance for MVI prediction.
• The model is effective for postoperative recurrence-risk stratification in HCC.
• It holds potential as an adjunct tool for preoperative risk assessment in HCC management.