Preoperative prediction of glypican-3 positive expression in solitary hepatocellular carcinoma on gadoxetate-disodium enhanced magnetic resonance imaging

Yidi Chen, Yun Qin, Yuanan Wu, Hong Wei, Yi Wei, Zhen Zhang, Ting Duan, Hanyu Jiang, Bin SongFrontiers in Immunology 2022 · DOI

Abstract

Based on serum AFP and five EOB-MRI features, we developed an easy-to-use and noninvasive risk score which could accurately predict positive GPC-3 HCC, which may help identify potential responders for GPC-3-targeted immunotherapy.

Introduction

Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide. A specific protein, glypican-3 (GPC-3), is expressed in approximately 70–80% of HCC cases and is closely associated with aggressive tumor behavior and poor clinical outcomes. As GPC-3 is highly specific to malignant cells, it has emerged as a primary target for next-generation immunotherapies, such as chimeric antigen receptor (CAR) T-cell therapy.

Currently, determining GPC-3 status requires invasive procedures like biopsy or surgical resection. Biopsies carry risks of complications and potential sampling errors. Consequently, there is a critical clinical need for non-invasive methods to assess the molecular profile of tumors preoperatively. This study, published in Frontiers in Immunology, explores the use of gadoxetate-disodium enhanced MRI (EOB-MRI) combined with serum markers to predict GPC-3 expression.

Methods and Study Design

The researchers conducted a retrospective analysis of 278 patients with solitary HCC. The study utilized EOB-MRI, a specialized imaging technique where contrast agents are selectively absorbed by healthy hepatocytes, providing detailed anatomical and functional insights.

Flowchart of the retrospective study cohort showing the selection process for the 278 participants.

As illustrated in the flowchart, the cohort was divided into three datasets for robust evaluation:

  • Training Set (n=156): Used for model development.
  • Internal Validation Set (n=39): Used for initial testing of the model within the same institution.
  • External Validation Set (n=83): Sourced from a different hospital to ensure the model's generalizability.

The team evaluated 37 imaging features based on the Liver Imaging Reporting and Data System (LI-RADS). Using multivariable logistic regression, they identified the most significant predictors of GPC-3 positivity to create a visual scoring tool (nomogram).

Key Insights and Predictive Model

The study demonstrated that molecular characteristics manifest as observable morphological features on MRI. A key finding was that an infiltrative appearance—where tumor margins appear ill-defined—was the strongest predictor of GPC-3 positivity (odds ratio: 9.3). Conversely, "iron sparing in solid mass" was a strong indicator of GPC-3 negative status.

The final non-invasive risk score incorporates six variables:

  1. Serum AFP > 10 ng/ml (+4 points)
  2. Tumor size > 3.0 cm (-3 points)
  3. Nonperipheral "washout" (+5 points)
  4. Infiltrative appearance (+10 points)
  5. Marked diffusion restriction (+5 points)
  6. Iron sparing in solid mass (-7 points)

Results

Using an optimal threshold of 5.5 points, the model achieved moderate-to-high accuracy across all datasets. The Area Under the Curve (AUC) was 0.775 in the training set, 0.726 in internal validation, and 0.681 in external validation. Notably, the model showed a high specificity of 87.8% in the internal validation set, effectively identifying GPC-3 negative patients and potentially preventing unnecessary targeted treatments.

Discussion and Future Directions

While the results are promising, the authors noted limitations, including the retrospective nature of the study and the focus on solitary tumors. Future research may involve combining these visual radiological features with quantitative radiomics data to further enhance predictive accuracy. This tool represents a significant step toward personalized, non-invasive molecular profiling in liver cancer management.

Summary updated

Abstract

Based on serum AFP and five EOB-MRI features, we developed an easy-to-use and noninvasive risk score which could accurately predict positive GPC-3 HCC, which may help identify potential responders for GPC-3-targeted immunotherapy.
BibTeX
@inproceedings{chen2022preoperative,
  author = {Yidi Chen and Yun Qin and Yuanan Wu and Hong Wei and Yi Wei and Zhen Zhang and Ting Duan and Hanyu Jiang and Bin Song},
  title = {{Preoperative prediction of glypican-3 positive expression in solitary hepatocellular carcinoma on gadoxetate-disodium enhanced magnetic resonance imaging}},
  booktitle = {Frontiers in Immunology},
  year = {2022},
  doi = {10.3389/fimmu.2022.973153},
}
Figure