Related Experiment Video
Updated: Aug 5, 2026
07:47
Non-Invasive PET/MR Imaging in an Orthotopic Mouse Model of Hepatocellular Carcinoma
Published on: August 31, 2022
Integrated Radioproteomic Modeling for Early Recurrence Prediction and Metabolic Characterization in Hepatocellular
Qiuyu Zhuang1,2, Tongtong Xu1,2, Xiaohua Xing1,2
1The United Innovation of Mengchao Hepatobiliary Technology Key Laboratory of Fujian Province, Mengchao Hepatobiliary Hospital of Fujian Medical University, Fuzhou, China.
JCO Precision Oncology
|July 29, 2026
Summary
An interpretable deep learning model predicts hepatocellular carcinoma (HCC) recurrence using imaging and multiomics. This approach identifies pyruvate metabolism as a key factor, suggesting stiripentol as a potential therapy for high-risk HCC patients.
Area of Science:
- Hepatocellular Carcinoma Research
- Artificial Intelligence in Oncology
- Radiomics and Multiomics Integration
Background:
- Hepatocellular carcinoma (HCC) presents a significant mortality challenge due to high recurrence rates post-surgery.
- Current deep learning (DL) models for HCC recurrence prediction lack clinical interpretability and mechanistic insight.
- Bridging AI-driven imaging with biological mechanisms is crucial for clinical translation.
Purpose of the Study:
- To develop an interpretable DL framework for predicting early HCC recurrence.
- To elucidate the biological mechanisms underlying recurrence prediction using integrated radiologic and multiomics data.
- To identify potential therapeutic targets for high-risk HCC patients.
Main Methods:
- Developed a DL framework using preoperative multiphase CT imaging to predict early HCC recurrence and generate an Early Recurrence Risk Score (ERRS).
- Integrated DL-derived features with proteomic data to identify metabolic alterations, validated by metabolomics, IHC, and enzymatic assays.
- Utilized patient-derived organoids (PDOs) to assess therapeutic targeting of identified metabolic alterations.
Main Results:
- The DL model demonstrated superior performance in predicting early HCC recurrence compared to conventional methods.
- High-ERRS patients exhibited poorer survival and more aggressive tumor features.
- Radioproteomic analysis linked ERRS to dysregulated pyruvate metabolism, with reduced pyruvate dehydrogenase and elevated lactate dehydrogenase (LDH) activity.
- HCC PDOs showed sensitivity to the LDH inhibitor stiripentol, with high-ERRS tumors being more vulnerable.
Conclusions:
- Developed a biologically interpretable DL model for HCC recurrence prediction, integrating AI imaging with mechanistic insights.
- Identified dysregulated pyruvate metabolism as a key feature of high-risk HCC, suggesting stiripentol repurposing.
- Proposed a framework linking noninvasive risk stratification with pathway-guided therapy for HCC.
