Development and Validation of Predictive Machine Learning Models for Postoperative Recurrence and Microvascular...

Area of Science:

• Metabolomics
• Biomarker Discovery
• Cancer Research

Background:

• Hepatocellular carcinoma (HCC) has a poor prognosis, highlighting the need for improved diagnostic and prognostic tools.
• Nuclear magnetic resonance (NMR)-based metabolomics shows promise for cancer biomarker discovery but is underexplored in HCC.
• Current management of HCC requires enhanced tools for predicting postoperative outcomes.

Purpose of the Study:

• To identify plasma metabolic biomarkers for HCC diagnosis and prognosis.
• To develop predictive models for postoperative recurrence and microvascular invasion (MVI) in HCC patients.
• To enhance clinical management of HCC through advanced predictive tools.

Main Methods:

• Untargeted NMR metabolomic profiling of plasma from 92 HCC patients and 92 controls.
• Identification of differential metabolites and assessment of diagnostic performance using receiver operating characteristic (ROC) curves.
• Development and validation of predictive models for recurrence and MVI using machine learning algorithms (e.g., random forest, support vector machine).

Main Results:

• Significant metabolic differences were observed, with 67 metabolites and blood-lipid indicators altered in HCC.
• Acetic acid, dimethylsulfone, glycerol, glycine, and LDL-3 cholesterol showed high discriminatory power (AUC ≥0.954).
• Recurrence prediction model achieved an AUC of 0.811; MVI prediction model showed superior performance (AUC = 0.957).

Conclusions:

• Developed and validated NMR-based models for HCC prognosis and MVI prediction.
• These models offer valuable tools for precision management of HCC.
• Further validation in larger prospective cohorts is warranted to confirm clinical utility.