Spatial omics-based machine learning algorithms for the early detection of hepatocellular carcinoma

Area of Science:

• Biomarker discovery
• Glycomics
• Proteomics
• Machine learning in oncology

Background:

• Hepatocellular carcinoma (HCC) is a leading cause of cancer death globally.
• Early-stage HCC is treatable, but effective biomarkers for early detection are lacking.
• Current omics approaches have not identified suitable HCC biomarkers.

Purpose of the Study:

• To develop a novel platform for identifying early hepatocellular carcinoma (HCC) biomarkers.
• To investigate altered N-glycan structures in HCC tissues and serum.
• To validate potential biomarkers for differentiating HCC from cirrhosis.

Main Methods:

• Utilized spatial omics N-glycan imaging on HCC tissues and paired serum.
• Employed glycoproteomics to identify glycoproteins associated with altered N-glycans.
• Developed machine learning algorithms using glycan and glycoprotein data.
• Tested algorithms on independent case-control sample sets.

Main Results:

• Identified thirteen specific N-glycans altered in HCC tissue and serum.
• Over 50 glycoproteins found to contain these altered N-glycan structures.
• Machine learning algorithms accurately differentiated HCC from cirrhosis (AUROC 0.88-0.97).

Conclusions:

• A new biomarker platform for early HCC detection has been developed.
• The platform effectively identifies biomarkers for early-stage and advanced HCC.
• This approach may be applicable to other diseases with altered N-linked glycosylation.