Identification of matrix stiffness-related molecular subtypes in HCC via integrating multi-omics analysis and machine learning algorithms

J Transl Med. 2025 Jul 1;23(1):716. doi: 10.1186/s12967-025-06733-7.

Abstract

Background: Matrix stiffness is strongly associated with hepatocarcinogenesis and significantly influences the properties of hepatocellular carcinoma (HCC). Investigating matrix stiffness-related signatures provides crucial insights into HCC prognosis and therapeutic response.

Methods: Multi-omics data from liver hepatocellular carcinoma (LIHC) were integrated using 10 clustering algorithms, identifying three subgroups with distinct survival outcomes and treatment responses. A matrix stiffness-related signature comprising 57 genes was constructed by evaluating 101 machine learning algorithm combinations. PPARG, the key gene with the greatest contribution to the model, was selected for validation. Single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) analyses assessed matrix stiffness activity scores across different cell subgroups and examined PPARG spatial localization within tissues. Experimental studies and bioinformatics analyses further explored the role of PPARG in HCC carcinogenesis and the immune microenvironment.

Results: The matrix stiffness-related signature demonstrated superior prognostic prediction performance in both training and validation cohorts compared to other existing HCC signatures. Distinct immune and mutation landscape characteristics were observed between patients categorized into high and low matrix stiffness groups. PPARG functioned in tumorigenesis through HSC activation and immune suppression. Furthermore, increased matrix stiffness was found to upregulate PPARG expression, promoting cell proliferation, activating lipid metabolism, and enhancing the stemness of HCC cells through the MAPK signaling pathway. Targeting PPARG with trametinib displayed an enhanced therapy response.

Conclusions: The matrix stiffness-related signature not only serves as a robust prognostic tool but also aids in identifying immune characteristics and optimizing therapeutic strategies, thus advancing personalized medicine for patients with HCC.

Keywords: Hepatocellular carcinoma; Machine learning; Matrix stiffness; Multi-omics; Signature.

MeSH terms

  • Algorithms*
  • Carcinoma, Hepatocellular* / genetics
  • Carcinoma, Hepatocellular* / pathology
  • Cluster Analysis
  • Extracellular Matrix* / metabolism
  • Extracellular Matrix* / pathology
  • Gene Expression Profiling
  • Gene Expression Regulation, Neoplastic
  • Humans
  • Liver Neoplasms* / classification
  • Liver Neoplasms* / genetics
  • Liver Neoplasms* / pathology
  • Machine Learning*
  • Male
  • Multiomics
  • PPAR gamma / genetics
  • PPAR gamma / metabolism
  • Prognosis
  • Reproducibility of Results

Substances

  • PPAR gamma