본문으로 건너뛰기
← 뒤로

Comprehensive multi-omics data to construct hepatocellular carcinoma pathway subtypes and classification model.

1/5 보강
Computational biology and chemistry 2026 Vol.120(Pt 2) p. 108753
Retraction 확인
출처

Ji L, Li X, Gao B

📝 환자 설명용 한 줄

The high heterogeneity of Hepatocellular Carcinoma (HCC) severely hampers clinical outcomes.

이 논문을 인용하기

BibTeX ↓ RIS ↓
APA Ji L, Li X, Gao B (2026). Comprehensive multi-omics data to construct hepatocellular carcinoma pathway subtypes and classification model.. Computational biology and chemistry, 120(Pt 2), 108753. https://doi.org/10.1016/j.compbiolchem.2025.108753
MLA Ji L, et al.. "Comprehensive multi-omics data to construct hepatocellular carcinoma pathway subtypes and classification model.." Computational biology and chemistry, vol. 120, no. Pt 2, 2026, pp. 108753.
PMID 41175568

Abstract

The high heterogeneity of Hepatocellular Carcinoma (HCC) severely hampers clinical outcomes. Current classifications based on gene expression profiles may not adequately capture patients' overall biological mechanisms and phenotypic changes. Therefore, this study constructed HCC subtypes by integrating multi-omics data based on gene set pathways. We identified transcriptionally dysregulated genes, generated pathway scoring matrices for four gene sets using GSVA, and incorporated mutation data (with network smoothing) to define subtypes. We successfully constructed three subtypes (PS1, PS2, and PS3) that exhibited distinct patterns in immune infiltration, biological pathway characteristics, and genomic instability. Among these, PS3 showed the worst prognosis and was more enriched in pathways related to cell proliferation. The higher silhouette coefficient confirmed the validity of the classification. Moreover, our analysis identified six subtype-specific drugs, such as KU_55933 and Cyclophosphamide, that were more sensitive to PS1. Screening the differential pathways, we identified nine core pathways and performed mutation profiling on the extension of telomeres pathway, successfully identifying three telomere-related biomarkers: POLD1, RFC1, and TERF1. Meanwhile, we constructed a nomogram with important clinical features and validated the subtype of independent prognostic significance. Finally, by integrating 15 machine learning algorithms, we established a reproducible classification model (NN_MLP_10: AUC = 0.930) based on 10 genes. In conclusion, this study successfully constructed and evaluated a pathway-based molecular subtype and classification model for HCC, thoroughly investigated the biological and multi-omics differences between subtypes. Additionally, the identification of three telomere-associated biomarkers offers guidance and a theoretical basis for personalized treatment and clinical use of drugs for HCC patients.

MeSH Terms

Carcinoma, Hepatocellular; Humans; Liver Neoplasms; Biomarkers, Tumor; Gene Expression Profiling; Multiomics

같은 제1저자의 인용 많은 논문 (5)