Scholarlink

<p>Hepatocellular carcinoma (HCC) remains a major global health burden with persistently high morbidity and mortality. Dysregulation of the tumor microenvironment (TME) associated with chronic hepatitis B virus (HBV)/hepatitis C virus (HCV) infection and inflammation contributes to tumor progression, therapeutic resistance, and adverse clinical outcomes. Many existing HCC prognostic models do not systematically integrate infection-associated molecular signatures with immune and genomic features, which may limit their predictive performance in infection-associated HCC populations. In this study, we established an integrated predictive framework based on multi-cohort transcriptomic, somatic mutation, immune infiltration, and clinical data, together with artificial intelligence (AI)-assisted analysis. Here, “infection-associated” refers to a combined gene set covering HBV/HCV infection, antiviral and inflammatory responses, and immune regulation. We first constructed a comprehensive 248-gene infection-associated set, identified two robust HCC molecular subtypes with distinct infection-status annotations, immune infiltration features, genomic alteration profiles, and clinical prognosis through unsupervised consensus clustering, and screened 12 core prognostic genes that were significantly enriched in key oncogenic pathways via three complementary machine learning algorithms. The prognostic risk model was developed in the TCGA-LIHC training cohort and externally validated in ICGC-LIRI-JP and GSE14520, with 1-year area under the curve (AUC) values reaching 0.82, 0.78, and 0.75, respectively; it also showed independent prognostic value for HCC (hazard ratio (HR) = 2.87, <i>P</i> &lt; 0.001) and higher clinical net benefit than traditional clinicopathological models in retrospective decision curve analysis (DCA). These results suggest associations among infection-associated molecular characteristics, TME remodeling, and HCC prognosis. Together, the infection-associated signature may support retrospective HCC risk stratification, although prospective validation is required before routine clinical application.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multiomics and artificial intelligence identify infection associated subtypes and prognostic signatures in hepatocellular carcinoma

  • 期刊论文

  • 发表日期：

    2026年7月22日

查看全文
  • Peng Wang,
  • Jiayan Fang,
  • Wenna Zhao,
  • Yijuan Huang

摘要

Hepatocellular carcinoma (HCC) remains a major global health burden with persistently high morbidity and mortality. Dysregulation of the tumor microenvironment (TME) associated with chronic hepatitis B virus (HBV)/hepatitis C virus (HCV) infection and inflammation contributes to tumor progression, therapeutic resistance, and adverse clinical outcomes. Many existing HCC prognostic models do not systematically integrate infection-associated molecular signatures with immune and genomic features, which may limit their predictive performance in infection-associated HCC populations. In this study, we established an integrated predictive framework based on multi-cohort transcriptomic, somatic mutation, immune infiltration, and clinical data, together with artificial intelligence (AI)-assisted analysis. Here, “infection-associated” refers to a combined gene set covering HBV/HCV infection, antiviral and inflammatory responses, and immune regulation. We first constructed a comprehensive 248-gene infection-associated set, identified two robust HCC molecular subtypes with distinct infection-status annotations, immune infiltration features, genomic alteration profiles, and clinical prognosis through unsupervised consensus clustering, and screened 12 core prognostic genes that were significantly enriched in key oncogenic pathways via three complementary machine learning algorithms. The prognostic risk model was developed in the TCGA-LIHC training cohort and externally validated in ICGC-LIRI-JP and GSE14520, with 1-year area under the curve (AUC) values reaching 0.82, 0.78, and 0.75, respectively; it also showed independent prognostic value for HCC (hazard ratio (HR) = 2.87, P < 0.001) and higher clinical net benefit than traditional clinicopathological models in retrospective decision curve analysis (DCA). These results suggest associations among infection-associated molecular characteristics, TME remodeling, and HCC prognosis. Together, the infection-associated signature may support retrospective HCC risk stratification, although prospective validation is required before routine clinical application.