Scholarlink

<p>Microorganisms play significant roles in gastric cancer (GC) progression. However, it is unknown whether circulating microbiome DNA (cmDNA) possesses GC specific features and could serve as diagnostic biomarker for GC detection. In this study, one cohort of 586 participants from Zhejiang Cancer Hospital were divided randomly into the training and testing datasets. Another cohort of 299 participants enrolled from three hospitals was used as an independent validation cohort. The cmDNA from plasma samples were analyzed by sequencing and various tools. The significant features of cmDNA were used as inputs to establish a machine learning diagnostic model (cmDNA-MLM). The cmDNA-MLM achieved the area under receiver operating characteristic curve (AUC) of 0.831 for GC across all stages in the testing cohort and 0.914 in the independent validation cohort. Notably, this model demonstrated strong performance in detecting early-stage GC, achieving an AUC of 0.792 for stage I GC in the validation cohort and exhibited favorable sensitivities across various molecular subtypes. The stage shift analysis showed a notable increase in the number of patients diagnosed at stage I. This cmDNA-MLM exhibited promising performance in early GC detection, which could be used as a clinical liquid biopsy methodology after more clinical validation studies.</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

Early detection of gastric cancer: a novel circulating microbiome DNA based liquid biopsy assay

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月29日

查看全文
  • Yongyi Chen,
  • Xinhong Han,
  • Miao Luo,
  • Liulin Luo,
  • Tingzhang Wang,
  • Congcong Kong,
  • Liuqing Ye,
  • Jiangang Jin,
  • Dingding Hou,
  • Haiqi Liao,
  • Zhonglin Wang,
  • Wei Xue,
  • Ziao Lin,
  • Songxiao Xu

摘要

Microorganisms play significant roles in gastric cancer (GC) progression. However, it is unknown whether circulating microbiome DNA (cmDNA) possesses GC specific features and could serve as diagnostic biomarker for GC detection. In this study, one cohort of 586 participants from Zhejiang Cancer Hospital were divided randomly into the training and testing datasets. Another cohort of 299 participants enrolled from three hospitals was used as an independent validation cohort. The cmDNA from plasma samples were analyzed by sequencing and various tools. The significant features of cmDNA were used as inputs to establish a machine learning diagnostic model (cmDNA-MLM). The cmDNA-MLM achieved the area under receiver operating characteristic curve (AUC) of 0.831 for GC across all stages in the testing cohort and 0.914 in the independent validation cohort. Notably, this model demonstrated strong performance in detecting early-stage GC, achieving an AUC of 0.792 for stage I GC in the validation cohort and exhibited favorable sensitivities across various molecular subtypes. The stage shift analysis showed a notable increase in the number of patients diagnosed at stage I. This cmDNA-MLM exhibited promising performance in early GC detection, which could be used as a clinical liquid biopsy methodology after more clinical validation studies.