Scholarlink

<p>Pneumonia remains a leading cause of global mortality. Conventional diagnostic approaches frequently fail to distinguish microbial colonization from true infection in the lower respiratory tract, complicating clinical decision-making and contributing to antibiotic overuse. Improved diagnostic strategies are urgently needed. In this prospective, single-center study, deep sputum specimens were collected from patients with respiratory colonization (n = 17) and infectious pneumonia (n = 27) admitted to the neurosurgical ICU of Huashan Hospital. Metagenomic next-generation sequencing (mNGS) and metatranscriptomic profiling were performed to characterize both the pulmonary microbiota and the host immune response. These features were subsequently integrated to construct a diagnostic model. Microbial community profiling revealed reduced alpha diversity and enrichment of metabolically active pathogenic taxa in the infection group, consistent with a dysbiotic state permissive to invasion. In contrast, the colonization group demonstrated a more balanced microbial ecosystem. Transcriptomic analyses identified 2232 differentially expressed host genes between the two groups. The colonization group showed marked activation of the Wnt, MAPK, chemokine, and focal adhesion pathways, which are functionally implicated in epithelial barrier maintenance and early immune homeostasis. A multi-omics diagnostic model incorporating seven gene features (ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3, and LRRC15) achieved a discrimination between colonization and infection (AUC = 0.951 in the training cohort; 0.875 in the validation set). By jointly analyzing the pulmonary microbiome and host transcriptome, this study provides insight into host–microbe interactions distinguishing colonization from infection and presents a predictive model with potential clinical relevance.</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

A diagnostic model based on pulmonary microbiota and host gene expression to distinguish colonization from pneumonia

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月3日

查看全文
  • Zhangfan Fu,
  • Yuhan Sun,
  • Haijun Yao,
  • Qihui Liu,
  • Qiran Zhang,
  • Jin Hu,
  • Yang Zhou,
  • Ning Jiang,
  • Jingwen Ai,
  • Jialin Jin,
  • Wenhong Zhang

摘要

Pneumonia remains a leading cause of global mortality. Conventional diagnostic approaches frequently fail to distinguish microbial colonization from true infection in the lower respiratory tract, complicating clinical decision-making and contributing to antibiotic overuse. Improved diagnostic strategies are urgently needed. In this prospective, single-center study, deep sputum specimens were collected from patients with respiratory colonization (n = 17) and infectious pneumonia (n = 27) admitted to the neurosurgical ICU of Huashan Hospital. Metagenomic next-generation sequencing (mNGS) and metatranscriptomic profiling were performed to characterize both the pulmonary microbiota and the host immune response. These features were subsequently integrated to construct a diagnostic model. Microbial community profiling revealed reduced alpha diversity and enrichment of metabolically active pathogenic taxa in the infection group, consistent with a dysbiotic state permissive to invasion. In contrast, the colonization group demonstrated a more balanced microbial ecosystem. Transcriptomic analyses identified 2232 differentially expressed host genes between the two groups. The colonization group showed marked activation of the Wnt, MAPK, chemokine, and focal adhesion pathways, which are functionally implicated in epithelial barrier maintenance and early immune homeostasis. A multi-omics diagnostic model incorporating seven gene features (ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3, and LRRC15) achieved a discrimination between colonization and infection (AUC = 0.951 in the training cohort; 0.875 in the validation set). By jointly analyzing the pulmonary microbiome and host transcriptome, this study provides insight into host–microbe interactions distinguishing colonization from infection and presents a predictive model with potential clinical relevance.