Back

Airway microbiome-host transcriptome networks link microbial dysbiosis to survival outcomes and therapeutic opportunities in severe COVID-19

Chen, K.-P.; Huang, C.-H.; Chen, K.-Y.; Hsu, C.-L.; Huang, H.-C.; Juan, H.-F.

2025-06-25 microbiology
10.1101/2025.06.24.661228 bioRxiv
Show abstract

The lower respiratory tract microbiome may critically influence outcomes in severe COVID-19, yet remains understudied. We analyzed bronchoalveolar lavage fluid from 86 severe patients requiring mechanical ventilation, stratified into two groups. We identified 213 bacteria with differential abundance, 11 bacteria associated with differentially expressed genes, and 39 of them were associated with patient survival. The more serious group exhibited reduced microbial network complexity, suggesting ecosystem disruption. To explore therapeutic potential, we employed our previously published model and predicted that compounds like palbociclib, XMD-1499, RS-17053, and RS-504393 may shift host gene expression favoring beneficial microbial signatures. These results highlight an important link between microbial dysbiosis and COVID-19 severity, offering promising directions for microbiome-informed therapeutic strategies. Our findings underscore the importance of the lung microbiome in modulating host responses and clinical outcomes, and suggest that modulating microbial-host interactions may serve as a novel adjunctive approach in the treatment of severe COVID-19. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=101 SRC="FIGDIR/small/661228v1_ufig1.gif" ALT="Figure 1"> View larger version (39K): org.highwire.dtl.DTLVardef@2fc879org.highwire.dtl.DTLVardef@1128d5dorg.highwire.dtl.DTLVardef@fef0eaorg.highwire.dtl.DTLVardef@17aa499_HPS_FORMAT_FIGEXP M_FIG C_FIG

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.