A new study investigated whether the two well-established molecular phenotypes of acute respiratory distress syndrome (ARDS)—hyperinflammatory and hypoinflammatory—are associated with distinct biological processes in the lung. Although these phenotypes were originally defined using plasma biomarkers and clinical variables and are known to differ in prognosis and treatment response, their pulmonary biology has remained poorly characterised. The authors aimed to address this gap by analysing lower respiratory tract transcriptomes using bulk and single-cell RNA sequencing approaches in both COVID-19 and non–COVID-19 ARDS.
The investigators studied tracheal aspirate (TA) samples from two observational cohorts. The first was the ALI (Acute Lung Injury in Critical Illness) cohort, comprising mechanically ventilated adults with non–COVID-19 ARDS. The second was the COMET (COVID-19 Multiphenotyping for Effective Therapies) cohort, which included patients with COVID-19 ARDS and a subset with non–COVID-19 ARDS. ARDS diagnosis was adjudicated using the Berlin definition, and molecular phenotype assignment was performed using validated plasma biomarker–based or clinical classifier models. Hyperinflammatory ARDS was defined by elevated inflammatory cytokines, lower protein C and bicarbonate levels, and higher illness severity.
Bulk RNA sequencing was performed on TA samples from both cohorts, while single-cell RNA sequencing (scRNASeq) was conducted on early TA samples from patients with non–COVID-19 ARDS in the COMET cohort. In addition, plasma proteomic profiling was available for a subset of patients in the ALI cohort. Differential gene expression, pathway enrichment, cell–cell communication, and proteomic analyses were undertaken to identify reproducible biological differences between phenotypes.
In bulk RNASeq analyses, the authors identified substantial transcriptomic differences between ARDS phenotypes. In the non–COVID-19 ALI cohort, 1,157 genes were differentially expressed between hyperinflammatory and hypoinflammatory ARDS, while 85 genes differed in the COVID-19 ARDS cohort. Despite these quantitative differences, 18 genes were reproducibly differentially expressed across both cohorts, indicating conserved biological features. Genes more highly expressed in hyperinflammatory ARDS included IL32, a pro-inflammatory cytokine; HSPA8, a heat-shock protein; and PPP3CC, a calcineurin subunit involved in T-cell activation. In contrast, hypoinflammatory ARDS showed greater expression of genes associated with tissue repair and innate immune sensing, such as FRAT1 and CD14.
Gene set variation and enrichment analyses further demonstrated consistent phenotype-specific patterns. A hyperinflammatory ARDS gene signature derived from the ALI cohort was significantly enriched in hyperinflammatory samples from the COMET cohort, including in sensitivity analyses restricted to the first available sample per patient. Across both cohorts, 195 Reactome pathways were reproducibly enriched between phenotypes. Hyperinflammatory ARDS was characterised by increased expression of pathways related to granulopoiesis, pro-inflammatory cytokine signalling, interferon (IFN) responses, T-cell receptor signalling, cell cycle activity, and the integrated stress response. In contrast, hypoinflammatory ARDS demonstrated greater enrichment of glycosaminoglycan metabolism and extracellular matrix organisation pathways, consistent with earlier engagement of repair and resolution processes.
Single-cell RNASeq provided additional mechanistic insight into these findings. Analysis of over 26,000 cells from patients with non–COVID-19 ARDS revealed that neutrophils were the dominant cell population in both phenotypes. However, hyperinflammatory ARDS showed cell-specific transcriptional changes, including increased expression of IL1R2 in neutrophils, a marker associated with emergency granulopoiesis and severe systemic inflammation. Pathway analyses at the single-cell level corroborated bulk RNASeq results, demonstrating enhanced IFN signalling in monocyte-derived macrophages, metabolic reprogramming in T cells, and activation of stress response pathways across multiple immune cell types in hyperinflammatory ARDS.
Cell–cell communication analysis using CellChat predicted substantially greater T-cell–centred signalling networks in hyperinflammatory ARDS. Increased interactions were observed between T cells and other immune cells, driven by pathways including MHC class I signalling to CD8 T cells and natural killer cells. In contrast, MHC class II signalling was more prominent in hypoinflammatory ARDS. Enhanced signalling via NAMPT from T cells and NK cells to myeloid cells was also identified in hyperinflammatory ARDS, a finding of interest given prior associations between NAMPT and ARDS risk.
Plasma proteomic analysis supported the transcriptomic findings. Among 73 analysable proteins, 28 were significantly elevated in hyperinflammatory ARDS, including established inflammatory markers such as IL-6 and IL-8. Proteins linked to IFN signalling (CXCL9, CXCL10) and T-cell activation (CD5, TNFRSF9/CD137) were also increased, reinforcing the central role of IFN-driven immunity and T-cell activation in the hyperinflammatory phenotype.
This is the first comprehensive demonstration that ARDS molecular phenotypes defined in blood are associated with distinct pulmonary biology. Hyperinflammatory ARDS, which carries higher mortality, is marked by dysregulated airspace inflammation characterised by granulopoiesis, IFN signalling, and T-cell activation, whereas hypoinflammatory ARDS shows features more consistent with repair and resolution. The study highlights the importance of compartment-specific immune responses, noting that IFN and T-cell pathways previously observed in blood differ in directionality when examined in the lung.
Source: AJRCCM
Image Credit: iStock
