Extracellular Vesicle miRNAs as Biomarkers of Asthma Severity

To the Editor,

Asthma is a heterogeneous chronic lung disease with challenges in diagnosis and management¹. Identifying reliable biomarkers to characterize the asthma severity is essential for improving patient outcomes. Extracellular vesicles (EV) play a key role in intercellular communication by facilitating interactions among various lung cell types, modulating inflammatory responses, and influencing cellular differentiation and proliferation². EVs carry microRNAs (miRNAs) that regulate immune functions and may serve as circulating biomarkers^3,4. Plasma-derived EVs are stable and may reflect physiological and pathological changes; however, their role in assessing asthma severity remains underexplored.

The primary aim of this study was to investigate whether circulating EV-associated miRNAs could serve as biomarkers to distinguish patients based on asthma severity. We hypothesized that specific EV miRNAs would be differentially abundant in severe asthma compared to non-severe asthma and healthy controls, potentially reflecting biological processes associated with disease severity.

This study utilized a well-characterized cohort from Severe Asthma Research Program (SARP), comprising healthy controls (HC), non-severe asthma (NSA), and severe asthma (SA) patients, classified according to the SARP guidelines. HC participants had no history of asthma, allergic disease, or chronic respiratory conditions (Table S1). In our cohort, HC subjects had an average FEV₁ of 101.02 ± 9.46%, low IgE levels (46.17 ± 51.09 KU/L), low blood and sputum eosinophil counts, and no corticosteroid use. SA was defined according to SARP III criteria as requiring high-dose inhaled corticosteroids (ICS; ≥880 mcg fluticasone-equivalent/day) for at least 6 of the past 12 months, with or without daily oral corticosteroids (OCS), and exhibiting poor asthma control or frequent exacerbations. In our cohort, all SA participants were on ICS (100%), with 30% receiving biologics and 10% on daily oral corticosteroids (Table S1–S2). They showed significantly impaired lung function (FEV₁ 66.31 ± 13.87%), elevated IgE levels, and increased blood and sputum eosinophil counts. NSA included patients with mild to moderate asthma who did not meet the criteria for severe asthma. Mild intermittent or persistent asthma (FEV₁ ≥80%, no or low-dose ICS) moderate asthma (FEV₁ <80%, low to medium-dose ICS or no controller therapy). In our cohort, 60% of NSA participants were on ICS, none were on biologics, and they used fewer controllers (1.4 ± 1.1). Their lung function was relatively preserved (FEV₁ 85.60 ± 11.44%). Blood eosinophils and IgE levels were elevated compared to HC, and sputum eosinophils were higher than in HC but lower than in the SA group (Table S1–S2).

To identify plasma-derived EV miRNA biomarkers, we isolated and characterized EVs from plasma samples and performed small RNA sequencing to profile their miRNA content. Physiochemical characterization revealed that the EV isolated from SA patient exhibited higher particle counts compared to HC and NSA patients. However, no significant differences were observed in EV size (Figure 1A–D), surface marker abundance (Figure S1–2), or RNA biotype distribution across asthma severity groups (Table S3). The average number of total reads mapped to small RNA biotypes per group ranged from 1.05 to 1.87 million. We observed a distinct pattern and distribution of small RNA biotypes among HC, NSA, and SA groups. miRNAs were the most abundant, comprising 86.71 ± 9.77% in HC, 88.37 ± 6.85% in NSA, and 89.45 ± 6.04% in SA, followed by protein-coding transcripts (10.69 ± 7.89%, 9.47 ± 5.59%, and 8.51 ± 4.98%, respectively). Other small RNA species, including misc_RNA (0.21–0.28%), snoRNA (0.08–0.14%), and snRNA (0.02–0.03%), were present in much lower proportions (Table S3). Based on these findings and the scope of this pilot study, we chose to focus on miRNAs, given their dominant representation and well-established role in EV-mediated signaling.

Figure 1. Physiochemical characterization of plasma-derived extracellular vesicles (EVs) and differential miRNA abundance.

(A) Mean particle size, (B) particle counts, and (C) protein concentration of plasma-derived EVs collected from healthy controls (HC), non-severe asthmatics (NSA), and severe asthmatics (SA). (E) Volcano plots and bar graphs showing pairwise comparisons of miRNA abundance (|log2(FC)| > 1.5 and P < 0.05). Upregulated miRNAs are shown in red, and downregulated miRNAs in blue. (F) Network visualization and pathway enrichment analysis (KEGG and Reactome) of predicted miRNA-mRNA targets using the STRING database. Both upregulated and downregulated target mRNAs from each pairwise comparison were included. The top 10 significantly enriched pathways from KEGG and Reactome are presented as a column graph. Data are shown as mean ± SEM. ** P < 0.01 vs. HC; ^# P < 0.05 vs. NSA.

Previous studies have reported differential abundant miRNA profiles across various asthma phenotypes^3,4. In this study, small RNA sequencing and differential abundance analysis identified several miRNAs that were significantly abundant in pairwise comparison between asthma groups (Figure 1E). To evaluate variation in miRNA abundance profiles across study groups, two separate principal component analyses (PCA) were performed. The first PCA included normalized abundance of all miRNAs to assess global patterns. The second PCA focused only on significantly differentially abundant miRNAs to evaluate their discriminatory power between groups (Figure S3). Additionally, heatmap analyses were generated to visualize normalized miRNA abundance across the groups. One heatmap included all detected miRNAs, providing an overview of global abundance trends, while the second focused on significantly differentially abundant miRNAs grouped by clinical status (Figure S4).

Prior research has shown that miRNA profiles in BALF-derived EV from SA patients are associated with impaired lung function and inflammation⁵. Specifically, miR-205–5p, miR-9–5p, miR-610, miR-330–3p, and miR-500a-5p were previously reported to be downregulated in SA vs. HC, with miR-9–5p inversely correlating with BALF neutrophil counts⁵. In our study, these miRNAs showed similar and distinct patterns. miR-205–5p was downregulated in NSA vs. HC. miR-9–5p was downregulated in both SA vs. HC and SA vs. NSA. miR-500a-5p was downregulated in SA vs. HC. miR-610 (NSA vs. HC) and miR-330–3p (SA vs. HC) were upregulated (Figure 1E).

Venn diagram analysis revealed both unique and shared miRNAs across the pairwise comparisons, both before (Figure S5A–D) and after (Figure S6A–B) exclusion of participants on OCS and/or biologics in the SA group. After excluding SA participants receiving OCS and/or biologics, differential abundance analysis still showed miR-330–3p as upregulated in SA vs. HC. Additionally, miR-548ae-5p, miR-5100, and miR-513c-5p were downregulated in SA vs. HC. In the SA vs. NSA comparison, miR-6501, miR-1299 were upregulated, while miR-133a-3p was downregulated (Figure S6A–B).

Ingenuity pathway analysis was used to identified experimentally validated mRNA targets of differentially abundant miRNAs. Subsequent analysis using STRING database revealed strong molecular interactions among these targets in both before (Figure 1F) and after exclusion of participants on OCS and/or biologics in the SA group (Figure S6C).

Pathway enrichment analysis of these mRNA targets (Table S4) highlighted several signaling pathways relevant to asthma pathobiology. These included NSA vs. HC (MAPK, cGMP-PKG, focal adhesion, PI3K-Akt, ERBB2, and IL-4/IL-13 signaling pathways), SA vs. HC (FOXO-mediated processes) and SA vs. NSA (Th1, Th2, Th17 cell differentiation)⁶. Importantly, even after excluding SA participants receiving OCS and/or biologics, the updated differential abundance analysis continued to show strong intermolecular interactions within the STRING database network. Enriched pathways included MAPK, ERBB, EGFR, TGFβ signaling, and cellular senescence (SA vs. HC), as well as cGMP-PKG, AMPK, FOXO, PI3K-AKT, IL-4/IL-13, and receptor tyrosine kinase signaling (SA vs. NSA) (Figure S6C and Table S5).

Spearman correlation analysis revealed that miR-330–3p was negatively correlated with lung function parameters, including FEV₁% Pred, FVC% Pred, and the FEV₁/FVC ratio. In contrast, miR-515–3p, miR-4492, miR-500b-5p, and miR-5100 positively correlated with FEV₁% and FVC% Pred. Additionally, serum IgE levels showed positive correlation with miR-610 (Figure 2A–C). Further analysis showed that miR-500a-5p and miR-548ae-5p were positively correlated with FEV₁% Pred, while miR-548t-5p was negatively correlated with the FEV₁/FVC ratio (Figure 2A–C). We also identified additional miRNAs associated with clinical variables such as miR-515–3p positively correlated with asthma exacerbation. miR-330–3p positively correlated with body mass index (BMI). miR-133a-3p, and miR-9–5p positively correlated with Asthma Control Test (ACT) scores (Figure S7).

Figure 2. Correlation analysis between differentially abundant miRNAs and key clinical variables.

(A) Correlation matrix illustrating Spearman correlation coefficients (positive and negative) between differentially abundant miRNAs and clinical variables, regardless of statistical significance. (B) Subset of statistically significant correlations (P < 0.05), highlighting key associations between miRNAs and disease-relevant clinical metrics (C) Individual correlation plots showing representative associations between specific miRNAs and lung function parameters (hsa-miR-330–3p, hsa-miR-500a-5p, hsa-miR-500b-5p, hsa-miR-515–3p, hsa-miR-548ae-5p, hsa-miR-548t-5p, hsa-miR-610, hsa-miR-4492, and hsa-miR-5100), as well as serum IgE levels (hsa-miR610). Each plot includes the Spearman correlation coefficients (R) and corresponding P-value.

To assess the robustness of these associations, we repeated the Spearman correlation analysis after excluding SA participants receiving OCS and/or biologics. miR-5100 remained positively correlated with FEV₁% and FVC% Pred. miR-610 continued to show a negative correlation with the FEV₁/FVC ratio and a positive correlation with serum IgE levels. Additionally, miR-133a-3p maintained a positive correlation with ACT scores (Figure S8A–C).

Our miRNA-mRNA association analysis revealed that several miRNAs, such as miR-9–5p, miR-515–3p, and miR-133a-3p, were downregulated in SA. This downregulation is expected to result in the upregulation of their experimentally validated target genes, including Pmp22, Zfp36, Btg1, and Fscn1 (Table S4 and Figure S9A–C). To validate the expression directionality of these targets, we examined transcriptomic data from the publicly available dataset GSE158752⁷, which includes bronchial epithelial brushings from patients with SA, NSA, and HC. The expression patterns of these genes in GSE158752 were consistent with our predictions, supporting the regulatory relationship between identified miRNAs and their targets.

In vitro studies with BEAS-2B human bronchial epithelial cells demonstrated significant upregulation of miR-513a-5p, miR-548t-5p, miR-330–3p, and miR-5189–5p following house dust mite (HDM) treatment, while miR-548b-5p was downregulated. These findings support the miRNA expression pattern observed in our asthma cohort. Interestingly, total RNA from BEAS-2B-derived EVs exhibited opposing trends. For example, miR-5100 upregulated in EVs post-HDM treatment, despite being downregulated in plasma-derived EVs from asthmatics (Figure S10A–B). These results suggest that altered plasma EV miRNA signatures in asthma may reflect contributions from multiple cell types, including immune cells (e.g., lymphocytes, dendritic cells) and structural cells (e.g., epithelial and endothelial cells), highlighting the complexity of EV origin and regulation in asthma pathobiology^3,8–10.

This study is limited by a relatively small cohort size, lack of an independent validation cohort, and potential seasonal or temporal variability in sample collection. Higher BMI in the severe asthma group may have influenced miRNA profiles, but small sample size prevented exclusion of high-BMI individuals. Plasma volume variability due to unadjusted hematocrit levels may have affected EV quantification. Although a secondary analysis excluded participants on biologics and/or oral corticosteroids (OCS), treatment effects, especially from ICS/OCS/biologics, cannot be fully ruled out. In vitro findings using BEAS-2B cells treated with HDM may not fully replicate the complex airway environment, and the HDM preparation used contained high endotoxin levels, which could influence results.

This study identifies plasma-derived EV miRNAs that are associated with asthma severity. The distinct miRNAs (e.g., miR-133a-3p, miR-330–3p, miR-610, miR-513c-5p, miR-548ae-5p, miR-1299, miR-5100, and miR-6501–3p) are associated with asthma severity and may be relevant to molecular mechanisms underlying asthma. Among the differentially abundant EV-associated miRNAs identified across the three pairwise comparisons, only five miRNAs (miR-9–5p, miR-133a-3p, miR-205–5p, miR-500a-5p, and miR-548b-5p) have previously been reported in circulating plasma or serum in various disease contexts such as non-small cell lung cancer, myocardial infarction, coronary artery disease, thoracic aortic aneurysm, astrocytoma, glioblastoma, systemic lupus erythematosus, and Hashimoto’s disease (Figure S11). The remaining EV-associated miRNAs identified in our study were not found in the plasmiR database¹¹, suggesting that plasma-derived EV miRNA profiles are at least partially distinct from free circulating miRNAs. These findings highlight the potential of EV-associated miRNAs to provide unique insights into asthma pathophysiology, particularly in differentiating NSA and SA from HC. Future studies will be needed to directly compare circulating vs. EV-derived miRNA profiles in plasma to validate their respective diagnostic and prognostic relevance in asthma severity. While the relatively small cohort size underscores the need for validation in a larger, independent population and further mechanistic studies, these findings lay a strong foundation for the future clinical application of circulating and EV miRNAs in personalized asthma care.

Data availability statement

All data necessary to evaluate the conclusions of this manuscript are provided in the main text and Supporting Information (Tables S1–S5 and Figures S1–S11). The raw datasets supporting this study will be available in the NCBI Gene Expression Omnibus repository, GEO Series accession number: GSE280322.