What this research found
Asthma is treated as one disease but behaves like several. Nasal airway epithelium RNA sequencing from 679 people — 433 with asthma and 246 controls — was used to sort patients by the molecular pathway apparently driving their disease. Type 2 (T2) inflammation scores split the cohort 44% T2-high to 56% T2-low, and unsupervised clustering then divided the T2-low majority into an immune-active and an immune-quiet endotype, each mapped to a different class of therapy.
- Of 23,651 genes tested, 147 were differentially expressed between asthma and control at a false discovery rate below 0.05 with an absolute log2 fold change above 0.5, split 90 up-regulated and 57 down-regulated. A total of 2,963 genes cleared the false discovery rate threshold on its own.
- The strongest up-regulated genes were CST1 at log2 fold change 2.18 (FDR 8.60e-05), CLCA1 at 1.86 (FDR 2.84e-04) and FETUB at 1.72 (FDR 6.73e-05), with the mast cell markers TPSAB1 and CPA3 also elevated.
- Scoring T2 pathway activation split the cohort into 299 T2-high patients (44.0%) and 380 T2-low patients (56.0%), close to the roughly 50% of patients who respond to T2-targeted biologics in clinical practice.
- K-means clustering divided the T2-low group in two: 200 immune-active patients (29.5% of the cohort) with high MHC class II antigen presentation and interferon signalling, and 180 immune-quiet patients (26.5%) with minimal inflammatory expression. The silhouette score for that split was 0.5434.
- Of 50 pathways enriched at a false discovery rate below 0.05, MHC class II protein complex assembly ranked first (GO:0002399, 5 of 14 genes overlapping, FDR 1.44e-05), and 10 were targetable by approved or investigational drugs. The KEGG asthma pathway itself came out enriched, which serves as a check on the method.
How it was done
Nasal airway epithelium RNA sequencing for 679 samples — 433 asthma and 246 control — was retrieved from the Gene Expression Omnibus, leaving 23,651 genes after quality filtering. Principal component analysis and hierarchical clustering characterised the variance structure, then differential expression testing with false discovery rate correction produced the asthma signature. Enrichr was used to test that gene set against the GO Biological Process, KEGG and Reactome libraries. Composite activation scores for Type 2 immunity pathways separated patients into T2-high and T2-low groups, and K-means clustering with silhouette-based selection subdivided the heterogeneous T2-low population. Each resulting group was then matched to candidate drug classes, and the analysis was delivered as an 18-slide presentation built around the analysis figures.
Data sources
- Gene Expression Omnibus accession GSE152004 — nasal airway epithelium RNA sequencing, 679 samples (433 asthma, 246 control), 23,651 genes after filtering
- Enrichr — GO Biological Process, KEGG and Reactome pathway libraries
Limitations
The design is cross-sectional and the dataset carries no treatment outcome data, so the therapies proposed for each endotype are hypotheses awaiting prospective trials. Cohort demographics are not fully described and the T2-high versus T2-low cut-off was data-driven, so both may need adjustment before clinical use.
Figures from this analysis
How this research was produced
K-Dense Web planned and ran this genomics investigation end to end — gathering the sources, carrying out the analysis, producing the figures, and drafting the report. The full session transcript, including every intermediate step, is available to view.
