Multidimensional endotyping using nasal proteomics predicts molecular phenotypes in the asthmatic airways

Area of Science:

• Biomarker discovery
• Asthma research
• Machine learning in medicine

Background:

• Unsupervised clustering of nasal fluid biomarkers is underexplored for asthma endotyping.
• Identifying distinct asthma patient clusters with specific lower airway molecular phenotypes is crucial.

Purpose of the Study:

• To evaluate if nasal fluid protein expression can identify distinct asthma patient clusters.
• To correlate these clusters with lower airway molecular phenotypes.

Main Methods:

• Applied unsupervised clustering and Shapley values to 168 nasal proteins in 43 severe asthma patients.
• Utilized a machine learning approach on nasal brushings transcriptomics from U-BIOPRED.
• Performed feature reduction, functional gene analysis, and gene set variation analysis.

Main Results:

• Identified 2 severe asthma endotypes using nasal protein data, replicated in transcriptomics.
• Cluster 1 showed significant airway obstruction, small airways disease, and oxidative stress.
• Proteomic and transcriptomic pathways linked to T_H1, T_H2, neutrophil, and inflammatory signaling.

Conclusions:

• Nasal protein and gene analysis can indicate lower airway molecular phenotypes in asthma.
• This approach may lead to a simple, non-invasive test for non-type 2 immune response asthma.