Proteomic profiling and machine learning for endotype prediction in chronic rhinosinusitis.

Area of Science:

• Immunology
• Proteomics
• Bioinformatics

Background:

• Chronic rhinosinusitis (CRS) is a prevalent, heterogeneous inflammatory upper airway disease affecting ~12% of the population.
• CRS is classified into subtypes: CRS without nasal polyps, CRS with nasal polyps, and nonsteroidal anti-inflammatory drug-exacerbated respiratory disease (N-ERD).

Purpose of the Study:

• To identify molecular signatures and biomarkers distinguishing CRS endotypes and controls.
• Utilized targeted proteomics, bioinformatics, and machine learning for molecular analysis.

Main Methods:

• Analyzed nasal secretions and serum from 80 patients (CRS without polyps, CRS with polyps, N-ERD, controls) using high-throughput targeted proteomics (Olink).
• Examined expression patterns of 161 (nasal) and 2677 (serum) proteins.
• Correlated protein expression with clinical data including polyp presence and smell test scores.

Main Results:

• Identified two distinct nasal secretion patterns: Type 2 inflammation markers elevated in CRS with polyps and N-ERD.
• Innate immunity proteins, including Toll-like receptor 4 signaling, decreased progressively from controls to N-ERD.
• Machine learning identified glial cell line-derived neurotrophic factor (nasal) and Charcot-Leyden crystal protein (serum) as potential nasal polyposis biomarkers.

Conclusions:

• Findings offer novel insights into CRS pathophysiology.
• Highlighted potential biomarkers for precise diagnosis and targeted treatment of CRS, particularly N-ERD.