Proteomic profiling of nasal fluids and serum for type 2 nasal polyps diagnosis

Type 2 chronic rhinosinusitis with nasal polyps (T2 CRSwNP) is often associated with severe symptoms and polyp recurrence. Machine learning (ML) framework of biomarkers derived from non-invasive samples have been less evaluated as tools for describing T2 CRSwNP. To systematically assess the predictive value of protein expression in nasal fluids (NFs) and serum for T2 CRSwNP. T2 and non-T2 CRSwNP were classified using clustering analysis of tissue biomarkers from 82 patients. The expression of 92 inflammation-related proteins was measured in NFs and serum samples from the matched patients using proximity extension assays. ML with 5-fold cross-validation was employed to develop a diagnostic model for T2 CRSwNP. Selected biomarkers were further validated using immunohistochemistry (IHC), single-cell RNA-sequencing (scRNA-seq) and RT-qPCR. After defining the T2 and non-T2 CRSwNP groups, we identified 23 dysregulated proteins in NFs and 16 in serum. Four biomarkers-Glial cell line-derived neurotrophic factor (GDNF), Monocyte Chemoattractant Protein-4 (MCP-4), Transforming Growth Factor Beta 1 (TGFB1), and Cystatin D (CST5)-were selected using Lasso regression to predict T2 CRSwNP based on NFs alone. Their expression was validated through IHC, scRNA-seq, and RT-qPCR. The predictive model achieved area under the curve values (AUC) of 0.91 for the training, 0.91 for the testing, and 0.92 for the validation datasets. GDNF and MCP-4 were identified as independent prognostic biomarkers for CRSwNP. Proteomic analysis combined with an ML framework identified inflammatory endotypes and recurrence patterns in nasal polyps, offering a simple and non-invasive approach for diagnosing T2 CRSwNP.