Abstract
BACKGROUND: Allergic rhinitis (AR), as a prevailing disease worldwide, is a common chronic inflammatory disease of the upper airway; however, its prognosis did not meet expectations of the general public. Regarding the importance of treatment for AR, biomarkers with high sensitivity and precision are in urgent need. METHODS: To detect the potential mechanisms associated with the AR pathology, the differentially expressed gene data from the clinical dataset GSE19187 and GSE46171 were used for weighted gene co-expression network analysis. Additionally, 5 machine-learning algorithms were integrated to screen for novel diagnostic biomarkers for AR. RESULTS: This analysis identified 133 genes enriched in the black and purple module pathways from weighted gene co-expression network analysis (WGCNA). The 4 key diagnostic genes-CST1, CST2, SERPINB4, and TOX-were identified through machine learning algorithms and validated for their diagnostic potential with high area under the curve (AUC) values in external datasets. Immune infiltration analysis indicated a higher proportion of resting mast cells in AR samples compared to controls, with CST1, CST2, and SERPINB4 showing high expression in these cells. Real-time PCR validation confirmed that CST1, CST2, and SERPINB4 were upregulated in AR samples while TOX was downregulated. CONCLUSION: Our study identifies critical molecular markers and immune cell alterations associated with AR, providing insights into its pathogenesis and potential diagnostic biomarkers for clinical application. Future research should focus on validating these findings in larger cohorts to facilitate the development of targeted therapies for AR.