Biomarkers to identify tissue eosinophils in wheezing children

Background: Eosinophilic airway inflammation is a key component of asthma. Biomarkers that could predict airway eosinophils are needed for optimal asthma phenotyping and management, particularly in the era of biologics. Aim: We compared the accuracy of 4 potential biomarkers (blood and BAL eosinophils, BAL ECP, serum IgE) to detect tissue eosinophils. Methods: We studied 103 children (58 wheezing/45no-wheezing) undergoing a clinically indicated bronchoscopy and bronchial biopsy. Tissue eosinophilia (>23 eosinophils/mm2) was present in 58/103 children (56%). Associations were evaluated by Spearman Rank; accuracy determined by ROC analyses. Results: In the whole cohort, tissue eosinophils were positively correlated with blood eosinophils (p=0.006;r=0.27), serum IgE (p=0.005;r=0.28), BAL eosinophils (p=0.0009;r=0.34), BAL-ECP (p=0.03;r=0.22); yet correlation coefficients were low. A multivariate analysis showed that among the biomarkers only BAL ECP could predict tissue eosinophils (O.R. 2.6, p=0.009). On ROC analyses serum IgE, blood eosinophils and BAL-ECP had only modest predictive accuracy (serum IgE: AUC 0.63,p=0.02; blood eosinophils: AUC 0.64,p=0.009; BAL-ECP: AUC 0.62; p=0.03). When the analyses were limited to wheezing children only (n=58), none of the 4 biomarkers predicted tissue eosinophils. For similar levels of tissue eosinophils, blood eosinophils, BAL eosinophils, BAL ECP were prominently increased in the atopic phenotype. Conclusions: Blood eosinophils, serum IgE, BAL eosinophils, BAL ECP are characteristic components of asthma pathogenesis, yet they are not accurate biomarkers to predict tissue eosinophils. Understanding their individual roles in the distinct asthma phenotypes could improve clinical management.