Bronchoalveolar lavage cytokine patterns in children with severe neutrophilic and paucigranulocytic asthma

Abstract

Background: Asthma is a complex heterogeneous disease occurring in adults and children that is characterized by distinct inflammatory patterns. While numerous studies have been performed in adults, little is known regarding the heterogeneity of severe asthma in children, particularly inflammatory signatures involving the air spaces.

Objective: We sought to determine the relationship of bronchoalveolar lavage (BAL) cytokine/chemokine expression patterns in children with severe therapy-resistant asthma stratified according to neutrophilic versus nonneutrophilic BAL inflammatory cell patterns.

Methods: Children with severe asthma with inadequate symptom control despite therapy underwent diagnostic bronchoscopy and BAL. Inflammatory cytokine/chemokine concentrations were determined using a multiplex protein bead assay.

Results: Analysis of BAL constituents with an unbiased clustering approach revealed distinct cytokine/chemokine patterns, and these aligned with pathways associated with type 2 innate lymphoid cells, monocytes, neutrophil trafficking, and T effector cells. All cytokines examined (n = 27) with 1 exception (vascular endothelial growth factor) were overexpressed with BAL neutrophilia compared with nonneutrophilic asthma, and this was confirmed in a cross-validation analysis. Cytokines specifically responsible for T_h17 (IL-17, IL-6, G-CSF) and T_h1 differentiation and expression (IL-12, TNF-α, IFN-γ) were enhanced in the neutrophilic cohorts. Neutrophilic groups were also characterized by higher prevalence of bacterial and viral pathogens; however, cytokine expression patterns manifested independently of pathogen expression.

Conclusions: The results demonstrate that children with refractory asthma and neutrophilic inflammation had a BAL cytokine pattern consistent with a mixed T_h17/T_h1/T_h2 response. In contrast, nonneutrophilic asthma presented independently of cytokine overexpression.

Keywords: BAL granulocytes; Severe asthma; asthma heterogeneity; asthma phenotypes; chemokine; cytokine; eosinophilic asthma; neutrophilic asthma; rhinovirus.

Figure 1.

Empirical cytokine/chemokine distributions stratified by granulocytic (neutrophil/pauci-granulocytic) phenotype (p values for all cytokines are ≤ 0.0005 (Table I).

Figure 1.

Empirical cytokine/chemokine distributions stratified by granulocytic (neutrophil/pauci-granulocytic) phenotype (p values for all cytokines are ≤ 0.0005 (Table I).

Figure 2.

Cross-validation estimates for accuracy (A) in classifying patients as neutrophilic or pauci-granulocytic. Proteins are ranked in descending ordered from the classifier with highest classification A to the classifier with the lowest A. Vertical lines identify the 95% confidence interval.

Figure 3.

Ranking of cytokines/chemokines as classifier of granulocytic (neutrophilic/pauci-granulocytic) status. Proteins are ranked in descending ordered from the classifier with highest Youden Index to the classifier with the lowest Youden Index. Note that the Youden Index is a dimensionless number ranging from 0 to 1, where the Youden Index = classification sensitivity + classification specificity − 1. The greater the Youden Index the smaller the overall classification error rate.

Figure 4.

Empirical CXCL8 distributions based on presence or absence of bacterial infection detected in BAL (differences are not significant).

Figure 5.

Hierarchical clustering of the cytokine/chemokine variables based on Spearman rank correlation.