Novel Treatment-Refractory Preschool Wheeze Phenotypes Identified by Cluster Analysis of Lung Lavage Constituents.

BACKGROUND

Preschool children with treatment-refractory wheeze often require unscheduled acute care. Current guidelines advise treatment of persistent wheeze with inhaled corticosteroids. Alternative treatments targeting structural abnormalities and specific inflammatory patterns could be more effective.

OBJECTIVE

To apply unsupervised analysis of lung lavage (bronchoalveolar lavage [BAL]) variables to identify clusters of preschool children with treatment-refractory wheeze.

METHODS

A total of 155 children 6 years or younger underwent bronchoscopy with BAL for evaluation of airway structure, inflammatory markers, and pathogens. Variables were screened with factor analysis and sorted into clusters by Ward's method, and membership was confirmed by discriminant analysis.

RESULTS

The model was repeatable in a 48-case validation sample and accurately classified 86% of cases. Cluster 1 (n = 60) had early-onset wheeze, 85% with structural abnormalities, mostly tracheamalacia, with low total IgE and agranulocytic BAL. Cluster 2 (n = 42) had later-onset wheeze, the highest prevalence of gastroesophageal reflux, little atopy, and two-third had increased BAL lipid-laden macrophages. Cluster 3 (n = 46) had mid-onset wheeze, low total IgE, and two-third had BAL viral transcripts, predominately human rhinovirus, with BAL neutrophilia. Cluster 4 (n = 7) was older, with high total IgE, blood eosinophilia, and mixed BAL eosinophils and neutrophils.

CONCLUSIONS

Preschool children with recurrent wheeze refractory to inhaled corticosteroid treatment include 4 clusters: airway malacia, gastroesophageal reflux, indolent human rhinovirus bronchoalveolitis, and type-2 inflammation. The results support the risk and cost of invasive bronchoscopy to diagnose causes of treatment-refractory wheeze and develop novel therapies targeting airway malacia, human rhinovirus infection, and BAL neutrophilia in preschool children.