Article Text
PDF
Abstract
Background Severe asthma is a heterogeneous disease. Defining its phenotypic heterogeneity is likely to shed light upon its immunopathogenesis and direct therapy. We sought to determine the relationship between phenotypes of severe asthma and sputum mediator profiles.
Methods Subjects were recruited from a Difficult Asthma Clinic at a single centre (n=164) and assessments of lung function, atopic status, asthma control and sputum induction were undertaken. Sputum was obtained and supernatants were analysed for 23 mediators using the Meso-Scale Discovery platform. We performed k-means cluster analysis to determine clinical clusters using the baseline characteristics and sputum differential counts. The pattern of mediator expression was determined by factor analysis to identify biological factors. The biological factors were related to the clinical clusters and subjects stratified by asthma control, exacerbation frequency, treatment and sputum cell counts. The repeatability of the individual clinical characteristics and biological mediators was assessed in paired samples in 106 subjects and in three samples in 66 subjects.
Results We identified four clinical clusters and five biological factors. The biological factors were differentially expressed in subjects stratified by sputum cell counts, asthma control and exacerbation frequency, but were not significantly different across the clinical clusters. The within subject repeatability of mediators was moderate; biological factors were consistent and tracked with sputum cell counts for the repeated visits.
Conclusions Sputum mediator profiling of severe asthma revealed repeatable biological factors that were strongly associated with cellular profiles and inform our understanding of asthma phenotypes.
Relationship between clinical clusters and biological factors. The clinical clusters are plotted in two dimensions with airway inflammation on the x axis and asthma control (ACQ) on the y axis. The size of the ellipsoid represents the number of subjects within each clinical cluster. The distribution of the five biological factors for each cluster is shown as the mean (SEM) factor scores.