Methods: The Local Ethics Committee approved the study. Induced sputum sampled outside an exacerbation was processed by dialysis (D-Tube Dialyzers Midi, Novagen) followed by ultrafiltration (Amicon Ultra 3K, Millipore, 3000 NMWL). Sputum cytokines were measured with the 27 Plex Human Cytokine Group I Bioplex (Biorad). Hierarchical cluster analysis with a wide array of variables was used. Difference between clusters was assesed by One-way Anova.
Results. 60 adult asthmatics, mean age 44±14 years old, 52% females were included. Three clusters were separated, significantly different for all sputum cytokines (except RANTES and FGF-b), asthma severity (GINA), onset of asthma, ACT score, lung function (LF), AHR, exacerbation frequency, near-fatal asthma (NFA), corticosteroid-resistance, blood eosinophils. Cluster 1 (n=27, 67% males) had the lowest frequency of asthma exacerbations, the best LF, predominant macrophages and lymphocytes in induced sputum and the highest sputum IL-4, IL-1RA, VEGF and IL-7. Cluster 2 (n=14, 64.3% females) had the longest asthma duration, the highest AHR and frequency of asthma exacerbations, corticosteroid-resistance and NFA, the worst LF and ACT score, the highest sputum and blood eosinophilia and the highest sputum cytokines. Cluster 3 (n=19, 63.2% females), had the shortest asthma duration, the highest FeNO and total serum IgE, the highest incidence of LF decline and atopy, sputum neutrophilia and the lowest values for all sputum cytokines.
Conclusion: Deep phenotyping incorporating phenotypic traits, longitudinal data and biomarkers of inflammation stratifies asthma into subclasses according to their biological basis.