Cluster Analysis to Identify Distinct Asthma Phenotypes in the ATLANTIS Cohort

Background: Previous cluster analyses have identified subgroups of asthma. However, only a few studies included parameters of small airways dysfunction (SAD), or gene expression profiles reflecting underlying disease mechanisms. We aimed to identify clinically distinct asthma phenotypes, beyond GINA asthma severity, using available data from the ATLANTIS study which focused on identifying the prevalence of SAD in asthma and its role in asthma control, exacerbations and quality of life. Methods: The ATLANTIS study included 773 asthma patients (mean age 44 years, 58% female, 76% never-smoker, GINA 1-5). Subjects were extensively characterized, including symptoms, parameters of large and small airways dysfunction, blood and sputum differential cell counts, and genome-wide gene expression profiling from nasal brushes. Clusters were generated using the Self-Organizing Map-Ward's method. Results: Four distinct clusters were identified: A (N = 62; 8%) characterized by the most frequent exacerbations, lower post-bronchodilator FEV1 % predicted, more small airways dysfunction, higher sputum and blood eosinophils, and high expression of asthma-related genes. B (N = 206; 27%) consisting of atopic patients with early-onset asthma, uncontrolled symptoms, and normal lung function and bronchial hyperresponsiveness, along with a high expression of asthma-related genes in the nasal epithelium. C (N = 277; 36%), predominantly male former smokers, with well-controlled asthma, mild obstructive lung disease, and relatively high neutrophil levels. D (N = 228; 29%), with normal lung function and low blood and sputum eosinophils. Conclusions: Four distinct clusters were identified, where the presence of SAD was associated with high type-2 inflammation, lower lung function, and frequent exacerbations. SAD may be a marker of poorly controlled asthma and should be considered as an important clinical trait.