Adult asthma phenotypes identified by a cluster analysis on clinical and biological characteristics

Abstract

A recent approach to tackle airway diseases recommends deconstructing them into identifiable and treatable traits [Pavord I, Lancet 2018]. We aimed at identifying adult asthma phenotypes using a cluster analysis, integrating for the first time inflammatory and oxidative stress-related biomarkers. We studied 318 adults with current asthma from the first follow-up of the Epidemiological study on the Genetics and Environment of Asthma (EGEA2). We considered 25 classical personal and asthma characteristics covering respiratory symptoms, asthma exacerbation and treatment, lung function, allergic sensitization, and added blood eosinophil and neutrophil counts, and fluorescent oxidation products (FlOPs) level. For cluster analysis, we used the mixture approach relying on the Expectation-Maximization algorithm which accommodates continuous and categorical variables. We identified 3 clusters. Cluster 1: 43 asthmatics (55 years old) predominantly with adult-onset asthma, poor lung function, treatments, cough and phlegm, exacerbations, high neutrophil count and high FlOPs level. Cluster 2: 178 asthmatics (41 years old) predominantly with paucigranulocytic asthma, normal lung function, rhinitis and low IgE level. Cluster 3: 97 asthmatics (27 years old) predominantly men with childhood-onset asthma, eosinophilic asthma, allergic comorbidities, and high IgE level. The clusters differed for other biomarkers: exhaled breath condensate 8-isoprostanes, blood leptin, IL-5, IL-8, IL-10, and C-reactive protein. Our clustering approach integrating biomarkers allowed us to identify 3 asthma clusters characterized by distinct traits. Their association with long term evolution of asthma is now under study.