Cytokine Profiling In Severe Asthma Subphenotypes Using Factor And Cluster Analysis

Understanding the phenotypic heterogeneity of asthma is likely to shed light upon its immunopathogenesis. Introduction: To examine patterns of cytokine expression in different clinical phenotypes of severe asthma. Rationale: We performed k-means clustering on demographic, lung function, allergen testing and sputum induction data of 164 patients Methods: attending our Difficult Asthma Clinic. Sputum supernatant was analysed for 23 mediators using Meso Scale Discovery platform. The data obtained was reduced using principal component analysis. Factorial patterns of mediator expression were compared across clinical phenotypes. Using unbiased data reduction tools we were able to identify severe asthma clinical sub-phenotypes. Cluster analysis identified 4 Results: clinical phenotypes: A. Discordant high symptoms/non-eosinophilic, obese, normal FEV ‚; B. Late onset, concordant 1 symptoms/eosinophilia, normal FEV ‚; C. Early onset, discordant low symptoms/eosinophilia, low FEV ; D. Early onset, concordant 1 1 symptoms/eosinophilia, normal FEV ‚. The individual mediators did not differ significantly across clusters. Four cytokine factors were 1 identified 1. IL6R, IL8, TNFRI; 2. CXCL11, IL6, CCL5; 3. CCL26, IL13, IL5 and; 4. IL10, IL17, IL4. The factors loaded onto the clusters as shown in the figure below where data are shown as the mean factor loading ± SEM. Graph showing cytokine factor loading across clinical clusters