Cluster analysis identifies distinct clinical phenotypes with poor treatment responsiveness in asthma.

Background: Treatment responsiveness, an important consideration in asthma management is different from asthma severity Objectives: Identify clinical phenotypes based on treatment responsiveness. Methodology: Retrospective study of 113 asthmatics on guideline based (GINA) treatment, minimum 6-month follow-up. Demographic characteristics, age at disease onset, disease duration, smoking and indoor air pollution, BMI, serial lung functions and allergen sensitization were collected. Treatment responsiveness was defined based on improvement in FEV1. R version 3.4.3 was used to perform statistical analysis. Ward’s minimum-variance hierarchical clustering method was performed using an agglomerative (bottom-up) approach and a dendrogram generated. To compare differences between clusters, analysis of variance using Kruskal-Wallis test (continuous variables) and chi-square test (categorical variables) was used. Results: Cluster analysis identified 4 different clusters. Cluster 1, largest cluster (N=40), is characterized by childhood onset of disease, normal weight, equal gender distribution and very high treatment responsiveness. Cluster 2, (N=16), adults (Mean age 41.7 years), more males, disease onset in adolescence, obese and poorly treatment responsive. Cluster 3 (N= 20), elderly (Mean age 61.2 years), more males, late onset of disease (Mean 51.9 years), obese and poorly treatment responsive. Cluster 4 (N=24), adults (mean age 38.5 years), predominantly female (75%), obese, onset of disease in adulthood (mean 31. 8 years) and highly treatment responsive. Conclusion: Cluster analysis identified 2 distinct clinical phenotypes who were poorly treatment responsive.