Identification of Asthma Phenotypes in the Spanish MEGA Cohort Study Using Cluster Analysis
Introduction: The definition of asthma phenotypes has not been fully established, neither there are cluster studies showing homogeneous results to solidly establish clear phenotypes. The purpose of this study was to develop a classification algorithm based on unsupervised cluster analysis, identifying clusters that represent clinically relevant asthma phenotypes that may share asthma-related outcomes.
Methods: We performed a multicentre prospective cohort study, including adult patients with asthma (N=512) from the MEGA study (Mechanisms underlying the Genesis and evolution of Asthma). A standardised clinical history was completed for each patient. Cluster analysis was performed using the kernel k-groups algorithm.
Results: Four clusters were identified. Cluster 1 (31.5% of subjects) includes adult-onset atopic patients with better lung function, lower BMI, good asthma control, low ICS dose, and few exacerbations. Cluster 2 (23.6%) is made of adolescent-onset atopic asthma patients with normal lung function, but low adherence to treatment (59% well-controlled) and smokers (48%). Cluster 3 (17.1%) includes adult-onset patients, mostly severe non-atopic, with overweight, the worse lung function and asthma control, and receiving combination of treatments. Cluster 4 (26.7%) consists of the elderly-onset patients, mostly female, atopic (64%), with high BMI and normal lung function, prevalence of smokers and comorbidities.
Conclusion: We defined four phenotypes of asthma using unsupervised cluster analysis. These clusters are clinically relevant and differ from each other as regards FEV1, age of onset, age, BMI, atopy, asthma severity, exacerbations, control, social class, smoking and nasal polyps.
Keywords: Asthma; Asthma endotypes; Asthma phenotypes; Clustering analysis.
keywords:
Publication: Article
1706096678250
January 24, 2024
/research/publications/identification-of-asthma-phenotypes-in-the-spanish-mega-cohort-study-using-cluster-analysis
Introduction: The definition of asthma phenotypes has not been fully established, neither there are cluster studies showing homogeneous results to solidly establish clear phenotypes. The purpose of this study was to develop a classification algorithm based on unsupervised cluster analysis, identifying clusters that represent clinically relevant asthma phenotypes that may share asthma-related outcomes.
Methods: We performed a multicentre prospective cohort study, including adult patients with asthma (N=512) from the MEGA study (Mechanisms underlying the Genesis and evolution of Asthma). A standardised clinical history was completed for each patient. Cluster analysis was performed using the kernel k-groups algorithm.
Results: Four clusters were identified. Cluster 1 (31.5% of subjects) includes adult-onset atopic patients with better lung function, lower BMI, good asthma control, low ICS dose, and few exacerbations. Cluster 2 (23.6%) is made of adolescent-onset atopic asthma patients with normal lung function, but low adherence to treatment (59% well-controlled) and smokers (48%). Cluster 3 (17.1%) includes adult-onset patients, mostly severe non-atopic, with overweight, the worse lung function and asthma control, and receiving combination of treatments. Cluster 4 (26.7%) consists of the elderly-onset patients, mostly female, atopic (64%), with high BMI and normal lung function, prevalence of smokers and comorbidities.
Conclusion: We defined four phenotypes of asthma using unsupervised cluster analysis. These clusters are clinically relevant and differ from each other as regards FEV1, age of onset, age, BMI, atopy, asthma severity, exacerbations, control, social class, smoking and nasal polyps.
Keywords: Asthma; Asthma endotypes; Asthma phenotypes; Clustering analysis. - Marcos Matabuena,Francisco Javier Salgado,Juan José Nieto-Fontarigo,María J Álvarez-Puebla,Ebymar Arismendi,Pilar Barranco (...) Francisco Javier González-Barcala
publications_en