Clinical Studies In COPD Phenotype By Principal Component Analysis And Cluster Analysis

Objective：Nowadays, it is widely recognized that COPD is a complex syndromecharacterized by numerous pulmonary and extrapulmonary manifestations. As a result,there is consensus that FEV1by itself does not adequately describe the complexity ofthe disease and that FEV1cannot be used in isolation for the optimal diagnosis,assessment, and management of the disease.Methods：COPD subjects recruited in the second affiliated hospital. Principalcomponent analysis (PCA) was performed using eight variablesselected for theirrelevance to COPD: age, cumulative smoking, forced expiratory volume in1s(FEV1)(%predicted), dyspnoea (modified Medical ResearchCouncil scale), comorbidities,IL-8,IL-32,TNF-α. Patient classification was performed using clusteranalysis basedon PCA-transformed data.Results：124COPD subjects were analysed:77.4%were male; median (interquartilerange) age was72.0(64.0–73.75) yrs; FEV1was38.04(34.1–66.3)%pred; and10,28,34and52subjects were classified in Global Initiative for Chronic Obstructive LungDisease (GOLD) stages1,2,3and4,respectively. PCA showed that threeindependent components accounted for73.1%of variance.PCA-based cluster analysisresulted in the classification of subjects into four clinical phenotypes。Conclusion：1、Using principal component analysis and cluster analysis may be able to identify theclinical phenotype that could not be identified by GOLD classification.2、Subjects with comparable airflow limitation (FEV1) belonged to differentphenotypes and had marked differences in smoking index,symptoms, comorbidities。...