PT  - JOURNAL ARTICLE
AU  - Hiroki Hayashi
AU  - Takeo Ishii
AU  - Nariaki Kokuho
AU  - Misuzu Kurahara
AU  - Takashi Motegi
AU  - Kumiko Hattori
AU  - Kouichi Yamada
AU  - Koichiro Kamio
AU  - Akihiko Gemma
AU  - Kozui Kida
TI  - Cluster analysis revealed differences on quality of life and susceptibility to exacerbation between subpopulations of smokers including COPD
DP  - 2011 Sep 01
TA  - European Respiratory Journal
PG  - p3566
VI  - 38
IP  - Suppl 55
4099  - http://erj.ersjournals.com/content/38/Suppl_55/p3566.short
4100  - http://erj.ersjournals.com/content/38/Suppl_55/p3566.full
SO  - Eur Respir J2011 Sep 01; 38
AB  - Background: It is necessary to categorize subpopulations of COPD or smokers by non-outcome phenotypes, such as emphysema without airflow obstruction.Methods: 730 current- or ex-smokers including 382 COPD subjects were studied (66.7±11.0 yrs, 602male/128 females). We collected the data for all the subjects on pulmonary function test, 6 minute walking test (6MWT), body mass index (BMI), dysnpea (modified Medical Research Council (MMRC) Dyspnea Scale, Oxygen Cost Diagram (OCD)), and the extent of emphysema and airway disease assessed by chest computed tomography (CT) (low attenuation area (LAA)% and wall area (WA)%), and we also studied the data on the score of St. George's Respiratory Questionnaire (SGRQ) for QOL [n = 361], and exacerbations [n = 178]. We performed a principal component analysis (PCA) and cluster analysis by k-means method.Results: PCA showed the major factors as follows: vital capacity (VC), LAA%, WA%, reversibility, PaO2, PaCO2, leg fatigue on 6MWT.Cluster analysis with these factors classified the subjects into four clusters as follows: Cluster 1; 254 cases with mild emphysema (LAA% 22.1±10.8, WA% 45.0±14.1), Cluster 2; 156 cases with airway disease (LAA% 15.7±11.3, WA% 58.3±12.8), Cluster 3; 152 cases with emphysema and airway disease (LAA% 20.1±13.2, WA% 61.5±11.5), Cluster 4; 168 cases with severe emphysema (LAA% 41.0±11.8, WA% 57.0±11.3).Cluster 4 has the highest SGRQ score (p < 0.0001). Cluster 2 and 4 were more prone to exacerbations (p < 0.01).Conclusions: PCA and cluster analysis revealed that chest CT contributes to the classification to subpopulations in smoker with or without COPD.