Significance of Low-Attenuation Cluster Analysis on Quantitative CT in the Evaluation of Chronic Obstructive Pulmonary Disease

OBJECTIVE: To assess clinical feasibility of low-attenuation cluster analysis in evaluation of chronic obstructive pulmonary disease (COPD). MATERIALS AND METHODS: Subjects were 199 current and former cigarette smokers that underwent CT for quantification of COPD and had physiological measurements. Quantitative CT (QCT) measurements included low-attenuation area percent (LAA%) (voxels ≤ −950 Hounsfield unit [HU]), and two-dimensional (2D) and three-dimensional D values of cluster analysis at three different thresholds of CT value (−856, −910, and −950 HU). Correlation coefficients between QCT measurements and physiological indices were calculated. Multivariable analyses for percentage of predicted forced expiratory volume at one second (%FEV1) was performed including sex, age, body mass index, LAA%, and D value had the highest correlation coefficient with %FEV1 as independent variables. These analyses were conducted in subjects including those with mild COPD (global initiative of chronic obstructive lung disease stage = 0–II). RESULTS: LAA% had a higher correlation coefficient (-0.549, p < 0.001) with %FEV1 than D values in subjects while 2D D(−910HU) (−0.350, p < 0.001) revealed slightly higher correlation coefficient than LAA% (−0.343, p < 0.001) in subjects with mild COPD. Multivariable analyses revealed that LAA% and 2D D value(−910HU) were significant independent predictors of %FEV1 in subjects and that only 2D D value(−910HU) revealed a marginal p value (0.05) among independent variables in subjects with mild COPD. CONCLUSION: Low-attenuation cluster analysis provides incremental information regarding physiologic severity of COPD, independent of LAA%, especially with mild COPD.