Subtyping COPD by Using Visual and Quantitative CT Imaging Features

Abstract

Background: Multiple studies have identified COPD subtypes by using visual or quantitative evaluation of CT images. However, there has been no systematic assessment of a combined visual and quantitative CT imaging classification. We integrated visually defined patterns of emphysema with quantitative imaging features and spirometry data to produce a set of 10 nonoverlapping CT imaging subtypes, and we assessed differences between subtypes in demographic features, physiological characteristics, longitudinal disease progression, and mortality.

Methods: We evaluated 9,080 current and former smokers in the COPDGene study who had available volumetric inspiratory and expiratory CT images obtained using a standardized imaging protocol. We defined 10 discrete, nonoverlapping CT imaging subtypes: no CT imaging abnormality, paraseptal emphysema (PSE), bronchial disease, small airway disease, mild emphysema, upper lobe predominant centrilobular emphysema (CLE), lower lobe predominant CLE, diffuse CLE, visual without quantitative emphysema, and quantitative without visual emphysema. Baseline and 5-year longitudinal characteristics and mortality were compared across these CT imaging subtypes.

Results: The overall mortality differed significantly between groups (P < .01) and was highest in the 3 moderate to severe CLE groups. Subjects having quantitative but not visual emphysema and subjects with visual but not quantitative emphysema were unique groups with mild COPD, at risk for progression, and with likely different underlying mechanisms. Subjects with PSE and/or moderate to severe CLE had substantial progression of emphysema over 5 years compared with findings in subjects with no CT imaging abnormality (P < .01).

Conclusions: The combination of visual and quantitative CT imaging features reflects different underlying pathological processes in the heterogeneous COPD syndrome and provides a useful approach to reclassify types of COPD.

Trial registry: ClinicalTrials.gov; No.: NCT00608764; URL: www.clinicaltrials.gov.

Keywords: COPD; CT imaging; epidemiology; heterogeneity; subtype.

Copyright © 2019 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1

Definition of CT imaging subtypes. Ten consensus CT imaging subtypes based on a combination of visual and quantitative CT imaging assessments. CLE = centrilobular emphysema; LAA = low attenuation area; PSE = paraseptal emphysema.

Figure 2

Kaplan-Meier curves of risk for all-cause death in smokers according to CT imaging subtypes. P values were calculated with the use of a log-rank test. BD = bronchial airway disease; diff = diffuse; mod = moderate; Normal = no emphysema or airway abnormality; quant = quantitative; quant w/o visual = discordant: quantitative emphysema without visual emphysema; SAD = small airway disease; visual w/o quant = discordant: visual emphysema without quantitative emphysema. See Figure 1 legend for expansion of other abbreviations.

Figure 3

Change in FEV1 (A) and adjusted lung density (B) in each pair of CT imaging subtypes. Values in each figure panel are coefficients from linear mixed regression for each comparison pair, adjusting for age, race, sex, current smoking status, cumulative smoking intensity, BMI, and FEV1. If the corrected P value is less than .05, the box is displayed in red or blue; a gray box indicates no statistical significance. Red (losing lung function or lung density faster) and blue (losing lung function or lung density slower) in each group of columns and in each group of rows indicates levels of statistical significance. For example, those with PSE had a coefficient of −0.68 for FEV1 decline, and −9.19 for change in adjusted lung density, indicating significantly faster decline for these parameters than for those with no CT imaging abnormality. See Figures 1 and 2 legends for expansion of abbreviations.

Figure 4

Cox regression proportional hazard model analysis of all-cause mortality: comparison of discordant groups with other subtypes. Adjusted hazard ratios (solid circles) and 95% CIs (horizontal lines) for death at 5-year follow-ups comparing each of the other CT imaging subtypes with the visual only (A) and quantitative only (B) subtypes. Hazard ratios are adjusted for race, sex, age, BMI, pack-years, current smoking status, and baseline FEV1. See Figures 1 and 2 legends for expansion of abbreviations.