Chest CT imaging phenotypes may explain COPD heterogeneity

CT measurements may support patient satisfaction along the continuum of COPD phenotypes, and imaging phenotypes could explain a proportion of COPD heterogeneity, researchers reported in Annals of the American Thoracic Society.

Researchers aimed to identify the primary variables that drive COPD heterogeneity by using principal component analysis. They defined disease axes and evaluated the prognostic value of the axes across COPD progression, exacerbation and mortality.

The study included 7,331 patients with COPD aged 39 to 85 years (45.8% female smokers; 40.3% Black) from the COPDGene Phase 1 cohort conducted from 2008 to 2011. All participants were current or former smokers across all COPD stages with a minimum of 10 pack-year smoking history (mean, 44.6 pack-years).

From a total of 916 phenotypes, researchers selected 147 continuous clinical, spirometric and CT features. For each principal component, researchers computed a principal component score based on feature weights and used these distributions to define disease axes and divide patients into quartiles.

“Leveraging the broadest range of imaging and clinical characteristics available in the COPDGene cohort, we observed that the top principal components that emerged from our unsupervised analysis were predominantly related to CT imaging and spirometric phenotypes,” Nancy F. Yuan, MS, of the department of biomedical informatics at the University of California, San Diego, and colleagues wrote.

The first principal component, which accounted for 43.7% of variance, was defined by CT measures of air trapping and emphysema. The second principal component, which accounted for 13.7% of variance, was defined by spirometric and CT vital capacity and lung volume measures. The third principal component, which accounted for 7.9% of variance, was defined by CT measures of lung mass, airway thickening and body habitus.

“We used these components to create disease axes. Each of these axes also individually demonstrate prognostic potential of their own, predictive of future FEV₁ decline, exacerbation and mortality,” the researchers wrote.

When patients were stratified across each disease axis, there were 3.2-fold greater odds of 5-year COPD progression, 5.4-fold greater odds of 5-year COPD exacerbation and a fivefold greater odds of 10-year mortality between those in the highest and lowest quartiles.

“This work shows that imaging phenotypes explain a significant proportion of COPD heterogeneity,” the researchers wrote. “Including CT measurements in COPD patient management may assist in identifying those most in need of therapeutic interventions early in their disease trajectory.”