Elastic parametric response mapping: quantitative CT scoring for local COPD severity.

Abstract

Current quantitative chest CT techniques improve chronic obstructive pulmonary disease (COPD) phenotyping but do not capture spatial variability and potentially reversible disease in local lung parenchyma. Applying elastic principal graphing to CT scans from Genetic Epidemiology of COPD study participants (age 45-80 years; ≥10 pack-years), we developed elastic parametric response mapping (ePRM), a tiered scoring system (tiers 0-3 and tier Op, ie, lung opacities) that classifies lung subvolumes based on their relative composition of normal lung, emphysema, small airways disease and parenchymal disease. For 3631 participants with longitudinal data, we evaluated how relative tier assignment and mean tier position of subvolumes changed over 5 years and how they associated with forced expiratory volume in 1 s (FEV₁) change. We stratified analyses by baseline spirometry: no airflow obstruction, Global Initiative for Chronic Obstructive Lung Disease (GOLD) 1-2 and GOLD 3-4. The proportion of tier 0 subvolumes decreased with worsening airflow obstruction, while tier 2 and 3 proportions increased. Tier 1 proportions were similar in GOLD 1-2 (25.7%) and GOLD 3-4 (28.1%), with over half of subvolumes remaining in tier 1 or reverting to tier 0 at year 5. In contrast to tiers 0 and 2, baseline mean tier 1 position was strongly predictive of reassignment to more advanced tiers at year 5 in participants without airflow obstruction, GOLD 1-2 and GOLD 3-4 (area under the curves (95% CIs) 0.86 (0.85 to 0.87), 0.90 (0.89 to 0.91) and 0.92 (0.90 to 0.93), respectively). A higher per cent volume of lung retained in tier 1 was associated with less FEV₁ decline in all groups. CT ePRM categorises local lung tissue into distinct and potentially reversible tiers of disease severity.