Upper-lobe CT imaging features improve prediction of lung function decline in COPD.

Area of Science:

• Pulmonary Medicine
• Radiology
• Medical Imaging Analysis

Background:

• Predicting lung function decline is crucial for managing chronic obstructive pulmonary disease (COPD).
• The utility of computed tomography (CT) imaging features, specifically from upper lobes versus global regions, for predicting lung function decline in individuals at risk for or with COPD remains unclear.

Purpose of the Study:

• To investigate whether prediction models for lung function decline using CT imaging-derived features from the upper lobes enhance performance compared to globally derived features.
• To evaluate these models in individuals at risk of and with COPD.

Main Methods:

• Utilized data from the Canadian Cohort Obstructive Lung Disease (CanCOLD) retrospective study.
• Extracted 103 CT features globally and regionally, combined with 12 clinical features (demographics, questionnaires, spirometry).
• Employed machine-learning models to predict rapid lung function decline, with performance assessed by area under the receiver operating characteristic curve (AUC) and DeLong's test.

Main Results:

• Included 780 participants: 276 at risk, 298 with Global Initiative for Chronic Obstructive Lung Disease (GOLD) 1 COPD, and 206 with GOLD 2+ COPD.
• For individuals at risk, the upper-lobe CT model achieved a significantly higher AUC (0.80) compared to lower-lobe (0.63) and global models (0.66).
• No significant differences in prediction performance were observed between upper-lobe, lower-lobe, or global CT features for individuals with COPD.

Conclusions:

• CT features from the upper lobes significantly improved prediction performance for rapid lung function decline in individuals at risk.
• Prediction performance was notably enhanced in early/mild COPD stages compared to global CT features.