There is increasing interest in using inhaled corticosteroids (ICS) as an option in Chronic obstructive pulmonary disease (COPD) treatment. However, COPD patients vary in their responsiveness to ICS. Extracellular vesicles (EVs), which mediate intercellular communication, could offer new opportunities for therapeutic targeting and disease monitoring.

Here we aimed to identify biomarkers predictive of disease severity and ICS responsiveness. In-depth proteomic analysis of EVs isolated from bronchoalveolar lavage fluid (BALF) in 34 COPD patients was conducted using Weighted Gene Co-expression Network Analysis (WGCNA). We identified two modules?red and salmon?that were highly correlated with lung function parameters before treatment. Within these modules, 25 from the red module and 11 proteins from the salmon module were identified as reliable predictors of lung function improvement, indicating ICS responsiveness after 6 months ICS use. A predictive model based on these 36 proteins demonstrated high accuracy, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.90 ± 0.2 and a Leave-One-Out (LOO) cross-validated accuracy of 81% (p < 0.001). Among these proteins, members of the CST family emerged as particularly robust predictors and displayed opposing trends in responder and non-responder groups during-ICS treatment, suggesting a close link to ICS sensitivity and type 2 inflammatory pathways.

These findings illustrate the BALF-derived EV profiles in COPD subjects and highlight the potential of EV proteins as biomarkers for predicting ICS responsiveness in COPD patients, paving the way to more personalized therapeutic strategies.