Introduction: Adverse effects of inhaled corticosteroids (ICS), such as pneumonia, are believed to be due to effects on airway bacterial load and the microbiome. Different ICS molecules may have distinct effects in COPD.

Objectives: To investigate whether commonly used ICS have different effects on the COPD airway microbiome.

Methods: Multicentre randomized controlled trial. After a 4-week washout patients with COPD (FEV1 <50% predicted or ?2 exacerbations per year) were randomized to one of 4 treatments: budesonide/formoterol 400/12 (BF), fluticasone/salmeterol 500 (FS500), fluticasone/salmeterol 250 (FS250) or aclidinium/formoterol (AF). Patients were treated for 3 months with monthly induced sputum, oropharyngeal (OP) and nasopharyngeal (NP) swabs for bacterial load and 16S sequencing. Inflammatory markers were measured in sputum. The primary outcome was bacterial load in OP swabs between BF and FS500, and the key secondary outcome was sputum bacterial load.

Results: Following washout of 122 patients, 61 were randomized. After 3 months treatment there was no significant difference in OP bacterial load between FS500 and BF (difference 0.26 logunits, 95%CI 0.65-1.18, p=0.56), but there was a significant increase in bacterial load in sputum at 3 months comparing FS500 and BF (difference 0.87 logunits, 95%CI 0.29-1.44, p=0.0037). BF treatment did not increase bacterial load compared to AF. There was no significant effect of different ICS on Shannon diversity index or inflammatory biomarkers.

Conclusion: Patients randomized to fluticasone/salmeterol at licensed doses had an increased bacterial load in sputum over time compared to budesonide/formoterol.