Background: The short-term benefits of inhaled corticosteroids for patients with chronic obstructive pulmonary disease (COPD) are greater in patients with evidence of eosinophilic airway inflammation. We investigated whether blood eosinophil count is a useful biomarker of the long-term effect of the inhaled corticosteroid fluticasone furoate on exacerbation frequency.

Methods: We did a post-hoc analysis of data from two replicate, randomised, double-blind trials of 12 months' duration (Sept 25, 2009 to Oct 21, 2011 and Oct 17, 2011) in which once a day vilanterol 25 μg was compared with 25 μg vilanterol plus 50 μg, 100 μg, or 200 μg fluticasone furoate in patients with moderate-to-severe COPD and a history of one or more exacerbation in the previous year. We compared exacerbation rates according to two baseline eosinophil cell count strata (<2% and ≥2%), and according to four baseline percentage groupings. We also assessed lung function and incidence of pneumonia per strata in treatment groups.

Findings: We included 3177 patients in the analyses, with 2083 patients (66%) having an eosinophil count of 2% or higher at study entry. Across all doses of inhaled corticosteroids, fluticasone furoate and vilanterol reduced exacerbations by 29% compared with vilanterol alone (mean 0·91 vs 1·28 exacerbations per patient per year; p<0·0001) in patients with eosinophil counts of 2% or higher, and by 10% (0·79 vs 0·89; p=0·2827) in patients with eosinophil counts lower than 2%. Reductions in exacerbations with fluticasone furoate and vilanterol, compared with vilanterol alone, were 24% in patients with baseline eosinophil counts of ≥2-<4%, 32% for those with counts of 4-<6%, and 42% for those with eosinophil counts of ≥6%. In patients treated with vilanterol alone, exacerbation rates increased progressively with increasing eosinophil count percentage category. Improvement in trough forced expiratory volume in 1 s (FEV1) and the increased risk of pneumonia with fluticasone furoate and vilanterol compared with vilanterol alone were not associated with eosinophil count.

Interpretation: Blood eosinophil count is a promising biomarker of response to inhaled corticosteroids in patients with COPD. Blood eosinophil count could potentially be used to stratify patients for different exacerbation rate reduction strategies.

Funding: GlaxoSmithKline (study ID 201595).