Some patients with chronic obstructive pulmonary disease (COPD) respond to corticosteroid therapy. Whether these patients have different airway pathology from other COPD patients is unclear. We tested the hypothesis that response to prednisolone is related to the presence of eosinophilic airway inflammation.

We did a randomised, double-blind, crossover trial. Patients who had COPD treated with bronchodilators only were assigned placebo and 30 mg prednisolone daily for 2 weeks each, in a random order, separated by a 4-week washout period. Before and after each treatment period, we assessed patients with spirometry, symptom scores, the chronic respiratory disease questionnaire (CRQ), incremental shuttle walk test, and induced sputum. Analysis was done by intention to treat.

83 patients were recruited, of whom 67 were randomised. The geometric mean sputum eosinophil count fell significantly after prednisolone (from 2·4% to 0·4%; mean difference six-fold [95% CI 3·1–11·4]) but not after placebo. Other sputum cell counts did not change. After stratification into tertiles by baseline eosinophil count, postbronchodilator forced expiratory volume in 1 s (FEV₁) and total scores on the CRQ improved progressively after prednisolone from the lowest to the highest eosinophilic tertile, compared with placebo. The mean change in postbronchodilator FEV₁, total CRQ score, and shuttle walk distance with prednisolone compared with placebo in the highest tertile was 0·19 L (0·06–0·32), 0·62 (0·31–0·93), and 20 m (5–35), respectively.

Our findings suggest that eosinophilic airway inflammation contributes to airflow obstruction and symptoms in some patients with COPD and that the shortterm effects of prednisolone are due to modification of this feature of the inflammatory response. The possibility that sputum eosinophilia identifies a subgroup of patients who particularly respond to long-term treatment with inhaled corticosteroids should be investigated.