Evaluation of sputum eosinophil count as a predictor of treatment response to mepolizumab

Predicted annualised rates of clinically significant exacerbations at week 52, by a) baseline sputum eosinophil count and b) baseline blood eosinophil count; c) rate ratio (mepolizumab/placebo) of clinically significant exacerbations, by sputum and blood eosinophil count thresholds. Shaded areas in panels a) and b) represent 95% confidence intervals for predicted rates; arrows represent predicted events per year at baseline eosinophil thresholds. The analyses of clinically significant exacerbations included all patients with sputum samples and a) sputum eosinophil data available at baseline (mepolizumab: n=62 (n=18, 23 and 21 in the 75, 250 and 750 mg intravenous (i.v.) arms, respectively); placebo: n=24) and b) blood eosinophil data available at baseline (mepolizumab: n=67 (n=20, 24 and 23 in the 75, 250 and 750 mg i.v. arms, respectively); placebo: n=27). Owing to small sample sizes, statistical testing was not performed on these data.