Title : Blood eosinophil count and prospective annual asthma disease burden : UK cohort

Background Elevated sputum eosinophil counts predict asthma exacerbations and responsiveness to inhaled corticosteroids but are impractical to measure in primary care. We investigated the relationship between blood eosinophil count and prospective annual asthma outcomes for a large UK cohort. Methods Historical database analysis utilising anonymised medical record data to identify primary care patients with asthma aged 12-80 years with 2 years of continuous data, including 1 year before (baseline) and 1 year after (outcome) their most recent eosinophil count. Negative binomial regression was used to compare outcome exacerbation rates and logistic regression to compare odds of asthma control for patients with blood eosinophils ≤400/μL vs. >400/μL, adjusting for age, sex, body mass index, smoking status, and Charlson comorbidity index. Findings Overall, 20 929 of 130 248 (16%) of patients had blood eosinophil counts >400/μL. During the outcome year, these patients experienced significantly more severe exacerbations (adjusted rate ratio [RR] 1·42; 95% CI 1·36-1·47) and acute respiratory events (RR 1·28; 1·241·33) than those with counts ≤400/μL. They also had significantly lower odds of achieving overall asthma control (OR 0·74; 0·72-0·77), defined as limited reliever use and no asthma-related hospital attendance/admission, acute course of oral corticosteroids, or prescription for antibiotics. Exacerbation rates increased progressively with nine ascending categories of blood eosinophil count as compared with a reference category of ≤200/μL. Interpretation Patients with asthma and blood eosinophil counts >400/μL experience more severe exacerbations and have poorer asthma control. Furthermore, a count-response relationship exists between blood eosinophil counts and asthma-related outcomes. Blood eosinophil counts could add predictive value to GINA control-based risk assessment. Funding Teva ClinicalTrials.gov: NCT02140541 Manuscript reference number: THELANCETRM-D-15-00283R1 Title: Blood eosinophil count and prospective annual asthma disease burden: UK cohort study Ms. Laura Feetham Senior Editor, The Lancet Respiratory Medicine Email: [email protected] Dear Ms. Feetham, Thank you for the second review of our manuscript (THELANCETRM-D-15-00283R1) and for giving us the opportunity to provide a revision. We have conducted the reanalysis requested by the reviewer and have tested the ELEN index as suggested.The results, summarised below, are now included inthe manuscript and the online supplement. We have submitted our revised paper as one "clean" copy and one copy where our changes are tracked, as instructed. In addition, we have provided a separate document listing the comments and our replies, point by point (below). Thank you for reconsidering our submission. Best wishes, David B. Price, for the authors Reviewers' comments: Reviewer #1: Reviewer comments: THELANCETRM-D-15-00283R1 Comments: 1. The authors have partially addressed my initial major comment # 1. This study pre-specified testing a binary cutoff of blood eosinophil at >400 cells/uL (which conforms to the RCTs of Teva'sreslizumab) for association with clinical endpoints. Alternatively, GSK initially conducted RCTs of mepolizumab with blood eosinophil cutoff at 300 cells/uL, and later modified the cutoff selection to also allow >150 cells/uL at screening. Various authors in the peer-reviewed literature have recommended other different cutoffs in peripheral blood eosinophils (usually ranging between 200 and 300 cells/uL) as the most accurate for identifying sputum eosinophilic asthmatics. Given the large sample size in this observational study, I believe this particular work has the potential to be a landmark publication for demonstrating an ideal blood eosinophil cutoff for identifying eosinophilic asthmatics in the general clinic to target optimal treatments. That is why I had previously recommended that the authors test associations between the clinical endpoints and blood eosinophil cutoffs at 200, 250, 300, 350, 400 cells/uL, and the ELEN Index. The authors only partially addressed my initial comment. In this revised submission, the authors tested blood eosinophil cutoffs at 300, 400, and 500 cells/uL to show association with the primary clinical endpoints (Table S3). A trend of increasing exacerbation risk *Reply to Reviewers Comments