Comparative effectiveness of LABA-ICS versus LAMA as initial treatment in COPD targeted by blood eosinophils: a population-based cohort study

Background Long-acting beta(2) agonists (LABAs) and long-acting inuscarinic antagonists (LAMAs) are the recommended initial maintenance treatment for chronic obstructive pulmonary disease (COPD), with almost all LABAs dispensed in fixed combination with inhaled corticosteroids (LABA-ICS). We compared the effectiveness and safety of LABA-ICS versus LAMA treatment initiation as a function of blood eosinophilia, a potential biomarker of ICS effectiveness, in a real-world setting. Methods In this population-based cohort study, we identified a cohort of patients with COPD initiating treatment with a LAMA or LABA-ICS during 2002-15, age 55 years or older, from the UK's Clinical Practice Research Datalink. We excluded patients who initiated treatment with both bronchodilators on the same date. All patients required at least 1 year of medical history and a measure of blood eosinophil concentration before cohort entry, defined by the date of the first cohort-defining bronchodilator prescription. Patients initiating a LAMA were matched on high-dimensional propensity scores with patients initiating a LABA-ICS. They were followed up for 1 year for the occurrence of a moderate or severe COPD exacerbation and for severe pneumonia. Sensitivity analyses included, among others, repeating the analysis among patients with two blood eosinophil concentration measures and stratification by concurrent asthma and previous exacerbations. Findings The base cohort included 539643 patients with a prescription for LABAs or LAMAs from an 1, 2002, to Dec 31, 2015, of whom 18 500 were initiated on LABA-ICS and 13 870 on LAMAs. Propensity score analysis resulted in 12366 initiators of LAMAs (mainly tiotropium) matched to 12366 initiators of LABA-ICS. The hazard ratio (HR) of COPD exacerbation associated with LABA-ICS initiation, relative to LAMA initiation, was 0.95 (95% CI 0.90-1.01). In patients with blood eosinophil concentrations of less than 2% of white blood cell count, the HR was 1.03 (95% CI 0.93-1.13) and for those with eosinophil concentrations of 2-4%, the HR was 1.00 (0.91-1.10). For patients with eosinophil concentrations of more than 4%, the HR was 0.79 (0. 70-0.88). The incidence of pneumonia increased with LABA-ICS initiation (HR 1.37 [95% CI 1.17-1.60]) and was similar across all eosinophil concentrations. Sensitivity analyses were consistent with these findings, but the incidence of exacerbation with LABA-ICS among the 2766 (11%) of all 24 732 patients with two or more COPD exacerbations during the baseline year was marginally lower (HR 0.87 [95% CI 0.79-0.97]). Interpretation In this real-world, dinical practice, observational study, initial COPD treatment with LABA-ICS inhalers was only more effective than with LAMAs in patients with high blood eosinophil concentrations (>4%) or counts (>300 cells per mu L) and possibly in frequent exacerbators. Because of the increased risk of pneumonia associated with the ICS component, initiation with a LAMA should be preferred in patients with blood eosinophil concentrations of less than 4%. Copyright (C) 2018 Elsevier Ltd. All rights reserved.