Predictors and associations of the persistent airflow limitation phenotype in asthma: a post-hoc analysis of the ATLANTIS study

Background: Persistent airflow limitation (PAL) occurs in a subset of patients with asthma. Previous studies on PAL in asthma have included relatively small populations, mostly restricted to severe asthma, or have no included longitudinal data. The aim of this post-hoc analysis was to investigate the determinants, clinical implications, and outcome of PAL in patients with asthma who were included in the ATLANTIS study. Methods: In this post-hoc analysis of the ATLANTIS study, we assessed the prevalence, clinical characteristics, and implications of PAL across the full range of asthma severity. The study population included patients aged 18-65 years who had been diagnosed with asthma at least 6 months before inclusion. We defined PAL as a post-bronchodilator FEV1/forced vital capacity (FVC) of less than the lower limit of normal at recruitment. Asthma severity was defined according to the Global Initiative for Asthma. We used Mann-Whitney U test, t test, or χ2 test to analyse differences in baseline characteristics between patients with and without PAL. Logistic regression was used for multivariable analysis of the associations between PAL and baseline data. Cox regression was used to analyse risk of exacerbation in relation to PAL, and a linear mixed-effects model was used to analyse change in FEV1 over time in patients with versus patients without PAL. Results were validated in the U-BIOPRED cohort. Findings: Between June 30, 2014 and March 3, 2017, 773 patients were enrolled in the ATLANTIS study of whom 760 (98%) had post-bronchodilator FEV1/FVC data available. Of the included patients with available data, mean age was 44 years (SD 13), 441 (58%) of 760 were women, 578 (76%) were never-smokers, and 248 (33%) had PAL. PAL was not only present in patients with severe asthma, but also in 21 (16%) of 133 patients with GINA step 1 and 24 (29%) of 83 patients with GINA step 2. PAL was independently associated with older age at baseline (46 years in PAL group vs 43 years in non-PAL group), longer duration of asthma (24 years vs 12 years), male sex (51% vs 38%), higher blood eosinophil counts (median 0·27 × 109 cells per L vs 0·20 × 109 cells per L), more small airway dysfunction, and more exacerbations during 1 year of follow-up. Associations between PAL, age, and eosinophilic inflammation were validated in the U-BIOPRED cohort, whereas associations with sex, duration of asthma, and risk of exacerbations were not validated. Interpretation: PAL is not only present in severe disease, but also in a considerable proportion of patients with milder disease. In patients with mild asthma, PAL is associated with eosinophilic inflammation and a higher risk of exacerbations. Our findings are important because they suggest that increasing treatment intensity should be considered in patients with milder asthma and PAL.