Eosinophilic airway inflammation measured by using induced sputum is an important treatment stratification tool in patients with severe asthma. In addition, sputum eosinophilia has been shown to be associated with severe exacerbations and airflow limitation.

We sought to identify whether eosinophilic inflammation in sputum is associated with FEV₁ decrease in patients with severe asthma and whether we could identify subgroups of decrease behavior based on the variation of eosinophilic airway inflammation over time.

Ninety-seven patients with severe asthma from the Glenfield Asthma Cohort were followed up with scheduled 3-month visits; the median duration of follow-up and number of visits was 6 years (interquartile range, 5.6-7.6 years) and 2.7 visits per year. Induced sputum was analyzed for eosinophilic inflammation at scheduled visits. Linear mixed-effects models were used to identify variables associated with lung function and overall decrease. In addition, using individual patients' mean and SD sputum eosinophil percentages over time, a 2-step cluster analysis was performed to identify patient clusters with different rates of decrease.

FEV₁ decrease was −25.7 mL/y in the overall population. Postbronchodilator FEV₁ was also dependent on exacerbations, age of onset, height, age, sex, and log₁₀ sputum eosinophil percentages (P < .001). Three decrease patient clusters were identified: (1) noneosinophilic with low variation (mean decrease, −14.0 mL/y), (2) eosinophilic with high variation (mean decrease, −40.9 mL/y), and (3) hypereosinophilic with low variation (mean decrease in lung function, −19.2 mL/y).

The amplitude of sputum eosinophilia was associated with postbronchodilator FEV₁ in asthmatic patients. In contrast, high variability rather than the amplitude at baseline or over time of sputum eosinophils was associated with accelerated FEV₁ decrease.