Introduction: The GOLD guidelines advocate not to institute inhaled corticosteroids (ICS) in patients with mild-to-moderate COPD. However, many patients do use ICS and in some patients, withdrawal is associated with deteriorating lung function and increased exacerbation rates. Unfortunately, physicians do not know in which patients they can stop ICS treatment safely.

Aim: To identify predictors of COPD exacerbations after ICS withdrawal.

Methods: During ICS treatment, post-bronchodilator spirometry, body plethysmography, and health status assessment were performed in 68 COPD patients using ICS. Additionally, sputum cell differentials, supernatant leukotriene B(4), eosinophilic cationic protein, and myeloperoxidase, serum C-reactive protein and soluble intracellular adhesion molecule, and urinary desmosine were assessed. Sputum was also analysed for mRNA levels of haemoxygenase-1, tumour necrosis factor-α, RANTES, interleukin 5(IL-5), IL-10, IL-12, IL-13, transforming growth factor-β, and interferon-γ.

Statistics: Cox regression analyses were performed using time to exacerbation as outcome variable to identify significant hazards for a COPD exacerbation after ICS withdrawal.

Results: Higher sputum % eosinophils, higher sputum MPO/neutrophil level, longer duration of COPD symptoms, <40 packyears smoking, and ICS withdrawal in November, December or January were significant hazards (all p<0.05) for experiencing a COPD exacerbation after ICS withdrawal in a monovariate model. In a multivariate model, all factors proved independent predictors except for sputum MPO/neutrophil level.

Conclusions: Decisions on whether or not inhaled corticosteroids can be safely withdrawn in mild-to-moderate COPD can be facilitated by assessment of sputum inflammation, particularly eosinophil numbers, next to packyears smoking, season, and duration of COPD symptoms.