RESUMEN
BACKGROUND: A multicenter clinical trial in patients with mild persistent asthma indicated that response to inhaled corticosteroids (ICS) is limited to those with sputum eosinophilia. However, testing for sputum eosinophilia is impractical in most clinical settings. OBJECTIVE: We examined associations between sputum eosinophilia and type 2 inflammatory biomarkers in untreated mild persistent asthma. METHODS: Induced sputum, blood eosinophil count (BEC), fractional exhaled nitric oxide (FeNO), and serum periostin were obtained twice during the 6-week run-in period in a clinical trial that enrolled patients 12 years and older with symptomatic, mild persistent asthma without controller therapy. The optimal threshold for each biomarker was based on achieving 80% or greater sensitivity. Performance of biomarkers (area under the receiver operating characteristics curve [AUC], range 0.0-1.0) in predicting sputum eosinophilia 2% or greater was determined; AUCs of 0.8 to 0.9 and more than 0.9 define excellent and outstanding discrimination, respectively. RESULTS: Of 564 participants, 27% were sputum eosinophilic, 83% were atopic, 70% had BEC of 200/uL or higher or FeNO of 25 ppb or greater; 64% of participants without sputum eosinophilia had elevated BEC or FeNO. The AUCs for BEC, FeNO, and both together in predicting sputum eosinophilia were all below the threshold for excellent discrimination (AUC 0.75, 0.78, and 0.79, respectively). Periostin (in adults) had poor discrimination (AUC 0.59; P = .02). CONCLUSIONS: In untreated mild persistent asthma, there is substantial discordance between sputum eosinophilia, BEC, and FeNO. Until prospective trials test the ability of alternative biomarkers to predict ICS response, BEC or FeNO phenotyping may be an option to consider ICS through a shared decision-making process with consideration of other clinical features.