Use of exhaled nitric oxide measurement to identify a reactive, at-risk phenotype among patients with asthma

Abstract

Rationale: Exhaled nitric oxide (Fe(NO)) is a biomarker of airway inflammation in mild to moderate asthma. However, whether Fe(NO) levels are informative regarding airway inflammation in patients with severe asthma, who are refractory to conventional treatment, is unknown. Here, we hypothesized that classification of severe asthma based on airway inflammation as defined by Fe(NO) levels would identify a more reactive, at-risk asthma phenotype.

Methods: Fe(NO) and major features of asthma, including airway inflammation, airflow limitation, hyperinflation, hyperresponsiveness, and atopy, were determined in 446 individuals with various degrees of asthma severity (175 severe, 271 non-severe) and 49 healthy subjects enrolled in the Severe Asthma Research Program.

Measurements and main results: Fe(NO) levels were similar among patients with severe and non-severe asthma. The proportion of individuals with high Fe(NO) levels (>35 ppb) was the same (40%) among groups despite greater corticosteroid therapy in severe asthma. All patients with asthma and high Fe(NO) had more airway reactivity (maximal reversal in response to bronchodilator administration and by methacholine challenge), more evidence of allergic airway inflammation (sputum eosinophils), more evidence of atopy (positive skin tests, higher serum IgE and blood eosinophils), and more hyperinflation, but decreased awareness of their symptoms. High Fe(NO) identified those patients with severe asthma characterized by the greatest airflow obstruction and hyperinflation and most frequent use of emergency care.

Conclusions: Grouping of asthma by Fe(NO) provides an independent classification of asthma severity, and among patients with severe asthma identifies the most reactive and worrisome asthma phenotype.

Figure 1.

Relevant variables (as outlined in Table E1 in the online supplement) in the database were analyzed on the basis of receiver operator characteristic (ROC) curves with Fe_NO as a continuous variable. The cutoff point for each variable was determined on the basis of these ROC curves. Shown here is the frequency distribution of all these cutoff points. The median of all cutoff points for the variables (both categorical and continuous) that showed a significant relationship with Fe_NO was 37 ppb. This provided support for the validity of our selection of 35 ppb as the cutoff point between high and low Fe_NO.