Clinical Outcome of Eosinophilic Airway Inflammation in Chronic Airway Diseases Including Nonasthmatic Eosinophilic Bronchitis

We enrolled patients with confirmed sputum eosinophilia who had visited our tertiary referral hospital between 2012 and 2015. We evaluated the incidence and predictors of exacerbations in patients with nonasthmatic eosinophilic bronchitis (NAEB), and investigated predictors of improvement in eosinophilic inflammation in chronic airway diseases with or without persistent airflow limitation. In total, 398 patients with sputum eosinophilia were enrolled. Of these, 152 (38.2%) had NAEB. The incidence rate of exacerbations requiring treatment with antibiotics, systemic corticosteroids, or hospital admission was 0.13 per patient-year (95% CI, 0.06–0.19) in NAEB. Inhaled corticosteroid (ICS) did not affect the risk of exacerbations, even in an analysis of propensity score. One hundred seventy-six patients had chronic airway diseases; in 37 of these (21.0%), sputum eosinophilia had improved at the 1-year follow-up. Patients who had persistent airflow limitation were less likely to show an improvement in eosinophilic inflammation (aOR, 0.26; 95% CI, 0.09–0.77) when they were treated with ICSs for less than 75% of the follow-up days. Exacerbations requiring systemic corticosteroids, antibiotics, or hospitalization did occur, although infrequently, in NAEB patients. Among patients with chronic airway diseases, those with persistent airflow limitation were less likely to show an improvement in eosinophilic airway inflammation.


Measurement.
The following clinical data were collected for analysis: age, sex, smoking status, and baseline symptom scores (cough score, COPD assessment test score 18 , and asthma control test score 19 ). The cough score was assessed using a numeric rating scale ranging from 0 to 5 (0 = no cough, 5 = cough all the time). Adherence to ICSs was assessed using the medication possession ratio (MPR), which estimates the percentage of days' supply obtained during the follow-up period.
A positive bronchodilator response was defined as an increase in FEV 1 of ≥12% and ≥200 mL from baseline after inhalation of 200 µg of salbutamol. Airway hyperresponsiveness was identified using bronchial provocation tests; these were considered positive if the FEV 1 fell by ≥20% after a methacholine dose of <16 mg/mL 20 , or if the FEV 1 had fallen by ≥15% before the last dose of mannitol (before a cumulative mannitol dose of 635 mg had been administered) 21 .
The patients' atopic status was determined using either skin prick testing to 55 common inhalant allergens or immunoglobulin E (IgE) specific to Dermatophagoides pteronyssinus and Dermatophagoides farinae measured by an ImmunoCAP 250 analyzer (ThermoFisher, Uppsala, Sweden). Specific IgE sensitization was dichotomized at a cut-off of 3.50 kU/L. Atopy was defined as either a positive skin prick test or a positive specific IgE measurement to above house dust mites.
Induced sputum tests were conducted as previously described 22 . Briefly, sputum was induced using a 4.5% hypertonic saline inhalation, administered through an ultrasonic nebulizer (Omron; Tokyo, Japan) for 5-20 min, with the output set at 4.5 mL/min. The sputum was mixed with an equal volume of 0.01 M dithioerythritol, filtered using a 100-µm cell strainer or mesh on a 15-mL tube, and centrifuged at 400-800 g for 10 min at 4 °C. The cell pellets were resuspended using phosphate-buffered saline. Cytospins were performed at 42 g for 5 min; the cells then underwent Diff-Quik staining. Differential cell counts were expressed as percentages of 300 non-squamous cells.
Data Analysis. Clinical characteristics were compared between groups using either the independent samples t-test or one-way analysis of variance for continuous variables. For categorical variables, they were compared using either the χ 2 test or Fisher's exact test.
Data analyses were carried out independently in two subcohorts. First, in a subcohort of patients who had NAEB diagnosed at baseline (subcohort 1), the incidence rate of moderate or severe exacerbations was estimated, and the incidence rate ratio was calculated-using a negative binomial regression model-between the group with an MPR for ICSs of <50% and that with an MPR for ICSs of ≥50%. An exacerbation was defined as any worsening of respiratory symptoms that led to treatment with systemic corticosteroids, antibiotics, or both (moderate), to hospital admission, or to emergency department visits (severe). This definition is widely applied in studies involving smokers 23 and chronic respiratory diseases (asthma, COPD, and destroyed lung by tuberculosis 24 ). The time to the first moderate or severe exacerbation was analyzed using the Kaplan-Meier method after propensity score (PS) matching, which provided 33 pairs of patients, each pair comprising one with an MPR for ICSs of <50% and the other with an MPR for ICSs of ≥50%. We also estimated the incidence rate of the development of chronic airflow obstruction in subcohort 1. Second, in a subcohort of patients with sputum eosinophilia who had been followed up with induced sputum tests for at least 1 year (subcohort 2), we used a logistic regression model to identify predictors of improvement in sputum eosinophilia. Improvement in sputum eosinophilia was defined as a decrease in induced sputum eosinophil count to <3%. The multivariable analysis was adjusted for persistent and variable airflow limitation, and variables with P values < 0.1 in univariable analyses considering collinearity. Persistent airflow limitation was identified in cases of postbronchodilator FEV 1 /FVC < 70% predicted, and SCIeNtIfIC REPORts | (2018) 8:146 | DOI:10.1038/s41598-017-18265-2 variable airflow limitation was defined in cases of positive bronchodilator response or airway hyperresponsiveness. The 1-year change in sputum eosinophils was analyzed using a random-slope linear mixed model after PS matching, which provided 47 pairs of patients, each pair comprising one with an MPR for ICSs of <75%, and the other with an MPR for ICSs of ≥75%. P values less than 0.05 were considered significant. Statistical analyses were performed using Stata statistical software (Version 14.2; StataCorp LP, College Station, TX). Data Availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Results
Patient Characteristics. A total of 560 patients with sputum eosinophilia were enrolled. Of these, 162 were excluded from the study (Fig. 1). Of the resulting 398 patients, 152 had NAEB (subcohort 1); Table 1 summarizes their clinical characteristics. Only one fourth of the NAEB patients had received ICS therapy for ≥50% of the follow-up days, and less than one sixth had received it for ≥75% of the follow-up days.
Exacerbations in NAEB patients. Of the 152 NAEB patients, 15 (9.9%) experienced 16 moderate or severe exacerbations during the follow-up period (median, 6.4 months; interquartile range [IQR], 0.7-15.3 months). The incidence rate of moderate or severe exacerbations was 0.13 per patient-year (95% CI, 0.06-0.19 per patient-year). To investigate whether ICSs affect exacerbations, PS matching was applied (MPR for ICSs, <50% vs ≥50%). The clinical characteristics of subcohort 1 were comparable after PS matching (S1 Table). The incidence rates of exacerbations in each group were not significantly different in the PS matched-cohort (incidence rate ratio, 1.72; 95% CI, 0.44-6.63; Table 2). Furthermore, the time to the first moderate or severe exacerbation did not differ significantly between the groups (log-rank P = 0.607; Fig. 2).

Chronic Airflow Obstruction in NAEB Patients.
Of the 48 NAEB patients who were followed-up with induced sputum tests for at least 1 year, 46 underwent additional yearly pulmonary function tests with bronchodilator response or provocation tests during the follow-up period (median, 16.2 months; IQR, 12.7-25.1 months). Among them, chronic airflow obstruction developed in five patients (asthma: three patients, COPD: two patients). The incidence rate of the development of chronic airflow obstruction was 0.07 per patient-year (95% CI, 0.03-0.16 per patient-year). The median time to the development of chronic airflow obstruction was 14.2 months (IQR, 12.5-25.1 months).

Characteristics and Outcomes of Patients With Sputum Eosinophilia.
The clinical characteristics of subcohort 2 are summarized in Table 3. About 4% of NAEB patients experienced moderate or severe exacerbations during the 1-year follow-up period, whereas about 27% of those with asthma and possible ACOS, and 12% of those with COPD did. Of the 176 patients in subcohort 2, 37 (21.0%) showed an improvement in sputum eosinophilia at the 1-year follow-up. Patients with NAEB were more likely to show an improvement in sputum eosinophilia than those with other chronic airway disease (NAEB: 35.4%, asthma: 18.2%, possible ACOS: 16.9%, and COPD: 12.2%; P = 0.044; Table 3). Table 4 shows the univariable analysis of factors associated with a 1-year improvement in sputum eosinophilia. Because there was a statistically significant interaction between persistent airflow limitation and MPR for ICSs (P for interaction = 0.027), the multivariable analysis was performed in each subgroup stratified in terms of MPR for ICSs (<75% vs ≥75%). After adjustment for blood eosinophils, serum IgE, FEV 1 % predicted, sputum eosinophils, and variable airflow limitation, patients who had persistent airflow limitation and had received ICS therapy for less than 75% of the follow-up days were less likely to show an improvement in eosinophilic airway inflammation (aOR, 0.26; 95% CI, 0.09-0.77; P = 0.015;    Table), ICS use had no significant impact on the decrease in sputum eosinophils in the linear mixed model (P for interaction = 0.101).

Discussion
In summary, the incidence rate of acute exacerbations in NAEB was 0.13 per patient-year. We did not find that ICS therapy reduced the exacerbation rate in patients with NAEB. NAEB rarely progressed to chronic airflow obstruction. One fifth of patients with chronic airway diseases showed an improvement in eosinophilic airway inflammation at the 1-year follow-up. Patients with persistent airflow limitation (COPD or possible ACOS) showed a lower probability of improvement in sputum eosinophilia than those without persistent airflow limitation (NAEB or asthma). Our study showed that NAEB patients did experience acute exacerbations during the follow-up period. As mentioned previously, NAEB patients could have respiratory symptoms other than cough, such as chest tightness with wheezing, shortness of breath, and sputum production [8][9][10] . Treatment with systemic corticosteroids is occasionally required to relieve these symptoms 11 . However, there were no studies investigating the incidence and predictors of acute exacerbations. Recently, the SPIROMICS cohort study-which used the same definition of an exacerbation-reported that symptomatic current or former smokers without COPD did experience exacerbations, and that their annualized exacerbation rate was significantly higher than those of asymptomatic current or former smokers and never-smokers (0.27, 0.08, and 0.03 events per year, respectively) 23 . The exacerbation rate of NAEB patients in our study was higher than that of asymptomatic smokers, but half that of symptomatic smokers in the SPIROMICS cohort who had preserved pulmonary function.
We did not find that ICSs prevented exacerbations in NAEB patients. In addition, the mean sputum eosinophil counts between baseline and the 1-year follow-up were not different regardless of ICS treatment (S3 Table and S1A Fig.). Little improvement in eosinophilic airway inflammation was in contrast to findings of previous prospective studies, in which all NAEB patients were treated with ICSs for at least 4 weeks 7,25 . The failure to show their effect on exacerbations and eosinophilic airway inflammation highlights that only a small number of patients have received adequate therapy in the real-world population of NAEB. In our study, only a quarter of NAEB patients were treated with ICSs for ≥50% of the follow-up days, and less than one sixth were treated with ICSs for ≥75% of the follow-up days. This might result in the lack of statistical power in the 1:1 PS matched analysis because only a small fraction of participants could be included in the analysis. Another potential cause of the real-world consequence in respect to lack of the efficacy of ICSs in preventing NAEB exacerbations is prevailing infectious triggers leading to aggravated symptoms. The definition of the exacerbation was non-discriminatory with respect to inherent worsening of eosinophilic airway inflammation versus infection by respiratory viruses or other infectious organisms. There may be a signal for a subset analysis where subjects with obvious infectious etiologies to exacerbations are excluded. However, due to the retrospective design of this study, we could not explicitly distinguish between exacerbations with and without infectious etiologies. Nevertheless, regardless of their effect on exacerbations and sputum eosinophils, ICSs played a role in improving symptoms in our study. The mean cough score had significantly reduced-from 2.3 to 1.6-at the 1-year follow-up in the eight NAEB patients whose MPR for ICSs was ≥50% (P = 0.049). In the 12 NAEB patients whose MPR for ICSs was <50%, the mean cough score had not changed significantly (P = 0.586; S3 Table and S1B Fig.).
We found that chronic airflow obstruction developed in relatively few NAEB patients, although it was limited by the relatively short follow-up period. According to previous studies by Berry et al. 5 and Park et al. 6 , persistent airflow obstruction developed in approximately 15% of NAEB patients. However, in a recent investigation by Lai et al. 7 , none of the NAEB patients developed persistent airflow obstruction. This inconsistency regarding COPD development may have arisen because the studies had different proportions of smokers. Specifically, approximately 20% of NAEB patients were current or former smokers in the present study, and in that by Berry et al. 5 . In the study by Park et al. 6 , 46% of the participants were smokers. However, only 6% of NAEB patients were smokers in the study by Lai et al. 7 .
In a subgroup analysis, we showed that patients with persistent airflow limitation were less likely to show improvements in sputum eosinophilia. It has been reported that ICSs reduce the number of inflammatory cells in the bronchial mucosa and sputum 26 , and that the presence of eosinophilia in sputum 12,14 and blood 27,28 is a predictor of response to ICSs in COPD patients. However, no studies have compared the treatment response between COPD and other chronic airway diseases. In the present study, only 12% of patients with COPD showed an improvement in sputum eosinophilia, while more than a third of those with NAEB did. Corticosteroid resistance in COPD 29,30 might explain the relatively poor treatment response to ICSs in our study.
The current study has several limitations, including its retrospective design of a cohort at a single institution. First, we could not standardize therapeutic plans of various chronic airway diseases, thus, therapies other than ICSs could affect airway eosinophilic inflammation. In addition, not all NAEB patients were evaluated with regard to whether their symptoms were improved by ICS therapy. Second, since we excluded patients with eosinophilic lung diseases based on chest radiographs, a possibility of systemic diseases such as vasculitis occurring without     variable airflow limitation, rather than disease entities, in subcohort 2. For this reason, our definition of ACOS (dubbed "possible ACOS") did not affect the validity of the analysis determining predictors of 1-year improvement in sputum eosinophilia.
In conclusion, exacerbations requiring systemic corticosteroids, antibiotics, or hospitalization did occur in NAEB patients, although infrequently. Among patients with chronic airway diseases, those with persistent airflow limitation were less likely to show improvement in eosinophilic airway inflammation.