We analyzed the positivity of 39 allergen-specific IgE in all COPD patients. Concerning the perennial aeroallergens, the positivity for moth (31.5%), Candida (23.7%), Dermatophagoides pteronyssinus (D. pteronyssinus, 22.4%), and house dust (22.4%) was greater than 20% (Figure 1). As for pollen allergens, the positivity for Japanese cedar (35.5%) and Japanese cypress (22.2%) exceeded 20%. As for foods or other allergens, the positivity for shrimp/lobster (15.5%), bananas (15.5%), and wheat (12.7%) was higher than 10%.
We also compared the characteristics of patients who are positive with those who are negative for specific IgE (Figure 2, Additional File 3). In patients with positive IgE for cockroach, residual volume (RV) was significantly higher (136.7% vs 110.2%), and FEV1 tended to be lower (69.5% vs 78.7%) compared with patients with negative IgE (Figure 2). Contrarily, in patients with positive IgE for Japanese cedar, the diffusion capacity of carbon monoxide/alveolar volume (DLCO/VA) was significantly higher than in patients with negative IgE (92.1% vs 73.2%) (Figure 2). The prevalence of perennial allergic rhinitis was significantly higher in patients with positive specific IgE for all allergens, except for Japanese cedar, which causes seasonal rhinitis only in the spring season (Additional File 3).
Furthermore, we compared patient characteristics between ACO and non-ACO COPD (Table 1). In ACO, body mass index and FeNO were significantly higher compared with non-ACO COPD. FEV1, %predicted also tended to be lower in ACO, and FeNO was significantly higher in patients with values that fell in the lower half of FEV1, %predicted (26.3 vs 40.8 ppb, P < 0.05). A significantly greater proportion of patients with ACO exhibited all the features of asthma, except for airway reversibility (Additional File 4).
Next, we compared the positivity of specific IgE between ACO and non-ACO COPD (Table 2). Among various aeroallergens, only the positivity for house dust and D. pteronyssinus was significantly higher in ACO. As for other perennial aeroallergens, the positivity for moth, Candida, and Malassezia was numerically higher in ACO without statistical significance. As for pollen, the positivity for Japanese cedar tended to be higher in ACO (P = 0.16). Concerning food allergens, the overall positivity for beef-specific IgE was low (8.5%); however, the positivity was significantly lower in ACO. We also analyzed the relationship between the value of IgE class and the proportion of patients diagnosed with ACO (Figure 3). Only house dust- and D. pteronyssinus-specific IgE showed significant relationship between the class of specific IgE and ACO diagnosis.
As the diagnostic criteria for ACO included FeNO and blood eosinophil count, in addition to IgE levels, we compared the utility of those biomarkers in ACO diagnosis by the ROC curve analysis. Although there were weak or no significant relationships between those biomarkers (Additional File 5), the area under the curve for ACO diagnosis was comparable among three biomarkers: 0.666 (95% CI 0.540–0.793, P = 0.013) for blood eosinophil count, 0.665 (95% CI 0.538–0.791, P = 0.014) for total IgE, and 0.703 (95% CI 0.578–0.829, P = 0.0027) for FeNO (Figure 4). There were no significant differences between the ROC curves for ACO diagnosis (IgE vs blood eosinophil count: P = 0.99, IgE vs FeNO P = 0.67, FeNO vs blood eosinophil count: P = 0.64). The best cutoff value for diagnosis was 234 count/μl for blood eosinophils, 158 IU/mL for serum total IgE, and 31.0 ppb for FeNO.
The contribution of blood eosinophils, FeNO, and IgE was analyzed to further determine the relative importance of the different biomarkers in ACO diagnosis. In univariate analysis, the criteria, except for blood eosinophils, significantly contributed to ACO diagnosis (FeNO: OR 9.58, 95% CI 2.45–37.4, P < 0.01, IgE: OR 3.93, 95% CI 1.45–10.68, P < 0.01). In addition, although specific IgE for house dust (P = 0.0146) and D. pteronyssinus (P = 0.0146) significantly contributed to the diagnosis of ACO, positivity for moth (P = 0.2564), cockroach (P = 0.5653), Japanese cedar pollen (P = 0.1593), and Candida (P = 0.2917) did not exhibit significant contributions. We also analyzed the contribution of the number of sensitized allergens, and the number did not contribute to the diagnosis of ACO (P = 0.98).
Concerning IgE factors, all of the following criteria significantly contributed to ACO diagnosis: total IgE > 100 IU/mL, positive for house dust-specific IgE, and positive for D. pteronyssinus-specific IgE. In the three models for multivariate analyses (Table 3), all IgE criteria significantly contributed to ACO diagnosis, in addition to FeNO > 35 ppb criterion.