Longitudinal Relationship between Sputum Eosinophils and Exhaled Nitric Oxide in Children with Asthma

To the Editor:

Measurement of fraction of exhaled nitric oxide (Fe_NO) is an established tool for the assessment of airway inflammation in asthma. Despite increasing evidence of a poor relationship between the two (1), Fe_NO continues to be widely regarded as a surrogate marker of sputum eosinophilia (2–4). Furthermore, recent American Thoracic Society (ATS) guidelines have suggested clinically significant cut points for low and high Fe_NO levels that can be used to indicate the presence or absence of sputum eosinophilia (5). It is known that levels of Fe_NO may be elevated even in the absence of eosinophilic inflammation and are strongly associated with atopy (6, 7). Furthermore, adult studies have also reported some patients with low Fe_NO levels despite high sputum eosinophils (8). These discordant relationships make interpretation of Fe_NO levels, in the individual, more difficult. In view of these published guidelines, we retrospectively analyzed data from two previous studies (9, 10) to assess the validity of these cut points in our population and whether in an individual the relationship between sputum eosinophils and Fe_NO would be consistent over time and thus aid the interpretation of Fe_NO results by identifying those in whom Fe_NO consistently reflects the presence or absence of sputum eosinophilia. Some of these results have been previously reported in the form of an abstract (11).

We analyzed paired sputum eosinophil and Fe_NO levels in children with a range of asthma severity up to four times per child in a 1-year period. A total of 79 school-aged children were recruited. Fifty-one had severe asthma and were recruited for a previously reported randomized controlled trial comparing a management strategy aimed at controlling sputum eosinophils with a conventional symptom-based strategy (NCT00262340) (9). Twenty-eight had mild to moderate asthma. They were prospectively recruited for a previously described observational study of inflammatory phenotypes (10). Their demographics are shown in Table 1. A total of 197 sputum and Fe_NO pairs were obtained.

TABLE 1.

DEMOGRAPHICS OF THE SEVERE ASTHMA AND MILD TO MODERATE ASTHMA COHORTS

	Severe Asthma (N = 51)	Mild to Moderate Asthma (N = 28)	P Value
Age, yr, mean (SD)	12.9 (2.7)	11.1 (2.2)	0.003
Sex, M/F	25/26	13/15	0.825
Baseline FEV1, % predicted, mean (SD)	79.4 (13.8)	76.9 (16.3)	0.473
Subjects with atopy, n (%)	43 (84%)	22 (79%)	0.523
Dose of ICS, FP equivalent, μg/d, median (IQR)	500 (500–1,000)	200 (100–344)	<0.001
Subjects prescribed ICS, n (%)	51 (100%)	23 (82%)	0.002
Subjects prescribed maintenance OCS, n (%)	8 (16%)	0	0.027
Subjects prescribed LABA, n (%)	51 (100%)	18 (64%)	<0.001
Subjects prescribed LRTA, n (%)	20 (39%)	3 (11%)	0.008

Definition of abbreviations: F = female; FP = fluticasone propionate; ICS = inhaled corticosteroids; IQR = interquartile range; LABA = long-acting β-agonist; LRTA = leukotriene receptor antagonist; M = male; OCS = oral corticosteroids; SD = standard deviation.

Significant differences are shown in italics.

Fe_NO was measured at a flow rate of 50 ml/s according to published European Respiratory Society/ATS guidelines (12). Spirometry, sputum induction, and processing were performed as previously described (13–15). Atopy was defined as one or more positive skin-prick tests (wheal ≥ 3 mm) or serum-specific IgE > 0.34 kU/L to a standard panel of six aeroallergens.

Concordance and discordance between sputum eosinophils and Fe_NO were defined using the Fe_NO cut points recommended by the recent ATS guidelines to define high and low Fe_NO levels (5). Concordance: high eosinophils and high Fe_NO (eos⁺Fe_NO⁺), eosinophils > 2.5%, Fe_NO ≥ 20 ppb; low eosinophils and low Fe_NO (eos⁻Fe_NO⁻), eosinophils ≤ 2.5%, Fe_NO ≤ 35 ppb. Discordance: high eosinophils and low Fe_NO (eos⁺Fe_NO⁻), eosinophils > 2.5%, Fe_NO < 20 ppb; low eosinophils and high Fe_NO (eos⁻Fe_NO⁺), eosinophils ≤ 2.5%, Fe_NO > 35 ppb. Fe_NO values between 20 and 35 ppb were always considered concordant whatever the eosinophil value.

The relationship between sputum eosinophils and Fe_NO was calculated using one pair per subject (the first available). There was a significant positive correlation between sputum eosinophils and Fe_NO (r = 0.49, P < 0.0001). Receiver operating characteristic curves were constructed to assess the utility of the ATS cut points for predicting sputum eosinophil counts of >2.5% and ≤2.5% (Figure E1 in the online supplement). Fe_NO levels of >35 ppb gave a sensitivity of 58%, specificity of 73%, positive predictive value of 67%, and negative predictive value of 65% for identifying a sputum eosinophil count of >2.5%. Fe_NO levels of <20 ppb gave a sensitivity of 51%, specificity of 76%, positive predictive value of 70%, and negative predictive value of 59% for identifying a sputum eosinophil count of ≤2.5%.

When all 179 samples were considered, 148 samples were concordant (eos⁺Fe_NO⁺ = 77; eos⁻Fe_NO⁻ = 71) and 49 (25%) were discordant (eos⁺Fe_NO⁻ = 25; eos⁻Fe_NO⁺ = 24) (Figure 1). Fifty-nine children produced two or more sputum samples. Of these, 31 children (53%) were consistently concordant, 24 (41%) had discordant eosinophil/Fe_NO levels in one sample but concordant levels on at least one other occasion (1 of whom showed discordance in both directions and concordance in a third sample), and only 4 (7%) demonstrated consistently discordant levels. Those who demonstrated the eos⁻Fe_NO⁺ relationship on at least one occasion were significantly more likely to have severe asthma (P = 0.027). There were no other differences between those who were consistently concordant and those who had discordant levels on at least one occasion in terms of age, sex, atopic status, or dose of inhaled corticosteroids (ICS) (Table E1).

Figure 1.

Relationship between sputum eosinophils and fraction of exhaled nitric oxide (Fe_NO) for all available samples. The dotted line denotes the 2.5% level for sputum eosinophils and the 20 and 35 ppb levels for Fe_NO. Levels of sputum eosinophils and Fe_NO have been log-transformed for clarity.

To our knowledge, this is the first study to assess over time the relationship between sputum eosinophils and Fe_NO. Almost half of all children did not have the expected relationship between sputum eosinophils and Fe_NO when more than one sputum eosinophil/Fe_NO pair per subject were included. Previous studies in children have reported high levels of Fe_NO in the absence of sputum eosinophilia; however, in our study, low levels of Fe_NO despite high levels of eosinophils were just as likely to occur. There is poor sensitivity and only moderate specificity at the ATS-suggested cut points for identifying the absence of sputum eosinophilia in children (20 ppb) and the presence of sputum eosinophilia (35 ppb). It was not possible in this study to identify those in whom Fe_NO would be a useful predictor of sputum eosinophilia.

A number of previous studies have reported an association between Fe_NO and atopy independent of asthma (6, 7), and high Fe_NO in the absence of eosinophils is well recognized (4, 16). This in part is due to the fact that Fe_NO is only indirectly related to airway eosinophilia and that NO production can be independent of eosinophilic inflammation (17). The presence of low levels of Fe_NO despite high eosinophil levels (eos⁺Fe_NO⁻) is much less well described and understood in childhood asthma and yet in our study was just as common as eos⁻Fe_NO⁺. This may be because of ICS causing Fe_NO suppression but only partial eosinophil suppression or because of stimulation of eosinophils by cytokines not involved in inducing iNOS. This study cannot answer these questions, but it does highlight that using Fe_NO to predict sputum eosinophilia is equally likely to overestimate and underestimate sputum eosinophilia.

The relationship between sputum eosinophils and Fe_NO was variable in almost half the children who produced more than one sputum sample. Children with severe asthma were more likely to have a variable relationship, but this in part was due to the fact that we collected more sputum samples per child in this group. It was not otherwise possible to identify a group in whom Fe_NO would consistently reflect eosinophilia. Each of these biomarkers measures different things and give us information that, although under some circumstances it is largely overlapping, is nonetheless independent of the other. This study cannot answer which is better, and further work is needed to establish the utility of Fe_NO, independent of sputum eosinophils, particularly as an indicator of steroid responsiveness in children already treated with ICS.