Determinants of increased exhaled nitric oxide in patients with suspected asthma
Dr Pekka Malmberg
Department of Allergy
Helsinki University Central Hospital
PO Box 160, 00029 HUS
Exhaled nitric oxide (FENO) has been proposed as a marker of asthmatic inflammation, but it is unclear whether FENO in clinical use selects patients primarily according to their atopic or asthmatic status. The aim of this study was to investigate the determinants of increased FENO in patients with suspected asthma, by means of multinomial logistic regression analysis. The FENO of 132 patients referred because of symptoms suggestive of asthma were studied, and the explanatory factors tested included atopy according to prick skin tests, clinical asthma according to lung function tests, sputum eosinophilia and bronchial hyperresponsiveness (BHR). Slightly elevated FENO levels were significantly explained only by sputum eosinophilia (OR: 3.7; 95% CI: 1.1–13.1; P = 0.04), but for high levels of FENO (≥3 SD of predicted), clinical asthma (OR: 16.3; 95% CI: 5.4–49.7; P < 0.0001) and sputum eosinophilia (OR: 12.0; 95% CI: 4.1–35.0; P > 0.0001) were the characteristics with the highest prediction, followed by atopy and BHR. A significant interaction between asthma and atopy was observed relating to the effect on high FENO, but further analyses stratified by atopy showed significant associations between asthma and high FENO both in atopic and nonatopic patients. We conclude that in patients with symptoms suggesting asthma, slightly elevated and high levels of FENO are associated with sputum eosinophilia, whereas asthma is significantly associated only with high levels of FENO, irrespective of atopy. The results suggest that FENO is primarily a marker of airway eosinophilia, and that only high values of FENO may be useful to identify patients with atopic or nonatopic asthma.
Endogenous nitric oxide (NO) is synthesized from l-arginine by isoenzymes of NO synthetase, some of which are induceable by proinflammatory cytokines. In the exhaled air, NO is detectable in small amounts in healthy individuals, but the fractional concentration (exhaled nitric oxide, FENO) has been shown to increase in patients with inflammatory disorders of the airways, such as bronchiectasis and lower respiratory tract infections, and especially in asthma (1). Therefore, FENO has been proposed as a noninvasive marker of airway inflammation in asthma. Accordingly, FENO has been shown to identify asthmatic adults (2) and children (3) from healthy subjects with high sensitivity and specificity.
Recent studies have shown the relationship between FENO and markers of mucosal eosinophilic inflammation in patients with asthma (4, 5), although the precise mechanism how these different inflammatory markers are related is not known. Particularly, FENO seems to be increased in atopic asthmatics, and increased levels of FENO have been reported also in patients with atopic rhinitis without asthma (6), atopic asthmatics in remission (4), and in atopic subjects without any atopic disease or asthma (7). Furthermore, Gratziou et al. (8) could not find any difference in FENO between nonatopic asthmatics and healthy controls, suggesting that it is rather the atopic status than asthma, which is mainly responsible for the higher NO production in the lower airways. These findings are somewhat contradictory, since mucosal eosinophilic inflammation is likely to be present also in nonatopic asthma (9–13).
We wanted to address the question whether FENO can be used clinically to identify asthma and airway inflammation in a series of patients undergoing diagnostic procedures because of symptoms suggesting asthma (wheeze, cough, dyspnea). We reasoned that these patients represent the most probable target for the clinical use of FENO. Logistic regression analysis was used to identify determinants for raised FENO such as clinical asthma, atopy, sputum eosinophilia, and bronchial hyperresponsiveness (BHR). Particularly, the role of FENO in atopic and nonatopic asthma was of interest.
Material and methods
We included 132 patients remitted to an asthma and allergy clinic because of symptoms suggesting asthma (wheeze, cough, or dyspnea). Their characteristics have been presented in Table 1. The patients were included consecutively provided that satisfactory FENO measurements and induced sputum specimen were obtained. Patients using anti-inflammatory therapy within 3 months of the study (inhaled or systemic corticosteroids, chromones or leukotriene antagonists), were excluded. At the time of testing, the patients were in a stable condition and none of the patients had experienced a respiratory tract infection in the preceding 2 weeks. Short-acting β2-agonists were withheld for at least 12 h preceding the tests.
Table 1. Characteristics of the patients with symptoms suggesting asthma
|Age (years)||41 (14–81)|
|Height (cm)||168 (145–188)|
|Duration of symptoms (years)||3.8 (1–20)|
|FEV1 (l)||3.3 (1.8–5.9)|
|FEV1 (% predicted)||95 (69–134)|
|PD15FEV1 (mg), median (range)||1.2 (0.1 to >1.6)|
|Sputum eosinophils (score)||0.4 (0–3)|
|FENO (ppb)||11.3 (2.2–74)|
The patients underwent a series of lung function and allergy testing, including skin prick tests, flow-volume spirometry, bronchial challenge test, measurement of FENO, and induced sputum analysis. Standardized questionnaires were used to record the primary lower respiratory symptoms prior to referral, such as wheeze, cough, and shortness of breath. In addition, symptoms of rhinitis and smoking habits were recorded as possible confounding factors for FENO (6, 8, 14–16). The clinical diagnosis of asthma was based on typical symptoms and on at least 20% diurnal variation in peak expiratory flow (PEF), or on at least 15% improvement in forced expiratory volume in 1 s (FEV1) or PEF in response to bronchodilators, or on at least 15% exercise-induced decrease in PEF or FEV1. Skin prick testing was carried out with the following common inhalant allergens: birch, grass, mugwort, cat, dog, cow, horse, Cladosporium herbarum, Dermatophagoides pteronyssinus, and latex (Soluprick SQ, ALK, Horsholm, Denmark). The patient was regarded as atopic if having one or more positive reactions, determined as a wheal diameter of 3 mm or greater in the presence of expected results in control solutions (histamine dihydrochloride 10 mg/ml as a positive and solvent as a negative control).
Flow-volume spirometry was performed according to European standards (17), and BHR was estimated using a dosimetric histamine challenge test (18). By using the dose–response curve, the provocative dose of inhaled histamine producing a decrease of 15% in FEV1 (PD15FEV1) was determined. According to PD15FEV1, the bronchial hyperreactivity was graded as mild (0.41–1.60 mg), moderate (0.11–0.40 mg), or severe (≤0.1 mg) (18). The patients with PD15FEV1≤1.6 mg were considered hyperreactive. Fractional FENO concentration was measured using a chemiluminescence analyzer (CLD 77 AM; Eco Physics, Duernten, Switzerland) connected to a computerized system (Exhaled Breath Analyzer, Aerocrine AB, Stockholm, Sweden), and calibrated with a certified NO calibration gas mixture. The online single exhalation technique with exhalation rate 100 (±10) ml/s and positive expiratory mouth pressure of 10–20 mmH2O was applied, according to recommendations in Ref. (19). The mean FENO of two to three acceptable end-expiratory plateau measurements was calculated. According to the distribution of FENO in healthy nonsmoking subjects when using similar technique and exhalation rate, the concentration was coded as normal if <12 ppb, slightly elevated if 12–15 ppb and high if higher than 15 ppb (20).
Sputum was induced by inhalation of 5 ml of 3% NaCl solution, using an ultrasonic nebulizer (Omron U1, Omron, Germany) for 15 min as previously described (21). Sputum samples were transferred to Petri dishes and examined against a dark surface. The more viscous parts were collected, using forceps, and mixed. Air-dried slides were stained using eosin and methylene blue. A sample was considered to be adequate, and to originate from the lower airways, if it contained macrophages and fewer than 50% of squamous epithelial cells. The success rate for representative sputum induction has been evaluated previously and found to be 89–92% (21, 22). All analyses were conducted unaware of the clinical characteristics of the subject. Cell proportions were assessed semiquantitatively, on a scale from 0 to 4, as previously described (22). Eosinophilia was graded as 0 (no or occasional eosinophils), 1 (scanty), 2 (moderate), 3 (numerous eosinophils), or 4 (predominance of eosinophils). Approximately, grade 0 corresponds to eosinophils accounting for fewer than 1% of all nonsquamous cells, grade 1 to eosinophils accounting for 1–5%, grade 2 for 5–10%, grade 3 for 10–50% and grade 4 to eosinophils accounting for more than 50% of all nonsquamous cells (21, 22). The patient was defined as having significant sputum eosinophilia when the sputum eosinophil grade of 2 or more was observed.
Comparisons between continuous variables were estimated with unpaired t-test or Mann–Whitney test, depending on the distribution of the variable. For correlation analyses, Spearman rank correlation was used. The relationship of categorical variables with increased FENO were analyzed by multinomial logistic regression analysis with fixed model design, and by stepwise design with the forward selection method. The effects were given as odds ratios (OR) with 95% confidence intervals (CI). Statistical analyses were performed using statview and SPSS software packages (version 11.0 for Windows, Chicago, IL, USA).
Of the 132 patients recruited, 61 were skin prick test-positive (atopic) and, based on the lung function criteria, 30 were found to have clinical asthma. In 21 patients, high levels of FENO were observed, in 13 patients the concentration was slightly elevated, and in 98 patients the FENO was regarded as normal. Atopic patients had higher mean concentration of NO in exhaled air (13.6 ppb) than nonatopic patients (9.2 ppb, P < 0.05). The mean FENO was markedly increased (21.5 pbb) in the patients with clinically confirmed asthma, compared with those in whom the diagnosis could not be assessed (8.3 ppb, P = 0.0006). Accordingly, in the subgroup of nonatopic patients (n = 71), the asthmatics (n = 11) showed higher mean FENO (19.6 ppb), compared with nonasthmatics (7.4 ppb, P < 0.0001). In the whole series of patients, there was a significant correlation between FENO and the grade of sputum eosinophilia (r = 0.42; P < 0.0001). Similarly, in nonatopic patients, the relationship was significant (r = 0.38; P < 0.002).
Age, sex and smoking habits were not significantly related to elevated FENO. Smokers (n = 15) and nonsmokers (n = 117) showed similar mean levels of FENO (11.1 vs 13.0 ppb; P = 0.58, respectively). About 82 patients reported symptoms of rhinitis. Although the atopic patients with rhinitis (n = 44) had higher FENO compared to the patients with rhinitis but without atopy (n = 38) (13.5 vs 7.6 ppb, P = 0.04), as a group the patients with and without rhinitis showed similar mean levels of FENO (10.8 vs 12.1 ppb; P = 0.55, respectively). Since smoking habits and symptoms of rhinitis per se did not have an independent effect on FENO in the regression analyses (P > 0.10), these factors were omitted from subsequent models and analyses.
According to the questionnaire, 38 patients reported wheeze, 94 cough and 83 shortness of breath as their lower respiratory symptom. Slightly elevated FENO was not significantly related to any of the reported symptoms. For high FENO, wheeze was the only symptom which was significantly predictive (Table 2).
Table 2. Relationship of respiratory symptoms with increased exhaled nitric oxide (FENO, slightly elevated or high) according to multinomial logistic regression analysis in the patients with symptoms suggesting asthma (N = 132)
|Slightly elevated (12–15 ppb)||Wheeze||2.3||0.7–7.7||ns|
|High (>15 ppb)||Wheeze||4.9||1.8–13.1|| 0.002|
The relationship of increased FENO with possible explanatory findings have been shown in Table 3. Slightly elevated FENO levels were associated with sputum eosinophilia (OR: 3.7; 95% CI: 1.1–13.1; P = 0.04), but not significantly with asthma, atopy, or BHR. Clinical asthma was the characteristic with the highest prediction of high FENO (OR: 16.3; 95% CI: 5.4–49.7; P < 0.0001), followed by sputum eosinophilia (OR: 12.0; 95% CI: 4.1–35.0; P > 0.0001). The relationship of atopy and BHR with high FENO was also significant but weaker than that of asthma and sputum eosinophilia. By introducing the main explanatory variables and their interaction terms in stepwise (forward selection) model, clinical asthma (OR: 16.4; 95% CI: 4.6–58.5; P < 0.0001) and sputum eosinophilia (OR: 11.7; 95% CI: 3.2–41.7; P < 0.0001) were the factors included in the model with the best fit, whereafter the other variables and terms remained insignificant. The logistic regression analyses showed significant interactions between asthma and atopy (P = 0.001) as well as between sputum eosinophilia and atopy (P = 0.003) relating to their effects on high FENO. However, further analyses stratified by atopy showed significant associations between asthma and high FENO both in atopic (OR: 13.4; 95% CI: 3.3–54.4; P = 0.0003) and nonatopic (OR: 17.7; 95% CI: 2.7–117.6; P = 0.003) patients.
Table 3. Relationship of explanatory findings and increased exhaled nitric oxide (FENO, slightly elevated or high) according to multinomial logistic regression analysis in the patients with symptoms suggesting asthma (N = 132)
|Slightly elevated (12–15 ppb)||Asthma||1.2||0.3–6.0||ns|
|High (>15 ppb)||Asthma||16.3||5.4–49.7||<0.0001|
The results showed that in patients with symptoms suggesting asthma, high levels of NO in exhaled air, but not slightly elevated ones, have a strong relationship to clinical asthma, based on conventional lung function criteria. This association was significant both in atopic and nonatopic patients. When asthma and airway eosinophilia were accounted for, atopy did not have independent effect on FENO. This finding contrasts with some previous reports which suggest that FENO is primarily a marker of atopy (6, 8, 23, 24). Differences in the study populations may explain some of the disagreement: previous studies investigated population based samples, whereas in the present study, all the subjects were remitted for diagnostic evaluation because of suspicion of asthma, were symptomatic at the time of testing, and steroid naïve. The sample we investigated represents more closely the population where diagnostic tools, such as the measurement of FENO, will be applied in clinical practice.
As possible confounding factors for FENO, smoking habits and symptoms of rhinitis were recorded. Smoking is known to be associated with increased BHR, as well as with increased total cell counts and eosinophils in the sputum (25), but with decreased FENO levels (14). On the contrary, raised FENO levels have been observed in patients with rhinitis, even in the absence of asthma (6, 8, 15, 16). In the present study, however, the effects of smoking habits and rhinitis were negligible, and they could be omitted from the regression models between FENO and the explanatory variables.
In the present study, the data was categorized in order to estimate the association between FENO and the explanatory variables. For FENO, instead of using a simple cutoff value, we used multinomial analysis to investigate determinants of both slightly elevated (12–15 ppb) and high (>15 ppb) FENO. The latter limit of FENO corresponds approximately to +3 SD of the distribution in healthy nonsmoking subjects, determined with similar technique and expiratory flows (20). Slightly elevated FENO was found to be associated with sputum eosinophilia, but not significantly with asthma, atopy, or BHR. This finding may suggest that slightly elevated FENO levels do not have any diagnostic value for asthma. Indeed, modest elevations of FENO have been reported to be associated in many inflammatory disorders, e.g. with chronic rhinitis, mild upper airway respiratory tract infections and atopy per se (6, 7, 26). On the contrary, asthma was the most important explanatory factor for high levels of FENO. However, the number of cases with high FENO was less than that with lung function changes diagnostic for asthma, suggesting low sensitivity of this grade of FENO elevation.
By using stepwise logistic regression analysis, clinical asthma and airway eosinophilia were the most important explanatory variables for high FENO. The relationship of FENO with mucosal eosinophilic inflammation has been shown in patients with asthma (4, 5), but the present study shows that this association prevails also in a broader range of patients with symptoms suggesting asthma. It is possible that some of the symptomatic patients showing sputum eosinophilia and increased FENO, but failing to demonstrate lung function changes diagnostic for asthma, may have a milder state which we have referred as ‘asthma-like inflammation’ (27). Similar patients have been shown to present with increased levels of FENO (28). Furthermore, these patients, like those with clinical asthma, are likely to respond favorably with inhaled corticosteroids (27).
Airway inflammation is one of the primary manifestations of asthma (29), and characteristically, eosinophils can be found in the airway mucosa both in atopic and nonatopic asthmatics (9, 13). Some investigators have found many similarities in the immunopathology of intrinsic and extrinsic asthma (10–12). These studies on bronchial biopsies have shown locally enhanced expression of Th2-type cytokines even in patients with nonatopic asthma, similar to that seen in patients with extrinsic type of asthma. Up-regulation of expression of iNOS may occur as a result of proinflammatory cytokines involved in asthmatic inflammation, common to both phenotypes of asthma. On the contrary, there are also reports that have described distinct patterns of inflammatory profile in allergic and nonallergic asthma (30). These findings could explain the differences of FENO levels between atopic and nonatopic asthmatics and subjects reported previously (6, 8, 23, 24). Accordingly in the present study, atopy was a significant factor, although not the strongest one, in explaining variation of FENO in the patients with symptoms suggesting asthma. The present results also suggest that the relationship between atopy and increased levels of FENO is based on the predisposition of eosinophilic airway inflammation, and when the latter is accounted for, atopy per se does not have any independent effect.
We conclude that in patients with symptoms suggesting asthma, slightly elevated and high levels of FENO are associated with sputum eosinophilia, whereas asthma is significantly associated only with high levels of FENO, irrespective of atopy. The results suggest that FENO is primarily a marker of airway eosinophilia, and that only high values of FENO may be useful to identify patients with atopic or nonatopic asthma.