Bronchial Responsiveness Is Related to Increased Exhaled NO (FENO) in Non-Smokers and Decreased FENO in Smokers

Rationale Both atopy and smoking are known to be associated with increased bronchial responsiveness. Fraction of nitric oxide (NO) in the exhaled air (FENO), a marker of airways inflammation, is decreased by smoking and increased by atopy. NO has also a physiological bronchodilating and bronchoprotective role. Objectives To investigate how the relation between FENO and bronchial responsiveness is modulated by atopy and smoking habits. Methods Exhaled NO measurements and methacholine challenge were performed in 468 subjects from the random sample of three European Community Respiratory Health Survey II centers: Turin (Italy), Gothenburg and Uppsala (both Sweden). Atopy status was defined by using specific IgE measurements while smoking status was questionnaire-assessed. Main Results Increased bronchial responsiveness was associated with increased FENO levels in non-smokers (p = 0.02) and decreased FENO levels in current smokers (p = 0.03). The negative association between bronchial responsiveness and FENO was seen only in the group smoking less <10 cigarettes/day (p = 0.008). Increased bronchial responsiveness was associated with increased FENO in atopic subjects (p = 0.04) while no significant association was found in non-atopic participants. The reported interaction between FENO and smoking and atopy, respectively were maintained after adjusting for possible confounders (p-values<0.05). Conclusions The present study highlights the interactions of the relationship between FENO and bronchial responsiveness with smoking and atopy, suggesting different mechanisms behind atopy- and smoking-related increases of bronchial responsiveness.


Introduction
Bronchial hyperresponsiveness is one of the hallmarks of asthma and measurement of bronchial responsiveness has been used clinically for over 30 years for asthma diagnosis and monitoring [1]. Exhaled nitric oxide has been introduced as a tool for asthma diagnosis in subjects with symptoms of asthma [2] and for the monitoring of asthma therapy [3]. Fraction of nitric oxide in the exhaled air (FE NO ) is a non-invasive marker of steroid-sensitive inflammation in the airways [4]. NO has also known bronchodilating and bronchoprotective physiological roles [5]. Apart from asthma, bronchial responsiveness and FE NO are also associated with other factors such as atopy and smoking. Atopy is related both to increased bronchial responsiveness [6] and increased FE NO [7], while smoking is associated with increased bronchial responsiveness [8] and decreased FE NO [9].
A positive correlation between bronchial responsiveness and FE NO has been found among subjects with allergic asthma [10] and in population-based studies of adults [11,12] and children [13]. In these studies, after stratification for atopy, the association between bronchial responsiveness and increased FE NO was statistically significant only among atopic individuals [11,13].
An interaction of bronchial responsiveness with smoking and atopy has been previously suggested in a Spanish population-based study [14] where current smoking was associated with increased bronchial responsiveness only in non-atopic subjects. On the other hand, FE NO is reduced to the same extent by current smoking in non-atopics and atopics [15]. This suggests that the association between FE NO and bronchial responsiveness is affected both by smoking and atopy. No previous studies have analyzed how smoking and smoking amount influences the relationship between bronchial responsiveness and FE NO .
The aim of the present study was to investigate the association between bronchial responsiveness and FE NO , with special regard to how this association is influenced by smoking, smoking amount and atopy.

Ethics Statement
Written informed consent was obtained from each subject before inclusion in the study. The protocol was approved by the Uppsala Ethics Committee (decision 131/1999 for Swedish multicentre application for Uppsala and Gothenburg) and Verona Ethics Committee (decision 74/1998 for Italian multicentre ECHRS II application including Turin).

Study participants
The European Community Respiratory Health Survey (ECRHS) is an international multicenter study of asthma and allergy. The first part, ECRHS I, was conducted in 1990-4 and the follow-up study, ECRHS II, in 1999-2001. The design of ECRHS I and II has been published in detail [16,17].
The present study included 468 subjects from the random sample of three of ECRHS II centers, Gothenburg (n = 225) and Uppsala (n = 175) (both Sweden) and Turin (n = 68) (Italy), who have undergone stage 2 of ECRHS I and in ECRHS II have answered the main questionnaire, performed measurements of FE NO , lung function tests and methacholine challenge. No subjects on daily inhaled steroids and/or oral antileukotrienes were included in the present analyses. Details regarding the selection of the subjects in these three centers are available in another publication [18].

Methacholine challenge
Methacholine challenge was carried out using a dosimeter (Mefar, Brescia, Italy). Methacholine challenge dose-response slope (''slope'') was calculated as the regression coefficient of percentage decline in FEV 1 on log dose of methacholine and then reciprocally transformed to satisfy statistical assumptions of multiple regression [19]. Its values range from 1 to 20. Two units of change in ''slope'' corresponds to one unit of change in log 10 (PD 20 ), or 3.32 doubling doses [20]. This relationship has been used to express the results in doubling doses in the manuscript. After transformation a low "slope'', like low PD 20 , was indicative of increased bronchial responsiveness. All subjects were instructed to refrain from smoking for at least 1 hour before lung function and methacholine reactivity measurements.

Exhaled NO
Exhaled NO measurements were done according to ATS/ERS recommendations [21]. Exhaled NO measurements were carried out on a different day than methacholine challenge. Different techniques and flow rates of measuring FE NO were used in different centers -offline measurements at 350 mL s 21 in Turin and online measurements at 50 mL s 21 in Uppsala and Gothenburg. The methods are described in more detail in another publication [18]. All subjects were instructed to refrain from smoking for at least two hours before measurements of exhaled NO, in order to exclude any potential confounding effects of acute smoking.

Smoking habits, atopy and asthma diagnosis
Smoking habits were questionnaire-assessed. A subject was considered as being a current smoker if he/she had been smoking for more than one year or at least 20 packs of cigarettes and was still smoking the month before the study. The number of smoked cigarettes per day and cigarette consumption in pack-years was also questionnaire-assessed.
Specific IgE was measured against Dermatophagoides pteronyssinus, cat, timothy grass and Cladosporium herbarum, using the Pharmacia CAP System (Pharmacia Diagnostics, Uppsala, Sweden). A person was defined as atopic if the titers against at least one of the tested allergens were $0.35 kU/L. Current asthma diagnosis was defined having self-reported physician-diagnosis of asthma and at least one asthma symptom or taking regular antiasthmatic medication during the last 12 months preceding the study.

Lung function
Forced expiratory volume in one second (FEV 1 ) was measured with a standardized method with different spirometers in different study centers, as previously described [18]. FEV 1 was expressed as % of the predicted value [22].

Statistics
Statistical analyses were performed using STATA 8.0 software (Stata Corp., 2001, Texas, USA). Different FE NO measurement techniques [23], NO analysers [24] and exhalation flow rates were used and we therefore divided FE NO in quartiles for each centre and pooled the data for the three centers instead of analyzing the absolute values of FE NO for each centre.
Trend tests were applied when analyzing the association between FE NO quartiles and other variables (Table 1). Simple linear regressions between slope values and FE NO quartiles were performed. Interactions with smoking and atopy were studied in multiple linear regression models where adjustments were made for factors known, from literature, to affect bronchial responsiveness and FE NO . The interactions were also tested by a metaanalysis of corresponding multiple regression linear models when using absolute value of FE NO instead of FE NO quartiles for the respective three study centers. Heterogeneity between centers regarding the interaction of smoking respectively atopy with the relation between FE NO and bronchial responsiveness was tested by means of a meta-analysis of the three centers. A p-value of ,0.05 was considered statistically significant.

Results
The characteristics of the study population are presented in Table 1. Subjects with higher FE NO levels were characterized by a higher prevalence of atopy and lower prevalence of current smoking, whereas no significant association was found between FE NO and slope values. Male gender, current asthma as well as increased height and weight, were associated with increased FE NO levels.

Selection bias -participants vs. non-participants
Participants who performed FE NO measurements were more likely to be men (50 vs. 44%, p = 0.02) and had a slightly higher mean age (43.260.3 vs. 41.260.3 years, p,0.0001) than participants who did not undergo FE NO measurements. No significant differences were found concerning bronchial responsiveness, smoking habits, atopy, physician diagnosed asthma, current asthma or body mass index between subjects who performed FE NO measurements and subjects who did not.

Effects of atopy and smoking on FE NO
Dividing the subjects after current smoking and atopy status (information available in 438 subjects), we obtained four groups: non-smoking non-atopic (n = 251), non-smoking atopic (n = 107), smoking non-atopic (n = 57) and smoking atopic subjects (n = 23). Comparing the distribution of subjects into different FE NO quartiles in the above mentioned groups, the group of nonsmoking non-atopic subjects had lower FE NO levels than the group of non-smoking atopic subjects (p = 0.01) and higher values than the smoking non-atopic subjects (p,0.001) ( Figure 1). No differences in FE NO were found between non-smoking non-atopics and the smoking atopic subjects (p = 0.96).

Effects of smoking status on the relationship between bronchial responsiveness and FE NO
Among non-smokers increased bronchial responsiveness was associated with increased FE NO values while an opposite trend was seen among current smokers ( Figure 2). There was a statistically significant difference in the association between slope and FE NO in non-and current smokers (p-value for interaction = 0.004).
In Table 2 the results are expressed as doubling doses of methacholine. The interaction between smoking and FE NO in relation to bronchial responsiveness remained significant after adjusting for gender, study centre, FEV 1 (%pred), age, height, weight, atopy, current asthma (Table 2). When stratifying for atopy, a significant interaction of smoking status with FE NO quartiles on airway responsiveness was found only among atopic subjects ( Table 2). No heterogeneity was found between centers regarding the interaction of current smoking with the association bronchial responsiveness and FE NO (p = 0.60). Significant trends for increasing bronchial responsiveness with increasing FE NO levels were found in all subjects (p = 0.009) and atopic subjects (p = 0.004) when a subanalysis was performed in Uppsala and Gothenburg centers. Moreover, the interactions with smoking remained statistically significant for all subjects (p = 0.012) and atopic subjects (p = 0.018).
The interaction between smoking and FE NO in relation to bronchial responsiveness was also found when FE NO was expressed as absolute FE NO values (p = 0.01).
The number of smoked cigarettes was correlated negatively to slope (p = 0.003) and current smokers who showed the lowest quartile of FE NO were those who were smoking more cigarettes and had a higher pack-years consumption (p = 0.002 and p = 0.003, see Table 1). Nevertheless, the cigarette consumption in pack-years was not significantly related to slope (p = 0.18). Dividing current smokers into two groups, a positive association between slope and FE NO quartile could be seen only in the group smoking less ,10 cigarettes/day (p = 0.008) and not in the group smoking $10 cigarettes/day (p = 0.81) (Figure 3). These relations were consistent after adjusting for pack-years consumption, and also after additional adjustments for age, gender, height, weight, lung function, current asthma, atopy, study centre (p = 0.03). Performing in such a model a test of interaction of ''light''/ ''heavy'' smoking with FE NO quartile on bronchial responsiveness a trend towards a significant interaction was found (p = 0.055). The positive association between slope and absolute levels of FE NO could be found in subjects smoking less than 10 cigarettes/day in Gothenburg and less than 13 cigarettes/day in Uppsala (both p,0.05) ( Table S1).

Effect of atopy on the relationship between bronchial responsiveness and FE NO
A positive correlation was found between increased bronchial responsiveness (decrease of slope) and increased FE NO among the atopic subjects whereas no significant correlation was found among the non-atopics (Figure 4, Table 3). The difference in association between bronchial responsiveness and FE NO in atopics and non-atopics was statistically significant and the interaction of atopy with FE NO quartiles on bronchial responsiveness remained statistically significant after adjusting for gender, study centre, FEV 1 (%pred), age, height, weight, atopy, current asthma (Table 3). No significant heterogeneity between centers was found regarding the interaction of atopy with the association between slope and FE NO (p = 0.13).
Dividing the participants into non-smokers and smokers the relationship between bronchial hyperresponsiveness and FE NO was found to be significant only among non-smoking subjects ( Table 3).
Significant trends for increasing bronchial responsiveness with increasing FE NO levels were found in all atopic subjects (p = 0.033) and all atopic, non-smoking subjects (p = 0.004) when a subanalysis was performed in Uppsala and Gothenburg centers. Moreover, the interactions with atopy remained statistically significant for atopic subjects (p = 0.04).
The interaction of atopy with the relationship between slope and FE NO was also found to be significant when using absolute FE NO values (p = 0.01).

Three-way interaction between atopy, smoking and FE NO on bronchial responsiveness
In a model where bronchial responsiveness was the outcome and three-way interactions between atopy, smoking and FE NO were tested, only the interaction between atopy with FE NO on bronchial responsiveness was significant (p = 0.005). This was consistent after adjusting for gender, study centre, FEV 1 (%pred), age, height, weight, atopy, current asthma (p = 0.003). The threeway interaction of smoking with atopy with FE NO on bronchial  responsiveness was not significant in unadjusted (p = 0.15) or adjusted model (p = 0.12).

Discussion
The main finding of the present study is that bronchial responsiveness is associated with increased FE NO levels in nonsmokers and with decreased FE NO levels in current smokers. Actually the inverse relationship between FE NO and bronchial responsiveness was significant only in ''light'' smokers, suggesting possible different mechanisms of bronchial responsiveness in ''light'' and ''heavy'' smokers. Increased bronchial responsiveness was associated with increased FE NO in atopic subjects while no such relationship could be seen in non-atopics. The nature of the interactions on the relationship between FE NO and bronchial responsiveness with smoking and atopy appears to be even more complex, since the interaction with smoking was seen only in atopics, while the interaction with atopy was seen only in nonsmokers.
We think there are many reasons why the inverse relationship between FE NO and bronchial responsiveness in smokers cannot be explained simply by considering the negative effect of smoking on FE NO , on one hand, and the favoring effect of smoking on bronchial hyperesponsiveness [25], on the other hand, without a causal relationship between the two effects. First, constitutively produced NO may play a bronchoprotective role, which should be lost in smokers, due to a lower NOS-production of NO [26,27] or an increased catabolism of NO [28,29]. Evidence for a bronchoprotective role of NO exist both in experimental animal studies [30] [31], but also in human studies performed in asthmatic subjects [32] [33] where administration of different non-selective iNOS inhibitors resulted in increased bronchial responsiveness. Another possible explanation could be related to the smoking-induced neutrophil inflammation. Sputum neutrophils count has been found to be negatively correlated to FE NO in smokers [34] and activation of neutrophils, in vitro, has been shown [35] to decrease NO, due to generation of peroxynitrite. Increased IL-16 has been linked with the neutrophilic inflammation [36], and IL-16 has been demonstrated to be increased in the airways of cigarette smokers, independent on the intensity of smoking [37]. Epithelial and subepithelial IL-16 immunoreactivity has been associated with increased bronchial responsiveness in humans with allergic asthma [38] and in an animal model of allergic asthma [39].
Moreover, in our study, decreased FE NO was associated with increased bronchial responsiveness only in ''light'' smokers, in whom the bronchoprotective effect of NO may be particularly valuable. Structural changes of small airways are related to smoking amount [40] and thus, in ''heavy'' smokers, bronchial hyperresponsiveness is best explained by structural changes of small airways and lung parenchyma [41]. However, we acknowledge the limitation that the different effects of ''light'' and ''heavy'' smoking on the association between FE NO and bronchial responsiveness could not be fully confirmed when performing a statistical interaction test (p = 0.055).
We were able to confirm in this large population sample that the previous reported association between FE NO and increased bronchial responsiveness in adults [11][12][13] was significant only in atopic subjects. Atopy-related increase in FE NO is due probably to the eosinophilic subclinical inflammation in the airways [42], as the link between FE NO and eosinophilic inflammation is well known [43][44][45]. The mechanism behind increased bronchial responsiveness in atopic subjects is most probably due to a combination of subclinical eosinophilic inflammation and remodeling changes described in the airways of atopic subjects [46]. A Th 2 -driven allergic response via IL-4-IL-13 cytokines could well result in both increased NO [47,48] and increased bronchial responsiveness [48,49].
The present study fills a gap regarding the effect of smoking on the association between bronchial responsiveness and FE NO and it  also made it possible to analyze the interactions of atopy and smoking on the association between bronchial responsiveness and FE NO . The only group where we did find an association of increased FE NO values with increased bronchial responsiveness was the group of non-smoking atopic individuals. We found similar levels of FE NO among the non-atopic non-smoking subjects and atopic smoking subjects due to the fact that FE NO is affected both by smoking and atopy.
The main weakness of the present study resides in the different methods to measure FE NO in the participating centers. We used quartiles of FE NO instead of absolute values of FE NO and no heterogeneity between centers was found regarding the interaction of smoking and atopy, respectively, with the relationship between FE NO and bronchial responsiveness. An indirect validation of this method of using FE NO quartiles in the present material is obtained by confirming the previous results on the relationship between FE NO and bronchial responsiveness [11][12][13]. The fact that in one center (Turin) FE NO was measured by higher flow-rate, which theoretically can sample to a slightly higher extent the peripheral airways, appears to be scarcely influent in this study, as atopy does not affect alveolar NO [50] and current smoking leads only to minor decrease of alveolar in comparison with bronchial contribution to exhaled NO [51]. Moreover, the main results could be confirmed in a subanalysis performed only in Gothenburg and Uppsala. In our population sample atopic subjects are underrepresented in the current smokers group, probably because the subjects with atopy and bronchial hyperresponsiveness might be less prone to start smoking. However this does not appear to confound our results, since the proportion of atopics increase with each FE NO quartile among the smokers without any corresponding increase in BR levels. COPD pathology is unlikely to have affected the results of the present study, as no subjects have a known COPD-diagnosis and only three subjects among the current smokers had a FEV1/FVC-ratio ,0.70.
The difference in the relationships between bronchial responsiveness and exhaled NO in smokers and atopics respectively suggests that atopy-and smoking cause bronchial hyperresponsiveness through different pathophysiological mechanisms. The nature of the interactions between bronchial responsiveness and exhaled NO is complex as the interaction with smoking could be seen only in atopics while the interaction with atopy could be seen only in non-smokers. Further studies are needed in order to understand the mechanisms explaining how smoking and atopy influence the relationship between bronchial responsiveness and exhaled NO. Table S1 The relation (beta coefficient from multiple linear regression models) between bronchial responsiveness (expressed as methacholine doubling dose) and FE NO in smoking subjects in Uppsala and Gothenburg centers # after dividing them for current cigarette consumption with different arbitrary cut-off levels. All the coefficients and p-values are adjusted for gender, FEV 1 (%pred), age, height, weight, atopy, current asthma. (DOCX) Table 3. The difference (D) in bronchial responsiveness (BR), expressed as doubling doses of methacholine 1 , between the first FE NO quartile (Q 1 ) and the other quartiles (Q 2 -Q 4 ) in all subjects, non-smokers and current smokers, after stratifying for atopy. Slope was the outcome of the regression model and doubling doses were obtained by multiplying the regression coefficients with 1.66, as described in the Methods. *p-value for trend represents the statistical significance for the association between bronchial responsiveness and FE NO quartile (used as a qualitative variable). # p-value for interaction represents the significance of interaction of atopy status with FE NO quartile on airways responsiveness. All the coefficients and p-values are adjusted for gender, study centre, FEV 1 (%pred), age, height, weight, atopy, current asthma. doi:10.1371/journal.pone.0035725.t003