Abstract
Age-related declines in forced expiratory volume in one second are accelerated in smokers. Smoking is associated with decreased exhaled nitric oxide fraction (FeNO). The aim of the present study was to determine the impact of age on FeNO in otherwise healthy smokers and nonsmokers.
FeNO and serum cotinine levels were measured in 994 healthy subjects aged 18–40 yrs. American Thoracic Society questionnaire data on smoking habits was used to validate serum cotinine levels as a surrogate marker for categorisation of smokers and nonsmokers in the cohort.
Serum cotinine levels were a good discriminator of smokers (n = 99) and nonsmokers (n = 895). FeNO levels were significantly lower in otherwise healthy smokers compared with nonsmokers. There was an inverse correlation of serum cotinine levels with FeNO. No correlation of age with FeNO was found in nonsmokers but an inverse correlation of FeNO with age in smokers was found. FeNO was significantly lower in smokers aged 21–40 yrs compared with nonsmokers aged 21–40 yrs, but was not lower in smokers aged 18–20 yrs compared with nonsmokers of the same age.
Smoking was associated with decreased exhaled nitric oxide. The greatest smoking-related declines in exhaled nitric oxide occurred in older subjects. This suggests that smoking is associated with age-related declines in exhaled nitric oxide and justifies future mechanistic studies that address the impact of exhaled nitric oxide decline on lung function.
Inhaled tobacco smoke has acute and chronic effects on exhaled nitric oxide fraction (FeNO). FeNO levels are lower in otherwise healthy subjects who habitually smoke tobacco compared with nonsmokers 1–8, and FeNO levels decrease in nonsmokers and smokers acutely after smoking a cigarette and after passive smoke exposure 1, 5, 9. There is an inverse relationship between the number of cigarettes smoked per day and FeNO levels 1. FeNO levels increase after smoking cessation but not to normal levels 2, 6, 10, which suggests that smoking-related declines in FeNO may be associated with permanent lung damage. As respiratory epithelium is the likely source of most FeNO 11–15 and is damaged by chronic smoke exposure 16–20, smoking-related declines in FeNO may be a marker of airway epithelial damage.
There are several reasons to believe that low FeNO levels in smokers may be detrimental. Ageing is associated with declines in lung function and age-related declines in lung function are accelerated in smokers 21, 22. Smoking-related declines in forced expiratory volume in one second (FEV1) with increasing age are important determinants of obstructive lung disease 23, 24. Nitric oxide (NO) promotes bronchodilation 25–28 and, as such, smoking-related declines in FeNO may contribute to bronchospasm or obstructive lung disease in smokers. NO is also important in resistance to infection 29, 30 and decreased FeNO may predispose to respiratory infection.
If smoking-related declines in FeNO were associated with respiratory epithelial damage, then otherwise healthy older smokers should have lower FeNO levels than younger smokers. Prior studies of FeNO in nonsmoking adults indicated that there was no association of age with FeNO levels 31. In a recent study, McSharry et al. 32 identified an association of decreasing FeNO levels with age in asthmatic smokers. Whether smoking results in age-related declines in FeNO in otherwise healthy subjects has not been determined.
Therefore, a secondary analysis was performed of a large cohort of otherwise healthy African-American smokers and nonsmokers who participated in a genetic study of lung function and FeNO to determine the effect of age on FeNO levels. The current authors found that smoking-related declines in FeNO are greatest in older subjects. This suggests that smoking is associated with age-related declines in FeNO and justifies future mechanistic studies that address the impact of FeNO decline on lung function.
METHODS
Subject recruitment and enrolment
The present study results represent a secondary analysis of data obtained from subjects who were recruited as part of a study of the relationship between FeNO and type II inducible NO synthase (NOS2) genotypes present in African-Americans 33. Healthy subjects aged 18–40 yrs were recruited from students and employees at local university campuses. In total, 994 subjects with adequate serum samples and FeNO levels were available for analysis. Informed consent was obtained as part of a protocol approved by the Duke University Institutional Review Board (Durham, NC, USA).
After giving informed consent, subjects were asked to provide their date of birth and declare that they were healthy (i.e. no chronic illnesses or chronic use of any medication except oral contraceptives), that they had no history of asthma, allergic rhinitis, hay fever or atopic dermatitis, that they were nonsmokers, and that they were of African ancestry. Ethnicity was based on a self-declared description of ethnicity and determined using a questionnaire based on that developed by the US Census bureau. Blood samples for preparation of serum were obtained. FeNO was measured and a modified American Thoracic Society (ATS) questionnaire 34 was also administered to 524 subjects at the time of enrolment. The modified questionnaire collected information on smoking, cough, phlegm production, wheezing, dyspnoea and asthma history. The questionnaire also collected information about rhinitis, eczema and other allergy symptoms.
Measurement of FeNO
FeNO levels were measured in triplicate and averaged using a Sievers 280i Nitric Oxide Analyzer (NOA; GE Analytical Instruments, Boulder, CO, USA) according to the manufacturer's instructions. FeNO was measured according to ATS recommendations 35 at a flow rate of 50 mL·s−1 and against enough resistance to maintain an oropharyngeal pressure of ≥5 cmH20 and thereby permit closure of the soft palate and exclusion of nasal NO. Ambient air NO was excluded by inclusion of an activated charcoal and potassium hydroxide filter unit attached to the air intake of the NOA flow meter. Subjects were asked to breathe at least three times through the NOA flow meter with the attached activated charcoal and potassium hydroxide filter unit to reduce airway NO from ambient air prior to each FeNO measurement.
Serum total immunoglobulin E level measurements
Total serum immunoglobulin (Ig)E levels were measured using the Pharmacia CAP System (IgE FEIA; Pharmacia Diagnostics, Uppsala, Sweden).
Serum cotinine levels
Serum cotinine levels were determined using an ELISA-based assay from OraSure Technologies (Bethlehem, PA, USA). Manufacturer-specified cut-points for serum cotinine levels were used to separate subjects into smokers (≥25 ng·mL−1) and nonsmokers (<25 ng·mL−1).
Statistical analysis
Continuous variables (age, FeNO, total IgE, eosinophil cationic protein (ECP) and C-reactive protein (CRP) levels) were not normally distributed and were log normalised prior to statistical analysis. For the comparison of FeNO levels in subjects with different cotinine levels, ANOVA testing was used with multiple pairwise comparisons using Tukey–Kramer tests. For the linear regression analysis of cotinine levels and FeNO levels, cotinine levels were log normalised to permit a parametric analysis. For the analysis of current smokers and subjects that denied being ever smokers, cotinine levels were not log normalised because many cotinine levels were 0 ng·mL−1. In this analysis, cotinine levels were not normally distributed and nonparametric statistics (Wilcoxon rank-sum test) were used for this analysis. For univariate analyses of FeNO levels, unpaired t-tests were used for comparisons of smokers and nonsmokers. Subjects were dichotomised into younger (18–20 yrs) and older (21–40 yrs) groups based on the median age (20 yrs) of the cohort. Pearson's product–moment correlation coefficient was used for linear regression analyses of FeNO levels versus age and serum cotinine levels. Multiple linear regression was used for comparisons of FeNO levels while controlling for age, smoking status, sex, cotinine and total IgE levels. There were no corrections for multiple testing and the reported r2 values were unadjusted.
RESULTS
Association of elevated cotinine levels with self-reported current cigarette smoking
The present authors’ original a priori study of genetic correlates of FeNO levels was designed to exclude smokers. However, given the significant association of cigarette smoking with lower FeNO levels 1–8, ATS questionnaire data on smoking 34 were collected and serum cotinine levels were measured 36, 37 as a way to rigorously exclude smoking as a potential confounder of FeNO levels. To confirm the validity of using serum cotinine levels as a surrogate marker of cigarette smoking, cotinine levels were compared in subjects who self-reported cigarette smoking on the ATS questionnaire with subjects that did not report smoking. Of the 524 subjects who were administered ATS questionnaires, seven reported being current smokers and 471 denied being ever-smokers (i.e. subjects who reported being either current smokers or past smokers). Cotinine levels were higher in subjects who reported being current smokers versus subjects that denied being ever-smokers (median (interquartile range) 29.4 ng·mL−1 (0.3–130 ng·mL−1) versus 0.1 ng·mL−1 (0–0.6 ng·mL−1); p = 0.0025, Wilcoxon rank-sum test). This analysis confirmed previous reports and validated the use of cotinine levels as a surrogate for cigarette smoking in the present cohort of subjects.
The manufacturer of the serum cotinine assay used in the present study recommends a cut-off of 25 ng·mL−1 for the identification of smokers. The current authors found that, despite self-reporting of nonsmoking by many subjects, 99 (10%) subjects had elevated serum cotinine levels (≥25 ng·mL−1) consistent with regular smoking. Given the inherent problems associated with self-reported information on smoking 38, the present authors elected to use serum cotinine levels as a surrogate for cigarette smoke exposure in the remainder of the analysis.
FeNO in subjects with elevated serum cotinine levels
FeNO levels were compared in subjects with serum cotinine levels <1 ng·mL−1 (subjects with minimal cigarette-smoke exposure), 1–25 ng·mL−1 (subjects with moderate cigarette-smoke exposure consistent with second hand-smoke exposure or occasional smoking) and ≥25 ng·mL−1 (subjects with high cigarette-smoke exposure and consistent with regular cigarette smoking). FeNO levels were significantly lower only in subjects with serum cotinine levels ≥25 ng·mL−1 (fig. 1⇓). Based on the results in figure 1⇓, subjects were dichotomised into smokers and nonsmokers for all further analyses using a cotinine concentration of 25 ng·mL−1 as a cut-off for the two subject groups.
Consistent with prior studies 1–8, it was found that FeNO levels were significantly lower in smokers compared with nonsmokers (geometric mean±sd): 13.9±18.0 ppb versus 20.5±21.3 ppb (p<0.0001); difference = 6.6 ppb (confidence interval (CI) 4.1–9.2). As shown in figure 2⇓, there was an association of FeNO levels with serum cotinine levels in smokers; serum cotinine levels were inversely correlated with FeNO levels in smokers (r2 = 0.13; p = 0.0003), suggesting that lower FeNO levels were associated with increased levels of tobacco smoke exposure.
Age-related declines in FeNO levels in smokers
The effect of age on FeNO levels was examined in the present cohort. Consistent with prior studies 31, no correlation of age with FeNO levels in nonsmokers was found (fig. 3a⇓). In contrast, analysis of the relationship between FeNO and age in smokers indicated that there was a significant decrease in FeNO levels in older smokers (fig. 3b⇓). A multiple linear regression model confirmed that age was significantly associated with FeNO levels in smokers while controlling for cotinine levels (table 1⇓).
In analyses stratified for age (based on the median age of the cohort, 20 yrs), comparison of FeNO levels in younger subjects (18–20 yrs) did not reveal significant differences in FeNO levels between smokers and nonsmokers, while FeNO levels in older subjects (21–40 yrs) were significantly different between smokers and nonsmokers (p<0.0001; fig. 4⇓). Among all subjects, the age-related difference in FeNO levels between smokers and nonsmokers (i.e. the difference between 18–20-yr-old smokers and nonsmokers compared with the difference between 21–40-yr-old smokers and nonsmokers) was significantly different (p<0.0001; fig. 4⇓) 39.
As there was an uneven distribution of subjects in each age group with regard to the variables listed in table 2⇓, the present authors also tested whether smoking status was significantly associated with FeNO levels in multiple linear regression models when subjects were stratified on the basis of age and while controlling for other variables (sex and total IgE levels; cotinine was not included because it is absent in most nonsmokers) associated with FeNO levels 31, 40. This analysis confirmed that smoking status was significantly associated with FeNO levels in older subjects (aged 21–40 yrs, p<0.0001; table 3⇓) but not younger subjects (aged 18–20 yrs, p = 0.16; table 4⇓). Likewise, when smoking status was included in a multiple linear regression model with age, sex and total IgE levels, there was a significant interaction of age and smoking status (p = 0.0008, data not shown).
DISCUSSION
The current authors believe that their study represents the largest analysis to date of the effect of smoking on FeNO levels in otherwise healthy smokers. The study confirms previous findings indicating that cigarette smoking was associated with lower FeNO levels 1–8 and that greater short-term (hours to days) exposure to inhaled tobacco smoke was associated with greater decreases in FeNO levels (fig. 2⇑) 1. The current study also addressed the hypothesis that smoking was associated with age-related declines in FeNO. Support for this hypothesis was demonstrated in two ways. First, an age-related decline in FeNO levels in smokers but not nonsmokers was demonstrated (fig. 3⇑). Secondly, it was demonstrated that FeNO levels in older (21–40 yrs of age) smokers but not younger (18–20 yrs of age) smokers were significantly different from comparably aged nonsmokers (fig. 4⇑).
Recent studies by McSharry et al. 32 and Malinovschi et al. 10 did not observe a similar age-related decline in FeNO levels in healthy nonasthmatic smokers. The current authors believe that the present results were different from the results of the other two other studies because they had smaller sample sizes and did not include subjects <21 yrs of age. For example, had the present study only analysed the correlation between age and FeNO in the subset of 54 smokers in the cohort aged >21 yrs, a correlation between age and FeNO levels would not have been observed (r2 = 0.0223, p = 0.2809). However, given the limited numbers of subjects in the current study who were in their late 30's, it remains possible that the analysis in figure 3a⇑ was unduly affected by low FeNO levels in some of the oldest smokers in the cohort.
Like the studies of McSharry et al. 32 and Malinovschi et al. 10, other studies of smoking and FeNO have not included subjects aged <21 yrs and, therefore, would not have reached the same conclusions as the present study regarding FeNO levels in young smokers compared with nonsmokers 1–4, 6. One possible explanation for the current findings regarding the lack of a difference in FeNO levels in young smokers compared with nonsmokers is related to the known association of increasing age in children (but not adults) with increasing FeNO 40–42. As such, continued increases in FeNO in adults <21 yrs of age may have counteracted smoking-related declines in FeNO in this age group and accounted for the findings regarding the lack of a difference in FeNO levels in young smokers compared with nonsmokers.
Long-term smoking leads to functional and structural changes in the lung. The present results on the age-related decline in FeNO suggest that the duration of tobacco-smoke exposure might be associated with decreased FeNO levels. The current data does not include self-reported information on the length of time that subjects smoked, although >80% of adult smokers start smoking before the age of 18 yrs 43, suggesting that subject age is a reasonable correlate of duration of tobacco-smoke exposure. While other factors besides smoking may account for the age-related decline in FeNO levels in smokers, the present authors believe this is unlikely because there were no age-related changes in FeNO levels in nonsmokers (fig. 3a⇑). The presence of mild chronic obstructive pulmonary disease (COPD) in older subjects may have influenced the results. Although Roy et al. 44 recently found that COPD was associated with decreased airway NO levels, most studies suggest that COPD is associated with increased FeNO levels 4, 45 and, if anything, this would have diminished the difference in FeNO levels between older smokers and nonsmokers. Taken together, the current authors believe that long-term smoking may result in permanent reductions in FeNO levels, possibly via airway epithelial lining changes and changes in NOS2 expression and NO production.
Tobacco-smoke exposure is associated with airway epithelial hyperplasia, mucus cell hyperplasia, airway metaplasia, epithelial damage and airway neutrophil infiltration 16, 17. Many of these changes persist after subjects quit smoking, including persistence of neutrophil inflammation 16, 17. The hypothesis that long-term cigarette smoking is associated with permanent reductions in FeNO is supported by studies on older smokers who quit smoking 2, 6. While these studies indicated that FeNO levels increased in subjects after they quit smoking, former smokers still had lower FeNO levels compared with healthy control subjects 2, 6. Likewise, Malinovschi et al. 10 found that ex-smokers had significantly lower FeNO levels than never-smokers. A prospective, longitudinal study of abstaining smokers who had previously smoked cigarettes for various durations would provide better evidence to support the hypothesis that long-term smoking may be associated with permanent reductions in FeNO levels.
The design of the present study did not allow direct examination of respiratory epithelium in smokers or the correlation of epithelial damage with FeNO. Therefore, other hypotheses that account for the age-related decline in FeNO in smokers are also possible. Smoking-related downregulation of epithelial NOS2 enzyme activity and smoking-related increases in airway enzymes and reactive oxygen species that react with or metabolise NO are possible aetiologies for the age-related decline in FeNO in smokers 10, 46. Another hypothesis relates smoking-induced pulmonary changes due to COPD as the cause of decreased FeNO in smokers. As stated earlier, while most studies found that FeNO levels increase in patients with COPD 4, 45, more recent data from Roy et al. 44, using multiple measurements of exhaled NO at different flow rates to model airway NO, suggests that FeNO levels are lower in COPD patients. Like the hypothesis about smoking-related respiratory epithelial damage, the COPD hypothesis would associate COPD-related inflammatory changes and parenchymal lung damage with reductions in FeNO. Although the current results do not provide definitive evidence to address these hypotheses as the cause of decreased FeNO levels in smokers, the results of the present study clearly support an association of an age-related decline in FeNO in smokers.
While the present authors focused on the possibility that smoking has significant age-related effects on airway respiratory epithelium, it is possible that other compartments within the lung may contribute to the lower FeNO levels associated with smoking. All three isoforms of NOS are expressed in the lung 11. Techniques that permit separate measurement of the airway and alveolar components of FeNO would be useful in this context to address the hypothesis that smoking primarily alters airway respiratory epithelial production of FeNO 47. Experiments using intravenous and inhaled inhibitors of the different isoforms of NOS would also be useful to delineate the relative contributions of neural NOS1, airway NOS2 and vascular NOS3 to the difference in FeNO between older smokers and nonsmokers 15, 48.
Measurement of serum cotinine, a nicotine metabolite, is an objective surrogate for quantifying tobacco-smoke exposure 49. Cotinine levels can be accurately measured in blood, urine, saliva and other body fluids. The half-life of cotinine in serum is 15–17 h, which allows for accurate estimation of daily tobacco-smoke exposure in randomly obtained blood specimens 49. Serum cotinine levels correlate well with the number of cigarettes smoked and with the amount of time since a cigarette was smoked 50–52. Information on the number of cigarettes that subjects smoked was not collected in the current study and, as such, relied on the use of serum cotinine levels as a surrogate marker of cigarette consumption.
The present authors believe that serum cotinine levels represent the best available assessment of smoking status in the cohort as subject self-reports of smoking habits did not adequately exclude smokers. Based on data provided by the manufacturer of the cotinine assay and published studies 53, it is believed that the manufacturer-recommended cut-off of 25 ng·mL−1 for serum cotinine levels is conservative and unlikely to include many nonsmoking subjects exposed to passive smoke. The significantly lower FeNO levels in subjects with serum cotinine levels ≥25 ng·mL−1 (fig. 2⇑) and the comparable FeNO levels in current smokers versus subjects with serum cotinine levels ≥25 ng·mL−1 also supports the rationale for selecting 25 ng·mL1 as a cut-off for classification of smoking status. The current data did not indicate lower FeNO levels in subjects with cotinine levels between 1–25 ng·mL−1 and there was only a weak correlation between FeNO and serum cotinine levels in subjects with cotinine levels between 1–25 ng·mL−1 (r2 = 0.03, p = 0.0410). Furthermore, serum cotinine concentrations >1 ng·mL−1 and <25 ng·mL−1 are consistent with exposure to second-hand smoke 53. In contrast to the declines in FeNO associated with cigarette smoking, a recent publication found that exposure to secondhand smoke was associated with modestly increased FeNO levels 53. Taken together with the present data on the relationship of FeNO levels and serum cotinine levels (fig. 2⇑), the current authors believe that this data supports dichotimisation of subjects into smokers and nonsmokers on the basis of a serum cotinine level of 25 ng·mL−1.
In summary, the present results indicated that smoking had different impacts on exhaled nitric oxide levels in subgroups of otherwise healthy smokers. Smoking-related declines in exhaled nitric oxide levels appeared to be disproportionately greater in older individuals. This suggested that smoking was associated with age-related declines in exhaled nitric oxide and justifies future mechanistic studies that address the impact of exhaled nitric oxide fraction decline on lung function.
Acknowledgments
The authors are grateful for the assistance of E. Scarlett and C. Foss during subject enrolment and sample collection. They also acknowledge C. Thomas and C. Tao for assistance with computer data entry. The use of J.B. Weinberg's nitric oxide analyser and W. Burks' (all affiliated with Duke University Medical Center, Durham, NC, USA) CAP-FEIA instrument for measurement of exhaled nitric oxide levels and total IgE levels, respectively, was greatly appreciated.
- Received July 12, 2007.
- Accepted September 17, 2007.
- © ERS Journals Ltd