Rationale: Chronic obstructive pulmonary disease (COPD) and asthma can exhibit overlapping clinical features. Exhaled air contains volatile organic compounds (VOCs) that may qualify as noninvasive biomarkers. VOC profiles can be assessed using integrative analysis by electronic nose, resulting in exhaled molecular fingerprints (breathprints).
Objectives: We hypothesized that breathprints by electronic nose can discriminate patients with COPD and asthma.
Methods: Ninety subjects participated in a cross-sectional study: 30 patients with COPD (age, 61.6 ± 9.3 years; FEV1, 1.72 ± 0.69 L), 20 patients with asthma (age, 35.4 ± 15.1 years; FEV1 3.32 ± 0.86 L), 20 nonsmoking control subjects (age, 56.7 ± 9.3 years; FEV1, 3.44 ± 0.76 L), and 20 smoking control subjects (age, 56.1 ± 5.9 years; FEV1, 3.58 ± 0.78). After 5 minutes of tidal breathing through an inspiratory VOC filter, an expiratory vital capacity was collected in a Tedlar bag and sampled by electronic nose. Breathprints were analyzed by discriminant analysis on principal component reduction resulting in cross-validated accuracy values (accuracy). Repeatability and reproducibility were assessed by measuring samples in duplicate by two devices.
Measurements and Main Results: Breathprints from patients with asthma were separated from patients with COPD (accuracy 96%; P < 0.001), from nonsmoking control subjects (accuracy, 95%; P < 0.001), and from smoking control subjects (accuracy, 92.5%; P < 0.001). Exhaled breath profiles of patients with COPD partially overlapped with those of asymptomatic smokers (accuracy, 66%; P = 0.006). Measurements were repeatable and reproducible.
Conclusions: Molecular profiling of exhaled air can distinguish patients with COPD and asthma and control subjects. Our data demonstrate a potential of electronic noses in the differential diagnosis of obstructive airway diseases and in the risk assessment in asymptomatic smokers.
Clinical trial registered with www.trialregister.nl (NTR 1282).
Lung diseases can change the molecular mixture of exhaled air. Electronic noses are suitable for the recognition of molecular patterns and provide an exhaled breathprint.
Patients with chronic obstructive pulmonary disease and asthma can be distinguished by their breathprints of exhaled air. This shows the potential of electronic noses in the discrimination of obstructive airways diseases.
The chronic airways inflammation in COPD and asthma features distinct cellular and molecular profiles (3, 7, 8). This provides opportunities to discriminate these disease entities based on composite molecular signatures. High-dimensional diagnostic techniques (e.g., genomics, metabolomics) allow assessment of biomarker profiles, resulting in a specific “fingerprint” of the disease (9). This may not only be applicable to serum (10) or bronchoalveolar lavage fluid (11) but also to noninvasive alternatives such as exhaled air. Metabolomics in exhaled breath condensate has successfully been applied to discriminate patients with asthma from control subjects (12). Exhaled air is also known to contain thousands of volatile organic compounds (VOCs) that are derived from various metabolic and inflammatory pathways in the lung and elsewhere in the human body (13–15). These VOCs can be used as biomarkers for diagnosing lung disease (15) as has been demonstrated in bronchial carcinoma (16–18). However, the need of gas chromatography and mass spectrometry (GC-MS) has limited widespread application in clinical practice.
Unlike GC-MS, electronic noses (eNoses) represent an integrative measurement of VOCs, allowing high-throughput analysis of complex gas mixtures. eNose technology is based on an array of nanosensors reacting to the different fractions of the VOC mixture in breath (19). When these sensor responses are combined, a specific fingerprint or “breathprint” for the disease is created, which is analyzed by pattern recognition algorithms (19, 20). Thus, this technique combines the noninvasiveness of measuring exhaled breath with real-time analysis of the complete spectrum of volatiles without individual determination of the molecular components. Therefore, eNoses may have potential as a diagnostic tool (21).
The diagnostic application of eNoses has mostly focused on lung cancer (18). Recently, Dragonieri and colleagues demonstrated that eNose assessment of exhaled air distinguished patients with lung cancer from those with COPD (22). Furthermore, breathprints were different between patients with asthma and healthy control subjects (23). When analyzing a young (27 ± 6 yr of age) and older (57 ± 7 yr of age) control group, there was no effect of age on breathprints (23). Considering the distinct inflammation profiles of asthma and COPD (7, 8), it is likely that these diseases will reveal different eNose breathprints of the exhaled air.
Therefore, we hypothesized that exhaled breath fingerprinting by eNose can distinguish COPD from asthma. The aim of the present study was to test this hypothesis by comparing patients with an established diagnosis of asthma and COPD in a cross-sectional design. To capture a potential confounding effect of smoking, we included nonsmoking and smoking control subjects. Exploratory objectives were to examine the influence of inhaled corticosteroid (ICS) use and smoking in COPD on eNose breathprints and the reproducibility and repeatability of measurements. Some of the results of these studies have been previously reported in the form of an abstract (24).
Ninety subjects, 18 to 87 years of age, were included in the study. Subjects were divided into four groups according to current standard diagnostic procedures and smoking status: (1) patients with moderate to severe COPD, (2) patients with mild to severe persistent asthma, (3) asymptomatic smoking control subjects, and (4) nonsmoking control subjects. Patients with asthma and COPD were recruited among those visiting the outpatient clinics of the participating centers and family care practices, whereas control subjects were recruited by advertisements in the hospitals and public newspapers.
The COPD group consisted of 30 patients with a smoking history of at least 15 pack-years, symptoms of dyspnea or chronic cough or chronic sputum production, a postbronchodilator FEV1 between 30 and 80% predicted (GOLD stage II or III according to GOLD guidelines ), and a FEV1/FVC ratio less than 70%. Patients with COPD on oral steroids were excluded.
The asthma group consisted of 20 patients with episodic chest symptoms and a documented reversibility of at least 12% or 200 ml in FEV1 after 400 μg of inhaled salbutamol or airway hyperresponsiveness (PC20 methacholine or histamine ≤4 mg/ml) (25). Patients with asthma who were on oral steroids were excluded.
Both control groups of 20 subjects each had a negative history on chest symptoms, a postbronchodilator FEV1 >80% predicted, FEV1/FVC >0.70, a less than 12% reversibility in FEV1 after 400 μg salbutamol, and absence of airway hyperresponsiveness (PC20 methacholine or histamine >4 mg/ml). Nonsmoking control subjects had a smoking history of 2 or fewer pack-years and had stopped smoking for at least 1 year. Smoking control subjects had a smoking history of at least 15 pack-years and current smoking of at least 10 cigarettes per day. Subjects of any group were excluded if they had current cancer, cardiovascular disease, systemic infection, diabetes, any other pulmonary disease, or pregnancy. In the case of a respiratory infection, a 4-week recovery period was taken into account.
The study was approved by the Medical Ethics Committees of all participating centers, and all subjects gave their written informed consent. The study was registered at www.trialregister.nl under NTR 1282.
The study had a cross-sectional, case-control design. The measurements were done at one (patients with asthma and patients with COPD) or two (control subjects) visits within 6 weeks as appropriate. Subjects were asked to refrain from eating, drinking, and smoking for at least 90 minutes and to stop short-acting bronchodilators and inhaled corticosteroids for at least 12 hours and long-acting bronchodilators for at least 24 hours on the day of the eNose measurement. Before the methacholine challenge, medication was stopped according to international standardized guidelines (26).
After 5 to 10 minutes of resting, exhaled breath was collected in duplicate with a 5-minute interval and sampled by two different eNose devices in random order (Figure 1). Spirometry was performed, followed by postbronchodilator spirometry or methacholine challenge (whichever was needed).
Symptoms of COPD and asthma were measured by questionnaire (27). Spirometry (MasterscreenPneumo; Jaeger; Würzburg, Germany) was performed by a trained lung function technician according to the latest ERS recommendations (28). FEV1 and FVC were measured before and 10 minutes after 400 μg of inhaled salbutamol. Airway hyperresponsiveness was assessed by methacholine challenge using the standardized tidal breathing method (26).
Exhaled breath analysis was done as previously described (23). In short, patients breathed normally through a mouthpiece with the nose clipped. The mouthpiece was connected to a three-way, nonrebreathing valve and an inspiratory VOC filter (A2; North Safety, Middelburg, The Netherlands) and an expiratory silica reservoir for 5 minutes. After a single deep inspiratory capacity maneuver, the patient exhaled a vital capacity volume into a 10-L Tedlar bag connected to the expiratory port. Within 30 minutes, the electronic nose was connected to the Tedlar bag, followed by a 1-minute sampling of the exhaled air in parallel to sampling a Tedlar bag with VOC-filtered room air for comparison. We used the Cyranose 320 (Smith Detections, Pasadena, CA), a handheld portable chemical vapor analyzer, containing a nanocomposite sensor array with 32 polymer sensors (19). The raw data (changes in electrical resistance of each of the 32 sensors) were stored in the onboard database, copied into an offline database, and used for further analysis with offline pattern-recognition software. Data from every first measurement was disregarded in the analysis because of deviant raw data.
The primary analysis was done by comparing the breathprints between patients with COPD and patients with asthma. A secondary analysis compared these groups with smoking and nonsmoking control subjects, respectively. An exploratory analysis was done by restricting the comparison of smoking control subjects versus patients with COPD to those patients who were current smokers. Similarly, we repeated the analysis in patients using the same maintenance treatment (ICS).
Offline analysis of raw data was performed using SPSS software (version 16.0). Data were reduced by principal component analysis (PCA) from the original 32 sensors to four principal components that captured 98.7% of the variance within the dataset. Using PCA factors based on the complete dataset results in a relatively stable assessment of the weight of the factors because of the variance-covariance structure of the signals. These PCA factors were used to perform a univariate analysis of variance, including a least-significance difference post hoc analysis to assess which PCA factors were discriminative between groups.
Linear canonical discriminant analysis was performed to classify cases into a categorical division. Based on the differentiating PCA factors, a discriminant function was calculated that best distinguished between categories. The accuracy of this model was defined as the percentage of correctly classified patients, cases, and control subjects combined, calculated in the training-set of data used to fit the discriminant analysis model and in the validation-set of data that was not used to fit the discriminant analysis model. Cross-validation using the leave-one-out method was used to calculate the cross-validated accuracy value (%). Within-day repeatability was assessed by comparison of sensor deflections from duplicate measurements using paired t tests and Bland-Altman analysis with Bonferroni correction for multiple testing. Intraclass correlation coefficients were calculated. In a subset of 18 control subjects, between-day biological repeatability was assessed by paired t testing of sensor deflections. Reproducibility was assessed by comparison of two eNose devices using Cohen's kappa analysis.
The sample size estimation was based on the following: The reliability of the percentage (p) correct classification and identification is dependent on the standard error (SE) of the percentage of correctly classified patients. In addition, the SE is a function of p: We aimed to estimate p with an SE less than 6%. If p is between 85 and 100% (22, 23), n = 20 per group is sufficient to achieve this level of SE. A P value less than 0.05 was considered significant.
The subject characteristics of the four groups are described in Table 1. As expected, patients with asthma were younger than patients with COPD, nonsmoking control subjects, and asymptomatic smokers (P < 0.05). There were no significant differences in age between the other three groups. Postbronchodilator FEV1 (% pred) in patients with asthma was significantly higher than in patients with COPD (P < 0.01) but significantly lower than in control subjects and asymptomatic smokers (P < 0.01). No significant differences in FEV1 were observed between smoking and nonsmoking control subjects.
Nonsmoking Control Subjects
Smoking Control Subjects
|Age, years||61.6 ± 9.3*||35.4 ± 15.1||56.7 ± 9.3||56.1 ± 5.9|
|FEV1 postbronchodilator, L||1.72 ± 0.69||3.32 ± 0.86||3.44 ± 0.76||3.58 ± 0.78|
|FEV1 postbronchodilator, % pred||57 ± 15||95 ± 18||116 ± 15||112 ± 14|
|FEV1/FVC||0.45 ± 0.12||0.76 ± 0.11||0.78 ± 0.07||0.77 ± 0.05|
|GINA-classification Mild/moderate/severe, n||NA||11/5/4||NA||NA|
|Pack-years||42.8 ± 17.7||0.6 ± 1.2||0.1 ± 0.2||37.9 ± 17.1|
|ICS use, n||13||18||0||0|
Principal component analysis showed that breathprints from patients with COPD and asthma could be very well distinguished (Figure 2). Subsequent canonical discriminant analysis showed a cross validated accuracy value of 96% (P < 0.0001) (Table 2).
Cross-validated Accuracy (%)
|Asthma–COPD no ICS||95||<0.0001|
|Asthma–nonsmoking control subjects||95||<0.0001|
|COPD–smoking control subjects||66||0.006|
|COPD smoking–smoking control subjects||72||0.018|
|COPD ex-smoking–smoking control subjects||61||0.026|
|COPD ICS–smoking control subjects||70||0.024|
|COPD no ICS–smoking control subjects||65||0.047|
|Control subjects–smoking control subjects||63||0.016|
Breathprints of patients with asthma could also be discriminated from those of nonsmoking control subjects (P < 0.0001; 95%) (Figure 3) and smoking control subjects (P < 0.0001; 93%). Exhaled breath profiles of patients with COPD were separated from those of asymptomatic smokers (P = 0.006) but led to a poor accuracy of cross-validation of 66% (Figure 4) (Table 2). COPD was not distinguishable from nonsmoking control subjects. When all four groups (patients with COPD, patients with asthma, and smoking and nonsmoking asymptomatic control subjects) were analyzed in one model, the cross-validated accuracy model reached only 56%, although this was significant (P < 0.0001).
Nine out of 30 patients with COPD were current smokers. When restricting the analysis to currently smoking patients with COPD, the accuracy of discrimination from asymptomatic smokers improved from 66 to 72% (P = 0.018). No difference in breathprints was found between current smokers and ex-smokers within the COPD group (P = 0.16). Smoking status in patients with COPD did not influence the discrimination from asthma, with 97 and 95% accuracy for smoking and ex-smoking COPD, respectively, compared with 96% for the complete COPD group (Table 2).
Thirteen out of 30 patients with COPD were on ICS treatment. On comparison, no difference in breathprints was found between patients with COPD on ICS treatment and patients not on treatment (P = 0.94). To assess the possible influence of ICS treatment on the breathprint, we restricted the analysis to patients with COPD on ICS treatment, which did not change the cross-validated accuracy of the model in discrimination from asthma (accuracy, 97%). Likewise, the accuracy did not change when patients with COPD not on ICS treatment were compared with asymptomatic smokers (accuracy, 65%) (Table 2).
Breathprints were repeatable and reproducible. Within-day repeatability was assessed by comparison of sensor deflections from two consecutive breathing and sampling procedures. The 95% limits of agreement of all sensors were small, and the sensor responses were not significantly different between the first and second measurements after Bonferroni correction. The intraclass correlation coefficients varied between 0.65 and 0.91 (mean, 0.80). Between-day repeatability in available data of a subset of the healthy control groups (eight smoking subjects, nine nonsmoking subjects) in breath samples taken 1 to 48 days apart (mean, 17 ± 17 d; median, 8 d) showed no differences in breathprints between the two visits. Reproducibility was assessed by comparison of test results from both procedures between the first and second eNose device. This resulted in a Cohen's kappa ranging from 0.75 to 0.91 for comparisons described in Table 2.
Our study shows that fingerprinting of exhaled air by eNose can adequately distinguish between patients with COPD and patients with asthma. Furthermore, the eNose could discriminate these patients from asymptomatic smoking and nonsmoking control subjects. Repeated measurements confirmed this distinction. This indicates that the VOC profiles in the exhaled breath differ between two inflammatory airways diseases, warranting further diagnostic validation of eNoses in COPD and asthma.
To our knowledge, this is the first study in which a group-to-group comparison of exhaled breath profiles was made between well-characterized patients with different obstructive inflammatory airways diseases. In one other study on breathprints in lung cancer, patients with asthma and COPD were included, but no direct comparison between these groups was reported (18). We observed a nearly complete separation of breathprints from patients with COPD and patients with asthma. Our study extends the results by Dragonieri and colleagues, who found that eNose breathprints from patients with severe and mild asthma were clearly distinguishable from those from nonsmoking control subjects (23). Breathprints from patients with COPD could also be distinguished from patients with lung cancer with an overall accuracy of 85% (22). The relatively low, but significant (P < 0.0001), cross-validated accuracy of 56% of the predictive model based on all four groups together in this study reflects the large dispersion of breathprints of COPD and healthy smoking and nonsmoking control subjects combined with a distinct signal for the asthma group. The current distinction of COPD and asthma suggests that eNose technology is a suitable technique for discriminating obstructive airway diseases, which may have clinical and pathophysiological implications.
In this study, we included two clinically relevant groups of patients as commonly seen in daily practice and two asymptomatic control groups. The current guidelines for evaluating diagnostic accuracy are recommended to evaluate the discriminative ability between a priori defined, gold-standard diseased and nondiseased subjects as the first step in the assessment of a novel test (29–31). Our subjects were well characterized according to internationally standardized and accepted guidelines (1, 2). Nevertheless, there were some inevitable differences between the groups. First, the patients with asthma were significantly younger than those with COPD, potentially introducing an age bias into the discrimination of the groups. However, we have recently shown that VOC patterns in exhaled air from young (26 ± 6 yr of age) and older (57 ± 7 yr of age) healthy control subjects could not be distinguished (23). This indicates that age is not a likely confounding factor when analyzing breath by eNose. Nevertheless, we cannot rule out the possibility that factors related to age affect the exhaled breathprints within a disease. Second, we cannot exclude the possibility that the difference in medication usage between the asthma and COPD groups affected the VOC profiles. However, when patients not on ICS treatment were excluded, VOC profiles of patients with asthma and patients with COPD were still separable. Third, even though smoking and nonsmoking control subjects could be separated, the distinction between COPD and asymptomatic smokers was sustained when limiting the analysis to smoking patients with COPD only. Therefore, we believe that the observed differences in VOC profiles between COPD and asthma are disease related.
The breathprints are critically dependent on the methods of collecting and sampling exhaled breath. We used previously validated breathing pattern, inspiratory VOC filtering, drying of the air, and sampling techniques (23) to minimize any influence on the exhaled VOC patterns by expiratory flow rate, humidity, contaminated material, or environmental VOCs. Furthermore, both eNose devices provided highly comparable results, confirming the correspondence between different devices. Even though the sensors of individual eNoses are not identical, such correspondence allows mapping and exchange between devices by integrating response patterns into a unified database (32). This can facilitate future wide-scale application of this technology.
How can we interpret these findings? COPD and asthma are inflammatory airways diseases, with distinct inflammation profiles (3, 7, 8, 33). It is likely that these inflammatory processes generate partly different metabolites, thereby providing different volatile biomarkers in the exhaled air. These are captured in an integrative fingerprint or breathprint by an eNose, making discrimination possible. eNoses cannot identify individual VOCs. eNose technology uses high-throughput methods that are comparable to “omics” techniques such as metabolomics. To unravel which VOCs drive the distinctive patterns between COPD and asthma, specific analysis of individual compounds is necessary. This requires GC-MS analysis (34) or nuclear magnetic resonance spectroscopy (12, 34, 35), which should be applied in studies focusing on pathophysiological pathways and in the development of more “specialized” eNoses detecting a smaller but more specific subset of VOCs for a certain disease.
The complex chronic airways inflammation in COPD and asthma is likely to change such metabolic pathways, thereby affecting the exhaled molecular markers (36). Airway inflammation in asthma is predominated by activated CD4-positive T cells, mast cells, and eosinophils, with an increase in Th2 cytokines such as IL-4, IL-5, and IL-13 (37). COPD, on the other hand, is characterized by CD8-positive T cells, increased numbers of neutrophils and macrophages, and an increase in inflammatory mediators such as IL-8, IL-1, leukotriene B4, and IFN-γ (38). Preliminary data from our group show an association between sputum mediators and exhaled breathprint in patients with mild to moderate COPD (39). These differences in inflammatory profile are a likely explanation for the difference in VOC profile between the diseases. It is also possible that part of the VOCs in exhaled breath are of systemic rather than pulmonary origin and are exhaled after alveolar diffusion (15). This view is in accordance with the hypothesis that asthma (40) and COPD (41, 42) can have systemic manifestations with a partially different profile rather than being localized in the airways only (43). Most likely, the ability of the eNose to distinguish exhaled breath VOC patterns from subjects with asthma and COPD and asymptomatic smoking and nonsmoking control subjects is based on a combination of the above-mentioned mechanisms.
Smoking itself is also known to induce an inflammatory response (44) in the airways, with recruitment of inflammatory cells such as neutrophils and macrophages. In addition, oxidative stress may also contribute to the exhaled VOC-profile. Reactive oxygen species produced by inflammatory cells or directly from cigarette smoke, result in lipid peroxidation producing ethane and pentane. These volatile hydrocarbons can be detected in exhaled breath (15, 45). The electronic nose was able to distinguish between asymptomatic smokers and smoking patients with COPD, even though the cross-validation value was only 72%. This suggests that either health-related or smoking-induced VOCs are partially similar to those produced in COPD. More importantly, the overlap between asymptomatic smokers and patients with COPD may be reflective of the presence of future patients with COPD among the “healthy” smokers, who are still asymptomatic with a normal lung function. This warrants assessing the prognostic value of breathprints in asymptomatic smokers in follow-up studies. On the other hand, when comparing breathprints from nonsmoking control subjects and patients with COPD, no difference was found. This is in agreement with other studies (18), and might be explained by the absence of actual active inflammation in this group of patients with COPD. Therefore, new studies including GC-MS should be undertaken to examine which components in the exhaled breath contribute to the strong signal in the active inflammatory state in asthma as compared with healthy control subjects and COPD.
What is the clinical implication of our findings? First, COPD and asthma are different diseases with partly overlapping characteristics (6). Using a handheld eNose device, it appears to be possible to distinguish both diseases in a noninvasive manner. This makes eNose technology an appealing alternative to metabolomic analysis of exhaled breath condensate, which requires highly sensitive laboratory assays (12, 35). Our results warrant the next step in the validation of the eNose for the diagnostic assessment of COPD and asthma by strictly following the STARD guidelines to test the diagnostic accuracy (29–31). Second, it should be examined whether clinically relevant (sub)phenotypes of COPD and asthma can be discriminated and whether the clinical course of the diseases can be adequately monitored. Third, our findings are also in support of validating breathprints in the risk assessment of asymptomatic smokers for COPD. If this is successful, eNoses have the potential to be widely used in the field of respiratory medicine because of their ease of use, noninvasive character, and rapid results (21). GC-MS analysis should be applied to unravel the pathophysiological pathways involved.
The authors thank the participants in the study, Marianne Smink, Haga Hospital; Den Haag and Kees Kanters, M.D.; Henk Brouwer, Ph.D.; and Prof. Patrick Bindels, M.D., Ph.D. from the Department of Primary Care, Academic Medical Centre, Amsterdam for their help in recruiting the patients.
|1.||Rabe KF, Hurd S, Anzueto A, Barnes PJ, Buist SA, Calverley P, Fukuchi Y, Jenkins C, Rodriguez-Roisin R, van Wheel C, Zielinski J. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med 2007;176:532–555.|
|2.||Bateman ED, Hurd SS, Barnes PJ, Bousquet J, Drazen JM, FitzGerald M, Gibson P, Ohta K, O'Byrne P, Pedersen SE, et al. Global strategy for asthma management and prevention: GINA executive summary. Eur Respir J 2008;31:143–178.|
|3.||Fabbri LM, Romagnoli M, Corbetta L, Casoni G, Busljetic K, Turato G, Ligabue G, Ciaccia A, Saetta M, Papi A. Differences in airway inflammation in patients with fixed airflow obstruction due to asthma or chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2003;167:418–424.|
|4.||Decramer M, Selroos O. Asthma and COPD: differences and similarities. With special reference to the usefulness of budesonide/formoterol in a single inhaler (Symbicort) in both diseases. Int J Clin Pract 2005;59:385–398.|
|5.||Guerra S. Overlap of asthma and chronic obstructive pulmonary disease. Curr Opin Pulm Med 2005;11:7–13.|
|6.||Gibson PG, Simpson JL. The overlap syndrome of asthma and COPD: what are its features and how important is it? Thorax 2009;64:728–735.|
|7.||Bhattacharya S, Srisuma S, Demeo DL, Shapiro SD, Bueno R, Silverman EK, Reilly JJ, Mariani TJ. Molecular biomarkers for quantitative and discrete COPD phenotypes. Am J Respir Cell Mol Biol 2009;40:359–367.|
|8.||Brasier AR, Victor S, Boetticher G, Ju H, Lee C, Bleecker ER, Castro M, Busse WW, Calhoun WJ. Molecular phenotyping of severe asthma using pattern recognition of bronchoalveolar lavage-derived cytokines. J Allergy Clin Immunol 2008;121:30–37.|
|9.||Weston AD, Hood L. Systems biology, proteomics, and the future of health care: toward predictive, preventative, and personalized medicine. J Proteome Res 2004;3:179–196.|
|10.||Crameri R. The potential of proteomics and peptidomics for allergy and asthma research. Allergy 2005;60:1227–1237.|
|11.||Sepper R, Prikk K. Proteomics: is it an approach to understand the progression of chronic lung disorders? J Proteome Res 2004;3:277–281.|
|12.||Carraro S, Rezzi S, Reniero F, Heberger K, Giordano G, Zanconato S, Guillou C, Baraldi E. Metabolomics applied to exhaled breath condensate in childhood asthma. Am J Respir Crit Care Med 2007;175:986–990.|
|13.||Pauling L, Robinson AB, Teranishi R, Cary P. Quantitative analysis of urine vapor and breath by gas-liquid partition chromatography. Proc Natl Acad Sci USA 1971;68:2374–2376.|
|14.||Moser B, Bodrogi F, Eibl G, Lechner M, Rieder J, Lirk P. Mass spectrometric profile of exhaled breath–field study by PTR-MS. Respir Physiol Neurobiol 2005;145:295–300.|
|15.||Buszewski B, Kesy M, Ligor T, Amann A. Human exhaled air analytics: biomarkers of diseases. Biomed Chromatogr 2007;21:553–566.|
|16.||Phillips M, Altorki N, Austin JH, Cameron RB, Cataneo RN, Kloss R, Maxfield RA, Munawar MI, Pass HI, Rashid A, et al. Detection of lung cancer using weighted digital analysis of breath biomarkers. Clin Chim Acta 2008;393:76–84.|
|17.||Gaspar EM, Lucena AF, Duro da Costa J, Chaves das Neves H. Organic metabolites in exhaled human breath–a multivariate approach for identification of biomarkers in lung disorders. J Chromatogr A 2009;1216:2749–2756.|
|18.||Machado RF, Laskowski D, Deffenderfer O, Burch T, Zheng S, Mazzone PJ, Mekhail T, Jennings C, Stoller JK, Pyle J, et al. Detection of lung cancer by sensor array analyses of exhaled breath. Am J Respir Crit Care Med 2005;171:1286–1291.|
|19.||Lewis NS. Comparisons between mammalian and artificial olfaction based on arrays of carbon black-polymer composite vapor detectors. Acc Chem Res 2004;37:663–672.|
|20.||Scott SM, James D, Ali Z. Data analysis for electronic nose systems. Microchim Acta 2006;156:183–207.|
|21.||Friedrich MJ. Scientists seek to sniff out diseases: electronic “noses” may someday be diagnostic tools. JAMA 2009;301:585–586.|
|22.||Dragonieri S, Annema JT, Schot R, van der Schee MP, Spanevello A, Carratu P, Resta O, Rabe KF, Sterk PJ. An electronic nose in the discrimination of patients with non-small cell lung cancer and COPD. Lung Cancer 2009;64:166–170.|
|23.||Dragonieri S, Schot R, Mertens BJ, Le CS, Gauw SA, Spanevello A, Resta O, Willard NP, Vink TJ, Rabe KF, et al. An electronic nose in the discrimination of patients with asthma and controls. J Allergy Clin Immunol 2007;120:856–862.|
|24.||Fens N, de Nijs SB, Roldaan AC, Dijkers E, Zwinderman AH, Bel EH, Sterk PJ. Successful distinction between asthma and COPD by molecular profiling of exhaled breath using electronic nose. Am J Respir Crit Care Med 2009;179:A5180.|
|25.||Bateman ED, Bousquet J, FitzGerald M, Haahtela T, O'Byrne P, Ohta K, Paggiaro P, Pedersen SE, Soto-Quiroz M, Tan W-C, et al.; Global Initiative for Asthma. GINA report, global strategy for asthma management and prevention [Internet]. Bethesda, MD: NHLBI; 2008 (accessed 2009 May 1). Available from:|
|26.||Crapo RO, Casaburi R, Coates AL, Enright PL, Hankinson JL, Irvin CG, MacIntyre NR, McKay RT, Wanger JS, Anderson SD, et al. Guidelines for methacholine and exercise challenge testing-1999. Am J Respir Crit Care Med 2000;161:309–329.|
|27.||Burney PG, Luczynska C, Chinn S, Jarvis D. The European Community respiratory health survey. Eur Respir J 1994;7:954–960.|
|28.||Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, Crapo R, Enright P, van der Grinten CP, Gustafsson P, et al. Standardisation of spirometry. Eur Respir J 2005;26:319–338.|
|29.||Knottnerus JA, Muris JW. Assessment of the accuracy of diagnostic tests: the cross-sectional study. J Clin Epidemiol 2003;56:1118–1128.|
|30.||Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, Moher D, Rennie D, de Vet HC, Lijmer JG. The STARD statement for reporting studies of diagnostic accuracy: explanation and elaboration. Ann Intern Med 2003;138:W1–W12.|
|31.||Gluud C, Gluud LL. Evidence based diagnostics. BMJ 2005;330:724–726.|
|32.||Shaham O, Carmel L, Harel D. On mappings between electronic noses. Sensors Actuators B 2005;106:76–82.|
|33.||Mauad T, Dolhnikoff M. Pathologic similarities and differences between asthma and chronic obstructive pulmonary disease. Curr Opin Pulm Med 2008;14:31–38.|
|34.||Van Berkel JJ, Dallinga JW, Moller GM, Godschalk RW, Moonen E, Wouters EF, Van Schooten FJ. Development of accurate classification method based on the analysis of volatile organic compounds from human exhaled air. J Chromatogr B Analyt Technol Biomed Life Sci 2008;861:101–107.|
|35.||de Laurentiis G, Paris D, Melck D, Maniscalco M, Marsico S, Corso G, Motta A, Sofia M. Metabonomic analysis of exhaled breath condensate in adults by nu2clear magnetic resonance spectroscopy. Eur Respir J 2008;32:1175–1183.|
|36.||Kharitonov SA, Barnes PJ. Exhaled markers of pulmonary disease. Am J Respir Crit Care Med 2001;163:1693–1722.|
|37.||Anderson GP. Endotyping asthma: 2new insights into key pathogenic mechanisms in a complex, heterogeneous disease. Lancet 2008;372:1107–1119.|
|38.||Chung KF, Adcock IM. Multifaceted mechanisms in COPD: inflammation, immunity, and tissue repair and destruction. Eur Respir J 2008;31:1334–1356.|
|39.||Fens N, de Nijs SB, Knobel H, Willard N, Verhaegh W, Vink A, Lutter R, Dijkers E, Krouwels FH, Sterk PJ. Exhaled breath molecular profiling using an electronic nose identifies eosinophilic inflammation in COPD [Abstract]. Am J Respir Crit Care Med 2009;179:A4324.|
|40.||Jousilahti P, Salomaa V, Hakala K, Rasi V, Vahtera E, Palosuo T. The association of sensitive systemic inflammation markers with bronchial asthma. Ann Allergy Asthma Immunol 2002;89:381–385.|
|41.||Higashimoto Y, Iwata T, Okada M, Satoh H, Fukuda K, Tohda Y. Serum biomarkers as predictors of lung function decline in chronic obstructive pulmonary disease. Respir Med (In press)|
|42.||Wouters EF. Chronic obstructive pulmonary disease. 5: systemic effects of COPD. Thorax 2002;57:1067–1070.|
|43.||Higashimoto Y, Yamagata Y, Taya S, Iwata T, Okada M, Ishiguchi T, Sato H, Itoh H. Systemic inflammation in chronic obstructive pulmonary disease and asthma: Similarities and differences. Respirology 2008;13:128–133.|
|44.||Morrison D, Rahman I, Lannan S, MacNee W. Epithelial permeability, inflammation, and oxidant stress in the air spaces of smokers. Am J Respir Crit Care Med 1999;159:473–479.|
|45.||Paredi P, Kharitonov SA, Barnes PJ. Elevation of exhaled ethane concentration in asthma. Am J Respir Crit Care Med 2000;162:1450–1454.|