Published ahead of print on February 15, 2007, doi:10.1164/rccm.200606-769OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200606-769OC
Metabolomics Applied to Exhaled Breath Condensate in Childhood Asthma1 Department of Pediatrics, University of Padova, Padova, Italy; and 2 European Commission, Joint Research Centre, Institute for Health and Consumer Protection, Physical and Chemical Exposure Unit, Ispra (VA), Italy Correspondence and requests for reprints should be addressed to Eugenio Baraldi, M.D., Department of Pediatrics, Allergy and Respiratory Medicine Unit, Via Giustiniani 3, 35128 Padova, Italy. E-mail: baraldi{at}pediatria.unipd.it
Rationale: Metabolomic analysis provides biochemical profiles of low-molecular-weight endogenous metabolites in biological fluids. Objectives: The aim of this study was to assess the feasibility of nuclear magnetic resonance (NMR)–based metabolomic analysis applied to exhaled breath condensate ("breathomics"). Information coming from NMR spectra was analyzed with a view to establish the NMR variables that best discriminate between children with asthma and healthy children. Methods: Twenty-five children with asthma (17 with persistent asthma treated with inhaled corticosteroids, 8 with intermittent asthma inhaled corticosteroid naive; age, 7–15 yr) and 11 healthy age-matched control subjects were enrolled. Every child performed exhaled nitric oxide measurement, spirometry, and condensate collection. Condensate samples were analyzed by means of NMR spectroscopy. Linear and partial least squares discriminant analyses were applied to data obtained from the NMR spectra.
Measurements and Main Results: The combination of exhaled nitric oxide and FEV1 discriminates children with asthma and healthy children with a success rate of approximately 81%, whereas selected signals from NMR spectra offer a slightly better discrimination ( Conclusions: Metabolomics can be applied to exhaled breath condensate, leading to the characterization of airway biochemical fingerprints. The presence of acetylated compounds suggests new metabolic pathways that may have a role in asthma pathophysiology.
Key Words: asthma • metabolomics • exhaled breath condensate • children • nuclear magnetic resonance
Asthma is a chronic inflammatory disease of the airways (1), and is currently considered a major health problem because its prevalence has increased in all populations, especially among children (2). The inflammatory state of asthmatic airways is weakly correlated with clinical symptoms and lung function parameters, and even clinically quiescent asthma can be associated with airway inflammation (3). Moreover, several inflammatory patterns are involved in asthma and contribute in characterizing different clinical phenotypes (4). The knowledge of airway inflammatory state therefore has a central role in asthma management. Recently, the availability of noninvasive techniques, such as fractional exhaled nitric oxide (FENO) measurement and exhaled breath condensate (EBC) analysis, has created new opportunities for investigating and monitoring airway inflammation. EBC is obtained by cooling exhaled air and it contains several biocompounds that are believed to reflect airway lining fluid composition (5). The analysis and interpretation of global metabolic data (which represent the expression of the multiparametric metabolic response of living systems to pathophysiological stimuli) by means of modern spectroscopic techniques and appropriate statistical approaches, are defined as "metabolic profiling," "metabonomics," or "metabolomics." The current article uses the term "metabolomics" (6–9). High-resolution proton nuclear magnetic resonance (1H-NMR) spectroscopy is one of the most powerful techniques for metabolite profile detection (10). This analytical technique enables the characterization of the most represented proton-containing low-molecular-mass compounds and their representation in a spectrum, providing a metabolic fingerprint of the sample analyzed. Multivariate statistical methods (usually termed as pattern recognition methods) are then applied to extract information from these complex NMR spectra (11). So far, EBC has been studied to detect known target compounds (e.g., leukotrienes, oxidative-stress markers, nitrogen oxide–related products) in subjects with asthma (5). Metabolomics may offer a completely new approach to EBC analysis, enabling not only the detection of known metabolites but also the prediction of unknown metabolites and novel biomarkers, which may provide insight into disease mechanisms. NMR may provide objective, reproducible patterns formed by the chemical constituents of the EBC. These patterns, though not always indicative of a specific identified chemical component, may nonetheless be useful in discriminating between disease and health. Moreover, the chemical characteristics revealed in the NMR profiles provide clues as to underlying metabolic derangements associated with asthma, and such findings may help us to subphenotype the asthma syndrome into its component disease processes. Metabolomics is usually conducted on urine and blood but other biofluids and cell cultures can be used. The aim of our study was to assess the feasibility of applying NMR-based metabolomic analysis to EBC. Information coming from NMR spectra was analyzed to establish the NMR variables that best discriminate between children with asthma and healthy children. Some of the results of this study have been previously reported in the form of an abstract (12).
Study Subjects Twenty-five children with controlled allergic asthma and 11 age-matched healthy control children were enrolled. Eight children with asthma had intermittent asthma and had received no inhaled corticosteroid treatment for at least a month. Seventeen children had persistent asthma and were treated with inhaled corticosteroids on a regular basis. The diagnosis of asthma was made by a pediatric respiratory physician and was based on clinical history (cough, shortness of breath, recurrent wheeze, chest tightness) and increase in FEV1 after bronchodilator of 12% or more, according to international guidelines (2). See the online supplement for additional information. At recruitment, children underwent physical examination and FENO measurement and spirometry were performed. EBC was collected, stored at –80°C, and subsequently analyzed by proton NMR. The ethics committee of our hospital reviewed and approved the protocol and all parents gave their informed consent.
FENO and Pulmonary Function Measurement
EBC Collection To assess within-day repeatability, six children with asthma were asked to undergo EBC collection twice within the same day.
NMR Measurements Spectra were segmented into 101 chemical shift regions ("bucketing" procedure), 0.04 ppm wide, using the AMIX (Bruker) software package, release 3.5.6. A region of 0.5–4.5 ppm was considered, excluding lactate signals (quartet at 4.12 ppm and doublet at 1.33 ppm). The resulting integrated regions (buckets) were used for statistical analysis. Each bucket was integrated and normalized using the total intensity of the spectrum. NMR analysis was performed without knowing which group (healthy or asthmatic) each child belonged to.
Statistical Analysis Within-day repeatability and technical repeatability of the measurements were evaluated using the method of Bland and Altman (18) and the coefficient of variation, respectively. See the online supplement for additional information.
Mean FEV1 was significantly lower in children with asthma (median, 87% predicted; interquartile range [IQR], 78–97%) than in healthy children (median, 103% predicted; IQR, 95–106%; p < 0.01). FENO was significantly higher in children with asthma (median, 30.5 ppb; IQR, 16.4–60.6 ppb) than in healthy children (median, 9.6 ppb; IQR, 6.8–12.9 ppb; p < 0.001). Linear discriminant analysis (LDA) demonstrated that, in combination, FENO and FEV1 have a success rate of approximately 81% in discriminating between children with asthma and healthy children, whereas selected NMR signals discriminate the two groups with a success rate of approximately 86%. Moreover, LDA using best-subset selection discriminated children with asthma and healthy children with a success rate of more than 95%. An even better classification can be achieved at the risk of overfit (by modeling the noise as well). The results were verified against another (independent) statistical technique (i.e., PLS), which is not sensitive to overfit, and clinical data were used in addition to all NMR variables. Considering three PLS components, we obtained a classification that was approximately 95% successful, with only one clearly misclassified sample: an asthmatic sample was classified as healthy (Figure 1).
We also considered performing a subgroup analysis to compare steroid-treated and untreated children, but we were unable to do so because the number of samples in each group was too small for reliable statistical analysis. On the basis of the results of the above statistical analysis, the most important signals for the classification of healthy patients and patients with asthma come primarily from the region of 1.7 to 2.2 ppm, and secondarily from the region of 3.2 to 3.4 ppm (Figure 2). The first set of signals due to the chemical shift and the shape of the single signals can be assumed to correspond to acetylated products. The second set can be attributed to oxygen-containing compounds.
A good within-day repeatability was demonstrated by means of the Bland-Altman test. Moreover, a good technical repeatability was shown, with a coefficient of variation of 1.07%. See the online supplement for additional information.
In our study, we showed that NMR-based metabolomic analysis can be applied to an EBC matrix (breathomics); to the best of our knowledge, there are no previously published data on the analysis of this biofluid using NMR-based spectroscopy. Within the NMR spectrum, we identified some profiles indicative of acetylated and oxidized compounds that significantly distinguished children with asthma from healthy children (Figure 2). The development and application of techniques enabling the study of underlying metabolic processes are important in improving our understanding of asthma (2). Although the gold standard for investigating airway inflammation in vivo is bronchoscopy with bronchoalveolar lavage and bronchial biopsy, this invasive technique is not suitable for routine use, particularly in children (19, 20). Growing interest has therefore focused recently on noninvasive techniques, such as exhaled NO and EBC analysis (19). The EBC technique is entirely noninvasive, safe, and easy to perform, even in children (20). Several studies have investigated EBC for the presence of known compounds believed to mirror different pathways involved in asthmatic airway inflammation and oxidative stress (19). However, the application of EBC is dependent on the development of sensitive and reproducible assays (21). Metabolomics has been gaining importance in quantitative measurement of the multivariate metabolic responses of whole organisms to pathophysiological stimuli (7–10), and it has recently been applied in studies investigating the metabolic response to various diseases, such as ovarian cancer (22), diabetes (23), and inherited metabolic disorders (24), but not asthma. NMR spectroscopy has recently been applied successfully to urine, demonstrating that drug-induced responses in individuals are potentially predictable from their predose metabolic profiles (pharmaco-metabolomics) (25). NMR spectroscopy is very powerful in providing overall biochemical profiles of low-molecular-weight endogenous metabolites in biological fluids, without requiring any preselection of measurable analytes (24, 26, 27). The biochemical compounds detected are represented in a spectrum consisting of thousands of signals, the intensity of each of which correlates with the metabolite's concentration. Metabolomics thus provides an overview of all the compound classes in a sample, drawing its metabolic fingerprint. Multivariate statistical methods (11) are applied to NMR-generated profiles to extract latent biochemical information from NMR spectra. Together with clinical symptoms, FEV1, a parameter of airflow limitation (2), and FENO, a marker of eosinophilic airway inflammation (28), are currently considered as functional, biological markers for asthma diagnosis and monitoring. Our data notably show that, in combination, FENO and FEV1 have a success rate of approximately 81% in discriminating between children with asthma and healthy children, whereas selected NMR signals discriminate between them with a success rate of approximately 86%. Moreover, LDA, using best-subset selection (a combination of FEV1 and NMR signals), produced a classification with a success rate of more than 95%. These observations suggest that NMR-based EBC analysis is effective in characterizing subjects with asthma and support a role for this technique in further studies aimed at investigating asthma pathobiology. The NMR variables (signals) identified by LDA and PLS are derived primarily from the region of 1.7 to 2.2 ppm (Figure 2), and can be attributed to acetylated products. To our knowledge, there are no published data on acetylated compounds in the exhaled air of subjects with asthma. We could not characterize the exact molecules represented by the peak identified in the NMR spectrum because the structural identification of a single metabolite requires more complex procedures, such as homo- and hetero-NMR bidimensional spectra or high-resolution mass spectrometry, and the application of such procedures was beyond the aim of the present study. We can, therefore, only make some speculations on this new finding. It has recently been demonstrated that, in response to several inflammatory signals, the high motility group box protein 1 (HMGB1) is acetylated and then actively secreted (29). In the extracellular space, HMGB1 behaves like a proinflammatory mediator (29). We hypothesize that the large number of inflammatory stimuli in the asthmatic airways might increase HMBG1 acetylation and release in the airway, where this protein might act as a proinflammatory agent. The peaks in the 3.2- to 3.4-ppm range of the NMR spectrum, which secondarily characterized children with asthma, are probably related to oxidized compounds. Oxidative stress (resulting from an imbalance between oxidants and antioxidants in the airways) has a significant role in asthma pathophysiology and lung damage (19). Our finding is consistent with numerous studies showing higher concentrations of oxidative stress markers in EBC collected from subjects with asthma (30), and it supports the need for further studies to investigate the therapeutic role of effective antioxidant agents. NMR metabolomics may reach a wider application in pulmonology. For example, further studies applying this method to the analysis of EBC might enable the identification of early markers of lung cancer or, as recently reported, the prediction of the response to drugs (25). When a new technique is described, the availability of repeatability data is a central issue (21). A recent study (31) addressed this problem concluding that NMR spectroscopy of biofluids combined with pattern recognition methods is a robust and precise approach for metabolomic studies. Our data confirm these observations, showing a good within-day repeatability when two EBC samples are collected from the same patient and a good technical repeatability when the NMR analysis is repeated eight consecutive times on the same sample. The present study determined the parts of the spectra that may help us discriminate between healthy and ill cases, but the small sample size prevented us from making any predictions in an independent group of subjects. We therefore recognize that our data are preliminary and further studies are needed to determine prospectively, in a separate group of patients, the utility of the model in discriminating between children with asthma and healthy children. In conclusion, the major novel finding emerging from our study is that metabolomic analysis can be applied to EBC, providing the opportunity to obtain overall information on the biochemical compounds in a biological fluid collected from the lower airway by noninvasive means. In patients with asthma, the application of metabolomics to the EBC matrix has the potential for representing a significant advance in defining the biochemical phenotype of the airways. The presence of acetylated compounds suggested by NMR profiling paves the way for the study of new metabolic pathways that may have a role in asthma pathophysiology.
The authors thank Dr. Manfred Spraul and Dr. Li-Hong Tseng of Bruker BioSpin GmbH for NMR measurements.
* These investigators contributed equally to the article.  The present address for S.R. is BioAnalytical Science, Metabonomics & Biomarkers, Nestlé Research Center, P.O. Box 44, CH-1000 Lausanne 26, Switzerland. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200606-769OC on February 15, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form June 9, 2006; accepted in final form February 15, 2007
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
86%). The selected NMR variables derive from the region of 3.2 to 3.4 ppm, indicative of oxidized compounds, and from the region of 1.7 to 2.2 ppm, indicative of acetylated compounds. 
