METABOLOMICS APPLIED TO EXHALED BREATH CONDENSATE IN

CHILDHOOD ASTHMA.

Silvia Carraro1, 4, Serge Rezzi2,3, 4, Fabiano Reniero2, Károly Héberger2,

Giuseppe Giordano1, Stefania Zanconato1, Claude Guillou2, Eugenio Baraldi1

1 Department of Pediatrics, University of Padova, Via Giustiniani, 3, 35128 Padova,

Italy 2 European Commission, Joint Research Centre, Institute for Health and Consumer

Protection, Physical and Chemical Exposure Unit, BEVABS T.P. 740, 21020 Ispra

(VA), Italy 3 Actual address: BioAnalytical Science, Metabonomics & Biomarkers, Nestlé

Research Center PO Box 44, CH-1000 Lausanne 26 4 These authors contributed equally to the work

Corresponding author: Dr. Eugenio Baraldi Department of Pediatrics Allergy and Respiratory Medicine Unit Via Giustiniani 3 35128 Padova Italy Phone: 049 8213560 Fax: 049 8213502 E-mail: baraldi@pediatria.unipd.it Impact of this research on clinical medicine Here we apply, for the first time, metabolomics in the analysis of exhaled breath condensate in asthmatic children. Metabolomics allows both the detection of known metabolites and the prediction of unknown metabolites. Characterizing the sample overall biochemical profile, this technique may represent a significant advance in defining the biochemical phenotype in asthmatic patients. Running title: Metabolomics in childhood asthma

Subject category number: 66 Word count for the body manuscript: 2419

This article has an online data supplement, which is accessible from this issue’s table

of content online at www.atsjournal.org.

AJRCCM Articles in Press. Published on February 15, 2007 as doi:10.1164/rccm.200606-769OC

Copyright (C) 2007 by the American Thoracic Society.

1

ABSTRACT Rationale and Objectives: Metabolomic analysis provides biochemical profiles of

low molecular weight endogenous metabolites in biological fluids. The aim of this

study was to assess the feasibility of Nuclear Magnetic Resonance (NMR) based

metabolomic analysis applied to exhaled breath condensate. Information coming

from NMR spectra was analyzed with a view to establish the NMR variables that best

discriminate between asthmatic and healthy children.

Methods: 25 children with asthma (17 with persistent asthma treated with inhaled

corticosteroids, 8 with intermittent asthma inhaled corticosteroid naïve, age 7-15

years) and 11 healthy age-matched controls were enrolled. Every child performed

exhaled nitric oxide measurement, spirometry and condensate collection.

Condensate samples were analysed by means of NMR spectroscopy (Bruker,

BioSpin, 600 MHz 3 mm cryogenic probe at 300 k). Linear and partial least squares

discriminant analysis were applied to data obtained from the NMR spectra.

Results: The combination of exhaled nitric oxide and FEV1 discriminates asthmatic

and healthy children with a success rate of ~81%, while selected signals from NMR

spectra offer a slightly better discrimination (~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.

Conclusion: 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.

Abstract Word count: 236

Key words: asthma, metabolomics, exhaled breath condensate, children, nuclear

magnetic resonance

2

INTRODUCTION

Asthma is a chronic inflammatory disease of the airways (1) that is nowadays

considered a major health problem since its prevalence has increased everywhere,

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 to characterize different

clinical phenotypes (4). The knowledge of airway inflammatory state has therefore a

central role in asthma management. In the last years, the availability of non-invasive

techniques, as exhaled NO (FENO) measurement and exhaled breath condensate

(EBC) analysis, has opened new prospective 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, expression of the

multiparametric metabolic response of living systems to pathophysiological stimuli, by

means of modern spectroscopic techniques and appropriate statistical approaches, is

defined as 'metabolic profiling’, 'metabonomics’ or 'metabolomics’. In the present

paper we use the term metabolomics (6-9). High resolution 1H nuclear magnetic

resonance (NMR) spectroscopy is one of the most powerful technique 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).

3

So far, EBC has been studied to detect known target compounds (e.g.

leukotrienes, oxidative-stress markers, nitrogen oxide related products) in asthmatic

subjects (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 syndrome asthma 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

with a view to establishing the NMR variables that best discriminate between

asthmatic and healthy children.

Some of the results of this study have been previously reported in the form of

an abstract (12).

METHODS

Study subjects

25 asthmatic allergic children with controlled disease and 11 age-matched healthy

controls were enrolled. Eight asthmatic children had intermittent asthma and had

received no inhaled corticosteroid (ICS) treatment for at least a month. Seventeen

4

children had persistent asthma and were treated with ICS on a regular base. 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 > 12%, according to international

guidelines (2). See the online supplement for additional information.

At recruitment children underwent physical examination and performed FENO

measurement and spirometry. 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.

Fractional exhaled nitric oxide (FENO) and pulmonary function measurement

FENO was measured with the NIOX system (Aerocrine, Stockholm, Sweden),

following the ERS/ATS guidelines for measuring FENO in children (13). Lung function

was measured by means of a 10-liter bell spirometer (Biomedin, Padova, Italy). See

the online supplement for additional information.

EBC collection

EBC was collected by means of a condenser consisting of five components:

a mouthpiece set up to work as a saliva trap too, a non-rebreathing polypropylene

valve, a 10 cm Tygon® tube, a 50 ml polypropylene vial and a Dewar flask

refrigerated with ice. Children breathed tidally through the mouth for 15 minutes,

while sitting comfortably and wearing a nose clip. They maintained a dry mouth

during collection by periodically swallowing excess saliva. EBC samples were

stored at –80°C in polypropylene tubes until assay.

To assess within-day repeatability, 6 asthmatic children were asked to

collect EBC twice within the same day.

5

NMR measurements

The EBC samples were dried with a speed vacuum system (UNIVAPO 150

ECH, UNIEQUIP, Munich, Germany) to small volume (10µl) and added to D2O for

NMR analysis (80 µl final volume). NMR measurements were taken on a 600MHz

Bruker BioSpin spectrometer, using a 3mm cryogenic probe, setting the temperature

at 300K. 1D-spectra were acquired with water suppression, using a modified 1-

dimensional Noesy sequence with 2 RF-Gradients (gradient ratio 50:10) in non-

spinning mode, acquiring 512 scans with a 32K time domain data size, a recycle time

of ~ 3.9 sec, and a mixing time of 100 msec. Samples were measured in 2.5 mm

NMR tubes using 80 microliters of sample, processed using an exponential

multiplication and a line-broadening factor of 1. Spectra were calibrated to TSP at 0

ppm (parts per million) (14). The resulting spectra were corrected for phase and

baseline using the TOPSPIN software package (Bruker), rel. 1.2.b.

Spectra were segmented into 101 chemical shift regions (“bucketing”

procedure) 0.04 ppm wide using the AMIX (Bruker) software package, rel. 3.5.6 and

considering the region 0.5 – 4.5 ppm, 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

Data from the NMR spectra were processed using the AMIX Principal

Component Analysis (PCA) routine and then, after exportation, using the Statistica

6.0 software package for Discriminant Analysis (DA) and Partial least squares (PLS)

(15-17).

6

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.

RESULTS

Mean FEV1 was significantly lower in asthmatic (median 87% pred.,

interquartile range (IQR) 78%-97%) than in healthy children (median 103% pred, IQR

95%-106%, p<0.01). FENO was significantly higher in asthmatic (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).

LDA (linear discriminant analysis) demonstrated that, in combination, FENO

and FEV1 have a success rate of ~81% in discriminating between asthmatic and

healthy children, while selected NMR signals discriminate the two groups with a

success rate of ~86%.

Moreover, LDA using best subset selection discriminated asthmatic and

healthy children with a success rate in excess of 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 as well as all

NMR variables. Considering three PLS components, we obtained a classification that

was around 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.

7

Based on the results of the above statistical analysis, the most important

signals for the classification of healthy and asthmatic patients 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.

DISCUSSION

In our study, we showed that NMR-based metabolomic analysis can be

applied to an EBC matrix and, 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 asthmatic from healthy children (figure 2).

The development and application of techniques enabling the study of

underlying metabolic processes are important to improve 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 non-invasive techniques, such as exhaled

NO and EBC analysis (19). The EBC technique is entirely non-invasive, 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

8

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 more and more 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 pre-dose

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 pre-selection 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.

Alongside clinical symptoms, FEV1 - a parameter of airflow limitation (2) - and

FENO - a marker of eosinophilic airway inflammation (28) - are nowadays considered

as functional and biological markers for asthma diagnosis and monitoring. Our data

notably show that, in combination, FENO and FEV1 have a success rate of ~81% in

discriminating between asthmatic and healthy children, while selected NMR signals

discriminate between them with a success rate of ~86%. Moreover, LDA (Linear

Discriminant Analysis) using best subset selection (a combination of FEV1 and NMR

signals) produced a classification with a success rate in excess of 95%. These

9

observations suggest that NMR-based EBC analysis is effective in characterizing

asthmatic subjects and supports a role for this technique in further studies aimed at

investigating asthma pathobiology.

The NMR variables (signals) identified by LDA and PLS, 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 asthmatic subjects. 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 aims 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 pro-inflammatory

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 pro-inflammatory agent.

The peaks in the 3.2-3.4 ppm range of the NMR spectrum, which secondarily

characterized asthmatic children, are probably related to oxidized compounds.

Oxidative stress (resulting from an imbalance between oxidants and anti-oxidants 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 asthmatic subjects (30), and it

supports the need for further studies to investigate the therapeutic role of effective

antioxidant agents.

10

In fact, 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 8 times in a row on the same sample.

A limit of the present study is that we could only evaluate which parts of the

spectra may help us to discriminate between healthy and ill cases, while the limited

number of samples 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 built model in discriminating between asthmatic and healthy children.

In conclusion, the major novelty emerging from our study is that metabolomic

analysis can be applied to EBC giving the opportunity to obtain overall information on

the biochemical compounds in a biological fluid collected from the lower airway by

non-invasive means. In asthmatic patients, 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 to the study of new metabolic pathways

that may have a role in asthma pathophysiology.

11

Acknowledgements

We thank Dr. Manfred Spraul and Dr. Li-Hong Tseng of Bruker BioSpin GmbH for

NMR measurements.

12

References

1. Lemanske RF, Busse WW. Astma. J Allergy Clin Immunol 2003;111:S502-

S519.

2. Global Initiative for Asthma. Global Strategy for Asthma Management and

Prevention. NIH-NHLI 02-3659, Bethesda, MD:2002 (updated 2005).

3. Van Den Toorn LM, Overbeek SE, de Jongste JC, Leman K, Hoogsteden HC,

Prins JB. Airway inflammation is present during clinical remission of atopic

asthma. Am J Respir Crit Care Med 2001;164:2107-2113.

4. Payne DNR, Wilson NM, James A, Hablas H, Agrafioti C, Bush A. Evidence for

different subgroups of difficult asthma in children. Thorax 2005;56:345-350.

5. Horvath I, Barnes P, Hunt J. Exhaled breath condensate: methodological

recommendations and unresolved questions. Eur Respir J 2005;26:523-548.

6. Nicholson JK, Wilson ID. Understanding ‘global’ systems biology :

metabonomics and the continuum of metabolism. Natur Rev Drug Discov

2003;2:668-676

7. Nicholson J, Lindon J, Holmes E. 'Metabonomics': understanding the metabolic

responses of living systems to pathophysiological stimuli via multivariate

statistical analysis of biological NMR spectroscopic data. Xenobiotica

1999;29:1181-1189.

8. Pattern recognition and biomarker validation using quantitative 1H-NMR-based

metabolomics. Expert Rev Mol Diagn 2006;6:717-31

9. Bollard M, Stanley E, Lindon J, Nicholson J, Holmes E. NMR-based

metabonomic approaches for evaluating physiological influences on biofluid

composition. NMR Biomed 2005;18:143-62.

10. Lindon J, Holmes E, Nicholson J. So what's the deal with metabonomics? Anal

Chem 2003;75:385A-391A.

13

11. Lindon JC, Holmes E, Nicholson J Pattern recognition methods and applications

in biomedical magnetic resonance. Progress in Nuclear Magnetic Resonance

Spectroscopy 2001;39:1-40

12. Carraro S, Rezzi S, Reniero F, Heberger K, Giordano G, Zanconato S, Guillou

C, Baraldi E. Metabonomics applied to exhaled breath condensate in childhood

asthma. Eur Respir J 2006;28 (supplement 50):123s.

13. Baraldi E, de Jongste J. European Respiratory Society, American Thoracic

Society. Measurement of exhaled nitric oxide in children, 2001.

Eur Respir J 2002;20:223-37.

14. Harris R, Becker E, Cabral de Menezes S, Goodfellow R, Granger P. NMR

nomenclature. Nuclear spin properties and conventions for chemical shifts

(IUPAC Recommendations 2001). Pure Appl Chem, 2001;73:1795

15. Vandeginste B, Massart D, Buydens L, Jong S, Lewi P, Smeyers-Verbeke J.

Supervised pattern recognition. In: Handbook of Chemometrics and

Qualimetrics: Part B. Amsterdam, The Netherlands: Elsevier; 1998. p. 207-223

16. Geladi P, Kowalski B. Partial least-squares regression: a tutorial. Anal Chim

Acta 1986;185:1-17.

17. Ståhle L, Wold S. Partial least squares analysis with cross validation for the

two-class problem: a Monte Carlo study. J Chemometrics 1987;1:185-196.

18. Bland J, Altman D. Statistical methods for assessing agreement between two

methods of clinical measurement. Lancet 1986;1:307-10

19. Kharitonov S, Barnes P. Exhaled markers of pulmonary disease.

Am J Respir Crit Care Med 2001;163:1693-722

20. Baraldi E, Ghiro L, Piovan V, Carraro S, Zacchello F, Zanconato S. Safety and

success of exhaled breath condensate collection in asthma. Arch Dis Child

2003;88:358-360

14

21. Zacharasiewicz A, Wilson N, Lex C, Li A, Kemp M, Donovan J, Hooper J,

Kharitonov SA, Bush A. Repeatability of sodium and chloride in exhaled breath

condensates. Ped Pulm 2004;37:273-275

22. Odunsi K, Wollam R, Ambrosone C, Huston A, McCann S, Tammela J, Geisler

J, Miller G, Sellers T, Cliby W, Qian F, Keitz B, Intengan M, Lele S, Alderfer J.

Detection of epithelial ovarian cancer using H-NMR-based metabonomics. Int J

Cancer 2005;113:782-788.

23. Messana I, Forni F, Ferrari F, Rossi C, Giardina B, Zuppi C. Proton nuclear

magnetic resonance spectral profiles of urine in type II diabetic patients. Cinical

Chem 1998;44:1529-1534.

24. Moolenaar S, Engelke U, Wevers R. Proton nuclear magnetic resonance

spectroscopy of body fluids in the field of inborn errors of metabolism. Ann Clin

Biochem 2003;40:16-24.

25. Clayton T, Lindon J, Cloarec O, Antti H, Charuel C, Hanton G, Provost J, Le Net

J, Baker D, Walley R, Everett J, Nicholson J. Pharmaco-metabonomic

phenotyping and personalized drug treatment. Nature 2006; 440:1073-77

26. Coen M, O'Sullivan M, Bubb W, Kuchel P, Sorrell T. Proton nuclear magnetic

resonance-based metabonomics for rapid diagnosis of meningitis and

ventriculitis. Clin Infect Dis 2005;11:1582-1590.

27. Brindle J, Antti H, Holmes E, Tranter G, Nicholson J, Bethell H, Clarke S,

Schofield P, McKilligin E, Mosedale D, Grainger D. Rapid and noninvasive

diagnosis of the presence and severity of coronary heart disease using 1H-

NMR-based Metabonomics. Nat Med 2002; 8:1439-1445

28. Berry M, Shaw D, Green R, Brightling C, Wardlaw A, Pavord I. The use of

exhaled nitric oxide concentration to identify eosinophilic airway inflammation:

an observational study in adults with asthma. Clin Exp Allergy 2005;35:1175-9.

15

29. Erlandsson Harris, H, Andersson U. The nuclear protein HMGB1 as a

proinflammatory mediator. Eur J Immunol 2004;34:1503-1512.

30. Paredi P, Kharitonov S, Barnes P. Analysis of expired air for oxidation products.

Am J Respir Crit Care Med 2002 15;166:S31-7.

31. Dumas M, Maibaum E, Teague C, Ueshima H, Zhou B, Lindon J, Nicholson J,

Stamler J, Elliott P, Chan Q, Holmes E. Assessment of Analytical

Reproducibility of 1H NMR Spectroscopy Based Metabonomics for Large-Scale

Epidemiological Research: the INTERMAP Study. Anal Chem 2006,78:2199.

16

FIGURE LEGENDS

Figure 1. Partial Least Squares (PLS) X scores discrimination for healthy and

asthmatic children. All variables are used and three PLS components are retained in

the model. X scores are the components in the input matrix that best describe

relevant variations in the input variables and correlate best with the target value. As

we cannot determine the number of PLS components by cross-validation, said

number was varied: the classification was 69% correct using one PLS component,

83% using two and 94% using three; since the fourth PLS component does not

improve the classification, it very probably gives rise to an overfit.

Figure 2. Examples of Nuclear Magnetic Resonance (NMR) spectra obtained in

asthmatic (A and B) and healthy (C) children (zoom between 1-4 ppm). Significant

signals are found in the 1.7-2.2 ppm region of the spectrum in asthmatic children

compared to healthy subjects. These signals suggest a significant presence of

acetylated compounds (2) in exhaled breath condensate of asthmatic subjects.

Asthmatic children are also characterized by the presence of signals in the 3.2-3.4

ppm regions, indicative of the presence of oxidized compounds (1).

17

FIGURE 1

18

FIGURE 2

METABONOMICS APPLIED TO EXHALED BREATH CONDENSATE

IN CHILDHOOD ASTHMA

Silvia Carraro, Serge Rezzi, Fabiano Reniero, Károly Héberger,

Giuseppe Giordano, Stefania Zanconato, Claude Guillou, Eugenio Baraldi

Online Data Supplement

METODHS

Study subjects

The study included 25 (20 males) asthmatic children (mean age 10.6 years, range 7-15)

and 11 (6 males) healthy controls (mean age 8.6 years, range 6-13). All the enrolled

children were Caucasian. The asthmatic patients were recruited among patients attending

the Pulmonology/Allergy outpatients clinic at the Pediatrics Department in Padova. 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 > 12%, according to international guidelines (E1).

Eight asthmatic children had intermittent asthma and had received no inhaled

corticosteroid (ICS) treatment for at least a month. Seventeen children had persistent

asthma and were treated with ICS on a regular base (average daily dose 200 mcg

fluticasone or equivalent).

Asthmatic children were in good clinical control of their disease. In addition they had

suffered no acute upper or lower airway infection in the last 3 weeks. All asthmatic children

were atopic, sensitized to at least one airborne allergen, as demonstrated by skin prick

test.

Healthy children were relatives of nurses and doctors of our hospital and they had no

history of atopy or respiratory diseases.

The Ethics Committee of our hospital reviewed and approved the protocol and all parents

gave their informed consent.

Fractional exhaled nitric oxide (FENO) measurement

FENO was measured with the NIOX system (Aerocrine, Stockholm, Sweden) using a

single-breath on-line method according to ERS/ATS guidelines for measuring FENO in

children (E2). Children inhaled NO-free air to total lung capacity and exhaled through a

dynamic flow restrictor with a target flow of 50 ml/sec for at least 6-7 seconds. No nose

clip was used. The NIOX system was calibrated using a 200 ppb NO tank (Lindegas Hoek

Loos Speciality gases, Amsterdam, Netherlands) according to the manufacturer’s

instructions.

Pulmonary function test

Pulmonary function parameters were measured with a 10-liter bell spirometer (Biomedin,

Padova, Italy) and the best of three maneuvers was expressed as a percentage (%) of

predicted reference values.

Statistical analysis

Mann-Whitney Rank Sum Test was applied to compare FENO and FEV1 in healthy and

asthmatic children, being the data not normally distributed.

Data from the NMR spectra were processed using the AMIX Principal Component Analysis

(PCA) routine and then, after exportation, using the Statistica 6.0 software package for

Discriminant Analysis (DA) and Partial least squares (PLS).

Linear Discriminant Analysis (LDA) (E3) is a supervised pattern recognition

technique. It determines the linear function of the variables that maximizes the ratio of

between-class variance and minimizes the ratio of within-class variance. The variables (up

to 4 or 6) with the greatest discriminant capacity were selected with the best subset

selection variant of discriminant analysis. Partial Least Squares Discriminant Analysis

(PLS DA) (E4-E5) determines the components in the input matrix (X) that describe as

much as possible of the relevant variations in the input variables, and that, at the same

time, correlate best with the target value in Y, without including the variations that are

irrelevant or noisy. The PLS analysis was conducted on the centered (without intercept)

and standardized (autoscaled) data. PLS DA can be used in the case of ‘landscape

matrices’ without variable selection. It is not sensitive to overfit (if proper number of PLS

components are selected) or collinearity of variables.

Within-day repeatability and technical repeatability of the measurements were

evaluated using the method of Bland and Altman (E6) and the coefficient of variation,

respectively.

RESULTS

Within day repeatability

I) 6 asthmatic children performed EBC collection twice within the same day. When

analyzing the repeatability of NMR spectra, the main problem is the need of

evaluating the simultaneous measurement of many variables. To overcome this

problem we divided each spectrum in 5 zones (zone 1: 4.2-3.7 ppm; zone 2: 3.7-3.2

ppm; zone 3: 2.5-1.5 ppm; zone 4: 1.5-1.0 ppm; zone 5: 1.0-0.5 ppm) and we

calculated the area under the curve in each zone, obtaining 5 integrals. We could,

therefore, express each spectrum through 5 numbers (the integrals). We then

evaluated the repeatability by means of the Bland and Altman test applied both at

each zone and at all the zones together.

Looking at each spectrum zone we found that all the values fall within the 2SD range,

demonstrating a good repeatability (figure a1-a5).

References

E1. Global Initiative for Asthma. Global Strategy for Asthma Management and

Prevention. NIH-NHLI 02-3659, Bethesda, MD:(2002 updated 2005).

E2. Baraldi E, de Jongste J, European Respiratory Society, American Thoracic Society.

Measurement of exhaled nitric oxide in children, 2001.

Eur Respir J. 2002;20:223-37.

E3. Vandeginste B, Massart D, Buydens L, Jong S, Lewi P, Smeyers-Verbeke J. Chapter

33. Supervised pattern recognition, pp. 207-223. In Handbook of Chemometrics and

Qualimetrics: Part B; Elsevier: Amsterdam, The Netherlands, 1998.

E4. Geladi P, Kowalski B. Partial least-squares regression: a tutorial. Anal. Chim. Acta

1986;185:1-17.

E5. Ståhle L, Wold S. Partial least squares analysis with cross validation for the two-class

problem: a Monte Carlo study. J Chemometrics 1987;1:185-196.

E6. Bland J, Altman D. Statistical methods for assessing agreement between two

methods of clinical measurement. Lancet 1986;1:307-10

Figure E1. Bland Altman plot for zone 1 (4.2-3.7 ppm)

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

0 5 10 15

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

)

+1.96 SD

-1.96 SD

-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

0 5 10 15

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

)

+1.96 SD

-1.96 SD

Figure E2. Bland Altman plot for zone 2 (3.7-3.2 ppm)ዊ�

-1

-0,8

-0,6

-0,4

-0,2

0

0,2

0,4

0,6

0,8

1

0 5 10 15 20 25 30 35 40 45 50

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

-1

-0,8

-0,6

-0,4

-0,2

0

0,2

0,4

0,6

0,8

1

0 5 10 15 20 25 30 35 40 45 50

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

Figure E3. Bland Altman plot for zone 3 (2.5-1.5 ppm)ዊ�

-0,8

-0,6

-0,4

-0,2

0

0,2

0,4

0,6

0,8

0 5 10 15 20

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

-0,8

-0,6

-0,4

-0,2

0

0,2

0,4

0,6

0,8

0 5 10 15 20

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

Figure E4. Bland Altman plot for zone 4 (1.5-1.0 ppm)

-0,9

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0,9

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

-0,9

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0,9

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

Figure E5a. Bland Altman plot for zone 5 (1.0-0.5 ppm)ዊ�

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0 5 10 15

Mean (t1+t2) /2

Diff

ere

nce

(t1-

t2) +1.96 SD

-1.96 SD

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0 5 10 15

Mean (t1+t2) /2

Diff

ere

nce

(t1-

t2) +1.96 SD

-1.96 SD

Also analyzing all the zones together we found a good repeatability with 29 out of 30

(97%) values falling within the 1,96*SD range (figure b)

Figure E5b. Bland Altman plot for the 5 zones considered altogether

-0,9

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0,9

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

-0,9

-0,7

-0,5

-0,3

-0,1

0,1

0,3

0,5

0,7

0,9

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

Mean (t1+t2) /2

Diff

eren

ce(t

1-t2

) +1.96 SD

-1.96 SD

Technical repeatability

II) To evaluate the repeatability of the technique, we have repeated the NMR analysis of

one sample 8 times in a row. We used the integral of the spectrum area between 3

and 4 ppm, finding a coefficient of variation of 1.07%.