0
Original Research: Genetic and Developmental Disorders |

Metabolomic Evaluation of Neutrophilic Airway Inflammation in Cystic FibrosisMetabolomics of Neutrophilic Airway Inflammation FREE TO VIEW

Charles R. Esther, Jr, MD, PhD; Raymond D. Coakley, MD; Ashley G. Henderson, MD; Yi-Hui Zhou, PhD; Fred A. Wright, PhD; Richard C. Boucher, MD
Author and Funding Information

From the Division of Pediatric Pulmonology (Dr Esther), the Cystic Fibrosis and Pulmonary Research and Treatment Center (Drs Coakley, Henderson, and Boucher), and the Department of Biostatistics (Drs Zhou and Wright), University of North Carolina at Chapel Hill, Chapel Hill, NC.

CORRESPONDENCE TO: Charles R. Esther Jr, MD, PhD, Pediatric Pulmonology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599; e-mail: Charles_Esther@med.unc.edu


Part of this article has been presented in abstract form (Zhou Y, Wright F, Boucher RC, Esther CR Jr. Metabolomic evaluation of cystic fibrosis airways disease. Am J Respir Crit Care Med. 2013;187:A1165).

FUNDING/SUPPORT: This study was supported by the National Institutes of Health (NIH) [Grant P42-ES005948 to Dr Zhou and Grant P30-ES010126 to Dr Wright]; the NIH/National Heart, Lung, and Blood Institute [Grant K23-HL089708 to Dr Esther and Grants HL34322, HL107168, P01-HL08808, P30-DK065988 (5-52184), P01-HL110873, and P50-HL107168 to Dr Boucher]; and the NIH/National Institute of Environmental Health Sciences [Grant P30-ES10126 to Dr Esther].

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details.


Chest. 2015;148(2):507-515. doi:10.1378/chest.14-1800
Text Size: A A A
Published online

BACKGROUND:  Metabolomic evaluation of cystic fibrosis (CF) airway secretions could identify metabolites and metabolic pathways involved in neutrophilic airway inflammation that could serve as biomarkers and therapeutic targets.

METHODS:  Mass spectrometry (MS)-based metabolomics was performed on a discovery set of BAL fluid samples from 25 children with CF, and targeted MS methods were used to identify and quantify metabolites related to neutrophilic inflammation. A biomarker panel of these metabolites was then compared with neutrophil counts and clinical markers in independent validation sets of lavage from children with CF and adults with COPD compared with control subjects.

RESULTS:  Of the 7,791 individual peaks detected by positive-mode MS metabolomics discovery profiling, 338 were associated with neutrophilic inflammation. Targeted MS determined that many of these peaks were generated by metabolites from pathways related to the metabolism of purines, polyamines, proteins, and nicotinamide. Analysis of the independent validation sets verified that, in subjects with CF or COPD, several metabolites, particularly those from purine metabolism and protein catabolism pathways, were strongly correlated with neutrophil counts and were related to clinical markers, including airway infection and lung function.

CONCLUSIONS:  MS metabolomics identified multiple metabolic pathways associated with neutrophilic airway inflammation. These findings provide insight into disease pathophysiology and can serve as the basis for developing disease biomarkers and therapeutic interventions for airways diseases.

Figures in this Article

Chronic neutrophilic airway inflammation is a hallmark of cystic fibrosis (CF) and contributes significantly to morbidity and mortality.1 This inflammation can alter cellular metabolism, including extracellular metabolic pathways active within the airway lumen that generate biologically active molecules capable of initiating and modulating inflammatory responses.24 Metabolomics studies, which seek to characterize the full range of metabolites within biologic samples, suggest that CF is associated with significant changes in the pattern and concentrations of metabolites in airway secretions, including elevated concentrations of amino acids and lactate in BAL fluid (BALF)5 and lipid mediators in sputum.6 Other studies have demonstrated that a metabolomic signal of CF airways disease can be detected noninvasively in exhaled breath condensate (EBC),7,8 although the specific metabolites responsible for the signal were not identified. Although these studies can provide new insights into pathophysiologic mechanisms and can identify new therapeutic targets and biomarkers of disease progression,9,10 studies to date are limited by a lack of independent validation to verify results.11

We hypothesized that a comprehensive metabolomic analysis of CF airways samples, obtained from subjects with various levels of inflammation, may provide insight into the pathways associated with inflammation and may identify new biomarkers and therapeutic targets. To this end, discovery mass spectrometry (MS) metabolomics was performed in BALF from 25 children with CF, and a biomarker panel of relevant metabolites was developed. The relationships between these metabolites and inflammation, infection, and lung function were then tested in independent validation sets of BALF from children with CF, as well as from adults with COPD and relevant control subjects.

Subjects and Samples

BALF from children with CF was collected during clinically indicated bronchoscopy via standardized protocols4 and was centrifuged at 11,000 × g for 5 min, and the supernatant was stored at −80°C. Fifty pediatric subjects were included: 25 for discovery and 25 for validation (Table 1). BALF from 10 subjects with COPD, 10 healthy smokers, and 10 healthy nonsmokers was obtained using a single lavage of 50 mL normal saline into the right middle lobe. Samples were placed on ice shortly after collection, centrifuged at 11,000 × g for 5 min at 4°C to remove cellular material, and then frozen at −80°C until analysis. Both studies were approved by the University of North Carolina institutional review board (IRB Nos. 11-0828 and 05-2876). Clinical data were abstracted from medical and research records.

Table Graphic Jump Location
TABLE 1 ]  Demographics of Subjects With CF

Data are presented as mean ± SD. CF = cystic fibrosis; PMN = polymorphonuclear leukocyte.

Metabolomic profiling was performed by the David H. Murdock Research Institute.12 Briefly, 500-μL aliquots BALF were lyophilized and reconstituted in 100 μL 0.1% formic acid/50% acetonitrile, filtered through 0.2-μm polytetrafluoroethylene, and analyzed on a UPLC-QtofMS system (ACQUITY UPLC-SYNAPT HDMS, Waters Corporation) operated in electrospray ionization positive mode. Chromatography used an ACQUITY UPLC BEH C18 1.7-μM VanGuard precolumn (2.1 × 5 mm) and an ACQUITY UPLC BEH C18 1.7-μM analytical column (2.1 × 100 mm) with gradients from 1% to 100% acetonitrile in 0.1% formic acid over 13 min. Data files were processed using MarkerLynx XS (Waters Corp), with peak detection and signal alignment as described.12 Analysis occurred in two batches of 10 and 15 samples, both of which included a range of neutrophilic inflammation.

Targeted MS was performed using a Quantum-Ultra triple quadrupole mass spectrometer (Thermo Finnigan LLC) in the multiple reaction monitoring mode. Chromatographic conditions similar to those of the metabolomics (UPLC BEH C18 column, acetonitrile/formic acid gradients) and previous work (UPLC T3 HSS C18 column, methanol/formic acid gradients13) were explored, with the T3 column judged superior and used for validation. Validation BALF samples were spiked with isotopically labeled internal standard14 and were filtered through a 10-kDa-size selection filter (EMD Millipore). Biomarker signals were defined as ratios to the internal standard with the closest column run time.

Statistical Analysis

Discovery metabolomic data were analyzed using robust principal component analysis (Y. H. Z. and J. S. Marron, PhD, unpublished data, 2014). Relationships to inflammation were tested using analysis within each of the two batches, with evidence summarized using Fisher combined P value. The false-discovery rate q value was used to correct for multiple comparisons. For validation, data not normally distributed (by D’Agostino and Pearson omnibus normality test) were log transformed and analyzed using parametric methods, including Student t tests, analysis of variance, and linear regression. Spearman correlations were used for relationships between percent neutrophils to metabolite to urea ratios. Categorical comparisons for demographic balance in discovery vs validation samples were made using the Fisher exact test. Statistical analyses were performed using GraphPad Prism v 5.0 (GraphPad Software, Inc) and SAFExpress software (Safexpress Pvt Ltd).15

Subjects and Samples

Two sets of BALF were obtained from children with CF: a discovery set of 25 samples and a validation set of 25 samples. The sets did not differ in terms of age, sex, or range of inflammation (Table 1), and there was no overlap in subjects. Bacterial or fungal pathogens were recovered in 20 discovery set and 21 validation set samples, with Staphylococcus, Pseudomonas, and Haemophilus species the most common. A second validation set was obtained from adults with COPD, healthy nonsmokers, and healthy smokers (n = 10 per group) (Table 2). Neutrophil counts and percent neutrophils in BALF were higher in those with COPD than in healthy control subjects, but were lower than in patients with CF. All subjects were fasting at the time of the procedure.

Table Graphic Jump Location
TABLE 2 ]  Demographics of Subjects With COPD

Data are presented as mean ± SD. ND = not determined; NS = not significant. See Table 1 legend for expansion of other abbreviations.

a 

All subjects with COPD had a history of smoking.

MS Metabolomics on Discovery Samples

Positive-mode MS metabolomics16 revealed 7,791 individual peaks, with each peak defined as a specific mass-to-charge (m/z) ratio and column run time. Total MS signal correlated positively with neutrophil counts (r = 0.71, P < .001) (Fig 1A). Principal component analysis revealed a clustering of low-inflammation samples (Fig 1B), with a greater dispersion among the high-inflammation samples.

Figure Jump LinkFigure 1 –  MS metabolomics. A, Relationship between neutrophil (PMN) count and total MS signal during discovery metabolomics. Signal for each peak was normalized such that the lowest signal in the dataset had a value of 1. A strong correlation (r = 0.70, P < .001) was observed between neutrophil counts and total MS metabolomics signal. B, Principal component analysis of metabolomic data. Low inflammation samples () generally clustered together away from high inflammation samples (), but with some overlap. MS = mass spectrometry; PC = principal component; PMN = polymorphonuclear leukocyte.Grahic Jump Location

Potential biomarkers of neutrophilic inflammation were identified by both group (samples divided into groups along median neutrophil count) and correlation analyses (linear regression to neutrophil counts). In group analysis, 267 peaks distinguished higher inflammation (neutrophil counts ≥ 2.0 million cells/mL [n = 13]) from lower inflammation (neutrophil counts ≤ 1.5 million cells/mL [n = 12]) groups (false discovery rate q < 0.05) (e-Table 1). For correlation analysis, 140 peaks correlated with neutrophil counts (false discovery rate q < 0.15). Sixty-nine peaks overlapped. This resulted in a total of 338 unique peaks identified as potential biomarkers. Several m/z ratios were present in multiple peaks, and 317 unique m/z ratios were represented.

Identification of Potential Biomarkers

Selected ion monitoring was utilized to determine which m/z ratios associated with putative biomarkers generated peaks that were larger in higher- vs lower-inflammation samples. Full scans of each peak were then utilized to determine the fragmentation pattern of the metabolite generating that peak. Metabolite identification was made by comparison with published literature17 and online resources such as the Human Metabolite Database (www.hmdb.ca) and was confirmed by comparison with appropriate standards. These analyses suggested that the putative biomarkers included the following metabolic pathways (e-Appendix 1).

Purine Metabolism:

The putative biomarker peak at m/z 137.1 was identified as hypoxanthine, a metabolic product of adenyl purines identified previously as neutrophilic inflammation biomarkers.4 In addition, several biomarker peaks observed at m/z 136.1 were consistent with adenine, the primary fragmentation ion of adenyl purines,13 which may have been generated by in-source fragmentation of adenyl purines.

Polyamine Metabolism:

Potential biomarker peaks at m/z 188.1, 146.1, and 89.1 were identified as N-acetyl-spermidine, spermidine, and putrescine, all part of the polyamine synthesis pathway. Furthermore, free adenine can be generated during polyamine synthesis,18 which may account for one of the previously noted peaks at m/z 136.1.

Protein Catabolism:

Several biomarker peaks had m/z ratios consistent with that of amino acids; these included proline (m/z 116.1), valine (m/z 118.1), threonine (m/z 120.1), leucine/isoleucine (m/z 132.1), glutamine (m/z 147.1), and phenylalanine (m/z 166.1). Analysis of the internal library suggested that m/z 188.1 represented an in-source fragmentation product of tryptophan. Free amino acids can be generated by protease activity in CF airways,19 suggesting that other peptide fragments may also be represented. The biomarker peak at m/z 229.1 fragmented into major daughter ions at m/z 116.1 and 132.1, suggestive of a leucine-proline dipeptide, and the peak coeluted with a leucine-proline standard. Other putative biomarkers consistent with dipeptides included those at m/z 173.1 (glycine-proline), 260.1 (lysine-leucine), 267.1, (threonine-phenylalanine), and 279.1 (leucine-phenylalanine).

Nicotinamide Metabolism:

The putative biomarker at m/z 123.1 was identified as nicotinamide based on an internal library, and its fragmentation pattern (major daughter ion at m/z 95.1) and chromotographic elution time were identical to those of authentic nicotinamide.

Validation of Potential Biomarkers

Based on these findings, an MS biomarker panel of 30 metabolites was created by modifying a previous MS method20 to include the putative biomarkers and other metabolites within related metabolic pathways. Correlations with neutrophil counts and clinical measures were then assessed in an independent validation set of BALF samples from 25 children with CF.

Purine Metabolism:

The purine nucleotide adenosine monophosphate (AMP) was positively correlated with neutrophilic counts, and adenosine exhibited a trend toward negative correlation (Table 3), both similar to previous observations.4 Downstream purine metabolites, including inosine, hypoxanthine, and xanthine, were strongly and positively correlated with neutrophilic counts, with hypoxanthine and xanthine exhibiting about three orders of magnitude concentration differences between the lowest and highest inflammation samples.

Table Graphic Jump Location
TABLE 3 ]  Correlations Between Metabolites Identified by MS Metabolomics and Neutrophils in Validation Dataset

Relationships to neutrophil counts were assessed using Pearson correlations after log transformation. P values for correlations are shown. AMP = adenosine monophosphate; Gly-Pro = glycine-proline; Leu-Phe = leucine-phenylalanine; Leu-Pro = leucine-proline; Lys-Leu = lysine-leucine; MS = mass spectrometry; NAD = nicotinamide adenine diphosphate; NMN = nicotinamide mononucleotide; Pro-Gly-Pro = proline-glycine-proline; Thr-Phe = threonine-phenylalanine.

a 

Significant after Bonferroni multiple-test correction.

Polyamine Metabolism:

The polyamines putrescine, spermidine, and spermine, as well as S-adenosyl-methionine and adenine, positively correlated with neutrophilic inflammation (Table 3). The spermidine to spermine ratio, reportedly reduced in CF exacerbation,21 exhibited a trend toward negative correlation with neutrophil counts (r = −0.37, P = .07). Of note, the peak associated with adenine eluted at a run time (about 4.2 min) substantially different from that of authentic adenine standard (about 2.5 min) and could not be confirmed.

Protein Catabolism:

Strong correlations with neutrophil counts were observed for multiple amino acids (Table 3), with MS signals varying by more than three orders of magnitude. Strong correlations were observed for multiple dipeptides, consistent with formation during protein catabolism.19 As a further control, previously described methods22 were used to demonstrate that the proline-glycine-proline proteolytic fragment of fibronectin, known to correlate with inflammation, also correlated with neutrophil counts (r = 0.84, P < .001).

Nicotinamide:

Nicotinamide and two related metabolites, nicotinamide adenine diphosphate and nicotinamide mononucleotide, were positively, albeit modestly, correlated with neutrophil counts.

Controlling for Dilution

Urea, commonly used as a BALF dilution marker, was included in the MS biomarker panel and correlated with neutrophil counts (r = 0.73, P < .001). Because this suggested that inflammation may have impacted the dilution of airway secretions in BALF, we performed a secondary analysis of each putative biomarker as a ratio to urea relative to percent neutrophils as a dilution-independent inflammatory marker (e-Table 2). Hypoxanthine, xanthine, and most amino acids and dipeptides remained correlated with neutrophils as ratios to urea, but polyamines and nicotinamide were not. However, these results should be interpreted with caution given the limitations of urea as a BALF dilution marker,23 particularly because urea is generated during polyamine synthesis.21

Impact of Infection

Neutrophilic inflammation correlates with the severity of infection,24 suggesting that airway pathogens may contribute to the metabolomic signal. In fact, biomarkers from each pathway were elevated in samples with bacterial pathogens (n = 15) relative to those with no pathogens (n = 5) (Fig 2). However, biomarker concentrations did not differ between samples that cultured exclusively gram-positive cocci (Staphylococcus aureus [n = 7]) vs only gram-negative rods (Pseudomonas, Burkholderia, or Stenotrophomonas species [n = 5]). Correlations between biomarker metabolites and bacterial counts (organisms/mL) were observed, although they were less robust than correlations with neutrophil counts, and most metabolites remained positively correlated with neutrophil counts even among samples that did not grow bacterial pathogens (Table 4).

Figure Jump LinkFigure 2 –  Biomarkers and infection. A, Differences in BAL fluid (BALF) neutrophil counts were observed among samples with no pathogens on culture (white, n = 5), fungal pathogens only (hatched, n = 5), or bacterial pathogens (gray, n = 15). *P < .05 by analysis of variance (ANOVA), Tukey posttest indicating differences between no pathogen and bacterial pathogen groups. B, Multiple metabolomics biomarkers also differentiated among BALF culture status, including some metabolites (**) that differentiated samples with bacterial pathogens from both no pathogen and fungal pathogen-only groups (Tukey posttest following ANOVA). Hyp = hypoxanthine; Leu = leucine; Leu-Pro = leucine-proline dipeptide; Nic = nicotinamide; NSp = N-acetyl spermidine; Phe = phenylalanine; Spm = spermine; Xan = xanthine. See Figure 1 legend for expansion of other abbreviations.Grahic Jump Location
Table Graphic Jump Location
TABLE 4 ]  Correlations Between Representative Metabolites and Airway Infection

See Table 1 and 3 legends for expansion of abbreviations.

Relationship to Clinical Markers

We observed a significant negative correlation between BALF neutrophil counts and percent predicted FEV1 in our validation set (r = −0.58, P < .01) (Fig 3A), similar to previous findings.3 Many biomarker metabolites, including hypoxanthine, xanthine, and several amino acids and dipeptides, were also negatively correlated with percent predicted FEV1 (Fig 3B). Less robust correlations for polyamines and a trend toward correlation for nicotinamide were also observed (Fig 3C). Biomarker metabolite concentrations were not predictive of changes in FEV1 following sample collection (average 3.0 ± 0.7 years of data), with the exception of adenosine, which demonstrated a significant negative correlation (Fig 3D). No metabolites were significantly correlated with BMI percentile as a nutritional marker.

Figure Jump LinkFigure 3 –  BALF biomarkers and lung function. A, Percent predicted FEV1 was significantly negatively correlated with BALF neutrophil (PMN) counts (r = −0.57, P < .01). B, Similar correlations were observed for BALF purines (Hyp, r = −0.56; Xan, r = −0.56; P < .01) and protein catabolism metabolites (Leu, r = −0.61; Phe, r = −0.59; Leu-Pro, r = −0.59; P < .01). C, Correlations with percent predicted FEV1 were observed for polyamines (Spm, r = −0.44, P < .05) with a trend for Nic (r = −0.36, P < .11). D, Ado, but no other metabolite, was negatively correlated with changes in predicted FEV1 over time (r = −0.71, P < .001). Ado = adenosine. See Figure 1 and 2 legends for expansion of other abbreviations.Grahic Jump Location
Biomarker Panel in COPD

Representative biomarker metabolites were elevated in subjects with COPD relative to healthy nonsmokers, and a subset differentiated those with COPD from healthy smokers (Fig 4A). All metabolites were significantly correlated with BALF neutrophil counts (r = 0.42-0.62, P < .05 for all), and negative correlations with percent predicted FEV1 were observed for hypoxanthine (r = −0.57), xanthine (r = −0.39), leucine (r = −0.44), phenylalanine (r = −0.48), and nicotinamide (r = −0.50) (Fig 4B), with a trend for spermine (r = −0.32, P = .053).

Figure Jump LinkFigure 4 –  Biomarker metabolites in COPD. A, Biomarker metabolites were measured in BALF from healthy nonsmokers (n = 10, white bars), healthy smokers (n = 10, hatched bars), and patients with COPD (n = 10, gray bars). Biomarker metabolites differentiated COPD samples from those of healthy nonsmokers (*) or from both healthy nonsmokers and healthy smokers (**); ANOVA with Tukey posttest. B, Biomarkers Hyp (r = −0.57, P < .001), Leu (r = −0.44, P < .01), and Nic (r = −0.50, P < .01) and others (not shown) were significantly correlated with percent predicted FEV1. See Figure 1 and 2 legends for expansion of abbreviations.Grahic Jump Location

Our data demonstrate that the inflamed CF airway is a metabolically rich environment. Although the airway metabolome reflects a complex interaction among airway epithelia, inflammatory cells, and microbial pathogens, our data suggest that the metabolites identified in this study primarily reflect neutrophilic inflammation. Correlations were strongest with neutrophil counts, and the purine metabolites hypoxanthine and xanthine, as well as several amino acids, appeared functionally equivalent to neutrophil counts as biomarkers. Measures of these metabolites using simple, commercially available assays could serve as an alternative means of assessing neutrophilic inflammation when standard cell counts are not available, although further investigation is needed to demonstrate the validity and reliability relative to established measures. These biomarkers may be broadly applicable, because they functioned similarly in subjects with CF and COPD.

Our metabolomics studies also identified novel pathways involved in disease pathogenesis. For example, we demonstrated that several adenosine metabolites strongly correlated with airway neutrophils, suggesting that the inflamed CF airway exhibits a robust extracellular metabolism of adenosine that may play a significant role in modulating purinergic signaling responses to inflammation.25 Accumulation of hypoxanthine in inflamed CF airways is consistent with findings from asthma,26 bronchopulmonary dysplasia,27 and ARDS,28 and may contribute to oxidative stress, because its metabolism to xanthine and uric acid generates oxygen superoxides.27,28 Thus, existing pharmacologic agents that block hypoxanthine formation29 may be considered as therapeutic candidates in CF and other airways diseases.

The correlations of airway neutrophils with amino acids and dipeptides are consistent with the known roles of extracellular airway proteases, including those derived from neutrophils, in CF airways disease.30 Although elevated amino acid concentrations in CF airway secretions has been described previously,5,31,32 the strong relationships between the markers of protein catabolism and the inflammatory, infection, and clinical measures observed here suggest that these metabolites could serve as biomarkers of in vivo protease activity in trials of protease inhibitors.

Elevated polyamine concentrations have been reported in CF sputum21 and may increase mucin secretion33 and aggregation.34 Interestingly, genetic studies suggest that activity of the methionine salvage pathway, which plays a critical role in polyamine synthesis, is increased in patients with CF who have more severe lung disease.35 These findings suggest that polyamines may be both a biomarker and a therapeutic target for CF airways disease and COPD.

Nicotinamide is known to have antiinflammatory properties, but an association between nicotinamide and airways disease has not been described previously. Free nicotinamide can be generated by extracellular ADP-ribosyl-transferases36 that modulate crosstalk between epithelia and neutrophils in inflamed airways.37 This pathway may be responsible for our findings, although further study is needed.

The metabolites identified in this study could also be used as noninvasive biomarkers of airways disease. MS methods similar to those used in this study have been used to assess EBC as a noninvasive airway sample.14,20,38 However, although biomarkers such as AMP function similarly in BALF and EBC,4,38 the relationships between inflammation and adenosine and phenylalanine in the current study differ from previous observations in EBC.14,20,38 Thus, EBC and BALF may differ in the metabolites that function optimally as biomarkers, perhaps reflecting different sites of origin of the airway secretions,39,40 and further study will be needed to identify the most effective metabolomic biomarkers for noninvasive detection.

Although we took steps to minimize enzymatic activity after sample collection, we cannot rule out the possibility that extracellular metabolism during sample collection or processing influenced our results. Similarly, although subjects were fasting at the time of sample collection, it is possible that dietary variations altered metabolite concentrations. However, given the strength of our findings, we do not believe that either of these potential confounders would significantly alter out conclusions.

In summary, we used MS metabolomics to identify the metabolites and metabolic pathways associated with neutrophilic airway inflammation in CF and validated these relationships in independent data sets from subjects with CF and COPD. These metabolites can serve as biomarkers of inflammation as well as therapeutic targets.

Author contributions: C. R. E. contributed to the design of the study and acquisition and analysis of the data; R. D. C. and A. G. H. contributed to the obtaining of lavage samples from subjects with COPD and control subjects; Y.-H. Z. and F. A. W. contributed to the biostatistical analysis of the data; R. C. B. contributed to the design of the study and analysis of the data; and C. R. E., R. D. C., A. G. H., Y.-H. Z, F. A. W., and R. C. B. contributed to the drafting, revising, and final approval of the article.

Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Role of sponsors: The funding bodies played no role in the design or execution of these studies beyond provision of financial support.

Other contributions: The authors thank the David H. Murdock Research Institute for performing the metabolomics analysis.

Additional information: The e-Appendix and e-Tables can be found in the Supplemental Materials section of the online article.

AMP

adenosine monophosphate

BALF

BAL fluid

CF

cystic fibrosis

EBC

exhaled breath condensate

MS

mass spectrometry

Rowe SM, Miller S, Sorscher EJ. Cystic fibrosis. N Engl J Med. 2005;352(19):1992-2001. [CrossRef] [PubMed]
 
Alton EW, Davies JC, Geddes DM. Biomarkers for cystic fibrosis: are we progressing? Am J Respir Crit Care Med. 2007;175(8):750-751. [CrossRef] [PubMed]
 
Mayer-Hamblett N, Aitken ML, Accurso FJ, et al. Association between pulmonary function and sputum biomarkers in cystic fibrosis. Am J Respir Crit Care Med. 2007;175(8):822-828. [CrossRef] [PubMed]
 
Esther CR Jr, Alexis NE, Clas ML, et al. Extracellular purines are biomarkers of neutrophilic airway inflammation. Eur Respir J. 2008;31(5):949-956. [CrossRef] [PubMed]
 
Wolak JE, Esther CR Jr, O’Connell TM. Metabolomic analysis of bronchoalveolar lavage fluid from cystic fibrosis patients. Biomarkers. 2009;14(1):55-60. [CrossRef] [PubMed]
 
Yang J, Eiserich JP, Cross CE, Morrissey BM, Hammock BD. Metabolomic profiling of regulatory lipid mediators in sputum from adult cystic fibrosis patients. Free Radic Biol Med. 2012;53(1):160-171. [CrossRef] [PubMed]
 
de Laurentiis G, Paris D, Melck D, et al. Metabonomic analysis of exhaled breath condensate in adults by nuclear magnetic resonance spectroscopy. Eur Respir J. 2008;32(5):1175-1183. [CrossRef] [PubMed]
 
Sofia M, Maniscalco M, de Laurentiis G, Paris D, Melck D, Motta A. Exploring airway diseases by NMR-based metabonomics: a review of application to exhaled breath condensate. J Biomed Biotechnol. 2011;2011:403260.
 
Sagel SD. Noninvasive biomarkers of airway inflammation in cystic fibrosis. Curr Opin Pulm Med. 2003;9(6):516-521. [CrossRef] [PubMed]
 
Sumner SJ, Burgess JP, Snyder RW, Popp JA, Fennell TR. Metabolomics of urine for the assessment of microvesicular lipid accumulation in the liver following isoniazid exposure. Metabolomics. 2010;6(2):238-249. [CrossRef] [PubMed]
 
Ransohoff DF. How to improve reliability and efficiency of research about molecular markers: roles of phases, guidelines, and study design. J Clin Epidemiol. 2007;60(12):1205-1219. [CrossRef] [PubMed]
 
Qiu Y, Cai G, Su M, et al. Serum metabolite profiling of human colorectal cancer using GC-TOFMS and UPLC-QTOFMS. J Proteome Res. 2009;8(10):4844-4850. [CrossRef] [PubMed]
 
Esther CR Jr, Jasin HM, Collins LB, Swenberg JA, Boysen G. A mass spectrometric method to simultaneously measure a biomarker and dilution marker in exhaled breath condensate. Rapid Commun Mass Spectrom. 2008;22(5):701-705. [CrossRef] [PubMed]
 
Esther CR Jr, Lazaar AL, Bordonali E, Qaqish B, Boucher RC. Elevated airway purines in COPD. Chest. 2011;140(4):954-960. [CrossRef] [PubMed]
 
Zhou YH, Barry WT, Wright FA. Empirical pathway analysis, without permutation. Biostatistics. 2013;14(3):573-585. [CrossRef] [PubMed]
 
Evans AM, DeHaven CD, Barrett T, Mitchell M, Milgram E. Integrated, nontargeted ultrahigh performance liquid chromatography/electrospray ionization tandem mass spectrometry platform for the identification and relative quantification of the small-molecule complement of biological systems. Anal Chem. 2009;81(16):6656-6667. [CrossRef] [PubMed]
 
Bajad SU, Lu W, Kimball EH, Yuan J, Peterson C, Rabinowitz JD. Separation and quantitation of water soluble cellular metabolites by hydrophilic interaction chromatography-tandem mass spectrometry. J Chromatogr A. 2006;1125(1):76-88. [CrossRef] [PubMed]
 
Kamatani N, Carson DA. Dependence of adenine production upon polyamine synthesis in cultured human lymphoblasts. Biochim Biophys Acta. 1981;675(3-4):344-350. [CrossRef] [PubMed]
 
Birrer P, McElvaney NG, Rüdeberg A, et al. Protease-antiprotease imbalance in the lungs of children with cystic fibrosis. Am J Respir Crit Care Med. 1994;150(1):207-213. [CrossRef] [PubMed]
 
Esther CR Jr, Olsen BM, Lin FC, Fine J, Boucher RC. Exhaled breath condensate adenosine tracks lung function changes in cystic fibrosis. Am J Physiol Lung Cell Mol Physiol. 2013;304(7):L504-L509. [CrossRef] [PubMed]
 
Grasemann H, Shehnaz D, Enomoto M, Leadley M, Belik J, Ratjen F. L-ornithine derived polyamines in cystic fibrosis airways. PLoS ONE. 2012;7(10):e46618. [CrossRef] [PubMed]
 
Gaggar A, Jackson PL, Noerager BD, et al. A novel proteolytic cascade generates an extracellular matrix-derived chemoattractant in chronic neutrophilic inflammation. J Immunol. 2008;180(8):5662-5669. [CrossRef] [PubMed]
 
Haslam PL, Baughman RP. Report of ERS Task Force: guidelines for measurement of acellular components and standardization of BAL. Eur Respir J. 1999;14(2):245-248. [CrossRef] [PubMed]
 
Muhlebach MS, Stewart PW, Leigh MW, Noah TL. Quantitation of inflammatory responses to bacteria in young cystic fibrosis and control patients. Am J Respir Crit Care Med. 1999;160(1):186-191. [CrossRef] [PubMed]
 
Eltzschig HK, Sitkovsky MV, Robson SC. Purinergic signaling during inflammation. N Engl J Med. 2012;367(24):2322-2333. [CrossRef] [PubMed]
 
Esther CR Jr, Peden DB, Alexis NE, Hernandez ML. Airway purinergic responses in healthy, atopic nonasthmatic, and atopic asthmatic subjects exposed to ozone. Inhal Toxicol. 2011;23(6):324-330. [CrossRef] [PubMed]
 
Boda D. Role of hyperuricaemia in critically ill patients especially newborns. Acta Paediatr Acad Sci Hung. 1984;25(1-2):23-32.
 
Quinlan GJ, Lamb NJ, Tilley R, Evans TW, Gutteridge JM. Plasma hypoxanthine levels in ARDS: implications for oxidative stress, morbidity, and mortality. Am J Respir Crit Care Med. 1997;155(2):479-484. [CrossRef] [PubMed]
 
Gandhi V, Balakrishnan K. Pharmacology and mechanism of action of forodesine, a T-cell targeted agent. Semin Oncol. 2007;34(6 suppl 5):S8-S12. [CrossRef] [PubMed]
 
Griese M, Kappler M, Gaggar A, Hartl D. Inhibition of airway proteases in cystic fibrosis lung disease. Eur Respir J. 2008;32(3):783-795. [CrossRef] [PubMed]
 
Thomas SR, Ray A, Hodson ME, Pitt TL. Increased sputum amino acid concentrations and auxotrophy ofPseudomonas aeruginosain severe cystic fibrosis lung disease. Thorax. 2000;55(9):795-797. [CrossRef] [PubMed]
 
Barth AL, Pitt TL. The high amino-acid content of sputum from cystic fibrosis patients promotes growth of auxotrophicPseudomonas aeruginosaJ Med Microbiol. 1996;45(2):110-119. [CrossRef] [PubMed]
 
Chen H, Macleod C, Deng B, et al. CAT-2 amplifies the agonist-evoked force of airway smooth muscle by enhancing spermine-mediated phosphatidylinositol-(4)-phosphate-5-kinase-gamma activity. Am J Physiol Lung Cell Mol Physiol. 2007;293(4):L883-L891. [CrossRef] [PubMed]
 
Chen EY, Wang YC, Chen CS, Chin WC. Functionalized positive nanoparticles reduce mucin swelling and dispersion. PLoS ONE. 2010;5(11):e15434. [CrossRef] [PubMed]
 
Wright FA, Strug LJ, Doshi VK, et al. Genome-wide association and linkage identify modifier loci of lung disease severity in cystic fibrosis at 11p13 and 20q13.2. Nat Genet. 2011;43(6):539-546. [CrossRef] [PubMed]
 
Koch-Nolte F, Fischer S, Haag F, Ziegler M. Compartmentation of NAD+-dependent signalling. FEBS Lett. 2011;585(11):1651-1656. [CrossRef] [PubMed]
 
Paone G, Stevens LA, Levine RL, et al. ADP-ribosyltransferase-specific modification of human neutrophil peptide-1. J Biol Chem. 2006;281(25):17054-17060. [CrossRef] [PubMed]
 
Esther CR Jr, Boysen G, Olsen BM, et al. Mass spectrometric analysis of biomarkers and dilution markers in exhaled breath condensate reveals elevated purines in asthma and cystic fibrosis. Am J Physiol Lung Cell Mol Physiol. 2009;296(6):L987-L993. [CrossRef] [PubMed]
 
Alexis NE, Hu SC, Zeman K, Alter T, Bennett WD. Induced sputum derives from the central airways: confirmation using a radiolabeled aerosol bolus delivery technique. Am J Respir Crit Care Med. 2001;164(10 pt 1):1964-1970. [CrossRef] [PubMed]
 
Fabian P, Brain J, Houseman EA, Gern J, Milton DK. Origin of exhaled breath particles from healthy and human rhinovirus-infected subjects. J Aerosol Med Pulm Drug Deliv. 2011;24(3):137-147. [CrossRef] [PubMed]
 

Figures

Figure Jump LinkFigure 1 –  MS metabolomics. A, Relationship between neutrophil (PMN) count and total MS signal during discovery metabolomics. Signal for each peak was normalized such that the lowest signal in the dataset had a value of 1. A strong correlation (r = 0.70, P < .001) was observed between neutrophil counts and total MS metabolomics signal. B, Principal component analysis of metabolomic data. Low inflammation samples () generally clustered together away from high inflammation samples (), but with some overlap. MS = mass spectrometry; PC = principal component; PMN = polymorphonuclear leukocyte.Grahic Jump Location
Figure Jump LinkFigure 2 –  Biomarkers and infection. A, Differences in BAL fluid (BALF) neutrophil counts were observed among samples with no pathogens on culture (white, n = 5), fungal pathogens only (hatched, n = 5), or bacterial pathogens (gray, n = 15). *P < .05 by analysis of variance (ANOVA), Tukey posttest indicating differences between no pathogen and bacterial pathogen groups. B, Multiple metabolomics biomarkers also differentiated among BALF culture status, including some metabolites (**) that differentiated samples with bacterial pathogens from both no pathogen and fungal pathogen-only groups (Tukey posttest following ANOVA). Hyp = hypoxanthine; Leu = leucine; Leu-Pro = leucine-proline dipeptide; Nic = nicotinamide; NSp = N-acetyl spermidine; Phe = phenylalanine; Spm = spermine; Xan = xanthine. See Figure 1 legend for expansion of other abbreviations.Grahic Jump Location
Figure Jump LinkFigure 3 –  BALF biomarkers and lung function. A, Percent predicted FEV1 was significantly negatively correlated with BALF neutrophil (PMN) counts (r = −0.57, P < .01). B, Similar correlations were observed for BALF purines (Hyp, r = −0.56; Xan, r = −0.56; P < .01) and protein catabolism metabolites (Leu, r = −0.61; Phe, r = −0.59; Leu-Pro, r = −0.59; P < .01). C, Correlations with percent predicted FEV1 were observed for polyamines (Spm, r = −0.44, P < .05) with a trend for Nic (r = −0.36, P < .11). D, Ado, but no other metabolite, was negatively correlated with changes in predicted FEV1 over time (r = −0.71, P < .001). Ado = adenosine. See Figure 1 and 2 legends for expansion of other abbreviations.Grahic Jump Location
Figure Jump LinkFigure 4 –  Biomarker metabolites in COPD. A, Biomarker metabolites were measured in BALF from healthy nonsmokers (n = 10, white bars), healthy smokers (n = 10, hatched bars), and patients with COPD (n = 10, gray bars). Biomarker metabolites differentiated COPD samples from those of healthy nonsmokers (*) or from both healthy nonsmokers and healthy smokers (**); ANOVA with Tukey posttest. B, Biomarkers Hyp (r = −0.57, P < .001), Leu (r = −0.44, P < .01), and Nic (r = −0.50, P < .01) and others (not shown) were significantly correlated with percent predicted FEV1. See Figure 1 and 2 legends for expansion of abbreviations.Grahic Jump Location

Tables

Table Graphic Jump Location
TABLE 1 ]  Demographics of Subjects With CF

Data are presented as mean ± SD. CF = cystic fibrosis; PMN = polymorphonuclear leukocyte.

Table Graphic Jump Location
TABLE 2 ]  Demographics of Subjects With COPD

Data are presented as mean ± SD. ND = not determined; NS = not significant. See Table 1 legend for expansion of other abbreviations.

a 

All subjects with COPD had a history of smoking.

Table Graphic Jump Location
TABLE 3 ]  Correlations Between Metabolites Identified by MS Metabolomics and Neutrophils in Validation Dataset

Relationships to neutrophil counts were assessed using Pearson correlations after log transformation. P values for correlations are shown. AMP = adenosine monophosphate; Gly-Pro = glycine-proline; Leu-Phe = leucine-phenylalanine; Leu-Pro = leucine-proline; Lys-Leu = lysine-leucine; MS = mass spectrometry; NAD = nicotinamide adenine diphosphate; NMN = nicotinamide mononucleotide; Pro-Gly-Pro = proline-glycine-proline; Thr-Phe = threonine-phenylalanine.

a 

Significant after Bonferroni multiple-test correction.

Table Graphic Jump Location
TABLE 4 ]  Correlations Between Representative Metabolites and Airway Infection

See Table 1 and 3 legends for expansion of abbreviations.

References

Rowe SM, Miller S, Sorscher EJ. Cystic fibrosis. N Engl J Med. 2005;352(19):1992-2001. [CrossRef] [PubMed]
 
Alton EW, Davies JC, Geddes DM. Biomarkers for cystic fibrosis: are we progressing? Am J Respir Crit Care Med. 2007;175(8):750-751. [CrossRef] [PubMed]
 
Mayer-Hamblett N, Aitken ML, Accurso FJ, et al. Association between pulmonary function and sputum biomarkers in cystic fibrosis. Am J Respir Crit Care Med. 2007;175(8):822-828. [CrossRef] [PubMed]
 
Esther CR Jr, Alexis NE, Clas ML, et al. Extracellular purines are biomarkers of neutrophilic airway inflammation. Eur Respir J. 2008;31(5):949-956. [CrossRef] [PubMed]
 
Wolak JE, Esther CR Jr, O’Connell TM. Metabolomic analysis of bronchoalveolar lavage fluid from cystic fibrosis patients. Biomarkers. 2009;14(1):55-60. [CrossRef] [PubMed]
 
Yang J, Eiserich JP, Cross CE, Morrissey BM, Hammock BD. Metabolomic profiling of regulatory lipid mediators in sputum from adult cystic fibrosis patients. Free Radic Biol Med. 2012;53(1):160-171. [CrossRef] [PubMed]
 
de Laurentiis G, Paris D, Melck D, et al. Metabonomic analysis of exhaled breath condensate in adults by nuclear magnetic resonance spectroscopy. Eur Respir J. 2008;32(5):1175-1183. [CrossRef] [PubMed]
 
Sofia M, Maniscalco M, de Laurentiis G, Paris D, Melck D, Motta A. Exploring airway diseases by NMR-based metabonomics: a review of application to exhaled breath condensate. J Biomed Biotechnol. 2011;2011:403260.
 
Sagel SD. Noninvasive biomarkers of airway inflammation in cystic fibrosis. Curr Opin Pulm Med. 2003;9(6):516-521. [CrossRef] [PubMed]
 
Sumner SJ, Burgess JP, Snyder RW, Popp JA, Fennell TR. Metabolomics of urine for the assessment of microvesicular lipid accumulation in the liver following isoniazid exposure. Metabolomics. 2010;6(2):238-249. [CrossRef] [PubMed]
 
Ransohoff DF. How to improve reliability and efficiency of research about molecular markers: roles of phases, guidelines, and study design. J Clin Epidemiol. 2007;60(12):1205-1219. [CrossRef] [PubMed]
 
Qiu Y, Cai G, Su M, et al. Serum metabolite profiling of human colorectal cancer using GC-TOFMS and UPLC-QTOFMS. J Proteome Res. 2009;8(10):4844-4850. [CrossRef] [PubMed]
 
Esther CR Jr, Jasin HM, Collins LB, Swenberg JA, Boysen G. A mass spectrometric method to simultaneously measure a biomarker and dilution marker in exhaled breath condensate. Rapid Commun Mass Spectrom. 2008;22(5):701-705. [CrossRef] [PubMed]
 
Esther CR Jr, Lazaar AL, Bordonali E, Qaqish B, Boucher RC. Elevated airway purines in COPD. Chest. 2011;140(4):954-960. [CrossRef] [PubMed]
 
Zhou YH, Barry WT, Wright FA. Empirical pathway analysis, without permutation. Biostatistics. 2013;14(3):573-585. [CrossRef] [PubMed]
 
Evans AM, DeHaven CD, Barrett T, Mitchell M, Milgram E. Integrated, nontargeted ultrahigh performance liquid chromatography/electrospray ionization tandem mass spectrometry platform for the identification and relative quantification of the small-molecule complement of biological systems. Anal Chem. 2009;81(16):6656-6667. [CrossRef] [PubMed]
 
Bajad SU, Lu W, Kimball EH, Yuan J, Peterson C, Rabinowitz JD. Separation and quantitation of water soluble cellular metabolites by hydrophilic interaction chromatography-tandem mass spectrometry. J Chromatogr A. 2006;1125(1):76-88. [CrossRef] [PubMed]
 
Kamatani N, Carson DA. Dependence of adenine production upon polyamine synthesis in cultured human lymphoblasts. Biochim Biophys Acta. 1981;675(3-4):344-350. [CrossRef] [PubMed]
 
Birrer P, McElvaney NG, Rüdeberg A, et al. Protease-antiprotease imbalance in the lungs of children with cystic fibrosis. Am J Respir Crit Care Med. 1994;150(1):207-213. [CrossRef] [PubMed]
 
Esther CR Jr, Olsen BM, Lin FC, Fine J, Boucher RC. Exhaled breath condensate adenosine tracks lung function changes in cystic fibrosis. Am J Physiol Lung Cell Mol Physiol. 2013;304(7):L504-L509. [CrossRef] [PubMed]
 
Grasemann H, Shehnaz D, Enomoto M, Leadley M, Belik J, Ratjen F. L-ornithine derived polyamines in cystic fibrosis airways. PLoS ONE. 2012;7(10):e46618. [CrossRef] [PubMed]
 
Gaggar A, Jackson PL, Noerager BD, et al. A novel proteolytic cascade generates an extracellular matrix-derived chemoattractant in chronic neutrophilic inflammation. J Immunol. 2008;180(8):5662-5669. [CrossRef] [PubMed]
 
Haslam PL, Baughman RP. Report of ERS Task Force: guidelines for measurement of acellular components and standardization of BAL. Eur Respir J. 1999;14(2):245-248. [CrossRef] [PubMed]
 
Muhlebach MS, Stewart PW, Leigh MW, Noah TL. Quantitation of inflammatory responses to bacteria in young cystic fibrosis and control patients. Am J Respir Crit Care Med. 1999;160(1):186-191. [CrossRef] [PubMed]
 
Eltzschig HK, Sitkovsky MV, Robson SC. Purinergic signaling during inflammation. N Engl J Med. 2012;367(24):2322-2333. [CrossRef] [PubMed]
 
Esther CR Jr, Peden DB, Alexis NE, Hernandez ML. Airway purinergic responses in healthy, atopic nonasthmatic, and atopic asthmatic subjects exposed to ozone. Inhal Toxicol. 2011;23(6):324-330. [CrossRef] [PubMed]
 
Boda D. Role of hyperuricaemia in critically ill patients especially newborns. Acta Paediatr Acad Sci Hung. 1984;25(1-2):23-32.
 
Quinlan GJ, Lamb NJ, Tilley R, Evans TW, Gutteridge JM. Plasma hypoxanthine levels in ARDS: implications for oxidative stress, morbidity, and mortality. Am J Respir Crit Care Med. 1997;155(2):479-484. [CrossRef] [PubMed]
 
Gandhi V, Balakrishnan K. Pharmacology and mechanism of action of forodesine, a T-cell targeted agent. Semin Oncol. 2007;34(6 suppl 5):S8-S12. [CrossRef] [PubMed]
 
Griese M, Kappler M, Gaggar A, Hartl D. Inhibition of airway proteases in cystic fibrosis lung disease. Eur Respir J. 2008;32(3):783-795. [CrossRef] [PubMed]
 
Thomas SR, Ray A, Hodson ME, Pitt TL. Increased sputum amino acid concentrations and auxotrophy ofPseudomonas aeruginosain severe cystic fibrosis lung disease. Thorax. 2000;55(9):795-797. [CrossRef] [PubMed]
 
Barth AL, Pitt TL. The high amino-acid content of sputum from cystic fibrosis patients promotes growth of auxotrophicPseudomonas aeruginosaJ Med Microbiol. 1996;45(2):110-119. [CrossRef] [PubMed]
 
Chen H, Macleod C, Deng B, et al. CAT-2 amplifies the agonist-evoked force of airway smooth muscle by enhancing spermine-mediated phosphatidylinositol-(4)-phosphate-5-kinase-gamma activity. Am J Physiol Lung Cell Mol Physiol. 2007;293(4):L883-L891. [CrossRef] [PubMed]
 
Chen EY, Wang YC, Chen CS, Chin WC. Functionalized positive nanoparticles reduce mucin swelling and dispersion. PLoS ONE. 2010;5(11):e15434. [CrossRef] [PubMed]
 
Wright FA, Strug LJ, Doshi VK, et al. Genome-wide association and linkage identify modifier loci of lung disease severity in cystic fibrosis at 11p13 and 20q13.2. Nat Genet. 2011;43(6):539-546. [CrossRef] [PubMed]
 
Koch-Nolte F, Fischer S, Haag F, Ziegler M. Compartmentation of NAD+-dependent signalling. FEBS Lett. 2011;585(11):1651-1656. [CrossRef] [PubMed]
 
Paone G, Stevens LA, Levine RL, et al. ADP-ribosyltransferase-specific modification of human neutrophil peptide-1. J Biol Chem. 2006;281(25):17054-17060. [CrossRef] [PubMed]
 
Esther CR Jr, Boysen G, Olsen BM, et al. Mass spectrometric analysis of biomarkers and dilution markers in exhaled breath condensate reveals elevated purines in asthma and cystic fibrosis. Am J Physiol Lung Cell Mol Physiol. 2009;296(6):L987-L993. [CrossRef] [PubMed]
 
Alexis NE, Hu SC, Zeman K, Alter T, Bennett WD. Induced sputum derives from the central airways: confirmation using a radiolabeled aerosol bolus delivery technique. Am J Respir Crit Care Med. 2001;164(10 pt 1):1964-1970. [CrossRef] [PubMed]
 
Fabian P, Brain J, Houseman EA, Gern J, Milton DK. Origin of exhaled breath particles from healthy and human rhinovirus-infected subjects. J Aerosol Med Pulm Drug Deliv. 2011;24(3):137-147. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Supporting Data

Online Supplement

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
Airways Inflammation and COPD*: Epithelial-Neutrophil Interactions
PubMed Articles
Airway inflammation in mild cystic fibrosis. J Cyst Fibros Published online Jun 10, 2016;
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543