Exhaled breath analysis for diagnosis and phenotyping in obstructive lung diseases

UoM administered thesis: Phd

  • Authors:
  • Baharudin Ibrahim

Abstract

Introduction: Asthma and chronic obstructive pulmonary disease (COPD) are heterogeneous diseases with a wide range of clinical manifestations not adequately described within the current diagnostic criteria. Exhaled breath analysis may provide a novel method for diagnosing and phenotyping these diseases. Our aim was to ascertain patterns of breath volatile organic compounds (VOCs) and nuclear magnetic resonance (NMR) spectral regions identifying diseased patients and subgroups determined by treatment requirement, asthma control, exacerbation frequency and inflammatory phenotypes. The validity and reproducibility of the methodology and the outcome were also investigated.Methods: Three separate clinical studies (two involving exhaled gas and one involving breath condensate) were conducted, as well as validation studies. In exhaled gas analysis, the adaptive breath sampler developed by Basanta et al was modified; efficiency of air supply and air filter and the reproducibility and stability of VOCs in storage were determined by comparing breath chromatograms. Concentrated late-expiratory breath samples were collected from asthmatics, COPD subjects and healthy controls. In the asthmatic group, sputum induction with hypertonic saline, fraction exhaled nitric oxide (FeNO) measurement and asthma control questionnaire (ACQ) were performed. In COPD subjects, sputum induction and exacerbation frequency were collected. In the exhaled breath condensate (EBC) study, similar data were collected in asthmatics and healthy controls. Breath samples were analysed using gas chromatography time-of-flight mass spectrometry (GC-TOF-MS) while EBC was analysed using NMR spectroscopy. Discriminatory compounds or NMR spectral regions were identified by univariate logistic regression, followed by multivariate analysis: 1. principal component analysis (PCA); 2. multivariate logistic regression; 3. receiver operating characteristic (ROC) analysis. The reproducibility was assessed using intraclass correlation coefficient (ICC).Results: In the COPD exhaled breath study, 11 VOCs significantly discriminated the COPD and healthy controls with AUROC of 0.74. The AUROC for phenotype discrimination was 0.83, 0.90, 0.94, 0.96 and 0.97 for inhaled corticosteroid (ICS) use, sputum eosinophilia (1% and 2% cut-off), neutrophilia (median cut-off) and exacerbation frequency respectively. In the asthma study, 15 VOCs significantly discriminated the two groups with AUROC of 0.93. The AUROC for phenotype discrimination was 0.96, 0.98, 0.90 and 0.97 for ICS use, eosinophils (2% cut-off), neutrophils (40% cut-off) and asthma control respectively. In EBC analysis, AUROC for asthmatics vs controls comparison was 0.96. Phenotyping results in this study were less good: only ICS use and sputum neutrophilia (65% cut-off) were clearly classified with AUROC of 0.89 and 0.88 while eosinophilia (3% cut-off) and asthma control had poor discrimination; 0.69 and 0.62 respectively. Breath VOC reproducibility varied greatly depending on the class of compounds studied, while for the EBC analysis, reproducibility was moderate to very good (ICCs in the range of 0.42-0.99).Conclusions: We have demonstrated the ability of breath analysis in discriminating asthmatics and COPD subjects from controls. Exhaled breath analysis was also able to phenotype these patients based on steroid treatment, sputum inflammatory cells, exacerbation frequency and asthma control. This metabolomic approach could provide a novel, non-invasive method of diagnosing and phenotyping obstructive lung diseases in the future.

Details

Original languageEnglish
Awarding Institution
Supervisors/Advisors
Award date1 Aug 2011