Analysis of volatile organic compounds in the breath of patients with stable or acute exacerbation of chronic obstructive pulmonary disease.

Chronic obstructive pulmonary disease (COPD) is a major cause of death worldwide. Acute exacerbations COPD (AECOPD), caused by infectious and non-infectious agents, contribute to an increase in mortality. The diagnostic procedure of AECOPD is mainly based on clinical features. The aim of this pilot study was to identify whether volatile organic compounds (VOCs) in breath could be used to discriminate for acute exacerbated COPD. Three patient groups were included in this controlled study: AECOPD patients (n = 14, age mean ± SD: 71.4 ± 7.46), stable COPD patients (n = 16, age mean ± SD: 66.9 ± 9.05) and healthy volunteers (n = 24, age mean ± SD: 28 ± 6.08). Breath samples were collected by optimizing a sampling strategy developed by us. These samples were then analyzed using a thermal desorption-gas chromatography-time of flight-mass spectrometer (TD-GC-ToF-MS). A total of 105 VOCs were identified in the breath samples. Relevant substances were subsequently selected by overall occurrence rate, the frequency of positive alveolar gradient (AG) (i.e. the difference in exhaled and inhaled VOCs concentration), exclusion of 'smoking related' VOCs and significant differences in AGs between the three groups. These steps dramatically reduced the number of relevant analytes and resulted in 12 key VOCs having discriminative values. The performance of patients' classification described by the Receiver Operating Characteristic (ROC) curve using all 12 substances delineates an area under the curve (AUC) of 0.97. A further reduction to four VOCs (AGs only different between AECOPD and COPD) delineates an AUC of 0.92. These results indicate that breath analysis with TD-GC-ToF-MS holds promise for an accurate and easy to perform differential diagnosis between AECOPD and COPD. In this regard, ketones were observed at the highest levels in exhaled breath of AECOPD, some of which are also related to potential bacterial pathogens. Using a set of VOCs that can discriminate for AECOPD, the calculated AUCs in ROC curve analysis show far superior results in comparison to serum AECOPD biomarkers, such as C-reactive protein. The identified VOCs should be further investigated in translational studies addressing their potential for developing highly specific nanosensors for breath gas analysis which would give clinicians a tool for non-invasive diagnosis of AECOPD at the point of care.