Breath volatile biomarkers are capable of distinguishing patients with various cancers. However, high throughput analytical technology is a prerequisite to a large-cohort study intended to discover reliable breath biomarkers for cancer diagnosis. Single-photon ionization (SPI) is a universal ionization technology, and SPI-mass spectrometry (SPI-MS) shows a remarkable advantage in the comprehensive detection of volatile organic compounds (VOCs), in particular, nonpolar compounds. In this study, we have introduced SPI-MS coupled with on-line thermal desorption (TD-SPI-MS) to demonstrate nontarget analysis of breath VOCs for gastric cancer patients. The breath fingerprints of the gastric cancer patients were significantly distinct from that of the control group. Acetone, isoprene, 1,3-dioxolan-2-one, phenol, meta-xylene, 1,2,3-trimethylbenzene, and phenyl acetate showed higher relative peak intensities in the breath profiles of gastric cancer patients. A diagnostic prediction model was further developed by using a training set (121 samples) and validated with a test set (53 samples). The predication accuracy of the developed model was 96.2%, and the area under the curve (AUC) of the receiver operator characteristic curve (ROC) was 0.997, indicating a satisfactory prediction ability of the developed model. Thus, by taking gastric cancer as an example, we have shown that TD-SPI-MS will be a promising tool for high throughput analysis of breath samples to discover characteristic VOCs in patients with various cancers.
Keywords: exhaled breath; gastric cancer; logistics regression; mass spectrometry; noninvasive diagnosis; on-line preconcentration.
© 2020 John Wiley & Sons, Ltd.