Discovering the metabolic pathway of liver disease by breath mass spectrometry combined with machine learning

Abstract

On account of the low concentration and complex background, most methods for detecting VOCs in exhaled breath samples required preconcentration prior to instrumental analysis. In the current study, a simple and rapid method is developed for analyzing the volatile organic compounds (VOCs) in exhaled breath by extractive electrospray ionization mass spectrometry (EESI-MS). Partial least square discriminant analysis (PLS-DA) using full scan mass spectrum data was used to distinguish exhaled breath of healthy volunteers, chronic hepatitis B patients, and liver failure patients. More VOCs were detected in exhaled breath samples from the volunteers including healthy adults, chronic hepatitis B patients, and liver failure patients by the enclosed EESI-MS method without any sample pretreatment in comparison to the traditional methods. The 9 new breath markers, including 2-hydroxyethanal, 2-methylpropene, butan-2-one, 2,5-dimethylfuran, 2-ethylprop-2-enoic acid, 2,4-dimethylpyridine, 2-methylbutanoic acid, cyclopentanone and 5-hexanolactone, were identified using high resolution tandem mass spectrometry, and some highly correlated metabolic pathways including tryptophan metabolism, tyrosine metabolism, butanoate metabolism, and fatty acid biosynthesis were found. EESI-MS² method, a promising non-invasive diagnostic tool, is robust and suitable for analysis of VOCs, and may provide further insight into the pathophysiologic processes and pathways leading to chronic hepatitis B and its complications.