GC-MS Based Hepato-toxicology Studies And Related Data-processing Methods With Metabonomics Approaches

Metabonomics, defined as "the quantitative measurement of the multiparametric time-related metabolic responses of a complex system to a pathophysiological intervention or genetic modification", has been widely used in the researches of pharmacology and biological sciences. Metabonomics encompasses comprehensive and simultaneous systematic profiling of metabolites and their systematic temporal changes in response to stimulus of xenobiotics, environment, and genetics in whole organisms. Gas chromatography/mass Spectrometry technique was employed; and appropriate data processing, pattern recognition, and statistical learning methods were used in this study. This methodology was applied for the liver tissue and plasma metabonome inversigation of the carbon tetrachloride (CCl₄) indued liver injure and the hepato-protective effect of silybin, as well as the serum metabolic profile analysis of HBV patients:Firstly, CCl₄ induced liver injury in mice was investigated on the basis of the metabolic profiling of the liver tissue and plasma. After the chemical derivatization and GC/MS analysis, the acquired GC/MS data was aligned by a home-designed algorithm. Correlation analysis and multivariant pattern recognition methods (principle component analysis and partial least square discriminant analysis) were employed to investigate the intrinsic associations among different metabolites and the patterns of metabonomic change. The applied GC/MS analysis was validated with acceptable linearity, precision and repeatability. For the liver tissue analysis, fumarate and malate, as well as four kinds of fatty acids, were detected increased while the CCl₄ intoxication. For the plasma analysis, the bio-effect of CCl₄ was characterized by the elevation of glutamate, citrate, serine and threonine, as well as the decrease ofα-glycerophosphate, docosahexaenoic acid, palmitic acid and oleic acid in plasma. Therefore, the related metabolic pathway perturbations were inferred.Secondly, on the basis of the effect of CCl₄, the hetato-protective effect of silybin is investigated in this study. Orthogonal partial least square-discriminant analysis (OPLS-DA) was processed to analysis the metabonome difference between the control and CCl₄ treated samples. 12 metabolites were selected both in mice liver tissue and plasma metabonome,respectively, via the parameter of variable importance in the projection (VIP). Principle component analysis (PCA) was employed to process the data from the preventive and therapeutic treatment of silybin for CCl₄ induced hepatotoxicity, and the hepato-protective effect of silybin was investigated through the selected responsor metabolites. Silybin could effectively alleviate the mitochondria dysfunction and the perturbation of amino acids metabolism. However, little effect of silybin was founded for the help of the disorders of lipids metabolism.Thirdly, this study developed a modeling approach with the combination of GC/MS and statistical learning techniques for the molecular classification of HBV-caused liver failure patients. K-nearest neighbor (KNN) and fuzzy support vector machine (FSVM) combined with exhaustive search (ES) and 5-fold cross validation, are employed to identify a subset of metabolites (features) that best predict liver failure. On the 3-class dataset with 47 samples and 19 features pre-selected by T-test, 89.36% by 9 features selected with KNN-ES and 93.62% by 6 features selected with FSVM-ES. By the analysis of the frequencies of features selected in multiple KNN-ES/FSVM-ES models, glyceric acid and citric acid are considered as potential diagnostic biomarkers.