| Part ? Plasma metabolomic analysis of patients with acute pancreatitis based on LC-MSObjective To analyze the plasma samples of patients with acute panereatitis of different severity by liquid chromatography and mass spectrometry(LC-MS),Screen out the characteristic ions of plasma metabolic profiles in patients with acute pancreatitis of different severity and identify the substance.To analyse the biological significance of differential metabolites through metabolic pathways,and preliminary assess the potential clinical value of these metabolites in the diagnosis and severity assessment of acute pancreatitis.Methods From July 2017 to Deceruber 2017,plasma samples from 87 patients with AP in the First Affiliated Hospital of Soochow University were collected and divided into 2 groups:48 patients with mild pancreatitis,39 patients with severe pancreatitis,and then 30 healthy volunteers were tanken as controls.Plasma small molecule metabolites were dectected by LC-MS.Principal component analysis(PCA),partial least squares discriminant analysis(PLS-DA),orthogonal-partial least squares discriminant analysis(OPLS-DA)were used to estimate the inter-group variation information of the sample.The models were validated by cross-validation,displacement test,and training set-test set to evaluate the classification effect and prediction ability of the classification models.The first principal component VIP value>1 was combined with Mann-Whitney-Wilcoxon Test pvalue<0.05 to screen for potential differential metabolites.The identification of differential metabolites was first confirmed based on the exact molecular weight,presuming its possible molecular formula.The differential metabolites were then confirmed based on the MS/MS fragmentation information comparison public database like Metlin,Human Metabolome Database(HMDB),massbank,LipidMaps,Mzclound,and so on.Result There were 1314 precursor molecules in positive ion mode and 1110 precursor molecules in negative ion mode,and data were derived to excel for subsequent analysis.Differences could be observed among different groups by the general ion flow base peak chromatograms(BPC).The PCA,PLS-DA,and OPLS-DA models were all with good interpretability(R2X>0.5).It could be observed from the PCA score map that the healthy volunteers,MAP,and SAP plasma samples had a certain clustering trend and a good classification model was obtained.In the PLS-DA model,the differences among groups were further highlighted,both in positive and negative ion mode,more than 95%of the samples were consistent with model discrimination(R2Y>0.95)and had high predictability(Q2>0.85).In the OPLS-DA model,62.5%of the samples(20 controls,22 SAPs,30 MAPs)were randomly selected as training sets for constructing the OPLS-DA model,extracting 38.5%(10 controls,17 SAPs,18 cases of MAP)were used as test sets to test the reliability of the model.There were 5 principal components(2 predicted principal components and 3 orthogonal principal components).It was clear that the three groups of clusters were more prominent on the OPLS-DA score map,and the model had good fitness and predictability(R2X=0.357,R2Y=0.948,Q2=0.839).The samples of the test set could be correctly classified into grouping clusters in the figure.The training set-test set model proved that the model had good classification and prediction ability.35 differential metabolites were screened out and identified,15 of which increased gradually with the severity of pancreatitis,suggesting that they might be potential biomarkers of the severity of AP.They were(R)-(+)-2-Pyrrolidone-5-carboxylic acid,(S)-(-)-2-hydroxyisocaproic acid,acetylcamitine,beta-Alanine,creatine,indolelactic acid,L-carnitine,kynurenine,L-palxnitoylcarnitine,octanoylcamitine,L-phenylalanine,malic acid,oxoglutaric acid,threonate,uric acid.The levels of another 13 metabolites,namely 2-aminoisobutyric acid,3-buten-1-amine,angelic acid,D(-)-beta-hydroxy butyric acid,L-glutamate,L-glutamine,L-histidine,L-valine,L-leucine,nonanedioic acid,succinic aldehyde,N,N-dimethylglycine,sphingosine-1-phosphate increased in AP compared to control group,which were significantly increased in MAP while the SAP group were lower than the MAP group,suggesting that these might be potential metabolite indicators for the early diagnosis of AP or characteristic indicators of MAP.The level of indoxyl sulfate in SAP was significantly higher than that in the other two groups,which might be a predictor of SAP.Conclusion LC-MS metabolomics can effectively identify the changes of plasma metabolites in patients with different severity of AP.The screening of differential metabolites has potential clinical application value for the classification of AP severity.Part ? Urine metabolomic analysis of patients with acute pancreatitis based on LC-MSObjective To analyze the urine samples of patients with acute pancreatitis of different severity by liquid chromatography and mass spectrometry(LC-MS),Screen out the characteristic ions of urine metabolic profiles in patients with acute pancreatitis of different severity and identify the substance.To analyse the biological significance of differential metabolites through ruetabolic pathways,and preliminary assess the potential clinical value of these metabolites in the diagnosis and severity assessment of acute pancreatitis.Methods From July 2017 to December 2017,urine samples from 67 patients with AP in the First Affiliated Hospital of Soochow University were collected and divided into 2 groups:40 patients with mild pancreatitis,27 patients with severe pancreatitis,and then 25 healthy volunteers were tanken as controls.Urine small molecule metabolites were dectected by LC-MS.Principal component analysis(PCA),partial least squares discriminant analysis(PLS-DA),orthogonal-partial least squares discriminant analysis(OPLS-DA)were used to estimate the inter-group variation information of the samples.The models were validated by cross-validation,displacement test,and training set-test set to evaluate the classification effect and prediction ability of the classification models.The first principal component VIP value>1 was combined with Mann-Whitney-Wilcoxon Test pvalue<0.05 to screen for potential differential metabolites.The identifieation of differential metabolites was first confirmed based on the exact molecular weight,Presuming its possible molecular formula.The differential metabolites were then confirmed based on the MS/M5 fragmentation information eomparison public database like Metlin,Human Metabolome Database(HMDB),massbank,LipidMaps,Mzclound,and so on.Result There were 893 precursor molecules in positive ion mode and 1878 precursor molecules in negative ion mode,and data were derived to excel for subsequent analysis.Differences could be observed among different groups by the general ion flow base peak chromatograms(BPC).The PC A,PLS-DA,and OPLS-DA models were all with good interpretability(R2X>0.5).It could be observed from the PCA score map that the healthy volunteers,MAP,and SAP urine samples had a certain clustering trend and a good classification model was obtained.In the PLS-DA model,the differences among groups were further highlighted,both in positive and negative ion mode,more than 80%of the samples were consistent with model discrimination(R2Y>0.8)and had high predictability(Q2>0.7).In the OPLS-DA model,60%of the samples(15 controls,15 SAPs,25 MAPs)were randomly selected as training sets for constructing the OPLS-DA model,extracting 40%(10 controls,12 SAPs,15 cases of MAP)were used as test sets to test the reliability of the model.There were 5 principal components(2 predicted principal components and 3 orthogonal principal components).It was clear that the three groups of clusters were more prominent on the OPLS-DA score map,and the model had good fitness and predictability(R2X=0.318,R2Y=0.891,Q2=0.697).The samples of the test set could be correctly classified into grouping clusters in the figure.The training set-test set model proved that the model had good classification and prediction ability.20 differential metabolites were screened out and identified,14 of which,namely 3-methyl pyruvic acid,Malic acid,L-Camitine,L-Methionine,L-Lysine,L-Phenylalanine,L-Leucine,L-Tryptophan,L-Histidine,Ketovaline,L-Tyrosine,gamma-Glutamylphenylalanine,D-Gluconic acid,Kynurenine,increased gradually with the severity of pancreatitis,while the eontent of adenosine decreased as the severity of pancreatitis progresses,suggesting that they might be potential biomarkers of the severity of AP.The levels of another 4 metabolites,namely D(-)-beta-hydroxy butyric acid,Succinic acid semialdehyde,CMPF,L-2-Aminoadipic acid,increased in AP compared to control group,which were significantly increased in MAP and subsequently decreased in SAP,suggesting that these might be potential metabolite indicators for the early diagnosis of AP or characteristic indicators of MAP.The level of Choline in SAP was significantly higher than that in the other two groups,which might be a predictor of SAP.Conclusion LC-MS metabolomics can effectively identify the changes of urine metabolites in patients with different severity of AP.The screening of differential metabolites has potential clinical application value for the classification of AP severity. |
PDF Full Text Request