| Objective:The prevention and treatment of Acute-on-chronic liver failure(ACLF)has attracted more attentionin China due to its srious harm to human health.In this dissertation,the serum prognostic metabolites of ACLF patients were firstly detected by nuclear magnetic resonance(~1HNMR)based metabolomics.Some metabolites were further quantified by ultra-high performance liquid chromatography-mass spectrometry(UPLC-MS/MS)for screening the diagnostic biomarkers of ACLF patients.Methods:1.~1HNMR based non-target metabolomics technique was used to detect and screen the differential metabolites in serum of 127 ACLF patients including 42patients who were dead in 28-day after being admitted to hospital and 85 28-day survival patients using the analysis of multi-statistical models.2.An UPLC-MS/MS quantitative analysis method was established to determine the contents of choline and purine metabolites in biological samples.Blood samples were subjected to protein precipitation with acetonitrile.The zic-hilic hydrophilic chromatography column and 10mmol/mL ammonium acetate-0.01%formic acid as mobile phase with gradient elution were used in the UPLC system.3.Another UPLC-MS/MS quantitative method was established to determine the serum levels of bile acid class in biological samples.Blood samples were subjected to protein precipitation with acetonitrile.A C18 column chromatography column,10mmol/ml ammonium acetate was used as mobile phase.The injection volume was3uL,the flow rate was 0.25 mL/min,and the column temperature was set at 45℃.4.In addition,an UPLC-MS/MS quantitative method was developed for measuring the urinary content of pterinates in ACLF patients.Results:1.The prognosis of ACLF by ~1HNMR non-target metabolomics showed that:the ratio of Glutamine/Creatinine(46.12±3.79 vs 61.11±5.65,AUROC=0.629),Carnitine/Creatinine(32.09±3.15vs19.88±2.33,AUROC=0.666),Acetate/Creatinine(7.61±0.57 vs 4.99±0.48,AUROC=0.656)in the serum metabolites of the ACLF survival and death groups were significantly different(independent sample T test,P<0.05).ROC analysis was carried out on the classification of ACLF death group and survival group with 21 different metabolites as variables.Logistic regression screened out combination of Glutamine/Creatinine、Carnitine/Creatinine、MELD Scores and INR variables to establish a logistic model.Using the model,the prediction of 28-day death and survial ACLF patient with the AUROC of 0.880,the sensitivity and specificity of 97.6%and 74.7%respectively.2.12 metabolites of choline pathway in serum of ALCF patients were quantitatively determined by UPLC-MS/MS.The results showed that there were significant differences between t 8 metabolites in the ACLF death and survival groups(independent sample T test,P<0.05),of which methionine had the highest diagnostic performance(AUROC=0.708),the sensitivity and specificity were76.70%and respectively of 60.80%,and the content of which was increased in28-day death group compared with the survival group.3.10 kinds of purine metabolites were measured by UPLC-MS/MS.The results showed that there were significant differences between the 5 metabolites in the ACLF death and survival groups(independent sample T test,P<0.05),of which xanthine had the highest diagnostic performance(AUROC=0.707).Thesensitivity and specificity were 66.40%and 74.10%respectively,and the content of which was increased in 28-day death group compared with the survival group.4.22 kinds of bile acids were quantitatively determined by UPLC-MS/MS.The results showed that there were significant differences between the 3 metabolites in the ACLF death and survival groups(independent sample T test,P<0.05).The Deoxycholic acid+Cheno-deoxycholic acid has the highest diagnostic performance(AUROC=0.622).The sensitivity and specificity of which were 82.20%and 10.40%,respectively,and and the content of which was increased in 28-day death group compared with the survival group.A binary logistic regression analysis was performed by integrate quantitative data of 33 metabolites including choline,purine and bile acids pathways.Xanthine、Deoxycholic acid、Cheno-deoxycholic acid were selected to establish the logistic model.The AUROC for ACLF prognosis was 0.762,and the sensitivity and specificity of 59.1%and 82.2%,respectively.The discriminant analysis model was established by linear combination of 3 variables including Xanthine、Cysta、Deoxycholic acid、Cheno-deoxycholic acid and the average classification accuracy was 73.3%.The AUROC was 0.788.The cross-validation results showed that the discriminant model was reliable.5.Another quantitative method by UPLC-MS/MS was established for the detection of 4 kinds of pterins metabolites in urine.Compared with the ACLF survival group,BH2/Creatinine(0.01±0.00 VS 0.02±0.00,AUROC=0.028),Neopterin/Creatinine(0.03±0.00 VS 0.04±0.00,AUROC=0.000),and Methotrexate/Creatine(189.69±16.69 VS 442.73±59.74,AUROC=0.000)had significant differences in the three types of pterins,of which the AUROC of Neopterin/Creatinine was 0.851.The logistic model established by combining Neopterin/Creatinine and Sepiapterin/Creatinine as variables for the 28-day prognosis of ACLF patients had an average accuracy of 79.5%,AUROC of 0.911,with the sensitivity and specificity of 90.90%and 81.80%,respectively.Conclusion:The ~1HNMR based metabolomics revealed that the short-term prognosis of ACLF patients had different serum metabolic phenotypes,including the alterations in multiple metabolic pathways.In this study,nearly 60 serum and urine ACLF-related metabolites were quantified and screened for classification of 28-day death and survial group in ACLF patients The results showed that the combination of Xanthine、Cysta、Deoxycholic acid、Cheno-deoxycholic acid and other three variables in serum has a highpredictive effect on the prognosis of ACLF patients.Neopterin/Creatinine and Sepiapterin/Creatinine have the highest predictive value for patients with ACLF in short-term prognosis.The metabolites in neopterin pathway were therefore deserved to be prognostic biomarkers,which should be further verified. |
PDF Full Text Request