| Objective:Meningitis is a common central nervous system infectious disease,this illness has two forms:bacterial meningitis(BM)and viral meningitis(VM).To date,few reports on meningitis diagnosis was employed based on metabolomics technology.Here,we combined non-target and target metabolomics techniques to investigate the potential biomarkers distinguishing meningitis from non-meningitis(control group,CG),BM and VM in cerebrospinal fluid(CSF)and serum samples in children.The correlation between the content of metabolites in cerebrospinal fluid and serum was determined.Method:1.67(20VM,28BM,19CG)serum samples and 58(16VM,23BM,19CG)cerebrospinal fluid samples were firstly investiganted by the nuclear magnetic resonance(~1H-NMR)based non-target metabolomics.2.An UPLC-MS/MS method was established for measuing the significantly related metabolites detected by ~1H-NMR.A total of 622(209VM,83BM,330CG)cerebrospinal fluid samples and 580(168VM,64BM,348CG)serum samples were employed to study.Logistic regression and receiver operating characteristic curve(ROC)analysis were used to screen the biomarkers and their combinations.3.The alteration of metabolites in serum and cerebrospinal fluid with age were analyzed;the correlation of 529 pairs of cerebrospinal fluid and matched serum data were analyzed,and the content differences of metabolites in serum and cerebrospinal fluid was compared;age-matched cerebrospinal fluid and serum samples were further analyzed to detect the diagnostic biomarkers.Result:1.~1H-NMR metabolomics results showed that acetic acid,citric acid and creatinine were statistically different in the cerebrospinal fluid samples between the children's meningitis group and the control group.Combination of these three variables to establish a Logistic model,the AUC of distinguishing above two groups was 0.794 and the classification accuracy was 74.1%;combination of alanine,isoleucine,lactic acid and valine levels in the cerebrospinal fluid samples,the AUC of distinguishing viral meningitis and bacterial meningitis was 0.891,and the classification accuracy was 74.4%;In the serum samples of meningitis and control group,the AUC was 0.755 when combination of acetic acid and formic acid,and the classification accuracy was 81.8%;the AUC was 0.848,and the classification accuracy was 76.6%in distinguishing the viral meningitis and bacterial meningitis group by combination of serum 3-hydroxybutyric acid,choline and alanine.Our result indicated that the differential metabolites contained choline substances both in the cerebrospinal fluid and serum samples.Therefore,choline substances are worthy of quantifation in further study.2.We used established UPLC-MS/MS methods and quantitatively analyzed 17metabolites in choline and bile acids pathways in children's cerebrospinal fluid and serum with large numbers of clinical samples.The results showed that the contents of betaine(43.77±0.71 vs 119.96±4.17)and choline(42.74±1.32 vs 147.23±6.64)in the cerebrospinal fluid samples of VM and BM were significantly different,and the diagnostic value of betaine was the highest with AUC value of0.980.Combined betaine,choline and taurodeoxycholic acid to achieve differential diagnosis,the AUC value is 0.993 withclassification accuracy of 96.6%,sensitivity and specificity of 96.4%and 95.2%,respectively;Among the cerebrospinal fluid samples of the control group and meningitis patients,taurocholic acid had the highest diagnostic value,with an AUC value of 0.706.Combined with creatinine,betaine,taurocholic acid and deoxycholic acid to achieve differential diagnosis,the AUC value is 0.862 with the classification accuracy of 76.2%,sensitivity and specificity of 82.2%and 74.5%,respectively.3.Among the serum samples of VM and BM,taurocholic acid has the highest diagnostic value,with an AUC value of 0.940.Combined with glycoursodeoxycholic acid,cholic acid and taurodeoxycholic acid as varibles to the differential diagnosis of VM and BM,the AUC is 0.972,the classification accuracy was 95.3%,and the sensitivity and specificity were 95.3%and 96.4%,respectively.Among the serum samples of the control group and meningitis patients,taurocholic acid had the highest diagnostic value,with an AUC value of 0.739.Combined betaine,carnitine,taurocholic acid and cholic acid as varibles to achieve differential diagnosis,its AUC is 0.814,with the classification accuracy of 72.2%,sensitivity and specificity of61.5%and 92.7%,respectively.4.253 children cerebrospinal fluid samples were grouped by age(15 groups),the results showed that large changes of choline,betaine,taurocholic acid,taurocholic acid,taurocholic acid and glycocholic acid in cerebrospinal fluid were observed by age,especially in patients whose age is less one year.Serum levels of glycosaminocholic acid,taurocholic acid,taurodeoxycholic acid,and glycocholic acid were also observed to be changed significantly with age.Most of metabolites has a positive correlation between serum and cerebrospinal fluid,and the Pearson correlation coefficient of taurocholic acid reaches 0.666.5.After being matched by age of the cerebrospinal fluid samples the diagnostic value between VM and BM by the content of taurocholic acid is the highest with the AUC of 0.983,In distinguishing between VM and control group,the diagnostic value of glycocholic acid is the highest with AUC value of 0.862.Combinion of contents of taurocholic acid,cholic acid,and glycocholic acid to establish a logistic model,the AUC value increases to 0.911 with the classification accuracy of 86.6%,the sensitivity and specificity of 82.2%and 90.0%,respectively.In distinguishing between BM and the control group,the diagnostic value of glycocholic acid is the highest with the AUC value of 0.852.Combinion of contents of carnitine,deoxycholic acid and glycocholic acid to establish a logistic model,the AUC value increases to 0.891 with the classification accuracy of 84.1%,the sensitivity and specificity of 85.5%and 82.7%,respectively.6.After being matched by age of the serum samples for grouping VM and BM,the diagnostic value of taurocholic acid is the highest(AUC=0.918).Combinion of serum levels of glycocholic acid and deoxycholic acid to establish a logistic model,the classification accuracy is 91.7%with the AUC value of 0.940,the sensitivity and specificity of 78.9%and 100.0%,respectively.While distinguishing between VM and the control group,the diagnostic value of acetylcholine is the highest with AUC of 0.838.The classification accuracy of logistic model in which serum levels of carnitine,acetylcholine and taurodeoxycholic acid was used as varibles is 91.9%with the AUC value of 0.908,the sensitivity and the specificity of 75.3%and100.0%,respectively.In the end,to distinguish BM from the control group,our results showed that serum taurocholic acid is the highest abiltiy with AUC value of0.935.The established logistic model by combination of serum glycocholic acid,taurocholic acid and glycocholic acid showed the classification accuracywas 94.6%with the AUC value of 0.971,the sensitivity and specificity of 91.1%and 93.0%,respectively.Conclusion:There are significant differences in abundance of metabolites in cerebrospinal fluid and serum samples between VM,BM and the control group,and some metabolites were significantly affected by age.Metabolites such as taurocholic acid and glycocholic acid contribute to the differential diagnosis of VM or BM,and our results indicated the feasibility of using bile acids for the diagnosis of meningitis.The alterations of serum levels of some metabolites in meningitis patients are helpful for diagnosis and classification of VM and BM,which can avoid the adverse reactions caused by lumbar puncture. |
PDF Full Text Request