Primary Exploration Of Children With Acute Lymphoblastic Leukemia Using Non-targeted And Targeted Metabolomics

Background and Objective: Metabolomics,by comprehensively qualitative and quantitative analysis of the concentration of metabolites in blood,urine,feces and tissues and changes in their metabolic pathways,can most directly and truly reflect the health and disease status of human body,and to early warn and reflect the degree of disease.Metabolomics is the most powerful technology to predict phenotypes in all "omics" studies of systems biology.In order to screen potential biomarkers that can guide clinical diagnosis and treatment,non-targeted and targeted metabolomics technology and theoretical knowledge were used to analyze the body fluid and feces samples of children with acute lymphoblastic leukemia(ALL)to study the changes of endogenous small molecule metabolites.Methods: In the first part,the subjects were divided into three groups: the control group(20cases),CSF samples with suspected encephalitis but normal cerebrospinal fluid(CSF)routine and biochemical examination;the ALL without central metastasis group(31 cases),among thse samples,15 were CSF pathological and flow cytometry both negative,16 were CSF pathological negative and CSF routine with 0 white blood cell count;the central nervous system leukemia(CNSL)group(31 cases),the results of pathological examination or flow cytometry were positive.In addition,CSF samples of one case of CNSL at different stages of chemotherapy were tracked and determined.Then we established an ultra-high performance liquid chromatography-Q extractive hybrid quadrupole orbitrap high-resolution accurate mass spectrometry(UPLC-QE/MS)based non-targeted metabolomics method to detect the above samples.Proteo Wizard software was used to convert the original data into mz ML format.XCMS was used to complete the data pre-processing process.OSI-SMMS software combined with the self-built database were used for metabolites identification.To eliminate individual differences of samples and systematic errors caused by the instrument,peak area normalization method was used.Multivariate statistical analysis was carried out by using multivariate statistical software SIMCA-P 14.1,and orthogonal projections to latent structures-Discriminant analysis(OPLS-DA)model was used to screen metabolites that contribute significantly to differentiating different groups of samples,and the variables importance in projection(VIP)greater than 1.0 wereselected as the differential metabolites.Finally,the Graph Pad Prism 8 software was used for One-way ANOVA of multiple comparisons.The metabolites with both VIP > 1.0 and P < 0.05 were identified as differential metabolites.The second part is about serum and feces metabolomics of 28 healthy children and 33 untreated ALL children.We established the ultra-high performance liquid chromatography-tandem mass spectrometry(UPLC-MS/MS)technology to detect the above samples.Data pretreatment was carried out with Analyst software,version 1.5.2.Multivariate statistical analysis was carried out by SIMCA-P 14.1 software,and the OPLS-DA was used to differentiate the normal control and ALL groups.Variables with VIP >1.0 were considered to have multivariate statistical significance.Wilcoxon rank-sum tests and logistic regression were carried out by SAS software for further analysis.Bile acids with VIP >1.0 and P < 0.05 were identified as differential bile acids.Results: In the first part,among the statistically significant variables identified using VIP values(VIP >1.0)in the OPLS-DA model and the T-test(p < 0.05),a total of metabolites13 and17 different metabolites were identified as potential biomarkers for ALL and CNSL,respectively.There were 13 different metabolites were found in CSF of children with CNSL compared with ALL children without central metastasis.One-way ANOVA of multiple comparisons showed that there were significant differences in nine metabolites in CSF of CNSL children compared with the control group and the ALL without central metastasis group,such as phenylalanine,lysine,tryptophan,leucine,alanyl carnitine,creatinine,inositol,indolin and arachidonic acid.The concentrations of these nine metabolites decreased gradually with the development of ALL central metastasis.Follow-up results of CSF samples from one child with CNSL at different stages of chemotherapy showed that the above nine metabolites in CSF increased significantly with the progress of chemotherapy.In the second part of the study,11 and 13 bile acids in serum and fecal samples were found significantly different from those of normal control group,respectively.Wilxocon rank sum test and logistic regression analysis showed that there were 6bile acids can be used as potential biomarkers for ALL(i.e.serum deoxycholic acid,glycocholic acid,taurine ?-rat bile acid and taurocholic acid,fecal ?-rat bile acid and 12-ketolithocholic acid).Conclusion: The non-targeted metabolomics method based on UPLC-QE/MS combined with multivariate statistical analysis can be used to screen small molecule metabolites closely related to central metastasis of ALL.The targeted metabolomics technology based on UPLC-MS/MS combined with multivariate statistical analysis can be used to screen key bile acids in serum and feces closely related to ALL.The metabolomics technology plays important roles in finding and discovering valuable biomarkers.