Metabolomic Biomarker Detection Of Serum And Tissue Of Pancreatic Cancer And Its Correlation Study With CT

Part Ⅰ:Serum and tissue metabolomic biomarker screening of pancreatic cancerObjective and Significance:Pancreatic cancer is a malignancy with high degree and is difficult to be diagnosed in early stage.Reliable biomarkers are urgently needed for the detection of PDAC in high-risk cohorts.In this study,non-targeted metabonomics was used to the serum and tissues of pancreatic cancer,with the purpose of screening serum biomarkers for the detection of pancreatic cancer with early stage,and explore the correlation between pancreatic cancer tissues and serum samples.Methods:1.In this study,the serum of 82 patients with PDAC(including 36 patients with early-stage and 46 patients with late-stage PDAC),36 patients with pancreatic benign cystic tumor(BP)and 49 healthy controls(Ctr)were analyzed by non-targeted metabolomics based on ultra-high performance liquid chromatography-mass spectrometry(UHPLC-MS)platform,including sample preparation,chromatography-mass spectrometry,metabolite identification and data analysis.Data analysis included univariate and multivariate statistical analysis.Univariate statistical analysis included Student’s t-test and fold change analysis.Multivariate statistical analysis included Principal Component Analysis(PCA),Partial Least Squares Discriminant Analysis(PLS-DA),and Orthogonal Partial Least Squares Discriminant Analysis(OPLS-DA).This study used a combination of univariate and multivariate analysis to preliminarily screen the common differential metabolites of PDAC vs Ctr and PDAC vs BP,as well as the common differential metabolites of early PDAC vs Ctr and early PDAC vs BP.The intersection of them were taken as the candidate differential metabolites of early PDAC,and then Lasso regression analysis was carried out on these metabolites using R software.Logical regression analysis was used to observe the influence of clinical confounding factors on differential metabolites.Correlation analysis was used to evaluate the correlation between CA19-9 and metabolites.Finally,differential metabolites that were not affected by clinical confounding factors and were not related to CA19-9 were selected to establish the diagnostic model.ROC curve,sensitivity and specificity were used to evaluate the predictive efficiency of the model and combined with CA19-9.Using MetaboAnalyst to realize Hierarchical Clustering Analysis(HCA).KEGG database was used for network and pathway analysis.2.Fifty-one PDAC tissues,40 non-tumor pancreatic tissues and 14 pancreatic cystic tumor tissues were analyzed by non-targeted metabonomics in this study.The analysis method was the same as serum metabonomics.In addition,the expression of screened serum metabolites in tissues was also evaluated.3.VENN diagram was used to compare the common differential metabolites and metabolic pathways of PDAC vs Ctr,PD AC vs BP of both serum and tissue samples,and to analyze the similarity of metabolomics results of the serum and tissue samples.Results:1.Multivariate analysis model had been established in PD AC serum metabolomics.There were significant metabolic differences between PDAC vs Ctr and PDAC vs BP.There were 22 differential metabolites in common both in PD AC vs Ctr and PDAC vs BP,mainly were fatty acids,amino acids,phosphatidylcholine and sphingomyelin.Of these 22 metabolites,19 metabolites were in common with early PDAC vs Ctr and early PDAC vs BP.Lasso regression analysis and logistic regression analysis were applied on the 19 different metabolites,and Embelin,L-pyroglutamic acid,1,2-dioleoyl-sn-glycerol-3-phosphatidylcholine,which can distinguish PDAC and Ctr were considered as the optimal model.AUC,sensitivity and specificity of this panel were 0.938,87.5%and 100%,respectively.The AUC,sensitivity and specificity for distinguishing early-stage PDAC from Ctr are 0.917,91.7 and 80%,respectively.A group of metabolites that can distinguish PDAC from BP were Embelin,D-proline,1,2-dioleoyl-sn-glycerol-3-phosphatidylcholine,and its AUC,sensitivity and specificity are 0.882,88.2%and 83.3%,respectively.After combination with CA19-9,the AUC,sensitivity and specificity for distinguishing PDAC and Ctr were 1.000,100%and 100%,respectively,and distinguishing PDAC and BP were 0.910,76.9%and 100%,respectively.2.Tissue metabolomics results of PDAC showed that OPLS-DA model can well distinguish PDAC vs Ctr,PDAC vs BP.There were 62 differential metabolites both in PDAC vs Ctr,PDAC vs BP,mainly including nucleotides,amino acids,fatty acids,choline phosphate and carbohydrates.Of the 62 metabolites,36 metabolites were in common in both early-stage PDAC vs Ctr,early-stage PDAC vs BP.In this study,the most significant metabolic characteristics of PDAC tissue were down-regulation of amino acids,up-regulation of fatty acids,down-regulation of nucleotides,up-regulation of glycolysis products,down-regulation of tricarboxylic acid circulating metabolites,etc.The screened serum metabolites L-Pyroglutamic acid and D-Proline were also differential metabolites in tissues,which were significantly lower in PDAC than non-tumor pancreatic tissues.3.Twenty-seven metabolites were identified both in the serum and tissues of PDAC vs Ctr,and 17 of them had the same direction of changes in serum and tissues.There were 11 important metabolic pathways both in the serum and tissues of PDAC vs Ctr,including tumor central carbon metabolism,protein digestion and absorption,mineral absorption,aminoacyl TRNA biosynthesis,gamma-aminobutyric acid synapse,unsaturated fatty acid biosynthesis,fatty acid biosynthesis,ABC transport,alanine,aspartic acid and glutamic acid metabolism,arginine metabolism,etc.There were 6 metabolites both in serum and tissues of PD AC vs BP,and only one metabolite with the same direction of change in serum and tissue.There are 7 important metabolic pathways both in the serum and tissues of PD AC vs BP,including ABC transport,tumor central carbon metabolism,tumor choline metabolism,linoleic acid metabolism,unsaturated fatty acid biosynthesis,glycerophospholipid metabolism and fatty acid biosynthesis.Conclusion:1.The panel of metabolites alone or in combination with CA19-9 had good performance.This panel of metabolites has the potential to be used as a supplementary diagnostic marker for PDAC.2.Reprogramming in tissue metabolism of PD AC in this study were obvious than serum.The level of amino acids and nucleotides was down-regulated,reflecting an increase in biosynthesis of proteins and nucleic acids.The up-regulation of fatty acid reflected the increased requirement of energy and tumor cell proliferation.The up-regulation of glycolysis products and the down-regulation of tricarboxylic acid circulating metabolites reflected the increase of aerobic glycolysis in PDAC.3.The consistency of differential metabolites and metabolic pathways between PDAC tissues and serum samples also confirmed the reliability and authenticity of the results of this study.Part Ⅱ Comparative study of CT evaluation and histopathology or surgery of pancreatic cancerObjective and Significance:To investigate the CT features of PDAC and assess the consistency of CT and pathology or surgery in evaluating tumor size,degree of vascular invasion,presence or absence of regional lymph node metastasis and distant metastasis.Methods:1.This study analyzed the CT characteristics of 82 patients with PDAC who were enrolled in serum metabolomics,including the tumor size,shape,margine,density,enhancement pattern,vessel invasion,lymphnode metastasis,distant metastasis,dilatation of common bile duct and main pancreatic duct,the invasion of adjacent structure.2.This study analyzed the consistency of CT with surgical exploration or histopathology in measuring tumor size and evaluating vascular invasion,presence or absence of regional lymph nodes and distant metastasis.Results:1.PD AC typically presented with solid masses,hypovascular in arterial phase and progressive enhancement in delay phase.Vascular invasion can be seen in 65.9%of patients,regional lymph node metastasis was seen in about 50%of the patients,and distant metastasis was seen in 25.6%of the patients.2.The tumor size measured at CT arterial phase images in this study was in good agreement with the size of the gross specimen,with ICC index of 0.773.In this study,a three-level rating method was used to evaluate vascular invasion.Imaging and surgical exploration were used to evaluate arteries and veins respectively.The weighted kappa consistency test showed that there was strong consistent between the two methods,with kappa values of 0.756 and 0.759 respectively.The kappa test results of imaging evaluation of lymph node metastasis and pathology showed that kappa value was 0.581,which was moderate consistent.The consistency test results of imaging evaluation of distant metastasis and pathology showed that kappa value was 0.871,which showed highly consistent.Conclusion:1.PD AC typically presented as a solid hypovascular tumor,with different digrees of vascular invasion and regional lymphnode metastasis or distant metastasis.2.CT had good performance in evaluating tumor size,vascular invasion and distant metastasis,and had limit value in evaluating regional lymphnode metastasis.Part Ⅲ Correlation study between CT features and serum metabonomics in pancreatic cancerObjective and Significance:This study attempts to combine imaging and metabolomics to explore metabolites characterizing different stages,with or without vascular invasion,lymph node metastasis and distant metastasis of PD AC.It is helful for clinicians to judge the progression of PD AC and to explore the invasive mechanism of PD AC.Methods:2.Combined with CT,surgical exploration and histopathological results of PD AC,82 patients were divided into groups according to the stage(early stage I,II;late stage III and IV),with or without vascular invasion,lymph node metastasis and distant metastasis respectively.Univariate and multivariate statistical methods were used.PCA and OPLS-DA models were constructed,and the standards of p<0.05 and VIP>2.0 were used to preliminarily screen the differential metabolites between groups.PCA was performed to screen metabolites,which was significant for the above groups.ROC curve,sensitivity and specificity were used to test the predictive efficacy of the metabolites.The KEGG map database was used for network and pathway analysis.Results:Combined with imaging,surgical exploration and pathological results,this study applied metabolomics to further distinguish PD AC staging(early or late stage),with or without vascular invasion,lymph node metastasis and distant metastasis and identified the metabolites that can distinguish these groups.Multivariate analysis results showed that L-isoleucine can distinguish early and late stage of PD AC,and it had better performance than CA19-9,with AUC 0.7812,sensitivity 100%,specificity 62.5%.Arachidonic acid can discriminate PD AC with or without vascular invasion,with AUC 0.679,sensitivity 50%,specificity 85.7%.1-streoayl-2-Hydroxy-sn-glycerol-3-phosphocholine can distinguish the PD AC patients with or without lymph node metastasis,with AUC 0.729,specificity 66.7%,sensitivity 87.5%.L-phenylalanine can discriminate PD AC with or without distant metastasis,with AUC 0.745,specificity 63.6%,sensitivity 100%.Conclusion:Combined with CT and metabolomics,metabolites indicating PD AC staging,with or without vascular invasion,lymph node metastasis and distant metastasis were identified in this study.