The Role And Mechanism Of Bile Acid Biomarkers In Differential Diagnoss Of Benign And Malignant Biliary Stricture

Part 1:Metabonomics biomarker screening of benign and malignant biliary strictureBackgroud:Preoperative diagnosis of indeterminate biliary stricture is very challenging.Even if combined with a variety of examination methods,there are still some cases can not determine the possible cause.Differentiating benign from malignant biliary stricture is of great significance for subsequent clinical treatment decision,especially in the case of extrahepatic bile duct with mass.It is necessary to balance the risk of surgical removal of benign stenosis or missed opportunity to remove potentially curable malignancies.It is very important to make corresponding clinical decisions according to the clinical situation of patients and the risk of benign and malignant biliary stricture.Aims:To screen biomarkers those can be used in the clinical differential diagnosis of benign and malignant biliary stricture,and form a combined labeling package to improve the diagnostic efficiency,and validate it.Methods and materials:The study consists of training sets and validation sets.Serum samples were collected from the multiple centers of three groups of people,normal healthy volunteers,benign biliary stricture（common bile duct stone,common bile duct inflammatory stenosis,etc.）and malignant biliary stricture（bile duct carcinoma,pancreatic cancer,carcinoma of the ampulla,hilus lymph node metastases,etc.）,excluding other diseases may affect the metabolism of bile or drug users,with each group 30 cases in the training sets and 20,20 and 27 cases in the validation sets.Bile acids were quantified by high performance liquid chromatography coupled with tandem mass spectrometry（UPLC-MS/MS）system（acquityuplc-xevotq-s,waters Corp.,Milford,Ma,USA）.Strict analytical quality control procedures were followed during the test,and all samples were analyzed within 48 hours.Three types of quality control samples were used,namely test mixture,internal standard of stable isotope labeling and three different levels of quality control（low,medium and high）.Internal standards were added to the test samples to monitor the analytical changes during the whole sample preparation and analysis process.Reagent blank samples are also used as system contamination alarms to wash columns and remove accumulated matrix effects.The calibrator consists of blank samples（matrix samples without internal standard treatment）,zero samples（matrix samples with internal standard treatment）and a series of seven concentrations,covering the expected range of metabolites present in specific biological samples.All samples are added to the Metabo profile LIMS system,and LIMS assigns a unique identifier,which is only associated with the original source identifier to avoid data mismatch.The original data files generated by UPLC-MS/MS were processed by quanmet software（v1.0,Metabo profile,Shanghai,China）.The peak integration,calibration and quantification of each metabolite were performed.Powerful R work package for statistical analysis.The measurement data were presented in the form of "mean（standard deviation）"or "median（interquartile spacing）";the count data were presented in the form of frequency distribution and corresponding percentage;the grade data were presented in the form of frequency distribution and corresponding percentage,as well as median and average rank;the qualitative data were presented in the form of positive rate,positive number and denominator.Spss22.0 software was used for statistical analysis.T test（normal distribution）or Mann Whitney U test（unnormal distribution）was used for measurement data;chi square test or Fisher exact test was used for count data;logistic regression was used for disease risk factor analysis.In the risk factor analysis,if P<0.05 in univariate analysis,the multivariate logistic regression analysis was further used.The significance test level was α=0.05.The diagnostic value of the candidate molecular markers was compared with the area under the curve（AUC）by ROC curve.Random forest method and support vector method were used to analyze the biomarkers detected by the above methods,and the biomarkers with more clinical significance were screened out to form a diagnostic combination of biomarkers.SPSS software was used to construct the combined diagnostic factors based on the top four bile acid subtypes,and logistic regression analysis was performed.Then read the Logsitic regression equation,and calculate a new variable joint factor in SPSS.ROC curve is drawn using the new joint factor.Serum bile acid subtypes of an independent validation cohort was quantitative detected using UPLC-MS/MS.Using the calculation formula of joint diagnostic markers obtained from the training set,calculate the value of the joint factor of the sample in the validation set,infer its population,and calculate the specificity,sensitivity,negative predictive value and positive predictive value of the new joint diagnostic factor.Through clinical follow-up,analyze the relationship between CDCA,bCDCA and TCA and the prognosis of malignant biliary stricture,and explore the possibility of their application in prognosis judgment.Results:There were three groups of normal volunteers,benign biliary stricture and malignant biliary stricture,with 30 cases in each group.The three groups in the validation set were 20 cases,20 cases and 27 cases respectively.1.The cluster analysis of bile acid subtypes in each group showed good clustering characteristics.2D-PLS-DA、3D-PLS-DA,PCA analysis and bile acid subtype thermogram drawn after the absolute value of bile acid subtypes was converted by Z-score normalization method all indicated that the three groups had good clustering characteristics.The Q2 values of opls-da analysis were-0.259,-0.229 and-0.503,which were less than 0,indicating that the model is reliable.2.There were significant differences in the composition of bile acid subtypes among the groups.The volcanic map can clearly show the difference of bile acid composition between the two groups.Based on one-dimensional and two-dimensional statistical analysis,significant differences in bile acids among groups can be effectively distinguished.The box diagram and violin diagram are intuitively displayed.Among them,there are 11 kinds of different bile acids in normal people and benign biliary stricture,which are DCA,BDCA and CDCA₃Gln,isoLCA,LCA,12_Ketolca,ghca,norca,bcdca,THCA and HDCA,P<0.01.There are 19 different bile acids between normal people and malignant biliary stricture₃Gln,bCDCA,DC A,bUDCA,CDCA,HDCA,bDCA,THCA,LCA,NorCA,12_ketoLCA,TCDCA,GCDCA,TBA,7_KetoLCA,GHCA,TCA,HCA and isoLCA,P<0.01.There were 10 kinds of different bile acids in benign and malignant biliary stricture group,which were CDCA₃Gln,HCA,TCA,7_KetoLCA,HDCA,THCA,LCA,isoLCA,bCDCA and CDCA,P<0.01.There were 6 bile acid subtypes in the three groups,which were bCDCA and CDCA₃gln,HDCA,isoLCA,LCA and THCA,P<0.01.3.Differential bile acid subtypes can be used to distinguish benign and malignant biliary strictures.For the different metabolites that meet the multi-dimensional and single dimensional screening criteria,we draw the receiver operating characteristic curve.In the normal people and benign biliary stricture group,the maximum AUC was 0.972,and the minimum was 0.723.By random forest method and support vector machine method,the top four differential bile acids were selected as the joint diagnostic indicators,which were NorCA,bCDCA,DCA and THCA,respectively.The AUC of the new indicators was 0.969.In the normal people and malignant biliary stricture group,the maximum AUC was 0.999 and the minimum was 0.957.TCA,NorCA,bCDCA and bDCA were selected as the combined diagnostic indexes,AUC=1.000.The AUC of biomarkers in benign and malignant biliary stricture was 0.892 and 0.733 respectively.CDCA,bCDCA,HCA and TCA were selected as the combined diagnostic indexes,AUC=0.871.4.Independent validation cohort test showed that the combined diagnostic indicators of bile acid subtypes had good performance in the differential diagnosis of benign and malignant biliary strictureThe multivariate quality control chart showed that the quantitative detection data of bile acids in the validation set were stable and reliable.The mixed detection of validation cohort and quality control samples found that most samples fluctuated up and down around the x-axis within 2 times the standard deviation,and only one sample was close to the control limit.The correlation heat map between quality control（QC）samples shows that the correlation coefficient of each QC sample is between 0.985-1.000,indicating that the data is stable and reliable.The PCA score chart with QC samples shows that QC sample points are close to each other and have a high degree of aggregation,suggesting that the instrument has good stability.Supervised PLS-DA analysis showed that each group in the validation set had good clustering characteristics,which was similar to the training set and comparable.Using the test set data,verify the logistic regression model established by the training set.In the differential diagnosis of benign and malignant biliary stricture,the AUC value was 0.844（95%CI 0.729-0.960）,the sensitivity was 0.888,the specificity was 0.650,the negative predictive value was 0.813 and the positive predictive value was 0.774.In the differentiation between normal population and benign biliary stricture,the AUC value was 0.888（95%CI 0.784-0.981）,the sensitivity was 0.600,the specificity was 0.950,the negative predictive value was 0.703 and the positive predictive value was 0.923.In the differential diagnosis of benign and malignant biliary stricture,the AUC value was 0.993（95%CI 0.978-1.000）,the sensitivity was 0.963,the specificity was 0.950,the negative predictive value was 0.950,and the positive predictive value was 0.963.5.Some bile acid subtypes are related to the prognosis of malignant biliary stricture.We analyzed the relationship between the prognosis of patients with malignant biliary stricture and the three bile acid subtypes of CDCA,bCDCA and TCA.The cut-off value was the optimal cut-off value according to the ROC curve of benign biliary stricture,in which the cut-off value of CDCA was 20.38,the cut-off value of bCDCA was 1.195,and the cut-off value of TCA was 2816.All three bile acid subtypes were significantly correlated with the prognosis of malignant biliary stricture（Fig.47）.The prognosis of patients with CDCA（16.57±3.40 vs 8.04±0.64,P<0.01）and bCDCA（16.37 ± 2.97 vs 7.73 ±0.59,P<0.01）was better than that of patients with TCA（8.38 ± 0.83 vs 13.89± 2.90,P<0.05）.Conclusions:The proportion of bile acid subtypes in patients with benign and malignant biliary stricture was significantly different;Bile acid subtypes such as CDCA,bcdca,HCA and TCA can be used as biomarkers for differential diagnosis.The new biomarker combined by screening has higher diagnostic efficiency,and its ROC-AUC reaches 0.871,which is significantly higher than the traditional biomarker CA19-9.It can be used as one of the means for differential diagnosis of benign and malignant biliary stricture before operation.Abnormal bile acid synthesis and secretion may be the cause of the difference of bile acid metabolism between cholangiocarcinoma and normal people.Some bile acid subtypes,such as CDCA,bCDCA and TCA,may be related to the prognosis of malignant biliary stricture.PART2:The role of bile acid synthesis and secretion signaling pathway in bile acid metabolism changes in cholangiocarcinomaBackgrouds:Cholangiocarcinoma is a common type of malignant biliary stricture.According to the results of the first part,it is suggested that bile acid subtypes can be used as biomarkers of benign and malignant biliary stricture,and the combination of multiple subtypes as a combined diagnostic index has higher diagnostic efficiency.Previous studies have shown that the changes of bile acid metabolism play an important role in cholangiocarcinoma.The main pathways of bile acid metabolism include bile acid synthesis and bile acid secretion,but its specific mechanism is unknown.Aims:To explore the role of primary bile acid synthesis and bile secretion signaling pathway in bile acid metabolism in cholangiocarcinoma.Methods and materials:1.Compare the composition of serum bile acid subtypes between cholangiocarcinoma and non cholangiocarcinoma volunteers.The sera of 20 cholangiocarcinoma and 20 non cholangiocarcinoma volunteers（benign biliary diseases and normal volunteers）were collected.The bile acids were analyzed quantitatively by high performance liquid chromatography coupled with tandem mass spectrometry（UPLC-MS/MS）system（acquityuplc-xevotq-s,waters Corp.,Milford,Ma,USA）,and the differences between the two groups were compared.2.Screening metabolic pathway changes by bioinformatics methodsTCGA database was used to analyze the reasons for the difference of bile acid between cholangiocarcinoma and normal people.The differential expression of mRNA was studied using the Limma software package of R software（version:3.40.2）.Adjusted P values were analyzed in TCGA to correct false-positive results " Calibration P<0.05 and |log2（fold change）|>1 "were defined as screening for threshold mRNA differential expression.The cluster profiler package in R was used to analyze the go function of potential mRNA and enrich KEGG pathway,so as to explore the signal pathway involved in different genes between cholangiocarcinoma and normal population.3.Establish the epithelial mesenchymal transformation model of human extrahepatic bile duct cells in vitro and verify its tumor biological behavior.Gradient concentrations of 10ng/ml（B）,20ng/ml（C）and 40ng/ml（D）TGF-β1 was used to stimulated the human primary extrahepatic bile duct epithelial cell line to establish an epithelial mesenchymal transformation（EMT）cell model.Compared with the normal control group（A）,the morphological changes of cells in different groups were observed;the ability of cell transfer and migration after EMT was verified by transwell test and scratch test;EMT biomarkers were detected by fluorescence quantitative detection;and the changes of cell cycle were detected by flow cytometry.4.Metabonomics and transcriptomics were used to verify whether the changes of bile acid metabolism and bile acid synthesis signal pathway were involved in the changes of bile acid metabolism in cholangiocarcinoma.Human extrahepatic bile duct epithelial cells transformed by EMT in each group were detected by LC-MS/MS for metabolomics and Ilumina sequencing platform for transcriptomics.The metabolic products and signal pathways in each group were compared and analyzed to verify the role of bile acid subtypes in the differential diagnosis of cholangiocarcinoma,and the role of abnormal bile acid synthesis and secretion signal pathways in the metabolic changes of bile acid subtypes of cholangiocarcinoma.Results:1.Some bile acid subtypes were significantly different between patients with cholangiocarcinoma and non tumor volunteers.Between the cholangiocarcinoma and non tumor volunteers,15 bile acid subtypes were significantly different,including TBA,THCA,TCA,NorCA and CDCA₃Gln expression decreased significantly,isoLCA,LCA,7_ketoLCA,CDCA,DCA,bUDCA,bCDCA,bDCA,HDCA and HCA increased significantly（P<0.05）.2.Biochemical analysis suggest that the abnormal bile acid synthesis and bile secretion might be the reason for the difference of bile acid metabolism between cholangiocarcinoma and normal population.There is differential gene expression between cholangiocarcinoma and adjacent tissues.TCGA gene expression was obtained from 36 cases of cholangiocarcinoma and 9 cases of adjacent tissues in January 2020.When |LogFC|≥1,P<0.05,adjust P<0.01,9887 differential genes met the standard.Volcano map and heat map show good gene clustering characteristics.The enrichment analysis of differential genes showed that there were 20 KEGG pathways,including primary bile acid biosynthesis and bile secretion,both of which were down regulated.Bile metabolic signaling pathways are different between cholangiocarcinoma and adjacent tissues.Enrichment analysis showed that the primary bile acid biosynthesis was down regulated.12 genes in bile acid biosynthesis pathway were down regulated,including ACOX2,AKR1C4,AK1D1,BAAT,CYP7A1,CYP8B1,CYP27A1,CYP39A1,HSD17B4,SCP2 and SLC27A5.Only HSD3B7 and ACOT8 were up-regulated.The expression of bile secretion signal pathway in cholangiocarcinoma was down regulated as a whole,but the bile reabsorption increased significantly.The expression of 11 genes in bile secretion signal pathway was up-regulated,including GNAS,SLC4A2,ABCC3,ADCY6,SLC9A1,ABCC4,CFTR,SLC2A1,AQP1,SCTR and SLC10A2,which were mainly involved in the secretion of HCP3-,water and the reabsorption of bile acids,glucose and other substances in bile;The expression of 12 genes was down regulated,including SLC10A1,SLC22A7,SLCO1B1,ABCG8,SLCO1B3,ABCB4,ABCG5,ABCB11,KCNN2,ABCC2,EPHX1 and ABCG2,which were mainly involved in the bile secretion and the bile acid reabsorption of hepatocytes in hepatointestinal circulation.3.After TGF-β1 stimulation,human bile duct epithelial cells can undergo epithelial mesenchymal transformation and have stable tumor biological behavior.After TGF-β1 stimulation,the morphological atypia of human extrahepatic bile duct epithelial cells increased under microscope,the cells were loose,and some cells grew in suspension.In Transwell experiment,the migration number of human extrahepatic bile duct epithelial cells increased significantly with the increase of stimulation concentration.The number of migration under 200 fold visual fields of group A,B,C and D were 38.3±8.4,40.3 ± 1.2,54.0±1.0 and 62.3±3.2,P<0.05,respectively.In the cell scratch experiment,the migration rate of human extrahepatic bile duct epithelial cells increased significantly with the increase of TGF-β1 concentration.The scratch width ratios of group A,B,C and D at 24h and 0h were 0.7722,0.7483,0.6861 and 0.4974 respectively,and the scratch width ratios at 48h and 0h were 0.4896,0.4798,0.4157 and 0.2562 respectively.Compared with the blank control group,there was no significant difference to group B,while great differences to the group C and D（P<0.05）.Relative fluorescence quantitative detection showed that,Fibronectin and a-SMA was significantly higher in TGF-β1 groups compared to the control group with no dose dependense,while E-cadherin decreased significantly（P<0.05）.Although the change of vimentin increased slightly,it was not statistically significant（P>0.05）.Similar results can also be observed in the Western blot experiment.Flow cytometry showed that the proportion of apoptotic cells in each group had no significant change（P>0.05）.4.In epithelial mesenchymal transformed bile duct epithelial cells,bile acid metabolism and bile acid synthesis signal pathway changed synchronously.Metabolic difference map and hierarchical cluster analysis thermodynamic map showed that there were significant differences in metabolites among groups and had good clustering characteristics.Comparing the data of bile acid subtypes detected by metabolomics in each group,it was found that TCA and THCA decreased significantly,and 7-ketolca,DCA and LCA increased significantly,which was consistent with the information of clinical samples,indicating that EMT transformed human extrahepatic bile duct epithelial cells have bile acid metabolism characteristics similar to cholangiocarcinoma.KEGG pathway differential abundance analysis showed that bile secretion signal pathway was up-regulated after EMT transformation of bile duct epithelial cells,but this up-regulation did not become stronger with the increasing concentration of TGF-β1.This is consistent with the overall down-regulation of the bile secret signal pathway obtained through the previous bioscience analysis,but the increase of inorganic salt secretion and the enhancement of bile acid reabsorption at the level of bile duct epithelial cells.Enrichment analysis of metabolic pathways of differential metabolites showed that the primary bile acid synthesis pathway changed after a low concentration of TGF-β1 stimulation,but the signal pathway did not show greater difference with the increase of concentration.The detected gene expression of each group tends to be saturated and does not increase with the increase of data,indicating that the detection results are reliable.The differentially expressed genes were screened by DEseq.The screening conditions of differentially expressed genes were as follows:the expression difference multiple |log2 FC |>1,with significance P<0.05.The results showed that compared with the control group,after different concentrations of TGF-β1 stimulation,the gene expression of human extrahepatic bile duct epithelial cells changed significantly,but the differential expression among groups was relatively small.Volcano map and hierarchical cluster analysis showed that the cells in each group had significant differences and had good clustering characteristics.The Read count of each group of samples was standardized by FPKM method.According to the KEGG signal pathway map,the differential genes in the primary bile acid synthesis and bile secretion signal pathway in each group were analyzed and compared.In the primary bile acid synthesis signaling pathway,the expression of ACOX2 and HSD17B4 were down-regulated.In the bile secretion signal pathway,the expressions of ABCC2,ABCG2 and EPHX1 were down-regulated,and the expressions of ADCY3,SLC2A1,GNAS and AQP1 were up-regulated,which was similar to the results of previous bioinformatics analysis.Conclusion:There were significant differences in bile acid metabolism between cholangiocarcinoma and non cholangiocarcinoma volunteers.The signal pathways of primary bile acid synthesis and bile acid secretion were significantly down regulated in cholangiocarcinoma compared to the adjacent tissues.After TGF-β1 stimulation,human extrahepatic bile duct cells can undergo stable epithelial mesenchymal transformation.Bile acid metabolism,primary bile acid synthesis and bile acid secretion signal pathway of bile duct epithelial cells after EMT were similar to those of clinical specimens.The abnormal signal pathway of primary bile acid synthesis and bile acid secretion may be the reason for the changes of bile acid metabolism in cholangiocarcinoma,which can provide direction for the study of pathogenesis of cholangiocarcinoma and drug therapy targeting bile acid metabolism in the future.