Study On Metabolic Markers Of Disease Progression In High Risk Population Of Esophageal Squamous Cell Carcinoma Based On Case Cohort Design

Background:Esophageal cancer is one of the most common malignant tumors in the world,mainly esophageal squamous cell carcinoma in China.accounting for more than 90%of esophageal cancer.The disease has no obvious symptoms in the early stage,it is mostly in the middle and late stages when it is discovered,with a poor prognosis and low 5-year survival rate.Therefore."early detection,early diagnosis,early treatment" is the key content of esophageal cancer prevention and treatment of esophageal cancer.In this study,our study population came from the esophageal cancer population screening and follow-up community cohort at the Feicheng City Demonstration Base for Early Diagnosis and Treatment of Esophageal Cancer.At present,although the rate of carcinogenesis is known,the process of metabolites in the body is not clear in the process of progression from precancerous lesions to cancer.In addition,metabolomics studies of esophageal cancer mostly identify the metabolites between cancer and healthy populations on the basis of cross-sectional study,and cannot predict the pathological progression of high-level populations.In view of the above scientific questions,this study intends to use the Ultra-High Performance Liquid Chromatography-Mass Spectrometry(UPLC-MS)platform based on the prospective case cohort study to perform non-targeted metabolomic measurements for patients with ESCC precancerous lesions progressing to esophageal carcinoma in situ,ESCC or gastric cancer,and screen for differential metabolites related to pathological progression,study related metabolic pathways,use differential metabolites to predict carcinogenesis prospectively,explore the changes of metabolites in the process of progression from precancerous lesions to cancer,and verify the screening effect of metabolites in the screening cohort.Methods:Based on the Feicheng City Demonstration Base for Early Diagnosis and Treatment of Esophageal Cancer,patients with esophagitis and dysplasia were used as the follow-up population,and patients who progressed to esophageal carcinoma in situ,ESCC or gastric cancer during the follow-up process were used as the case group,and the study population was randomly selected as a sub-cohort according to 1:1 in the total cohort,and the case cohort study was constructed,UPLC-MS non-targeted metabolomics measurements were performed,and the metabolic profile data were preprocessed.Quality control of metabolomics data by Relative Standard Deviation(RSD).Principal Component Analysis(PCA)and Partial Least Squares Discrimination Analysis(PLS-DA)were used to explore whether there was a classification trend among different groups of subiects,and combine the Wilcoxon rank sum test to screen differential metabolites.A random forest prediction model based on differential metabolites was constructed,and the prediction model was evaluated by using the leave one out cross validation.Use the Metaboanalyst platform for pathway and enrichment analysis.To further verify the screening effect of metabolites in the screening cohort,differential metabolites were first screened based on the pathological progression cohort(P<0.05).Then,the screening cohort was divided into training set and validation set according to the 7:3 ratio,and metabolites with the same retention time and mass-to-charge ratio as differential metabolites were selected in the screening cohort training set,and finally differential metabolites were screened according to the Least Absolute Shrinkage and Selection Operator(Lasso).In the training set of the screening cohort,a univariate logistic regression model was constructed for lifestyle risk factors,and significant variables were screened out as lifestyle risk factors.A random forest prediction model was constructed and evaluated in the screening cohort validation set.Results:A total of 77 case groups and 77 cohort populations were included for UPLC-MS non-targeted metabolomics detection,and the RSD results showed good quality control,good reproducibility of metabolomics experiments,and reliable data quality.The PCA and PLS-DA score plots showed that the metabolic profile separation trend was obvious.Nine differential metabolites were screened according to VIP>1 and FDR<0.05 criteria,including caprylic acid,valproic acid,pelargonic acid,2-(2-butoxyethoxy)ethanol,undecanoic acid,diethyl toluamide,dodecanoic acid,pentylethylene glycol,and hexaethylene glycol.Metabolic pathways include steroid hormone biosynthesis,histidine metabolism,primary bile acid biosynthesis,fatty acid biosynthesis,and glutathione metabolism.In the prediction model based on nine differential metabolites,the AUC value of the prediction model constructed by the case cohort study was 0.90(95%CI=0.85～0.95).Further verification of metabolite screening effect in the screening cohort,eight differential metabolites were screened,including pelargonic acid,monoethylglycine xylidine,cis-9-paImitoIeic acid,cortisol,LPC(0-16:0),LPC(17:0/0:0),LPC(18:3),and PC(18:0).In the training set of the screening cohort,a univariate logistic regression model was constructed for lifestyle risk factors,and statistical differences in water source,smoking and alcohol consumption were found,and a random forest model was constructed based on metabolites and lifestyle risk factors.The results showed that when there were only lifestyle risk factors,the AUC value was 0.82(95%CI=0.74～0.89),and the AUC value in the prediction model constructed by combining lifestyle and metabolites was 0.93(95%CI=0.88～0.97),which could further increase the predictive effect of the model after increasing metabolites compared with lifestyle alone.Conclusions:In this study,a total of 9 differential metabolic markers of pathological progression were found(caprylic acid,valproic acid,pelargonic acid,2(2-butoxyethoxy)ethanol,undecanoic acid,diethyl toluamide,dodecanoic acid,pentylethylene glycol,and hexaethylene glycol).The bile acid biosynthesis,steroidgenesis,fatty acid biosynthesis,and beta oxidation of very long chain fatty acids metabolic pathways were different in esophageal cancer.The prediction model based on 9 differential metabolic markers has a good prediction effect on the patients with precancerous lesions progressing to cancer.A total of 25 differential metabolites were found in the pathologic progression cohort and screening cohort,and the prediction model constructed by Lasso screening of differential metabolites showed good prediction effect.The results of this study provide a good direction and basic experimental basis for the clinical practice of using the changes of metabolites in the process of canceration to predict canceration.