Association Between Plasma Exosomal Protein And Lymph Node Metastasis Of Esophageal Squamous Cell Carcinoma

Purpose The purpose of the current study was to assess the association between plasma exosomal proteins and lymph node metastasis of esophageal squamous cell carcinoma(ESCC)and explore the mechanism of plasma exosomal proteins in ESCC lymph node metastasis.Methods1.Association between lifestyle behavior and lymph node metastasis in ESCC: ESCC patients were divided into lymph node metastasis group and non metastasis group according to lymph nodes metastasized status,logistic regression was undertaken to identify the associations between lifestyle behavior and ESCC lymph node metastasis.2.Association between plasma exosomal protein and lymph node metastasis of ESCC:Plasma exosomal protein were extracted from ESCC with lymph node metastasis in earlystage,gender,and age-matched patients with ESCC with non-lymph node metastasis in late-stage,and normal control individuals.Expression profiling data of plasma exosomal protein were identified by proteomics.Besides,we validated the association of plasma exosomal proteins with lymph node metastasis in larger ESCC samples.3.The mechanism of exosomal protein promoting lymph node metastasis of ESCC was explored in vitro experiments: The CORO1 C,LIMS1,SULT1A3 overexpression,NC and knockdown expression vectors were constructed and transfected into ESCC cells,and exosomes were isolated from the culture supernatant by ultrafiltration.The extracted exosomes were cocultured with Het-1A cell,and the effects of exosomal proteins on the phenotype of Het-1A cells were investigated using transwell migration assay,invasion assay,and CCK8 proliferation assay.The interactions proteins were screened for target proteins by co-immunoprecipitation,and the underlying mechanism of target proteins in ESCC lymph node metastasis was evaluated.4.The predictive model for lymph node metastasis was established in ESCC: Logistic regression analysis was applied to integrate clinical factors,radiomic features,and plasma exosomal transcriptomic to predict the lymph node metastasis in ESCC based on plasma exosomal proteomics.In order to reduce model parameters,filter the best predictive model,based on the above analysis,lasso regression,random forest and weighted quantile regression(WQS)were used to screen plasma exosomal proteomics,clinical factors,radiomic features,and plasma exosomal transcriptomic to construct a prediction model for lymph node metastasis in ESCC.The receiver operating characteristic curve was generated to confirm models’ diagnostic value,and the area under the model curve was compared by the Delong method.Results1.Association between lifestyle behavior and lymph node metastasis in ESCC: Compared with ESCC non-lymph node metastasis,hot tea drinking(OR =1.72,95% CI: 1.19～2.48),and eating moldy food(OR =2.19,95% CI: 1.32～3.65)were associated with increased the risk of ESCC lymph node metastasis.However,eating vegetable was associated with reduced risk of ESCC lymph node metastasis significantly(OR =0.59,95% CI:0.41～0.87).Besides,eating moldy food was associated with increased the risk of earlystage ESCC lymph node metastasis(OR =2.87,95% CI: 1.26～6.56).2.Association between plasma exosomal protein and lymph node metastasis of ESCC:Three plasma exosomal protein,CORO1 C,LIMS1,and SULT1A3 were selected after venn diagram analysis of the three groups.In a larger sample size,three plasma exosomal protein expression levels of ESCC with lymph node metastasis were significantly increased compared with ESCC with no lymph metastasis(P <0.05).High expression of CORO1 C,LIMS1,and SULT1A3 protein in plasma exosomes increases the risk of ESCC with lymph node metastasis was found by using multivariate logistic regression analysis,the corresponding ORs(95% CI)were 2.99(1.22,7.35),4.09(1.45,11.52),and 5.70(2.24,14.51),respectively.3.The mechanism of exosomal protein promoting ESCC lymph node metastasis was explored in vitro experiments: Overexpressed expression of exosomal CORO1 C,LIMS1,and SULT1A3 proteins could promote migration and invasion of Het-1A cells.Results of co-immunoprecipitation assays indicated that exosomal CORO1 C protein interacts with ZFP91 protein and ACIN1 protein to promote the proliferation of esophageal epithelial cells by regulating the apoptosis pathway,changing the microenvironment of lymph nodes.Exosomal LIMS1 protein interacts with SUMO4 protein to change the microenvironment of esophageal epithelial cells by regulating the NFKB2 pathway,which is beneficial to ESCC lymph node metastasis.Exosomal SULT1A3 protein interacts with S100A10 protein.S100A10 is related to the ECM pathway and can change the microenvironment of esophageal epithelial cells through the expression of MMP9 and promote lymph node metastasis.4.The predictive model for lymph node metastasis was established in ESCC: Plasma exosomal protein expression could be applied as a parameter to discriminate between ESCC non-lymph node metastatic and lymph node metastatic with an area under the curve(AUC)of 0.854(95% CI: 0.787,0.921),with 72.4% sensitivity at 83.9% specificity.Using plasma exosomal protein expression combined with clinical factors to establish ESCC lymph node metastasis prediction model with 0.891(95% CI:0.830,0.953),with 77.8%sensitivity at 91.8% specificity.AUC=0.9853(0.965,1.000),with 77.8% sensitivity at91.8% specificity(based on manual segmentation to extract radiomic features),AUC=0.941(0.902,0.981),with 94.2% sensitivity at 88.3% specificity(based on convolutional neural network to extract radiomic features)of ESCC lymph node metastasis prediction model constructed using plasma exosomal protein combined with clinical factors and radiomics features.Plasma exosomal protein combined with clinical factors,radiomic features,and plasma exosomal transcriptomics were used to establish ESCC lymph node metastasis prediction model with AUC=1.000(1.000,1.000),with100.0% sensitivity at 100.0% specificity(based on manual segmentation to extract radiomic features),AUC=0.980(0.956,1.000),with 92.7% sensitivity at 95.2% specificity(based on convolutional neural network(CNN)to extract radiomic features).Plasma exosomal proteomics combined with multi-omics features to construct a prediction model for ESCC lymph node metastasis has better predictive performance than the prediction model established using only plasma exosomal proteins(Delong method,all P <0.05).These models were validated in the training and validation set,and the constructed ESCC lymph node metastasis model had stable results in the training and validation set.Furthermore,in order to reduce model parameters,filter the best predictive model,based on the above analysis,using WQS to screen the above omics features,the variables included after WQS screening are less than the model that each omics feature is included one by one,but the AUC=1.000(1.000,1.000),with 100.0% sensitivity at 100.0%specificity of the ESCC lymph node metastasis prediction model was established.Conclusions1.Hot tea drinking,and eating moldy food were associated with increased the risk of ESCC lymph node metastasis.However,eating vegetable significantly was associated with reduced the risk of ESCC lymph node metastasis,2.Plasma exosomal CORO1 C,LIMS1,and SULT1A3 protein were associated with ESCC lymph node metastasis and could be used to identify ESCC lymph node metastasis.3.Exosomal CORO1 C,LIMS1,SULT1A3 protein could promote ESCC lymph node metastasis by promoting the migration and invasion of esophageal epithelial cells.Exosomal CORO1 C,LIMS1,SULT1A3 protein could regulate the apoptosis pathway,and ECM pathway,changing the microenvironment of esophageal epithelial cells and promoting lymph node metastasis.4.The sensitivity,specificity and AUC were high of model for ESCC lymph node metastasis established by using plasma exosomal proteomics combined with clinical factors,radiomic features,and plasma exosomal transcriptomics,which could be used to identify ESCC lymph node metastasis.