The Screening And Analysis Of The Protein Signatures Associated With Signal Pathway In Lymph Node Metastasis Of Gastric Cancer

Objection: Gastric cancer is the fourth most common malignancy and the secondleading cause of cancer-related death, after lung cancer, in the world. Most gastric cancerpatients are diagnosed at stage III or IV, with most of patients presenting with local lymphnode metastasis. The preoperative determination of lymph node status is critical in tumorstaging and in planning optimal management of gastric cancer patients. Currently,preoperative assessment of lymph node status is mainly based on imaging studies. However,many studies show that lymph node size determined by imaging techniques is not a reliableindicator of lymph node metastasis. Therefore, preoperative markers that can reliably predictlymph node metastasis in gastric cancer patients are urgently needed. The propose of thisstudy is to screen diagnostic and prognostic proteins for gastric cancers and to find theirmolecular biological mechanisms, and furthermore to identify the signal transductionpathways in which these proteins involved and to define a risk model to predict lymph nodemetastasis.Methods: The Protein Pathway Array (PPA) were used to assess the190protein andphosphoproteins expression in a total of130freshly frozen gastric cancer samples in the testcohort. The differentially expressed proteins were identified between samples with lymphnode metastasis (n=50) and those without lymph node metastasis (n=40) in the training set,inorder to identify a robust set of differentially expressed proteins for classification, a K-foldcross validation (K=10) using support vector machine (SVM) was performed in the trainingset. This study also tested these differentially expressed proteins using clustering analysis andperformed proteomic-wide expression analysis using Ingenuity Pathway Analysis (IPA)software to investigate the significant proteins’ expression in lymph node metastasis signalingnetwork of gastric cancer. To identify the proteins associated with other importantclinicopathologic characteristics, such as tumor size, histologic differentiation and vascular/ lymphatic invasion in addition to lymph node metastasis, SAM analysis was performed on thegastric cancer samples with lymph node metastasis (n=73) and those without lymph nodemetastasis (n=57) in the test cohort. For predicting the lymph node metastasis risk in gastricancer patients, we established a lymph node metastasis model for gastric cancer usingdifferentially expressed proteins and clinicopathologic characteristics, calculated the rate ofcorrect classification, sensitivity and specificity of the lymph node metastasis modelpredicting the lymph node metastasis, and furthermore, the receiver operating characteristic(ROC) curve was used to determine the optimal model to predict lymph node metastasis. Weused the lymph node metastasis model to predict the overall survival of130cases to find riskfactors affecting survival. To further confirm the ability of the risk score model to predictlymph node metastasis and survival, an independent validation cohort of gastric cancerpatients was analyzed using Western Blot technology.Results: In our study we found that27proteins were differentially expressed betweenlymph node metastasis positive gastric cancers and lymph node metastasis negtive gastriccancers, between these27proteins,12were selected as the best classifiers by SVM and weresupposed as a robust set of proteins for classification. The rate of correct classification was86%and the sensitivity and specificity were88%and82.5%, respectively. Lmph nodemetastasis positive gastric caners and lmph node metastasis negtive gastric caners could beseparated into two categories by hierarchical clustering analysis and the12proteins could alsobe separated into the upregulated proteins and downregulated proteins. Based on thedifferentially expressed proteins, IPA system proposed the lymph node metastasis signalingtransduction network of gastric cancer. IPA systems analysis results showed that theseproteins were closely related with cellular movement, cell death and survival, cellulardevelopment, cellular growth and proliferation and gene expression in addition to cancer,respiratory system disease, reproductive system disease, gastrointestinal disease andendocrine system disorders. The upstream regulative proteins may be the therapy target oflymph node metastasis in gastric cancer. The differentially expressed proteins existed indifferent tumor size, histologic differentiation and vascular/lymphatic invasion and could bethe developmental and prognostic molecular markers. Our study also showed that five proteins (Factor XIII B, TFIIH p89, ADAM8, COX-2and CUL-1) and vascular/lymphaticinvasion were independent risk factors of lymph node metastasis of gastric cancer. Combinedthe five proteins and vascular/lymphatic invasion, we established a lymph node metastasismodel of gastric cancer, the rate of correct classification, sensitivity and specificity to predictthe lymph node metastasis were84.6%、96.3%and91.2%, respectively. The model could alsopredict the prognosis of gastric cancer patients. The five proteins were evaluated by WesternBlot technology in an independent validation cohort of gastric cancer patients to validate theability of lymph node metastasis model, the results were consistent with the results in testcohort.Conclusion: Our study focus on the lymph node metastasis signaling pathway in gastriccancer, we did PPA analysis and protein screening with lymph node metastasis positive gastriccancers and lymph node metastasis negtive gastric cancers and to study the function of proteinsignal transduction pathway in lymph node metastasis of gastric cancer in tumor cellsdevelopment, movement, proliferation/growth and apoptosis. Our data demonstrated a broaddysregulation of signaling proteins in the lymph node metastasis process of gastric cancer,Some of the dysregulation of signaling proteins were correlated with tumor size, histologicdifferentiation and vascular/lymphatic invasion, these proteins could be the developmentaland prognostic molecular markers. Specially, the lymph node metastasis model of gastriccancer based on Factor XIII B, TFIIH p89, ADAM8, COX-2, CUL-1and vascular/lymphaticinvasion established an effective detection system for predicting the metastasis status andprognosis of gastric cancer patients.