Classification And Prognosis Analysis Of Pancreatic Cancer Based On DNA Methylation Profile And Clinical Information

Pancreatic cancer is a malignant tumor of the digestive tract.It is the seventh leading cause of cancer death worldwide.Its early diagnosis is difficult,the prognosis is poor,and different patients respond differently to treatment,however,there is no widely used molecular typing related to pancreatic cancer prognosis in clinical practice.Studies have shown that the occurrence and development of tumors is accompanied by changes in the epigenome.As an epigenetic regulatory factor,DNA methylation can participate in the regulation of gene activity and cell differentiation.One of the early manifestations of many tumors is the level of tumor suppressor gene methylation.Increased,abnormal DNA methylation changes and specific methylation patterns can be used as biomarkers of cancer.This article obtained clinical information and molecular information of 194 pancreatic cancer patients from the TCGA database,including m RNA expression profile and DNA methylation data,and collected clinical data of 205 389 pancreatic cancer patients from the SEER database,based on pancreatic cancer DNA methylation Analyze data,mine specific molecular markers related to the prognosis of pancreatic cancer,and construct a pancreatic cancer classification prediction model.In this study,we use the DNA methylation profile of patients with pancreatic cancer to obtain 1 235 differentially methylated genes between cancer and paracancerous samples,and construct univariate and multivariate COX proportional hazard regression models to screen 78 independently affecting the prognosis of pancreatic cancer Methylation molecular markers,these genes have the potential to serve as molecular markers of pancreatic cancer and explain the heterogeneity of pancreatic cancer.Based on these genes,two subtypes of pancreatic cancer were identified through consistent clustering.Survival analysis and clinical characteristics test showed that the prognosis of the subtype named cluster2 was significantly better than that of cluster1,and there were significant differences between the two subtypes in gender,clinical tumor stage,T stage,and tumor differentiation.Mining the subtype of each subtype The combination of specific clinical features,verified using large samples from SEER,proves that subtype differentiation based on DNA methylation patterns has unique clinical features and prognosis.Studies combined with transcriptome data show that there are 301 cancer driver genes that are differentially expressed between the two subtypes and the degree of immune cell infiltration differs significantly between the subtypes.In addition,this paper identified 14 specific high/low methylation genes that characterize the unique methylation patterns of the subtypes between the two subtypes,and constructed a Bayesian network-based prognostic prediction model for typing.The model is in the training set The accuracy of the test set is 93.75%,and the AUC value is 0.937.The model is applied to the test set,and the sample is successfully classified into the two methylation subtypes obtained in this study.The prognostic analysis result of the test set is also similar to the training set.In summary,this dissertain uses DNA methylation data combined with bioinformatics algorithms to mine pancreatic cancer DNA methylation molecular markers and identify pancreatic cancer DNA methylation subtypes related to prognosis,and construct a pancreatic cancer classification and prognosis prediction model,Analyzed the heterogeneity of pancreatic tumors from the perspective of epigenetics,and provided new ideas and new targets for personalized treatment plan evaluation and precision medicine for patients with pancreatic cancer.