Early Diagnosis And Drug Resistance Analysis Of Pancreatic Cancer Based On Universally Differential Expressed Genes

Objective:Pancreatic ductal adenocarcinoma(PDAC),which accounts for about90% of pancreatic cancer patients,is an aggressive form of cancer with a high mortality rate.Early diagnosis combined with radical surgical resection is the only treatment option for pancreatic cancer,but most patients are diagnosed at an advanced stage,which does not meet the surgical resection principle.Therefore,the main treatment for pancreatic cancer is chemotherapy.However,at present,the response rates of most patients to chemotherapy are very low,because the treatment is effective in the early stages but gradually loses effectiveness.This study aims to identify common characteristics among patients with pancreatic cancer in order to construct an early diagnosis model of pancreatic cancer and to explain why pancreatic cancer is universally drug resistant.Method: We first calculated the differential genes in each pancreatic cancer sample relative to its paired normal sample.In unpaired samples,Rank Comp was used to screen for differential genes.The dysregulation frequencies of differentially expressed genes were calculated in unpaired and paired samples,respectively.The genes with dysregulation frequencies over 85% in both samples were defined as universally differential expressed genes(UDEGs).To explore the expression of UDEGs at the single-cell level,we analyzed the differences in cell composition and function between pancreatic cancer and normal samples using the single-cell pancreatic cancer dataset CRA16001 from SRA database,and analyzed the specific expression of UDEGs in specific cell types.Finally,based on these UDEGs,we conducted the following analyses:(I)Application of UDEG in the diagnosis of pancreatic cancer.The main difficulty in early diagnosis of pancreatic cancer is the difference between chronic pancreatitis and pancreatic cancer.Besed on normal pancreas and chronic pancreatitis and pancreatic ductal adenocarcinoma samples,the UDEGs are role as a random forest characteristics to construct an early diagnosis model.The training set of the maximum AUC value reaches the optimal model,and the independent validation set are used to calculate the AUC value and F-score to evaluate the performance.(II)Correlation analysis of drug resistance.The drug sensitivity data and gene expression profiles of pancreatic cancer cell lines provided by GDSC database were used to analyze the relationship between UDEGs and drug resistance of different anticancer drugs,and the drug resistance-gene expression correlation was verified by gemcitabine response information of pancreatic ductal adenocarcinoma of TCGA.Finally,the drug-gene interaction relationship of drug-gene interaction database was used to analyze the network,and molecular docking were used to further identify possible interactions and screen the drugs that might be useful to treat pancreatic cancer.(III)Relationship between immune-related genes and immunotherapy.Using TCGA data of samples of pancreatic ductal adenocarcinoma tissue expression profiles,the application of deconvolution algorithm CIBERSORT calculation of 22 kinds of immune cells,and analyze the immune related UDEG expression related with immune treatment response,resistant immune cells proportion of relevance,thus UDEGs and the relationship between immune treatment response of patients with pancreatic cancer.Results: We obtained sixteen up-regulated genes(ANLN,CEACAM5,CORO2 A,ESM1,GALNT5,GPRC5 A,LAMC2,LRP8,OASL,PCDH7,SCEL,SDR16C5,SERPINB5,SFN,SLC6A14 and TMPRSS4)and three down-regulated genes(ACADM,CYB5 A and EPHX1),thirteen of which were confirmed to be cancer-related.These UDEGs and their interacting genes are enriched in the cell cycle,focus adhesion,Erb-βsignaling pathway,and PI3 K signaling pathway.These results suggest that changes in cell proliferation,migration,and the immune system all play important roles in the pathogenesis and progression of pancreatic cancer.To explore the expression pattern of UDEGs in various cell types,we analyzed single-cell data from pancreatic cancer and normal samples and found that the expression patterns of UDEGs in cancer-related ductal cells in cancer samples are same as those found in bulk cancer tissue and show similar function with the bulk of pancreatic cancer tissue,such as cancer and cell proliferation.Based on these UDEGs,the expression profile data of 60 non-cancer samples and 159 cancer samples in the training data were used to construct an early diagnosis model using the random forest algorithm.In the training data,the precision,recall,F-score,and AUC were 93.69%,96.24%,94.94%,and 0.9546(CI: 0.9230-0.9862),respectively.For 242 non-cancer samples and 647 cancer samples in the validation data,precision,recall,F-score,and AUC were 95.01%,95.50%,95.27%,and0.9659(CI: 0.9505-0.9814),respectively.Then,using the drug sensitivity data provided by GDSC,we found that drug resistance was mostly positively correlated with the expression levels of ANLN,GPRC5 A,and SERPINB5,and that high expression of these three genes was correlated with a poor prognosis in patients,indicating that high expression of these genes was closely related to the mechanism of drug resistance in pancreatic cancer.This suggests that targeting these genes may improve drug sensitivity in pancreatic cancer cells and improve patient survival.Through immunoinfiltration analysis,we found that the expression pattern of four immune genes,the high expression of ANLN,LAMC2,and OASL,and the low expression of EPHX1,in pancreatic cancer predicted a low proportion of immune cells related to immunotherapy response and predicted the low response rates to immunotherapy.Conclusion: The UDEGs identified in this study can be applied to improve the detection rate of pancreatic cancer through early diagnosis.The expressions of UDEGs are associated with chemotherapy and immunotherapy resistance in PDAC,suggesting their potential as therapeutic targets in pancreatic cancer and providing important clues for the analysis of drug repurposing in pancreatic cancer.