| Obejectives:Pancreatic cancer is a highly malignant tumor with insidious early symptoms,difficult diagnosis,easy metastasis and poor prognosis.Ferroptosis is a novel form of programmed cell death which occurs in highly metastatic tumor cells to affect tumor progression.Given the important role of ferroptosis in the progression of highly metastatic tumors,this study aimed to investigate the prognostic value of ferroptosis-related genes and to predict the mechanisms associated with ferroptosis in the progression of pancreatic cancer.Methods:1.Using the dataset of pancreatic cancer from The Cancer Genome Atlas(TCGA),the training and validation sets were randomly selected 1:1.And the 60 reported ferroptosis-related genes were combined with Cox regression analysis and LASSO regression analysis,to screen prognosis-related genes and construct a risk regression model of pancreatic cancer.The risk values of each sample were calculated and z-scored to classify the total samples of pancreatic cancer into high and low risk groups with a cut-off value of "0".2.The time-dependent receiver operating characteristic curve(ROC)was used to assess the predictiveness of the model for 1-,2-and 3-year survival of pancreatic cancer,and survival curves were plotted for the high-and low-risk groups.3.The same risk regression coefficient calculation formula was adopted to verify the model robustness using the validation set binned by TCGA,TCGA-PAAD and the Gene Expression Omnibus(GEO)database,respectively,and to compare the predictive ability of the same type of models.4.Analyzed the relationship between risk score and the prognosis of pancreatic cancer patients using clinical data to draw nomogram.5.Analyzed the differential genes,functional enrichment and differences in immune characteristics between high and low risk subgroups using the R package.Results:1.Model construction:Regression analysis showed that the expression of 6 genes(CD44,MT1G,PTGS2,SAT1,TFRC,STEAP3)was significantly and positively correlated with prognosis of pancreatic cancer(P<0.05).The predictiveness of the 6-gene model in the training set reached 0.66,0.78 and 0.77 at 1,2 and 3 years,respectively,and the validation set demonstrated that the model had high robustness and outperformed existing reported models.The model could effective classify patients in high-and low-risk groups,and high-risk patients were positively associated with poor prognosis of pancreatic cancer(P<0.01).There are significant differences in the risk scores of pancreatic cancer with different lymph node metastasis,Stage and Grade(P<0.05).Risk Type was an independent risk factor for prognosis of pancreatic cancer(HR=1.52,95%CI=1.1-2.11,P<0.05).2.Mechanisms explored:There were 1329 differentially expressed genes between high-and low-risk groups,including 1287 up-regulated genes and 42 down-regulated genes.The up-regulated genes in the high-risk group were mainly enriched in extracellular matrix(ECM)receptor-related pathways.High risk subtypes of pancreatic cancer were enriched in mismatch repair,cell cycle and other tumor-related pathways.Immune cell infiltration scores showed that neutrophil infiltration was significantly increased,and B,T,NK cells and other immune checkpoint genes were suppressed in the high-risk group(P<0.05).Immune checkpoint genes such as CD274,CD276,CD44,CD80,IDO1 and PDCD1LG2 were suppressed in the high-risk group.The expression of marker genes of the ECM receptor pathway was positively correlated with the risk score(r>0.5)Conclusions:1.The prognostic model of pancreatic cancer constructed in this study has good stability and validity,and is superior to the existing reported models.2.The effect of ferroptosis on pancreatic cancer may depend on ECM receptor-related pathways and the interaction between various immune cell subsets in the tumor microenvironment.3.Ferroptosis inducers in combination with immune checkpoint inhibitors are expected to be an effective treatment strategy to improve the survival of pancreatic cancer patients. |
PDF Full Text Request