Construction Of Prognostic Model Of Colorectal Cancer Based On Single Cell RNA Sequencing

Background:A colon cancer prognosis risk prediction model was constructed based on single-cell RNA sequencing technology.and on this basis,the relevant biological characteristics and clinical application value of the model were explored to provide a reference for clinical evaluation of patient prognosis.Methods:In this study,we downloaded single cell transcriptome sequencing(scRNA-seq)data from Gene Expression Omnibus(GEO)database,and the cell trajectories of the data set were analyzed.The differentially expressed colon cancer differentiation-related genes(CDRG)were selected as candidate genes for the predictive model,and the corresponding expression data and clinical information of colon cancer differentiation-related genes were downloaded from The Cancer Genome Atlas(TCGA)and GEO databases.Then the prognostic risk model was constructed by using Least Absolute Shrinkage and Selection Operator(LASSO)regression algorithm and Cox proportional hazard regression model,and the prediction accuracy of the risk model was evaluated by the area under the curve(AUC)of the receiver operating characteristic curve(ROC).In this study,CIBERSORT algorithm was also used to evaluate the tumor microenvironment score and immune cell infiltration level,as well as the difference of immune checkpoint gene(ICG)expression in high/low risk groups of colorectal cancer samples.And the immunophenotypic score of cancer immunotherapy atlas(TCIA)database was used to analyze the potential relationship between risk model and immunotherapy.Drug semi-inhibitory concentration(IC50)was used to evaluate the difference in sensitivity to common chemotherapeutic drugs between high-risk and low-risk groups.Gene cluster enrichment analysis(GSEA)was used to evaluate the potential biological differences between high-risk and low-risk groups.Finally,multivariate Cox regression analysis was used to verify whether the prognostic risk model was an independent prognostic factor for colorectal cancer.Based on the risk score of colorectal cancer patients and other clinical factors in the model,a nomograms was drawn to predict the survival probability of patients.Results:This study finally included the single-cell RNA sequencing data of 12499 colorectal cancer cells from 4 colorectal cancer samples.823 differentially expressed genes(CDRG)in different differentiated tumor cells were obtained by cell trajectory analysis.The results of WGCNA combined with TCGA database showed that the blue module was closely related to the survival time of colon cancer,including 615 CDRG,of which 169 genes were differentially expressed in tumor samples and normal samples,56 genes were up-regulated in tumors and 113 genes were down-regulated in tumors(|log2(FC)|>1,FDR<0.05).Through further LASSO regression algorithm and Cox regression analysis,a prognostic risk model of colorectal cancer was constructed,which was composed of two CDRG(TLN1 and TIMP1).Survival analysis showed that the overall survival time(OS)of colorectal cancer patients in the high-risk group was significantly lower than that in the low-risk group(P=0.002).The AUC values of 1-year,3-year and 5-year survival ROC curves were 0.599,0.619 and 0.677.The model was validated in the GSE94437 data set,and the results showed that the overall survival rate of colorectal cancer patients in the high risk group was significantly shorter(P=0.006).The AUC values of 1-year,3-year and 5-year survival ROC curves were 0.560,0.587 and 0.574.CIBERSORT analysis showed that the stromal scor,immune score and estimate score were higher in the high-score risk group(P<0.05).The number of B cells,T cells,macrophages and NK cells in the types of tumor infiltration is relatively large.There were significant differences in 10 kinds of immune cells between the two groups(P<0.05).The expression of plasma cells,silent CD4+T cells,dendritic cells and neutrophils in the low risk group was significantly higher than that in the high risk group,while the expression of silent NK cells,M0 type macrophages,M1 macrophages,silent mast cells and eosinophils was higher in the high risk group.In addition,through the analysis of immune checkpoint-related gene expression and immunophenotypic score among patients,we found that CD274,CTLA4,HAVCR2,IDO1,LAG3,PDCD1,PDCD1LG2 gene expression was higher in high risk group,and it was more sensitive to immunotherapy.In addition,the IC50 of erlotinib,bleomycin,dasatinib,doxorubicin,gemcitabine,rapamycin,nilotinib,pazopanib,Sunitinib,etoposide,pyrimidine,shikonin and paclitaxel were lower in the high-risk group,while metformin and all-trans retinoic acid were the opposite.Gene enrichment analysis showed that there were significant differences in many biological processes such as extracellular matrix structure,function,metabolic degradation and cell interaction between high and low risk groups.Finally,multivariate Cox regression analysis proved that the risk model could be used as an independent prognostic factor for colorectal cancer(P<0.001).Based on the risk score of the model and the patient’s age and tumor stage,a nomograms was further established.The calibration curve indicates that it has good reliability in predicting the 1-,3-and 5-year overall survival rate of colorectal cancer.Conclusions:In this study,we constructed a prognostic risk model of colorectal cancer based on two CDRG.The risk model can be used as a prognostic factor for colorectal cancer patients independent of other clinicopathological features,and it can also be used as an evaluation tool to screen patients who may benefit from immunotherapy and predict chemotherapeutic drug sensitivity.to provide some reference for individualized treatment.The established nomograms can predict the survival probability of patients with colorectal cancer.