The process is shown in Fig. 1. We searched the TCGA database for 385 cases of COAD and downloaded them. The clinical data and gene expression data in the download data were integrated, and the Limme package of R language was used to extract differentially expressed genes (DEGs). Using the immune gene names provided by IMMPORT, we can easily screen out the differential gene immune genes from DEGs. In the same way, we download the names of transcription factors from the CISTROME database and screen out the differentially expressed transcription factors from DEGs (screening conditions: LogFC༞| 2 |, FDR < 0.05). We then conduct further analysis of differentially expressed immune genes (screening conditions: LogFC༞| 2 |, FDR < 0.05) and differentially expressed transcription factors(screening conditions: LogFC༞| 1 |, FDR < 0.05). The differentially expressed immune genes were analyzed by univariate Cox regression using the survival package of R language and clinical data to obtain immune genes related to prognosis. Prognostic related immune genes are as follows: SLC10A2, CXCL3, NOX4, CCL19, IGHG1, IGHV5-51, IGKV1-33, INHBA, STC1, UCN, VIP and OXTR (Fig. 3A). We performed an interaction network analysis of prognostic-associated immune genes and differentially expressed transcription factors, and the results are shown in (Fig. 3B).
We excluded samples with a survival time of less than 90 days from the downloaded clinical data, and assessed the survival prognosis by the level of riskScore. The results showed that patients with high risk scores had significantly worse survival prognosis than patients with low risk scores, P = 8.876e-04 (Fig. 4A). The ROC curve showed that AUC = 0.749, and the riskScore and prognosis model were more reliable (Fig. 4B). This allows us to group patients according to risk scores in clinical work to predict their prognosis. According to Fig. 5A and Fig. 5B, we can find that as the riskScore increases, the survival time of the patient decreases. The heat map of Fig. 5C also shows that the genes that construct risk scores have higher expression levels in the high-risk score array. We included clinical data on COAD and the riskScore evaluated in this study into the Cox regression analysis. The results showed that stage, T, M, N, and riskScore were statistically significant and clinically significant in the survival prognosis of the patients in the univariate Cox regression analysis. However, in the results of multivariate Cox regression analysis, age, stage, T, and riskScore have statistical significance and clinical significance. Based on the analysis of the seven genes and clinical data used to construct the riskScore model, the results show that the expression of CXCL3 gene in M, N, and stage is higher than that in late stage . This is likely to be related to the mechanism of the immune microenvironment. Studies on the immune microenvironment have shown that some genes that construct the immune microenvironment can promote tumor progression (Fig. 7). Some genes that make up the tumor microenvironment, such as B, CD4, CD8, Dendritic, Macrophage, and Neutrophil cells, have been shown in research to be correlated with the survival prognosis of many types of tumor patients . We downloaded the data of these cells through the TIMER database and performed correlation analysis with the riskScore model we constructed. The results showed that the expression of CD4, CD8, Dendritic, Macrophage, and Neutrophil cells increased with the increase of the riskScore. This also confirms on the side that the risk score model we constructed has a certain predictive ability for the clinical prognosis of patients.
The formation of the tumor microenvironment is closely related to the occurrence and development of tumors . By studying the cells that constitute the tumor microenvironment, we can effectively find many cells or genes that are closely related to the clinical prognosis of patients. In order to evaluate the difference in immune cell content between samples with high RiskScore and samples with low RiskScore, we further evaluated them in the EPIC database. We selected 20 sets of samples, which were selected from the 10 samples with the highest RiskScore and the 10 samples with the lowest RiskScore, which passed the EPIC database . Calculate the difference in the amount of their direct immune cells. The results showed that in 10 samples with high RiskScore, the content of CAFs cells was significantly higher than that with 10 samples with low RiskScore. From this we can know that CAFs cells play an important role in RiskScore. The related literature reports that CAFs cells can promote tumor deterioration in multiple tumors. We judge that the poor prognosis of patients is related to the excessive expression of CAFs cells .
In this study, we constructed an immune gene risk score model for 385 COAD patients through correlation analysis. Through a series of analyses of the disease, it was found that the riskScore is closely related to the survival prognosis of patients. In future clinical treatments, we can use the risk score model to effectively predict the survival prognosis of patients with COAD, and we can do targeted immunotherapy for 7 immune genes (SLC10A2, CXCL3, IGHV5-51, INHBA, STC1, UCN, and OXTR) that constitute the riskScore to improve the prognosis of patients and improve the treatment effect.
SLC10A2 is mainly used to mediate the bile in the intestinal circulation and assist colonization of the intestinal flora. Recently, some research reports have reported that slc10a2 / PPARγ / PTEN / mTOR Signaling Pathway is related to the development of lung cancer [10, 11]. CXCL3 is related to the occurrence and development of prostate cancer, colon cancer, and breast cancer. There are also reports in the literature that the effect of suppressing the development of colon cancer can be achieved by immunosuppression of CXCL3 [12–16]. INHBA has a significant relationship with the occurrence and development of gastric, esophageal, and ovarian cancers, and studies have reported that the immunosuppressive treatment of INHBA can reduce the rate of deterioration of gastric and ovarian cancers [17–19]. STC1、OXTR, and UCN can promote the metastasis of colon cancer [20–23].