基于炎症反应相关基因构建结肠癌预后模型及验证

Construction and validation of a prognostic model for colon cancer based on inflammatory response-related genes

目的应用生物信息学方法筛选结肠癌预后相关的炎症反应相关差异表达基因,构建并验证结肠癌预后模型。方法从癌症基因组图谱(TCGA)数据库中检索472例结肠癌患者及41名健康人正常结肠组织的RNA测序和临床数据。从国际癌症基因组联盟(ICGC)数据库中检索结肠癌预后相关基因表达及临床数据。检索时间均为建库至2022年11月。通过基因集富集分析(GSEA)数据库获取200个与炎症反应相关的基因,将其与TCGA数据库中获得的结肠癌和正常结肠组织的RNA测序基因数据集进行对比,获得炎症反应相关的差异表达基因。采用Cox比例风险模型分析评估TCGA数据库中与预后相关的差异表达基因,炎症反应相关的差异表达基因与预后相关的差异表达基因取交集,获得预后相关的炎症反应相关差异表达基因。通过LASSO Cox回归构建结肠癌预后模型。计算风险评分,按风险评分的中位值将TCGA数据库结肠癌患者分为低风险(<中位值)和高风险(≥中位值)两组。对两组患者进行主成分分析(PCA),采用Kaplan-Meier法进行生存分析。基于R软件timeROC程序包分析风险评分预测TCGA数据库结肠癌患者总生存(OS)的效能。应用ICGC数据库中的临床数据对构建的预后模型进行外部验证,依据TCGA数据库结肠癌患者中位风险评分将ICGC数据库结肠癌患者分为高、低风险组。使用R软件,对TCGA数据库中预后相关的炎症反应相关差异表达基因进行免疫细胞及免疫功能的单样本基因集富集分析(ssGESA)、免疫分型差异分析、免疫微环境相关性分析、免疫检查点基因差异分析。结果共获得60个炎症反应相关的差异表达基因,12个与预后相关的差异表达基因,取交集获得6个预后相关的炎症反应相关差异表达基因(CCL24、GP1BA、SLC4A4、SRI、SPHK1、TIMP1)。通过LASSO Cox回归分析,基于6个预后相关的炎症反应相关差异表达基因构建结肠癌预后模型,风险评分=-0.113×CCL24+0.568×GP1BA+(-0.375)×SLC4A4+(-0.051)×SRI+0.287×SPHK1+0.345×TIMP1。PCA结果显示,结肠癌患者可以较好地分为2个集群。TCGA数据库中高风险组OS较低风险组差(P<0.001);预后风险评分预测患者1、3、5年OS率的曲线下面积(AUC)分别为0.701、0.685、0.675。ICGC数据库中低风险组的OS优于高风险组,预后风险评分预测1、2、3年OS率的AUC分别为0.760、0.788、0.743。ssGSEA分析结果显示,TCGA数据库中高风险组免疫细胞浸润水平较高,尤其是活化树突细胞、巨噬细胞、中性粒细胞、浆细胞样树突细胞、T辅助细胞、滤泡辅助性T细胞得分均高于低风险组,辅助性T细胞2(Th2)得分低于低风险组(均P<0.05);免疫功能方面,高风险组抗原呈递细胞(APC)共抑制、APC共刺激、免疫检查点、人类白细胞抗原(HLA)、促进炎症、副炎症、T细胞刺激、Ⅰ型干扰素(IFN)反应、Ⅱ型IFN反应得分均高于低风险组(均P<0.05)。免疫分型分析结果显示,IFN-γ为主型(C2)的炎症反应评分最高,分别与创伤愈合型(C1)、炎症反应型(C3)比较差异均有统计学意义(均P<0.05)。免疫微环境基质细胞、免疫细胞均与预后风险评分呈正相关(r值分别为0.35、0.21,均P<0.01)。免疫检查点差异分析结果显示,高、低风险组程序性死亡受体配体1(PD-L1)表达量比较,差异有统计学意义(P=0.002),且PD-L1表达量与预后风险评分呈正相关(r=0.23,P<0.01)。结论炎症反应相关基因在结肠癌肿瘤免疫中可能发挥重要作用,可以用于结肠癌患者的预后分析及免疫治疗。

Objective To screen the differentially expressed genes(DEG)related to inflammatory response associated with the prognosis of colon cancer based on the bioinformatics approach,and to construct and validate a prognostic model for colon cancer.Methods RNA sequencing and clinical data of 472 colon cancer patients and normal colon tissues of 41 healthy people were retrieved from the Cancer Genome Atlas(TCGA)database.Gene expression related to prognosis of colon cancer and clinical data were retrieved from the International Cancer Genome Consortium(ICGC)database.The retrieval time was all from the establishment of library to November 2022.A total of 200 genes associated with inflammatory response obtained from the Gene Set Enrichment Analysis(GSEA)database were compared with the RNA sequencing gene dataset of colon cancer and normal colon tissues obtained from the TCGA database,and then DEG associated with inflammatory response were obtained.The prognosis-related DEG in the TCGA database were analyzed by using Cox proportional risk model,and the inflammatory response-related DEG were intersected with the prognosis-related DEG to obtain the prognosis-related inflammatory response-related DEG.The prognostic model of colon cancer was constructed by using LASSO Cox regression.Risk scores were calculated,and colon cancer patients in the TCGA database were divided into two groups of low risk(<the median value)and high risk(≥the median value)according to the median value of risk scores.Principal component analysis(PCA)was performed on patients in both groups,and survival analysis was performed by using Kaplan-Meier method.The efficacy of risk score in predicting the overall survival(OS)of colon cancer patients in the TCGA database was analyzed based on the R software timeROC program package.Clinical data from the ICGC database were applied to externally validate the constructed prognostic model,and patients with colon cancer in the ICGC database were classified into high and low risk groups based on the median risk score of patients with colon cancer in the TCGA database.By using R software,single-sample gene set enrichment analysis(ssGESA),immunophenotyping difference analysis,immune microenvironment correlation analysis,and immune checkpoint gene difference analysis of immune cells and immune function were performed for prognosis-related inflammation response-related DEG in the TCGA database.Results A total of 60 inflammatory response-related DEG and 12 prognosis-related DEG were obtained;and 6 prognosis-related inflammatory response-related DEG(CCL24,GP1BA,SLC4A4,SRI,SPHK1,TIMP1)were obtained by taking the intersection set.LASSO Cox regression analysis showed that a prognostic model for colon cancer was constructed based on 6 prognosis-related inflammatory response-related DEG,and the risk score was calculated as=-0.113×CCL24+0.568×GP1BA+(-0.375)×SLC4A4+(-0.051)×SRI+0.287×SPHK1+0.345×TIMP1.PCA results showed that patients with colon cancer could be better classified into 2 clusters.The OS in the high-risk group was worse than that in the low-risk group in the TCGA database(P<0.001);the area of the curve(AUC)of the prognostic risk score for predicting the OS rates of 1-year,3-year,5-year was 0.701,0.685,and 0.675,respectively.The OS of the low-risk group was better than that of the high-risk group in the ICGC database;AUC of the prognostic risk score for predicting the OS rates of 1-year,2-year,3-year was 0.760,0.788,and 0.743,respectively.ssGSEA analysis showed that the level of immune cell infiltration in the high-risk group in the TCGA database was high,especially the scores of activated dendritic cells,macrophages,neutrophils,plasmacytoid dendritic cells,T helper cells,and follicular helper T cells in the high-risk group were higher than those in the low-risk group,while the score of helper T cells 2(Th2)in the high-risk group was lower compared with that in the low-risk group(all P<0.05);in terms of immune function,the high-risk group had higher scores of antigen-presenting cell(APC)co-inhibition,APC co-stimulation,immune checkpoint,human leukocyte antigen(HLA),promotion of inflammation,parainflammation,T-cell stimulation,typeⅠinterferon(IFN)response,and typeⅡIFN response scores compared with those in the low-risk group(all P<0.05).The results of immunophenotyping analysis showed that IFN-γ-dominant type(C2)had the highest inflammatory response score,and the differences were statistically significant when compared with trauma healing type(C1)and inflammatory response type(C3),respectively(all P<0.05).Immune microenvironment stromal cells and immune cells were all positively correlated with prognostic risk scores(r values were 0.35 and 0.21,respectively,both P<0.01).The results of immune checkpoint difference analysis showed there was a statistically significant difference in programmed-death receptor ligand 1(PD-L1)expression level between high-risk group and low-risk group(P=0.002),and PD-L1 expression level was positively correlated with prognostic risk score(r=0.23,P<0.01).Conclusions Inflammatory response-related genes may play an important role in tumor immunity of colon cancer and can be used in the prognostic analysis and immunotherapy of colon cancer patients.