机器学习算法确定的核心衰老基因在非远处转移性结直肠癌的预后和免疫微环境相关性分析
Correlation analysis of prognosis and immune microenvironment of core senescence genes identified by machine learning in non-distant metastatic colorectal cancer
李世森 1余鹏飞 1乔一桓 1王珂 2李丹 2李云龙1
作者信息
- 1. 空军军医大学西京医院消化外科,陕西西安 710032
- 2. 空军军医大学西京医院消化外科手术室,陕西西安 710032
- 折叠
摘要
目的 建立非远处转移性结直肠癌(CRC)衰老基因模型和确定核心的衰老基因,分析其与肿瘤微环境的作用.方法 下载TCGA和GEO数据库的非远处转移性CRC数据库,单因素Cox分析确定预后相关的衰老基因,多因素Cox分析建立非远处转移CRC的衰老模型.ESTIMATE算法计算出样本的免疫微环境评分(间质评分与免疫评分),并且研究衰老模型与其关系.采用随机生存森林算法用于计算出衰老基因的重要性,确定衰老核心基因.使用CIBERSORT算法评估肿瘤免疫细胞浸润情况,相关性分析研究衰老核心基因与细胞浸润的关系.结果 多因素Cox回归分析建立由6个衰老基因组成的模型,在TCGA和GEO中都表明高风险人群的预后远远差于低风险组(P＜0.001).该衰老模型与CRC微卫星状态、肿瘤免疫评分和间质评分相关.随机生存森林算法确定3个核心的衰老基因(TERT、SNAI1和CSNK1A1),且与肿瘤免疫细胞浸润相关.此外,在免疫治疗人群中,SNAI1(P=0.016)和CSNK1A1(P＜0.001)低表达组生存时间更长.结论 肿瘤细胞衰老的程度越大,代表其增殖能力越弱,非远处转移性CRC患者的预后可能越佳.TERT、SNAI1和CSNK1A1作为衰老的核心基因,与肿瘤微环境相关,可能作为免疫治疗的生物学标志物.
Abstract
Objective To establish a senescence gene model for non-distant metastatic colorectal cancer(CRC),identify core senescence genes,and analyze their role in the tumor microenvironment.Methods Non-distant metastatic CRC databases of TCGA and GEO databases were downloaded,univariate Cox analysis was performed to identify prognostic-related senescence genes,and multivariate Cox analysis was performed to establish a senescence model for non-distant metastatic CRC.The ESTIMATE algorithm calculated the immune microenvironment score(stromal and immune score)of the cancerous sample and investigated the relationship between the aging model and it.The random survival forest algorithm was used to calculate the importance of aging genes and determine the core genes of aging.The CIBERSORT algorithm was used to evaluate the infiltration of tumor immune cells,and correlation analysis was conducted to study the relationship between core genes of aging and cell infiltration.Results Multivariate Cox analysis established a model composed of 6 senescence genes.Both TCGA and GEO showed that the prognosis of high-risk group was much worse than that of low-risk group(P＜0.001).The aging model was correlated with CRC microsatellite status,tumor immune score and stromal score.Three core senescence genes(TERT,SNAI1 and CSNK1A1)were identified by random survival forest algorithm and were associated with tumor immune cell infiltration.In addition,in the immunotherapy population,those with low expression of SNAI1(P=0.016)and CSNK1A1(P＜0.001)had longer survival times.Conclusion The greater the degree of senescence of tumor cells,the weaker their ability to proliferate,and the better the prognosis of non-distant metastatic CRC patients may be.As the core genes of senescence,TERT,SNAI1 and CSNK1A1 are related to the tumor microenvironment and may be used as biomarkers for immunotherapy.
关键词
衰老基因/非远处转移结直肠癌/肿瘤免疫微环境/免疫细胞Key words
senescence genes/non-distant metastatic colorectal cancer/tumor immune microenvironment/immune cells引用本文复制引用
基金项目
国家自然科学基金青年科学基金(82100680)
出版年
2024
国家科技期刊平台
NSTL
万方数据