Identification of New Gene Labels for Colon Cancer Prognosis Based on Random Survival Forest Model

Objective To investigate the gene expression profile of mRNA in colon cancer and determine the optimal prognostic markers.Method The colon cancer dataset TCGA-COAD was downloaded from the Cancer Genome Atlas (TCGA) database as the training cohort. The random survival forest (RSF) model was used to determine gene labels, and the obtained gene labels were analyzed using the Cox model to construct risk scores. The colon cancer dataset GSE17536 was downloaded from the Gene Expression Database (GEO) as the validation cohort to validate the model, and compared horizontally with similar studies in the past year. Exploring the relationship between gene tags and immune cells through immune cell infiltration.Result A total of 11 gene tags were screened, and the risk score constructed by the multi factor Cox model was an independent prognostic indicator for colon cancer patients. The comparative development of this model is superior to previous studies. Immune cell infiltration revealed a significant correlation (P<0.05) between monocytes and the gene labels used in this study.Conclusion This study identified 11 gene markers with prognostic value for colon cancer, and monocytes may serve as potential therapeutic targets for colon cancer.

研究目的 探究结肠癌中mRNA的基因表达谱，并确定最优预后标志物。

研究方法 从癌症基因组图谱（The Cancer Genome Atlas, TCGA）数据库下载结肠癌数据集TCGA-COAD作为训练队列；采用随机生存森林（Random Survival Forest, RSF）模型确定基因标签，利用Cox模型分析所得基因标签以构建风险评分；从基因表达数据库（Gene Expression Database, GEO）下载结肠癌数据集GSE17536作为验证队列验证模型，并与过去一年的同类研究进行横向比较；通过免疫细胞浸润分析探究基因标签与免疫细胞的关系。

研究结果 共筛选出11个基因标签，多因素Cox模型构建的风险评分是结肠癌患者的独立预后指标；本模型的表现优于过去一年的同类研究；免疫细胞浸润分析显示，单核细胞与本研究使用的基因标签存在显著相关性（P<0.05）。

研究结论 本研究鉴定出11个具有结肠癌预后价值的基因标志物，单核细胞或可作为结肠癌的潜在治疗靶点。