Investigation of the association between endometrial cancer immune microenvironment and gene expression based on machine learning and its predictive value for prognosis(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2024年第12期
- Page:
- 1568-1577
- Research Field:
- 医学生物信息
- Publishing date:
Info
- Title:
- Investigation of the association between endometrial cancer immune microenvironment and gene expression based on machine learning and its predictive value for prognosis
- Author(s):
- LIN Haihong1; GUO Yuanli2; PAN Ru1; LEI Nanxiang1; ZENG Weihong1
- 1. Department of Gynaecology, Meizhou Peoples Hospital, Meizhou 514130, China 2. Department of Gynaecology, the First Affiliated Hospital of Guangdong Pharmaceutical University, Guangzhou 510699, China
- Keywords:
- Keywords: endometrial cancer machine learning the Cancer Genome Atlas program immune gene prognosis
- PACS:
- R318；R711
- DOI:
- DOI:10.3969/j.issn.1005-202X.2024.12.016
- Abstract:
- Abstract: Objective To investigate the association between the immune microenvironment and gene expression in endometrial cancer (EC) and discuss its predictive value for prognosis, and to identify critical immune-related genes through bioinformatics analysis and machine learning techniques and construct a prognostic model for providing new directions for personalized treatment of EC. Methods Based on data from the Cancer Genome Atlas (TCGA) program, tools such as DESeq2, edgeR, and limma were utilized to screen for differentially expressed genes. Immune-related genes were selected by integrating data from the Immunology Database and Analysis Portal (ImmPort). Machine learning algorithms including Lasso regression, univariate feature selection, Boruta and random forest were employed to refine the selection of feature genes. Univariate and multivariate Cox regression analyses were conducted to assess the prognostic value of the genes, followed by construction of a risk score model. Additionally, tumor immune infiltration was analyzed using CIBERSORT algorithm, and key gene expressions were validated through immunohistochemistry. Results The intersection of 3 difference analysis results and immune-related genes identified 62 differentially expressed immune genes, and 25 potential biomarkers which were selected by a variety of machine learning models were considered as prognosis related genes. Univariate and multivariate Cox regression analyses confirmed that INHBE, SLURP1 and TNFSF11 genes were significantly associated with survival in EC patients. The constructed risk score model effectively distinguished the survival rate of different prognostic groups, and was related to the degree of immune cell infiltration. Immunohistochemical analysis further verified the differences in the expression of these genes between tumor and normal tissues. Conclusion INHBE, SLURP1 and TNFSF11 are key prognostic biomarkers in EC immune microenvironment, and their expression levels are closely associated with immune cell infiltration and patient survival rate, providing theoretical basis for EC precision medicine.
Last Update: 2024-12-20