Immune Infiltration-Related Genes Regulate the Progression of AML by Invading the Bone Marrow Microenvironment

Background: Acute myeloid leukemia has a high degree of clinical heterogeneity, posing a severe threat to human health. The relationship between immune infiltration and the bone marrow microenvironment is not clear. It is essential to find the pathogenic role of immune-related genes in the bone marrow microenvironment of AML, providing a new perspective for clinical risk stratification and predicting prognosis. Methods: WGCNA was utilized to screen modules related to immune infiltration in the TCGA-LAML database. Unsupervised clustering and mutation analysis are used to obtain candidate genes. LASSO regression analysis is used to conduct risk stratification assessments. GO, KEGG, GSEA, and GSVA are used to elucidate the biological processes of DEGs and high-risk groups, respectively. The TIMER database, ImmuCellAI, and ssGSEA are used for immune infiltration analysis. Validation of relative expression levels of hub genes using RT-qPCR. Results: Through WGCNA, seven modules were obtained, among which the turquoise module containing 1793 genes was highly correlated with the immune infiltration score. By unsupervised clustering, the turquoise module was divided into two clusters. The intersection of clinically significant genes in the TCGA and DEGs to obtain 178 genes for mutation analysis, followed by obtaining 17 genes with high mutation frequency. Subsequently, 17 genes were subjected to LASSO regression analysis to construct a riskscore model of 8 genes. The TIMER database, ImmuCellAI portal website, and ssGSEA elucidate that hub genes and risk scores are closely related to immune cell infiltration into the bone marrow microenvironment. In addition, we also validated the relative expression levels of hub genes using the TCGA database and GSE114868, additional expression levels of hub genes in AML cell lines in vitro. Conclusion: We constructed an immune infiltration-related gene model with good risk stratification and predictive prognosis through WGCNA, unsupervised clustering, mutation analysis, and LASSO regression analysis, which is expected to become a potential prognostic tool for AML.