Establishment and Validation of Immune Risk Score for Predicting Survival of Patients with Acute Myeloid Leukemia

Abstract

To establish an immune gene prognostic model of acute myeloid leukemia (AML) and explore its correlation with immune cells in bone marrow microenvironment. Gene expression profile and clinical data of TCGA-AML were downloaded from TCGA database. Immune genes were screened by LASSO analysis to construct prognosis prediction model, and prediction accuracy of the model was quantified by receiver operating characteristic curve and area under the curve. Survival analysis was performed by Log-rank test. Enriched pathways in the different immune risk subtypes were evaluated from train cohort. The relationship between immune prediction model and bone marrow immune microenvironment was verified by flow cytometry in the real world. Patients with low-risk score of immune gene model had better prognosis than those with high-risk score. Multivariate analysis showed that the immune gene risk model was an independent prognostic factor. The risk ratio for AML patients in the training concentration was HR=24.594 (95%CI: 6.180-97.878), and the AUC for 1-year, 3-year, and 5-year overall survival rate was 0.811, 0.815, and 0.837, respectively. In addition, enrichment analysis of differential gene sets indicated activation of immune-related pathways such as cytokines and chemokines as well as autoimmune disease-related pathways. At the same time, real world data showed that patients with high immune risk had lower numbers of CD8+T cells and B lymphocytes compared with low immune risk patients. We constructed a stable prognostic model for AML, which can not only predict the prognosis of AML, but also reveal the dysregulation of immune microenvironment.