Identification of an Immune-Related Gene Signature Based on Immunogenomic Landscape Analysis to Predict the Prognosis of Adult Acute Myeloid Leukemia Patients

Ruiqi Zhu; Huishan Tao; Wenyi Lin; Liang Tang; Yu Hu

doi:10.3389/fonc.2020.574939

Identification of an Immune-Related Gene Signature Based on Immunogenomic Landscape Analysis to Predict the Prognosis of Adult Acute Myeloid Leukemia Patients

Front Oncol. 2020 Nov 20:10:574939. doi: 10.3389/fonc.2020.574939. eCollection 2020.

Authors

Ruiqi Zhu¹, Huishan Tao², Wenyi Lin¹, Liang Tang¹, Yu Hu¹

Affiliations

¹ Institute of Hematology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
² Department of Gynecology and Obstetrics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.

Abstract

Acute myeloid leukemia (AML) is a hematopoietic malignancy characterized by highly heterogeneous molecular lesions and cytogenetic abnormalities. Immune disorders in AML and impaired immune cell function have been found to be associated with abnormal karyotypes in AML patients. Immunotherapy has become an alternative therapeutic method that can improve the outcomes of AML patients. For solid tumors, the expression patterns of genes associated with the immune microenvironment provide valuable prognostic information. However, the prognostic roles of immune genes in AML have not been studied as yet. In this study, we identified 136 immune-related genes associated with overall survival in AML patients through a univariate Cox regression analysis using data from TCGA-AML and GTEx datasets. Next, we selected 24 hub genes from among the 136 genes based on the PPI network analysis. The 24 immune-related hub genes further underwent multivariate Cox regression analysis and LASSO regression analysis. Finally, a 6 immune-related gene signature was constructed to predict the prognosis of AML patients. The function of the hub IRGs and the relationships between hub IRGs and transcriptional factors were investigated. We found that higher levels of expression of CSK, MMP7, PSMA7, PDCD1, IKBKG, and ISG15 were associated with an unfavorable prognosis of AML patients. Meanwhile, patients in the TCGA-AML datasets were divided into a high risk score group and a low risk score group, based on the median risk score value. Patients in the high risk group tended to show poorer prognosis [P = 0.00019, HR = 1.89 (1.26-2.83)]. The area under the curve (AUC) was 0.6643. Multivariate Cox Regression assay confirmed that the 6 IRG signature was an independent prognostic factor for AML. The prognostic role of the immune related-gene signature was further validated using an independent AML dataset, GSE37642. In addition, patients in the high risk score group in the TCGA dataset were found to be of an advanced age, IDH mutation, and M5 FAB category. These results suggested that the proposed immune related-gene signature may serve as a potential prognostic tool for AML patients.

Keywords: acute myeloid leukemia; data mining; immune-related gene signature; immunogenomic landscape; prognosis prediction.