Comprehensive Analysis of Prognostic Markers for Acute Myeloid Leukemia Based on Four Metabolic Genes

Yuanyuan Zhang; Shengling Ma; Moran Wang; Wei Shi; Yu Hu

doi:10.3389/fonc.2020.578933

Comprehensive Analysis of Prognostic Markers for Acute Myeloid Leukemia Based on Four Metabolic Genes

Front Oncol. 2020 Sep 29:10:578933. doi: 10.3389/fonc.2020.578933. eCollection 2020.

Authors

Yuanyuan Zhang¹, Shengling Ma², Moran Wang², Wei Shi^{2

3}, Yu Hu²

Affiliations

¹ Department of Oncology, Tongji Medical College, Tongji Hospital, Huazhong University of Science and Technology, Wuhan, China.
² Institute of Hematology, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
³ Fred Hutchinson Cancer Research Center, Seattle, WA, United States.

Abstract

Background: Metabolic reprogramming is the core characteristic of tumors during the development of tumors, and cancer cells can rely on metabolic changes to support their rapid growth. Nevertheless, an overall analysis of metabolic markers in acute myeloid leukemia (AML) is absent and urgently needed. Methods: Within this work, genetic expression, mutation data and clinical data of AML were queried from Genotype-Tissue Expression (GTEx) database, The Cancer Genome Atlas (TCGA) database and Gene Expression Omnibus (GEO) database. The tumor samples of TCGA were randomly divided into a training group (64 samples) and an internal validation group (64 samples) at one time, and the tumor samples of GEO served as two external validation groups (99 samples, 374 samples). According to the expression levels of survival-associated metabolic genes, we divided all TCGA tumor samples into high, medium and low metabolism groups, and evaluated the immune cell activity in the tumor microenvironment of the three metabolism groups by single-sample gene set enrichment analysis (ssGSEA) algorithm. Finally, we examined the mutations and prognostic effects of each model gene. Results: Four metabolism-related genes were screened and applied to construct a prognostic model for AML, giving excellent results. As for the area under the curve (AUC) value of receiver operating characteristic (ROC) curve, the training group was up to 0.902 (1-year), 0.81 (3-year), and 0.877 (5-year); and the internal and external validation groups also met the expected standards, showing high potency in predicting patient outcome. Univariate and multivariate prognostic analyses indicated that the riskScore obtained from our prognostic model was an independent prognostic factor. ssGSEA analysis revealed the high metabolism group had higher immune activity. Single and multiple gene survival analysis validated that each model gene had significant effects on the overall survival of AML patients. Conclusions: In our study, a high-efficiency prognostic prediction model was built and validated for AML patients. The results showed that metabolism-related genes could become potential prognostic biomarkers for AML.

Keywords: acute myeloid leukemia; bioinformatic analysis; metabolism; prognosis; survival.