Background: Recently, an increasing number of studies have suggested dual-specificity phosphatase 23 (DUSP23) is a critical factor in the development of diffuse connective tissue disease and may be a valuable biomarker for primary human cancers. However, there is a lack of comprehensive studies on the prognostic significance of DUSP23 expression in acute myeloid leukemia (AML).
Methods: RNA sequencing data from The Cancer Genome Atlas (TCGA) (AML = 173), Genotype-Tissue Expression (GTEx) (healthy controls = 70) and GEO (AML = 461, healthy controls = 76) databases were used to compare DUSP23 expression between AML patients and healthy controls. The overall survival (OS) of DUSP23 in AML was evaluated using Kaplan-Meier Cox regression. Furthermore, univariate Cox regression and multivariate Cox regression analysis were used to determine whether DUSP23 was an independent prognostic factor for AML. We then verified the expression level and prognostic significance of DUSP23 in our cohort (AML = 128, healthy controls = 31). In addition, functional enrichment analysis of DUSP23-related DEGs was performed through gene set enrichment analysis (GSEA) and protein-protein interaction (PPI) network analysis.
Results: The expression level of DUSP23 is significantly higher in AML patients than in healthy controls in TCGA, GTEx, GEO databases and our cohort. By multivariate analysis, high expression of DUSP23 is a poor prognostic indicator of OS in the TCGA database. Next, we verified the role of DUSP23 as an adverse prognostic biomarker in our cohort. Enrichment analysis of related genes showed that DUSP23 may regulate important signal pathways in hematological tumors including the MAPK pathways. It is suggested by the PPI network that DUSP23, along with IMP3, MRPL4, MRPS12, POLR2L, and ATP5F1D may play a role in the process of AML.
Conclusion: The study demonstrated high expression of DUSP23 could serve as a poor independent prognostic biomarker in AML.

Keywords: DUSP23, expression, prognosis, AML, bioinformatics