A novel risk score model based on seven protein-coding genes and three pseudogenes predicts overall survival in acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is the most common hematologic malignancy in adults and is associated with poor clinical outcomes. Accurate prognostic stratification remains essential for improving patient management and identifying potential therapeutic targets. Here, we analyzed RNA-seq data from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) and The Cancer Genome Atlas (TCGA) cohorts to identify genes associated with overall survival (OS) in patients with AML. Using penalized regression modeling and integrative survival analyses, we identified a prognostic signature composed of seven protein-coding genes (ACOT7, SLC35E4, SELPLG, CCND3, RRAS, ITGAX, and COMTD1) and three processed pseudogenes (FDPSP2, UBE2V1P13, and AL158214.1) consistently associated with OS across independent AML cohorts. Based on these genes, we developed a novel risk score model that stratified AML patients into low- and high-risk groups with significantly different survival outcomes. The model demonstrated reproducible prognostic performance in TARGET and TCGA cohorts. In addition, several genes included in the prognostic signature were significantly dysregulated in AML compared with healthy samples. Functional enrichment analyses further revealed that high-risk AML was associated with suppression of translational and RNA-processing pathways, together with cohort-specific enrichment of metabolic and immune-related programs. Overall, this study presents a novel prognostic risk score integrating protein-coding genes and pseudogenes that robustly predicts OS in AML, providing a framework for improved prognostic stratification and potential biological insights into AML progression.