Derivation of a Novel Multi‐Gene Prognostic Model Based on Regulated Cell Death Pathways in Acute Myeloid Leukemia: A Comprehensive Bioinformatic Analysis Integrating Gene Expression, Mutation Profiling, and Immune Infiltration Publisher Pubmed



A Ahmadi ALI ; Aa Navidinia Amir ABBAS ; D Bashash DAVOOD ; B Poopak BEHZAD ; S Esmaeili SHADI
Source: PLOS ONE Published:2025

Abstract

Background Acute myeloid leukemia (AML) is a highly aggressive hematologic malignancy with dismal survival outcomes, where dysregulation of regulated cell death (RCD) pathways plays a pivotal role in leukemogenesis and therapeutic resistance. Methods Differential expression analyses were performed comparing AML samples with healthy bone marrow. Diagnostic differentially expressed genes (DEGs) were then intersected with curated gene sets representing apoptosis, pyroptosis, autophagy, necroptosis, and ferroptosis to derive an RCD-based gene signature. Prognostic markers were identified by univariate Cox regression, and these markers were refined using LASSO regression to construct a multi-gene prognostic model that generated an individual risk score (RS) for each patient. The performance of the model was validated internally through Kaplan–Meier survival analyses and receiver operating characteristic (ROC) curves for 1-, 3-, and 5-year survival, and externally confirmed in an independent TARGET-AML cohort. In addition, mutation analysis was conducted using the maftools package, and immune infiltration profiling was performed with CIBERSORT and xCell to characterize the molecular landscape of the risk groups. Results Our integrative approach yielded a four-gene prognostic model incorporating ARHGEF35, GSN, ELANE, and AKT3. High RS was strongly associated with adverse overall survival, with Kaplan–Meier analyses showing p-value<0.0001 in the training cohort and p-value=0.0026 in the testing cohort. The model demonstrated robust predictive accuracy with AUC values of 82%, 87%, and 91% for 1-, 3-, and 5-year survival in the training set, and 65%, 81%, and 94% in the testing set. Mutation analysis revealed that DNMT3A and RUNX1 mutations were significantly enriched in high-RS patients (p-value=0.0015 and p-value=0.0086, respectively), whereas KIT mutations were more prevalent in low-RS patients (p-value=0.0058). Immune profiling indicated that high-RS patients had increased M2 macrophage infiltration (p-value=0.0027) and reduced resting mast cells (p-value=0.0033). Conclusion These findings establish that an RCD-based multi-gene risk model can robustly stratify AML patients by prognosis and illuminate underlying genomic and immunologic mechanisms, thereby offering promising avenues for personalized therapeutic strategies. © 2025 Elsevier B.V., All rights reserved.