Abstract
Acute myeloid leukemia (AML) represents as a prevalent and formidable hematological malignancy, characterized by notably low 5-year survival rates. Ferroptosis has been found to be correlated with cancer initiation, therapeutic response, and clinical outcome. Nevertheless, the involvement of Ferroptosis-related genes (FRGs) in AML remains ambiguous. Five independent AML cohorts totaling 1,470 (GSE37642, GSE12417, GSE10358, Beat-AML, and TCGA-AML) patients with clinical information were used to systematically investigated the influence of these FRGs expression on outcome and tumor microenvironment. The integration of these datasets led to the subdivision into training and validation sets. Nineteen FRGs were identified as correlated with the overall survival (OS) of AML patients, primarily enriched in ferroptosis, fatty acid metabolism, and leukemia-related signaling pathways. The prognostic signature, consisting of 11 FRGs, was formulated using LASSO-Cox stepwise regression analysis. Patients with high-risk scores exhibited reduced survival compared to those in the low-risk group. The receiver operating characteristic (ROC) analysis underscored the signature's robust predictive accuracy. The high predictive efficacy was confirmed by both internal and external validation datasets. Leukemia and signaling related to immune regulation were mainly enriched pathways of the differentially expressed genes by comparing high- and low-risk groups. The immune composition deconvolution might indicate an immunosuppressive niche in the high-risk patients. The pRRophetic algorithm exploration unveiled chemical drugs with potentially sensitivity among patients in both groups. Collectively, our study developed a ferroptosis-derived prognostic signature that provides the OS prediction and identifies the immune microenvironment for AML patients on large-scale AML cohorts.