Abstract
BACKGROUND: Acute Myeloid Leukemia (AML) is an aggressive hematologic malignancy with significant clinical challenges due to its heterogeneity and high relapse rate. Dysregulated ribosome biogenesis is a recognized driver of tumorigenesis and therapy resistance, but its comprehensive impact on AML prognosis and the immune microenvironment remains to be fully elucidated. METHODS: We constructed an AML single-cell atlas and quantified ribosome biogenesis activity. A prognostic ribosome biogenesis signature (RBS) was developed through integrated machine learning and interpreted using the SHapley Additive exPlanations (SHAP) framework. We evaluated the RBS associations with prognosis, tumor immune microenvironment, immunotherapy response, and drug sensitivity. Key findings were validated by qPCR, Western blot, and molecular docking. RESULTS: Single-cell analysis revealed significantly elevated ribosome biogenesis activity scores (RAS, a composite metric quantifying cellular ribosome synthesis capacity) in AML stem and progenitor cells. The RBS model, established using a Random Survival Forest algorithm, demonstrated robust prognostic power. SHAP interpretation identified EXOSC2 as the top predictor. Low-RBS patients displayed an immune-activated microenvironment and heightened immunotherapy response. Molecular docking indicated high-affinity binding of ouabain and digoxin to EXOSC2. CONCLUSION: Our study delineates the critical role of ribosome biogenesis in AML progression and establishes an interpretable RBS prognostic signature. This tool effectively assesses immunotherapy responsiveness and reveals novel targeting opportunities, providing valuable insights for clinical decision-making to improve AML outcomes.