Integrated m6A reader network in acute myeloid leukemia: prognostic modeling, immune modulation, and functional validation of YTHDF3.

Emerging evidence highlights RNA N6-methyladenosine (m6A) modifications as pivotal regulators of tumorigenesis, yet the synergistic roles of m6A readers in the pathogenesis and prognosis of acute myeloid leukemia (AML) remains unclear. By leveraging multiomics data from 2680 AML patients, we systematically mapped the dysregulated m6A reader network, revealing the coordinated upregulation of readers and their associations with aggressive AML subtypes. We devised a m6A reader activity score (m6ARS), and an elevated m6ARS is associated with excessive proliferation, impaired CD8+ T cell infiltration, and PD-L1-mediated immune evasion. A machine learning-derived 28-gene prognostic model demonstrated robust performance across 10 independent cohorts and outperformed traditional clinicopathological factors in multivariate analysis. High-risk patients exhibited heightened sensitivity to targeted therapies (e.g., dasatinib and olaparib), suggesting actionable therapeutic vulnerabilities. Functional studies identified YTHDF3 as a novel oncogenic driver that is significantly overexpressed in AML, where its knockdown suppressed proliferation, induced apoptosis, and stabilized leukemogenic transcripts (ZZZ3, KLHL11) via the regulation of mRNA stability. This integrative study reveals the m6A reader network as a central orchestrator of AML progression, provides a validated prognostic framework for risk-adapted therapy, and positions YTHDF3 as a potential diagnostic and therapeutic target, bridging RNA epigenetics with clinical translation in AML.