Multidimensional immune subtyping reveals clinicogenetic molecular functional and pharmacologic heterogeneity in acute myeloid leukemia

Acute myeloid leukemia (AML) is characterized by substantial immune heterogeneity, yet current disease classification is dominated by cytogenetic and molecular features and does not systematically incorporate the tumor immune microenvironment (TIME). Here, we developed an AML-specific TIME classification based on 1503 curated immune-related genes and identified five immune subtypes (IS1–IS5) with distinct clinical, molecular, immunologic, and pharmacologic features. The subtype structure was established in GEO datasets and further evaluated in TCGA-LAML and Beat AML. TIME subtype was associated with survival, FAB morphology, ELN risk, mutational patterns, and established HOX–MEIS, NPM1-associated, and FLT3-ITD transcriptional programs, while remaining only partially explained by these conventional AML features. In Beat AML, addition of TIME subtype to age, sex, and ELN2017 significantly improved prognostic model fit and modestly increased discrimination, supporting its value as a complementary immune layer for risk assessment. Immune profiling identified IS4 as the most immune-engaged subtype, whereas IS1 and IS5 displayed relatively immune-low or exclusion-prone states. Projection of the pembrolizumab-treated GSE183415 cohort further supported the immunotherapy relevance of this immune-state axis. Importantly, primary AML samples provided functional validation: IS4 showed higher CD274, HLA-DRA, CXCL9, CXCL10, and LGALS9 expression, increased PD-L1/HLA-DR positivity, enhanced T-cell-mediated blast killing, and greater CD8 + T-cell activation compared with IS1. Beat AML drug-screen analysis additionally revealed subtype-specific ex vivo drug sensitivities, several of which persisted after adjustment for ELN2017. These findings establish a multidimensional AML TIME framework that complements existing clinicogenetic classification and provides a basis for refined immune stratification, prognostic assessment, and subtype-informed therapeutic investigation.