Single-Cell RNA sequencing analysis of cellular heterogeneity and microenvironment-driven signaling networks in acute myeloid leukemia

Acute myeloid leukemia (AML) exhibits cellular heterogeneity that complicates tumor micro-environment (TME) characterization. We integrated single-cell RNA sequencing (scRNA-seq) with BayesPrism bulk transcriptomic deconvolution to compare cellular composition by cytogenetic risk (favorable versus adverse). scRNA-seq identified immune and malignant hematopoietic populations, including leukemia stem cell (LSC) and granulocyte-monocyte-progenitor clusters, with recurrent alterations involving chromosomes 5, 7, and 8. TCGA-LAML deconvolution reconstructed cell-type-resolved profiles and linked adverse-risk AML to poorer overall survival. CellChat showed expanded cell-cell communication in adverse-risk disease involving SPP1, CXCL, GRN, TENASCIN, VCAM, and WNT signaling, particularly in leukemic stem-like and megakaryocytic compartments. Pathway analysis implicated differentiation arrest and immune exhaustion. In-silico repurposing nominated belinostat: ASGARD ranked it the top HDAC inhibitor, and scRANK characterized its HDAC1-centered target network. Gene-set analysis showed consistent interferon-gamma responses across key cell populations. These findings provide a transcriptome-based, hypothesis-generating framework linking cellular heterogeneity, microenvironmental signaling, and therapeutic vulnerabilities in adverse-risk AML.