A single-cell comprehensive atlas of leukemia and its microenvironment

Acute myeloid leukemia (AML) remains highly lethal, with treatment failure and relapse driven by dynamic cellular states and evolving interactions within the leukemic ecosystem. In this study, we built a comprehensive atlas of leukemia and its microenvironment (CALM) by integrating single-cell profiling from 73 AML samples and 9 healthy donors, bulk RNA sequencing (RNA-seq) from 87 newly diagnosed patients, and 4 paired bone marrow and peripheral blood single-cell data. All datasets were linked to clinical, cytogenetic, molecular, and treatment-response annotations. Immune analysis revealed diverse T cell distribution across disease status, where patients with persistent disease exhibited greater clonal expansion. Machine learning modeling on conventional T cells identified a T-cell signature pertinent to AML and outcomes, which were validated in two cohorts. CALM is a publicly accessible and interactive portal (calm.mdanderson.org) for exploring cellular compartments, cell-cell interactions, and longitudinal transcriptional and clonal dynamics to enable hypothesis generation, biomarker discovery, and target validation.