2023 – COORDINATED INTERACTIONS ACROSS BIOLOGICAL SCALES DEFINE DISEASE TRAJECTORIES IN ACUTE MYELOID LEUKEMIA

Abstract

Acute myeloid leukemia (AML) exhibits extensive heterogeneity across genetic, cellular, and microenvironmental scales, yet how these layers interact to shape disease phenotypes and clinical trajectories remains poorly understood. Here, we present a multiscale single-cell dataset of 184 treatment-naïve AML patients spanning all major genetic subtypes, coupled with detailed immune microenvironment profiling and comprehensive clinical annotation and follow-up. We define ten AML archetypes based on recurrent genotype–phenotype relationships within leukemic stem and progenitor cells, refining established classifications and resolving heterogeneity within NPM1- and myelodysplasia-related AML. To systematically resolve how biological layers profiled at diagnosis interact and shape individual clinical endpoints, we develop an analytical framework based on quantitative modeling. We demonstrate that distinct clinical outcomes arise from specific interactions operating within and across these layers. Response to induction chemotherapy is primarily determined by the interaction between genetic drivers and leukemic differentiation state, captured by AML archetypes. In contrast, relapse following chemotherapy is driven predominantly by nongenetic leukemia-intrinsic programs linked to metabolic state. Relapse after allogeneic stem cell transplantation is shaped mainly by microenvironmental factors and the extent of residual healthy hematopoiesis. Finally, overall survival was determined by a complex interplay of genetic and nongenetic determinants, alongside largely independent effects such as systemic interferon signaling. Together, our study provides a framework to resolve intrapatient and interpatient heterogeneity in AML and supports a unifying model in which distinct clinical outcomes are governed by specific, recurrent patterns of interaction across biological layers.

Acute myeloid leukemia (AML) exhibits extensive heterogeneity across genetic, cellular, and microenvironmental scales, yet how these layers interact to shape disease phenotypes and clinical trajectories remains poorly understood. Here, we present a multiscale single-cell dataset of 184 treatment-naïve AML patients spanning all major genetic subtypes, coupled with detailed immune microenvironment profiling and comprehensive clinical annotation and follow-up. We define ten AML archetypes based on recurrent genotype–phenotype relationships within leukemic stem and progenitor cells, refining established classifications and resolving heterogeneity within NPM1- and myelodysplasia-related AML. To systematically resolve how biological layers profiled at diagnosis interact and shape individual clinical endpoints, we develop an analytical framework based on quantitative modeling. We demonstrate that distinct clinical outcomes arise from specific interactions operating within and across these layers. Response to induction chemotherapy is primarily determined by the interaction between genetic drivers and leukemic differentiation state, captured by AML archetypes. In contrast, relapse following chemotherapy is driven predominantly by nongenetic leukemia-intrinsic programs linked to metabolic state. Relapse after allogeneic stem cell transplantation is shaped mainly by microenvironmental factors and the extent of residual healthy hematopoiesis. Finally, overall survival was determined by a complex interplay of genetic and nongenetic determinants, alongside largely independent effects such as systemic interferon signaling. Together, our study provides a framework to resolve intrapatient and interpatient heterogeneity in AML and supports a unifying model in which distinct clinical outcomes are governed by specific, recurrent patterns of interaction across biological layers.