State of the art, state of the cell: State-transition modeling of hematopoietic stem cells in CML and AML

Normal hematopoiesis is organized as a hierarchical and dynamic system in which multipotent hematopoietic stem cells (HSCs) reside within specialized bone marrow niches that regulate quiescence, self-renewal, and differentiation. Rather than occupying fixed identities, HSCs continuously transition between functional states to sustain lifelong hematopoiesis while limiting stem cell exhaustion. Disruption of these tightly regulated state transitions can lead to malignant transformation and hematologic disease. Leukemia, including chronic myeloid leukemia (CML) and acute myeloid leukemia (AML), is driven by leukemic stem cells (LSCs) that retain key features of normal HSCs, including quiescence, self-renewal, and multipotency, while exhibiting dysregulated differentiation and altered cellular-state dynamics. Although CML and AML differ in genetic drivers, disease kinetics, and clinical behavior, both are recognized as disorders of disrupted cellular-state regulation rather than solely arising from genetic alterations. Recent advances in single-cell and multi-omics technologies have shaped our understanding of hematopoiesis by enabling high-resolution mapping of cellular heterogeneity and developmental trajectories. These approaches challenge traditional hierarchical models and instead support a view of hematopoiesis and leukemogenesis as a continuous movement through cellular state-space. In line with this view, state-transition theory provides a quantitative modeling framework of disease progression defined by temporal transitions between stable and unstable cellular states. Integration of mathematical modeling, computational methods, and artificial intelligence (AI) facilitates identification of latent cellular states, modeling of transition probabilities, and detection of rare populations that drive disease evolution, treatment resistance, and relapse. In this review, we explore the current understanding of HSC dysregulation in CML and AML and propose state-transition theory as a unifying framework for leukemogenesis. We discuss how integrating single-cell technologies with computational modeling and AI may enable predictive mapping of cellular state-space, offering new opportunities for early disease detection, therapeutic targeting, and prevention of malignant transformation. We further propose that longitudinal multi-omic state-transition frameworks provide a quantitative approach for modeling how therapy reshapes cellular state-space. From this perspective, treatment acts as a perturbation that remodels the underlying state-space landscape, while state-transition theory provides a framework for measuring these changes, allowing for the prediction of treatment outcomes, and informing the development of more effective therapeutic strategies.