The intellectual backbone of the review is a distinction that the authors insist has been too often blurred in the literature. Bulk acute myeloid leukemia blasts and leukemic stem cells are not interchangeable entities. The stem cell concept traces back to landmark xenotransplantation studies from the 1990s showing that only a minor fraction of primary leukemia cells could initiate and serially propagate the disease in mice. Because surface markers alone do not prove stemness, the authors reserve the term leukemic stem cell only for studies that directly demonstrated functional self-renewal or serial repopulation. This matters therapeutically, because bulk blasts and stem cells remodel the marrow through partly overlapping but non-identical programs, and a drug validated against one population may fail against the other.

To appreciate what leukemia destroys, one must first understand what the healthy niche does. The normal bone marrow contains interconnected endosteal and perivascular niches where mesenchymal stromal cells, endothelial cells, osteolineage cells, macrophages, adipocytes, and sympathetic nerves deliver spatially and temporally orchestrated signals. Stem cell factor and the chemokine CXCL12, produced by stromal and vascular populations, keep hematopoietic stem cells retained, quiescent, and ready for regeneration. Metabolically, quiescent stem cells favor glycolysis in low-oxygen, low-reactive-oxygen-species environments, while activation pushes them toward mitochondrial oxidative phosphorylation. Even sympathetic neural signaling contributes, driving circadian stem cell mobilization through regulation of stromal CXCL12. It is precisely this finely tuned machinery that leukemia dismantles.

The review catalogs how bulk leukemia cells vandalize these compartments. Classical models demonstrate leukemia-induced sympathetic neuropathy, endosteal vascular remodeling, and disruption of the adipocytic niche, all of which impair normal blood production. At the stromal level, acute myeloid leukemia cells can hijack bioelectric signaling through CaV1.2-dependent mechanisms and induce p16INK4a-driven senescence in mesenchymal stromal cells, converting them into senescent factories that secrete pro-leukemic inflammatory factors such as IL-6 and CCL2. AML-derived BMP6 induces the transcription factor ID1 in stromal cells, which stabilizes SP1 and boosts production of ANGPTL7, further supporting leukemia progression. Meanwhile, TWIST1-high clonal cells redirect stromal differentiation toward an adipogenic, leukemia-supportive state, and TGF-beta1 functionally disables the stromal support of healthy hematopoiesis.

The leukemic stem cell evidence is narrower but mechanistically striking. In one key study, TET2-mediated mRNA demethylation regulated expression of TSPAN13 and thereby tuned the CXCR4/CXCL12 axis that governs stem cell homing to protected marrow regions and their self-renewal. Another investigation defined an ALDH2-PKCdelta-SHMT2 axis that sustains metabolic plasticity and allows stem-enriched populations to evade chemotherapy. ALDH3A2 has been shown to protect AML cells from oxidative death and ferroptosis-inducing stress, although the authors note that this study did not formally define the protected cells as leukemic stem cells. In the KG-1a cell-line model, CX43-mediated mitochondrial transfer from stromal cells preserved stem-like properties, while primary patient studies documented chemotherapy-associated mitochondrial transfer to bulk leukemic cells, a form of cellular altruism that literally donates energy machinery to surviving tumor cells.

Drug resistance emerges repeatedly as a niche-driven phenomenon. CD44 engagement on leukemia cells increases the avidity of the integrin VLA-4, strengthening VCAM-1-dependent adhesion and pro-survival signaling within the marrow. Stromal cells express CYP3A4, a drug-metabolizing enzyme that can chemically inactivate FLT3 inhibitors, effectively turning healthy tissue into a shield against targeted therapy. In FLT3-mutant AML, longitudinal profiling revealed a stepwise, niche-catalyzed evolution toward resistance to gilteritinib, including early dependence on AURKB, and BTK-dependent autophagy contributes to FLT3-inhibitor failure. TGF-beta receptor-SMAD3 signaling enhances DNA double-strand break repair and confers PARP-inhibitor resistance in marrow-based models, though the authors classify this as a leukemia-cell rather than stem-cell mechanism. Even venetoclax, the flagship BCL-2 inhibitor that preferentially eradicates quiescent leukemic stem cells by disrupting oxidative phosphorylation in combination with azacitidine, can be undermined by stromal signaling that raises alternative anti-apoptotic proteins and metabolic flexibility.

The review also surveys how immune and metabolic warfare unfold in the leukemic marrow. T cells from AML patients’ bone marrow show reversible functional suppression, elevated checkpoint expression, and defective immune-synapse formation, while AML-cell IDO1 expression shifts the B-cell compartment toward regulatory populations. Metabolic competition compounds the damage: the voracious glycolysis and oxygen consumption of leukemic cells starve effector immune cells of substrate, and accumulating lactate drives GPR81-dependent polarization of macrophages toward an immunosuppressive phenotype. Loss of ANXA5 in leukemia cells amplifies inflammatory conditioning through the COX2/PGE2 axis. On the defensive side, activating the NLRP3 inflammasome in marrow dendritic cells increased IL-1beta and Th1/IFN-gamma responses with antileukemic activity in preclinical models, hinting that reawakening local immunity is feasible.

To contextualize AML biology, the authors carefully compare other marrow-resident malignancies while insisting their findings not be transplanted wholesale. In B-cell acute lymphoblastic leukemia, stromal cells convert to cancer-associated fibroblast-like states and adipogenic progenitors support residual disease through SDF-1/CXCR4 and IL-7/VCAM1 pathways; in T-ALL, dexamethasone-induced adipocytes protect cells via a CXCL13-associated program. In chronic myeloid leukemia, TNF-alpha expands stromal precursors that promote leukemic stem cell growth through CXCL1/CXCR2, and stromal transfer of metabolic proteins drives imatinib resistance. Multiple myeloma remodels adipocytes into senescent, metabolically dysfunctional cells that support steroid resistance. These parallels reveal conserved principles, such as chemokine-mediated retention and bidirectional metabolic exchange, but the authors treat the specific pathways as comparative hypotheses rather than established AML targets.

On the therapeutic frontier, the review summarizes an array of niche-directed strategies, most still preclinical. Bone-marrow-targeted yolk-shell nanoparticles loaded with a LOXL2 inhibitor disrupt the collagen scaffold that anchors leukemia cells, detaching them from matrix protection and sensitizing them to chemotherapy. A marrow-responsive polymeric system co-delivering a drug and siRNA against CXCR4 and mitophagy reduced stem cell homing, stemness, and relapse-related phenotypes. Bisphosphonate-modified nanocomposites targeting the C-Myb/Survivin axis and CD123-targeted aptamer nanoparticles delivering EZH2 siRNA demonstrated activity against MLL-AF9-rearranged AML in cell, animal, and patient-derived xenograft models. The BET protein degrader ARV-825 simultaneously attacked leukemic stem and progenitor cells and eroded stromal protection, while arsenic trioxide has been reported to reprogram the microenvironment while directly killing blasts. VLA-4 blockade sensitized leukemic and myeloma cells to CD3-redirection therapy in cross-disease preclinical models, and a phase 3 transplantation trial showed that prophylactic N-acetyl-L-cysteine improved endothelial function and hematopoietic reconstitution after haploidentical stem cell transplantation, underscoring that rebuilding the niche matters as much as attacking it.

The authors close with a sober assessment of what stands between promising preclinical results and clinical reality: model dependence and the cellular and spatial heterogeneity of the human marrow. They propose five research priorities, including spatially resolved mapping of leukemia cell and stem cell states across marrow subniches, functional validation of putative stem-cell-specific mechanisms, humanized organoid and organ-on-chip platforms, rational scheduling of niche-directed agents with chemotherapy or immunotherapy, and strategies that restore normal vascular and stromal function after remission. Recent work showing that venetoclax and azacitidine can themselves damage sinusoidal endothelium and alter immune subsets is a pointed reminder that even our best drugs reshape the battlefield. If the marrow niche can be mapped, decoded, and selectively disarmed, the sanctuary that has long shielded acute myeloid leukemia from cure may finally become its Achilles’ heel.

Subject of Research: How the bone marrow microenvironment is remodeled in acute myeloid leukemia and how niche-directed therapies may overcome treatment resistance

Article Title: Malignant remodeling of the bone marrow microenvironment in acute myeloid leukemia: mechanisms and therapeutic implications

Article References: Xu, W., Han, Q., Sun, C., & Li, D. (2026). Malignant remodeling of the bone marrow microenvironment in acute myeloid leukemia: mechanisms and therapeutic implications. Frontiers in Oncology, 16, Article 1903019. https://doi.org/10.3389/fonc.2026.1903019

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1903019

Keywords: acute myeloid leukemia, bone marrow microenvironment, leukemic stem cells, treatment resistance, mesenchymal stromal cells, CXCL12-CXCR4 axis, venetoclax, immunosuppression, metabolic reprogramming, nanomedicine, drug resistance, hematopoiesis