Remodeling the marrow fat niche: BMAT in cancer bone metastases and hematological malignancies Authors: Izabela Podgorski Published: 2026 ABSTRACT Bone marrow adipose tissue (BMAT) is a dynamic component of the marrow microenvironment that influences tumor persistence, therapy response, and skeletal integrity. Recent work shows that marrow adipocytes function not only as metabolic partners for malignant cells but also as stress-responsive components of bone marrow niche remodeled by tumor progression and cancer therapy. Reciprocal signaling between tumor cells and adipogenic lineage populations alters lipid mobilization, redox balance, and stromal differentiation, creating microenvironments that support tumor cell survival under metabolic and therapeutic stress. Cancer therapies can reprogram adipocyte and stromal populations, generating inflammatory or senescent niches that persist beyond active disease and influence residual tumor behavior. At the same time, adipogenic cells contribute to marrow repair, highlighting the need to distinguish regenerative from tumor-permissive states. Spatial heterogeneity across skeletal sites, and its evolution during treatment, adds an additional layer of complexity to how BMAT regulates tumor behavior in bone. Here, we review recent advances that are redefining the role of BMAT in cancer progression and therapy response. Introduction Bone marrow adipose tissue (BMAT) is a metabolically active and spatially heterogeneous component of the marrow microenvironment that influences hematopoiesis, bone remodeling, immune cell metabolism, and tumor behavior [1–5]. Its role as a metabolic fuel source and pro-survival niche for marrow-resident metastatic solid tumors and hematologic malignancies has become increasingly appreciated [2,4,6–8]. Over the past several years, accumulating evidence has broadened the recognized role of BMAT in cancer progression and therapy response. BMAT is now viewed as a stress-responsive and plastic niche that is dynamically remodeled by both tumor growth and associated treatment. This review highlights recent advances that redefine our understanding of the interplay between BMAT remodeling and tumor behavior within malignant bone microenvironments (Table 1). New insights into tumor-driven BMAT remodeling It is now widely accepted that BMAT remodeling by both the hematological malignancies and bone-metastatic tumors is reciprocal: tumor cells actively reprogram marrow adipocytes by inducing lipolysis, while adipocytes enhance lipid uptake and transport in tumor cells, collectively reshaping tumor metabolism [7,9]. Mechanistically, tumor-induced lipolysis is driven by a combination of soluble factors, stress signaling, and bidirectional tumor-adipocyte communication [10]. BMAT responses vary by skeletal site. Although marrow adipocytes are plastic and can undergo remodeling, delipidation, or functional reprogramming, as observed in leukemia [11,12], multiple myeloma [13], and bone metastasis [14], they are generally less susceptible to lipolysis than white adipocytes from peripheral adipose depots, and this resistance is not uniform [15]. In AML, regulated adipocytes in hematopoietic-rich regions (rBMAds) are depleted in response to leukemic cells, whereas constitutive BMAds (cBMAds) are comparatively resistant to delipidation, highlighting depotspecific BMAd plasticity [12]. These differences raise important questions about how tumors access lipid stores in more stable marrow niches. A recent important study from the Scheller laboratory suggests that the apparent lipolytic resistance of cBMAT is not permanent but rather reflects a homeostatic state [16]. Although sympathetic signaling has historically been viewed as the principal regulator of adipocyte lipolysis, this elegant study identifies adaptive, catecholamine-independent pathways capable of mobilizing lipids in response to systemic or local stress. They show that stable adipocyte depots, including cBMAT, can undergo ATGL-dependent lipolysis independent of local sympathetic input when nutrient and endocrine conditions shift, such as during hypoglycemia, reduced insulin signaling, or tumor-induced cachexia. These mechanisms may be particularly relevant in tumor-bearing marrow, where inflammatory cytokines, nutrient competition, and hypoxia could engage non-adrenergic lipolytic programs to sustain lipid mobilization. Notably, a direct transfer of lipolytic cAMP signaling from breast cancer cells to white adipocytes by GJB3 gap junctions has recently been reported [17], suggesting that contact-dependent mechanisms may further amplify lipid mobilization. Whether similar mechanisms contribute to lipid mobilization by tumors residing within the marrow niche remains to be determined. Emerging mechanisms shaping BMAT-tumor crosstalk A current view of the field is that metabolic coupling, adipokine signaling, and microenvironment-mediated resistance are central features of adipocyte—tumor interactions in bone [10]. Systemic states such as obesity, age, as well as conditions of energy deficit such as anorexia influence marrow adiposity [1], while adipokine-driven signaling links marrow fat to bone remodeling and tumor-supportive niche function [18–20]. This evolving perspective frames BMAT as a dynamic and state-dependent regulator of tumor behavior shaped by tumor-derived signals and host metabolism. Mechanistically, adipocyte-derived fatty acids remain front and center in supporting tumor survival, operating through multiple metabolic routes, including fatty-acid oxidation, membrane synthesis, lipid storage, and stress-protective remodeling [7]. Increasingly, however, lipid-handling pathways are being recognized not simply as passive fuel sources but as components of coordinated stress-adaptation programs that enable tumor persistence within adipocyte-rich marrow niches. Under hypoxic conditions, chronic myeloid leukemia cells were reported to increase glutamine-dependent CD36-mediated fatty acid uptake, a response associated with enhanced survival and maintenance of leukemic stem-cell populations in the adipocyte co-culture systems [21]. In multiple myeloma cells, inhibition of acyl-CoA synthetase long-chain (ACSL) enzymes was recently shown to suppress proliferation and induce mitochondrial dysfunction, highlighting the importance of converting fatty acids into metabolically useable acyl-CoA pools to sustain tumor fitness in marrow environments [22]. Complementing this work, an in vivo CRISPR activation screen identified acyl-CoA-binding protein (ACBP/DBI) as a driver of bone metastatic growth, linking acyl-CoA availability to mitochondrial function and tumor survival in the skeletal niche [23]. Furthermore, targeting fatty acid binding proteins (FABPs) was shown to disrupt MYC signaling and cell-cycle progression in pre-clinical models of multiple myeloma [24], reinforcing the role of lipid trafficking in coordinating proliferative and metabolic programs within adipocyte-rich niches [25,26]. Beyond lipid acquisition and trafficking, lipid-modifying enzymes that buffer metabolic stress are emerging as important regulators of tumor adaptation within marrow niches. In bone-metastatic prostate cancer models, inhibition of the ER-resident desaturase stearoyl-CoA desaturase (SCD) suppresses tumor progression in adipocyte-rich bone through mechanisms involving ER stress, mTOR dysregulation, and DNA-damage signaling [27]. This is consistent with recent work showing that high SCD expression is associated with adverse outcome and enrichment of unsaturated fatty acid biosynthesis in relapsed or chemo-refractory AML, and that pharmacologic SCD inhibition induces lipotoxicity and cell death across genetically diverse AML models [28]. Notably, even in models with limited single agent sensitivity, SCD inhibition increased chemotherapy-induced DNA damage and synergized with conventional cytotoxic therapy to prolong tumor cell survival in vivo, indicating that metabolic perturbation can prime leukemic cells for a second therapeutic hit. Together with observations in pre-clinical lung cancer bone metastasis models in which combined inhibition of fatty acid synthesis and chemotherapy restored chemosensitivity in adipocyte-rich bone [29], these findings suggest that targeting tumor lipid handling enhances therapeutic response across marrow malignancies. This may occur through limiting desaturation or de novo synthesis, thereby destabilizing adipocyte-supported metabolic buffering. Therapy-driven BMAT remodeling Therapy-associated increases in marrow adiposity have long been observed, but recent studies now clarify the mechanisms and functional consequences of these changes, positioning BMAT as an active component of therapy-remodeled marrow niches that impact tumor progression and therapeutic response [10]. Compelling recent work demonstrates that cytotoxic therapy can induce senescence preferentially within murine marrow adipogenic lineage populations, including CXCL12-abundant reticular (CAR) cells and bone marrow adipocytes [30]. These cells upregulate senescence associated secretory phenotype (SASP) factors, including RANKL, to drive osteoclastogenesis while simultaneously reducing osteoblastogenic capacity, thereby uncoupling bone remodeling. Importantly, targeting stress-activated MK2 signaling (a downstream effector of p38 MAPK that regulates inflammatory SASP phenotype) or pharmacologically clearing senescent adipo-lineage cells (e.g., using senolytics) mitigated bone loss without compromising chemotherapy efficacy. These findings indicate that therapy-associated BMAT remodeling is not a transient phenomenon but a persistent stromal response with functional consequences. In this context, therapy-induced senescence may create a stromal environment that supports survival of residual tumor cells during post-therapy adaptation. Radiation and chemotherapy can also bias mesenchymal stromal progenitors toward adipogenesis at the expense of osteoblastogenesis, contributing to sustained marrow adiposity and impaired bone regeneration. Although therapy-induced adipogenesis has been recognized for years [9,10,31,32], recent lineage-resolved studies in mice demonstrate that these adipogenic shifts arise from defined stromal populations and persist in marrow niches that are also exploited by malignant cells [33,34]. Importantly, adipogenic lineage expansion is not uniformly detrimental. Marrow adipogenic lineage precursors (MALPs) can support marrow recovery following chemotherapy [33], as these late-stage adipogenic progenitors resist early cytotoxic depletion, transiently expand, and adopt proliferative and myofibroblast-like states that promote vascular and stromal repair. Genetic ablation of MALPs impairs hematopoietic and microenvironmental recovery after treatment, and similar transient expansion and myofibroblast conversion of MALPs has been observed after radiation injury, where their ablation was shown to impair restoration of marrow cellularity and vasculature [34]. This duality suggests that therapeutic strategies targeting BMAT must consider timing and cellular state, as interventions that broadly suppress adipogenesis could impair marrow recovery while still leaving processes that support tumor cell survival intact. Together, these findings highlight BMAT as a niche that can be simultaneously reparative and permissive under conditions of therapeutic pressure. Other disease-relevant therapies further illustrate how treatment can reshape marrow niches. Bone-targeted agents such as radium-223, used in prostate cancer bone metastases, alter bone remodeling dynamics and have been associated with longer-term microenvironmental changes including increased marrow adiposity [35]. This suggests that therapies directed at skeletal lesions can also remodel stromal and adipogenic compartments in ways that influence residual tumor cells, immune populations, and subsequent treatment responsiveness. Endocrine therapies represent another major clinical axis of BMAT remodeling: androgen deprivation therapy increases marrow adiposity and suppresses bone formation [36], and estrogen depletion in breast cancer similarly favors adipogenic allocation and contributes to skeletal fragility [37–39]. Glucocorticoids, widely used in ALL, lymphoma, and myeloma regimens, are potent pharmacologic drivers of adipogenesis that promote adipocyte differentiation while suppressing osteoblast programs [40]. Because all of these agents are often administered repeatedly or chronically, their effects may accumulate and compound BMAT remodeling due to therapy, with potential consequences for marrow recovery, niche signaling, and tumor persistence. The clinical relevance of therapy-associated BMAT remodeling was recently reinforced by the phenome-wide association and Mendelian randomization analysis of UK Biobank participants [41], which demonstrated associations between cancer therapy exposure, including chemotherapy and radiation, and increased marrow fat fraction across skeletal sites. Notably, higher bone marrow fat fraction showed site-specific associations with malignancies including breast and prostate cancer bone metastases and multiple myeloma, aligning with preclinical evidence that marrow adipocytes can promote tumor cell survival and therapeutic resistance. These findings suggest that BMAT remodeling is a clinically measurable consequence of treatment exposure and a reflection of altered marrow niche function. Defining when such remodeling supports tissue repair versus tumor persistence will be critical for understanding how treatment shapes long-term disease outcomes and skeletal recovery. Emerging mechanisms responsible for persistence and therapy evasion As BMAT is remodeled by both tumor growth and treatment, adipocyte-rich marrow niches may become key determinants of how tumor cells survive therapy and persist. Single-cell analyses in murine myeloma models demonstrate that malignant cells induce inflammatory and oxidative stress programs across non-hematopoietic niche populations and bias mesenchymal differentiation toward adipogenic lineages that support tumor growth [42]. Therapy-associated BMAT changes may therefore build upon pre-existing stromal polarization induced by the tumor, rather than arising de novo. In this setting, marrow niches rich in adipocytes are likely to influence how tumor cells respond to therapy. Adipocytes can reduce effective drug exposure by metabolizing lipophilic agents such as daunorubicin, while simultaneously supplying fatty acids that sustain mitochondrial metabolism and redox balance during therapy-induced stress [43]. Tumor-induced adipocyte lipolysis may further promote resistance to agents such as docetaxel by increasing local lipid availability and reinforcing stress-adaptive metabolic programs [27,44,45]. These lipid-supported tumor survival pathways should be considered not only as drivers of proliferation but also as mechanisms that allow tumor cells to remain viable under therapeutic pressure without requiring active cell cycling. Consistent with this interpretation, recent work in human T-ALL samples and murine models reported enrichment of quiescent tumor populations within adipocyte-dense marrow regions [46,47], suggesting that such niches may support low-cycling states compatible with minimal residual disease. Bone marrow adipocytes have also been implicated as regulators of disseminated tumor cell (DTC) reactivation by providing free fatty acids that fuel mitochondrial metabolism and support proliferative outgrowth of previously dormant cells [48]. Marrow adipocyte interaction with other components of bone marrow niche, including osteoclast-mediated remodeling, loss of osteoblast-derived quiescence signals, and inflammatory cues, collectively shape dormancy escape [48]. This extends the role of BMAT beyond therapy tolerance, suggesting that the same metabolic and signaling interactions that support survival during treatment may also influence the timing and probability of dormancy escape. Therapy-driven BMAT remodeling may therefore shape not only whether tumor cells survive treatment but also how and when they re-enter the cell cycle. A key priority moving forward will be to define how lipid-rich marrow niches influence tumor cell tolerance to metabolic and oxidative stress, and how adaptation to stress induced by lipids supports survival of dormant populations versus their eventual reawakening. Biogeography of bone marrow niche - spatial tools, models and imaging approaches A major emerging direction in the BMAT-cancer field is the need to understand marrow niches in a spatially resolved and human-relevant context. Bone marrow adipose tissue is not uniformly distributed across the skeleton [49], and recent work suggests that tumor cell behavior and fate, including persistence, dormancy, and therapy responsiveness may depend on the specific microenvironments in which tumor cells reside [8,10]. Spatial and single-cell approaches are therefore becoming essential for defining the biogeography of marrow niches and for determining whether adipocyte-rich regions possess distinct immune, stromal, or vascular signatures that influence tumor fate. Recent mapping of human marrow niche architecture using single-cell transcriptomics and multiplexed imaging has begun to define location-specific niche states and their association with disease progression and therapy exposure [50]. These spatial analyses enable investigation of whether BMAT-rich regions harbor unique inflammatory, metabolic, or senescence-associated programs and whether these signatures correlate with quiescence or relapse risk. As spatially resolved datasets become more comprehensive, it should become possible to define how niche location influences tumor persistence and therapeutic response. Despite rapid mechanistic progress, translation of BMAT biology into clinically relevant systems remains a major challenge. While metabolic coupling and adipocyte-driven signaling are now well established, major limitations remain in existing model systems, particularly in recapitulating human marrow architecture, adipocyte heterogeneity, and vascular context [8]. Recent methodological advances have begun to address this gap. New three-dimensional adipose culture systems and extracellular matrix-mimicking hydrogels are emerging as experimental models for testing of adipocyte-tumor interactions in therapy-relevant settings, including endocrine and metabolic stress conditions [51]. Emerging human bone marrow organoid systems that incorporate mesenchymal, endothelial, and hematopoietic lineages, provide increasingly physiologic platforms for modeling tumor-niche interactions and therapy response [52]. However, current organoids largely lack adipocyte compartments, in part because adipogenic differentiation requires specific metabolic and mechanical conditions, mature adipocytes are fragile and difficult to maintain in long-term culture, and BMAT is tightly integrated with bone, hematopoietic, and vascular niche components [53]. Incorporating adipogenic lineages while preserving physiologic niche architecture therefore remains a key technical barrier for next generation BMAT-based models [53]. Continued development of physiologically faithful models will be essential for linking BMAT heterogeneity with clinical outcomes [4]. Advances in quantitative imaging are also beginning to bridge preclinical and clinical studies. Proton density fat fraction (PDFF) MRI and related techniques provide non-invasive methods to quantify marrow adiposity and detect dynamic changes associated with aging, radiation, and therapy exposure, including increases in marrow fat fraction and decreases during marrow regeneration or hematopoietic recovery [54]. Specifically, in preclinical mouse models, PDFF measurements correlate strongly with histologic adiposity and capture spatial and temporal variability in marrow fat content [54]. Ongoing clinical studies are now extending these approaches to patients, including longitudinal imaging of marrow fat fraction in individuals undergoing chemotherapy and transplantation (e.g., NCT03422731), supporting their translational relevance. In parallel, spatial transcriptomic approaches, applied in both preclinical models and human tissue samples, enable high-resolution mapping of stromal lineage heterogeneity within the marrow, including adipogenic progenitor populations that localize to distinct niches and support leukemic cell survival and chemoresistance in B-cell acute lymphoblastic leukemia [55]. Population-level and clinical imaging analyses further support the concept that marrow adiposity is a quantifiable and therapy-responsive phenotype [56,57], raising the possibility that BMAT could serve as a biomarker of marrow niche state and long-term treatment effects. Outlook and future directions Over the past several years, BMAT has shifted from being viewed primarily as a passive lipid reservoir to a dynamic, state-dependent component of the bone marrow niche that shapes tumor behavior, therapy response, and skeletal health (Figure 1). This shift highlights several outstanding questions. First, a clearer understanding of how tumor- and therapy-driven signals converge over time to remodel BMAT, and how long these changes persist, will be important for establishing how residual tumor cells adapt and survive in bone marrow niche. Second, because adipocyte-rich marrow niches appear to support tumor persistence by buffering metabolic and oxidative stress and stabilizing low-proliferative states, identifying the specific lipid and stress adaptation pathways that support tumor survival in these environments will be an important next step. Third, adipogenic lineage populations can also contribute to marrow repair. Distinguishing reparative from tumor-permissive BMAT states will therefore be critical when considering therapeutic intervention. Finally, because marrow adiposity varies across skeletal sites and niche states appear to differ in their capacity to support persistence or repair, defining when BMAT is tumor-permissive versus adaptive will be central to future work. Linking mechanistic insights from pre-clinical studies to patient-relevant systems will be essential for determining whether adipocyte-rich niches represent targetable vulnerabilities or reflect broader remodeling of bone marrow microenvironment that supports tumor progression and therapy evasion. In this context, the widespread clinical use of metabolic therapies such as GLP-1 receptor agonists underscores the need to understand how systemic metabolic interventions influence bone microenvironment and skeletal health [58]. Although current data primarily address skeletal and metabolic outcomes, their potential to impact BMAT remodeling or tumor—niche interactions is plausible and represents an important, largely unexplored area for future investigation.