Acute myeloid leukemia (AML) is the most aggressive form of hematologic malignancies in adults, driven by somatically acquired mutations in genes regulating myeloid lineage differentiation, proliferation, and survival.1 Among cytogenetically normal AML (CN-AML), which accounts for 40%–50% of cases, recurrent mutations in NPM1, FLT3, and DNMT3A are most common.2 Mutations in NPM1 nuclear localization signal disrupt its nuclear function, but on their own are usually insufficient to induce AML, and cooperating lesions such as FLT3-Internal tandem duplication (ITD) mutations are often required to promote leukemic transformation.3 FLT3-ITD mutations lead to constitutive kinase activation, resulting in increased blast survival and more aggressive disease progression.1, 3 Several hematologic malignancies show persistent inflammation arising from both leukemia-intrinsic and microenvironmental sources.4 Importantly, several AML-associated mutations have been linked not only to leukemogenesis but also to dysregulation of inflammatory pathways. Somatic mutations can activate innate immune pathways, such as TLR–NF-κB signaling, inflammasome activation, and type-I interferon responses, within malignant cells, while altered stromal and immune cells in the bone marrow further amplify these inflammatory circuits. Elevated plasma levels of proinflammatory cytokines such as IL-6, IL-1β, TNF-α, and IFN-γ, which are commonly observed in AML patients, contribute to leukemic survival and resistance to therapies.4, 5 Epidemiologic data show increased AML incidence in individuals with autoimmune conditions,6, 7 indicating that pre-existing inflammation may promote leukemogenesis. On the other hand, AML itself can act as a driver of systemic inflammation. While these findings highlight the importance of inflammation in AML, it remains unclear whether AML-associated inflammatory signals extend beyond the hematopoietic system to affect other organs, such as the central nervous system (CNS). This question is particularly relevant considering recent work showing that clonal hematopoiesis of indeterminate potential (CHIP) can drive systemic inflammation and influence neurological outcomes. For example, TET2-driven CHIP lowers Alzheimer's disease (AD) risk,8 and brain-infiltrating Tet2 mutant monocytic cells were protective in a mouse model of AD.9 In contrast, DNMT3A-mutant monocyte-derived microglia accumulate in nigrostriatal regions and induce Parkinson's disease-like motor deficits,10 demonstrating that premalignant hematopoietic clones can modulate neuroinflammatory processes and neurological function. In addition, hematologic therapies, such as involving CAR-T cells, can induce CNS inflammation and associated neurotoxicity, which arises from excessive cytokine release with endothelial activation and blood–brain barrier (BBB) disruption.11 Bleeding and coagulation abnormalities common in AML can also promote endothelial activation and BBB alterations, thereby contributing to inflammatory responses in the CNS. Together, these observations underscore the vulnerability of the CNS to inflammatory stimuli originating from hematopoietic cells or their therapies. Yet, whether inflammation driven by AML cells themselves contributes to CNS dysfunction has not been systematically investigated. This question is particularly relevant, given the emerging role of inflammation in neuropsychiatric and neurological disorders.12 Elevated levels of cytokines such as IL-1β, IL-6, and TNF-α have been associated with major depressive disorder and anxiety, and are thought to influence CNS function by altering neurotransmitter metabolism, synaptic plasticity, and behavior.13 Psychiatric disorders such as anxiety and depression affect approximately 20%–30% of patients with AML, a rate lower than that in patients with high-prevalence solid tumors but higher than that in most chronic diseases.14 Although the psychological impact of a life-threatening diagnosis is a clear risk factor for these conditions, it is still unknown whether the inflammatory milieu accompanying AML contributes biologically to these symptoms or to broader CNS dysfunction. To investigate this possibility, we used a murine model of NPMc⁺/Flt3-ITD AML, which rapidly develops leukemia and closely recapitulates human disease.15 Because previous studies have shown that altered NPM1 function can trigger aberrant inflammasome activation, while FLT3-ITD promotes cytokine network alterations and a proinflammatory bone-marrow microenvironment,16, 17 we hypothesized that AML-related inflammation caused by these leukemogenic mutations may also induce neuroinflammation. Therefore, we evaluated CNS inflammatory responses in this genetically defined model of AML. To this end, we first purified and quantified leukocytes from whole-brain homogenates of leukemic mice defined as such by elevated peripheral white blood cell (WBC) counts and reduced hemoglobin levels compared to age-matched healthy controls (Supporting Information S1: Tables 1 and 2). A significant increase in the total leukocyte count was detected in the brains of leukemic mice compared to controls (Figure 1A). Flow cytometric analysis revealed significantly increased numbers of CD4⁺ and CD8⁺ T cells, CD11b⁺CD45hi myeloid cells, and CD11b⁺CD45lo microglia in leukemic mice (Figure 1B,C). As normal monocytic differentiation is largely impaired in the NPMc⁺/Flt3-ITD model, the expanded CD11b⁺CD45hi population represents leukemic myeloid cells rather than bona fide monocytes. Consistently, CD11b⁺CD45hi cells from leukemic brains showed increased forward scatter, consistent with enlarged cell size typical of leukemic cells (Figure 1D), and expressed Mac-1 and Gr-1, but not the markers of leukemia stem cells (Supporting Information S1: Figure S1A–J). The presence of CNS-infiltrating leukemic blasts was confirmed by detection of the Npm1c transgene expression by qPCR (Figure 1E) and by immunofluorescence staining in spleen and brain parenchyma in NPMc⁺/Flt3-ITD but not in control animals (Figure 1F,G). A significant positive correlation between peripheral WBC counts and CNS infiltration was observed (Spearman test, P = 0.018) (Supporting Information S1: Figure S1K). These results suggest that leukemic progression is associated with increased immune and leukemic cell infiltration into the brain. Consistent with a systemic inflammatory state, mRNA expression of the proinflammatory cytokines Il1b, Il6, and Tnfa was increased in spleens from NPMc⁺/Flt3-ITD mice compared with controls (Figure 1H). The frequency of IL-6–producing CD11b⁺ cells was also increased in peripheral blood of NPMc+/Flt3-ITD as compared to control mice (Figure 1I). The blood–brain barrier (BBB) plays a crucial role in maintaining CNS homeostasis by regulating immune cell trafficking.18 Tight junctions (TJs), primarily composed of occludin and claudin-5, are essential for preserving BBB integrity by controlling intercellular permeability. Degradation of the TJ components, often mediated by matrix metalloproteinases (MMPs), compromises the BBB, increasing permeability and leukocyte access, thereby contributing to neuroinflammation and neurological disorders. Infections and inflammatory states are known to challenge BBB integrity.18 To assess BBB integrity in the brain of WT and leukemic mice, we evaluated the expression of the TJ component occludin via western blot analysis. Notably, occludin protein levels were significantly reduced in the brains of NPMc⁺/Flt3-ITD mice compared to controls (Figure 1J and Supporting Information S1: Figure S2A–C). We next performed functional Evans Blue (EB) permeability assays to validate BBB disruption. Both EB accumulation in the brain and BBB permeability (brain-to-blood EB concentration ratio), expressed as fold change relative to WT controls, were significantly increased in AML mice (Figure 1K). Circulating EB levels did not differ significantly between groups (Supporting Information S1: Figure S2D,E), indicating that the elevated brain EB is not due to differences in systemic dye availability but reflects a functional decrease in BBB integrity. To determine whether dysregulated MMP expression may contribute to TJ degradation, we measured mRNA levels of Mmp2 and Mmp9 by quantitative real-time PCR (qPCR). Our results showed a significant increase in Mmp9 expression levels in NPMc⁺/Flt3-ITD mice compared to controls, while Mmp2 levels did not differ significantly between the groups (Figure 1L). These findings suggest that MMP9-mediated TJ degradation may contribute to BBB disruption and facilitate leukocyte infiltration into the brain. To investigate whether the enhanced leukocyte infiltration contributes to the establishment of a neuroinflammatory microenvironment in the brains of the leukemic mice, we measured the mRNA expression levels of the expression of key inflammatory mediators in whole-brain tissue from WT and leukemic mice. Specifically, we measured mRNA levels of pro- and anti-inflammatory cytokines (Il1b, Tnfa, Il6, and Tgfb), chemokines involved in immune cell recruitment (Ccl2 and Ccl4), homeostatic and neuroprotective chemokines (Cx3cl1, Cxcl12, and Cxcl9), and type I interferons (Ifna and Ifnb). Leukemia-bearing mice showed significantly elevated expression of Il1b, Tnfa, Il6, Tgfb, Ccl2, and Ccl4, while Cx3cl1, Cxcl12, and Cxcl9 were markedly downregulated; however, expression of Ifna and Ifnb remained unchanged (Figure 2A). These data suggest a brain-restricted inflammatory response marked by immune cell recruitment, glial activation, and loss of homeostatic signals, rather than systemic immune activation. Collectively, these results indicate a selective upregulation of proinflammatory cytokines in the brains of leukemic mice, consistent with a brain-associated innate inflammatory response. To determine whether glial activation accompanies this inflammatory state, we examined astroglial and microglial activation in the cortex and hippocampus of WT and leukemia-bearing mice by immunofluorescence staining of brain sections with glial fibrillary acidic protein (GFAP), ionized calcium-binding adaptor molecule 1 (Iba1), and DAPI to label astrocytes (green), microglia (red), and nuclei (blue), respectively, as we previously described.19 The number of GFAP-positive astrocytes (asterisks) and Iba1-positive microglia cells (arrows) was significantly increased in cortex sections of NPMc+/Flt3-ITD mice compared to controls (Figure 2B,C and data not shown), suggesting reactive astrogliosis and microglia activation. Similar glial alterations were observed in the hippocampus. Low-magnification images show that the hippocampal architecture was preserved in both groups (Figure 2D,E), while high-magnification images of the dentate gyrus (DG) revealed increased numbers of GFAP-positive astrocytes and Iba1-positive microglia in NPMc⁺/Flt3-ITD mice compared to controls (Figure 2F,G and data not shown). The reactive status of astrocytes in leukemic mice was characterized by enhanced expression of GFAP and more ramified morphology (arrows). Objective quantification of images corroborated the histological findings, revealing a significant increase in the proportion of Iba1⁺ microglia cells (Figure 2H) and GFAP expression (Figure 2I) in both the cortex and the hippocampus of NPMc⁺/Flt3-ITD mice compared to controls. To investigate the phenotypic features associated with microglial activation, we performed flow cytometric analysis of CD11b⁺CD45lo microglia isolated from the brains of WT and NPMc⁺/Flt3-ITD mice. Microglia from leukemic mice showed a significant increase in the forward scatter area (FSC-A), suggesting a hypertrophic, activated state (Figure 2J). In addition, the expression of FcγRIII/II (CD16/32), a marker associated with proinflammatory microglial activation,20 was significantly upregulated in microglia from NPMc⁺/Flt3-ITD mice compared to controls (Figure 2K). Taken together, these data suggest that leukemic mice show a glial phenotype indicative of ongoing neuroinflammation. In addition, to explore whether human AML cells may directly lead to microglial activation, we used a co-culture system using HMC3 human microglia and OCI-AML3 AML cell lines (carrying NPM1 mutation A and WT FLT3). The two cell lines were co-cultured either in direct cell-to-cell contact or in an indirect trans-well setting, allowing for the exchange of soluble factors without direct cell contact. Consistent with microglial activation, AML-conditioned media induced IL-1β and IL-6 production (Figure 2L–N) and increased expression of CD11b, CD14, and CD80 in HMC3 cells (Figure 2O–Q, Supporting Information S1: Figure S3A–C), whereas CD86 expression remained unchanged (Supporting Information S1: Figure S3D). Moreover, morphological assessment of HMC3 microglia cells exposed to OCI-AML3- or MOLM-13 (NPM1 WT and FLT3-ITD-positive)-conditioned media showed marked morphological transformation characterized by a more rounded shape, consistent with the activated state (Figure 2R). Thus, the data obtained in vitro using human cell lines further suggest that AML cells may cause microglial activation via secretion of soluble factors. To extend our in vitro observations to the clinical setting, we assessed whether MMP9 expression is also increased in human AML patients. For this purpose, we compared the MMP9 mRNA expression in peripheral blood mononuclear cells isolated from AML patients and healthy donors. Consistent with murine data, MMP9 expression was significantly elevated in AML samples compared to controls (Figure 2S). Although differences in MMP9 protein levels or activity await confirmation, the upregulation of MMP9 mRNA both in mice and in human AML suggests that it may play a role in BBB disruption. The observed upregulation of MMP9 transcription supports our hypothesis and strengthens the translational relevance of our murine findings. Finally, to determine whether the AML-associated inflammatory responses may contribute to CNS alterations in AML patients, we compared levels of known neuronal injury markers in plasma of AML patients and healthy controls using multiplex technology. We detected significantly elevated levels of neuron-specific enolase (NSE) and neurofilament heavy chain (NF-H) in plasma of AML patients compared to those of healthy controls (Figure 2T and Supporting Information S1: Figure S4A). Importantly, mRNA expression of these neuronal markers did not significantly differ in PBMC from AML patients and healthy controls (Supporting Information S1: Figure S4B). The increase in plasma levels of two independent protein markers of neuronal damage suggests the presence of a possible AML-associated neuroinflammation in human AML. Our findings provide evidence that AML progression is associated with neuroinflammatory changes characterized by BBB disruption, increased leukocyte infiltration, glial activation, and elevated levels of pro-inflammatory cytokines. In addition, both in vitro data and patient-derived evidence support the potential involvement of AML in the onset of neuroinflammation and its consequent neurological manifestations. Understanding how systemic malignancies like AML induce brain-specific inflammatory responses may provide new avenues for neuroprotective strategies aimed at improving cognitive outcomes and the overall quality of life in leukemia patients. The selective cytokine profile points to specific inflammatory pathways that might be therapeutically targeted to mitigate CNS involvement in leukemia. Future studies should explore whether modulating neuroinflammation can alleviate these neurological effects and examine the long-term impact of leukemia-induced glial activation on brain structure and function. Open access publishing facilitated by Universita degli Studi di Perugia, as part of the Wiley - CRUI-CARE agreement. Marta Febo: Investigation; writing—original draft; writing—review and editing; data curation; methodology. Daniele Sorcini: Investigation; resources; methodology; writing—review and editing; data curation. Alessandra Mirarchi: Investigation; methodology. Paolo Cogliati: Investigation; methodology; validation. Arianna Stella: Investigation; writing—review and editing; methodology. Paolo Sportoletti: Writing—review and editing; methodology; resources. Olga Ermakova: Writing—review and editing; methodology; formal analysis. Martina Mandarano: Investigation; methodology. Maria Cristina Marchetti: Investigation; methodology; data curation. Carlo Riccardi: Writing—review and editing; formal analysis. Brunangelo Falini: Methodology; writing—review and editing; resources; project administration. Graziella Migliorati: Writing—review and editing; project administration; supervision. Cataldo Arcuri: Investigation; conceptualization; writing—review and editing; methodology; data curation; formal analysis. Stefano Bruscoli: Investigation; writing—review and editing; methodology; formal analysis; data curation; supervision; funding acquisition. Oxana Bereshchenko: Investigation; conceptualization; writing—original draft; writing—review and editing; validation; formal analysis; project administration; data curation; supervision; methodology. The authors declare no conflict of interest. This research was funded by the Italian Ministry of University and Research grant (PRIN-202039WMFP) to S.B. The data that support the findings of this study are available from the corresponding author upon reasonable request. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Febo et al. (Mon,) studied this question.