Glioblastoma is characterized by diffuse brain invasion, yet the subcellular mechanisms enabling this aggressive behavior remain poorly understood. A subpopulation of glioblastoma cells forms invasive tumor microtubes (TMs), neurite-like extensions that drive whole-brain colonization. Here, we establish local protein translation as a fundamental driver of TM dynamics and invasive cell states. Developing a subcellular transcriptomics approach - integrating subcellular organelle organization with spatially resolved transcriptomics and functional readouts - we reveal that TM gene expression drives cell state identity. Invasive cells further exhibit significantly elevated local translation in protruding TMs, directly linking subcellular protein synthesis to functional invasive states associated with neurodevelopmental programs of axonal growth cones. Targeted disruption of TM-localized translation via photoswitchable puromycin, and specific knockdowns of the TM-enriched proteins GPM6A and GAP43, impaired TM dynamics, suppressed invasion, and reduced tumor growth. Together, these findings define local translation as a key determinant of tumor heterogeneity and glioblastoma invasion.

Background: Supramaximal resection beyond the contrast-enhancing margin is associated with prolonged survival in IDH-wild type glioblastoma, but whether this benefit is uniform is unknown, and no biomarker identifies responders. We tested whether on a neural methylation-based signature (neural-high vs neural-low) determines benefit and can be obtained intraoperatively by nanopore sequencing for surgical stratification. Methods:In discovery and validation cohorts of newly diagnosed IDH-wildtype glioblastoma treated with chemoradiotherapy, Cox models tested the interaction between RANO class 1 vs 2-3 resection and neural-high vs neural-low tumours for overall survival (primary) and progression-free survival (secondary); sensitivity analyses included residual-volume modelling. In the prospective EpiGuide study, patients undergoing glioblastoma surgery had intraoperative nanopore sequencing with real-time neural-class prediction compared with matched EPIC array classification; the primary endpoint was a valid neural class within a prespecified 30-min on-flow-cell window. Findings: The retrospective analysis included 247 (discovery) and 303 (validation) patients. RANO class interacted with neural score in both cohorts (p=0·02 discovery, p=0·20 validation): supramaximal resection benefited neural-high tumours (adjusted HR 0·35 [95% CI 0·18-0·67] discovery; 0·46 [0·29-0·74] validation) but had an inconsistent effect in neural-low tumours (1·22 [0·53-2·82]; 0·75 [0·46-1·21]); sensitivity analyses were consistent. In EpiGuide, 47 of 67 enrolled had paired nanopore and array data; intraoperative classification agreed with the array in 39 (83·0%) of 47 (AUC 0·89; sensitivity 79·3%, specificity 88·9%). Presequencing took 42·4 min (SD 12·43); median on-flow-cell time to a confident, correct call was 11 min (IQR 6-20), and 34 (72·3%) reached this within the window.Interpretation. The neural signature identified IDH-wildtype glioblastomas in which RANO class 1 resection was associated with longer survival, with no reproducible benefit in neural-low tumours. Intraoperative nanopore sequencing enabled neural-class prediction within a feasible surgical timeframe, concordant with the methylation-array reference. These findings support prospective evaluation of biomarker-stratified extent-of-resection strategies but do not yet establish clinical utility for routine surgery.Funding. German Research Foundation (DFG), Else-Kröner-Frisenius Foundation, Illumina, Fördergemeinschaft Kinderkrebszentrum Hamburg, TRANSCAN (BMBF).Registration.This prospective, non-interventional, blinded intraoperative feasibility and concordance study did not require registration.

The aim of this study was to characterize the anatomical and imaging features of adult-type diffuse gliomas across histomolecular subtypes. Clinical 3-Tesla brain MRI images from 644 patients with pathologically confirmed adult-diffuse glioma before treatment was retrospectively evaluated: 527 IDH-wildtype glioblastoma, 71 astrocytoma, and 46 oligodendroglioma. Pre- and post-contrast T1-weighted, T2-weighted and FLAIR sequences were part of the MRI protocol. Contrast-enhancing tumors and non-enhancing lesions (NEL) were automatically segmented using HD-GLIO. We used a voxel-wise Fisher-exact-test followed by random-permutation (ADIFFI) to identify regions with higher occurrence of tumor associated with IDH-mutation status or 1p/19q-codeletion status. Mann-Whitney-U-test was used to compare signal intensities in CET and NEL across the three different subtypes of adult-diffuse glioma investigated here. We observed a significant correlation of IDH-mutant gliomas with a predominance in the frontal lobe adjacent to the rostral extension of the lateral ventricles. IDH-wildtype tumors had larger NEL volumes than IDH-mutant gliomas (p < 0.0001). Signal intensity analysis demonstrated consistently lower T1w and T1-CE values and higher T2w and FLAIR values in IDH-mutant gliomas (all p < 0.0001). Oligodendrogliomas showed higher signal intensity on T1-CE images compared to astrocytomas (p = 0.035). We analyzed a large radio-genomic patient cohort consisting of glioblastoma, astrocytoma and oligodendroglioma. Our findings are in line with previously analyzed smaller patient cohorts. Our findings underline the importance of the IDH-mutation for determining tumor location and potentially point to a cell of origin along the rostral extension of the lateral ventricles. Adult-type diffuse gliomas with IDH-mutation showed a predominance in the frontal lobe adjacent to the rostral extension of the lateral ventricles. Signal intensity analysis demonstrated consistently lower T1w and T1-CE values and higher T2w and FLAIR values in IDH-mutant gliomas. Oligodendrogliomas showed higher mean signal intensity of NEL on T1-CE images than astrocytomas.

Importance Glioblastoma (GBM) cells integrate into neuronal circuits, and preclinical work implicates multiple neurotransmitter (NT) networks as key drivers of invasion and treatment resistance. Whether the integration of GBM within NT-defined large-scale brain networks conveys prognostic information for overall survival (OS) is unknown. Objective To determine whether NT-specific network involvement of GBM is associated with OS in patients with newly diagnosed Isocitrate dehydrogenase (IDH)-wildtype(wt) GBM. Design, Setting, and Participants In this observational multicenter cohort study, we analyzed two independent cohorts of adults with histopathologically confirmed IDH-wt GBM. Cohort 1 included 153 patients treated at the University Medical Center Hamburg-Eppendorf, Germany (2012-2024), and cohort 2 comprised 264 patients from the University of Pennsylvania Health System, USA (2006-2018). Preoperative contrast-enhanced MRI was used to derive individual tumor masks, which were spatially mapped onto normative NT-informed structural connectomes spanning 19 receptor and transporter systems. Exposures Preoperative contrast-enhancing GBM lesions, quantified as patient-specific involvement scores (0-1) within each NT-defined brain network. Statistics We used partial least-squares regression for variable selection and multivariable Cox proportional-hazards models alongside regularized logistic regression with out-of-sample prediction, adjusted for age, methylguanine methyltransferase ( MGMT) promoter methylation, and extent of resection, to test associations between NT-specific GBM network involvement and OS. Results Across 417 patients in two cohorts, greater GBM involvement within cholinergic networks, defined by normative vesicular acetylcholine transporter (VAChT)-weighted as well as dopaminergic D2 receptor involvement, was consistently associated with reduced OS, independent of age, MGMT status, and resection extent. Further, cholinergic network involvement showed the strongest contribution to the prediction models. Other NT networks did not show reproducible prognostic effects across cohorts. Tumor-intrinsic hypomethylation of acetylcholine receptor-associated regions correlated with imaging-based cholinergic network involvement and mirrored its prognostic relevance. Conclusion and Relevance Tumor integration into neurotransmitter-specific brain networks is an independent predictor of poorer survival in GBM. By combining routine clinical MRI with normative NT-informed connectome data, this approach delineates a novel systems-level marker of tumor aggressiveness and supports cholinergic inhibition as a putative therapeutic target in GBM.

Glioblastoma (GB) integrates into the human brain by exploiting spatially restricted neuronal microenvironments at tumor margins. Combining magnetoencephalography-based functional mapping in patients with GB, spatially resolved biopsies, single-cell RNA sequencing, spatial transcriptomics, and electrophysiological profiling in human cortical slice models, we identify peritumoral connectivity hubs enriched for synaptogenic and immature neuronal programs. Connectivity-rich regions preferentially involve projection-capable excitatory neurons and display convergent activation of the WNK-SPAK-NKCC1 chloride homeostasis, consistent with a reduction in inhibition via GABAergic transmission in peritumoral neurons. GB-conditioned neurons exhibit developmental regression, impaired inhibitory control, and immature dendritic spine remodeling while network-level recordings reveal a GABA-sensitive reorganization of circuit topology. NKCC1 blockade reduces tumor-neuron synaptic integration and dampens functional network connectivity. Clinically, peritumoral degree centrality is an independent predictor of survival beyond established covariates. These findings establish neuronal immaturity as a functional substrate for tumor-neuron connectivity and define actionable biomarkers and pathways at the tumor-brain interface.

Rationale:Magnetic resonance imaging (MRI) is essential for visualizing the healthy and diseased brain, yet the cellular basis of MRI signal and how it changes over time remain poorly understood. Methods:Here, we present BRIDGE (Brain Radiological Imaging with Deep-learning based Ground-Truth Exploration), a platform integrating in vivo MRI with in vivo two-photon (2P) and ex vivo super-resolution microscopy using a multi-step, iterative co-registration pipeline. It enables in vivo, longitudinal and voxel-precise mapping of MRI signals to their biological ground truth for the first time. The registered overlay reveals the cellular and anatomical origins of MRI signals and enables training of convolutional neural networks to enhance the effective resolution of MRI. Results:Using BRIDGE, we identified a microenvironmental vessel biomarker for early metastatic colonization in patient-derived xenograft models of breast cancer brain metastasis. In particular, we found that distinct T2*-weighted hypointense lesions correspond to reduced blood flow and erythrostasis in perimetastatic capillaries. In glioma, longitudinal intravital studies further demonstrated direct correlations between non-vasogenic T2-weighted signal changes and patient-dependent tumor growth dynamics. Conclusions:Taken together, BRIDGE advances radiological interpretation by establishing a microscopic ground truth for MRI signatures over time, enabling deep learning-based predictive histology and providing cellular level insights into tumor microenvironment with direct clinical imaging implications.

The Takeda Innovators in Science Award with Nature supports groundbreaking research by early-career scientists in neuroscience. Varun Venkataramani was shortlisted for the award in 2026. We spoke to him about his research and the growing field of cancer neuroscience.

Both the nervous system and cancer-intrinsic neural features can govern cancer initiation, growth, progression, metastasis, and treatment resistance, while cancer can likewise influence the nervous system, promoting neural reprogramming and neuropsychiatric symptoms that worsen patient outcomes. The field of cancer neuroscience seeks to unravel this complex neuro-cancer crosstalk and holds the promise to develop neuroscience-instructed cancer therapies that improve disease control and quality of life. Here, we summarize the key discoveries of neuro-cancer crosstalk to date, including neuron-to-cancer synapses and paracrine and neuro-immuno-oncological interactions, and then explore emerging topics such as downstream effects on cancer cell pathophysiology, circadian influences, brain-body-cancer communication, and neural regulation of the metastatic cascade and the tumor microenvironment. Finally, we distill overarching principles, highlight relevant ongoing research, and outline conclusions to guide the development of cancer neuroscience, proposing hypotheses for future experimental validation.

Supratentorial ependymomas are aggressive childhood brain cancers that retain features of neurodevelopmental cell types1 and segregate into molecularly and clinically distinct subgroups2,3, suggesting different developmental roots. The developmental signatures, as well as microenvironmental factors, underlying aberrant cellular transformation and behaviour across each supratentorial ependymoma subgroup are unclear. Here we integrated single-cell and spatial transcriptomics, as well as in vitro and in vivo live-cell imaging, to define supratentorial ependymoma cell states, spatial organization and dynamic behaviour within the neural microenvironment. We find that individual tumour subgroups have two distinct progenitor-like cell states-neuroepithelial-like and embryonic-like-that are reminiscent of early human brain development and diverge in the extent of their neuronal or ependymal differentiation. We further identify several modes of spatial organization of these tumours, including a high-order architecture that is influenced by mesenchymal and hypoxia signatures, and local neighbourhood structures. Finally, we identify a role for brain-resident cells in shifting supratentorial ependymoma cellular heterogeneity towards neuronal-like cells that co-opt immature neuronal morphology and migratory mechanisms, and a subset of neuroepithelial-like cells that are both proliferative and highly migratory. Collectively, these findings provide a multidimensional framework to integrate transcriptional and phenotypic characterization of tumour heterogeneity in supratentorial ependymoma and its potential clinical implications.

Communication in multicellular networks is a cancer-intrinsic neural feature and crucial for primary brain tumor growth and resistance, but it is unclear whether brain metastases (BrM), the most common and deadliest brain malignancy, are also driven by communicating cancer networks. Using intravital two-photon microscopy in awake mice, clinical specimens, and Ca 2+ integrators, we demonstrate that brain-colonizing breast and lung cancer and melanoma cells display gap-junction-dependent, coordinated Ca 2+ activity in multicellular, cancer-cell intrinsic networks, which drives their proliferation. Mechanistically, Ca 2+ oscillations induce transcription of immediate early genes, adoption of a neuronal expression profile, and cell cycle progression. While many of those features are enriched in BrM, all investigated cancer cell lines showed collective Ca 2+ activity. Therapeutically, blocking Ca 2+ activity with gap junction inhibitors reduces BrM burden in mouse models. Here we show communicating cancer cell syncytia as drivers of BrM growth, pointing to a targetable pathomechanism, and potentially a new pan-cancer hallmark.

BACKGROUND:Scarce T cell infiltration, immunosuppressive tumor-associated macrophages, and ineffective drug delivery drive glioma progression and limit treatment efficacy. Mapping immunotherapy distribution by multimodality imaging might be a biomarker that could aid tumor monitoring and guide therapy development. METHODS:To assess drug delivery, we developed a MRI-lightsheet microscopy platform (MR-LSM) to monitor immunotherapy at the cellular level in two immunocompetent glioma models (Gl261, SB28). The atezolizumab (PD-L1 inhibitor) subgroup of the multicenter N2M2/NOA20 trial in MGMT unmethylated GBM patients was assessed by CNN analysis and correlated to progression-free survival. RESULTS:In contrast to the conventional Gl261 glioma model, SB28 gliomas are characterized by poor immunogenicity and resistance to Toll-like receptor (TLR) 7 targeted therapy delivered by CDNP-R848 nanoparticles. SB28 resistance is driven by microvascular pathology, vasogenic edema, and drug off-targeting to peritumoral edema and white matter tracts. Vascular endothelial growth factor (VEGF) inhibition in conjunction with irradiation and dual immunotherapy (DIR) targeting innate (CDNP-R848) and adaptive immunity (anti-CTLA-4) breaks resistance, increases survival, and reverses drug off-targeting. Mechanistically, tumor control is orchestrated by vascular normalization, enhanced CD8+ T cell influx, and a proinflammatory shift of myeloid cells along with strong IL-12/IL-13 upregulation. In a translational analysis of the multicenter N2M2/NOA20 trial, we validate that edema and microvascular pathology are associated with poor prognosis in glioblastoma patients treated with checkpoint immunotherapy and that patients without edema have increased PFS. CONCLUSIONS:. We develop a customizable imaging platform to map drug delivery to glioma with broad applicability in neuroscience and neuro-oncology.

Glioblastoma (GBM) is notoriously resistant to treatment. Scarce T cell infiltration, immunosuppressive tumor-associated macrophages (TAMs) and ineffective drug delivery drive tumor progression. To overcome resistance, suitable imaging biomarkers that guide therapy development are essential to ultimately improve outcomes. We find that in contrast to the susceptible Gl261 glioma model, SB28 gliomas are characterized by poor immunogenicity due to low MHC expression and resistance to Toll-like receptor (TLR) 7 targeted therapy by CDNP-R848 nanoparticles. SB28 resistance is driven by strong microvascular pathology, vasogenic edema and drug off-targeting to tumor adjacent white matter tracts. To tackle therapeutic resistance and map drug delivery to the TME, we developed a 3D MRI-lightsheet microscopy platform (MR-LSM) to monitor immunotherapy distribution at the cellular level. Using this platform we find that vascular endothelial growth factor (VEGF) inhibition in conjunction with irradiation and dual immunotherapy (DIR) targeting innate (CDNP-R848) and adaptive immunity (anti-CTLA-4) breaks resistance, increases survival and reverses off-targeting of immunotherapies. Mechanistically, tumor control is orchestrated by vascular normalization, enhanced cytotoxic CD8+ T cell influx and a proinflammatory shift of myeloid cells along with strong IL-12 /IL-13 upregulation and normalization of drug delivery. In a translational analysis of the multicenter N2M2/NOA20 trial we validate that edema and microvascular pathology are also associated with poor prognosis in glioblastoma patients treated with anti-PD-L1 immunotherapy and that patients without edema have increased progression free survival (PFS). In summary, we develop a customizable imaging platform (3D-MR-LSM) to three-dimensionally map drug delivery to the central nervous system (CNS) with broad applicability in neuroscience and oncology.

Deepening our understanding of neuro-cancer interactions can innovate brain tumor treatment. This mini review unfolds the most relevant and recent insights into the neural mechanisms contributing to brain tumor initiation, progression, and resistance, including synaptic connections between neurons and cancer cells, paracrine neuro-cancer signaling, and cancer cells' intrinsic neural properties. We explain the basic and clinical-translational relevance of these findings, identify unresolved questions and particularly interesting future research avenues, such as central nervous system neuro-immunooncology, and discuss the potential transferability to extracranial cancers. Lastly, we conceptualize ways toward clinical trials and develop a roadmap toward neuroscience-instructed brain tumor therapies. Significance: Neural influences on brain tumors drive their growth and invasion. Herein, we develop a roadmap to use these fundamentally new insights into brain tumor biology for improved outcomes.

Purpose:Identifying radiomics features that help predict whether glioblastoma patients are prone to developing epilepsy may contribute to an improvement of preventive treatment and a better understanding of the underlying pathophysiology. Materials and methods:In this retrospective study, 3-T MRI data of 451 pretreatment glioblastoma patients (mean age: 61.2 ± 11.8 years; 268 men, 183 women) were analyzed. Three hundred thirty-six patients reported no epilepsy, while 115 patients were diagnosed with symptomatic epilepsy. A total of 1,546 radiomics features were extracted from contrast-enhancing tumor, peritumoral regions, and normal-appearing white matter as regions of interest using PyRadiomics. The dataset was initially split into a training (70%) and a validation (30%) cohort. The training cohort was used for feature selection with ElasticNet and model optimization. Various machine learning models, including logistic regression (LR), were used to predict epilepsy status. The models' performances were evaluated with the validation cohort, and the area under the curve of the receiver operating characteristics (AUC) was used as a measure. For identifying relevant features, permutation feature importance was applied. Results:The performance of LR using radiomics features from only a single ROI in the validation cohort was AUC = 0.83 (95% CI: 0.76-0.91) and AUC = 0.77 (95% CI: 0.69-0.85) for the peritumoral and white matter regions, respectively. The most important features in peritumoral regions were shape features, while for the white matter region, higher-order features from FLAIR were most relevant. Conclusion:Radiomics features from peritumoral and normal-appearing white matter can be associated with epilepsy status at diagnosis, suggesting an important role of these regions for the development of epilepsy in glioblastoma patients.

Tumor cell networks formed by tumor microtubes (TMs) may play a key role in the development of therapy resistance in glioblastoma (GB). TM-mediated detoxification from radiation-induced reactive oxygen species (ROS) may infer radioresistance. We hypothesize that high linear energy transfer (LET) radiation, which describes the amount of energy deposited by radiation per unit length, interacts directly with the DNA backbone to induce complex lesions and thus might be less dependent on TM-mediated resistance mechanisms. Therefore, we sought to systematically investigate the impact of LET-induced complex DNA damage on TM formation and GB survival. To this end, the formation of TMs, radiation-induced nuclear DNA damage repair foci (RIF), and GB survival were correlated with a gradual increase in LET using a dose series of clinical protons (low), helium (intermediate), and carbon (high) ion beams. Consistent with conventional photon/X-rays, low-LET proton irradiation promoted TM formation in a dose-dependent manner. In contrast, an anti-correlation between LET and TM induction was found, i.e., a decreased network connectivity with gradual increase of LET and formation of complex DNA damage. Consequently, LET increase correlated with reduced cell survival, with the most pronounced cell killing observed after high-LET carbon irradiation. Moreover, the inverse correlation between LET and TM density was further confirmed for a broad range of LET modulated within the carbon ion irradiation spectrum. This is the first report on the relevance of LET as a novel mean to overcome TM network-mediated radioresistance in GB, with ramifications for the clinical translation of high-LET particle radiotherapy to further improve outcome in this still devastating disease. ### Competing Interest Statement JD reports grants from CRI The Clinical Research Institue GmbH grants from View Ray Inc., grants from Accuray International Sarl, grants from Accuray Incorposrated, grants from RaySearch Laboratories AB, grants from Vision RT limited, grants from Merck Serono GmbH, grants from Astellas Pharma GmbH, grants from Astra Zeneca GmbH, grants from Siemens Healthcare GmbH, grants from Merck KGaA Accounts Payable, grants from Solution Akademie GmbH, grants from Ergomed PLC Surrey Research Park, grants from Siemens Healthcare GmbH, grants from Quintiles GmbH, grants from Pharmaceutecal Research Associates GmbH, grants from Boehringer Ingelheim Pharma GmbH Co, grants from PTW-Freiburg Dr. Pychlau GmbH. AA report grants and other from Merck and EMD, grants and other from Fibrogen, other from BMS, other from BioMedX, other from Roche, outside the submitted work. All other authors declare no competing interests.

Brain-invasion of meningioma is a hallmark of malignant behavior and linked to higher risk of surgical resection and poor clinical outcome. Here, we hypothesized that meningioma tumor ecosystems with invasive behavior enhance cellular communication with residual brain cells allowing an integration of tumor cells into neuronal circuits. To this end, we profiled 42 tumors with and without brain invasive pattern leveraging single cell (Xenium) and array-based (Visium) spatially resolved transcriptomics followed by a spatial graph attention model (sGAT) trained to identify the distance of tumor ecosystems to the brain-tumor boarder. Leveraging explainable AI techniques (integrated gradients and attention) we identified enhanced SPP1+ myeloid cells enriched at the infiltrative boarder. Cellular neighborhood analysis identified enriched proximity of pro-inflammatory myeloid cells, activated astrocytes and excitatory neurons within a distance of 50-100µm to the infiltrative boarder followed by OPC- and oligodendrocytes enhancement peaking at at 200µm distance. To further investigate the tumor-neuronal interaction we leveraged tumor neuronal co-culture systems with genetically diverse meningioma cell models followed by rabies-ΔG-GFP retrograde tracing model demonstrating high degree of neuronal connectivity after 5 days co-culture. To confirm electrically synaptic transmission, patch-clamp and ca2+ imaging revealed functional synaptic contact predominantly activated by acetylcholine (puff). Cross-validation with spatial and single cell transcriptomic data revealed a meningioma neuron-like cell population enriched for acetylcholine receptor expression and localized mainly in the invasive front. Using super-resolution microscopy in brain-invasive meningioma specimens, we demonstrated putative neuron-to-tumor synapses at brain-tumor interfaces. In summary, invasive meningioma are marked by inflammatory reshape activating microglia and astrocytes and accumulation neuronal input on meningioma tumor cells.