Neuro-Oncology Advances

BackgroundPaediatric and adolescent gliomas and glioneuronal tumours remain challenging to assess non-invasively. Amide proton transfer (APT) chemical exchange saturation transfer (CEST) MRI has shown promise in adult gliomas but has not been well studied in younger patients. PurposeTo assess whether APT CEST signal can act as a non-invasive surrogate of tumour proliferation in adolescent CNS tumours by correlating it with 18F-choline PET uptake (SUV) as a proxy for membrane synthesis / proliferative activity. MethodsTen adolescent patients (14-19 yrs) with confirmed or suspected gliomas / glioneuronal tumours underwent simultaneous APT CEST and 18F-choline PET-MRI. Regions of interest (ROIs) corresponding to non-enhancing, enhancing, necrotic tumour, and contralateral white matter were delineated. Mean APT signal intensity (SI) and PET SUV were extracted per ROI. Nonparametric statistics and Spearmans correlation analyses were performed. ResultsAPT SI was significantly elevated in enhancing, non-enhancing, and necrotic tumour ROIs compared to normal white matter (p<0.001). 18F-choline SUV was elevated in enhancing and necrotic ROIs vs white matter, but not significantly so for non-enhancing tumour (p=0.02). A strong correlation between whole-tumour APT SI and 18F-choline SUV was seen (Spearman {rho}=0.86, p<0.001). ConclusionOur results indicate that APT CEST is feasible in adolescents and may reflect proliferative tumour burden. The detection of elevated APT SI even in non-enhancing tumour regions suggests potential utility in monitoring non-contrast-enhancing disease. Larger cohorts and multimodal correlation (e.g. Ki-67, amino acid PET) are warranted to confirm and extend these findings.

BackgroundThis study identified a clinically significant subset of glioma patients with tumor outside of contrast-enhancement present at autopsy, and subsequently developed a method for detecting non-enhancing tumor using radio-pathomic mapping. We tested the hypothesis that autopsy-based radio-pathomic tumor probability maps would be able to non-invasively identify areas of infiltrative tumor beyond traditional imaging signatures. MethodsA total of 159 tissue samples from 65 subjects were aligned to MRI acquired nearest to death for this study. Demographic and survival characteristics for patients with and without tumor beyond the contrast-enhancing margin were computed. An ensemble algorithm was used to predict pixelwise tumor presence from pathological annotations using segmented cellularity (Cell), extracellular fluid (ECF), and cytoplasm (Cyt) density as input (6 train/3 test subjects). A second level of ensemble algorithms were used to predict voxel-wise Cell, ECF, and Cyt on the full dataset (43 train/22 test subjects) using 5-by-5 voxel tiles from T1, T1+C, FLAIR, and ADC as input. The models were then combined to generate non-invasive whole brain maps of tumor probability. ResultsTumor outside of contrast was identified in 41.5 percent of patients, who showed worse survival outcomes (HR=3.90, p<0.001). Tumor probability maps reliably tracked non-enhancing tumor in the test set, external data collected pre-surgery, and longitudinal data to identify treatment-related changes and anticipate recurrence. ConclusionsThis study developed a multi-1 stage model for mapping gliomas using autopsy tissue samples as ground truth, which was able to identify regions of tumor beyond traditional imaging signatures.

Abstract and KeywordsO_ST_ABSBackground and ObjectivesC_ST_ABSPreoperative prediction of functional outcomes in contrast-enhancing glioma could support surgical decision-making and patient counseling, yet most existing models incorporate histopathological or postoperative variables unavailable before surgery. Our objectives were to develop a preoperative-only prediction model for one-year functional status and evaluate the added value of MRI-based tumor characteristics beyond clinical predictors. MethodsWe conducted a retrospective cohort study of consecutive adults ([&ge;] 18 years old) undergoing first resection of supratentorial contrast-enhancing glioma (WHO grade [&ge;] 2, histopathologically confirmed postoperatively) at a single center, with one-year follow-up. The primary outcome was functional status classified as mortality (Karnofsky Performance Score (KPS) = 0), functional dependence (KPS 10-60), or functional independence (KPS [&ge;] 70). In addition to clinical variables (age, sex, preoperative KPS, preoperative seizures), a deep learning tool was used to extract structural MRI-based tumor characteristics as predictors. A machine-learning model was developed and conformal prediction was applied to stratify patients by prediction confidence level. Results552 patients were included (median age: 60 years, range: 18-84; median contrast-enhancing volume: 24 mL, IQR: 10-43; median preoperative KPS: 80, range: 30-100; retrospectively confirmed 88% glioblastoma). Most MRI-based predictors did not improve performance as the best-performing model included three predictors: age at diagnosis, contrast-enhancing volume, preoperative KPS. Bootstrapped areas under the curves were 0.77 (95% confidence interval 0.70-0.84) for mortality, 0.64 (0.52-0.77) for functional dependence, and 0.71 (0.63-0.79) for functional independence. F1 scores per class were 0.65, 0.24, 0.65, respectively. Conformal prediction provided reliable predictions for 18% patients, moderate uncertainty for 57%, and identified 25% with genuinely unpredictable outcomes. DiscussionOur preoperative machine-learning model predicted one-year functional status in contrast-enhancing glioma with functional independence being the most reliably classified outcome (ROC-AUC = 0.77, F1 score = 0.65) and functional dependence the most challenging to predict (ROC-AUC = 0.64, F1 score = 0.24). A small set of three preoperative predictors drove model performance, supporting generalizability to broader patient populations. Our open-source model enables individualized risk stratification and may help clinicians identify patients with uncertain prognoses warranting more intensive preoperative counseling or follow-up planning.

BackgroundIsocitrate dehydrogenase (IDH) mutation and 1p19q codeletion are important beneficial prognosticators in glioma. IDH and 1p19q diagnosis requires tissue sampling and there are likely benefits of presurgical diagnosis. Research supports the potential of MRI-based IDH and 1p19q diagnosis, however there is a paucity of external validation outside the widely used The Cancer Imaging Archive (TCIA) dataset. We present a combined IDH and 1p19q classification algorithm and assess performance on a local retrospective cohort (NZ) and the Erasmus Glioma Database (EGD). Methods2D convolutional neural networks are trained to provide IDH and 1p19q classification. Inputs are T1 post-contrast, T2, and FLAIR sequences. Training data consists of preoperative imaging from the TCIA dataset (n=184) and a locally obtained NZ dataset (n=349). Evaluation data consists of the most recent cases from the NZ dataset (n=205) and the EGD (n=420). ResultsIDH classification accuracy was 93.3% and 91.5% on the NZ and EDG, with AUC values of 95.4% and 95.8%, respectively. 1p19q accuracy was 94.5% and 87.5% with AUC values of 92.5% and 85.4% on the NZ and EGD datasets. Combined IDH and 1p19q accuracy was 90.4% and 84.3% on the NZ and EGD, with AUC values of 92.4% and 91.2%. ConclusionsHigh IDH and 1p19q classification performance was achieved on the NZ retrospective cohort. Performance generalised to the EGD demonstrating the potential for clinical translation. This method makes use of readily available imaging and has high potential impact in glioma diagnostics. Key Points- IDH and 1p19q are the main molecular markers in glioma. - Accurate predictions can be obtained from preoperative MRI without changes to imaging protocols. - Non-invasive diagnosis will likely enhance treatment planning and facilitate targeted preoperative therapies. Importance of the StudyThe 2021 WHO CNS tumour classification system formalises the increasing recognition of molecular factors like IDH and 1p19q in the prognostication and treatment of glioma. Emerging research shows the potential of artificial intelligence methods applied to preoperative MRI sequences to noninvasively predict molecular status. A limitation of the literature published to date is a lack of generalisation and external validation outside the widely used TCIA dataset. Here we present the performance of an MRI-based IDH and 1p19q classification tool evaluated on a large consecutive cohort from New Zealand and an independent publicly available dataset of MR images from the Netherlands. We demonstrate high predictive performance with robust generalisation, indicating the potential usefulness of this method in the workup of glioma. Reliable preoperative tumour characterisation may facilitate tailored treatment approaches and early decision making without the need for additional imaging.

BackgroundGlioblastoma (GBM) is a highly aggressive cancer with near-universal recurrence, often due to residual tumor cells that persist after aggressive standard of care treatment. This study aimed to characterize tumor infiltration and microenvironment in the GBM periphery. MethodsWe prospectively collected 161 radiologically guided biopsies from 45 GBM patients and conducted an immunohistochemical analysis. We also integrated single-cell RNA sequencing data to select specific markers for immune, glial, and vascular cells. We measured the expression of these genes in samples from contrast-enhancing (CE) and non-enhancing (nCE) tumor areas, vasogenic edema, and radiologically normal tissue. Correlations with resection extent and clinical outcomes were evaluated. ResultsnCE biopsies exhibited neoplastic features similar to those of the tumor core. However, tumor infiltration was also found in regions classified radiologically as edema, particularly in elderly patients. We found important differences in the composition of the peripheral microenvironment between male and female GBM patients. Prognostic associations with specific cell types, such as myeloid cells, showed intertumor heterogeneity, with variations depending on patient sex, age and extent of resection. Furthermore, in our cohort, minimal residual CE tumor following surgery was associated with significantly poorer patient survival. ConclusionsThe GBM periphery includes regions of active tumor growth that are visible on MRI, as well as infiltrated areas that resemble edema radiologically. Tumor infiltration and microenvironmental features are influenced by patient sex and age, which has major implications for recurrence rates, highlighting the need to tailor surgical and therapeutic strategies based on tumor biology and patient subgroup. Key pointsnCE areas show similar neoplastic traits to the CE tumor. GBM infiltrates edema tissue, predominantly in older adult patients. Prognostic value of the peritumoral phenotype depends on resection extent and patient age/sex. Importance of the studyThis study shows that the impact of peripheral and core cellular features on prognosis differs between patients who undergo complete versus incomplete CE tumor resection. Our results suggest a paradigm shift in the classification and management of these patients, encouraging the inclusion of detailed post-surgical MRI analyses to guide the design of future clinical trials according to the nature and extent of the residual disease. Furthermore, our data confirm the presence of tumorigenic features in non-enhancing areas, supporting the benefits of supratotal resections according to the new RANO classification. Our findings also underscore the need to refine surgical and therapeutic strategies based on a more detailed understanding of the tumor microenvironment beyond the GBM core. This understanding may help identify novel targets for more effective and personalized GBM therapies.

Vestibular schwannoma (VS) is a benign, slow growing tumor that may affect hearing and balance. It accounts for 7-8% of all primary brain tumors. Gamma knife radiosurgery (GKRS) is a common treatment option for VS. Magnetic resonance imaging (MRI) is employed for diagnosis, surgery planning, and follow-up of VS. Long-term follow-up determines efficacy of VS treatment. Identifying MRI-derived markers to improve management of VS is challenging. This study describes MRI processing pipeline that automatically segments VS and investigates stability and outcome predictive power of radiomic MRI features. We first preprocessed and segmented available pre-GKRS T1-weighted post-contrast MRI images in VS patients, using a Convolutional Neural Network (CNN) developed on DeepMedic framework. Then, we compared CNN and manual segmentations, extracted radiomic features from both manual and CNN segmentations of VS, and, finally, evaluated robustness of extracted features and clinical outcome analyses based thereof. We found that homogeneity, robust maximum intensity and sphericity were the most robust across segmentations. We also found that maximum and minimum intensities were most predictive of tumor growth across all segmentation methods and subject cohorts. We used retrospective post-GK SRS data collected in our institution to build the processing pipeline for unsupervised segmenting of VS. This pipeline is released as a Docker image integrated with XNAT (extensible neuroimaging archive toolkit), an established open research imaging database platform15. Generated segmentations can be viewed and edited in the XNAT-based online OHIF (Open Health Imaging Foundation) viewer16 in real time.

BackgroundGlioblastoma (GBM), Isocitrate dehydrogenase-wildtype (IDH-wt) is characterised by diffuse infiltration, with progression often arising from perilesional tissue and occult white-matter damage. We investigated whether radiomics from the T2/FLAIR-defined oedema and the structural disconnectome improve prediction of progression-free survival (PFS). MethodsWe retrospectively analysed 387 adults with newly diagnosed GBM, IDH-wt treated at a single tertiary centre (2005-2020). A deep-learning pipeline segmented enhancing tumour, non-enhancing tumour, and oedema on pre-operative MRI; lesion masks were propagated to normative tractography to derive disconnectome maps. 3-D shape radiomic features extracted for each segmented region underwent appropriate feature selection. Finally, 10 tumour and 9 oedema radiomics were combined with 6 clinical features to train 3 survival models (Random Survival Forest (RSF), XGBoost, Cox proportional hazards (CPH)) that were evaluated on a held-out 20% test set using Harrells C-index, Kaplan-Meier risk stratification and time-dependent ROC curves. ResultsThe best performance was achieved by RSF using all clinical and radiomic features (C-index 0.665 vs 0.595 for clinical features only, p=0.088). Models including oedema radiomics outperformed those using tumour radiomics alone, and disconnectome features, derived from both tumour and oedema regions, were repeatedly selected among the top predictors across algorithms. Combining radiomic and clinical features improved risk stratification and 12-month early-versus-late recurrence classification (AUC 0.704 vs 0.582 for clinical features alone). ConclusionsIntegrating perilesional oedema and white-matter disconnectome MR features with clinical and molecular data enhances prediction of PFS in GBM, IDH-wt. These network-aware, multimodal survival models may support personalised risk-adapted treatment strategies pending external validation. Key Points- GBM IDH-wt exhibits a high recurrence rate despite aggressive treatment. - Addition of high-dimensional oedema and disconnectome radiomic features to clinical features showed consistent improvement in the test performance of 3 ML models. - This can support informed clinical decision-making. Importance of the StudyPrediction of progression free survival (PFS) for a patient with highly recurrent glioblastoma IDH-wt traditionally relies on clinical history, demographics, and molecular markers of the tumour. Recent literature reveals the tumours disruptive nature through its invasion of white-matter tracts and identifies its microenvironment, particularly the perilesional oedema, as a harbour of treatment resistant tumour cells. This study is the first to combine high-dimensional radiomic features of the tumour, the oedema, and their disconnectome with clinical and treatment factors to predict PFS. Using 3 model architectures (XGBoost, RSF, and CoxPH), we demonstrate consistent directional improvements in performance, on addition of radiomic features to clinical baseline models. Furthermore, oedema and disconnectome radiomics are identified as top predictor features across algorithms. This proof-of-concept study provides a reproducible multimodal pipeline, reaffirms the usability of MR radiomics, and identifies features of the oedema and the structural connectome as promising biomarkers, demanding large-scale external validation.

BackgroundBRAF status is crucial for treating pediatric low-grade gliomas (pLGG) and can be assessed non-invasively from segmented tumor regions on MRI using machine learning (ML). However, there are inherent limitations to manual and automated tumor segmentations. PurposeTo assess the performance of automated segmentation algorithms and to develop and assess a segmentation-free ML classification pipeline that identifies BRAF status from whole-brain FLAIR MRI sequences. Materials and MethodsIn this REB-approved retrospective study, molecularly-characterized tumors and whole-brain FLAIR MR images were collected from 455 patients with pLGG treated between 1999 and 2023 at a single tertiary care childrens hospital. We trained and evaluated three medical segmentation models, TransBTS, MedNeXt, and MedicalNet. Next, we developed a model to identify BRAF status from whole-brain FLAIR MRI, without any reliance on manual or automated segmentations. We then implemented a novel pretraining regimen that embedded segmentation knowledge into the whole-brain FLAIR MRI classification model. Finally, we trained and evaluated a baseline model that used manual segmentations as inputs. All ML models were trained and evaluated under a nested-cross validation scheme, and mean performance across all test folds was compared using the corresponding t-test. ResultsThe MedNeXt segmentation model (mean Dice score: 0.555) outperformed both the convolutional neural network (CNN) based MedicalNet (0.516) and the CNN-transformer hybrid TransBTS (0.449) (p <0.05 for all comparisons). The MedNeXt style classification model achieved a one-vs-rest area under the ROC curve of 0.741 using the whole brain FLAIR sequence as an input, without any segmentation knowledge. This was improved to 0.772 through pretraining on the segmentation task, which was not significantly different from the baseline manual segmentation-based model (0.756, p-value: 0.141). ConclusionBRAF status can be assessed non-invasively using ML models based on whole-brain FLAIR sequences. Dependence on inconsistent manual or automated segmentations can be reduced by integrating tumor region information into the model through pretraining.

BackgroundGlioblastoma (GBM) remains one of the most aggressive primary brain tumors, with limited survival despite maximal safe resection and chemoradiotherapy. Neoadjuvant bevacizumab (BEV) has been proposed to reduce peritumoral edema, improve functional status, and potentially enhance progression-free survival (PFS). However, its survival benefit in newly diagnosed, surgically resectable GBM remains unclear. ObjectiveTo systematically review and quantitatively synthesize the evidence on neoadjuvant BEV in adult patients with newly diagnosed, resectable GBM, focusing on survival and functional outcomes. MethodsFollowing PROSPERO registration (CRD420251078761), we searched PubMed, Embase, Scopus, Web of Science, and Cochrane Library up to July 20, 2025, without language restrictions. Eligible randomized trials, non-randomized trials, and cohort studies compared neoadjuvant BEV (alone or with other therapies) to standard care without BEV. Primary outcomes were overall survival (OS) and PFS; secondary outcomes included Karnofsky Performance Status (KPS), steroid use, radiological response, and biomarkers. Data were pooled using a random-effects model. ResultsThirteen studies (2 RCTs, 7 non-randomized trials, 4 cohorts) met the inclusion criteria; four (n=751) were eligible for meta-analysis. Pooled HR for OS was 0.72 (95% CI: 0.42-1.25, p=0.246) and for PFS was 0.72 (95% CI: 0.42-1.22, p=0.220), both with low heterogeneity (I2=0%). Functional outcomes suggested improved KPS and reduced steroid dependence, but certainty was low. Biomarker and radiological findings were inconsistent. ConclusionsNeoadjuvant BEV in resectable GBM does not significantly improve OS or PFS but may offer symptomatic and functional benefits. Current evidence is limited by small sample sizes, heterogeneous protocols, and low methodological quality. Well-designed multicenter RCTs are warranted.

PurposeGlioblastoma (GBM) is the most common adult primary brain tumor for which new therapeutic strategies are desperately needed. Monopolar spindle 1 (MPS1) is a mitotic kinase that plays a pivotal role in the spindle assembly checkpoint (SAC). GBM appears to be dependent on SAC fidelity, as MPS1 is overexpressed in many GBM patients. Thus, inhibiting MPS1 seems a viable therapeutic strategy to enhance mitotic cell death by attenuating SAC fidelity. NTRC 0066-0 is an MPS1 inhibitor that combines low nanomolar potency with a relatively long on-target residence time. MethodsWe here investigate the potential of NTRC 0066-0 as monotherapy and in combination with chemo-radiation for treatment of GBM using various in vitro and orthotopic in vivo models. ResultsWe show that NTRC 0066-0 efficiently induces GBM cell death in vitro, following continuous exposure with IC50s in the low nanomolar range. In contrast to previous reports of studies with other MPS1 inhibitors, we did not observe synergy in vitro with anti-microtubule drugs, such as docetaxel and vincristine. We demonstrate that NTRC 0066-0 has a high brain penetration, despite being a substrate of the efflux transporter P-glycoprotein. However, even when using recipient Abcb1a/b;Abcg2-/- mice with superior brain penetration and administering NTRC 0066-0 using a dose-dense regimen, we did not observe antitumor efficacy against an orthotopic GBM mouse model, neither as monotherapy nor in combination with standard-of-care temozolomide chemotherapy and radiotherapy. ConclusionThese data indicate that GBM is probably not a suitable indication for developing MPS1 inhibitors.

IntroductionSingle-cell RNA sequencing (scRNA-seq) has helped to elucidate the cellular composition of cancer and its microenvironment. Recent scRNA-seq studies have highlighted the heterogeneity of glioblastoma (GBM). Moreover, single-cell GBM analyses have proposed resemblance of GBM cells to radial glia and outer radial glia supporting the hypothesis that remnants of developmental tissue get reactivated in cancer. A recent study isolated neural progenitor cells (NPCs) from developing fetal human brain (gestational week 17-19) and classified NPCs based on their expression of THY1, CD24 and EGFR. Ventricular radial glia are THY1-CD24-EGFR+ whereas outer radial glia are THY1-CD24-EGFR-. Early neuron precursors are CD24+THY1-EGFR+ and glial progenitor cells (GPCs) are THY1+EGFR+. GPCs give rise to THY1+EGFR+PDGFRA+ pre-oligodendrocyte progenitor cells. The importance of EGFR in NPCs again highlights the resemblance to glioma. MethodsWe aimed to apply the classification above in IDH mutant astrocytoma and oligodendroglioma as well as IDHwt glioblastoma samples. We used three publicly available datasets: Wang (paired 74 IDHwt primary and recurrent samples), Tirosh (6 primary oligodendroglioma samples) and Venteicher (10 primary IDH mutant astrocytoma). ResultsIn IDH mutant astrocytoma, 82.63% of cells express THY1+ (mostly EGFR+PDGFRA+) and 10.76% of cells are THY1-CD24-EGFR+. In oligodendroglioma, 75% of cells are THY1+ (mostly EGFR+PDGFRA+) and 12.07% are THY1-CD24-EGFR+. In IDHwt EGFR amplified primary GBM samples, 87.5% of cells are THY1-CD24-EGFR+. This percentage drops to 70.4% in the recurrent setting. THY1-CD24-EGFR-cells increase from 9.7% to 23.1% at recurrence. In IDHwt EGFRwt primary GBM samples, 48.6% of cells are THY1-CD24-EGFR+ and 44.15% are THY1-CD24-EGFR-. In the recurrent setting, 43.26% of cells are THY1-CD24-EGFR+ and 49.58% are THY1-CD24-EGFR-. ConclusionIDH mutant gliomas and IDHwt glioblastoma express different progenitor cell markers. THY1 is highly expressed in IDH mutant gliomas.

PurposeArtificial intelligence (AI)-automated tumor delineation for pediatric gliomas would enable real-time volumetric evaluation to support diagnosis, treatment response assessment, and clinical decision-making. Auto-segmentation algorithms for pediatric tumors are rare, due to limited data availability, and algorithms have yet to demonstrate clinical translation. MethodsWe leveraged two datasets from a national brain tumor consortium (n=184) and a pediatric cancer center (n=100) to develop, externally validate, and clinically benchmark deep learning neural networks for pediatric low-grade glioma (pLGG) segmentation using a novel in-domain, stepwise transfer learning approach. The best model [via Dice similarity coefficient (DSC)] was externally validated and subject to randomized, blinded evaluation by three expert clinicians wherein clinicians assessed clinical acceptability of expert- and AI-generated segmentations via 10-point Likert scales and Turing tests. ResultsThe best AI model utilized in-domain, stepwise transfer learning (median DSC: 0.877 [IQR 0.715-0.914]) versus baseline model (median DSC 0.812 [IQR 0.559-0.888]; p<0.05). On external testing (n=60), the AI model yielded accuracy comparable to inter-expert agreement (median DSC: 0.834 [IQR 0.726-0.901] vs. 0.861 [IQR 0.795-0.905], p=0.13). On clinical benchmarking (n=100 scans, 300 segmentations from 3 experts), the experts rated the AI model higher on average compared to other experts (median Likert rating: 9 [IQR 7-9]) vs. 7 [IQR 7-9], p<0.05 for each). Additionally, the AI segmentations had significantly higher (p<0.05) overall acceptability compared to experts on average (80.2% vs. 65.4%). Experts correctly predicted the origins of AI segmentations in an average of 26.0% of cases. ConclusionsStepwise transfer learning enabled expert-level, automated pediatric brain tumor auto-segmentation and volumetric measurement with a high level of clinical acceptability. This approach may enable development and translation of AI imaging segmentation algorithms in limited data scenarios. SummaryAuthors proposed and utilized a novel stepwise transfer learning approach to develop and externally validate a deep learning auto-segmentation model for pediatric low-grade glioma whose performance and clinical acceptability were on par with pediatric neuroradiologists and radiation oncologists. Key PointsO_LIThere are limited imaging data available to train deep learning tumor segmentation for pediatric brain tumors, and adult-centric models generalize poorly in the pediatric setting. C_LIO_LIStepwise transfer learning demonstrated gains in deep learning segmentation performance (Dice score: 0.877 [IQR 0.715-0.914]) compared to other methodologies and yielded segmentation accuracy comparable to human experts on external validation. C_LIO_LIOn blinded clinical acceptability testing, the model received higher average Likert score rating and clinical acceptability compared to other experts (Transfer-Encoder model vs. average expert: 80.2% vs. 65.4%) C_LIO_LITuring tests showed uniformly low ability of experts ability to correctly identify the origins of Transfer-Encoder model segmentations as AI-generated versus human-generated (mean accuracy: 26%). C_LI

BackgroundGlioblastoma is a highly aggressive brain cancer and, unlike many other cancers types, the median survival for patients after treatment (14.6 months) has barely improved in the last 20 years. Infiltrative growth into the surrounding brain parenchyma facilitates tumor recurrence and ultimately the death of the patient - novel therapies targeting this process are desperately needed. Lysyl oxidase inhibition has been shown to decrease invasive growth in a variety of solid tumours and is a potential therapy for glioblastoma patients. MethodsGenes highly expressed in the mesenchymal subtype of glioblastoma were analyzed in a data set from the Cancer Genome Atlas and tissue microarrays. Two patient-derived human glioblastoma stem cell lines were used to assess the involvement of lysyl oxidase (LOX). The effect of LOX on infiltration was examined in an organotypic brain slice assay and in an orthotopic mouse model. Chemotactic assays, protease and cleavage arrays were used to assess the underlying mechanism behind LOX-mediated infiltration. The orthotopic model was used to evaluate potential clinical utility of targeting LOX in glioblastoma. ResultsLOX is overexpressed in the mesenchymal glioblastoma subtype and strongly associated with poor patient survival. LOX expression upregulates MMP7 expression, which subsequently cleaves the vascular matrix resulting in increased chemotaxis of glioblastoma cells. ConclusionsWe have uncovered a novel mechanism of glioblastoma infiltration and suggest that targeting LOX represent an effective therapeutic approach blocking glioblastoma infiltration. Importance of the studyThe ability of glioblastoma cells to infiltrate the surrounding normal brain tissue facilitates their evasion of current therapies, leading to tumor recurrence and ultimately the death of the patient. To improve targeted therapies for glioblastoma patients we need to understand the molecular mechanisms of glioblastoma cell infiltration and how cells interact with the unique microenvironment of the brain. We have identified a novel mechanism whereby tumor-derived LOX mediates chemotaxis of glioblastoma cells to the laminin rich perivascular niche, enabling infiltrative growth. Inhibiting this infiltrative pathway is a potential anti-invasive therapy that is desperately needed for glioblastoma patients.

BackgroundGlioblastoma is an aggressive and incurable brain tumor. Clinical trials of immune checkpoint inhibitors showed no clinical benefit in glioblastoma when given after surgery. However, a clinical trial in which PD1 inhibition was given prior to second surgery did show pharmacodynamic evidence for activity. This suggests the possibility that immune checkpoint inhibitors may be more effective in a setting where large tumors are present. Here we have studied immune responses to large tumors in an autochthonous mouse model of glioblastoma. MethodsGlioblastoma was induced by transfection with oncogenic plasmids injected directly into the lateral ventricle of neonatal mice. Immune responses were assessed using a combination of spectral flow cytometry and immunohistochemistry. ResultsThere was a marked immune response to large tumors, with significant increases in CD4 T cells and dendritic cells. T cell changes occurred primarily at leptomeningeal/perivascular border sites. A large proportion of CD4 T cells expressed PD1 and half of these were regulatory T cells. NK cells were also increased in mice with large tumors, but were predominantly in immature states. The mouse model accurately recapitulates the formation of palisading necroses. These contain apoptotic cells and avidly recruit myeloid cells that are induced to express large amounts of TGF{beta}. ConclusionsLarge glioblastoma tumors generate a border site population of PD1 positive T cells that may explain the pharmacodynamic response in neoadjuvant trials, and a palisading necrosis-driven immunosuppressive mechanism that may explain why responses are insufficient to provide a significant clinical benefit. KEY POINTSThe SB mouse model accurately recapitulates immune features of human glioblastoma Large tumors induce a significant border site immune response Palisading necroses in large tumors counter this with a strong immunosuppressive response IMPORTANCE OF STUDYImmune checkpoint inhibitors have not shown efficacy in glioblastoma when used post-surgery, but do show pharmacodynamic activity when used in patients prior to second surgery (i.e. neoadjuvant). This suggest the possibility that immune checkpoint inhibition is more effective when large tumors are present. Using a clinically-relevant autochthonous mouse model, we show here that large tumors induce an immune response that is evident in leptomeningeal border sites. Large tumors in this mouse model also generate palisading necroses, a well-known diagnostic feature in glioblastoma tumors. These palisading necroses generate large amounts of TGF{beta}, providing a mechanism by which large tumors can suppress border site immune responses. This further supports the concept that palisading necroses are drivers of glioblastoma malignancy and suggests novel strategies to enhance responses to immune checkpoint inhibition in this cancer.

Brain tumors are among the most lethal cancers with gliomas representing the most morphologically complex type. Precise and time efficient glioma segmentation and classification are essential for accurate diagnosis, treatment planning, and patient monitoring. Magnetic resonance imaging (MRI) remains the primary imaging modality for noninvasive glioma assessment. This review systematically analyzes deep learning (DL) and artificial intelligence (AI) approaches for brain tumor segmentation and classification. Thirty one studies, out of 310 published between 2022 and 2025, met the inclusion criteria, among which 8 performed both segmentation and classification tasks. For segmentation, most of the studies used publicly available multiparametric MRI datasets. Segmentation performance varied by model and tumor region, with those focused on the whole tumor region achieving the highest Dice Score Coefficient (DSC). Classical U Nets achieved DSC scores around 80%, while advanced models integrating residual or attention modules exceeded 90%. Two main classification tasks were performed: tumor type and glioma staging. Classification models primarily relied on learned features extracted from multiparametric MRI using DL models, reporting an accuracy from 91.3% to 99.4%, with sensitivity and specificity typically above 95%, indicating robust predictive performance. Surprisingly, explainable AI approaches were infrequently applied, highlighting the persistent need for greater model transparency to foster clinical trust. Overall, these results demonstrate the strong potential of current AI based segmentation and classification pipelines. These methods can help clinicians accelerate the decision making process, increasing both the accuracy and efficiency of brain tumor diagnosis. These approaches may also support the development of personalized treatment plans tailored to each patient.

Background and purposeTimely identification of local failure after stereotactic radiotherapy for brain metastases allows for treatment modifications, potentially improving outcomes. Previous studies showed that adding radiomics or Deep Learning (DL) features to clinical features increased Local Control (LC) prediction accuracy. However, no study has integrated radiomics, DL, and clinical features into machine learning algorithms to predict LC. We examined whether a model using all these features achieves better accuracy than models using only a subset. Materials and methodsWe collected pre-treatment brain MRIs and clinical data for 129 patients at the Gamma Knife Center of the Elisabeth-TweeSteden Hospital. Radiomics features (extracted using the Python radiomics feature extractor) and DL features (extracted using a 3D ResNet model) were combined with clinical features. Performance of a Random Forest classifier was compared across four models trained with: clinical features only; clinical and radiomics features; clinical and DL features; and clinical, radiomics, and DL features. ResultsThe prediction model utilizing only clinical variables provided an Area Under the receiver operating characteristic Curve (AUC) of 0.82 and an accuracy of 75.6%. Adding radiomics features increased the AUC to 0.88 and accuracy to 83.3%, while adding DL features resulted in an AUC of 0.86 and accuracy of 78.3%. The best performance came from combining clinical, radiomics, and DL features, achieving an AUC of 0.89 and accuracy of 87%. ConclusionIntegrating radiomics and DL features with clinical characteristics improves LC prediction after stereotactic radiotherapy for brain metastases. These findings demonstrate the potential for early outcome prediction, enabling timely treatment modifications to improve patient management. Clinical and Translational Impact StatementOur study holds great clinical value, as the increased prediction accuracy can lead to tailored and effective interventions, resulting in better outcomes for brain metastases patients treated with stereotactic radiotherapy.

Background: Extent of resection remains central to meningioma management, yet Simpson grading is subjective and may not reflect measurable postoperative residual disease. We compared surgeon-reported Simpson grade, report-derived radiological grading, and residual tumour volumetry across a multicentre cohort. Methods: We performed a retrospective study across two tertiary neurosciences centres comprising four hospitals, including patients undergoing primary cranial meningioma resection from 2006 to 2025. Postoperative magnetic resonance imaging (MRI) reports were harmonised using weakly supervised natural language processing based on term frequency-inverse document frequency (TF-IDF) and a linear support vector machine classifier. Residual tumour volume was segmented from contrast-enhanced postoperative MRI and log-transformed. Concordance between Simpson and radiological gross-total/subtotal resection classification was assessed using absolute agreement and prevalence-adjusted bias-adjusted kappa (PABAK). Cox models assessed recurrence-free survival, with bootstrap validation and anatomical and scan-timing sensitivity analyses. Results: Among 912 patients, recurrence or residual progression occurred in 281. Surgical-radiological agreement was substantial but imperfect (absolute agreement 74%; PABAK 0.61), with lower agreement in skull-base and parafalcine-parasagittal tumours. In adjusted models, recurrence hazard increased with Simpson grade (hazard ratio 1.54, 95% confidence interval 1.37-1.72), radiological grade (1.92, 1.68-2.20), and log-transformed residual volume (1.20, 1.16-1.24; all p<0.0005). Optimism corrected concordance increased from Simpson grade to radiological grade and log-volumetry (0.692, 0.733, and 0.748), with this ranking preserved across sensitivity analyses. Conclusions: Imaging-based postoperative residual disease measures outperformed Simpson grade. TF-IDF-assisted report-derived grading provides a scalable bridge to volumetry, while quantitative residual volume offers the strongest prognostic representation.

IntroductionGlioblastoma is a devastating brain tumor with poor prognosis despite current treatment modalities. Chimeric antigen receptor T-cell (CAR T-cell) therapy has shown promise in other cancers but has yielded mixed results in glioblastoma. This augmented meta-analysis aims to address the limitations of previous studies and evaluate the safety and efficacy of CAR T-cell therapy for recurrent glioblastoma. MethodsWe followed PRISMA guidelines, including specific inclusion and exclusion criteria, for our literature review. Eight studies with 108 patients were included. We used standard and augmented meta-analyses to assess outcomes, complications, and publication bias. ResultsIt was found that the mean overall survival for glioblastoma patients who underwent CAR T-cell therapy was 6.49 months, demonstrating no significant deviation from the median survival observed in those following the standard protocol. CAR T-cell therapy did not lead to a statistically significant improvement in achieving complete responses, with only 80% of patients exhibiting this outcome. Conversely, 44% of patients experienced stable disease, while 58% faced disease progression after CAR T-cell therapy. Adverse events were notable, with CAR T cell therapy-related encephalopathy affecting 37% of treated patients, while cytokine release syndrome was a rare event, observed in only 3% of cases. ConclusionsTo our knowledge, this is the first study that utilizes this novel statistical technique to predict the outcomes of CAR T-cell therapy for recurrent glioblastoma. The results of this study are predictive rather than confirmatory. CAR T-cell therapy for glioblastoma was not predicted to significantly improve survival or achieve substantial complete responses. Stable disease rates are modest, while disease progression is notable. Adverse events, especially CAR T-cell therapy-related encephalopathy, raise safety concerns. Further trials and refinements are needed to enhance CAR T-cell therapys effectiveness and safety in glioblastoma treatment, Manuscript Click here to view linked References potentially through optimizing administration routes and target antigens or combining it with other therapies. This challenging disease necessitates continued research to improve patient outcomes.

BackgroundPrecise delineation of non-contrast-enhancing tumor (nCET) in glioblastoma (GB) is critical for maximal safe resection, yet routine imaging cannot reliably separate infiltrative tumor from vasogenic edema. The aim of this study was to develop and validate an automated method to identify nCET and assess its prognostic value. MethodsPre-operative T2-weighted and FLAIR MRI from 940 patients with newly diagnosed GB in four multicenter cohorts were analyzed. A deep-learning model segmented enhancing tumor, edema and necrosis; a non-local spatially varying finite mixture model then isolated edema subregions containing nCET. The ratio of nCET to total edema volume--the Diffuse Infiltration Index (DII)--was calculated. Associations between DII and overall survival (OS) were examined with Kaplan-Meier curves and multivariable Cox regression. ResultsThe algorithm distinguished nCET from vasogenic edema in 97.5 % of patients, showing a mean signal-intensity gap > 5 %. Higher DII is able to stratify patients with shorter OS. In the NCT03439332 cohort, DII above the optimal threshold doubled the hazard of death (hazard ratio 2.09, 95 % confidence interval 1.34-3.25; p = 0.0012) and reduced median survival by 122 days. Significant, though smaller, effects were confirmed in GLIOCAT & BraTS (hazard ratio 1.31; p = 0.022), OUS (hazard ratio 1.28; p = 0.007) and in pooled analysis (hazard ratio 1.28; p = 0.0003). DII remained an independent predictor after adjustment for age, extent of resection and MGMT methylation. ConclusionsWe present a reproducible, server-hosted tool for automated nCET delineation and DII biomarker extraction that enables robust, independent prognostic stratification. It promises to guide supramaximal surgical planning and personalized neuro-oncology research and care. Key Points- KP1: Robust automated MRI tool segments non-contrast-enhancing (nCET) glioblastoma. - KP2: Introduced and validated the Diffuse Infiltration Index with prognostic value. - KP3: nCET mapping enables RANO supramaximal resection for personalized surgery. Importance of the StudyThis study underscores the clinical importance of accurately delineating non-contrast-enhancing tumor (nCET) regions in glioblastoma (GB) using standard MRI. Despite their lack of contrast enhancement, nCET areas often harbor infiltrative tumor cells critical for disease progression and recurrence. By integrating deep learning segmentation with a non-local finite mixture model, we developed a reproducible, automated methodology for nCET delineation and introduced the Diffuse Infiltration Index (DII), a novel imaging biomarker. Higher DII values were independently associated with reduced overall survival across large, heterogeneous cohorts. These findings highlight the prognostic relevance of imaging-defined infiltration patterns and support the use of nCET segmentation in clinical decision-making. Importantly, this methodology aligns with and operationalizes recent RANO criteria on supramaximal resection, offering a practical, image-based tool to improve surgical planning. In doing so, our work advances efforts toward more personalized neuro-oncological care, potentially improving outcomes while minimizing functional compromise.

BACKGROUNDImaging-based monitoring of gliomas is limited by treatment-related changes. D-2-hydroxyglutarate (D-2-HG), produced by the isocitrate dehydrogenase (IDH) mutation, is detectable in cerebrospinal fluid (CSF) that can be accessed from various anatomic compartments. We evaluated CSF D-2-HG as a serially accessible biomarker for IDH-mutant gliomas. METHODSA CLIA-approved gas chromatography mass spectrometry assay was developed for CSF D- and L-2-HG. Lumbar and cranial CSF samples were collected from patients with IDH-mutant gliomas or IDH-wild-type brain tumors and non-tumor pathologies via surgical field collection, lumbar punctures, Ommaya reservoirs, and ventriculoperitoneal shunts. RESULTSCSF D-2-HG was significantly higher in cranial than lumbar samples from IDH-mutant glioma patients (median lumbar=0.20 M, cranial = 1.72 M; p<0.0001). Cranial, but not lumbar, CSF D-2-HG distinguished primary IDH-mutant gliomas from IDH-wild type lesions (cranial AUC= 0.89, 95% confidence interval (CI)= 0.80-0.97); lumbar AUC= 0.52, 95% CI=0.28-0.76). When evaluated in recurrent lesions as a separate validation cohort, this finding was also reproduced in this group (cranial AUC=0.97, 95% CI= 0.94-1.00; lumbar AUC=0.60, 95% CI=0.38-0.83). Cranial CSF D-2-HG levels decreased to 0.54x of baseline with resection in seventeen patients (p=0.0129) but did not decrease significantly with chemoradiation in five patients (p=0.6250). Longitudinal anatomical changes, such as cavity collapse, influenced serial sample interpretation. In grade 4 IDH-mutant astrocytomas, serial cranial CSF D-2-HG increased with disease progression and differentiated stability from pseudoprogression when tumor-CSF contact was sufficient. CONCLUSIONSSerial cranial CSF D-2-HG shows promise as a monitoring biomarker in patients with IDH-mutant gliomas when anatomic variables remain constant. KEY POINTSO_LICranial CSF D-2-HG levels exceed that of lumbar CSF in patients with IDH-mutant gliomas. C_LIO_LICranial CSF D-2-HG may discriminate disease stability vs. treatment effects, although post-resection anatomical changes can impact monitoring. C_LI IMPORTANCE OF THE STUDYImproved glioma monitoring is needed due to challenges distinguishing disease progression from treatment-related changes on imaging. Toward this goal, we evaluated CSF D-2-HG as a biomarker of IDH-mutant gliomas using a CLIA-approved assay. This study answers whether D-2-HG can identify IDH-mutant gliomas via either cranial or lumbar CSF. Importantly, in seventeen patients, we demonstrate that CSF D-2-HG is responsive to cytoreduction via resection, but not chemoradiation in five patients. This is also the first study to demonstrate that longitudinal anatomical changes can impact evaluation of CSF D-2-HG as a monitoring biomarker. Finally, the study demonstrates that serial CSF D-2-HG can increase with disease progression, but not pseudoprogression or stable disease, in five patients with grade 4 IDH-mutant astrocytomas. These findings support the potential of CSF D-2-HG as a monitoring biomarker in patients with IDH-mutant gliomas, particularly when there are minimal changes to the anatomy of the resection cavity.