N6-methyladenosine (m6A) is the most prevalent internal mRNA modification and is enriched in the central nervous system (CNS), yet its role in glioma remains incompletely defined. Using long-read direct RNA sequencing, we mapped transcriptome-wide m6A modifications in a single glioma cell line following targeted knockdown of the m6A reader IGF2BP2, writer METTL3, and eraser ALKBH5. Across perturbations, the global architecture of m6A, including transcript class, positional enrichment, and site multiplicity, was largely preserved, while differential methylation was weakly coupled to gene expression. In contrast, m6A regulator perturbation coincided with widespread isoform switching and with changes in untranslated regions, coding potential, and predicted transcript fate, largely independent of bulk gene expression changes. Public glioma datasets further supported the relevance of isoform-specific changes in glioma. Together, these findings highlight isoform-level transcript variation associated with m6A regulator perturbation and support the use of long-read, isoform-resolved approaches to study RNA regulatory states in glioma.

OBJECTIVE:Large language models (LLMs) have shown promising performance on medical licensing examinations, but their ability to excel in subspecialty domains and their robustness under adversarial conditions remain unclear. Herein, the authors present AtlasGPT, a subspecialty-focused LLM for neurosurgery, and evaluate its performance on a benchmark multiple-choice question bank and under adversarial testing, as well as its ability to generate high-quality explanations. METHODS:AtlasGPT was built by fine-tuning GPT-4 architecture and retrieval-augmented generation from neurosurgical knowledge sources. Its performance was compared with that of GPT-4 and Gemini Advanced on a 149-question neurosurgery examination. Adversarial testing assessed robustness to misinformation. Answer explanations were rated by 15 independent neurosurgeons and compared with the question bank. RESULTS:Across all 149 questions and on text-only questions, AtlasGPT (96%) outperformed Gemini Advanced (93%) and GPT-4 (88%) in accuracy. In adversarial testing, under which AtlasGPT was tasked with identifying medical misinformation, it was fooled 14% of the time, compared with 44% for GPT-4 and 68% for Gemini Advanced. Neurosurgeons rated AtlasGPT's answer explanations as significantly more comprehensive, relevant, and better referenced than the question bank's explanations of the responses (p < 0.001). AtlasGPT did not demonstrate any evidence of hallucination or other content that would be harmful for patient care or the surgeon's clinical decision. CONCLUSIONS:AtlasGPT demonstrates the potential of subspecialty-focused LLMs to outperform general models, exhibit robustness to misinformation, and generate high-quality explanations. Domain-specific LLMs may improve medical knowledge, decision-making, and educational materials in complex fields like neurosurgery.

Mutations in isocitrate dehydrogenase (IDH) are a significant prognostic and biological factor leading to slower growth and T cell suppression within diffuse gliomas. However, the effect of IDH and its downstream metabolites specifically on intratumoral myeloid cells remains underexplored. Utilizing patient tumor samples, we performed RNA-sequencing and quantitative immunofluorescence on IDH-wildtype glioblastoma and IDH-mutant grade 4 astrocytoma cases. We then engineered the murine GL261 glioma cell line to harbor mutant IDH, comparing transcriptomic and cell-level changes in IDH-wildtype versus IDH-mutant murine tumors. We identified greater hallmarks of productive inflammation in IDH-mutant tumors compared to IDH-wildtype tumors. We also saw transcriptomic enrichment of suppressive macrophage and myeloid-derived suppressor cell (MDSC) signatures in IDH-wildtype tumors, which was confirmed at the cellular level. Furthermore, engineering the IDH mutation into murine tumors appeared sufficient to recapitulate many of the transcriptomic and cellular shifts observed among patient samples. Our data show that mutant IDH is associated with greater inflammatory signatures and fewer suppressive myeloid cells in human gliomas, and that delivering mutant IDH to murine tumors is sufficient to drive these microenvironment changes. This work advances our understanding of key myeloid cell populations that may be targeted by future immunotherapy strategies.

N6-methyladenosine (m6A) is the most prevalent internal mRNA modification, enriched in the CNS yet poorly characterized in glioma. Using long-read RNA sequencing, we mapped m6A in an in vitro glioma model following knockdown (KD) of the reader IGF2BP2, writer METTL3, and eraser ALKBH5, with naive glioma cells and astrocytes as controls. Glioma cells exhibited a two-fold reduction in global m6A, suggesting progressive loss from healthy to malignant states. Integrated analysis revealed that m6A mediated control of gene expression is influenced by modification topology (CDS:3'UTR), transcript biotype, and length. Regulator KD, particularly ALKBH5 induced redistribution of m6A toward 3'UTR with consequent gene upregulation. We also identified m6A-mediated isoform switching, with a higher usage of retained intron and nonsense-mediated decay isoforms. Structural and splicing alterations at the isoform level were identified unique to each KD condition indicating m6A driven aberrant alternative splicing. At the functional level, KD specific remodeling of oncogenic signaling was also observed. ALKBH5 KD suppressed MYC targets and pro-apoptotic signaling while METTL3 KD enhanced mTOR and PI3K-AKT signaling. Collectively, these results demonstrate that m6A mediated regulation in glioma is highly context-dependent, defining distinct clinically relevant phenotypes. This has implications for future biomarker discovery and development of targeted therapeutics.

Background:The glioma immune repertoire has emerged as a vital point of interest, particularly in the context of immunotherapeutics development and as a key player for prognostic and diagnostic biomarker identification. Methods:Tumor tissue was collected from glioma patients and targeted immune repertoire sequencing of tumor infiltrating lymphocytes (TIL) from each of the four collected glioma tumor subtypes was performed. Gliomas were stratified based on WHO21 classification to map the TCR landscape of astrocytomas (grade II/III, grade IV), glioblastomas, and oligodendrogliomas. Results:Following stratification of TCR repertoires and complete clonotype, V-J cassette, and CDR3 analysis, we identified cohort-specific levels of diversity, clonotype sharing, and conservation. Partitioning of these repertoires based on TCR diversity revealed significant influence on patient survival. Furthermore, mapping of CDR3 binding regions to antigens and their origins highlighted prognostic biomarkers and identified sequences binding to viral signatures associated with patient clinical outcomes. Conclusion:These findings underscore the importance of characterizing TCR repertoires in the context of the patient clinical condition. These unique repertoire signatures and correlated antigens may facilitate patient outcome prognostication and serve as a potential foundation for immunotherapeutic applications.

Enlargement of the cerebrospinal fluid (CSF)-filled brain ventricles (ventriculomegaly) is a defining feature of congenital hydrocephalus (CH) and an under-recognized concomitant of autism. Here, we show that de novo mutations in the autism risk gene PTEN are among the most frequent monogenic causes of CH and primary ventriculomegaly. Mouse Pten-mutant ventriculomegaly results from aqueductal stenosis due to hyperproliferation of periventricular Nkx2.1+ neural progenitor cells (NPCs) and increased CSF production from hyperplastic choroid plexus. Pten-mutant ventriculomegalic cortices exhibit network dysfunction from increased activity of Nkx2.1+ NPC-derived inhibitory interneurons. Raptor deletion or postnatal everolimus treatment corrects ventriculomegaly, rescues cortical deficits and increases survival by antagonizing mTORC1-dependent Nkx2.1+ NPC pathology. Thus, PTEN mutations concurrently alter CSF dynamics and cortical networks by dysregulating Nkx2.1+ NPCs. These results implicate a nonsurgical treatment for CH, demonstrate a genetic association of ventriculomegaly and ASD, and help explain neurodevelopmental phenotypes refractory to CSF shunting in select individuals with CH.

Glioma represents the most common CNS neoplasm in adults. Current classification utilizes molecular alterations, particularly IDH1.R132H, to stratify lesions into distinct prognostic and therapeutic protocol groups. Identification of the single nucleotide variant through traditional tissue biopsy assessment poses procedural risks and may not fully reflect the heterogeneous and evolving tumor landscape. Initial optimization included, comparing the sensitivity of detection in RNA compared to DNA across different sample types. Following nucleic acid selection, key variables were optimized to enhance the signal-to-noise ratio using droplet digital PCR (ddPCR) technology and included: (i) amplicon size, (ii) cycling conditions, (iii) preamplification PCR, and (iv) gating threshold. Rigorous, blinded validation testing was performed in n = 133 patients comprising of IDH1.R132H gliomas (n = 80), IDH1 wild-type gliomas (n = 44), and age matched healthy controls (n = 9). Results from plasma based diagnostic testing yielded an overall sensitivity of 75.0% (95% CI: 64.1%-84.0%), specificity 88.7% (95% CI: 77.0%-95.7%), positive predictive value 90.9%, and negative predictive value 70.1% compared to the tissue gold standard. Findings indicated a correlation between IDH1.R132H mutant allele frequency (MAF; tumor tissue, plasma) and important clinical variables including: (i) overall survival, (ii) tumor shedding rate, (iii) treatment outcomes. Longitudinal monitoring using the mt-IDH1dx assay (n = 8, 39 timepoints) demonstrated excellent concordance with radiological findings and varying disease outcomes (progression, stable disease, treatment response). In addition to baseline diagnosis, mt-IDH1dx assay can be utilized for blood-based prognostication and longitudinal monitoring. The proposed workflow enables rapid and efficient completion of both tumor tissue and plasma testing in under 4 hours from the time of sampling.

Gliomas are biologically heterogeneous brain tumors with marked differences in clinical behavior based on the IDH1 mutation status. While epigenetic dysregulation is well characterized, the contribution of RNA modifications, particularly N6-methyladenosine (m6A), remains underexplored. Using direct RNA nanopore sequencing of patient-derived gliomas, we generated the first isoform-resolved m6A maps across IDH1-mutant and wild-type tumors. IDH1-mutant gliomas exhibited globally elevated m6A methylation, along with increased expression of methyltransferases (METTL3, METTL14) and stabilizing readers (YTHDF3). In contrast, wild-type glioblastomas showed enhanced expression of m6A erasers (ALKBH5, FTO) and RNA decay factors (YTHDF2). These subtype-specific differences in m6A architecture impacted transcript stability, isoform usage, and gene expression. Isoform-level analyses revealed stronger prognostic associations than gene-level parameter, including for IGF2BP2-202, PUF60-202, and GLUL-203. Our study establishes m6A as a critical, subtype-specific layer of RNA regulation in glioma with clinical and therapeutic implications.

INTRODUCTION: Large language models (LLMs) have shown promising performance on medical licensing exams, but their ability to excel in subspecialty domains and their robustness under adversarial conditions remain unclear. METHODS: AtlasGPT was built using GPT-4 with retrieval-augmented generation from expert-verified neurosurgical knowledge sources. Its performance was compared to GPT-4 and Gemini Advanced on a 149-question neurosurgery exam. Adversarial testing assessed robustness to misinformation. Answer explanations were rated by 15 independent neurosurgeons and compared to the question bank. RESULTS: Across all 149 questions, AtlasGPT achieved 90.6% accuracy, outperforming GPT-4 (80.5%, P=0.020) and Gemini Advanced (80.5%, P=0.020). On text-only questions, AtlasGPT, Gemini Advanced and GPT-4 achieved 95.6%, 92.9% and 87.8% accuracy, respectively. In adversarial testing, AtlasGPT was fooled 14% of the time, compared to 44% for GPT-4 and 68% for Gemini Advanced. Neurosurgeons rated AtlasGPT's answer explanations as significantly more comprehensive, relevant, and better-referenced than the question bank's explanations (P<0.001). CONCLUSIONS: AtlasGPT demonstrates the potential of subspecialty-focused LLMs to outperform general models, exhibit robustness to misinformation, and generate high-quality explanations. Domain-specific LLMs may improve medical knowledge, decision-making, and educational materials in complex fields like neurosurgery.

BACKGROUND:Meningioma is the most common primary CNS tumor, with high-grade cases exhibiting aggressive behavior, frequent recurrence, and poor prognosis. Currently, no systemic therapies are approved for recurrent or malignant meningiomas. Chimeric antigen receptor (CAR) T-cell therapy has shown efficacy in hematologic malignancies and promise for solid tumors, but its use for meningiomas has been underexplored. Mesothelin, a glycoprotein overexpressed in several solid tumors of mesodermal origin, may serve as a viable immunotherapeutic target. This study aimed to evaluate mesothelin as a CAR T-cell target in meningiomas. METHODS:Mesothelin expression was analyzed in patient-derived meningioma samples using immunohistochemistry, flow cytometry, and droplet digital PCR. Mesothelin-specific CAR T-cells were generated and evaluated invitro, exvivo using patient-derived organotypic tumor spheroids (PDOTS), and invivo using orthotopic meningioma mouse models of human xenografts. Cytotoxicity, T-cell proliferation, cytokine secretion, and tumor clearance were assessed. RESULTS:Mesothelin was detected in a subset of tumors across all meningioma grades at the transcript and protein levels, with surface expression confirmed in patient-derived primary cells. Mesothelin-specific CAR T-cells exhibited potent and specific cytotoxicity, T-cell activation, and cytokine secretion in vitro and effectively eliminated PDOTS. In orthotopic human xenograft models, mesothelin CAR T-cell therapy led to significant tumor regression and prolonged survival. CONCLUSIONS:Mesothelin is a viable CAR T-cell target for meningiomas, and mesothelin-specific CAR T-cell therapy shows strong preclinical efficacy. These findings provide a rationale for early-phase clinical trials of mesothelin CAR T-cell therapy in patients with refractory meningiomas.

High-grade gliomas (HGGs) are aggressive, infiltrative brain tumors with poor survival outcomes and limited therapeutic options. Their spatial and molecular heterogeneity necessitates large, well-curated datasets integrating imaging, clinical, and genomic information to enable precision medicine and drive novel treatment development. Using the Medtronic StealthStation neuronavigation system, we collected over 1,000 spatially localized intraoperative biopsy samples from more than 250 patients undergoing resection for glioma at Mayo Clinic and partnering sites. Each biopsy site was recorded with 3D coordinates, labeled in real time, flash-frozen, and annotated with metadata including collection time, anatomical location, and surgical plan. Samples were processed and entered into a secure database along with the corresponding sampling coordinates, frozen pathology, whole-exome and RNA sequencing data, 5-ALA fluorescence status, patient demographics, and T1, T1Gd, T2, and T2-FLAIR MRI. This multimodal dataset supports radiogenomic research by enabling precise mapping of imaging features to molecular characteristics. A subset of this data, including 268 biopsies from 52 patients (25 male, 27 female) with reported intraoperative 5-ALA fluorescence, was used to develop a radiomics model predicting locoregional patterns of 5-ALA positivity based on preoperative MRI. Image features were extracted from the biopsy sample coordinates across multiple MRI sequences and paired with corresponding 5-ALA fluorescence status to train the model for regional fluorescence prediction. The final model achieved 83% accuracy on the training set and 86% on the validation set using a 70-15-15 data split; this included 208 samples for training and 42 for validation, stratified by patient to ensure no overlap of samples from the same individual across sets. These findings demonstrate the feasibility of predicting 5-ALA fluorescence using radiomic features from standard preoperative MRI, offering a promising noninvasive strategy for improving intraoperative decision-making and enabling more precise, personalized resections in patients with high-grade gliomas.

OBJECTIVES:To determine if the chronic use of bevacizumab, prior to withholding for elective surgery in the preoperative period, increases the risk of hemorrhage during vestibular schwannoma (VS) resections. METHODS:Retrospective reviews of estimated intraoperative blood loss volume during surgical resection of 50 bevacizumab-treated and 56 bevacizumab-untreated VS from patients with NF2-related schwannomatosis (NF2-SWN). Two index cases are included. Additionally, masked histopathologic examination of tissues from bevacizumab-treated and untreated VS was evaluated. RESULTS:The mean blood loss during VS resection was statistically increased in patients with prior bevacizumab treatment (421 mL) compared to untreated patients (285 mL) (p = 0.015). When accounting for both tumor size and the amount of tumor resected, blood loss per percent of tumor resected was also elevated in the bevacizumab treated group (15.78 mL/%resection) when compared to the untreated group (12.97 mL/%resection) (p = 0.024). Intraoperative hemostasis using standard techniques was difficult, and some cases resulted in early termination of the surgical procedure prior to total resection. Eighteen percent of bevacizumab-treated patients had gross total or near total tumor resection, while untreated tumors had 52% gross total or near total resections (p-value = 0.001). The length of time on bevacizumab preoperatively correlated with intraoperative blood loss (p = 0.049). The median time patients were treated with bevacizumab prior to surgery was 23.5 months and ranged from 3 to 121 months. The median time off bevacizumab prior to surgery was 11.5 months and ranged from 0.75 to 87 months. Histopathologic examination revealed no readily apparent difference in blood vessel density, integrity, or growth patterns. CONCLUSIONS:Bevacizumab treatment prior to excision of NF2-SWN VS in this retrospective case series is associated with increased intraoperative hemorrhage. Bleeding was difficult to control using standard techniques, thus warranting preoperative surgical awareness. Underlying pathologic mechanisms are not yet understood and are under further investigation. LEVEL OF EVIDENCE: 3:

INTRODUCTION: Chimeric antigen receptor (CAR) T cells represent a promising approach to cancer and have proven efficacy against hematological malignancies, for which they have become the standard of care. However, the use of CAR T cells in solid tumors has been limited. METHODS: This is a nonrandomized, open-label, single-site Phase I clinical trial. Three patients with EGFRvIII-positive recurrent glioblastoma were enrolled in a safety run-in cohort. Patients were treated with 10 million CARv3-TEAM-E T cells and monitored for toxic effects. Cerebrospinal fluid (CSF) and blood were sampled longitudinally and subjected to correlative analyses. RESULTS: No dose-limiting toxic effects were noted. Radiographic tumor regression occurred in all three patients within days after treatment, but this response was transient in two of the patients. Tumor regression correlated with decreased detection of antigen-specific RNA derived from extracellular vesicles (EVs) in both CSF and peripheral blood. CONCLUSIONS: Early data suggest safety and anti-tumor activity of CARv3-TEAM-E T cells in recurrent glioblastoma. EV-based liquid biopsy may assist in monitoring response to cell therapy. Ongoing enrollment has been modified to enhance durability using lymphodepletive chemotherapy. Additional arms will evaluate this approach in the setting of EGFRvIII-negative tumors and newly-diagnosed disease.

INTRODUCTION: Fetal ventriculomegaly, the most common antenatally-diagnosed brain abnormality, is the defining feature of congenital hydrocephalus (CH). Fetal ventriculomegaly is also an overlooked associated finding in neuropsychiatric disorders, including autism spectrum disorder (ASD), which is diagnosed at a 10-fold higher rate in CH patients than in the general population. METHODS: We subjected 2,978 parent-trio probands with primary ventriculomegaly, including shunted, sporadic CH, to whole exome sequencing (WES). Using mouse molecular genetics, we generated a novel CH mutant mouse model via prenatal, genetic deletion of a WES-identified CH gene. MRI, measurement of CSF secretion, and cortex-wide, mesoscopic Ca2+ imaging were performed in mice. RESULTS: We identify phosphatase and tensin homolog (PTEN) to be the most frequently mutated gene in primary human ventriculomegaly. Integrative analysis of the human fetal brain revealed PTEN was most highly expressed in NKX2.1+ neuroprogenitor cells (NPCs) and their post-natal interneuron descendants. Pten mutant mice with Nkx2.1-specific Pten deletion exhibited neonatal-onset obstructive hydrocephalus, resulting from aqueductal stenosis due to mTor-activated hyperproliferation of NPCs, and CSF hypersecretion due to inflammation-driven choroid plexus hyperplasia. Hydrocephalic Pten mutants also exhibit autism-like hypersynchronization of the somatosensory cortices due to impaired activity of interneurons. Strikingly, genetic or pharmacologic mTORC1 inhibition (everolimus) corrects ventriculomegaly and rescues cortical pathology of Pten mutants. CONCLUSIONS: Our data demonstrate that PTEN, a commonly mutated ASD gene, is also the most frequently mutated gene in primary ventriculomegaly. To attenuate the pathologically entangled enlargement of the ventricular system and intrinsic neuronal deficits within the surrounding cortical mantle, the use of rapamycin analogs has high translational potential as an adjunct therapy to neurosurgical CSF diversion in ventriculomegalic patients harboring PTEN mutations. Ventriculomegaly may also be a useful radiographic biomarker for early referral for exome sequencing and formal neurodevelopmental assessments.

Gene markers for Xenium in situ cell type identification.

Glioma represents the most common central nervous system neoplasm in adults. Current classification scheme utilizes molecular alterations, particularly IDH1.R132H, to stratify lesions into distinct prognostic groups. Identification of the single nucleotide variant through traditional tissue biopsy assessment poses procedural risks and does not fully reflect the heterogeneous and evolving tumor landscape. Here, we introduce a liquid biopsy assay, mt-IDH1dx. The blood-based test allows minimally invasive detection of tumor-derived extracellular vesicle RNA using only 2 ml plasma volume. We perform rigorous, blinded validation testing across the study population (n = 133), comprising of IDH1.R132H patients (n = 80), IDH1 wild-type gliomas (n = 44), and age matched healthy controls (n = 9). Results from our plasma testing demonstrate an overall sensitivity of 75.0% (95% CI: 64.1%-84.0%), specificity 88.7% (95% CI: 77.0%-95.7%), positive predictive value 90.9%, and negative predictive value 70.1% compared to the tissue gold standard. In addition to fundamental diagnostic applications, the study also highlights the utility of mt-IDH1dx platform for blood-based monitoring and surveillance, offering valuable prognostic information. Finally, the optimized workflow enables rapid and efficient completion of both tumor tissue and plasma testing in under 4 hours from the time of sampling.

Analysis of RNA Tirosh exhaustion and inhibitory scores expressed by T cells from cohort 1.