Understanding the association between fatigue and neurocognitive functioning in patients with glioma: A cross-sectional multinational study

Background

Fatigue and neurocognitive impairment are highly prevalent in patients with glioma, significantly impacting health-related quality of life. Despite the presumed association between these two factors, evidence remains sparse. Therefore, we aimed to investigate this relationship using multinational data.

Methods

We analyzed data on self-reported fatigue and neurocognitive outcomes from postoperative patients with glioma from the University of California San Francisco (n = 100, UCSF) and Amsterdam University Medical Center (n = 127, Amsterdam UMC). We used multiple linear regression models to assess associations between fatigue and seven (sub)domains of neurocognitive functioning and latent profile analysis to identify distinct patterns of fatigue and neurocognitive functioning.

Results

UCSF patients were older (median age 49 vs. 43 years, P = .002), had a higher proportion of grade 4 tumors (32% vs. 18%, P = .03), and had more neurocognitive deficits (P = .01). While the number of clinically fatigued patients was similar between sites (64% vs. 58%, P = .12), fatigue and the number of impaired neurocognitive domains were not correlated (P = .16-.72). At UCSF, neurocognitive domains were not related to fatigue, and at Amsterdam UMC attention and semantic fluency explained only 4-7% of variance in fatigue. Across institutions, we identified four distinct patterns of neurocognitive functioning, which were not consistently associated with fatigue.

Conclusions

Although individual patients might experience both fatigue and neurocognitive impairment, the relationship between the two is weak. Consequently, both fatigue and neurocognitive functioning should be independently assessed and treated with targeted therapies.

A prognostic neural epigenetic signature in high-grade glioma

Neural-tumor interactions drive glioma growth as evidenced in preclinical models, but clinical validation is limited. We present an epigenetically defined neural signature of glioblastoma that independently predicts patients' survival. We use reference signatures of neural cells to deconvolve tumor DNA and classify samples into low- or high-neural tumors. High-neural glioblastomas exhibit hypomethylated CpG sites and upregulation of genes associated with synaptic integration. Single-cell transcriptomic analysis reveals a high abundance of malignant stemcell-like cells in high-neural glioblastoma, primarily of the neural lineage. These cells are further classified as neural-progenitor-cell-like, astrocyte-like and oligodendrocyte-progenitor-like, alongside oligodendrocytes and excitatory neurons. In line with these findings, high-neural glioblastoma cells engender neuron-to-glioma synapse formation in vitro and in vivo and show an unfavorable survival after xenografting. In patients, a high-neural signature is associated with decreased overall and progression-free survival. High-neural tumors also exhibit increased functional connectivity in magnetencephalography and resting-state magnet resonance imaging and can be detected via DNA analytes and brain-derived neurotrophic factor in patients' plasma. The prognostic importance of the neural signature was further validated in patients diagnosed with diffuse midline glioma. Our study presents an epigenetically defined malignant neural signature in high-grade gliomas that is prognostically relevant. High-neural gliomas likely require a maximized surgical resection approach for improved outcomes.

Perivascular NOTCH3+ stem cells drive meningioma tumorigenesis and resistance to radiotherapy

Meningiomas are the most common primary intracranial tumors. Treatments for patients with meningiomas are limited to surgery and radiotherapy, and systemic therapies remain ineffective or experimental. Resistance to radiotherapy is common in high-grade meningiomas and the cell types and signaling mechanisms that drive meningioma tumorigenesis and resistance to radiotherapy are incompletely understood. Here we report NOTCH3 drives meningioma tumorigenesis and resistance to radiotherapy and find that perivascular NOTCH3+ stem cells are conserved across meningiomas from humans, dogs, and mice. Integrating single-cell transcriptomics with lineage tracing and imaging approaches in genetically engineered mouse models and xenografts, we show NOTCH3 drives tumor initiating capacity, cell proliferation, angiogenesis, and resistance to radiotherapy to increase meningioma growth and reduce survival. To translate these findings to patients, we show that an antibody stabilizing the extracellular negative regulatory region of NOTCH3 blocks meningioma tumorigenesis and sensitizes meningiomas to radiotherapy, reducing tumor growth and improving survival.

The Association Between Task Complexity and Cortical Language Mapping Accuracy

BACKGROUND AND OBJECTIVES

Direct cortical stimulation (DCS) mapping enables the identification of functional language regions within and around gliomas before tumor resection. Intraoperative mapping is required because glioma-infiltrated cortex engages in synchronous activity during task performance in a manner similar to normal-appearing cortex but has decreased ability to encode information for complex tasks. It is unknown whether task complexity influenced DCS mapping results. We aim to understand correlations between audiovisual picture naming (PN) task complexity and DCS error rate. We also asked what functional and oncological factors might be associated with higher rates of erroneous responses.

METHODS

We retrospectively reviewed intraoperative PN and word reading (WR) task performance during awake DCS language mapping for resection of dominant hemisphere World Health Organization grade 2 to 4 gliomas. The complexity of word tested in PN/WR tasks, patient characteristics, and tumor characteristics were compared between correct and incorrect trials.

RESULTS

Between 2017 and 2021, 74 patients met inclusion criteria. At median 18.6 months of follow-up, 73.0% were alive and 52.7% remained recurrence-free. A total of 2643 PN and 978 WR trials were analyzed. A greater number of syllables in PN was associated with a higher DCS error rate (P = .001). Multivariate logistic regression found that each additional syllable in PN tasks independently increased odds of error by 2.40 (P < .001). Older age was also an independent correlate of higher error rate (P < .043). World Health Organization grade did not correlate with error rate (P = .866). More severe language impairment before surgery correlated with worse performance on more complex intraoperative tasks (P < .001). A higher error rate on PN testing did not correlate with lower extent of glioma resection (P = .949).

CONCLUSION

Word complexity, quantified by the number of syllables, is associated with higher error rates for intraoperative PN tasks but does not affect extent of resection.

Quantitative Assessment of Preanalytic Variables on Clinical Evaluation of PI3/AKT/mTOR Signaling Activity in Diffuse Glioma

Biomarker-driven therapeutic clinical trials require the implementation of standardized, evidence-based practices for sample collection. In diffuse glioma, phosphatidylinositol 3 (PI3)-kinase/AKT/mTOR (PI3/AKT/mTOR) signaling is an attractive therapeutic target for which window-of-opportunity clinical trials could facilitate the identification of promising new agents. Yet, the relevant preanalytic variables and optimal tumor sampling methods necessary to measure pathway activity are unknown. To address this, we used a murine model for isocitrate dehydrogenase (IDH)-wildtype glioblastoma (GBM) and human tumor tissue, including IDH-wildtype GBM and IDH-mutant diffuse glioma. First, we determined the impact of delayed time-to-formalin fixation, or cold ischemia time (CIT), on the quantitative assessment of cellular expression of 6 phosphoproteins that are readouts of PI3K/AK/mTOR activity (phosphorylated-proline-rich Akt substrate of 40 kDa (p-PRAS40, T246), -mechanistic target of rapamycin (p-mTOR; S2448); -AKT (p-AKT, S473); -ribosomal protein S6 (p-RPS6, S240/244 and S235/236), and -eukaryotic initiation factor 4E-binding protein 1 (p-4EBP1, T37/46). With CITs ≥ 2 hours, typical of routine clinical handling, all had reduced or altered expression with p-RPS6 (S240/244) exhibiting relatively greater stability. A similar pattern was observed using patient tumor samples from the operating room with p-4EBP1 more sensitive to delayed fixation than p-RPS6 (S240/244). Many clinical trials utilize unstained slides for biomarker evaluation. Thus, we evaluated the impact of slide storage conditions on the detection of p-RPS6 (S240/244), p-4EBP1, and p-AKT. After 5 months, storage at -80°C was required to preserve the expression of p-4EBP1 and p-AKT, whereas p-RPS6 (240/244) expression was not stable regardless of storage temperature. Biomarker heterogeneity impacts optimal tumor sampling. Quantification of p-RPS6 (240/244) expression in multiple regionally distinct human tumor samples from 8 patients revealed significant intratumoral heterogeneity. Thus, the accurate assessment of PI3K/AKT/mTOR signaling in diffuse glioma must overcome intratumoral heterogeneity and multiple preanalytic factors, including time-to-formalin fixation, slide storage conditions, and phosphoprotein of interest.

Defining interventions and metrics to improve diversity in CNS clinical trial participation: A SNO and RANO effort

Despite major strides in cancer research and therapy, these advances have not been equitable across race and ethnicity. Historically marginalized groups (HMG) are more likely to have inadequate preventive screening, increased delays in diagnosis, and poor representation in clinical trials. Notably, Black, Hispanic, and Indigenous people represent 30% of the population but only 9% of oncology clinical trial participants. As a result, HMGs lack equitable access to novel therapies, contradicting the principle of distributive justice, as enshrined in the Belmont report, which demands the equitable selection of subjects in research involving human subjects. The lack of clinical trial diversity also leads to low generalizability and potentially harmful medical practices. Specifically, patients with brain cancer face unique barriers to clinical trial enrollment and completion due to disease-specific neurologic and treatment-induced conditions. Collectively, the intersection of these disease-specific conditions with social determinants of health fosters a lack of diversity in clinical trials. To ameliorate this disparity in neuro-oncology clinical trial participation, we present interventions focused on improving engagement of HMGs. Proposals range from inclusive trial design, decreasing barriers to care, expanding trial eligibility, access to tumor profiling for personalized medical trials, setting reasonable metrics and goals for accrual, working with patient community stakeholders, diversifying the neuro-oncology workforce, and development of tools to overcome biases with options to incentivize equity. The diversification of participation amongst neuro-oncology clinical trials is imperative. Equitable access and inclusion of HMG patients with brain tumors will not only enhance research discoveries but will also improve patient care.

Next Directions in the Neuroscience of Cancers Arising outside the CNS

SUMMARY

The field of cancer neuroscience has begun to define the contributions of nerves to cancer initiation and progression; here, we highlight the future directions of basic and translational cancer neuroscience for malignancies arising outside of the central nervous system.

Asleep triple-modality motor mapping for perirolandic gliomas: an update on outcomes

OBJECTIVE

Maximal safe resection of gliomas near motor pathways is facilitated by intraoperative mapping. Here, the authors review their results with triple-modality asleep motor mapping with motor evoked potentials and bipolar and monopolar stimulation for cortical and subcortical mapping during glioma surgery in an expanded cohort.

METHODS

This was a retrospective analysis of patients who underwent resection of a perirolandic glioma near motor pathways. Clinical and neuromonitoring data were extracted from the electronic medical records for review. All patients with new or worsened postoperative motor deficits were followed for at least 6 months. Regression analyses were performed to assess factors associated with a persistent motor deficit.

RESULTS

Between January 2018 and December 2021, 160 operations were performed in 151 patients with perirolandic glioma. Sixty-four patients (40%) had preoperative motor deficits, and the median extent of resection was 98%. Overall, patients in 38 cases (23.8%) had new or worse immediate postoperative deficits by discharge, and persistent deficits by 6 months were seen in 6 cases (3.8%), all in patients with high-grade gliomas. There were no new persistent deficits in low-grade glioma patients (0%). The risk factors for a persistent deficit included an insular tumor component (OR 8.6, p = 0.01), preoperative motor weakness (OR 8.1, p = 0.03), intraoperative motor evoked potential (MEP) changes (OR 36.5, p < 0.0001), and peri-resection cavity ischemia (OR 7.5, p = 0.04). Most persistent deficits were attributable to ischemic injury despite structural preservation of the descending motor tracts. For patients with persistent motor deficits, there were 3 cases (50%) in which a change in MEP was noted but subsequent subcortical monopolar stimulation still elicited a response in the corresponding muscle groups, suggesting axonal activation distal to a point of injury.

CONCLUSIONS

Asleep triple motor mapping results in a low rate of permanent deficits, especially for low-grade gliomas. Peri-resection cavity ischemia continues to be a significant risk factor for permanent deficit despite maintaining appropriate distance for subcortical tracts based on monopolar feedback.

A combinatory vaccine with IMA950 plus varlilumab promotes effector memory T-cell differentiation in the peripheral blood of patients with low-grade gliomas

BACKGROUND

Central nervous system (CNS) WHO grade 2 low-grade glioma (LGG) patients are at high risk for recurrence and with unfavorable long-term prognosis due to the treatment resistance and malignant transformation to high-grade glioma. Considering the relatively intact systemic immunity and slow-growing nature, immunotherapy may offer an effective treatment option for LGG patients.

METHODS

We conducted a prospective, randomized pilot study to evaluate the safety and immunological response of the multipeptide IMA950 vaccine with agonistic anti-CD27 antibody, varlilumab, in CNS WHO grade 2 LGG patients. Patients were randomized to receive combination therapy with IMA950 + poly-ICLC and varlilumab (Arm 1) or IMA950 + poly-ICLC (Arm 2) before surgery, followed by adjuvant vaccines.

RESULTS

A total of 14 eligible patients were enrolled in the study. Four patients received pre-surgery vaccines but were excluded from postsurgery vaccines due to the high-grade diagnosis of the resected tumor. No regimen-limiting toxicity was observed. All patients demonstrated a significant increase of anti-IMA950 CD8+ T-cell response postvaccine in the peripheral blood, but no IMA950-reactive CD8+ T cells were detected in the resected tumor. Mass cytometry analyses revealed that adding varlilumab promoted T helper type 1 effector memory CD4+ and effector memory CD8+ T-cell differentiation in the PBMC but not in the tumor microenvironment.

CONCLUSION

The combinational immunotherapy, including varlilumab, was well-tolerated and induced vaccine-reactive T-cell expansion in the peripheral blood but without a detectable response in the tumor. Further developments of strategies to overcome the blood-tumor barrier are warranted to improve the efficacy of immunotherapy for LGG patients.

Genomic alterations associated with postoperative nodular leptomeningeal disease after resection of brain metastases

OBJECTIVE

The relationship between brain metastasis resection and risk of nodular leptomeningeal disease (nLMD) is unclear. This study examined genomic alterations found in brain metastases with the aim of identifying alterations associated with postoperative nLMD in the context of clinical and treatment factors.

METHODS

A retrospective, single-center study was conducted on patients who underwent resection of brain metastases between 2014 and 2022 and had clinical and genomic data available. Postoperative nLMD was the primary endpoint of interest. Targeted next-generation sequencing of > 500 oncogenes was performed in brain metastases. Cox proportional hazards analyses were performed to identify clinical features and genomic alterations associated with nLMD.

RESULTS

The cohort comprised 101 patients with tumors originating from multiple cancer types. There were 15 patients with nLMD (14.9% of the cohort) with a median time from surgery to nLMD diagnosis of 8.2 months. Two supervised machine learning algorithms consistently identified CDKN2A/B codeletion and ERBB2 amplification as the top predictors associated with postoperative nLMD across all cancer types. In a multivariate Cox proportional hazards analysis including clinical factors and genomic alterations observed in the cohort, tumor volume (× 10 cm3; HR 1.2, 95% CI 1.01-1.5; p = 0.04), CDKN2A/B codeletion (HR 5.3, 95% CI 1.7-16.9; p = 0.004), and ERBB2 amplification (HR 3.9, 95% CI 1.1-14.4; p = 0.04) were associated with a decreased time to postoperative nLMD.

CONCLUSIONS

In addition to increased resected tumor volume, ERBB2 amplification and CDKN2A/B deletion were independently associated with an increased risk of postoperative nLMD across multiple cancer types. Additional work is needed to determine if targeted therapy decreases this risk in the postoperative setting.

Glioblastoma evolution and heterogeneity from a 3D whole-tumor perspective

Treatment failure for the lethal brain tumor glioblastoma (GBM) is attributed to intratumoral heterogeneity and tumor evolution. We utilized 3D neuronavigation during surgical resection to acquire samples representing the whole tumor mapped by 3D spatial coordinates. Integrative tissue and single-cell analysis revealed sources of genomic, epigenomic, and microenvironmental intratumoral heterogeneity and their spatial patterning. By distinguishing tumor-wide molecular features from those with regional specificity, we inferred GBM evolutionary trajectories from neurodevelopmental lineage origins and initiating events such as chromothripsis to emergence of genetic subclones and spatially restricted activation of differential tumor and microenvironmental programs in the core, periphery, and contrast-enhancing regions. Our work depicts GBM evolution and heterogeneity from a 3D whole-tumor perspective, highlights potential therapeutic targets that might circumvent heterogeneity-related failures, and establishes an interactive platform enabling 360° visualization and analysis of 3D spatial patterns for user-selected genes, programs, and other features across whole GBM tumors.

Does waiting for surgery matter? How time from diagnostic MRI to resection affects outcomes in newly diagnosed glioblastoma

OBJECTIVE

Maximal safe resection is the standard of care for patients presenting with lesions concerning for glioblastoma (GBM) on magnetic resonance imaging (MRI). Currently, there is no consensus on surgical urgency for patients with an excellent performance status, which complicates patient counseling and may increase patient anxiety. This study aims to assess the impact of time to surgery (TTS) on clinical and survival outcomes in patients with GBM.

METHODS

This is a retrospective study of 145 consecutive patients with newly diagnosed IDH-wild-type GBM who underwent initial resection at the University of California, San Francisco, between 2014 and 2016. Patients were grouped according to the time from diagnostic MRI to surgery (i.e., TTS): ≤ 7, > 7-21, and > 21 days. Contrast-enhancing tumor volumes (CETVs) were measured using software. Initial CETV (CETV1) and preoperative CETV (CETV2) were used to evaluate tumor growth represented as percent change (ΔCETV) and specific growth rate (SPGR; % growth/day). Overall survival (OS) and progression-free survival (PFS) were measured from the date of resection and were analyzed using the Kaplan-Meier method and Cox regression analyses.

RESULTS

Of the 145 patients (median TTS 10 days), 56 (39%), 53 (37%), and 36 (25%) underwent surgery ≤ 7, > 7-21, and > 21 days from initial imaging, respectively. Median OS and PFS among the study cohort were 15.5 and 10.3 months, respectively, and did not differ among the TTS groups (p = 0.81 and 0.17, respectively). Median CETV1 was 35.9, 15.7, and 10.2 cm3 across the TTS groups, respectively (p < 0.001). Preoperative biopsy and presenting to an outside hospital emergency department were associated with an average 12.79-day increase and 9.09-day decrease in TTS, respectively. Distance from the treating facility (median 57.19 miles) did not affect TTS. In the growth cohort, TTS was associated with an average 2.21% increase in ΔCETV per day; however, there was no effect of TTS on SPGR, Karnofsky Performance Status (KPS), postoperative deficits, survival, discharge location, or hospital length of stay. Subgroup analyses did not identify any high-risk groups for which a shorter TTS may be beneficial.

CONCLUSIONS

An increased TTS for patients with imaging concerning for GBM did not impact clinical outcomes, and while there was a significant association with ΔCETV, SPGR remained unaffected. However, SPGR was associated with a worse preoperative KPS, which highlights the importance of tumor growth speed over TTS. Therefore, while it is ill advised to wait an unnecessarily long time after initial imaging studies, these patients do not require urgent/emergency surgery and can seek tertiary care opinions and/or arrange for additional preoperative support/resources. Future studies are needed to explore subgroups for whom TTS may impact clinical outcomes.

"De novo replication repair deficient glioblastoma, IDH-wildtype" is a distinct glioblastoma subtype in adults that may benefit from immune checkpoint blockade

Glioblastoma is a clinically and molecularly heterogeneous disease, and new predictive biomarkers are needed to identify those patients most likely to respond to specific treatments. Through prospective genomic profiling of 459 consecutive primary treatment-naïve IDH-wildtype glioblastomas in adults, we identified a unique subgroup (2%, 9/459) defined by somatic hypermutation and DNA replication repair deficiency due to biallelic inactivation of a canonical mismatch repair gene. The deleterious mutations in mismatch repair genes were often present in the germline in the heterozygous state with somatic inactivation of the remaining allele, consistent with glioblastomas arising due to underlying Lynch syndrome. A subset of tumors had accompanying proofreading domain mutations in the DNA polymerase POLE and resultant "ultrahypermutation". The median age at diagnosis was 50 years (range 27-78), compared with 63 years for the other 450 patients with conventional glioblastoma (p < 0.01). All tumors had histologic features of the giant cell variant of glioblastoma. They lacked EGFR amplification, lacked combined trisomy of chromosome 7 plus monosomy of chromosome 10, and only rarely had TERT promoter mutation or CDKN2A homozygous deletion, which are hallmarks of conventional IDH-wildtype glioblastoma. Instead, they harbored frequent inactivating mutations in TP53, NF1, PTEN, ATRX, and SETD2 and recurrent activating mutations in PDGFRA. DNA methylation profiling revealed they did not align with known reference adult glioblastoma methylation classes, but instead had unique globally hypomethylated epigenomes and mostly classified as "Diffuse pediatric-type high grade glioma, RTK1 subtype, subclass A". Five patients were treated with immune checkpoint blockade, four of whom survived greater than 3 years. The median overall survival was 36.8 months, compared to 15.5 months for the other 450 patients (p < 0.001). We conclude that "De novo replication repair deficient glioblastoma, IDH-wildtype" represents a biologically distinct subtype in the adult population that may benefit from prospective identification and treatment with immune checkpoint blockade.

The surgical management of diffuse gliomas: Current state of neurosurgical management and future directions

After recent updates to the World Health Organization pathological criteria for diagnosing and grading diffuse gliomas, all major North American and European neuro-oncology societies recommend a maximal safe resection as the initial management of a diffuse glioma. For neurosurgeons to achieve this goal, the surgical plan for both low- and high-grade gliomas should be to perform a supramaximal resection when feasible based on preoperative imaging and the patient's performance status, utilizing every intraoperative adjunct to minimize postoperative neurological deficits. While the surgical approach and technique can vary, every effort must be taken to identify and preserve functional cortical and subcortical regions. In this summary statement on the current state of the field, we describe the tools and technologies that facilitate the safe removal of diffuse gliomas and highlight intraoperative and postoperative management strategies to minimize complications for these patients. Moreover, we discuss how surgical resections can go beyond cytoreduction by facilitating biological discoveries and improving the local delivery of adjuvant chemo- and radiotherapies.

Multi-class glioma segmentation on real-world data with missing MRI sequences: comparison of three deep learning algorithms

This study tests the generalisability of three Brain Tumor Segmentation (BraTS) challenge models using a multi-center dataset of varying image quality and incomplete MRI datasets. In this retrospective study, DeepMedic, no-new-Unet (nn-Unet), and NVIDIA-net (nv-Net) were trained and tested using manual segmentations from preoperative MRI of glioblastoma (GBM) and low-grade gliomas (LGG) from the BraTS 2021 dataset (1251 in total), in addition to 275 GBM and 205 LGG acquired clinically across 12 hospitals worldwide. Data was split into 80% training, 5% validation, and 15% internal test data. An additional external test-set of 158 GBM and 69 LGG was used to assess generalisability to other hospitals' data. All models' median Dice similarity coefficient (DSC) for both test sets were within, or higher than, previously reported human inter-rater agreement (range of 0.74-0.85). For both test sets, nn-Unet achieved the highest DSC (internal = 0.86, external = 0.93) and the lowest Hausdorff distances (10.07, 13.87 mm, respectively) for all tumor classes (p < 0.001). By applying Sparsified training, missing MRI sequences did not statistically affect the performance. nn-Unet achieves accurate segmentations in clinical settings even in the presence of incomplete MRI datasets. This facilitates future clinical adoption of automated glioma segmentation, which could help inform treatment planning and glioma monitoring.

Epigenetic neural glioblastoma enhances synaptic integration and predicts therapeutic vulnerability

Neural-tumor interactions drive glioma growth as evidenced in preclinical models, but clinical validation is nascent. We present an epigenetically defined neural signature of glioblastoma that independently affects patients' survival. We use reference signatures of neural cells to deconvolve tumor DNA and classify samples into low- or high-neural tumors. High-neural glioblastomas exhibit hypomethylated CpG sites and upregulation of genes associated with synaptic integration. Single-cell transcriptomic analysis reveals high abundance of stem cell-like malignant cells classified as oligodendrocyte precursor and neural precursor cell-like in high-neural glioblastoma. High-neural glioblastoma cells engender neuron-to-glioma synapse formation in vitro and in vivo and show an unfavorable survival after xenografting. In patients, a high-neural signature associates with decreased survival as well as increased functional connectivity and can be detected via DNA analytes and brain-derived neurotrophic factor in plasma. Our study presents an epigenetically defined malignant neural signature in high-grade gliomas that is prognostically relevant.

Glioma-neuronal circuit remodeling induces regional immunosuppression

Neuronal activity-driven mechanisms impact glioblastoma cell proliferation and invasion 1-7 , and glioblastoma remodels neuronal circuits 8,9 . Distinct intratumoral regions maintain functional connectivity via a subpopulation of malignant cells that mediate tumor-intrinsic neuronal connectivity and synaptogenesis through their transcriptional programs 8 . However, the effects of tumor-intrinsic neuronal activity on other cells, such as immune cells, remain unknown. Here we show that regions within glioblastomas with elevated connectivity are characterized by regional immunosuppression. This was accompanied by different cell compositions and inflammatory status of tumor-associated macrophages (TAMs) in the tumor microenvironment. In preclinical intracerebral syngeneic glioblastoma models, CRISPR/Cas9 gene knockout of Thrombospondin-1 (TSP-1/ Thbs1 ), a synaptogenic factor critical for glioma-induced neuronal circuit remodeling, in glioblastoma cells suppressed synaptogenesis and glutamatergic neuronal hyperexcitability, while simultaneously restoring antigen-presentation and pro-inflammatory responses. Moreover, TSP-1 knockout prolonged survival of immunocompetent mice harboring intracerebral syngeneic glioblastoma, but not of immunocompromised mice, and promoted infiltrations of pro-inflammatory TAMs and CD8+ T-cells in the tumor microenvironment. Notably, pharmacological inhibition of glutamatergic excitatory signals redirected tumor-associated macrophages toward a less immunosuppressive phenotype, resulting in prolonged survival. Altogether, our results demonstrate previously unrecognized immunosuppression mechanisms resulting from glioma-neuronal circuit remodeling and suggest future strategies targeting glioma-neuron-immune crosstalk may open up new avenues for immunotherapy.

Association of CDKN2A alterations with increased postoperative seizure risk after resection of brain metastases

OBJECTIVE

Seizures are common and significantly disabling for patients with brain metastases (BMs). Although resection can provide seizure control, a subset of patients with BMs may continue to suffer seizures postoperatively. Genomic BM characteristics may influence which patients are at risk for postoperative seizures. This work explores correlations between genomic alterations and risk of postoperative seizures following BM resection.

METHODS

All patients underwent BM resection at a single institution, with available clinical and sequencing data on more than 500 oncogenes. Clinical seizures were documented pre- and postoperatively. A random forest machine learning classification was used to determine candidate genomic alterations associated with postoperative seizures, and clinical and top genomic variables were correlated with postoperative seizures by using Cox proportional hazards models.

RESULTS

There were 112 patients with BMs who underwent 114 surgeries and had at least 1 month of postoperative follow-up. Seizures occurred preoperatively in 26 (22.8%) patients and postoperatively in 25 (21.9%). The Engel classification achieved at 6 months for those with preoperative seizures was class I in 13 (50%); class II in 6 (23.1%); class III in 5 (19.2%), and class IV in 2 (7.7%). In those with postoperative seizures, only 8 (32.0%) had seizures preoperatively, and preoperative seizures were not a significant predictor of postoperative seizures (HR 1.84; 95% CI 0.79-4.37; p = 0.156). On random forest classification and multivariate Cox analysis controlling for factors including recurrence, extent of resection, and number of BMs, CDKN2A alterations were associated with postoperative seizures (HR 3.22; 95% CI 1.27-8.16; p = 0.014). Melanoma BMs were associated with higher risk of postoperative seizures compared with all other primary malignancies (HR 5.23; 95% CI 1.37-19.98; p = 0.016). Of 39 BMs with CDKN2A alteration, 35.9% (14/39) had postoperative seizures, compared to 14.7% (11/75) without CDKN2A alteration. The overall rate of postoperative seizures in melanoma BMs was 42.9% (15/35), compared with 12.7% (10/79) for all other primary malignancies.

CONCLUSIONS

CDKN2A alterations and melanoma primary malignancy are associated with increased postoperative seizure risk following resection of BMs. These results may help guide postoperative seizure prophylaxis in patients undergoing resection of BMs.

Identification of risk factors associated with leptomeningeal disease after resection of brain metastases

OBJECTIVE

Resection of brain metastases (BMs) may be associated with increased risk of leptomeningeal disease (LMD). This study examined rates and predictors of LMD, including imaging subtypes, in patients who underwent resection of a BM followed by postoperative radiation.

METHODS

A retrospective, single-center study was conducted examining overall LMD, classic LMD (cLMD), and nodular LMD (nLMD) risk. Logistic regression, Cox proportional hazards, and random forest analyses were performed to identify risk factors associated with LMD.

RESULTS

Of the 217 patients in the cohort, 47 (21.7%) developed postoperative LMD, with 19 cases (8.8%) of cLMD and 28 cases (12.9%) of nLMD. Six-, 12-, and 24-month LMD-free survival rates were 92.3%, 85.6%, and 71.4%, respectively. Patients with cLMD had worse survival outcomes from the date of LMD diagnosis compared with nLMD (median 2.4 vs 6.9 months, p = 0.02, log-rank test). Cox proportional hazards analysis identified cerebellar/insular/occipital location (hazard ratio [HR] 3.25, 95% confidence interval [CI] 1.73-6.11, p = 0.0003), absence of extracranial disease (HR 2.49, 95% CI 1.27-4.88, p = 0.008), and ventricle contact (HR 2.82, 95% CI 1.5-5.3, p = 0.001) to be associated with postoperative LMD. A predictive model using random forest analysis with an area under the receiver operating characteristic curve of 0.87 in a test cohort identified tumor location, systemic disease status, and tumor volume as the most important factors associated with LMD.

CONCLUSIONS

Tumor location, absence of extracranial disease at the time of surgery, ventricle contact, and increased tumor volume were associated with LMD. Further work is needed to determine whether escalating therapies in patients at risk of LMD prevents disease dissemination.

Genomic alterations associated with rapid progression of brain metastases

OBJECTIVE

The aim of this study was to investigate associations between genomic alterations in resected brain metastases and rapid local and distant CNS recurrence identified at the time of postoperative adjuvant radiosurgery.

METHODS

This was a retrospective study on patients who underwent resection of intracranial brain metastases. Next-generation sequencing of more than 500 coding genes was performed on brain metastasis specimens. Postoperative and preradiosurgery MR images were compared to identify rapid recurrence. Genomic data were associated with rapid local and distant CNS recurrence of brain metastases using nominal regression analyses.

RESULTS

The cohort contained 92 patients with 92 brain metastases. Thirteen (14.1%) patients had a rapid local recurrence, and 64 (69.6%) patients had rapid distant CNS progression by the time of postoperative adjuvant radiosurgery, which occurred in a median time of 25 days (range 3-85 days) from surgery. RB1 and CTNNB1 mutations were seen in 8.7% and 9.8% of the cohort, respectively, and were associated with a significantly higher risk of rapid local recurrence (RB1: OR 13.6, 95% CI 2.0-92.39, p = 0.008; and CTNNB1: OR 11.97, 95% CI 2.25-63.78, p = 0.004) on multivariate analysis. No genes were found to be associated with rapid distant CNS progression. However, the presence of extracranial disease was significantly associated with a higher risk of rapid distant recurrence on multivariate analysis (OR 4.06, 95% CI 1.08-15.34, p = 0.039).

CONCLUSIONS

Genomic alterations in RB1 or CTNNB1 were associated with a significantly higher risk of rapid recurrence at the resection site. Although no genomic alterations were associated with rapid distant recurrence, having active extracranial disease was a risk factor for new lesions by the time of adjuvant radiotherapy after resection.

Molecular Features of Resected Melanoma Brain Metastases, Clinical Outcomes, and Responses to Immunotherapy

Importance

Central nervous system (CNS)-penetrant systemic therapies have significantly advanced care for patients with melanoma brain metastases. However, improved understanding of the molecular landscape and microenvironment of these lesions is needed to both optimize patient selection and advance treatment approaches.

Objective

To evaluate how bulk and single-cell genomic features of melanoma brain metastases are associated with clinical outcome and treatment response.

Design, Setting, and Participants

This cohort study analyzed bulk DNA sequencing and single nuclear RNA-sequencing data from resected melanoma brain metastases and included 94 consecutive patients with a histopathologically confirmed diagnosis of melanoma brain metastasis who underwent surgical resection at a single National Comprehensive Cancer Network cancer center in San Francisco, California, from January 1, 2009, to December 31, 2022.

Exposure

A Clinical Laboratory Improvement Amendments-certified targeted sequencing assay was used to analyze tumor resection specimens, with a focus on BRAF V600E alteration. For frozen pathologic specimens from CNS treatment-naive patients undergoing surgical resection, commercial single nuclear RNA sequencing approaches were used.

Main Outcomes and Measures

The primary outcome was overall survival (OS). Secondary outcomes included CNS progression-free survival (PFS), microenvironmental composition with decreased T-cell and macrophage populations, and responses to immunotherapy.

Results

To correlate molecular status with clinical outcome, Kaplan-Meier survival analysis of 94 consecutive patients (median age, 64 years [range, 24-82 years]; 70 men [74%]) with targeted BRAF alteration testing showed worse median intracranial PFS (BRAF variant: 3.6 months [IQR, 0.1-30.6 months]; BRAF wildtype: 11.0 months [IQR, 0.8-81.5 months]; P < .001) and OS (BRAF variant: 9.8 months [IQR, 2.5-69.4 months]; BRAF wildtype: 23.2 months [IQR, 1.1-102.5 months]; P = .005; log-rank test) in BRAF V600E variant tumors. Multivariable Cox proportional hazards regression analysis revealed that BRAF V600E status was an independent variable significantly associated with both PFS (hazard ratio [HR], 2.65; 95% CI, 1.54-4.57; P < .001) and OS (HR, 1.96; 95% CI, 1.08-3.55; P = .03). For the 45 patients with resected melanoma brain metastases undergoing targeted DNA sequencing, molecular classification recapitulated The Cancer Genome Atlas groups (NRAS variant, BRAF variant, NF1 variant, and triple wildtype) with no subtype enrichment within the brain metastasis cohort. On a molecular level, BRAF V600E variant lesions were found to have a significantly decreased tumor mutation burden. Moreover, single nuclear RNA sequencing of treatment-naive BRAF V600E variant (n = 3) brain metastases compared with BRAF wildtype (n = 3) brain metastases revealed increased immune cell populations in BRAF wildtype tumors (mean [SD], 11% [4.1%] vs 3% [1.6%] CD45-positive cells; P = .04). Survival analysis of postoperative immunotherapy responses by BRAF status revealed that BRAF wildtype lesions were associated with a response to checkpoint inhibition (median OS: with immunotherapy, undefined; without immunotherapy, 13.0 months [range, 1.1-61.7 months]; P = .001; log-rank test) while BRAF variant lesions (median OS: with immunotherapy, 9.8 months [range, 2.9-39.8 months]; without immunotherapy, 9.5 months [range, 2.5-67.2 months]; P = .81; log-rank test) were not.

Conclusions and Relevance

This molecular analysis of patients with resected melanoma brain metastases found that BRAF V600E alteration is an important translational biomarker associated with worse clinical outcomes, differential microenvironmental composition, and benefit from immunotherapy. Patients with BRAF V600E variant melanoma brain metastases may thus benefit from alternative CNS-penetrant systemic regimens.

NOTCH3 drives meningioma tumorigenesis and resistance to radiotherapy

Meningiomas are the most common primary intracranial tumors1-3. Treatments for patients with meningiomas are limited to surgery and radiotherapy, and systemic therapies remain ineffective or experimental4,5. Resistance to radiotherapy is common in high-grade meningiomas6, and the cell types and signaling mechanisms driving meningioma tumorigenesis or resistance to radiotherapy are incompletely understood. Here we report NOTCH3 drives meningioma tumorigenesis and resistance to radiotherapy and find NOTCH3+ meningioma mural cells are conserved across meningiomas from humans, dogs, and mice. NOTCH3+ cells are restricted to the perivascular niche during meningeal development and homeostasis and in low-grade meningiomas but are expressed throughout high-grade meningiomas that are resistant to radiotherapy. Integrating single-cell transcriptomics with lineage tracing and imaging approaches across mouse genetic and xenograft models, we show NOTCH3 drives tumor initiating capacity, cell proliferation, angiogenesis, and resistance to radiotherapy to increase meningioma growth and reduce survival. An antibody stabilizing the extracellular negative regulatory region of NOTCH37,8 blocks meningioma tumorigenesis and sensitizes meningiomas to radiotherapy, reducing tumor growth and improving survival in preclinical models. In summary, our results identify a conserved cell type and signaling mechanism that underlie meningioma tumorigenesis and resistance to radiotherapy, revealing a new therapeutic vulnerability to treat meningiomas that are resistant to standard interventions.

Glioblastoma remodelling of human neural circuits decreases survival

Gliomas synaptically integrate into neural circuits1,2. Previous research has demonstrated bidirectional interactions between neurons and glioma cells, with neuronal activity driving glioma growth1-4 and gliomas increasing neuronal excitability2,5-8. Here we sought to determine how glioma-induced neuronal changes influence neural circuits underlying cognition and whether these interactions influence patient survival. Using intracranial brain recordings during lexical retrieval language tasks in awake humans together with site-specific tumour tissue biopsies and cell biology experiments, we find that gliomas remodel functional neural circuitry such that task-relevant neural responses activate tumour-infiltrated cortex well beyond the cortical regions that are normally recruited in the healthy brain. Site-directed biopsies from regions within the tumour that exhibit high functional connectivity between the tumour and the rest of the brain are enriched for a glioblastoma subpopulation that exhibits a distinct synaptogenic and neuronotrophic phenotype. Tumour cells from functionally connected regions secrete the synaptogenic factor thrombospondin-1, which contributes to the differential neuron-glioma interactions observed in functionally connected tumour regions compared with tumour regions with less functional connectivity. Pharmacological inhibition of thrombospondin-1 using the FDA-approved drug gabapentin decreases glioblastoma proliferation. The degree of functional connectivity between glioblastoma and the normal brain negatively affects both patient survival and performance in language tasks. These data demonstrate that high-grade gliomas functionally remodel neural circuits in the human brain, which both promotes tumour progression and impairs cognition.

Interactive Effects of Molecular, Therapeutic, and Patient Factors on Outcome of Diffuse Low-Grade Glioma

PURPOSE

In patients with diffuse low-grade glioma (LGG), the extent of surgical tumor resection (EOR) has a controversial role, in part because a randomized clinical trial with different levels of EOR is not feasible.

METHODS

In a 20-year retrospective cohort of 392 patients with IDH-mutant grade 2 glioma, we analyzed the combined effects of volumetric EOR and molecular and clinical factors on overall survival (OS) and progression-free survival by recursive partitioning analysis. The OS results were validated in two external cohorts (n = 365). Propensity score analysis of the combined cohorts (n = 757) was used to mimic a randomized clinical trial with varying levels of EOR.

RESULTS

Recursive partitioning analysis identified three survival risk groups. Median OS was shortest in two subsets of patients with astrocytoma: those with postoperative tumor volume (TV) > 4.6 mL and those with preoperative TV > 43.1 mL and postoperative TV ≤ 4.6 mL. Intermediate OS was seen in patients with astrocytoma who had chemotherapy with preoperative TV ≤ 43.1 mL and postoperative TV ≤ 4.6 mL in addition to oligodendroglioma patients with either preoperative TV > 43.1 mL and residual TV ≤ 4.6 mL or postoperative residual volume > 4.6 mL. Longest OS was seen in astrocytoma patients with preoperative TV ≤ 43.1 mL and postoperative TV ≤ 4.6 mL who received no chemotherapy and oligodendroglioma patients with preoperative TV ≤ 43.1 mL and postoperative TV ≤ 4.6 mL. EOR ≥ 75% improved survival outcomes, as shown by propensity score analysis.

CONCLUSION

Across both subtypes of LGG, EOR beginning at 75% improves OS while beginning at 80% improves progression-free survival. Nonetheless, maximal resection with preservation of neurological function remains the treatment goal. Our findings have implications for surgical strategies for LGGs, particularly oligodendroglioma.

Abstract 2497: Glioma-induced neuronal remodeling promotes regional immunosuppression

Recent studies have elucidated that gliomas remodel neuronal circuits, and distinct intratumoral regions maintain functional connectivity through a subpopulation of synaptogenic malignant cells expressing Thrombospondin-1 (TSP-1, encoded by the THBS1 gene). Single-cell RNA sequencing analyses of our primary glioblastoma patient samples identified a significant downregulation of immune response signatures in myeloid cells and lymphoid cells in functionally connected intratumoral regions characterized by upregulated THBS1. Understanding the biological significance of immunosuppression within functionally connected intratumoral regions may uncover therapeutic vulnerabilities. Here, we investigate glioma-neuronal-immune crosstalk across clinical tumor specimens and preclinical syngeneic models through bulk and single-cell RNA sequencing (13,670 cells), flow cytometry, and spatial transcriptomics. Using an SB28 murine glioma cell line that endogenously expresses Thbs1 at high levels, we generated a CRISPR Thbs1-knock-out (KO) cell line. In bulk RNA-sequencing data performed on orthotopic tumor models, Thbs1-KO tumors were characterized by the downregulated expression of synapse-associated genes and synaptogenic factors and the recovered expression of genes related to immune response. It highlighted the important role of Thrombospondin-1 in synaptogenesis and immunosuppression consistent with the observation from our primary patient tumor samples. Flow cytometry revealed that tumor-associated macrophages isolated from Thbs1-KO tumors were more polarized toward the pro-inflammatory “M1-like” phenotype (median M1/M2 ratio = 0.6 [WT] vs. 1.34 [KO], p &lt; 0.006). Unbiased gene expression program analysis using spatial transcriptomics for in vivo tumor-harboring mouse brains demonstrated a significant negative correlation between signatures of synaptogenesis (represented by Ntng1 and Nlgn3 genes) and those of immune response (represented by Nfkb1 and Cd83 genes). SB28-Thbs1-KO syngeneic models demonstrated slower tumor growth and significantly longer survival compared to Thbs1-WT counterparts (19 days [WT] vs. 25 days [KO], p &lt; 4.5E-5). The survival difference was abrogated in B6-SCID immunodeficient mice, indicating the critical role of adaptive immunity in the survival advantage associated with Thrombospondin-1 downregulation. Our results identify previously unknown immunosuppression mechanisms in the context of glioma-induced intratumoral connectivity via Thrombospondin-1. Future therapeutic strategies targeting this glioma-neuronal-immune crosstalk may open up new avenues for glioblastoma immunotherapy.

Citation Format: Takahide Nejo, Saritha Krishna, Christian Jimenez, Akane Yamamichi, Jacob S. Young, Tiffany Chen, Senthilnath Lakshmanachetty, Payal Watchmaker, Abrar Choudhury, Hirokazu Ogino, David R. Raleigh, Shawn L. Hervey-Jumper, Hideho Okada. Glioma-induced neuronal remodeling promotes regional immunosuppression [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2497.

102 AI-Based Molecular Classification of Diffuse Gliomas using Rapid, Label-Free Optical Imaging

INTRODUCTION

Molecular classification has transformed the management of brain tumors by enabling more accurate prognostication and personalized treatment. Access to timely molecular diagnostic testing for brain tumor patients is limited, complicating surgical and adjuvant treatment and obstructing clinical trial enrollment.

METHODS

By combining stimulated Raman histology (SRH), a rapid, label-free, non-consumptive, optical imaging method, and deep learning-based image classification, we are able to predict the molecular genetic features used by the World Health Organization (WHO) to define the adult-type diffuse glioma taxonomy, including IDH-1/2, 1p19q-codeletion, and ATRX loss. We developed a multimodal deep neural network training strategy that uses both SRH images and large-scale, public diffuse glioma genomic data (i.e. TCGA, CGGA, etc.) in order to achieve optimal molecular classification performance.

RESULTS

One institution was used for model training (University of Michigan) and four institutions (NYU, UCSF, Medical University of Vienna, and University Hospital Cologne) were included for patient enrollment in the prospective testing cohort. Using our system, called DeepGlioma, we achieved an average molecular genetic classification accuracy of 93.2% and identified the correct diffuse glioma molecular subgroup with 91.5% accuracy within 2 minutes in the operating room. DeepGlioma outperformed conventional IDH1-R132H immunohistochemistry (94.2% versus 91.4% accuracy) as a first-line molecular diagnostic screening method for diffuse gliomas and can detect canonical and non-canonical IDH mutations.

CONCLUSIONS

Our results demonstrate how artificial intelligence and optical histology can be used to provide a rapid and scalable alternative to wet lab methods for the molecular diagnosis of brain tumor patients during surgery.

Preclinical modeling of lower-grade gliomas

Models for human gliomas prove critical not only to advancing our understanding of glioma biology but also to facilitate the development of therapeutic modalities. Specifically, creating lower-grade glioma (LGG) models has been challenging, contributing to few investigations and the minimal progress in standard treatment over the past decade. In order to reliably predict and validate the efficacies of novel treatments, however, LGG models need to adhere to specific standards that recapitulate tumor genetic aberrations and micro-environment. This underscores the need to revisit existing models of LGG and explore prospective models that may bridge the gap between preclinical insights and clinical translation. This review first outlines a set of criteria aimed to address the current challenges hindering model development. We then evaluate the strengths and weaknesses of existing preclinical models of LGG with respect to these established standards. To conclude, the review discusses potential future directions for integrating existing models to maximize the exploration of disease mechanisms and therapeutics development.

CDKN2A/B co-deletion is associated with increased risk of local and distant intracranial recurrence after surgical resection of brain metastases

Background

While genetic alterations in brain metastases (BMs) have been previously explored, there are limited data examining their association with recurrence after surgical resection. This study aimed to identify genetic alterations within BMs associated with CNS recurrence after surgery across multiple cancer types.

Methods

A retrospective, single-center study was conducted with patients who underwent resection of a BM with available clinical and gene sequencing data available. Local and remote CNS recurrence were the primary study outcomes. Next-generation sequencing of the coding regions in over 500 oncogenes was performed in brain metastasis specimens. Cox proportional hazards analyses were performed to identify clinical features and genomic alterations associated with CNS recurrence.

Results

A total of 90 patients undergoing resection of 91 BMs composed the cohort. Genes most frequently mutated in the cohort included TP53 (64%), CDKN2A (37%), TERT (29%), CDKN2B (23%), NF1 (14%), KRAS (14%), and PTEN (13%), all of which occurred across multiple cancer types. CDKN2A/B co-deletion was seen in 21 (23.1%) brain metastases across multiple cancer types. In multivariate Cox proportional hazard analyses including patient, tumor, and treatment factors, CDKN2A/B co-deletion in the brain metastasis was associated with increased risk of local (HR 4.07, 95% CI 1.32-12.54, P = 0.014) and remote (HR 2.28, 95% CI 1.11-4.69, P = 0.025) CNS progression. Median survival and length of follow-up were not different based on CDKN2A/B mutation status.

Conclusions

CDKN2A/B co-deletion detected in BMs is associated with increased CNS recurrence after surgical resection. Additional work is needed to determine whether more aggressive treatment in patients with this mutation may improve outcomes.

Adverse radiation effect and freedom from progression following repeat stereotactic radiosurgery for brain metastases

OBJECTIVE

The authors previously evaluated risk and time course of adverse radiation effects (AREs) following stereotactic radiosurgery (SRS) for brain metastases, excluding lesions treated after prior SRS. In the present analysis they focus specifically on single-fraction salvage SRS to brain metastases previously treated with SRS or hypofractionated SRS (HFSRS), evaluating freedom from progression (FFP) and the risk and time course of AREs.

METHODS

Brain metastases treated from September 1998 to May 2019 with single-fraction SRS after prior SRS or HFSRS were analyzed. Serial follow-up magnetic resonance imaging (MRI) and surgical pathology reports were reviewed to score local treatment failure and AREs. The Kaplan-Meier method was used to estimate FFP and risk of ARE measured from the date of repeat SRS with censoring at the last brain MRI.

RESULTS

A total of 229 retreated brain metastases in 124 patients were evaluable. The most common primary cancers were breast, lung, and melanoma. The median interval from prior SRS/HFSRS to repeat SRS was 15.4 months, the median prescription dose was 18 Gy, and the median duration of follow-up imaging was 14.5 months. At 1 year after repeat SRS, FFP was 80% and the risk of symptomatic ARE was 11%. The 1-year risk of imaging changes, including asymptomatic RE and symptomatic ARE, was 30%. Among lesions that demonstrated RE, the median time to onset was 6.7 months (IQR 4.7-9.9 months) and the median time to peak imaging changes was 10.1 months (IQR 5.6-13.6 months). Lesion size by quadratic mean diameter (QMD) showed similar results for QMDs ranging from 0.75 to 2.0 cm (1-year FFP 82%, 1-year risk of symptomatic ARE 11%). For QMD < 0.75 cm, the 1-year FFP was 86% and the 1-year risk of symptomatic ARE was only 2%. Outcomes were worse for QMDs 2.01-3.0 cm (1-year FFP 65%, 1-year risk of symptomatic ARE 24%). The risk of symptomatic ARE was not increased with tyrosine kinase inhibitors or immunotherapy before or after repeat SRS.

CONCLUSIONS

RE on imaging was common after repeat SRS (30% at 1 year), but the risk of a symptomatic ARE was much less (11% at 1 year). The results of repeat single-fraction SRS were good for brain metastases ≤ 2 cm. The authors recommend an interval ≥ 6 months from prior SRS and a prescription dose ≥ 18 Gy. Alternatives such as HFSRS, laser interstitial thermal therapy, or resection with adjuvant radiation should be considered for recurrent brain metastases > 2 cm.

CNSC-07. MECHANISMS OF GLIOBLASTOMA-INDUCED CORTICAL REMODELING IDENTIFY THERAPEUTIC VULNERABILITIES

Prior work demonstrated synaptic integration of malignant gliomas into neural circuits induces local hyperexcitability and tumor proliferation. However, prognostication and therapeutic vulnerabilities are lacking from preclinical models. Here, we integrate in vivo and in vitro neurophysiology spatially matched with gene expression programs and protein signaling mechanisms across 66 IDH WT glioblastoma patients to identify thrombospondin-1 (TSP-1) as a molecular driver of glioma-induced network remodeling. Bulk and single cell RNA-sequencing of 11 intratumoral regions maintaining functional connectivity (13,730 cells analyzed) revealed a distinct neurogenic signature enriched for the synaptogenic factor TSP-1. Mechanistic and functional studies validating therapeutic vulnerabilities to TSP-1 silencing by shRNA knockdown and FDA-approved inhibitors (Gabapentin and LSKL) of excitatory synapse formation through the gabapentin a2d-1 receptor was performed in vitro and in vivo. Glioma-neuron co-culture of TSP-1 overexpressing cells demonstrated increased Ki67 proliferation and tumor microtube (TMT) formation when cultured in the presence of neurons. Pharmacological inhibition of TSP-1 using gabapentin or TSP-1 shRNA inhibited the proliferation and TMT-mediated expansion. Hippocampal xenografted mice with TSP-1 over expressing primary patient cultures demonstrated shorter survival and gabapentin treatment of xenografted mice significantly reduced the proliferation of TSP-1 overexpressing cells in vivo. Electrophysiological properties of glioma-neuron co-cultures analyzed using multi-electrode array (MEA) demonstrated increased neuronal spiking activity and network burst synchrony in the presence of TSP-1 over expressing cells. Strikingly, these increases were eliminated in the presence of gabapentin. We modeled survival risk in patients incorporating the effects of glioma intrinsic neuronal activity, molecular, therapeutic, and clinical factors on overall survival by recursive partitioning. Three risk groups were identified based on tumor intrinsic neuronal activity, pre- and post-operative tumor volume with shortest overall survival in patients with glioma intrinsic neuronal activity. These data identify glioma-induced secretion of TSP-1 as a key contributor of tumor proliferation shedding light on new therapies.

BIOM-02. MUTATIONAL ANALYSIS AND SINGLE CELL SEQUENCING OF MELANOMA BRAIN METASTASES REVEALS BRAF STATUS CORRELATES WITH CLINICAL OUTCOME AND DIFFERENTIAL IMMUNE POPULATIONS

Understanding the molecular landscape and microenvironment of melanoma brain metastases is critical to devise improved treatments. Here, we perform bulk and single cell genomic analysis of melanoma brain metastases to identify molecular correlates of clinical outcome. 84 consecutive patients who underwent surgical resection at a single institution with a histo-pathologically confirmed diagnosis of melanoma brain metastasis were retrospectively identified. In 60 patients (71%) with sufficient brain metastasis tissue for targeted next generation sequencing, DNA mutations were assessed with a CLIA certified sequencing assay. Single nuclear RNA sequencing using the 10x platform was performed on n=6 samples from treatment naïve patients. Overall survival (OS) and CNS progression free survival (CNS PFS) from time of brain metastasis diagnosis were estimated using the Kaplan-Meier method. The median patient age was 62 years old (range: 25-78 years), and the median clinical follow up was 17 months. A total of 33 patients (39%) had BRAFV600E melanoma brain metastases. Multivariate analysis incorporating age, performance status, and extracranial disease revealed BRAF status was an independent prognostic factor for OS (p&lt; 0.05). In patients undergoing targeted next generation sequencing, the most common pathogenic variant was TERT promoter mutation (n=44; 73%). With regard to TCGA molecular melanoma subgroups, NRAS mutant (n=22; 37%) brain metastases were most common followed by BRAF mutant (n=20; 33%), NF1 mutant (n=11; 18%), and triple wildtype (n=7; 12%). Evaluation of clinical outcomes in the context of next generation sequencing results revealed no differences by TERT status but demonstrated worse overall survival in the BRAF mutant molecular group (p&lt; 0.01, log-rank test). Single nuclear sequencing of 36,115 nuclei across 6 samples revealed BRAF wildtype tumors exhibit greater infiltrating immune cell populations including microglia and T cell subtypes. Future work will require integration of these findings with different systemic therapy regimens and across larger, prospective, multi-institutional cohorts.

NIMG-46. TOWARDS PREDICTING TUMOR AGGRESSIVENESS WITH RADIOPATHOMIC ANALYSIS OF MULTI-PARAMETRIC ANATOMICAL, DIFFUSION-WEIGHTED, AND METABOLIC MRI IN PATIENTS WITH NEWLY-DIAGNOSED GLIOMAS

INTRODUCTION

Pathologically aggressive tumor biology can extend beyond the contrast-enhancing or non-enhancing anatomical lesions in patients with glioma. Identification of malignant regions can help guide diagnosis and subsequent treatment planning. This study leverages a unique multi-parametric MRI dataset with tissue samples of known spatial coordinates to noninvasively predict cellular proliferation (KI-67) and a novel index of tumor aggressiveness (TAI), that combines proliferation, cellularity, and tumor-score.

METHODS

420 tissue samples were collected from 162 patients with newly-diagnosed glioma (47% IDH-wildtype). Clinical imaging consisted of T2-weighted, T2-FLAIR, T1-weighted pre- and post-contrast images, and apparent diffusion coefficient (ADC) and fractional anisotropy (FA) from diffusion-weighted imaging. Mean normalized imaging metrics were quantified from 5mm spheres centered at the location of the tissue sample. A single spectrum was reconstructed at the location of each tissue sample from 3D 1H-MR Spectroscopic Imaging (MRSI) before quantifying normalized metabolite peak-heights for choline, creatine, NAA, lactate/lipid, and relative indices. Univariate mixed-effects linear regression models were employed and features with p&lt; 0.2 were selected for subsequent model building. Support vector machine (SVM), random forest, and gradient boosting machine-learning algorithms were trained and tested on a ⅔-⅓ train-test split with 4-fold cross-validation in training to predict a high/low KI-67 and TAI.

RESULTS

Although none of the individual imaging metrics were significantly associated with KI-67 in the univariate analysis, all diffusion and several MRSI metrics (ncholine, nNAA, CNI, excess choline and creatine) were significantly associated with cellularity. Preliminary multivariate analyses to date suggest that the best radiopathomic model performance is achieved when an SVM was used along with T1-precontrast, nADC, and all metabolite levels (mean cross-validation AUC=0.73 and accuracy=.77).

CONCLUSION

Our results suggest that multi-parametric physiologic and metabolic MRI are useful for radiopathomic-mapping of tumor aggressiveness and are currently being optimized in a larger cohort.

CSIG-36. NOTCH3 DRIVES MENINGIOMA TUMORIGENESIS AND RESISTANCE TO RADIOTHERAPY

There are no approved targeted therapies for meningiomas and the cell types underlying meningeal tumorigenesis are incompletely understood. To address these limitations, we performed single-cell RNA sequencing of 57,114 cells from 8 human meningiomas and 54,607 cells from 3 canine meningiomas. Pseudotime, gene ontology, and copy number variant analyses revealed a population of pericyte-like meningioma cells that were conserved across human and canine tumors and were enriched in expression of Notch3 and other cancer stem cell genes. Deconvolution of cell types from bulk RNA sequencing and DNA methylation profiling of 200 human meningiomas integrated with immunohistochemistry (IHC), immunofluorescence (IF), and RNAScope demonstrated Notch3+ pericytes and Notch3 expression were enriched in high grade or Immune-enriched meningiomas, which were distinguished from other meningioma DNA methylation groups by genes driving vasculature development. IHC and IF of human meninges integrated with lineage tracing approaches using Notch3-CreERT2 ROSAmT/mG alleles in mice demonstrated Notch3 expression was restricted to the perivascular stem cell niche during meningeal development and homeostasis. Mice harboring Notch3-CreERT2 Nf2fl/fl alleles developed meningeal hyperproliferation. Overexpression of constitutively activated Notch3 (Notch3AICD) in Immune-enriched human meningioma cells increased the expression of cancer stem cell genes, driving clonogenic growth in vitro, limiting dilution tumor-initiating capacity in vivo, and resistance to radiotherapy in vivo. A selective Notch3 neutralizing antibody (αNRR3) blocked meningioma cell proliferation and expression of Notch3 target genes, inhibiting meningioma xenograft growth and prolonging overall survival. Single-cell RNA sequencing of 187,366 cells from meningioma xenografts after αNRR3 or radiotherapy treatment ± Notch3AICD overexpression revealed distinct meningioma cell-intrinsic or cell-extrinsic mechanisms driving responses to radiotherapy or αNRR3, respectively. Combined treatment with αNRR3 and radiotherapy additively blocked meningioma xenograft growth and extended survival benefit. In sum, these data shed light on a novel cell type, molecular mechanism, and therapeutic vulnerability in the most common primary intracranial tumor.

SURG-04. CDK2NA/B, SMAD4, AND PIK3R1 MUTATIONS ARE ASSOCIATED WITH INCREASED RISK OF LOCAL RECURRENCE AFTER SURGICAL RESECTION OF BRAIN METASTASES

BACKGROUND

While resection of brain metastases (BMs) offers favorable local disease control, factors and genetic alterations associated with local recurrence are not well defined. This study examined patient, tumor, treatment, and genetic factors associated with local recurrence.

METHODS

A retrospective, single-center study was conducted with patients who underwent resection of a BM with available clinical outcome and genetic data available. Local recurrence was the primary outcome of the study. Next-generation sequencing of coding regions in over 500 cancer genes was performed to detect mutations. Cox proportional hazards analysis was performed to identify patient, tumor, treatment, and genetic factors associated with local recurrence.

RESULTS

We identified 91 patients who underwent surgical resection of a BM with available data, of which 80 (87.1%) underwent preoperative radiotherapy or received some form of adjuvant radiation to the resection cavity. Primary pathologies in the cohort included non-small cell lung cancer (24.2%), melanoma (24.2%), breast cancer (16.5%), gastrointestinal cancers (13.2%), gynecologic cancers (4.4%), renal cell carcinoma (4.4%), and other cancers (13.2%). Eleven patients (12.1%) developed postoperative recurrence at the surgical site with 6- and 12-month PFS of 91.2% and 84.2%. Multivariate Cox proportional hazard analysis identified cancer type (Gyn and RCC vs Others: OR 39.4, p=0.002), CDK2NA/B co-deletion (OR 37.52, p=0.0009), PIK3R1 mutation (OR 56.88, p=0.003), and SMAD4 mutation (OR 134.8, p=0.0007) to be associated with time to local recurrence. Overall survival and length of follow-up were not different based on mutational status of these genes, demonstrating that results were not due to survival bias.

CONCLUSIONS

Genetic alterations within BMs impact clinical outcomes after surgical resection. Further work is needed to determine if targeted therapies for BMs with these alterations can decrease rates of local recurrence.

DISP-14. DEFINING INTERVENTIONS AND METRICS TO IMPROVE DIVERSITY IN CNS CLINICAL TRIAL PARTICIPATION: A SNO AND RANO EFFORT

Despite major strides in cancer research, care, and therapy, these advances have not been equitable across race and ethnicity. Groups underrepresented in medicine (URM) are more likely to have inadequate preventive screening, increased delays in diagnosis, and poor representation in clinical trials. Notably, Black and Latino Americans represent 29% of the population but only reflect 8% of oncology clinical trial participants. Recent studies have shown that this disparity is also present in neuro-oncology as women, Black, and Latino Americans remain under-accrued in clinical trials. Brain tumor patients already face unique barriers to clinical trial enrollment and completion due to disease-specific conditions–such as impaired motor function, cognition, language deficits, and caregiver dependency–which pose additional difficulties in clinical trial consent, enrollment, and adherence. As part of this collaborative initiative, we evaluated the impact of how a lack of diversity in cancer research is detrimental to scientific discovery and propose interventions focused on improving URM engagement with clinical research. Recommendations include the creation of inclusive trial design at the onset, decreasing barriers to care, expanding trial eligibility, and equitable access to tumor profiling for personalized medical trials. Additionally, setting reasonable metrics and goals for accrual and engagement with patient and community stakeholders will ultimately help to diversify trial participants. Lastly, diversification and inclusion practices within the neuro-oncology workforce, including all personnel involved in clinical research, will help to minimize bias, promote concordant care, and assist with developing sustainable solutions. The diversification of participation in neuro-oncology clinical trials is imperative. The lack of diversity in clinical trials can contribute to improper generalizability of treatment efficacy, resulting in potentially harmful practices. Equitable access and inclusion of URM brain tumor patients will not only enhance research discoveries but will also result in improved patient care for all cancer patients.

NIMG-61. IMPROVED GENERALIZABILITY OF RADIOPATHOMIC PROBABILISTIC MAPPING OF TREATMENT-INDUCED EFFECTS WITH PHYSIOLOGIC MR IMAGING AND DEEP LEARNING IN PATIENTS WITH RECURRENT GLIOBLASTOMA

Although physiologic (diffusion-weighted and perfusion-weighted) MRI has shown promise in identifying regions of recurrent tumor (rTumor) in patients with glioblastoma suspected of progression, distinguishing treatment-induced effects (TxE) from rTumor on anatomical MRI remains a challenge. Whereas prior larger-scale machine learning (ML)-based studies mostly utilize anatomical imaging alone and/or perform lesion-level predictions, this study aimed to develop a non-invasive, radiopathomic tool for regional probabilistic mapping of TxE using 208 tissue-samples (55 pathologically-confirmed TxE, 153 recurrent glioblastoma) acquired from 107 patients with known spatial coordinates on pre-surgical MRI. We tested the hypothesis that applying a deep-learning (DL) model that included physiological MRI can: 1) more accurately identify areas of TxE that mimic rTumor on anatomical MRI and 2) better generalize to an independent test set than ML-models or a DL-model that uses anatomical MRI alone. An 80/20 split for training/validation was used after 1/3 of the patients were withheld for testing. Oversampling of TxE samples was employed to address class imbalance and an equal proportion of TxE samples was maintained across all datasets. Three ML-models, their ensemble, and a deep 4D-convolutional-neural-network were trained based on normalized anatomical (post-contrast T1, T2-FLAIR), diffusion-weighted (ADC, FA), and DSC perfusion-weighted (PeakHeight, %recovery) images cropped to 10mm-cubic patches centered on the coordinates from where tissue was obtained. Although Random Forest and voting-ensembled ML-models using all imaging and the anatomical DL-model had the best validation performance (AUC=0.81-0.82), these models did not generalize (test AUC=0.58-0.59). The DL-model including physiologic images had slightly lower validation AUC (0.78) but the best overall test AUC (0.795), indicating superior generalizability. Elevated blood volume (nPeakHeight) was the most important feature. Our DL-model’s interpretability was also demonstrated by disrupting class separation after shuffling voxels within each input patch. These results suggest that using deep-learning with physiologic MRI can improve intratumoral classification of TxE from rTumor.

SURG-05. SUPERVISED MACHINE LEARNING IDENTIFIES RISK FACTORS ASSOCIATED WITH LEPTOMENINGEAL DISEASE AFTER SURGICAL RESECTION OF BRAIN METASTASES

INTRODUCTION

Predictors of postoperative leptomeningeal disease (LMD) after resection of brain metastases (BMs) are not well defined.

OBJECTIVE

This study examined rates and predictors of LMD, including subtypes, in patients who underwent resection of a BM followed by postoperative radiation.Method: A retrospective, single-center study was conducted examining overall LMD, classical LMD (cLMD), and nodular LMD (nLMD) risk. Logistic regression and a Cox proportional hazards analyses were performed to identify risk factors associated with LMD. Random forest models were constructed to predict LMD and differentiate cLMD versus nLMD. Accuracy and the area under the receiver operating characteristic curve (AUROC) were calculated to evaluate the models.Result: Of the 217 patients in the cohort, 47 (21.7%) developed postoperative LMD with 19(8.8%) cLMD cases and 28(12.9%) nLMD cases . Six-, 12-, and 24-month LMD-free survival rates were 92.3%, 85.6%, and 71.4%, respectively. Patients with cLMD had worse survival outcomes from LMD diagnosis compared to nLMD (2.4 vs 6.9 mo, Log-rank p=0.02), and treatment of LMD was associated with improved survival for both cLMD and nLMD subtypes. Multivariate Cox hazard analysis identified cerebellar/insular/occipital location (HR 3.25, 95% CI 1.73-6.11, p=0.0003), absence of extracranial disease (HR 2.49, 95% CI 1.27-4.88, p=0.008), and ventricle contact (HR 2.82, 95% CI 1.5-5.3, p=0.001) to be associated with postoperative LMD. A predictive model using random forest analysis with an AUROC of 0.87 in a test cohort identified tumor location, systemic disease status, and tumor volume as the most important factors associated with LMD. Both regression analysis and random forest analysis identified postoperative systemic therapy exposure as the main factor differentiating cLMD from nLMD development.

CONCLUSION

Tumor location, absence of extracranial disease at the time of surgery, contact with a ventricle, and increased tumor volume are associated with LMD. Classical LMD is associated with worse prognosis compared to nLMD.

QOL-10. NOVEL MULTIMODAL STUDY OF THREE COGNITIVE REHABILITATION INTERVENTIONS IN LOWER GRADE GLIOMA

BACKGROUND

Grade 2 and 3 glioma survivors (LrGG) are living longer, yet experience cognitive impairments with diminished quality of life (QOL). We present a novel multimodal study of three cognitive rehabilitation interventions in stable LrGG survivors.

METHODS

Participants were radiologically stable adult LrGG patients who were off medical treatment for ≥ 6 months with subjective and objective cognitive impairments ( &gt;1SD in 2 or more domains). Patients were offered either In-person cognitive rehabilitation (strategy training including telehealth), or randomized to App-based cognitive rehabilitation (retraining and strategy training) versus Text messaging (strategy training). Intervention duration was 3 months. Neuropsychological testing (with parallel forms) and QOL assessments were conducted at baseline (T1), immediate post intervention (T2), and 6-month follow-up (T3), and analyzed with repeated measures regression or Wilcoxon signed rank tests.

RESULTS

Of the 33 analyzed (enrollment ongoing); 15/17 In-person, 5/8 App-based, and 8/8 Texting completed ≥ 80% or greater of interventions. Demographic and clinical characteristics were similar between cohorts. Median age was 48 years (range 27-63), 58% astrocytoma, 30% oligodendroglioma, 15% other (1 pilocytic astrocytoma, 4 diffuse glioma NOS), and 76% had prior radiotherapy. Rehabilitation interventions showed improvements in auditory working memory (T1-T2 In-person p= 0.02, eta2= 0.32-medium effect), verbal learning (T1-T3 App-based p= .06, eta2= 0.54-large effect; T1-T3 Texting p= .01, eta2= 0.75-large effect), and verbal memory (T1-T3 App-based p= .06, rho=0.31-medium effect).

CONCLUSION

Significant improvements in cognitive impairments were found with medium to large treatment effects within each cohort. Cognitive rehabilitation via In-person and Texting showed strongest feasibility and acceptability. In-person cognitive rehab showed earlier posttreatment improvements whereas treatment effects for App-based and Texting were noted, but took longer to realize gains. These interventions may show promise for addressing cognitive impairments in LrGG survivors and warrant further investigation.

CNSC-19. GLIOMA-INDUCED NEURONAL REMODELING PROMOTES REGIONAL IMMUNOSUPPRESSION

Gliomas remodel neuronal circuits and distinct intratumoral regions maintain functional connectivity through a subpopulation of synaptogenic malignant cells expressing Thrombospondin-1 (TSP-1, encoded by the Thbs1 gene). Single-cell RNA sequencing analyses of primary patient samples identified a significant downregulation of immune response in myeloid cells and lymphoid cells in high functionally connected intratumoral regions characterized by upregulated TSP-1. Understanding the functional significance of immunosuppression within functionally connected intratumoral regions may uncover therapeutic vulnerabilities. Here, we investigate glioma-neuronal-immune crosstalk across clinical tumor specimens and preclinical syngeneic models through bulk and single-cell RNA sequencing (13,670 cells), flow cytometry, and spatial transcriptomics. Using an SB28 murine glioma cell line with endogenous Thbs1 stably expressed, we generated a CRISPR Thbs1-knock-out (KO) cell line. Bulk RNA-sequencing demonstrated that Thbs1-WT tumors exhibited gene expression programming consistent with synapse-associated genes and synaptogenic factors, recapitulating enriched connectivity in primary patient samples. Flow cytometry of brain-infiltrating leukocytes revealed that macrophages isolated from Thbs1-KO tumors were more frequently polarized into the pro-inflammatory “M1-like” phenotype (median M1/M2 ratio = 0.6 [WT] vs 1.34 [KO], p &lt; 0.006). Unbiased gene expression program analysis using spatial transcriptomics for in vivo tumor-harboring mouse brains demonstrated a significant spatial overlap of signatures of synaptogenesis (represented by Ntng1 and Nlgn3 genes) and downregulated immune response (represented by Nfkb1 and Cd83 genes). SB28-Thbs1-KO syngeneic models demonstrated slower tumor growth and significantly longer survival compared to Thbs1-WT counterparts (19 days [WT] vs 25 days [KO], p &lt; 4.5E-5). The survival difference was abrogated in the B6-SCID immunodeficient mice, indicating the critical role of adaptive immunity in the survival advantage associated with TSP-1 inhibition. Our results identify previously unknown immunosuppression mechanisms in the context of glioma-induced intratumoral connectivity via Thbs1. Future therapeutic strategies targeting this glioma-neuronal-immune crosstalk may open up new avenues for glioblastoma immunotherapy.

CNSC-23. NEURONAL ORIGIN INFLUENCES SPONTANEOUS NETWORK ACTIVITY IN GLIOBLASTOMA-NEURON COCULTURES

Neuronal activity is emerging as a driver of cancer initiation and proliferation. Several models have been developed to recapitulate cancer-neuron interactions within the tumor microenvironment; however, there is substantial diversity of neuronal subtypes within the murine and human cortex. Differences in the neuronal microenvironment may influence glioma cell behavior and fail to recapitulate in vivo mechanisms. Here, we investigate the electrophysiological properties of cortical neurons and activity-regulated paracrine signaling across co-culture conditions using microelectrode arrays (MEA). Cortical neurons were obtained from three conditions, (1) human prenatal tissues, (2) murine embyronic tissue (E18), and (3) murine postnatal tissue (P1.5), then co-cultured with primary patient-derived glioblastoma cultures. Neuronal activity was assessed using weighted mean firing rate (WMFR; spike rate multiplied by number of active electrodes) and network burst synchrony index. Characterization of the activity-dependent paracrine signaling was performed using proteomics. Initiation of spontaneous firing activity differed across neuronal conditions, with activity occurring earlier in postnatal P1.5 cultures (day in vitro [DIV] 7), whereas E18 and human prenatal cultures demonstrated delayed patterns of activity (DIV 14 and DIV 21, respectively). Glioblastoma cells were added on the first day of consistent spiking. All neuronal co-culture conditions demonstrated glioma-induced hyper-activity (increased WMFR), however in different patterns. E18 co-cultures exhibited glioma-induced hyperexcitability within 24 hours with a subsequent decrease in activity over 14 days with minimal synchrony. P1.5 demonstrated no glioma-induced hyper-excitability at 24 hours; however, WMFR peaked by 7 days and remained elevated through co-culture day 14 (including concurrent increase in synchrony). Human postnatal neurons demonstrated a steady and consistent increase in WMFR as early as 24 hours post co-culture and remained active through day 14 with no synchrony change. Our data suggest that careful consideration should go into selecting neuronal co-culture conditions for mechanistic studies.

CNSC-05. ELECTROPHYSIOLOGICAL PATTERNS OF GLIOMA-INDUCED NEURONAL NETWORK REMODELING ARE CONSERVED ACROSS TUMOR SUBTYPE

Recent evidence indicates that diffuse gliomas engage with neurons at the single-unit and circuit level through differing mechanisms. Certain malignant gliomas form glioma-neuron excitatory glutamatergic synapses and modulate neuron-neuron synapses through activity-dependent paracrine signaling, while others establish glioma-glioma connections via tumor microtubes. It is therefore possible that diffuse gliomas remodel neuronal circuits in a defined and predictable manner and demonstrate distinct electrophysiological profiles with prognostic and therapeutic significance. Here we apply machine learning principles in 140 patients across glioma subtypes to uncover unique electrophysiological features non-invasively via magnetoencephalography (discovery dataset) followed by feature validation using subdural electrocorticography (validation dataset). Following spatial-temporal registration, we fit an elastic net logistic regression classifier to distinguish between power spectra arising from glioma-remodeled cortex and within-subject control conditions. Model significance was determined non-parametrically by re-training each model 1,000 times with randomly permuted class labels and testing the true phi coefficient against the null distribution. In the discovery dataset, we were able to classify glioma infiltration based on tumor intrinsic neuronal activity (p &lt; 0.05) in 127 patients (90.7%). We identified 30 electrophysiological features which revealed increased power in the delta range (1-4 Hz) and decreased power in the beta range (12-20 Hz) as a unique signature of glioma remodeling (p &lt; 0.05) which was preserved in the validation dataset as well as across WHO 2021 diffuse glioma subtypes. In order to identify gene expression programs and signaling mechanisms that may contribute to glioma-induced remodeling but are potentially not identified in the current clinical classification scheme, we assessed targeted, next generation sequencing and DNA mutations as covariates, which again demonstrated the significance of the delta-beta spectral features. These data support converging mechanisms of glioma-induced neuronal network remodeling across tumor subtypes, setting the stage for novel therapies such as neuromodulation.

SURG-19. PROGNOSTIC VALIDATION OF A NEW CLASSIFICATION SYSTEM FOR EXTENT OF RESECTION IN GLIOBLASTOMA: A REPORT OF THE RANO RESECT GROUP

BACKGROUND

Terminology to describe extent of resection in glioblastoma is inconsistent across clinical trials. A surgical classification system was previously proposed based upon residual contrast-enhancing (CE) tumor. We aimed to (I) explore the prognostic utility of the classification system and (II) define how much removed non-CE tumor translates into a survival benefit.

METHODS

The international RANO resect group retrospectively searched the databases from seven neuro-oncological centers in the USA and Europe for patients with newly diagnosed glioblastoma per WHO 2021 classification. Clinical and volumetric information from pre- and post-operative MRI were collected.

RESULTS

We collected 1021 patients with newly diagnosed glioblastoma, including 1008 IDHwt patients. 744 IDHwt glioblastomas were treated with radiochemotherapy per EORTC 26981/22981 (TMZ/RT→TMZ) following surgery. Among such homogenously treated patients, lower absolute residual tumor volumes (in cm3) were favorably associated with outcome: patients with ‘maximal CE resection’ (class 2) had superior outcome compared to patients with ‘submaximal CE resection’ (class 3) or ‘biopsy’ (class 4) (median OS: 19 versus 15 versus 10 months; p=0.001). Extensive resection of non-CE tumor (≤ 5 cm3 residual non-CE tumor) provided an additional survival benefit in patients with complete CE resection, thus defining class 1 (‘supramaximal CE resection’) (median OS: 24 versus 19 months; p=0.008). The prognostic value of the resection classes was retained on multivariate analysis when adjusting for molecular and clinical markers including MGMT promotor status. Relative tumor reduction (in percentage) was not prognostic for outcome on multivariate analysis, and inter-rater agreement for CE and non-CE tumor on post-operative MRI was sufficient.

CONCLUSION

The proposed “RANO categories for extent of resection in glioblastoma” are highly prognostic and may serve for stratification of clinical trials. Removal of non-CE tumor beyond the CE tumor borders translates into additional survival benefit, providing a rationale to explicitly denominate such a ‘supramaximal CE resection’.

NIMG-65. IMPROVED SPATIAL MAPPING OF TUMOR AGGRESSIVENESS WITH 1H MAGNETIC RESONANCE SPECTROSCOPY AND DEEP LEARNING IN PATIENTS WITH NEWLY-DIAGNOSED GLIOMA

INTRODUCTION

Noninvasive, radiopathomic mapping of tumor aggressiveness can benefit patients with glioma by guiding the selection of tissue samples for diagnosis, increasing extent of resection, and non-invasively characterizing residual tumor burden for subsequent treatment. Although prior studies have demonstrated the utility of metabolic metrics quantified from 1H-MR Spectroscopy (MRS) in probing tumor pathology, this study evaluated the benefit of using the entire 1D-spectrum and deep learning for radiopathomic mapping of intratumoral cellularity, proliferation (ki-67), and a new tumor aggressiveness index (TAI) defined as log((n(ki−67)+n(cellularity))*tumor-score).

METHODS

Multi-voxel 1H-MRS was acquired on 281 patients newly diagnosed with a glioma (47% IDH-wildtype) immediately before surgical resection. After reconstructing individual spectra at the locations where tissue samples were obtained during surgery and normalizing by NAA in contralateral normal-appearing-white-matter, 607 spectra with corresponding histopathology were deemed of sufficient quality for analysis. A 1D convolutional-neural-network with bidirectional long- and short-term memory deep-learning model using the entire spectrum (0.6-3.6ppm) was compared to mixed-effects regression (with choline-to-NAA index[CNI]) and Random Forest (with CNI+normalized peak heights) models for predicting ki-67, cellularity, and TAI. Results &

DISCUSSION

Using deep-learning on the entire spectrum resulted in 10.3%-22.1% lower mean absolute error (MAE) and 0.32-0.37 higher R2 values compared to using CNI alone or a random forest model with multiple metabolic metrics. MAE values for all 3 deep-learning models were 26-44% &lt; 1 standard deviation of the ground truth, demonstrating reasonable prediction accuracy within the test data set. Although the lowest MAE (0.16) and highest R2 (0.41) was attained when predicting TAI with deep-learning, the prediction of cellularity resulted in the lowest %MAE. Colormaps of predicted pathology identified regions of heightened aggressiveness surrounding tissue samples with most abnormal pathological features that sometimes extended beyond the non-enhancing lesion. Current work is evaluating the clinical utility of our deep-learning model and predicted maps of aggressiveness.

NIMG-30. AI-BASED MOLECULAR CLASSIFICATION OF DIFFUSE GLIOMAS USING RAPID, LABEL-FREE OPTICAL HISTOLOGY

INTRODUCTION

Molecular classification has transformed the management of brain tumors by enabling more accurate prognostication and personalized treatment. Access to timely molecular diagnostic testing for brain tumor patients is limited, complicating surgical and adjuvant treatment and obstructing clinical trial enrollment.

OBJECTIVE

We aim to develop a rapid (&lt; 90 seconds), AI-based diagnostic screening system that can provide molecular classification of diffuse gliomas and report its use in a prospective, multicenter, international clinical trial of diffuse glioma patients (n = 153).

METHODS

By combining stimulated Raman histology (SRH), a rapid, label-free, non-consumptive, optical imaging method, and deep learning-based image classification, we are able to predict the molecular genetic features used by the World Health Organization (WHO) to define the adult-type diffuse glioma taxonomy, including IDH-1/2, 1p19q-codeletion, and ATRX loss. We developed a multimodal deep neural network training strategy that uses both SRH images and large-scale, public diffuse glioma genomic data in order to achieve optimal molecular classification performance.

RESULTS

One institution was used for model training (University of Michigan) and four institutions (NYU, UCSF, Medical University of Vienna, and University Hospital Cologne) were included for prospective patient enrollment and model testing. Using our system, called DeepGlioma, we achieved an average molecular genetic classification accuracy of 93.2% and identified the correct diffuse glioma molecular subgroup with 91.5% accuracy. DeepGlioma outperformed conventional IDH1-R132H immunohistochemistry (94.2% versus 91.4% accuracy, respectively) as a first-line molecular diagnostic screening method for diffuse gliomas, detecting canonical and non-canonical IDH mutations with high accuracy.

CONCLUSION

Our results demonstrate how artificial intelligence and optical histology can be used to provide a rapid and scalable alternative to wet lab methods for the molecular diagnosis of brain tumor patients during surgery.

Seizure outcome after resection of insular glioma: a systematic review, meta-analysis, and institutional experience

OBJECTIVE

Gliomas arising from the insular cortex can be epileptogenic, with a significant proportion of patients having medically refractory epilepsy. The impact of surgery on seizure control for such tumors is not well established. In this study, the authors aimed to investigate seizure outcomes after resection of insular gliomas using a meta-analysis and institutional experience.

METHODS

Three databases (Ovid MEDLINE, Embase, and Cochrane Central Register of Controlled Trials) were systematically searched for published studies of seizure outcomes after insular glioma resection from database inception to March 27, 2021. In addition, data were retrospectively collected on all adults (age > 17 years) who had undergone insular glioma resection between June 1997 and June 2015 at the authors' institution. Primary outcome measures were seizure freedom rates at 1 year and the last follow-up. Secondary outcome measures consisted of persistent postoperative neurological deficit beyond 90 days, mortality, and tumor progression or recurrence.

RESULTS

Eight studies reporting on 453 patients who had undergone 460 operations were included in the meta-analysis. The pooled mean age of the patients was 42 years. The pooled percentages of patients with extents of resection (EORs) ≥ 90%, 70%-89%, and < 70% were 55%, 33%, and 11%, respectively. The pooled seizure freedom rate at 1 year was 73% for Engel class IA and 78% for Engel class I. The pooled seizure freedom rate at the last follow-up was 60% for Engel class IA and 79% for Engel class I. The pooled percentage of persistent neurological deficit beyond 90 days was 3%. At the authors' institution, 109 patients had undergone resection of insular glioma. A greater EOR was the only significant independent predictor of seizure freedom after surgery (HR 0.290, p = 0.017). The optimal threshold for seizure freedom corresponded to an EOR of 81%. Patients with an EOR > 81% had a significantly higher seizure freedom rate (OR 2.16, p = 0.048).

CONCLUSIONS

Maximal safe resection can be performed with minimal surgical morbidity to achieve favorable seizure freedom rates in both the short and long term. When gross-total resection is not possible, an EOR > 81% confers the greatest sensitivity and specificity for achieving seizure freedom. Systematic review registration no.: CRD42021249404 (https://www.crd.york.ac.uk/prospero/).