IDH Wildtype Research Articles

CTIM-12. A RETROLECTIVE ANALYSIS ON THE SAFETY, EFFICACY AND IMMUNOGENICITY OF 203 GLIOBLASTOMA PATIENTS TREATED WITH PERSONALIZED THERAPEUTIC CANCER VACCINES (PTCV)

Abstract BACKGROUND Glioblastoma multiforme (GBM) is the most common and malignant primary brain tumor. Despite of the current SoC, patients have a poor prognosis with a median OS of about 14-17 months.Method, RESULTS We report retrospectively analyzed data from 203 GBM patients (IDH wildtype) who added PTCV derived from patient individual tumor mutations to their treatment regimen in the setting of individual treatment attempts (ITA) in Germany. Data were analyzed re. vaccine induced T cell responses and safety acc. to CTCAE terminology version 5.0. Subsets of patients with and without hypermethylation of the MGMT promoter were further evaluated re. efficacy. Treated patients had a median age of 53 years at diagnosis, started vaccination after a median of 11 months after diagnosis, received a median of 8 immunizations with a median of 20 neoepitope peptides, and had a median follow-up time since first vaccination of 21 months. They usually received SoC (Stupp scheme), often accompanied by additional therapies. In general, the PTCV was well tolerated, and most patients experienced only Grade 1 and 2 AEs. Most frequently injection site reactions were observed (67%). Seven patients developed Grade 3 allergic or anaphylactic reactions after 7 or more vaccinations. For 119 patients T-cell responses against the vaccinated neoepitopes were investigated. Almost all patients developed vaccine induced T-cell responses (97%). 71% of patients showed both CD4+ and CD8+ T-cell responses towards vaccinated peptides. Immune responses usually increased over time peaking within 3-6 months. Median OS of the whole cohort was 26 mo (95% CI 24, 31). Primary GBM patients without recurrence events before the start of vaccination and with or without hypermethylated MGMT promoter status showed a mOS of 41 months (95% CI: 32, NA) or 23 months (95% CI: 20, NA), respectively. CONCLUSION The retrospective data analysis from a real-world cohort of GBM patients suggests that PTCV are safe, immunogenic, and may prolong the survival of patients.

BIOS-06. MACHINE LEARNING MODEL INDEPENDENTLY IDENTIFIES CLINICAL FEATURES ASSOCIATED WITH WILD-TYPE GLIOBLASTOMA VERSUS MUTANT GRADE 4 ASTROCYTOMA

Abstract BACKGROUND Glioblastoma (GBM), the most common and aggressive primary brain tumor in adults, presents significant clinical challenges due to its invasive nature and high recurrence rates. Understanding the clinical demographic factors associated with IDH wild-type (IDH-wt) versus mutant grade 4 astrocytoma (IDH-Mut) is critical for prognosis and treatment strategies. This study aimed to utilize a novel machine learning method to identify key clinical and demographic factors linked to IDH status. METHODS The Cancer Imaging Archive dataset included 617 patients but required preprocessing and cleaning. The cleaning process ensured data completeness and consistency by removing non-relevant variables like date and time, resulting in 20 potential predictors and 317 observations. A random forest model using Gini impurity score was built to classify patient IDH status. The dataset was split into a training set (two-thirds) and a test set (one-third). A variable importance plot identified significant variables based on the mean decrease in the Gini index. RESULTS Our machine learning model achieved 98% accuracy on the test set, outperforming the previously published model of 80% accuracy. A variable importance plot identified age at diagnosis, gender, ethnicity, and race as key predictors of IDH status, with age being the most important. Age was the most important predictor, but race (Caucasian) and ethnicity (not Hispanic or Latino) were also significant. CONCLUSIONS We present a highly accurate machine learning model for differentiating glioma IDH molecular status based only on patient clinical and demographic factors; improving upon prior published models. Future work will use a generalized logistic model to explore variable interactions and predict IDH-wt versus IDH-Mut status. These steps are vital for enhancing predictive accuracy and supporting personalized treatment strategies. Furthermore, we will define associations and interactions between these variables with other biological and therapeutic variables in predicting clinical outcome.

DNAR-13. LONG TERM FOLLOW-UP OF A LYNCH SYNDROME-ASSOCIATED GLIOBLASTOMA RESISTANT TO STANDARD-OF-CARE CHEMORADIATION

Abstract BACKGROUND Lynch syndrome is caused by pathogenic germline variants in the DNA mismatch repair (MMR) genes, predisposing individuals to malignancies, including glioblastoma (GBM). MMR-deficient tumors are resistant to monofunctional alkylators such as temozolomide but not to bifunctional alkylators such as the nitrosourea lomustine. Anti-programmed cell death protein 1 (PD1) immune checkpoint inhibitor (ICI) pembrolizumab represents another treatment option, which is approved by the United States Food and Drug Administration (FDA) for the treatment of mismatch repair deficient cancers. METHODS We present a patient with multifocal GBM and Lynch syndrome who experienced rapid progression following standard-of-care chemoradiation, followed by durable responses to lomustine and then to pembrolizumab. RESULTS A 58-year-old man with Lynch syndrome (MSH2 mutation) and prior history of colon cancer presented with seizures. Magnetic resonance imaging (MRI) brain identified right frontal and left parietal masses. Right frontal tumor was resected; pathology was consistent with GBM, IDH wildtype, MGMT-unmethylated. Both lesions were treated with radiation therapy (60 Gy, 30 fractions) with concurrent and adjuvant temozolomide. After two cycles of adjuvant temozolomide, his left parietal tumor progressed and was subtotally resected, followed by four cycles of lomustine (stopped due to myelotoxicity). He remained progression-free for 15 months. Surveillance MRI identified new enhancement involving the left frontal lobe, left periventricular white matter, and right internal auditory canal, concerning for leptomeningeal dissemination. MRI spine identified no additional lesions. Cerebrospinal fluid cytology was negative for tumor cells on two occasions. Patient was treated with pembrolizumab. He remains on treatment with stable disease four months after ICI initiation and 31 months after GBM diagnosis. CONCLUSIONS Patients with MMR-deficient GBM resist standard chemoradiation. Alternatives such as nitrosourea or ICI therapy may achieve durable responses. While data is insufficient to administer these agents as upfront therapy in this specific setting, both should be considered as options for recurrent disease.

EPID-32. ELEVATED SERUM IGE IS LINKED TO IMPROVED GLIOBLASTOMA SURVIVAL, PARTICULARLY AMONG FEMALES

Abstract Prior epidemiological research shows a correlation between serum immunoglobulin E (IgE) levels and decreased risk of glioma. However, the relationship between IgE and the prognosis of glioma remains poorly understood. This study seeks to investigate how factors such as sex, tumor subtype, and IgE class influence the association of serum IgE levels with both glioma risk and survival outcomes. METHODS- In this investigation, we conducted a case-control study utilizing participants enrolled in the UCSF Adult Glioma Study from 1997 to 2010. We measured serum IgE levels for total, respiratory, and food allergy in adults diagnosed with glioma (n=1,696), and controls (n=1,135) matched based on age, sex, and race/ethnicity. Logistic regression, adjusted for patient demographics, was employed to evaluate the correlation between IgE levels and glioma risk. Multivariable Cox regression, adjusted for patient-specific and tumor-specific factors, was utilized to compare survival outcomes between the elevated and normal IgE groups. RESULTS- Increased levels of total IgE were linked to a decreased risk of both IDH wildtype (OR=0.65, 95% CI: 0.54-0.78) and IDH mutant glioma (OR=0.65, 95% CI: 0.50-0.85). In multivariable Cox regression analysis, heightened respiratory IgE levels were associated with enhanced survival among individuals with IDH wildtype glioma (HR=0.78, 95% CI: 0.67-0.91). Notably, this improved survival was more pronounced in females (HR=0.71, 95% CI: 0.53-0.96) compared to males (HR=0.80, 95% CI: 0.66-0.97), resulting in median survival improvements of 6.2 months (P<.001) and 1.6 months (P=0.003), respectively. CONCLUSION- Increased levels of serum IgE were associated with reduced risk of glioma overall, and improved prognosis for IDH wildtype glioma, which was particularly strong in females. These findings hint at potential sexual dimorphism in the antitumor activity of IgE-mediated immune responses, suggest an important role for IgE assessment in glioblastoma clinical trial design, and provide justification for investigations into IgE-based therapeutics.

PATH-31. COMPLETE TUMOR DISAPPEARANCE ON BRAIN MRI AFTER STANDARD CHEMORADIATION THERAPY FOR A MOLECULAR GLIOBLASTOMA

Abstract INTRODUCTION Glioblastoma (GBM) is the most aggressive brain tumor. The WHO CNS5 classification has newly incorporated three molecular criteria in diagnosing molecular GBM, including a concurrent gain of chromosome 7 and loss of chromosome 10 (+7/-10), TERT promoter mutation, or EGFR amplification. Despite new classification, the Stupp protocol remains the standard of care for all GBMs. However, it is unclear if the treatment response of the molecular GBM with low grade features is the same as the limited response of the traditional histopathological GBM. CASE REPORT A 74-year-old male presented in August of 2023 with new onset left-sided facial numbness. The numbness resolved spontaneously that day. Brain MRI was notable for a diffuse left frontal T2/Flair hyperintense lesion extending across the corpus callosum into the right frontal lobe without definite enhancement. Brain biopsy from 8 sites of the tumor demonstrated a subset of areas with mild hyper-cellularity, suggesting a low-density infiltrating glioma. However, TERT promoter mutation (-124C>T), was detected along with wild-type IDH1, supporting molecular GBM. Studies also found unmethylated MGMT gene promotor. Brain MRIs after 3 and 6 months showed no tumor progression. The patient has remained asymptomatic apart from baseline mild forgetfulness. He was started on the 6-week standard concomitant temozolomide and radiation therapy. Follow-up brain MRI after the above treatment showed complete imaging resolution of the tumor. DISCUSSION It is very rare for complete imaging resolution of GBM on brain MRI after standard 6-week chemoradiation therapy. Our case suggests that molecular GBM with low-grade features on imagine and pathology might have a better response to the standard treatment than traditional histopathological GBM even though the molecular GBM patient has unfavorable prognostic factors such as unmethylated MGMT and older age (>70 years old). Our experience emphasizes the heterogenous response of molecular GBM to the current standard treatment.

EPCO-29. PATHOMICS ANALYSIS IDENTIFIES POTENTIAL SUBTYPES OF DIFFUSE GLIOMAS IN H&E-STAINED WHOLE-SLIDE IMAGES

Abstract IDH status is the first criteria of diagnostic definition for diffuse gliomas in the 2021 WHO classification. While sufficient number of molecular tools are available to identify IDH1 and/or IDH2 mutations, they impose a time and cost burden on patients. We previously trained deep learning models visually differentiate IDH wild-type (IDHwt) and IDH mutant (IDHmut) gliomas in a small set of H&E-stained whole image slides (WSIs) from TCGA using an Uncertainty-Aware Convolutional Neural Network (CNN) at 10x magnification. In this follow-up study, we aimed to improve our models by expanding our annotations of a more representative subset of the TCGA public data, increasing the sampling magnification to 20x, redesigning the preprocessing filters, adjusting the training pipeline to reduce overfitting, and assessing the confidence in model prediction via uncertainty metrics. METHODS By modifying the pre-processing methods from our pilot study, we increased the number of expert-annotated WSIs. Using a 2:1:2 train/validation/test split, we fine-tuned a ResNet18 deep learning model with a pre-trained feature extractor and Monte Carlo dropouts. To identify image features critical for separating by IDH status, we extracted the learned embeddings and performed post-hoc analyses (PCA, t-SNE, and UMAP). RESULTS The new set of annotations resulted in ~15-fold increase in the number of WSIs (i.e.,>2.5 million tiles), without any data augmentation. Our prototype model demonstrated clear separation of cases according to IDH status (sensitivity = 0.63, specificity = 0.91, weighted F1 = 0.81). CONCLUSION We successfully improved the prediction of IDH status in H&E-stained WSIs and identified potential subtypes of IDHwt and IDHmut diffuse gliomas after significantly increasing the dataset size with appropriate expert annotations. We are currently addressing the data imbalance by annotating additional IDHmut glioma WSIs. Furthermore, we will integrate other biomarkers (1p19q, EGFR etc.) to increase the robustness of histo-pathomics as a surrogate for molecular tests.

TMET-12. TARGETING AMINO ACID METABOLIC VULNERABILITIES IN GLIOMAS

Abstract Isocitrate dehydrogenase (IDH) wild-type gliomas and IDH mutant gliomas are similar in histologic appearance but differ in their genetic and metabolic backgrounds. This study aimed to reveal the difference in metabolites between IDH-wildtype glioma and IDH-mutant glioma, and to find the effective treatment according to the status of IDH in gliomas. Two artificial cell lines made from human astrocyte were used: NHAE6E7hTERTRas (IDH-wildtype) and NHAE6E7hTERTIDHmut (IDH-mutant). Using capillary electrophoresis- and ion chromatography–coupled mass spectrometry, the amount of asparagine was lower in NHAE6E7hTERTRas cells compared with NHAE6E7hTERTIDHmut cells, whereas the amount of glutamine, glutamate and 2-oxoglutarate in NHAE6E7hTERTIDHmut cells was lower than NHAE6E7hTERTRas cells. L-asparaginase, which converts asparagine into aspartate, was more effective in NHAE6E7hTERTRas cells than NHAE6E7hTERTIDHmut cells. L-asparaginase induced autophagy and inhibition of autophagy by 3-MA suppressed L-asparaginase-induced antitumor effect, which meant that the antitumor effect was at least partially due to the L-asparaginase-induced autophagy. Pharmacological or genetical inhibition of GLUD1 which converts glutamate to 2-oxoglutarate, suppressed proliferation of the cells by inducing ROS and apoptosis in NHAE6E7hTERTIDHmut cells. ROS inhibitor, NAC suppressed GLUD1 inhibitor-induced ROS, apoptosis, and cytotoxicity in NHAE6E7hTERTIDHmut cells, revealing that cytotoxicity by GLUD1 inhibitor was at least partially due to the inhibitor-induced ROS. The experiments using mutant IDH1 overexpressed glioma cell line showed similar sensitivity to GLUD1 inhibitor to NHAE6E7hTERTIDHmut, which suggested that the difference of sensitivity to GLUD1 inhibitor was due to the status of mutant IDH. In vivo experiments, L-asparaginase suppressed the growth of xenografted glioma cells without mutant IDH, whereas GLUD1 inhibitor inhibited the proliferation of xenografted glioma cells with overexpressed mutant IDH1. In conclusion, L-asparaginase and Glud1 inhibitor will be new therapeutic option for IDH-wildtype glioma and IDH-mutant glioma, respectively.

CTNI-29. A PHASE 1B STUDY TO EVALUATE THE SAFETY, PHARMACOKINETICS, AND OBJECTIVE RESPONSE OF LP-184 ALONE AND IN COMBINATION WITH SPIRONOLACTONE IN IDH WILD TYPE GLIOBLASTOMA AT FIRST PROGRESSION

Abstract LP-184 is a fully synthetic small molecule alkylator of the acylfulvene therapeutics class that induces DNA double-strand breaks and improves survival in orthotopic GBM xenograft models. In preclinical studies, LP-184 is brain penetrant with a brain tumor/plasma concentration ratio of 0.2. DNA damage induced by acylfulvenes is repaired via transcription-coupled nucleotide excision repair (TC-NER) mediated by ERCC complexes. GBM with intrinsic or pharmacologically induced DNA Damage Repair deficiency (dDDR) may preferentially respond to LP-184, independent of MGMT status. LP-184 is currently under investigation in a first-in-human Phase 1a dose-escalation safety study (NCT05933265) in adult patients with advanced solid tumors including GBM, with no dose-limiting toxicities to date at dose level 6. Following determination of the Maximum Tolerated Dose and/or Recommended Phase 2 Dose from Phase 1a, the selected expansion dose will be assessed in a Phase 1b study to evaluate the safety, pharmacokinetics, and objective response of LP-184 alone and in combination with spironolactone in IDH wild type glioblastoma at first progression. Spironolactone is a brain penetrant FDA approved agent that proteolytically degrades the TC-NER component ERCC3, thereby creating dDDR. Two rGBM cohorts will be enrolled according to a Simon’s 2-stage optimal design: (i) LP-184 alone (days 1 and 8 in a 21-day cycle, as intravenous infusion over 30 minutes) and (ii) combination of LP-184 and spironolactone (200 mg orally daily in conjunction and before the LP-184 infusion). In the first stage, 10 response evaluable subjects will be enrolled. If 1 or more of the first 10 response evaluable subjects achieve an objective response, 19 further subjects may be accrued in the second stage. In each cohort, at least 5 subjects will undergo planned tumor resection and will be treated with LP-184 or the combination the day prior to surgery to evaluate PK/PD.

NIMG-44. ELEVATED TRANSLOCATOR PROTEIN 18KDA BINDING IN LOW MEAN DIFFUSIVITY CLUSTERS IS A FEATURE OF THE TRANSFORMING GLIOMA AND IDH MUTANT GLIOMA MICROENVIRONMENT

Abstract Around 95% of low-grade gliomas undergo anaplastic de-differentiation into high-grade gliomas, a process only partially understood from in vivo clinical imaging. 11 patients with suspected transforming gliomas underwent a combined magnetic resonance imaging (MRI) and positron emission tomography (PET) protocol, incorporating post-contrast T1-weighted/FLAIR acquisition for tumour delineation and diffusion tensor imaging. The PET radiotracer [11C]-(R)-PK11195 assessed translocator protein 18kDa (TSPO) distribution, whose upregulation is associated with inflammation/glioma progression. [11C]-(R)-PK11195 binding potential (BPND) maps were generated using the simplified reference tissue model, with the grey-matter cerebellar time-activity curve serving as the reference tissue input function. KMeans clustering of mean diffusivity (MD) intensity values defined spatially distinct “habitats” of high/low MD voxels within the glioma microenvironment. BPND values within high/low MD clustering habitats were evaluated, across glioma grades and isocitrate dehydrogenase (IDH) mutational status. Across all grade 2 and 3 gliomas (n= 6), mean TSPO BPND values in low MD habitats (0.0383 ±0.102) were significantly greater than high MD habitats (-0.045 ±0.142) (p= 0.003). In grade 4 gliomas (n= 5), no significant difference was observed in mean BPND values between low (0.514 ±0.176) and high (0.465 ±0.296) MD habitats (p= 0.543). IDH mutant gliomas (n= 7) exhibited significantly higher mean BPND in low MD habitats (0.116 ±0.225), relative to high habitats (0.052 ±0.287) (p= 0.021). IDH wildtype gliomas (n= 4) showed no significant differences between BPND values in low (0.498 ±0.199) and high MD habitats (0.424 ±0.325) (p= 0.467). Our novel findings demonstrate spatial heterogeneity in TSPO expression within 2/3/IDH mutant gliomas, a characteristic absent from grade 4/IDH wildtype varieties. Regions of restricted diffusion with high TSPO signal in grade 2/3/IDH mutant tumours indicate the heterogenous distribution of active TSPO-expressing immune/neoplastic cell populations, contrasting the homogenous distribution in grade 4/IDH wildtype gliomas, highlighting differences between their inflammatory profile and pathological features.

TMET-25. WILD-TYPE IDH1 INHIBITS FERROPTOSIS IN GLIOBLASTOMA

Abstract De novo lipogenesis and the defense against oxidative lipid damage require large amounts of cytosolic NADPH. Our group has recently found that glioblastoma (GBM) up-regulate wild-type Isocitrate dehydrogenase 1 (referred to hereafter as ‘wt-IDH1high GBM) that, through oxidative decarboxylation of isocitrate, generates large quantities of alpha-ketoglutarate and cytosolic NADPH. We demonstrated that RNAi-mediated knockdown of wt-IDH1, alone and in combination with RT, slows the growth of patient-derived HGG xenografts, while overexpression of wt-IDH1 promotes intracranial HGG growth. Here, we demonstrate that wt-IDH1high GBM produce excess NADPH, which is a rate-limiting reductant that drives de novo fatty acid biosynthesis. Isotope tracer and liquid chromatography-based lipidomic studies indicate that wt-IDH1 supports the de novo biosynthesis of mono-unsaturated fatty acids (MUFAs), promotes the incorporation of monounsaturated, and reduces the abundance of oxidizable polyunsaturated (PUFA) phospholipids in the cell membrane. Conversely, the genetic ablation of wt-IDH1 or the pharmacologic inhibition of wt-IDH1 in tumor cells, using a brain-penetrant wt-IDH1-specific small molecule, revealed a significant increase in PUFA-containing lipid species harboring oxidized arachidonic tails. Of note, two lipid species, (PE16:0_20:4) and (PE18:0_22:4), previously reported to act as substrates for 15-lipoxygenase (15-LOX) - a key enzyme that facilitates lipid peroxidation - are enriched in tumor cells with wt-IDH1 compromise. Further, we found that the enhanced NADPH production in wt-IDH1high GBM, together with heightened biosynthesis of carbon precursors required for glutathione biosynthesis, increased levels of reduced glutathione (GSH), activate the phospholipid peroxidase glutathione peroxidase 4 (GPX4)-driven lipid repair pathway, which scavenges PUFA-containing lipid peroxides, known executioners of ferroptosis. These findings support elevated wt-IDH1 expression, the ensuing effect on the tumor lipidome, redox homeostasis, and protection against lipid peroxidation as contributors to GBM tumor growth and therapy resistance and point to the inhibition of wt-IDH1 as a ferroptosis-inducing strategy for the treatment of GBM.

TMET-11. IMPACT OF METABOLIC RISK FACTORS ON SURVIVAL IN GLIOBLASTOMA IDH-WILD TYPE

Abstract Altered cellular metabolism is a hallmark of cancer. Systemic metabolic dysregulation has been associated with the pathogenesis of cancer, including glioblastoma. In particular, the metabolic syndrome has been found to increase the risk of developing various systemic cancers, but it is not known how often the metabolic syndrome occurs in patients with glioblastoma or how it affects their clinical outcomes. This study retrospectively investigates the prevalence and impact of metabolic syndrome, and its component risk factors, on survival in patients with glioblastoma IDH-wild type. Interim analysis includes results from 73 patients with anticipation of over 100 patients who received treatment at UC Davis between 2018-2024. Metabolic syndrome prevalence prior to diagnosis was 41% (30/73), higher than the prevalence of 33% in the general population. Patients with metabolic syndrome showed a trend towards reduced overall survival. A significant dose-response association was found between the number of metabolic risk factors and overall survival (p=0.03). Hyperglycemia demonstrated worsened survival (p=0.04), aligning with prior studies. Obesity, hypertension, and dyslipidemia were not significantly associated with survival. These findings highlight the therapeutic potential for targeting systemic metabolic dysregulation, and hyperglycemia in particular, in patients with glioblastoma.

BIOM-18. MAPPING TUMOR HABITATS IN IDH-WILD TYPE GLIOBLASTOMA: INTEGRATING MR IMAGING, PATHOLOGIC, AND RNA DATA FROM IVY GLIOBLASTOMA ATLAS PROJECT

Abstract BACKGROUND To identify biologic correlates and spatially validate intratumoral subregions (tumor habitat) using diffusion-weighted and dynamic susceptibility contrast-imaging on a comprehensive pathology map of the isocitrate dehydrogenase (IDH)-wild type whole glioblastoma sample using the Ivy Glioblastoma Atlas Project (Ivy GAP). METHODS Complete data of 20 patients with 168 slides of IDH-wild type glioblastoma were obtained from the IvyGAP (http://glioblastoma.alleninstitute.org/). On MRI, tumor habitats were defined using voxel-wise clustering of apparent diffusion coefficient (ADC) and cerebral blood volume (CBV) maps for contrast-enhancing lesion (CEL) and peritumoral non-enhancing lesion (NEL). On pathology slides, normalized areas of leading edge (LE), infiltrating tumor (IT), cellular tumor (CT), hypervascular lesion (CThypervascular), and perinecrotic lesion (CTperinecrotic) were obtained. Gross specimen pictures were co-registered on MRI images for co-localization and correlation between MRI tumor habitats. Pathologic areas were calculated using Pearson’s correlation coefficient. RNA sequencing of 67 RNA-seq samples was assessed using 4 Neftel subtypes and further correlated with pathology. RESULTS Six tumor habitats were correlated: hypervascular, hypovascular cellular, and hypovascular hypocellular habitats for CEL and NEL. CT was strongly correlated with hypovascular cellular habitat in CEL (C2, r = 0.238, p =.005). IT was strongly correlated with hypovascular cellular habitat in NEL (C5, r = 0.294, p =.017). CThypervascular was correlated with hypervascular habitat in NEL (r = 0.195, p =.023). CTperinecrotic was correlated with imaging necrosis (r = 0.199, p =.005). Astrocyte-like subtypes were positively correlated with IT (r = 0.256, p <.001), while mesenchymal-like subtypes were positively correlated with CTperinecrotic area (r = 0.246, p <.001). CONCLUSION Pathologically matched tumor subregions were cellular tumor with hypovascular cellular habitat in CEL and infiltrative tumor with hypovascular cellular habitat in NEL. Identification of the most aggressive tumor portion, and infiltrative tumor portion using non-invasive imaging can be achieved using MRI tumor habitats.

SURG-50. CLINICAL EXPERIENCE AND PATHOLOGICAL FEATURES IN AUTOPSY CASES TREATED WITH TUMOR TREATING FIELDS

Abstract BACKGROUND Since 2017 Tumor Treating Fields (TTF) was approved for primary glioblastoma in Japan. In this study, we reviewed clinical experience and the pathological characteristics of autopsy cases treated with TTF. METHODS From March 2018 to April 2022, 67 patients (36 males and 31 females, median age 68 years) with primary glioblastoma, IDH-wild type treated at our institution were included in the analysis. RESULTS Twenty-eight patients (19 males and 9 females, median age 66 years) met the TTF indication criteria for induction, 19 cases (67.9%) were 65 years or older, and 14 cases (50.0%) indicated MGMT promoter methylation. The mean duration of TTF treatment was 7.2 months. The main reasons for discontinuation were tumor recurrence in 11 patients (40.7%), skin symptoms in 8 patients (29.6%), and psychological distress in 6 patients (22.2%). Only 12 patients (42.9%) underwent TTF treatment over 18 hours per day, because of discomfort of wearing pad and concern about srrounding. Median PFS and OS in the used patients were 9.9 and 17.7 months, respectively. Autopsies were performed in 3 of the 16 deaths. Histopathological examination of tumor specimen indicate that the Ki-67 labeling index was higher in brainstem (33.6%) than in the supratentorial primary lesion (23.6%). DISCUSSION Comparing to the previous report (Stupp R et al., JAMA. 2017), our cases contained a large number of patients over 65 years old with shorter mean duration of TTF treatment, and shorter OS. After TTF treatment, the Ki-67 labeling index was higher on infratentorial than on supratentorial in the autopsy specimens, suggesting the anti-tumor effect of TTF for the supratentorial lesion. RESULTS High proliferative activity of infratentrial lesion was suggested as a typical form of progression for TTF treatment.

BIOM-63. LONG-TERM SURVIVAL AND MOLECULAR CHARACTERIZATION OF IDH-WILDTYPE GLIOBLASTOMA IN AN ASIAN COHORT

Abstract BACKGROUND Glioblastoma (GB), the most aggressive form of brain tumor, presents a significant clinical challenge with a devastating prognosis. Globally, the median overall survival (OS) for GB is typically 12-15 months, largely due to its inherent resistance to chemotherapy and radiotherapy. However, some patients exhibit prolonged survival, and their specific demographic and genomic characteristics remain poorly defined. Understanding the clinical profile of these exceptional responders holds great promise for informing improved management strategies for GB. METHODS In a retrospective two-center study conducted in Hong Kong (IRB: UW 21-189), we recruited 80 patients diagnosed with IDH wildtype glioblastoma. All patients underwent next-generation sequencing as part of their treatment. Long-term survivor (LTS) was defined as >=36 months from diagnosis, while medium-term survivor (MTS) was defined as >=18 months, and <18 months as short-term survivors (STS). When appropriate, descriptive statistics were used to report frequency with percentage or median with range. Fisher’s exact test was employed to analyze categorical variables, while continuous variables were assessed using the Kruskal-Wallis test. All statistical analyses were conducted using R (version 4.2.3). Result: Among the cohort of 80 recruited patients, the median survival was 18.8 months (95% CI 17.2-24.8). Notably, 21% (n=17) were LTS, 31% (n=25) were MTS, and 48% (n=38) were STS. A statistically significant variation (p < 0.05) was evident across three groups: gender distribution, the number of resections performed, the proportion of patients receiving genomic-guided treatment, and their participation in clinical trials. However, no significant difference was observed in MGMT promoter status and ECOG score. CONCLUSION Identifying these clinical factors enhances the clinical management of GB. The implications of these findings underscore the potential importance of incorporating genomic-guided treatment in both local and global clinical practices in GB.

SURG-47. SUPRAMAXIMAL RESECTION IMPROVES OVERALL SURVIVAL IN PATIENTS WITH MGMT-UNMETHYLATED GLIOBLASTOMA

Abstract INTRODUCTION Glioblastoma is highly resistant to radio- and chemotherapy and its heterogeneity is a therapeutic challenge. O6-methylguanine–DNA methyltransferase (MGMT) promotor methylation is an established biomarker for favorable effect of chemotherapy. Supramaximal resection has proven predict the best overall survival (OS), but whether this benefit applies equally to patients with different MGMT-promotor methylation status is still to be debated. MATERIAL AND METHODS Population based retrospective study that examined all adult patients in South-Eastern Norway that underwent resection for IDH wildtype glioblastoma between January 2019 and December 2021. Extent of resection (EOR) was classified according to RANO resect classification and MGMT-promotor methylation status by pyrosequencing-qPCR. RESULTS The study included 281 patients. MGMT-promotor methylated patients (MGMT+) had a median OS of 18.4 months compared to 13.1 months in MGMT-unmethylated (MGMT-) patients (p<0.0001). Patients with supramaximal resection had a median OS of 20.4 months compared to 14.9 months (p<0.01) with maximal resection. For patients with supramaximal resection we found no significant difference in median OS (25.5 vs. 18.3 months (p=0.24)) dependent on MGMT-status. Separated analysis between MGMT-status and EOR showed a median survival of 18.3, 13.4 (p<0.001), and 8.5 (p<0.001) months for MGMT-, and 25.5, 18.4 (p=.99), and 9.0 (p<0.001) months for MGMT+ in patients where supramaximal-, maximal-, and submaximal resection was achieved, respectively. CONCLUSIONS Patients with supramaximal resection and MGMT+ have a 50% 2-years survival probability. In patients where supramaximal resection was achieved, survival benefit was not dependent on MGMT-status. There was a significant improved survival in MGMT- patients between supramaximal and maximal resection, that was not present for MGMT+ patients. Intraoperative molecular diagnosis has the potential to guide the surgical strategy dependent on MGMT-promotor methylation status, especially in tumors where supramaximal resection has a higher risk of new neurological deficits.

RADT-17. FRACTIONATED REIRRADIATION OF RECURRENT IDH WILDTYPE GLIOBLASTOMA: HIGHER REIRRADIAITON DOSE AND SAFETY

Abstract The optimal regimen, normal tissue tolerances, and appropriate indications for reirradiation for recurrent glioblastoma (GBM) are uncertain. We evaluate outcomes and toxicity after fractionated reirradiation. Patients with recurrent IDH wildtype GBM after prior standard radiotherapy during 2007-2022 were retrospectively reviewed. All patients initially received conventionally fractionated radiotherapy. Radiation necrosis (RN) was determined by imaging or pathology. Overall survival (OS) was estimated using Kaplan-Meier methodology. Univariate and multivariate Cox proportional hazard models for OS were constructed. Bivariable associations with RN were estimated using fisher exact tests. 99 patients (median age 56 yrs, median KPS 80) were treated with reirradiation (median followup 7.4 mo). Median time between initial and reirradiation was 20 months (range, 2 – 214 mo). Median received dose at reirradiation was 41.4 Gy (30.6 – 60 Gy) to a median planning target volume (PTV) of 183 cc (16 – 825 cc). Median cumulative maximum dose to the optics and brainstem were 54 Gy (4 – 103 Gy) and 76 Gy (5 – 146 Gy), respectively. Median OS was 9 mo (95% CI 6.2 – 10.6 mo). On univariate analysis, time between initial and reirradiation ≥24 mo (HR 0.58, CI 0.36 – 0.94) and reirradiation dose ≥41.4 Gy (HR 0.56, CI 0.35 – 0.88) were associated with longer OS. On multivariate analysis, interval between radiation sessions (HR 0.56, CI 0.34 – 0.91) and higher reirradiation dose (HR 0.55, CI 0.35 – 0.87) remained associated with OS. No cases of radiation injury to the optics or brainstem were reported. 10% of patients experienced RN. Reirradiation dose ≥41.4 Gy, reirradiation PTV ≥200 cc and interval between radiation sessions < 24 mo were not significantly associated with RN. These data support the safety and efficacy of fractionated reirradiation for recurrent GBM including for large tumors. This study suggests that higher reirradiation dose may be feasible, including for large treatment volumes.

NIMG-51. PROSPECTIVE CASE CONTROL STUDY OF HYPOXIA FMISO PET/MRI AS A DIAGNOSTIC FEATURE OF GLIOBLASTOMA IMMUNOTHERAPY FAILURE

Abstract BACKGROUND Hypoxia is a prognostic biological feature of glioblastoma. 18F-fluoromisonidazole (FMISO) PET quantifies tumor hypoxia. Efficacy of checkpoint immunotherapy is being investigated, but specific response assessment criteria remain unknown. We hypothesized that tumor hypoxia is a defining feature of immunotherapy failure in patients with glioblastoma. METHODS A prospective case control study was undertaken in patients with newly diagnosed IDH wildtype glioblastoma (NCT03649880). Cases were concurrently enrolled in a phase II study (NCT03347617) of pembrolizumab immunotherapy (200 mg IV every 3 weeks) combined with Stupp protocol temozolomide-based chemoradiotherapy following maximal safe resection. Controls did not receive pembrolizumab. FMISO PET/MRI (3.7 MBq/kg, 90 min uptake) was performed at time of mRANO presumed progression. Hypoxic volume (HV) was defined by pixels with FMISO uptake ≥1.2x that of mean cerebellar uptake. MRI-Gd enhancing volume was segmented using a semi-automated approach. Hypoxic fraction (HF) was defined as the HV-to-MRI enhancing volume ratio. Disease progression or pseudoprogression (Psp) was confirmed through surgery or follow-up MRI. T-test and receiver operator characteristic (ROC) analysis assessed differences and diagnostic performance. RESULTS 18 patients (14 cases [11 progression and 3 Psp] and 4 controls [3 progression and 1 Psp]) provided evaluable disease. For all patients, HF was significantly elevated with progression (1.44 ± 0.8; N=14) compared to Psp (0.31 ± 0.3; N= 4; P=0.01). HV was not different (P=0.10). No difference in HV or HF was observed between Cases and Controls (P> 0.42). Use of HF yielded an AUROC of 0.96 with an optimal cut off of 0.74, to differentiate progression from Psp. CONCLUSION At mRANO presumed progression HF with FMISO PET/MRI accurately distinguishes tumor progression from neuroinflammatory Psp, irrespective of prior pembrolizumab immunotherapy. The incorporation of FMISO PET/MRI in larger prospective therapeutic clinical trials in patients with glioblastoma is underway and likely to improve response assessment criteria.

NIMG-82. DEVELOPMENT AND VALIDATION OF AN MRI-BASED DEEP LEARNING MODEL TO DIFFERENTIATEIDH-WILDTYPE GLIOBLASTOMA AND TUMEFACTIVE MULTIPLE SCLEROSIS

Abstract Clinical management of IDH wildtype glioblastoma (GBM) and tumefactive multiple sclerosis (tMS) is drastically different. GBM requires maximal safe resection followed by chemoradiation, while tMS outcome is worsened by surgery and radiotherapy. Noninvasive methods are needed to help with accurate diagnosis of tumor and non-tumor etiologies. To develop an MRI-based classification model, tMS subjects diagnosed prior to January 1, 2020, were matched to tMS by age at diagnosis, sex, index MRI date, and 2D/3D acquisition. Inclusion criteria included one cm minimal lesion size and pre-operative post-contrast T1 and T2 images available for analysis. A 3D-DenseNet121 was used to develop a classification model using prespecified parameters: 650 epochs, batch size 16, learning rate 10-3, cross-entropy loss, and AdamW optimizer. The stopping rule was defined as three sequential differences in epoch cross-entropy loss <0.02. Models were developed using both T1gd and T2, as well as from only T1gd and only T2. Training included 220 subjects (110 GBM, 110 tMS). A 2-stage validation design was used, which included both retrospective and prospective cohorts. Stage 1 consisted of 272 retrospective GBM (diagnosed prior to January 1, 2020). Stage 2 consisted of 69 and 34 prospective (diagnosed after January 1, 2020) GBM and tMS, respectively. External validation on the 272 retrospective GBM demonstrated accuracy of 91%, 84%, and 78% for T1gd+T2, T1gd only, and T2 only, respectively. External validation on the 69 prospective GBM demonstrated an accuracy of 87%, 64%, and 67% for T1gd+T2, T1gd only, and T2 only, respectively. The 34 prospective tMS demonstrated accuracy of 76%, 76%, and 82% for T1gd+T2, T1gd only, and T2 only, respectively. This shows the feasibility of deep learning to aid in differential diagnosis of brain lesions. Future work entails the integration of germline variants into the classification model, including variants associated with the risk of glioma or MS.

BIOM-21. SMALL EXTRACELLULAR VESICLES AS A NOVEL LIQUID BIOPSY APPROACH FOR GLIOBLASTOMA

Abstract Glioblastoma (GBM) remains the most common and aggressive primary malignant brain tumor, with limited treatment options and poor overall survival. Diagnosis, prognosis, and assessment of treatment effectiveness are hampered by the lack of sensitive, tumor-specific, non-invasive blood-based assays which could be followed serially. Small extracellular vesicles (sEV) are nano-sized (≤200 nm) membrane-bound bodies secreted by all cells, encapsulating cargo reflective of their parent cells. sEVs have emerged as promising molecular disease indicators. Here, we report the feasibility of isolating glioma-specific sEV (sEVglioma) from plasma and characterizing them for GBM-specific molecular biomarkers. Plasma samples were collected from 31 glioma patients (grades 3 and 4 GBM, grades 2 and 3 astrocytoma) and 9 healthy individuals. Total sEVs (TE) were isolated from plasma by a modified precipitation (ExoQuick™) method, followed by immunoprecipitation to isolate sEVglioma employing surface markers targeting the cellular origin of gliomas, including astrocyte (GLAST and EAAT2), oligodendrocyte precursor cell (OSP and MOG), and neural stem cell (CD133). The average size of sEVglioma was less than 200 nm. Importantly, compared to TE, sEVglioma showed significant enrichment for glioma-specific biomarkers such as ephrin type-A receptor 2 (14.7-fold), tenascin-C (22.7-fold), and glial fibrillary acidic protein (8.4-fold). Similarly, sEVglioma demonstrated higher expression of the GBM biomarker EGFRvIII (3.6-fold). These results were confirmed by confocal microscopy. Interestingly, the expression of specific miRNAs (miR-9a-5p, miR-16-5p, miR-21-5p) in sEVglioma was higher in glioma patients with shorter survival (<12 months) compared to longer survival (>20 months). Lastly, we successfully detected the existence of wild-type IDH1 and absence of mutated IDH1 R132H in sEVglioma from GBM patients, validated with cell line experiments. In conclusion, we present a novel liquid biopsy approach for isolating sEVglioma from blood, providing valuable molecular and genetic information. This approach promises early detection, potential to distinguish pseudo-progression, and assessment of treatment effectiveness with remarkable precision.

TMIC-66. A NEW MOUSE MODEL THAT COMBINES DRIVER MUTATIONS TO ACCURATELY RECAPITULATE THE DIVERSE PHENOTYPES OF HUMAN GLIOBLASTOMA

Abstract To effectively beat glioblastoma, new pre-clinical models are needed: to dissect the biology of processes like migration and therapy resistance; to understand immune suppression and immunotherapy; and to use as a model for testing in therapeutic development. We have developed a new panel of glioblastoma mouse models, which engineer well-characterised glioblastoma driver mutations into an immortalised eGFP murine astrocyte-like cell. These “GEM-CLeM” are Genetically Engineered Mouse Cell Line Models. When transplanted intra-cranially into immunocompetent mice, these lines formed tumors with the key characteristics of glioblastoma – migratory and invasive, with rapid proliferation and infiltration of suppressive immune cells. The eGFP expression allowed detection of glioblastoma cells in situ and ex vivo, and the ability to readily distinguish tumor and stromal components of the tumors. Importantly, different combinations of driver mutations led to diverse tumor phenotypes. Loss of Pten with over-expression of mutant RAS generated tumors with 100% efficiency and the same time to tumor onset as the commonly used GL261 cell line. However, unlike GL261 tumors the Pten-RASV12 tumors had a highly invasive edge and extensive migration, and were full of pro-tumorigenic macrophages and monocytes, with very few T-cells or dendritic cells. Together, this recapitulated the phenotype of human wildtype IDH1 glioblastoma. In contrast, the very common EGFRvIII mutant did not drive tumor formation in combination with Pten loss. We also modelled mutant IDH glioblastoma by combining loss of p53 with overexpression of IDH1R132H. These were slow growing, densely packed tumors with extensive microglial involvement in addition to pro-tumorigenic macrophages. This GEM-CLeM system combines the benefits of the traditional genetically engineered mouse models with accessible, rapid, reproducible tumor formation. The GEM-CLeM lines are amenable to further genetic manipulation, so cells can be customised with other mutations to ask important biological questions in the presence of an intact tumor immune microenvironment.