Abstract Background: Cerebrospinal fluid (CSF) is an increasingly important liquid biopsy for brain tumors, offering improved sensitivity over plasma due to proximity to the tumor and reduced peripheral DNA background. However, cfDNA concentrations in CSF are low, and extraction inefficiencies can limit the reliability of downstream molecular assays. This study evaluates cfDNA recovery performance from CSF-like matrices to support sensitive molecular profiling relevant to brain cancer applications. Methods: Synthetic CSF samples spiked with fragmented cfDNA reference material were processed using the nRichDX Revolution Max20 workflow. Low-input volumes representative of clinical CSF collections were evaluated. cfDNA yield and recovery were quantified using fluorometric and quantitative PCR-based measurements, and fragment size distributions were assessed to confirm recovery of short cfDNA fragments typical of CSF-derived material. Results: The extraction workflow consistently recovered cfDNA from synthetic CSF with high reproducibility across replicates and input volumes. Quantitative measurements demonstrated reliable cfDNA yields suitable for sensitive downstream molecular analyses. Fragment size analysis confirmed preservation of short cfDNA fragments, including those within the range relevant for mutation detection and copy number analysis. Minimal variability was observed between runs, indicating robustness of the extraction process under low-input conditions. Conclusions: Efficient and reproducible cfDNA extraction from CSF-like matrices enables molecular profiling approaches that are critical for brain cancer research and clinical translation. The performance demonstrated in this study supports clinical applications including detection of tumor-derived genetic alterations, longitudinal disease monitoring, and assessment of treatment response in central nervous system malignancies. Reliable cfDNA recovery from limited CSF volumes is a key enabling step for integrating CSF-based liquid biopsy into precision oncology workflows for brain cancer. Generative AI Disclosure: Generative AI was used solely to assist with editing and improving the clarity of this abstract; all scientific content and conclusions were reviewed and validated by the authors prior to submission. Citation Format: Nafiseh Jafari, Jason Saenz, Carlos Hernandez, Daniel Cedeno, Cameron Van Dieren, Mayer Saidian. Efficient recovery of cell-free DNA from synthetic cerebrospinal fluid enables sensitive molecular profiling in brain cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A052.

Abstract Brain tumors, particularly high-grade gliomas such as glioblastoma multiforme (GBM), remain among the most aggressive and treatment-resistant malignancies. Their poor prognosis is driven by extreme cellular heterogeneity, rapid proliferation, infiltrative growth, and the presence of therapy-resistant cancer stem cells (CSCs). Increasing evidence highlights a central role for epigenetic dysregulation—especially abnormal histone deacetylase (HDAC) activity—in tumor initiation, progression, and therapeutic resistance. HDAC overexpression, particularly HDAC1, HDAC2, HDAC6, and several SIRT family members, is frequently observed in GBM and correlates with tumor aggressiveness, metabolic plasticity, and enhanced survival signaling. Histone deacetylase inhibitors (HDACi) have emerged as promising therapeutic agents capable of targeting multiple malignant features simultaneously. Preclinical studies demonstrate that HDACi can reduce CSC viability, promote apoptosis and autophagy, induce cell-cycle arrest, and impair angiogenesis and tumor invasion. HDACi also modulate pathways such as STAT3, SIRT, and SHH, thereby disrupting essential networks that support glioma growth. Importantly, several HDAC inhibitors have demonstrated the ability to cross the blood–brain barrier, increasing their potential for use in primary brain tumor therapy. Furthermore, combining HDACi with radiotherapy or chemotherapy has shown synergistic effects, improving tumor sensitivity and overcoming resistance mechanisms. Although current clinical progress is preliminary, accumulating data suggest that HDAC inhibitors may represent a valuable addition to future multimodal strategies for GBM and other central nervous system tumors. Continued investigation is needed to optimize their therapeutic integration and identify the most responsive patient populations. Citation Format: Laila T A. Metwaly. Histone deacetylase inhibitors as emerging therapeutic strategies for glioblastoma and other brain tumors [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A037.

Abstract Background: Recurrent pediatric ependymoma remains a major therapeutic challenge with few evidence-based systemic options. NCCN guidelines emphasize maximal safe resection and radiotherapy, without systemic chemotherapy and which should generally be reserved for recurrence due to limited and unpredictable benefit. However, personalized functional profiling may provide actionable, patient-specific evidence by directly measuring tumor drug sensitivity and resistance ex vivo. Methods: Fresh surgical specimens from three pediatric recurrent ependymomas were processed into viable tumor fragments and 3D microtumors analyzed using the SAGE Oncotest™, a multiplexed ex- vivo functional profiling assay measuring both cytotoxicity and antiproliferation. NCCN-relevant agents were tested at standardized and physiologically representative concentrations over several logs. Cytotoxicity (loss of viability) and antiproliferation (inhibition of DNA synthesis) were quantified using our proprietary algorithmic drug sensitivity score (DSS) with percent effect relative to untreated controls. Results: Responses were highly patient-specific and frequently uncoupled between growth arrest and cell death, indicating distinct underlying dependencies. Cytotoxic chemotherapies: Topoisomerase I–based therapy showed the most consistent dual activity, consistent with replication stress vulnerability. Irinotecan produced cytotoxicity across cases (20–77%) with strong antiproliferation (85–99%). Gemcitabine produced strong antiproliferation (96–99%) with variable cytotoxicity (0–45%), consistent with S-phase suppression without uniform tumor kill. CDK4/6 inhibitors: Intra-class divergence suggested differences in target selectivity and pathway reliance. Ribociclib produced strong antiproliferation (72–96%) with minimal cytotoxicity (5–33%), consistent with cytostatic CDK4/6 blockade, whereas abemaciclib showed in one tumor a striking dual activity (cytotoxicity 81%, antiproliferation 97%), consistent with broader kinase activity beyond CDK4/6. With the exception of venetoclax, most other targeted agents were largely inactive or cytostatic only. Combinations: Combinations did not reliably exceed the best single agent and occasionally reduced cytotoxicity, consistent with independent and sometimes antagonistic interactions. Conclusions: Current NCCN guidelines for ependymoma emphasize surgical resection and radiotherapy without routine systemic therapy, a strategy supported by broad drug resistance across most tumors. However, multiplex ex vivo profiling revealed patient-specific vulnerabilities, including strong antiproliferative effects and, in selected cases, measurable cytotoxicity. Functional profiling may therefore enable rational selection of individualized systemic therapies in recurrent ependymoma, with the potential to improve quality of life, prolong progression-free survival, and possibly overall survival. Citation Format: Rajeshwar Nitiyanandan, Ivan Trus, Chiara Maestri, Shawn Mahoney, Ricardo J. Parker, John Norko, Sheila Singh, William Cance, Chris Apfel. Functional precision oncology in recurrent ependymoma: Multiplexed functional profiling reveals unexpected cytotoxic and anti-proliferative drug activity [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A054.

Abstract Glioblastoma (GBM) is the most common and lethal primary brain malignancy in adults, with a median survival of 12–14 months and a 5-year survival rate of ∼6%. Despite decades of effort, patient outcomes have improved minimally, underscoring an urgent need for more effective and targeted therapies. We identified the bromodomain-containing chromatin regulator BRD8 as an essential dependency in TP53-wild-type (TP53WT) GBM, which comprises ∼71% of cases. BRD8 sustains malignancy by disabling p53-mediated tumor suppression through a previously unrecognized epigenetic mechanism: BRD8 reprograms the p53 regulatory network via the EP400 histone acetyltransferase complex and bromodomain-directed occupancy of the histone variant H2AZ at p53 target genes, enforcing a repressive chromatin state that blocks p53-dependent transactivation. Targeting BRD8 remodels chromatin by evicting H2AZ and increasing accessibility, and restores p53 binding and transcriptional activity. This reactivation of p53 restores gene expression, induces cell-cycle arrest, suppresses gliomagenesis, and prolongs survival in TP53WT GBM xenograft models. Our recent work further demonstrates that pharmacological targeting of BRD8 selectively restores p53 tumor-suppressive activity in TP53WT GBM cells, but not in non-malignant brain cells. Notably, BRD8 inhibition synergizes with multiple therapeutic pathways and enhances the efficacy of standard-of-care GBM treatments in patient-derived TP53WT GBM models. Together, these findings uncover a previously unappreciated epigenetic mechanism by which GBM cells evade p53-mediated tumor suppression, establish BRD8 as a tractable epigenetic vulnerability in TP53WT GBM, and nominate rational combination therapeutic strategies for GBM treatment. This work reveals new principles of chromatin-based tumor suppression and offers promising therapeutic avenues for this devastating disease. Citation Format: Xueqin Sherine Sun. An epigenetic dependency on BRD8 in TP53-wild-type glioblastoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr B008.

Abstract Melanoma brain metastasis (MBM) is linked to dismal prognosis and onset of brain metastasis is increasing in frequency in all cancer types. Circulating Tumor Cells (CTCs) are "seeds" of fatal metastasis and smallest functional units of cancer. We previously employed an unbiased, multipronged approach to discover a unique ribosomal protein large/small subunit (RPL/RPS) CTC gene signature associated with MBM. In this study, we hypothesized that MBM-CTC drives secondary metastasis (“metastasis of metastasis” per clinical scenarios), and has targeted organ specificity for the liver. We injected parallel cohorts of immunodeficient and newly developed humanized NBSGW (huNBSGW) mice with cells from CTC-derived melanoma brain metastasis to identify secondary metastatic patterns. We found the presence of a melanoma brain-liver metastasis axis in huNBSGW mice. Furthermore, RNA sequencing analysis of tissues showed a significant upregulation of the RPL/RPS CTC gene signature linked to metastatic spread to the liver. Additional RNA sequencing of CTCs from huNBSGW blood revealed extensive CTC clustering with human B cells in these mice. CTC:B-cell clusters were also upregulated in the blood of patients with primary melanoma and maintained either in CTC-driven MBM or MBM CTC-derived cells promoting extensive liver metastasis. CTC-generated tumor tissues were interrogated at single-cell gene and protein expression levels (10x Genomics Xenium and HALO spatial biology platforms, respectively). Collectively, our findings suggest the presence of a CTC brain-liver metastasis axis, and that heterotypic CTC:B-cell interactions can be critical at multiple stages of metastasis. Significance: This study provides important insights into the relevance of prometastatic CTC:B-cell clusters in melanoma progression, extends the importance of the CTC RPL/RPS gene signature beyond primary metastasis/MBM by driving targeted organ specificity for liver metastasis (“metastasis of metastasis”), and identifies new targets for therapies against clinical melanoma metastasis, eg, ongoing studies involcving cryogenic electron microscopy analyses of CTC ribosomes. Citation Format: DARIO MARCHETTI. The Melanoma CTC Brain-Liver Metastasis Axis [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr B067.

Abstract Background In this pooled analysis of 7 multicenter prospective studies, Pooled Analysis of 7 Multicenter Prospective Studies (n=212) Using an Integrated Multi-Compartment Biomarker Ecosystem to Differentiate True Progression from Treatment Effect in High-Grade Glioma we confront the pressing clinical dilemma of distinguishing true tumor progression from treatment effect (pseudoprogression/radiation necrosis). We analyzed 212 patients (140 newly diagnosed, 72 recurrent) with high-grade glioma (160 IDH-wildtype GBM; 52 other HGG). Conventional MRI remains frequently ambiguous and blood–brain barrier dynamics limit single-compartment assays, so a multimodal ecosystem may be required. Methods We integrated plasma cfDNA methylation profiling, paired CSF and plasma proteomics, MRI radiomics, and an AI ensemble classifier trained on multimodal features. Tissue/CSF concordance was assessed (κ = 0.78) and all cases underwent blinded adjudication by two neuro-oncologists. Prespecified thresholds used the Youden index for primary discrimination and a 90% sensitivity rule to prioritize clinical safety. Feasibility: cost ≈ $250/sample, turnaround 48–72 hours, and scalable lab and computational workflows. Thresholding balanced false negatives against harms of unnecessary intervention. Results / Key Findings / New Concepts The ensemble achieved AUC 0.92, sensitivity 91%, and specificity 84%, with a median lead time gain of 6.2 weeks over radiographic consensus. Proteomic signals (IL-6, CXCL1) and a myeloid inflammation signature were enriched in treatment effect and correlated modestly with tumor mutational burden (r≈0.42). We introduce the dynamic biomarker axis and ecosystem monitoring—longitudinal, multi-compartment integration that appears to resolve many ambiguous cases where single modalities fail. Conclusions This pooled, multimodal ecosystem may enable earlier therapy adaptation, reduce unnecessary surgery and steroid exposure, and provide adaptive trial endpoints. Prospective validation and biomarker-guided therapeutic algorithms are warranted; if confirmed, this approach is likely to be paradigm-shifting in neuro-oncology. Citation Format: Mohamed Tharwat. Kamouna, Abdelrahman Wael . Ibrahim, Abdullah Wael . Ibrahim. Pooled Analysis of 7 Multicenter Prospective Studies (n=212) Using an Integrated Multi-Compartment Biomarker Ecosystem to Differentiate True Progression from Treatment Effect in High-Grade Glioma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A065.

Abstract Brain tumors may start long before diagnosis, but the earliest genetic event and the first mutant “normal” cell remain unclear. We used multi-regional surgical sampling plus single-cell/spatial genomics and mouse models to map early initiation steps in glioblastoma (GBM), IDH-mutant glioma, and meningioma. We collected matched tumor and non-tumor tissues from patients. For GBM, we profiled tumors and matched non-tumor regions, including the tumor-free subventricular zone (SVZ) and peritumoral cortex, using single-nucleus RNA sequencing (snRNA-seq) with copy-number inference (13 tumors, 10 SVZ, 2 controls). For IDH-mutant glioma, we collected tumor, tumor-free SVZ, peritumoral cortex (n=18), and blood, and performed deep sequencing, droplet digital PCR (ddPCR), and spatial mapping. For meningioma, we analyzed paired meningioma tumors and pathologically tumor-free dura (n=22) using whole-exome sequencing (WES), amplicon sequencing, long-read sequencing, clonal and mutational-signature analyses, and single-nucleus DNA sequencing (snDNA-seq). In parallel, we tested SVZ-specific tumorigenicity in a mouse GBM model and built an oligodendrocyte progenitor cell (OPC)–targeted Idh1(R132H) CRISPR model with Trp53/Atrx/Nf1. GBM: Putative mouse originating cell candidate (mGBM-OCs) formed spheres, showed neuro/glial differentiation, and were highly invasive, but tumorigenicity was niche dependent: cortex implantation produced no tumors (0/20), whereas SVZ implantation produced tumors (11/29). In patients, SVZ neural stem cell–like, astrocyte-like, and OPC-like subsets showed low-level early copy-number signals consistent with chromosome 7 gain and/or chromosome 10 loss, suggesting these subsets as putative human GBM-OCs. IDH-mutant glioma: Tumor-free SVZ showed no shared mutations with tumor. Low-level IDH1(R132H) was detected in peritumoral normal cortex in 11/18 patients (61.1%) and localized mainly to OPC and astrocyte-precursor–like cells, supporting peritumoral cortical glial progenitor lineages as the first mutant cells. In mice, adding Idh1(R132H) increased tumor incidence (OPC-ITAN 18/29, 62.1% vs OPC-TAN 3/7, 42.9%) and worsened survival. Meningioma: Driver alterations were detected in tumor-free dura in 18/22 patients (82%) at very low variant allele fractions, including NF2 and non-NF2 drivers (TRAF7/KLF4/AKT1-related); remaining cases were supported by long-read detection of structural drivers. Dura patterns were inconsistent with tumor microinvasion and supported clonal evolution from dural developmental mosaicism. snDNA-seq identified NF2-mutant cells with chromosome 22q loss in both dura and tumor, indicating prenatal biallelic NF2 inactivation. Shared clonal mutations were enriched for SBS1/5, whereas tumor-private subclonal mutations were enriched for SBS15, consistent with later postnatal steps. These data reveal tumor-type–specific initiation programs and support origin-based therapeutic strategies targeting origin-like cells and their niches, beyond tumor-mass–focused treatment. Generative AI was used only for English-language editing. Citation Format: Hyeong-Chol Oh, Jung Won Park, Young-soo Chung, Ran Joo Choi, Hye Joung Cho, Se-Young Jo, Eunchae Yeo, Jihwan Yoo, Tae Hoon Roh, Jong Hee Chang, Joo Ho Lee, Hoon Kim, Jeong Ho Lee, Sangwoo Kim, Seok-Gu Kang. Multi-regional surgical sampling reveals tumor origins and early developmental pathways in glioblastoma, IDH-mutant glioma, and meningioma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr B039.

Abstract Glioblastoma (GBM) evolves within a microenvironment abundant in oligodendrocyte-lineage (OL) cells. In this study, we utilized single-cell and spatial transcriptomics from primary and recurrent GBM tumors, immunohistochemistry, cytokine profiling, and migration assays to show that GBM cells recruit OLs to the tumor border via CX3CL1/CX3CR1 signaling. A pan-disease human OL meta-atlas and syngeneic mouse models reveal an interferon-induced reactive OL state, akin to those seen in demyelinating inflammatory and traumatic injury, which is enriched in CNS malignancies. These reactive OLs secrete pro-tumorigenic cytokines, notably CCL5, that promote GBM tumor cell growth through CCR5 signaling. CCR5 is preferentially expressed in glioma stem-like cells (GSCs) and upregulated at recurrence. Targeting CCR5 with genetic knockdown or FDA-approved drug Maraviroc impairs GSC stemness and prolongs survival in GBM models. Our work highlights the functional interplay between OLs and GBM cells and positions the CCL5/CCR5 axis as a druggable target in GBM. Citation Format: Nicholas Mikolajewicz, Kui Zhai, Anish Puri, Petar Miltec, Nazanin Tatari, Jiarun Wei, Neil Savage, Zhi Huang, Qian Huang, Seon Yong Lee, Mahta Jan-Ahmadnejad, Roseanne Nguyen, David Chen, Tiegan Korman, Daniel Mobilio, Maxwell Topley, Jack Qinyu Lu, Matthew R. Voisin, Zsolt Zador, Shawn C. Chafe, Chitra Venugopal, Kevin R. Brown, Gelareh Zadeh, Hong Han, Julien Muffat, Shideng Bao, Sheila K. Singh, Jason Moffat. Reactive oligodendrocytes promote glioblastoma progression through CCL5/CCR5-mediated glioma stem cell maintenance [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A040.

Abstract Glioblastoma (GB) remains one of the most lethal brain tumors, largely due to extensive inter- and intratumoral heterogeneity, which limits the efficacy of standard therapies and contributes to frequent recurrence. Tumor Treating Fields (TTFields) provide clinical benefit but fail to prevent tumor relapse, suggesting the presence of resistant subpopulations. Here, we modeled prolonged TTFields exposure in patient-derived GB cultures across multiple subtypes, revealing that a fraction of cells survives and adopts shared features of resistance, including enhanced OXPHOS metabolism and elevated tmCLIC1 expression, a regulator of GB cell metabolism and survival. We demonstrated that tmCLIC1 represents a targetable vulnerability in these resistant populations. Transcriptomic analysis of paired patient tumors pre- and post-TTFields treatment corroborated these findings, highlighting conserved molecular adaptations across subtypes. Pharmacological inhibition of tmCLIC1 with metformin synergizes with TTFields, suppressing proliferation in vitro and reducing tumor growth in vivo, even at low drug concentrations facilitated by TTFields-enhanced blood-brain barrier permeability. Collectively, our results indicate that tmCLIC1-targeted therapy can overcome resistance in heterogeneous GB populations and suggest a rational combinatorial strategy to improve clinical outcomes. Citation Format: Francesca Cianci, Guido Rey, Elisa Meraviglia, Antonio Maria. Polito, Michele Mazzanti. Targeting tmCLIC1 to overcome resistance across heterogeneous glioblastoma subtypes by exploiting metabolic vulnerabilities with synergistic combinatorial therapy [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr B024.

Abstract Glioblastoma (GBM) is the most aggressive primary tumor of the central nervous system, with a median survival of approximately 14 months despite standard-of-care treatment consisting of maximal surgical resection followed by radiotherapy and temozolomide (TMZ) chemotherapy. Although TMZ remains the most commonly used chemotherapeutic agent for GBM, patient responses vary widely. In clinical practice, its benefit is often more limited in older patients (≥65 years) and in those with poorer performance status, who also tend to experience greater treatment-related toxicity. These observations suggest that age-related differences in tumor biology and cellular stress responses may influence therapeutic outcomes and highlight the need for molecular markers that better predict TMZ sensitivity. To explore age-associated transcriptional differences following TMZ treatment, the public bulk RNA-sequencing dataset GSE263890 was reanalyzed, a transcriptomic dataset of GBM tumors from younger and senior patients after TMZ. Raw gene expression data were processed, normalized, and log-transformed, with low-expression and unannotated transcripts removal to improve data quality. Differentially expressed genes (DEGs) were identified using thresholds of |log2 fold change| > 0.2 and FDR < 0.05. Heatmaps and volcano plots were used to visualize expression patterns, and Gene Ontology (GO) enrichment analysis was performed to identify biological pathways. DEGs analysis between younger and older patients revealed 402 up-regulated genes and 522 down-regulated genes in tumors from senior patients. Prominent up-regulated genes included ZYG11A, HMX1, AP1M2, KERA, GARIL6, USP6, SFTPA1, PAX9, TSPAN19, and TPH2, whereas down-regulated genes included MEST, FLNC, H3C12, NDUFB2, CAV1, GAS2L1, ID4, ATP6V1F, POLR2J, DENND2A, EDNRB, TREM2, RARRES2, and CDC42EP1. Gene Ontology enrichment analysis showed that up-regulated genes were significantly associated with pathways related to cilium organization, microtubule bundle formation, axoneme assembly, ciliary movement, and calcium ion transport, pointing to changes in cytoskeletal structure and intracellular signaling in older patients following TMZ exposure. In contrast, down-regulated genes were enriched in pathways involved in cytoplasmic ribosome biogenesis, peptide chain elongation, eukaryotic translation initiation and termination, nonsense-mediated mRNA decay, and EIF2AK4 (GCN2)-mediated amino acid stress responses, suggesting an overall reduction in protein synthesis and translational activity. In conclusion, these results reveal age-associated transcriptional differences in GBM tumors following TMZ treatment. The observed alteration markers in cellular structure, signaling pathways, and translational regulation may help explain differences in therapeutic response and tolerance between younger and older patients. This study provides molecular insight into age-related variability in TMZ efficacy and supports the development of age-informed biomarkers to improve outcomes for patients with GBM. Citation Format: Catherine Chen, Wlliam Xu, Qian Wang. Age-Associated Transcriptomic Alterations in Glioblastoma Following Temozolomide Treatment [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A049.