Abstract Background Fifty percent of SHH-medulloblastoma harbor identical somatic point mutations in a non-coding small nuclear RNA called U1 (r.3A>G). This point mutation is found in the majority of SHH-α tumors (in children and toddlers) with TP53 mutations and in over 95% of SHH-α tumors (in adults). The U1 mutation results in the spliceosome failing to anneal to the proper intron-exon boundaries and instead recognizing false splice junctions inside introns. These aberrantly spliced RNA species include the 5’ end of introns, meaning that portions of the intron are translated into protein until the ribosome encounters a stop codon. This means that the tumors exhibit a unique form of post-transcriptional hypermutation. Therefore, novel epitopes may be identified in these tumors. Hypermutant tumors are responsive to immune checkpoint inhibitors (ICI). However, data for patients with relapsed medulloblastoma treated with ICI are scarce. Methods We established a registry study (NCT07242963) to enroll patients with relapsed SHH-medulloblastoma to assess the incidence of the U1 mutation and compare outcomes of patients with relapsed SHH-medulloblastoma harboring the U1 mutation who receive different therapies, including ICI. Results We activated the registry study (NCT07242963) at Texas Children’s Hospital. The study is actively enrolling patients. The registry’s primary aim is to report on the incidence of U1 mutation in patients with relapsed/refractory SHH-medulloblastoma, and their response to ICI. Our secondary aims include identifying cryptic, novel tumor antigenic epitopes by developing a first long-read sequencing and surfaceome dataset for SHH-medulloblastoma, and developing cellular and immunotherapies targeting the U1-mutation-driven tumor antigenic epitopes. Conclusions A subset of SHH-medulloblastomas may exhibit post-transcriptional hypermutation secondary to aberrant splicing caused by the U1 (r.3A>G) mutation. Given the small number of these patients, our registry study enables multicenter and multinational collaboration to test the effectiveness of ICI in them.

Glioblastoma is the most common malignant brain tumor in adults, and radiotherapy (RT) is the most effective postoperative treatment for patients with glioblastoma. Nevertheless, glioblastoma recurrence after treatment is nearly universal, and cell state plasticity and intratumor heterogeneity underlie glioblastoma evolution and resistance to treatment. Here we integrate in vitro genome-wide CRISPR interference screens and in vivo perturb-seq in preclinical models with single-nucleus and spatial transcriptomic sequencing of human tumors to identify therapeutic vulnerabilities that overcome glioblastoma resistance to RT. Gene regulatory network modeling identifies DNA-PKcs as a RT-sensitizing target in glioblastoma cells in vitro and in vivo. Small molecule inhibition of DNA-PKcs plus RT improves survival and reprograms tumor cell states and immune microenvironment composition compared to DNA-PKcs inhibition or RT monotherapy. Bioinformatic and imaging analyses of patient-matched glioblastoma samples before and after DNA-PKcs inhibition and RT show that combination therapy drives inflammatory gene expression programs in tumor cells that recruits pro-inflammatory myeloid cells to the tumor microenvironment. Using this framework to inform rational sequential therapy in preclinical models, we show that immunomodulation in response to genomic stress after DNA-PKcs inhibition and RT primes glioblastoma for response to cGAS/STING activation. These data show that DNA-PKcs modulates tumor cell states and immune microenvironment cell types to drive resistance to RT in glioblastoma, and that targeting DNA-PKcs sensitizes glioblastomas to RT and cGAS/STING activation.

2053 Background: Diffuse gliomas are the most common primary brain tumors, requiring multimodal therapy. Radiotherapy (RT) represents a key pillar of treatment; however, radiation-induced brain toxicities remain a major challenge. Radiation necrosis (RN) is a delayed complication that can impair clinical outcomes and quality of life. Emerging evidence suggests that tumor biology, together with clinical factors, can influence radiation toxicity in a variety of cancer entities. However, in diffuse gliomas, tumor-intrinsic biomarkers predisposing to radiation toxicities remain poorly defined. We hypothesized that integrating clinical features with tumor genomic data could identify patients at increased risk of developing RN. Methods: We analyzed a cohort of 943 adult patients with diffuse gliomas treated between 2014 and 2024 at MD Anderson. All tumors were classified according to the WHO 2021 criteria into three diagnoses: GBM, astrocytoma (astro), and oligodendroglioma (OD). RN occurring at least 12 weeks after completion of RT was identified based on pathology and/or advanced brain tumor imaging (ABTI), which included conventional MRI, perfusion imaging and spectroscopy. A multivariate regression model was employed to identify clinical or genetic predictors for RN. Included covariates were demographics, diagnosis, and common genetic alterations. To address missing values, multiple imputation was utilized. Results: Sixty patients with RN were identified: 51 GBM (85%), 6 astro (10%), and, 3 OD (5%); compared with 883 controls: 682 GBM (77%), 140 astro (16%), and 61 OD (7%). Age, gender, and diagnoses were not associated with RN. We evaluated the top 22 altered genes and MGMT status. In the overall cohort, IDH1 (OR 0.15 [95% CI, 0.02–0.93]; p=0.042) and MDM2 (OR 0.11 [95% CI, 0.01–1.00]; p=0.050) alterations were associated with reduced RN risk, while ATR alterations were associated with increased RN risk (OR 6.05 [95% CI, 0.85–33.1]; p=0.049) on multivariable analysis. In the GBM subgroup, CDK4 (OR 3.32 [95% CI, 1.04–9.97]; p=0.042) and MSH6 alterations (OR 6.05 [95% CI, 1.12–33.1]; p=0.037) were associated with increased RN risk, whereas MDM2 alterations remained associated with reduced risk (OR 0.07 [95% CI, 0.01–0.69]; p=0.023). In the IDH -mutant diffuse glioma subgroup (RN n=9; controls n=201), ATRX alterations were associated with lower RN risk (OR 0.10 [95% CI, 0.01–0.73]; p=0.023). MGMT status was not associated with RN risk. Conclusions: In this cohort, RN risk was not associated with demographic factors, tumor diagnosis, or MGMT status, but was associated with specific tumor genomic alterations. The enrichment of DNA damage response–related genes among those associated with RN suggests a potential biological link between tumor genomics and RN toxicity. Prospective validation will be required to determine the utility of these findings for genome-based risk stratification.

Introduction: Pituitary carcinoma (PC) is a rare, aggressive endocrine neoplasm characterized by metastasis and challenging clinical management. The transformation from pituitary adenoma (PA) to PC is poorly understood, and predictors of metastasis remain elusive. This study evaluates the clinical course, surgical outcomes, and molecular characteristics of PC. Methods: We retrospectively reviewed patients with PC treated at the M. D. Anderson Cancer Center between 1993 and 2023. Primary outcomes included metastasis-free survival and overall survival (OS). Clinical features, radiographic findings, surgical strategies and outcomes, immunohistochemical profiles, and MIB-1 were analyzed. Results: The cohort (n = 20) had a median age at PA and PC diagnosis of 33.9 and 43.3 years, respectively. The median metastasis-free interval was 7.4 years. GH- and ACTH-secreting tumors showed shorter times to PC diagnosis, while nonfunctioning PAs had longer metastasis-free survival. PAs with MIB-1 > 10% had shorter survival. Dura was the most common site of metastasis within the CNS, and bone was the most common outside the CNS. Leptomeningeal disease was seen in six patients. PAs became aggressive > five years after initial surgical resection (n = 13) or metastasized early within the first five years (n = 7). Median OS from PA diagnosis was 13.7 years, and 8.6 years from PC diagnosis. A total of 102 neurosurgical procedures were performed, with a median of five per patient; the median was similar in patients surviving longer than five years vs. those whose survival was shorter (5.0 vs. 4.5 procedures, p = 0.661). Most surgical interventions post-PC diagnosis were for optic decompression or metastasectomy. All long-term survivors (at least five years after PC diagnosis) received temozolomide-based therapy, with most also receiving radiotherapy. Conclusions: PC shows a variable clinical course, with some PAs progressing to PC after years, while others transform rapidly. All long-term survivors received temozolomide-based therapy, most in combination with radiotherapy and repeated surgical intervention, suggesting that aggressive multimodal management may be associated with prolonged survival. Future research will focus on identifying reliable predictors of metastasis at different time points in the complex clinical evolution of these tumors.

Capicua (CIC) mutations occur in most 1p/19q-codeleted oligodendrogliomas and lead to loss of transcriptional repressor function, resulting in upregulation of proliferative genes and impaired cellular differentiation. CIC mutations are associated with more aggressive tumor behavior and poor outcomes. Noninvasive predictors of CIC mutations are lacking. We evaluated the potential of MRI-based radiomics combined with machine learning (ML) for the noninvasive prediction of CIC mutation and histologic grade in oligodendroglioma. We conducted a retrospective study of oligodendroglioma with CIC mutation testing. Of 65 identified cases, 15 were excluded due to limited imaging, yielding 50 patients with preoperative MRI for analysis. Regions of interest (ROIs) including tumor, edema, necrosis, and hemorrhage were segmented using 3D Slicer. First-order features describing intensity distributions, and second-order features capturing pixel spatial relationships across four orientations derived from the Grey-Level Co-occurrence Matrix (GLCM) were extracted. To minimize noise, each ROI was quantized at multiple gray levels, and GLCM features were extracted from each quantized ROI. All features were normalized using patient-specific white matter and ROI volume. A total of 400 features were computed per ROI. After removing highly correlated features (Pearson’s correlation coefficient > 0.85), feature selection was performed using ElasticNet and Ridge regularization methods. Model performance and feature significance were evaluated via area under curve (AUC). For grade status prediction, the best predictive model was obtained using seven texture features from Ridge method with an accuracy of 74%, sensitivity of 70%, specificity of 78%, and AUC of 69%. For CIC mutation status (wildtype vs mutant), the best model was obtained using five texture features selected via ElasticNet method with an accuracy of 84%, sensitivity of 83%, specificity of 86%, and AUC of 83%. MRI-based radiomics combined with ML offers a promising noninvasive approach for predicting CIC mutation status and histologic grade in oligodendroglioma.

TPS2101 Background: Patients with recurrent glioblastoma have limited effective treatment options due to the highly immunosuppressive microenvironment and rapid proliferation fueled by neoangiogenesis. Anti-angiogenic therapy, including targeting vascular endothelial growth factor (VEGF) with bevacizumab, and immune checkpoint inhibition with programmed cell death protein 1 (PD-1) inhibitors, have independently had limited efficacy in these tumors. Ivonescimab is a humanized tetravalent bispecific antibody against PD-1 and VEGF, which has demonstrated cooperative binding in vitro leading to increased binding of PD-1 in the presence of VEGF and vice-versa 1 . Ivonescimab has shown activity in multiple phase 3 trials conducted in China in non-small cell lung cancer, including one trial which demonstrated activity in patients with brain metastases, but has not yet been evaluated in patients with primary brain tumors. This trial evaluates ivonescimab in patients with recurrent glioblastoma. Methods: This investigator-initiated study consists of a phase I and II component; the primary objectives are safety and tolerability for phase I and determining progression-free survival for phase II. The phase I component evaluates 3 dose levels of ivonescimab (7.5, 10, and 20 mg/kg every 3 weeks), employing a Bayesian optimal interval (BOIN) design for assessing toxicity. Once the recommended phase II dose is determined, the phase II portion will follow a Bayesian optimal phase II (BOP2) design, with interim analyses at pre-specified enrollment points allowing for monitoring of efficacy as well as ongoing evaluation of toxicity. The maximum accumulative sample size at the target dose will be 30 patients. Radiographic assessment will utilize the Response Assessment in Neuro-Oncology 2.0 criteria. Key eligibility criteria include adults with recurrent glioblastoma, IDH-wildtype (by WHO CNS 2021 classification) at first or second recurrence with Karnofsky Performance Scale ≥60 and normal blood counts and organ function. Prior therapy with anti-angiogenic agents (including bevacizumab) or check-point inhibitors is excluded, as well as concurrent corticosteroids ≥ 2 mg/day dexamethasone or equivalent. Samples of archival tumor, blood and stool microbiome will be collected for correlative studies as an exploratory evaluation of predictive biomarkers of response or resistance to ivonescimab. The study has been approved by the institutional review board and accrual to phase I will commence in the first quarter of 2025. 1. Zhong T, Huang Z, Pang X, et al. 1194 Mechanism of action of ivonescimab (AK112/SMT112): a first-in-class tetravalent Fc-silent bispecific antibody with dual blockade of PD-1 and VEGF that promotes cooperative biological effects. Journal for ImmunoTherapy of Cancer 2023;11:doi: 10.1136/jitc-2023-SITC2023.1194. Clinical trial information: NCT06672575 .

PRMT5 inhibitors has shown promising clinical activity in MTAP deficient tumors. Ongoing studies focus on tumors with homozygous MTAP deletions; but that doesn’t span the whole spectrum of MTAP dysfunctional tumors. Herein, we tried to characterize the pan-cancer prevalence of MTAP oncogenic alterations. We used the AACR GENIE database v.16.1 to extract data on the prevalence of MTAP alterations across different tumor types. OncoKB was used to obtain data on functional significance. We reviewed data from 214, 475 samples obtained from 184, 980 patients. Structural variants, mutations, and copy number alterations in MTAP were profiled in 55, 021, 58, 144, and 57, 382 samples across different cancers; respetively. A total of 46 MTAP fusions were identified; all were likely oncogenic per OncoKB including 12 with level 4 alterations. The most common fusions were MTAP intragenic, MTAP::CDKN2A, and MTAP::CDKN2B fusions. The highest rates of MTAP structural variants were observed in CNS cancers (0.4%; n=1), gastrointestinal neuroendocrine tumor (0.3%; n=1), hepatobiliary cancer (0.2%; n=3), melanoma (0.2%; n=3), and glioma (0.1%; n=5). A total of 356 MTAP mutations were identified; including 82 that were likely oncogenic per OncoKB. The most common mutations that were annotated as likely oncogenic were MTAP A191Rfs*6, MTAP X231_splice, and MTAP X116_splice. The highest rates of MTAP mutations were observed in endometrial cancer (1.6%; n=36), small bowel cancer (1.6%; n=3), non-melanoma skin cancer, (1.5%; n=6), colorectal cancer (1.2%; n=72), and bladder cancer (1%; n=20). The highest rates of MTAP homozygous deletions were observed in glioma (24.6%; n=899), blastic plasmacytoid dendritic cell neoplasm (23.1%; n=3), nerve sheath tumor (18.3%; n=28), mesothelioma (17.7%; n=79), and bladder cancer (16.9%; n=350) (Table 1). The spectrum of MTAP alterations expands beyond MTAP deletions; and could be explored in clinical trials targeting MTAP loss. Mohamed A. Gouda, Camila B. Xavier, Carlos Torrado, Pooja A. Shah, Kunal Rai, Natalie Vokes, Omar Alhalabi, Mark T. Bedford, Jianjun Gao, Nazanin K. Majd, Jenny Li, Milind Javle, Funda Meric-Bernstam, Jordi Rodon. Pan-cancer prevalence of unusual MTAP alterations: data from the AACR GENIE database [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1121.

Purpose While immunotherapy has transformed treatment in multiple solid tumors, its efficacy in high-grade glioma remains limited. Understanding the molecular and clinical factors that influence glioma’s response to immunotherapy is essential to improving outcomes. Methods We identified patients with recurrent glioblastoma or astrocytoma, IDH-mutant grade 4, who had been treated with checkpoint inhibitor (CPI), virus therapy, or cell therapy and determined the association between their molecular, clinical, and demographic characteristics and survival outcomes. Results We identified 66 patients, 57 glioblastoma and 9 astrocytoma, IDH-mutant grade 4; 38 were treated with CPI, 22 with virus therapy, and 6 with cell therapy. PIK3CA mutation was associated with shorter PFS and OS (p = 0.022, 0.073, respectively) among all patients and a shorter OS among CPI-treated patients (p = 0.004). Tumor tissue without the mutation had less PD-1 expression in CD3+/CD8 + T cells. In CPI-treated patients, IDH1/2 mutation was associated with a shorter OS (p = 0.002), and mutations in RB1 and TERT promoter were associated with a shorter PFS (p-value = 0.00056, 0.022, respectively). Length of CPI therapy of more than 6 months was associated with increased PFS and OS (p = 0.048, 0.062, respectively), while steroid use at baseline was associated with a shorter OS (p = 0.00023). Multifocal disease was associated with shorter PFS and OS durations (p = 0.0017, 0.0013, respectively) among all patients. Conclusions In high-grade glioma, PIK3CA, IDH1/2, RB1, and TERT promoter mutations may be associated with a poor response to immunotherapy. Our results may provide the rationale for clinical trials combining PI3K and IDH inhibitors with CPI in high-grade glioma.

Introduction: Autologous stem cell transplant (ASCT) is widely accepted as the standard consolidation therapy for primary CNS lymphoma (PCNSL) in first remission, with whole brain radiation therapy (WBRT) used in selected cases. However, older and frail patients with PCNSL often cannot tolerate ASCT or WBRT due to increased risk of treatment-related toxicities, such as infections and neurocognitive impairment. Without consolidation, survival outcomes are poor, with a median progression free survival (PFS) ranging from 10 to 20 months. Therefore, novel therapeutic approaches are needed for older PCNSL patients. Nivolumab, an anti-PD1 antibody, has demonstrated promising activity in relapsed or refractory PCNSL, and we hypothesized that adding nivolumab in the frontline setting would be safe and improve survival in older PCNSL patients who are ineligible for ASCT or WBRT. Methods: In this multicenter phase 1/1B study (NCT 04022980), patients aged ≥ 65 with previously untreated PCNSL received up to 6 cycles of nivolumab consolidation after receiving at least 3 cycles of high-dose methotrexate (HD-MTX)-containing induction chemotherapy. Patients with systemic lymphoma or known ocular involvement were excluded. Stage 1 aimed to evaluate the safety of nivolumab consolidation following HD-MTX-containing induction chemotherapy. A 3+3 design was used for the safety run-in, enrolling 6 patients on nivolumab at the FDA-approved single-agent dose and monitoring for dose-limiting toxicities (DLTs) during Cycle 1. Stage 2 assessed the efficacy of nivolumab consolidation based on 2-year PFS. Neurological function was evaluated at baseline and throughout the study follow-up using the Neurologic Assessment in Neuro-Oncology (NANO) scale. Kaplan-Meier methods were used to estimate overall survival (OS) and PFS, with both time-to-event endpoints calculated from the initiation of nivolumab consolidation therapy. Results: Fourteen patients were enrolled across four U.S. sites between April 2020 and December 2022. The median age was 72 years (range 65-79), 71% were female, and 36% had an ECOG score of 2-3 at enrollment. Induction chemotherapy regimens included 36% R-MPV (rituximab, methotrexate, procarbazine, and vincristine), 21% MRT (methotrexate, rituximab, and temozolomide), 21% MR (methotrexate and rituximab), and 7% MATRix (methotrexate, cytarabine, thiotepa, and rituximab). Following induction chemotherapy, 64% achieved a complete response (CR), 29% achieved a partial response (PR), and 7% had a stable disease (SD). Patients received a median of 6 nivolumab cycles (range 1 – 6). No DLTs occurred during Cycle 1 of nivolumab in Stage 1. The most common grade ≥ 3 adverse events (AEs) were neutropenia (14%) and fatigue (7%). One patient developed grade 4 Stevens-Johnson syndrome, leading to treatment discontinuation after Cycle 1. Among those receiving ≥ 1 cycle of nivolumab consolidation, the overall response rate was 86%, with a CR rate of 79%. Three patients improved from PR or SD post-induction to CR after nivolumab consolidation. At a median follow-up duration of 33 months, median OS and PFS were not reached. The 2-year OS and PFS rates were 79% and 63%, respectively. All deaths (3/3) were due to disease progression. Two patients were found to have active lymphoma during or immediately after Cycle 1 and were subsequently discontinued from the trial. In one case, intraocular involvement was diagnosed during Cycle 1, likely undetected at screening due to its occult presentation. Eleven patients completed all planned cycles of nivolumab consolidation, and 73% remain in remission at the last follow-up. Conclusion: This study demonstrates favorable safety and clinical outcomes with nivolumab consolidation in older PCNSL patients unsuitable for ASCT or WBRT. No DLTs were observed during the safety run-in phase, and there were no unexpected toxicities associated with nivolumab, although one early discontinuation occurred due to Stevens-Johnson syndrome, a rare but known AE of nivolumab. Overall, the absence of DLTs and encouraging survival outcomes support further investigation. Ongoing analyses aim to identify predictive biomarkers, including 9p24.1 and 6p21.3 amplifications, that may correlate with response to nivolumab in PCNSL. Additionally, NANO assessments are being analyzed to measure neurological changes in this population.

10017 Background: Dordaviprone (ONC201), a first in class imipridone, has demonstrated safety and efficacy in an integrated analysis of patients with recurrent H3 K27M-mutant diffuse midline glioma across clinical studies. Here, we report efficacy and safety findings from two prospectively defined clinical trial arms that evaluated single-agent dordaviprone response in recurrent H3 K27M-mutant glioma. Methods: Phase 2 trialONC013 (Arm B) and Phase 1 trial ONC014 (Arm F) were designed to evaluate the objective response rate (ORR) by RANO-HGG criteria of dordaviprone in adult and pediatric patients, respectively, with recurrent H3 K27M-mutant diffuse glioma. Open label dordaviprone was administered once weekly at 625 mg for adults and at a dose scaled by body weight for pediatrics. Responses were investigator-assessed by RANO criteria. Eligibility required measurable enhancing recurrence by RANO-HGG criteria, radiotherapy completed ≥90 days prior to dordaviprone unless unequivocal progression qualified per RANO, Karnofsky or Lansky performance status >60. DIPG, spinal tumors, leptomeningeal disease, and CSF dissemination were excluded. Results: ONC013 Arm B enrolled 30 patients (median age 32, range, 21-66 years) with the majority having a primary midline non-brainstem tumor (n = 19, 63.3%) and one prior recurrence (n = 22, 73.3%). The ORR was 16.7% (95% CI, 5.6-34.7) with 5 partial responses (PR). The median duration of response (DOR) and time to response (TTR) were 15.1 months (7.5-not reached) and 3.8 months (1.8-4.6), respectively. Three patients experienced a grade ≥3 treatment-related adverse events (TR-AE), none had treatment-related serious AEs (TR-SAEs), and 1 had TR-AE leading to dose reduction (ALT increase). ONC014 Arm F enrolled 11 patients (median age 14, range 11-19 years). Most had a primary midline non-brainstem tumor (n = 7, 63.6%) and 1 prior recurrence (n = 6, 65.6%). Two (18.2%) radiographic responses were reported, 1 response (9.1%) qualified by RANO criteria. One PR occurred with > 95% tumor regression and an 8.5-month DOR (1.9-month TTR). Another patient experienced a > 50% tumor regression (4.3-month TTR) that did not meet RANO PR criteria due to initiation of 2.5 mg dexamethasone post-baseline. 12-month PFS rate was not reached; 12-month OS rate was 27.3% (6.5, 53.9). One patient experienced a grade ≥3 TR-TEAE (9.1%); no TR-SAEs, treatment-related deaths, or TR-AE leading to treatment discontinuation occurred. Conclusions: In prospective clinical trials designed to evaluate ORR, single-agent dordaviprone response and safety in adult and pediatric recurrent H3 K27M-mutant diffuse glioma were similar to previously pooled analyses. Clinical trial information: NCT03295396 and NCT03416530 .

The prognosis of patients with glioblastoma (GBM) remains poor despite current treatments. Targeted therapy in GBM has been the subject of intense investigation but has not been successful in clinical trials. The reasons for the failure of targeted therapy in GBM are multifold and include a lack of patient selection in trials, the failure to identify driver mutations, and poor blood-brain barrier penetration of investigational drugs. Here, we describe a case of a durable complete response in a newly diagnosed patient with GBM with leptomeningeal dissemination and PTPRZ1-MET fusion who was treated with tepotinib, a brain-penetrant MET inhibitor. This case of successful targeted therapy in a patient with GBM demonstrates that early molecular testing, identification of driver molecular alterations, and treatment with brain-penetrant small molecule inhibitors have the potential to change the outcome in select patients with GBM.

Leptomeningeal metastases (LM) are an aggressive and rare complication of renal cell carcinoma (RCC), with few cases described in the literature. Data regarding this presentation is currently limited. We describe our single center experience with patients with RCC and cerebrospinal fluid cytology confirmed LM from January 2013 to July 2024. Five cases were identified, with a median age of 51 years. Most patients had clear cell histology (4/5). LM presented with neurological symptoms in most (4/5), and included headache, cranial nerve IV and VII palsies, unilateral leg weakness, paraparesis, and urinary/bowel dysfunction. MRI revealed findings concerning for LM in both brain and spine in most patients (4/5). KPS was above 90 in 3 patients, and 50 and 60 in the other patients. Systemic disease was stable in most patients (4/5), and most were on active systemic treatment (4/5). Three patients had concomitant parenchymal metastases, in which 2 were present prior to the diagnosis of LM. Median time from RCC to LM diagnosis was 40.1 months (CI 23.3-not reached months). One patient was treated with bevacizumab followed by whole brain radiation, one with pazopanib, and one with focal radiation. One patient remains on treatment (intrathecal and intravenous nivolumab plus cabozantinib). Three patients were transitioned to hospice shortly after the diagnosis. Median survival after LM diagnosis was 3.4 months (CI 1.4- not reached months). LM from RCC is exceedingly rare. Survival following a diagnosis of LM is typically poor even with stable systemic disease, although one patient in our series continues treatment 5 months following the diagnosis. Our data provides the experience of a large cancer center in the US, and highlights some of the challenges of this rare diagnosis.

DNA-dependent protein kinase (DNA-PK) facilitates DNA repair and reduces the effectiveness of DNA-damaging treatments such as radiation therapy (RT), making it an attractive target in glioblastoma. Peposertib, a selective DNA-PK inhibitor, has demonstrated preclinical efficacy in sensitizing GBM to RT, resulting in tumor regression. 21 adult patients with newly diagnosed GBM received peposertib concurrently with 60Gy RT over 6 weeks, followed by at least 6 cycles adjuvant TMZ (NCT04555577). RT-induced changes (i.e. pseudoprogression and radiation-necrosis) were assessed through conventional MRI and confirmed with spectroscopy and perfusion or pathology. Fractional Tumor Burden (FTB) were generated from Dynamic Susceptibility Contrast-(DSC)-MRI using relative cerebral blood volume thresholds to differentiate viable tumor from treatment-effect. γ-H2AX immunohistochemistry was performed on all available samples. The maximum-tolerated dose of peposertib with RT was 300mg. After a median follow-up of 16.9 months (13.9 – 23.2), mPFS was 10.7 months (95%CI: 8.9-18.2) and mOS was 19.5 months (95%CI: 14.3-30.4). Among 18 evaluable patients, 6 (33%) had distant and 12 (66%) had local recurrence. RT-induced changes occurred in 9 of 19 patients (47%) and were associated with longer PFS (log-rank p = 0.031). In 9 patients with available DSC-MRI, FTB map analysis showed a trend toward increased treatment-effect volume at 4 weeks post-RT compared to immediate post-RT (median +9.4%, p = 0.195). γ-H2AX staining in 5 recurrent samples compared to 20 archival samples showed a trend toward higher post-treatment positivity (+14.4%, p = 0.112). Baseline γ-H2AX expression was not associated with survival outcomes. Updated outcomes remain encouraging and support further development of peposertib. High rate of distant recurrence and positive association between RT-induced changes and PFS suggest that peposertib is active as a radiosensitizer. Further genomic, transcriptomic, and spatial proteomic analyses, are underway and will be presented at the conference. The study was supported by EMD Serono (CrossRef Funder ID: 10.13039/100004755).

Abstract Overexpression of MAPK (Ras-Raf-MEK-ERK) signaling is well known in NF1-associated tumors. MEK inhibition has shown promising results in pediatrics NF1-associated tumors. In addition, we have observed augmented MAPK signaling and anti-tumor activity with MAPK inhibition in preclinical models of CIC-mutated vs. wild-type oligodendroglioma. Ulixertinib is a small molecule inhibitor of ERK that is being developed as a novel anti-cancer drug. A phase I study of ulixertinib in advanced solid malignancies had shown two central nervous system responses. We hypothesized that ulixertinib crosses the blood-brain barrier (BBB) and it will result in improved responses in MAPK-activated lower-grade gliomas in adults (recurrent NF1-mutated gliomas grade 1,2,3 and oligodendrogliomas grade 2,3 which are enriched in CIC-mutation). This is a window-of-opportunity study of ulixertinib in patients ≥18 yo. Twenty patients who are candidates for non-emergent surgical resection will be enrolled (10 in each cohort). Patients will receive neoadjuvant ulixertinib at 600mg BID. Selected patients will undergo CSF collection 1-week after and all will undergo surgical intervention 2-weeks after initiation of ulixertinib. Patients will then continue treatment until disease progression or unacceptable toxicity. Primary endpoints are ulixertinib tumor concentration and tumor/plasma ratio in enhancing and non-enhancing disease. Secondary endpoints include median PFS, overall response rate (ORR) and disease control rate at 12mo, duration of response, time-to-next intervention and time-to-response, safety profile, and ulixertinib CSF and tumor/CSF ratio. Exploratory endpoint includes ORR of plexiform neurofibromas in NF1 patients. To date, 4 patients have been dosed; 3 CIC-mutated oligodendroglioma and 1 NF1-mutated glioma. A fifth patient with NF1-mutated glioma has been consented. Upon collection of tumor tissue on all 5 patients, initial assessment of tumor and CSF drug concentration will be conducted and presented at the conference. Ulixertinib’s ability to cross the BBB could influence what tumor types may be explored for further patient benefit.

Background:Leptomeningeal metastases (LM) in gastroesophageal (GE) malignancies are exceedingly rare. Historically, treatment for LM has included steroids, radiation, chemotherapy, and intrathecal (IT) chemotherapy. However, the outcomes in GE malignancies with LM remain poor. Unfortunately, clinical trials in GE malignancies have traditionally excluded those with LM, limiting advances in therapeutic strategies. Given that LM poses potentially devastating neurologic and psychologic sequelae, there is an urgent need for more effective treatments. Case Description:Patient 1 is a 44-year-old woman with localized esophageal adenocarcinoma who undergoes neoadjuvant chemoradiation followed by esophagectomy. Seven months following surgery, she develops ataxia, weakness, and nausea/vomiting. Magnetic resonance imaging (MRI) reveals intracranial disease that is subsequently successfully resected and then treated with gamma knife (GK) radiation. Pathology confirms metastases. Three months later she is found to have LM. She receives palliative whole brain radiation therapy as well as focal radiation to the spine. Following this she transitioned to concurrent IT topotecan plus intravenous (IV) ipilumumab/nivolumab with durable response beyond 14 months. Patient 2 is a 71-year-old man with de novo metastatic esophageal adenocarcinoma with durable response to 5-fluorouracil plus irinotecan. Asymptomatic intracranial metastases are detected on surveillance scans 2 years after initial diagnosis for which he receives GK. Follow up MRI identifies new LM. As such, to treat the LM, he was transitioned to IT topotecan and IV pembrolizumab with good response for 6 months until death from a gastrointestinal bleed. Conclusions:We present two cases of LM in patients with GE adenocarcinoma who had longer survival than what has been reported. They were treated with combination IT topotecan and IV checkpoint inhibition. Further studies evaluating the central nervous system tumor immune-microenvironment can help expand our understanding of how this combination has worked well in our patients and how to care for others with similar scenarios.

Background:We observed rapid tumor progression following COVID-19 infection among patients with glioblastoma and sought to systematically characterize their disease course in a retrospective case-control study. Methods:Using an institutional database, we retrospectively identified a series of COVID-19-positive glioblastoma cases and matched them by age and sex 1:2 to glioblastoma controls who had a negative COVID-19 test during their disease course. Demographic and clinical data were analyzed. Hyperprogression was defined using modified response evaluation criteria in solid tumors criteria. Time to progression and overall survival were estimated using the Kaplan-Meier method. Results:Thirty-two glioblastoma cases with positive COVID-19 testing were matched to 64 glioblastoma controls with negative testing; age, sex, and molecular profiles did not differ between groups. Progression events occurred in 27 cases (84%) and 46 controls (72%). Of these, 14 cases (52%) presented with multifocal disease or leptomeningeal disease at progression compared with 10 controls (22%; P = .0082). Hyperprogression was identified in 13 cases (48%) but only 4 controls (9%; P = .0001). Cases had disease progression at a median of 35 days following COVID-19 testing, compared with 164 days for controls (P = .0001). Median survival from COVID-19 testing until death was 8.3 months for cases but 17 months for controls (P = .0016). Median overall survival from glioblastoma diagnosis was 20.7 months for cases and 24.6 months for controls (P = .672). Conclusions:Patients with glioblastoma may have accelerated disease progression in the first 2 months after COVID-19 infection. Infected patients should be monitored vigilantly. Future investigations should explore tumor-immune microenvironment changes linking tumor progression and COVID-19.

Purpose: Treatment options for leptomeningeal metastasis (LM) are limited. A recent phase 2 study found that proton craniospinal irradiation (pCSI) was well-tolerated and improved survival. We report our experience with pCSI for solid-tumor LM. Methods and Materials: This is a retrospective review of patients treated with pCSI for solid-tumor LM from December 2020 to January 2024 at our center. Patient characteristics were summarized using descriptive statistics. Median overall survival and median central nervous system progression-free survival from the first day of pCSI were estimated using Kaplan-Meier survival curves. Results: We identified 45 patients who completed pCSI. The median age was 54 years (range, 23-79); 73% were female, and 53% lived more than 100 miles from our center. Breast cancer (53%), lung cancer (20%), and melanoma (9%) were the most common primary cancers; 51% of patients had stable systemic disease at LM diagnosis. All had imaging evidence of LM, and 64% of cases were confirmed using cytologic examination of the cerebrospinal fluid. Eighty percent had symptomatic LM, and the median Karnofsky performance scale at LM diagnosis was 80. The median time from primary cancer diagnosis to LM detection was 23.1 months (range, 0-221.3). Fifty-three percent of patients had active brain metastasis at LM diagnosis; 33% of all patients had received prior intracranial radiation. The median time from simulation to pCSI start was 12 days. At the first visit following pCSI, the median Karnofsky performance scale score was 70. During or right after radiation, 76% of patients reported nausea, 51% headache, and 31% fatigue. Following pCSI, 4% received intrathecal chemotherapy, 67% systemic therapy, and 9% hospice care; 18% were observed and 2% lost to follow-up. Median overall survival was 13.7 months (95% confidence interval [CI], 11.2 to not reached), and median progression-free survival was 6.5 months (95% CI, 4.9-12.8). Conclusions: The outcomes in our cohort are comparable to those recently reported in a phase 2 trial. Further study is indicated to determine the optimal candidates for pCSI and sequential therapies.

Abstract Current standard-of-care for glioblastoma (GBM) includes surgery followed by radiation with concurrent and adjuvant temozolomide. There are limited treatment options available upon progression or recurrence of disease. Options often involve participation in clinical trials with promising new therapies. However, patients may not meet the strictly defined entry criteria to participate in these clinical trials. Under an Expanded Access (EA) program, we have treated 14 recurrent GBM patients with VAL-083, they were not eligible to participate in other clinical trials. The median time from last progression to start of VAL-083 was 1.0 month (95%CI:0.09-2.67). All patients had received chemoradiation with temozolomide. Five (5/14;36%) patients had ≥ 2 recurrences, 8/14 (57%) had multifocal disease, and the median KPS was 80 (95%CI:70-90). All patients had unmethylated MGMT promoter, 13/14 (93%) were IDH-WT, 9/14 (64%) had a TERT promoter mutation, 6/14 (43%) had a PTEN mutation, and 4/14 (28%) had an EGFR mutation. All patients received treatment with VAL-083 at 30 mg/m2 x 3 days every 21 days. Five patients with cerebral edema refractory to steroids received bevacizumab (BEV;10 mg/kg) concurrently with VAL-083, and 8 patients received dexamethasone (Dex; >4 mg/day) with VAL-083. This is the first report of the use of VAL-083 in combination with BEV (VAL-03/BEV).The main adverse event was thrombocytopenia consistent with prior experience. Four (4/14; 28%) patients had a dose reduction, 3 of which were due to thrombocytopenia. No patients had a dose reduction while receiving VAL-083/BEV. Two patients who progressed on VAL-083, were later treated with CCNU, and myelosuppression was not observed.Median progression-free survival (mPFS) and median overall survival (mOS) from last progression were 5.7 months (95%CI:1.3-7.9) and 8.3 months (95%CI:3.0-14.3), respectively. Additional treatment, safety and outcome data will be presented at the meeting.Clinicaltrials.gov Identifier: NCT03138629. EA treatment plans were approved by MD Anderson Cancer Center IRB.