Androgen loss weakens anti-tumor immunity and accelerates brain tumor growth

Many cancers, including glioblastoma (GBM), have a male-biased sex difference in incidence and outcome. The underlying reasons for this sex bias are unclear but likely involve differences in tumor cell state and immune response. This effect is further amplified by sex hormones, including androgens, which have been shown to inhibit anti-tumor T cell immunity. Here, we show that androgens drive anti-tumor immunity in brain tumors, in contrast to its effect in other tumor types. Upon castration, tumor growth was accelerated with attenuated T cell function in GBM and brain tumor models, but the opposite was observed when tumors were located outside the brain. Activity of the hypothalamus-pituitary-adrenal gland (HPA) axis was increased in castrated mice, particularly in those with brain tumors. Blockade of glucocorticoid receptors reversed the accelerated tumor growth in castrated mice, indicating that the effect of castration was mediated by elevated glucocorticoid signaling. Furthermore, this mechanism was not GBM specific, but brain specific, as hyperactivation of the HPA axis was observed with intracranial implantation of non-GBM tumors in the brain. Together, our findings establish that brain tumors drive distinct endocrine-mediated mechanisms in the androgen-deprived setting and highlight the importance of organ-specific effects on anti-tumor immunity.

Abstract 4740: Molecular effects of histone deacetylase inhibitor Quisinostat on diffuse midline glioma of the pons

Diffuse Midline Glioma of the pons (DMG) is a lethal, aggressive heterogeneous brain stem tumor. Median overall survival is less than a year, with radiation as the only standard treatment. Recently, mutations in DMGs have arisen as potential therapeutic targets, specifically a mutation in one of the histone H3 genes, resulting in methionine substituted for lysine at site 27 (H3K27M). H3K27M induces a marked reduction in global acetylation of histone tails, altering chromatin structure and causing aberrant gene expression. Histone deacetylase inhibitors, HDACis, are epigenetic drugs that show anticancer activity. In our studies, Quisinostat (Quis) is used to preserve histone acetylation. Using two H3K27M-DMG treatment-naive preclinical models (PBT22 and PBT29) we detected a 100-fold differential response to the histone deacetylase inhibitor, Quisinostat. PBT-22 harbors mutations in H3F3A, TP53, and ASXL2, while PBT-29 has mutations in H3F3A, TP53, PIK3CA and FGFR1. Following Quis treatment (48 hrs) in both preclinical models, total H3K27ac protein abundance increased 3-fold, suggesting HDACi stabilizes or impedes turnover of K27 acetylated H3 histone. We are pursuing studies to test whether sensitivity to HDACi in DMGs is determined by a shift in relative or total abundance of respective H3wt- and H3K27M- histones. We posit that changes in H3K27ac manifest as shifts in nucleosome integration with genes responsible for cell survival/death. This project will profile differentially expressed genes (DEGs) from Quis-treated PBT22 and PBT29. Additionally, total and relative abundance of H3 proteins (wt and mt, me, me2, me3, and ac) from treated cells will be determined. Gene ontology analysis will focus on pathways accounting for chromatin remodeling, cell death, and growth arrest. RNA (qRT-PCR) and proteins (western blot) from analytes from a larger panel of DMG cell lines and neural stem cells treated with Quis will be used to validate the findings. Overall, the data depicts DMG preclinical models with large differential sensitivity to Quis, which may be partially due to different oncoprints between the models. The markedly different sensitivity of these models enables mechanistic study of the consequences of elevated abundance of histone 3 acetylation. The long term goal is to discover a molecular profile of DMGs indicative of the vulnerability to HDACi.

Citation Format: Danyelle Paine, Nanyun Tang, Yue Hao, Matt Biery, Carrie Meyers, Alyssa Noll, Nicholas Vitanza, Michael Berens. Molecular effects of histone deacetylase inhibitor Quisinostat on diffuse midline glioma of the pons. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4740.

Abstract 1601: Potentiation of arsenic trioxide in glioblastoma

Glioblastoma (GBM) is the most prevalent type of malignant tumor within the central nervous system. Possessing a five-year survival rate of 6.8%, new therapeutic options are a necessity in order to increase patient quality of life and survival outcomes. Arsenic trioxide (ATO) is an FDA approved drug for the treatment of relapsed or refractory acute promyelocytic leukemia. Clinical trials of ATO in combination with radiation and temozolomide show therapeutic effects in a modest subset of glioblastoma patients. ATO resistance in GBM cell lines is associated with MNK1-eIF4E upregulation. Additionally, diminished response to ATO is linked to resistance to oxidative stress via increased glutathione levels and NrF2 pathway expression. We sought to target each of these proposed mechanisms of resistance in order to develop a suite of compounds for use in combination therapies, with the goal of sensitizing GBM models to ATO. Six GBM PDX models were dosed with ATO at varying concentrations and monitored for the IC50 value of treatment. The cell lines exhibited a 20-fold difference in sensitivity, indicative of an underlying innate resistance to ATO. We performed drug dose response studies of patient-derived glioma cells treated with ATO in combination with the nutraceutical compounds Chrysin and Silibinin. Combination treatments using Chrysin and Silibinin showed potentiation of the cytotoxic effects of ATO. Confirmation of the mechanism for induced sensitivity to ATO will be explored by measuring shifts in glutathione content and susceptibility to oxidative stress in response to treatment using nutraceutical compounds. In addition, drug dosing with combination therapies utilizing the MAPK interacting serine/threonine kinase 1 (MNK1) inhibitors eFT-508 and EFT-206 also show synergistic effects in GBM cell lines. The eIF4E phosphorylation response to treatment of both ATO alone ATO MNK1i combination therapies will be monitored and analyzed for potential translational effects. Each of these compounds were selected based on safe use determined in clinical trials (MNK1 inhibitors) or over-the-counter availability (nutraceuticals). All four companion approaches show low pharmacological liabilities in crossing the blood brain barrier. Future studies involving these compounds will consist of testing these suggested combination therapies in vivo by observing their effectiveness in orthotopic, GBM tumor bearing mice. Identification and development of sensitization targets for ATO will allow for greater effectiveness of treatment and a greater potency of the drug in combating resistance which arises in response to treatment. Discovery of a predictive molecular signature of synergy from ATO and other targeted agents may pave the way for a successful clinical trial of these combinations in an identifiable subpopulation of GBM patients.

Citation Format: Charles Shaffer, Nanyun Tang, Yue Hao, Karen Fink, George Snipes, Bruce Mickey, Michael Berens. Potentiation of arsenic trioxide in glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1601.

Abstract 6779: Spatial and single-nucleus transcriptomics of human glioblastoma

Glioblastoma (GBM) multiforme is the most aggressive type of brain cancer. These lethal brain tumors are characterized by the inter- and intra-tumor molecular heterogeneity and diverse cell plasticity. We sought to interrogate the immune landscape of GBM, tease out the distinct biological and therapeutic niches in GBM and to understand how the tumor microenvironments mediated the heterogeneity of GBM. We have collected six human GBM samples for spatial profiling and single-nucleus RNA sequencing. Multiple sets of spatial gene panels were designed to capture the cell identities (cancer stem cells, astrocytes, oligodendrocytes, microglia, macrophages, endothelial cells, etc.) and the spatially-resolved expression patterns of genes indicative of different functional states. We also aim to find altered neurovascular units by examining the spatial correlation between cells that highly express angiogenesis markers such as VEGFA with tumor-associated macrophages, which could be recruited to the angiogenic niche by GBM tumor cells. From the spatial RNA dataset of the first two samples, we observed close proximity of tumor-associated macrophages and endothelial cells that form the microvasculature. Our preliminary results from the single-nucleus RNA sequencing and spatial transcriptomics of two samples shed light on the complex cell landscape and the dynamic interaction between tumor cells and infiltrating or resident immune cells in the GBM tumor tissue.

Citation Format: Yue Hao, Angela Baker, Brian Hilbush, Marcos Otero, Chris Streck, Karen L. Fink, George J. Snipes, Bruce E. Mickey, Michael Berens. Spatial and single-nucleus transcriptomics of human glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6779.

Interrogating the Human Diplome: Computational Methods, Emerging Applications, and Challenges

Human DNA sequencing protocols have revolutionized human biology, biomedical science, and clinical practice, but still have very important limitations. One limitation is that most protocols do not separate or assemble (i.e., "phase") the nucleotide content of each of the maternally and paternally derived chromosomal homologs making up the 22 autosomal pairs and the chromosomal pair making up the pseudo-autosomal region of the sex chromosomes. This has led to a dearth of studies and a consequent underappreciation of many phenomena of fundamental importance to basic and clinical genomic science. We discuss a few protocols for obtaining phase information as well as their limitations, including those that could be used in tumor phasing settings. We then describe a number of biological and clinical phenomena that require phase information. These include phenomena that require precise knowledge of the nucleotide sequence in a chromosomal segment from germline or somatic cells, such as DNA binding events, and insight into unique cis vs. trans-acting functionally impactful variant combinations-for example, variants implicated in a phenotype governed by compound heterozygosity. In addition, we also comment on the need for reliable and consensus-based diploid-context computational workflows for variant identification as well as the need for laboratory-based functional verification strategies for validating cis vs. trans effects of variant combinations. We also briefly describe available resources, example studies, as well as areas of further research, and ultimately argue that the science behind the study of human diploidy, referred to as "diplomics," which will be enabled by nucleotide-level resolution of phased genomes, is a logical next step in the analysis of human genome biology.

BIOM-45. DETERMINANTS OF GBM VULNERABILITY TO INHIBITORS OF NEDD8 ACTIVATING ENZYME (NAE)

Advancing new therapeutics to FDA approval for GBM is hampered by intra- and inter-tumoral heterogeneity. Design of clinical trials to enroll patients likely to respond to a new treatment seems critical to build the arsenal of agents against this disease. Neddylation is a protein modification pathway aligned to protein degradation and cytological localization within the ubiquitin/proteasome system. Components of the neddylation pathway are frequently overexpressed or hyperactivated in GBM, and whose upregulation has been associated with glioma progression and worse survival. Among glioblastoma cell lines and patient-derived stem cells, subsets were found to be selectively susceptible to NAE inhibitors. We previously showed that PTEN signaling, DNA replication, and chromatin instability pathways are the most significant differentiators between MLN4924 sensitive vs. non-sensitive models. To test a candidate molecular “signature of vulnerability”, two subsets of PDX models (PTEN wildtype and PTEN del or mutated) were treated with a panel of 3 NAE inhibitors. 5 GBM PTENwt models were shown vulnerable to MLN4924 and TAS4464 with average EC50 values 5-10 folds lower than 6 PTENmt/del models. CDC activity was greater against PTENwt, but only 2-fold different vs PTENmt/del. Transcriptomic data mining uncovered additional determinants of response to NAE inhibitors. To disclose the relationship between constitutive expression of three gene sets: DNA Replication Fork Progressing, Replication Fork Protection, and Replication Fork Protection Complex, we found that the genes involved in these are significantly under expressed in PTENmt models when comparing to PTENwt models. The molecular determinants of drug response to NAE inhibitors will be further verified in the extended preclinical models and lead to a patient-enrollment “signature of vulnerability” to increase the likelihood of demonstrating therapeutic efficacy in the early stage of clinical trials.

Upfront Biology-Guided Therapy in Diffuse Intrinsic Pontine Glioma: Therapeutic, Molecular, and Biomarker Outcomes from PNOC003

PURPOSE

PNOC003 is a multicenter precision medicine trial for children and young adults with newly diagnosed diffuse intrinsic pontine glioma (DIPG).

PATIENTS AND METHODS

Patients (3-25 years) were enrolled on the basis of imaging consistent with DIPG. Biopsy tissue was collected for whole-exome and mRNA sequencing. After radiotherapy (RT), patients were assigned up to four FDA-approved drugs based on molecular tumor board recommendations. H3K27M-mutant circulating tumor DNA (ctDNA) was longitudinally measured. Tumor tissue and matched primary cell lines were characterized using whole-genome sequencing and DNA methylation profiling. When applicable, results were verified in an independent cohort from the Children's Brain Tumor Network (CBTN).

RESULTS

Of 38 patients enrolled, 28 patients (median 6 years, 10 females) were reviewed by the molecular tumor board. Of those, 19 followed treatment recommendations. Median overall survival (OS) was 13.1 months [95% confidence interval (CI), 11.2-18.4] with no difference between patients who followed recommendations and those who did not. H3K27M-mutant ctDNA was detected at baseline in 60% of cases tested and associated with response to RT and survival. Eleven cell lines were established, showing 100% fidelity of key somatic driver gene alterations in the primary tumor. In H3K27-altered DIPGs, TP53 mutations were associated with worse OS (TP53mut 11.1 mo; 95% CI, 8.7-14; TP53wt 13.3 mo; 95% CI, 11.8-NA; P = 3.4e-2), genome instability (P = 3.1e-3), and RT resistance (P = 6.4e-4). The CBTN cohort confirmed an association between TP53 mutation status, genome instability, and clinical outcome.

CONCLUSIONS

Upfront treatment-naïve biopsy provides insight into clinically relevant molecular alterations and prognostic biomarkers for H3K27-altered DIPGs.

Abstract 195: A focused ultrasound blood brain barrier disruption model to test the influence of tight junction genes to treat brain tumors

A major hindrance to advances in the care of patients with malignant gliomas is the presence of the blood brain barrier (BBB) and blood-brain tumor barrier (BBTB) that greatly restricts drug access from the plasma to the tumor cells. Bubble-assisted Focused Ultrasound (BAFUS) has proven effective in opening the BBB for treatment of glial tumors in adults and pediatric cases. BAFUS has been previously shown to disrupt noninvasively, selectively, and transiently the BBB in small animals in vivo. However, there is a lack of an in vitro preclinical model suitable for testing the genetic determinants of endothelial cell tight junction integrity and vulnerability to the physical disruption. Our BBB organ-on-chip platform will enable precision medicine of brain cancers through identifying patient-specific parameters by which to open the BBB allowing use of drugs and drug combinations otherwise unsuitable. We intend to sequence these in vitro models to verify that the genotype (alleles/SNPs) of tight junction proteins contribute to BBB structure and integrity. To initiate this effort, we report the development of an ultrasound transparent organ-on-chip model populated by iPSC-derived endothelial cells (iPSC-EC) co-cultured with astrocytes. Western blot, immunocytochemistry, permeability, and transelectrical endothelial resistance (TEER) studies all convey expression of key EC proteins and marked barrier integrity. Further benchmarking of device-ultrasound interactions, successful iPSC differentiation, tight junction formation, and fabrication of nanobubbles and their assistance in ultrasound BBB disruption will be presented. Efforts are underway to analyze nine characteristic BBB tight junction genes from WGS data to determine associations between iPSC-EC genotype and phenotype.

Citation Format: Jayashree Iyer, Adam Akkad, Nanyun Tang, Sen Peng, Michael Berens, Frederic Zenhausern, Jian Gu. A focused ultrasound blood brain barrier disruption model to test the influence of tight junction genes to treat brain tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 195.

DIPG-31. Prognostic and predictive biomarkers of response in children and young adults with H3K27M-altered diffuse intrinsic pontine glioma: results from a multi-center, interventional clinical trial (PNOC003)

BACKGROUND: Diffuse intrinsic pontine glioma (DIPG) is a fatal brain tumor. Herein, we report on novel prognostic and predictive genomic biomarkers identified in PNOC003, a multi-center precision medicine trial for children and young adults diagnosed with DIPG. METHODS: Patients aged 3-25 years were enrolled on PNOC003 based on radiographic diagnosis of DIPG. Pre-treatment tumor biopsies were analyzed using tumor-normal whole-exome sequencing and mRNA-tumor sequencing to determine biology-informed, multi-agent therapy following radiation therapy (RT). Whole-genome sequencing was performed as an exploratory study aim. Genomic biomarkers were investigated to identify predictors of RT response and overall survival (OS) in patients with confirmed H3K27M-altered DIPG. Prognostic biomarkers were verified in a retrospective, H3K27M-altered diffuse midline glioma cohort (n=22) from the Children’s Brain Tumor Network (CBTN). RESULTS: Thirty patients enrolled on PNOC003 met molecular criteria for H3K27M-altered DIPG. TP53 was the most frequently altered driver gene (73%). Somatic alterations in PTEN>TP53>PDGFRA were independently associated with OS (P<0.05, in order of negative impact on survival). TP53 mutations associated with worse OS (TP53mut 11.1 mo [95% CI 8.7, 14]; TP53wt 13.3 mo [95% CI 11.8, NA]; P=3e-2), chromosomal instability (P=3e-3), and resistance to RT (P=6e-4). Moreover, loss of chromosome 10q, encoding tumor suppressor PTEN, was associated with worse OS, co-occurred with PTEN alterations, biallelic PTEN inactivation and loss of gene expression. The combination of TP53 alterations and loss of 10q/PTEN in H3K27M-altered DIPG was associated with the worst OS in a combined PNOC003 and CBTN cohort (TP53mut/10qdel, n=14, OS 8.4 mo [95% CI 7.4, 15.8]; TP53mut/10qwt, n=20, OS 13.1 mo [95% CI 10.1, 17.2]; TP53wt/10qwt, n=14, OS 15.5 mo [11.8, 29.4]; P=2e-3). CONCLUSION: PNOC003, a tissue-driven clinical trial, provided insights into prognostic and predictive genomic biomarkers and informed a novel molecular tumor classification system for H3K27M-altered DIPGs.

TAMI-54. SEXUAL DIMORPHISM IN IRON ACQUISITION IN GLIOBLASTOMA

Sexual dimorphism in incidence and the clinical outcomes of Glioblastoma (GBM) has been reported, however, our knowledge of contributing biological mechanisms is limited. Iron acquisition is key to robust tumor growth. Upregulation of Transferrin (TF, iron transport protein)/Transferrin receptor (TFR) is critical for found in multiple different cancers, specifically, we have identified H-ferritin (FTH1) as a contributor to iron transport and protection in cancer stem cells. To interrogate brain tumor iron uptake mechanisms,we performed binding studies on homogenized samples of human male and female GBM tissue samples using 125I labeled TF and FTH1. Tumors from males had a ̴ 3.8-fold increased binding of both proteins compared to tumors from females. We interrogated iron uptake in a syngeneic orthotopic mouse model (GL261 cells) using male and female mice. After the tumors were established, radioactive 125I labeled TF and FTH1 proteins were injected retro-orbitally in the mice. After 24 hours, tumors wereremoved, and analyzed for TF and FTH1 uptake. Male tumors showed an increased uptake, of ̴ 3.2-fold, as compared to female tumors. There was no significant difference in TF uptake between male and female tumors nor between tumor and matched non-tumor brain tissue. We next queried role of FTH1 in the context of sexual dimorphism in GBM in a FTH1+/- mouse strain developed in our laboratory. Survival was monitored in the mice which were injected with GL261 cells at 3 months. Male mice that had reduced expression of FTH1 had poorer survival as compared to the male wild type controls whereas wild type and FTH+/- females had no major differences in survival outcomes. In summary, this study demonstrates sexual dimorphism in iron acquisition in GBM and animal models further suggesting a pathophysiological role of iron metabolism in GBM development and its possible role in prognosis.

EXTH-17. A FOCUSED ULTRASOUND BLOOD BRAIN BARRIER DISRUPTION MODEL TO TEST THE INFLUENCE OF TIGHT JUNCTION GENES TO TREAT BRAIN TUMORS

Treating primary or metastatic tumors in the brain (glioblastomas, melanoma, lung cancer, breast cancer) proves challenging by virtue of the protective function of the blood brain barrier (BBB). The tight junction proteins (TJPs) binding the specialized endothelial cells of the BBB largely contribute to the limited permeability of cancer-therapeutic drugs. In both preclinical and clinical models, low intensity focused ultrasound (LIFU) coupled with microbubbles has been proven to safely and transiently open the BBB. Despite this method being established, potential genetic influences on the durability and vulnerability of tight junctions to LIFU have not been elucidated, nor have the determinants of tight junction repair post LIFU been thoroughly investigated. We report the development of an ultrasound transparent organ-on-chip model populated by iPSC-derived endothelial cells (iPSC-EC) co-cultured with astrocytes. We aim to probe the contributions of various tight junction genes to barrier integrity along with the subsequent protein topology involved in reassembly post ultrasound. Thus, this model serves to determine parameters for ultrasound disruption for precision opening of the BBB. The BBB-On-Chip was successfully fabricated and assembled with an optimized technique that has an 80% yield of leak-free devices, with stable cavitation post nanobubble injection. Furthermore, Western blots show expression of claudin-5, a key TJP, in our iPSC-ECs. We have also demonstrated by confocal microscopy that another component of the TJP complex, ZO-1, can be visualized at iPSC-derived cell junctions. Further benchmarking of device-ultrasound interactions, successful iPSC differentiation, tight junction formation, and fabrication of nanobubbles and their assistance in ultrasound BBB disruption will be presented. Efforts are underway to characterize the contributions of tight junction genes and their variations to the integrity and disruption of the BBB.

CSIG-14. PTEN AS A SIGNATURE OF VULNERABILITY OF GBM TO NEDDYLATION INHIBITION

Neddylation is a specific pathway within the ubiquitin/proteasome system that is overactive in GBM, and whose upregulation has been associated with glioma progression and worse survival. Pevonedistat (MLN4924) is a first-in-class small-molecule neddylation inhibitor shown to inhibit growth of GBM cells by impacting protein degradation in culture and orthotopic xenografts. However, the determinants of vulnerability are not fully understood. Because the molecular heterogeneity within and across GBM patients obscures therapeutic targets and obfuscates signals of efficacy in clinical trials, we pursue the use of molecular “signatures of vulnerability” to targeted agents in subsets of preclinical models. Selective vulnerability to pevonedistat was shown in a subset of GBM; notably, models with mutations or copy number deletions of PTEN are associated with de novo resistance to pevonedistat. Time-course studies of sensitive and non-sensitive GBM cells using transcriptomics and proteomics/phosphoproteomics uncovered additional determinants of response to pevonedistat. Our results demonstrate that in GBM, resistance to pevonedistat is driven by reduced PTEN-chromatin binding (loss-of-function or lower expression) that is also independent of PTEN’s lipid phosphatase activity (i.e., PI3K/AKT signaling). Across 25 glioma cell lines, we found that PTEN signaling, DNA replication, and chromatin instability pathways are the most significant differentiators between pevonedistat sensitive vs. non-sensitive models. In GBM models with modest to low sensitivity to pevonedistat, TOP2A expression was elevated. Combination treatment with the TOP2A inhibitor, etoposide, proved synergistic with pevonedistat. We report that PTEN status both serves as a novel biomarker for GBM sensitivity to pevonedistat and reveals a synergistic vulnerability to TOP2A inhibitors in combination with pevonedistat. Paired use of GBM PDX models of varying sensitivity with drug development testing allows the advancement of a promising agent as well as a patient-enrollment “signature of vulnerability” likely to increase the likelihood of demonstrating therapeutic efficacy in early stage clinical trials.

EPID-08. IMPORTANCE OF THE INTERSECTION OF AGE AND SEX TO UNDERSTAND VARIATION IN INCIDENCE AND SURVIVAL FOR PRIMARY MALIGNANT GLIOMAS

BACKGROUND

Gliomas are the most common type of malignant brain and other CNS tumors, accounting for 80.8% of malignant primary brain and CNS tumors. They cause significant morbidity and mortality. This study investigates the intersection between age and sex to better understand variation of incidence and survival for glioma in the United States.

METHODS

Incidence data from 2000-2017 were obtained from the Central Crain Tumor Registry of the United States, which obtains data from the CDC’s National Program of Cancer Registries and NCI’s Surveillance Epidemiology and End Results Program (SEER), and survival data from the CDC’s NPCR Registries. Age-adjusted incidence rates and rate ratios per 100,000 were generated to compare male-to-female incidence by age group. Cox proportional hazard models were performed by age group, generating hazard ratios to assess male-to-female survival differences.

RESULTS

Overall, glioma incidence was higher in males. Male-to-female incidence was lowest in ages 0-9 years (IRR: 1.04, 95% CI:1.01 - 1.07, p=0.003), increasing with age, peaking at 50-59 years (IRR:1.56, 95% CI: 1.53 - 1.59, p< 0.001). Females had worse survival for ages 0-9 (HR:0.93, 95% CI:0.87-0.99), though male survival was worse for all other age groups, with the difference highest in those 20-29 years (HR:1.36, 95% CI:1.28-1.44). Incidence and survival differences by age and sex also varied by histological subtype of glioma.

CONCLUSION

To better understand the variation in glioma incidence and survival, investigating the intersection of age and sex is key. The current work shows that the combined impact of these variables is dependent on glioma subtype. These results contribute to the growing understanding of sex and age differences that impact cancer incidence and survival.

EXTH-58. THERAPEUTIC HDAC INHIBITION IN HYPERMUTANT DIFFUSE INTRINSIC PONTINE GLIOMA

Diffuse intrinsic pontine glioma (DIPG) continues to carry a dismal prognosis despite a growing understanding of its epigenetic regulation. While generally reclassified as diffuse midline glioma, H3 K27M-mutant (DMG), a subgroup of DIPGs do not harbor the classic histone mutation, with a further subset exhibiting a hypermutant phenotype. To evaluate whether hypermutant DIPG shares transcriptional vulnerabilities with H3K27M-mutant DMG, we screened a biopsy-derived treatment-naive PMS2 mutant DIPG model (PBT-24FH) for sensitivity to a panel of HDAC inhibitors (HDACi). In vitro evaluation of cell viability revealed the low nanomolar IC50 of quisinostat (50nM) and romidepsin (2nM). Dose-dependent increases in H3 acetylation and c-PARP were confirmed by western blot. Despite romidepsin’s superior potency in vitro, quisinostat demonstrated greater efficacy in an in vivo PBT-24FH flank study. 42 days following drug initiation, quisinostat-treated mice displayed dramatic tumor regression (mean volume= 33mm3, n= 7) compared to mice treated with romidepsin (mean volume= 669mm3, n= 7)(p= 0.005), or vehicle (mean volume= 990mm3, n= 6)(p< 0.001). Immunohistochemistry of quisinostat-treated tumors revealed few residual tumor cells displaying a low proliferative index. To evaluate cross-resistance, romidepsin-treated mice (mean volume= 1158mm3, n= 2) were switched to quisinostat treatment and displayed swift tumor regression (mean volume after 25 days of quisinostat= 419mm3), emphasizing quisinostat’s in vivo cytotoxic effect against both large tumors and tumors previously treated by another HDACi. To evaluate quisinostat’s effect on other hypermutant tumors, we tested HCT-116, a colon cancer cell line bearing a biallelic MLH1 deletion and observed similar cytotoxicity. We also aim to repeat these studies utilizing additional pediatric hypermutant high grade glioma models. Transcriptomic and proteomic investigations are underway to identify the mechanism of action underlying quisinostat-induced cytotoxicity. Ultimately, we are the first to demonstrate in vivo efficacy of the HDACi quisinostat against hypermutant DIPG, supporting further investigation and clinical advancement.

RTID-08. MACHINE LEARNING TO UNCOVER SIGNATURES OF VULNERABILITY IN GLIOBLASTOMA UMBRELLA SIGNATURE TRIAL (GUST)

Glioblastoma is characterized by intra- and inter-tumoral heterogeneity. A glioblastoma umbrella signature trial (GUST) posits multiple investigational treatment arms based on corresponding biomarker signatures. A contingency of an efficient umbrella trial is a suite of orthogonal signatures to classify patients into the likely-most-beneficial arm. Assigning optimal thresholds of vulnerability signatures to classify patients as “most-likely responders” for each specific treatment arm is a crucial task. We utilized semi-supervised machine learning, Entropy-Regularized Logistic Regression, to predict vulnerability classification. By applying semi-supervised algorithms to the TCGA GBM cohort, we were able to transform the samples with the highest certainty of predicted response into a self-labeled dataset and thus augment the training data. In this case, we developed a predictive model with a larger sample size and potential better performance. Our GUST design currently includes four treatment arms for GBM patients: Arsenic Trioxide, Methoxyamine, Selinexor and Pevonedistat. Each treatment arm manifests its own signature developed by the customized machine learning pipelines based on selected gene mutation status and whole transcriptome data. In order to increase the robustness and scalability, we also developed a multi-class/label classification ensemble model that’s capable of predicting a probability of “fitness” of each novel therapeutic agent for each patient. Such a multi-class model would also enable us to rank each arm and provide sequential treatment planning. By expansion to four independent treatment arms within a single umbrella trial, a “mock” stratification of TCGA GBM patients labeled 56% of all cases into at least one “high likelihood of response” arm. Predicted vulnerability using genomic data from preclinical PDX models correctly placed 4 out of 6 models into the “responder” group. Our utilization of multiple vulnerability signatures in a GUST trial demonstrates how a precision medicine model can support an efficient clinical trial for heterogeneous diseases such as GBM.

Surgical Therapies

CSIG-31. MULTI-OMICS TO EDGE INTO PRECISION MEDICINE FOR DIPG

DIPG is an incurable pediatric brain tumor with 80% of patients harboring H3F3A (H3.3) mutation that substitutes methionine for lysine at position 27 (K27M), resulting in global depletion of H3.3K27 me3 (trimethylation). These histone mutations modify the epigenome and alter oncogenic transcription, causing oncogenic insults to progenitor cells in early neurodevelopment (1). To determine the reprogramming pathways in the cell context of H3.3K27M tumors, we conducted LC-MS based proteomic and phosphoproteomic analysis on seven patient-derived DIPG cell lines. Three normal neuronal stem cell lines were included as non-tumor brain cells for comparison. Pathway analysis identified 29 pathways that are significantly altered in DIPG compared to normal brain cells at both the protein abundance and phosphosite level. Notably, AKT and MAPK associated PI3K signaling, VEGF signaling, mTOR signaling, and HIF1a signaling were differentially active in H3.3K27M tumors compared to healthy control cell lines. We saw significantly higher activity of multiple kinases involved in axon guidance and cytoskeletal remodeling in DIPG, such as PTK2B, DYRK2, TTBK2 and MARK2. This is the first time to report an increased abundance and kinase activity of PYK2 protein (coded by PTK2B), a close homologue of FAK and its associated signaling in DIPG. PYK2 has been proposed to act in concert with Src to link Gi- or Gq-coupled receptors with the mitogen-activated protein (MAP) kinase signaling pathway (2). Because of the shared signaling across kinase pathways, targeting activated PYK2 in DIPG may complement inhibitors of other dysregulated signaling networks in DIPG such as MAPK2, VEGFR, PI3K and Src. Our data also found that IL13RA2 was upregulated in DIPG. We conclude that for H3 K27M DIPG tumors, campaigns to target PYK2, MAPK2, VEGFR, PI3K, Src and IL13Ra2 using small molecules that traverse the blood brain barrier loom as promising opportunities for drug development.

CTIM-27. LOCOREGIONAL B7-H3-SPECIFIC CAR T CELLS FOR CHILDREN AND YOUNG ADULTS WITH DIPG: INTERIM REPORT OF BRAINCHILD-03 ARM C

Following preclinical optimization of B7-H3-specific CAR T cells against pediatric brain tumor models, we opened BrainChild-03 (NCT04185038), a phase 1 clinical trial of repeatedly dosed, outpatient, locoregional B7-H3-specific CAR T cells for children with recurrent/refractory central nervous system (CNS) tumors or diffuse intrinsic pontine glioma (DIPG). Here, we report the interim findings from patients enrolled on Arm C, dedicated to DIPG. The primary endpoints are feasibility and safety, with secondary endpoints of disease response. We utilize second-generation CAR T cells with a 4-1BB costimulatory domain and a methotrexate-resistant human DHFR mutein (huDHFRFS; L22F,F31S), allowing for methotrexate selection. We do not deliver conditioning chemotherapy. The first three evaluable patients with DIPG all met feasibility for generating a balanced CD4:CD8 CAR T cell product, with 3.85x109 CAR T cells generated for S005, 4.29x109 for S008, and 2.45x109 for S012, allowing for greater than 6 months of biweekly dosing for each patient. All subjects were treated at Dose Level 1 (1x107 CAR T cells). S005 received 10 doses before clinical progression greater than 2 years from diagnosis, S008 has received 10 doses and continues on therapy with decreased tumor volume, and S012 has received 5 doses and continues on study with stable disease. There have been no dose limiting toxicities (DLT). 3/3 patients exhibited post infusion fever, headache, and elevated serum CRP but no evidence of cytokine release syndrome (CRS) or systemic CAR T cells. 0/3 patients required PICU admissions. In the cerebrospinal fluid (CSF), 2/3 patients have had elevations of cytokines such as CXCL10 and CCL2, as well as circulating CSF CAR T cells. Advanced serial patient CSF proteomic and transcriptomic profiling are underway. Ultimately, this report provides preliminary evidence that outpatient locoregional B7-H3 CAR T cells for children with DIPG may be feasible and tolerable.

CLRM-01. MACHINE LEARNING TO UNCOVER SIGNATURES OF VULNERABILITY IN GLIOBLASTOMA UMBRELLA SIGNATURE TRIAL (GUST)

Glioblastoma is characterized by intra- and inter-tumoral heterogeneity. A glioblastoma umbrella signature trial (GUST) posits multiple investigational treatment arms based on corresponding biomarker signatures. A contingency of an efficient umbrella trial is a suite of orthogonal signatures to classify patients into the likely-most-beneficial arm. Assigning optimal thresholds of vulnerability signatures to classify patients as “most-likely responders” for each specific treatment arm is a crucial task. We utilized semi-supervised machine learning, Entropy-Regularized Logistic Regression, to predict vulnerability classification. By applying semi-supervised algorithms to the TCGA GBM cohort, we were able to transform the samples with the highest certainty of predicted response into a self-labeled dataset and thus augment the training data. In this case, we developed a predictive model with a larger sample size and potential better performance. Our GUST design currently includes four treatment arms for GBM patients: Arsenic Trioxide, Methoxyamine, Selinexor and Pevonedistat. Each treatment arm manifests its own signature developed by the customized machine learning pipelines based on selected gene mutation status and whole transcriptome data. In order to increase the robustness and scalability, we also developed a multi-class/label classification ensemble model that’s capable of predicting a probability of “fitness” of each novel therapeutic agent for each patient. Such a multi-class model would also enable us to rank each arm and provide sequential treatment planning. By expansion to four independent treatment arms within a single umbrella trial, a “mock” stratification of TCGA GBM patients labeled 56% of all cases into at least one “high likelihood of response” arm. Predicted vulnerability using genomic data from preclinical PDX models correctly placed 4 out of 6 models into the “responder” group. Our utilization of multiple vulnerability signatures in a GUST trial demonstrates how a precision medicine model can support an efficient clinical trial for heterogeneous diseases such as GBM.

Rigorous approach for unifying wave optics and state of the art rendering techniques

With the capabilities of diffractive optics there is a rising demand for determining the light interaction of diffractive elements with arbitrary illumination and scenery. Since the structured surfaces' scale lies within the visible wavelengths and below, the light's interaction cannot be simulated with state of the art geometric optic rendering approaches. This paper presents a new model for the inclusion of wave-optical effects into Monte Carlo path rendering concepts. The derived method allows the coupling of a rigorous full-field approach with the concept of backward ray propagation through complex scenes. Therefore, the rendering of arbitrarily structured periodic optical components is now possible for complex sceneries for design, verification and testing purposes. The method's performance is demonstrated by comparing rendering results of complex sceneries including CDs with corresponding photographs.

Abstract 2534: Underlying mechanism of response to neddylation inhibition in a subset of glioblastoma

Glioblastoma (GBM) is the most common and aggressive primary malignant brain tumor in adults with a 5-year survival rate of ~3%. The ubiquitin/proteasome system maintains intracellular homeostasis via degradation of unwanted proteins. Bortezomib is a first-in-class, non-specific proteasome inhibitor exploiting this system, although it has poor penetration across the blood brain barrier, and its lack of specificity is accompanied by adverse events. Neddylation is a specific pathway within the ubiquitin/proteasome system that is overactive in glioblastoma (GBM), and whose upregulation has been associated with glioma progression and worse survival. Pevonedistat is a first-in-class small-molecule neddylation inhibitor shown to impact protein degradation, leading to elevated abundance of some tumor suppressor proteins (Wee1, others), which then inhibit growth of GBM cells in culture and orthotopic xenografts. Because the molecular heterogeneity within and across GBM patients obscures therapeutic targets and obfuscates signals of efficacy in clinical trials, we propose the use of molecular “signatures of vulnerability” to targeted agents in subsets of preclinical models. We and others have shown that pevonedistat interferes with the growth of multiple types of cancers, including GBM, but the determinants of vulnerability are not fully understood. Here, we report a selective vulnerability to pevonedistat in a subset of GBM, specifically, instances with mutations or copy number deletions of PTEN are associated with de novo resistance to pevonedistat. Time-course studies of sensitive and non-sensitive GBM cells using transcriptomics and proteomics/phosphoproteomics enable independent discovery and testing for determinants of response to pevonedistat. Our results demonstrate that in GBM, resistance to pevonedistat is driven by reduced PTEN-chromatin binding (loss-of-function or lower expression) that is also independent of PTEN's lipid phosphatase activity (i.e., PI3K/AKT signaling). Across 25 glioma cell lines, we found that PTEN signaling, DNA replication, and chromatin instability pathways are the most significant differentiators between pevonedistat sensitive vs. non-sensitive models. In GBM models with modest to low sensitivity to pevonedistat, TOP2A expression was elevated. Combination treatment with the TOP2A inhibitor, etoposide, proved synergistic with pevonedistat. We report, for the first time, that PTEN status both serves as a novel biomarker for GBM sensitivity to pevonedistat and reveals a synergistic vulnerability of TOP2A inhibitors in combination with pevonedistat. Paired use of GBM PDX models of varying sensitivity with drug development testing allows the advancement of a promising agent as well as a patient-enrollment “signature of vulnerability” likely to increase the likelihood of demonstrating therapeutic efficacy in early clinical trials.

Citation Format: Shayesteh Ferdosi, Brett Taylor, Matthew Lee, Nanyun Tang, Sen Peng, Rita Bybee, George Reid, Lauren Hartman, Krystine Garcia-Mansfield, Ritin Sharma, Patrick Pirrotte, Frank Furnari, Harshil Dhruv, Michael Berens. Underlying mechanism of response to neddylation inhibition in a subset of glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2534.

IMMU-11. CLINICAL UPDATES AND CORRELATIVE FINDINGS FROM THE FIRST PATIENT WITH DIPG TREATED WITH INTRACRANIAL CAR T CELLS

We report preliminary data for the first subject with diffuse intrinsic pontine glioma (DIPG) treated with intracranial CAR T cells. BrainChild-03 (NCT04185038) is a phase 1 trial of repetitively-dosed locoregional B7-H3-specific CAR T cells for children with recurrent/refractory central nervous system (CNS) tumors or DIPG. DIPG patients enroll on Arm C, on which B7H3CARs are delivered into the ventricular system via a CNS reservoir catheter. This study does not use lymphodepletion. Primary endpoints are feasibility and safety, with second endpoints of disease response. This 18-year-old female (BrainChild-03 S005) with radiographically-classic DIPG and biopsy-confirmed H3 K27M mutation enrolled on Arm C after progression 552 days from diagnosis following focal radiation and temozolomide, irinotecan, and bevacizumab. Apheresis and manufacturing produced 4.2x10⁹ second-generation B7H3CARs with a methotrexate-resistant human DHFR mutein (huDHFR^FS; L22F,F31S) in a single transcript in combination with the B7-H3-specific CAR and EGFRt, each separated by a T2A linker, allowing methotrexate selection and enrichment. At time of submission, she has received 10 every-other-week outpatient infusions of 1x10⁷ B7H3CARs (first dose on October 2, 2020). She has had no DLTs, but has experienced grade 2 fever and grade 2–3 headache peaking ~12–48 hours after each infusion. Following the 8^th CAR T cell infusion, she experienced increased focal weakness and dysarthria at ~72 hours with resolution after 48 hours. She has not experienced cytokine release syndrome (CRS). She has stable disease 138 days post-initial CAR T cell infusion. Frequently collected correlative studies have detected viable B7H3CARs in the CSF post-infusion via flow cytometry. CSF cytokine analysis has revealed elevations of CXCL10, GM-CSF, and G-CSF following B7H3CAR infusions, without correlation in the serum. A second evaluable subject with DIPG has also received 4 locoregional doses of 1x10⁷ B7H3CARs without a DLT. She also has stable disease and detectable viable B7H3CARs in the CSF.

JAM-A functions as a female microglial tumor suppressor in glioblastoma

BACKGROUND

Glioblastoma (GBM) is the most aggressive primary brain tumor and has a dismal prognosis. Previously, we identified that junctional adhesion molecule A (JAM-A), a cell adhesion molecule, is highly elevated in human GBM cancer stem cells and predicts poor patient prognosis. While JAM-A is also highly expressed in other cells in the tumor microenvironment, specifically microglia and macrophages, how JAM-A expression in these cells affects tumor growth has yet to be determined. The goal of this study was to understand the role of microenvironmental JAM-A in mediating GBM growth.

METHODS

Male and female wild-type (WT) and JAM-A-deficient mice were transplanted intracranially with the syngeneic glioma cell lines GL261 and SB28 and were assessed for differences in survival and microglial activation in tumors and in vitro. RNA-sequencing was performed to identify differentially regulated genes among all genotypes, and differences were validated in vitro and in vivo.

RESULTS

We found that JAM-A-deficient female mice succumbed to GBM more quickly compared with WT females and JAM-A-deficient and male WT mice. Analysis of microglia in the tumors revealed that female JAM-A-deficient microglia were more activated, and RNA-sequencing identified elevated expression of Fizz1 and Ifi202b specifically in JAM-A-deficient female microglia.

CONCLUSIONS

Our findings suggest that JAM-A functions to suppress pathogenic microglial activation in the female tumor microenvironment, highlighting an emerging role for sex differences in the GBM microenvironment and suggesting that sex differences extend beyond previously reported tumor cell-intrinsic differences.

Freeform optics design for extended sources in paraxial approximation exploiting the expectation maximization algorithm

Freeform optics generating specific irradiance distributions have been used in various applications for some time now. While most freeform optics design algorithms assume point sources or perfectly collimated light, the search for algorithms for non-idealized light sources with finite spatial as well as angular extent is still ongoing. In this work, such an approach is presented where the resulting irradiance distribution of a freeform optical surface is calculated as a superposition of pinhole images generated by points on the optical surface. To compute the required arrangement of the pinhole images for a prescribed irradiance pattern, the expectation maximization algorithm from statistics is applied. The result is then combined with a ray-targeting approach for finding the shape of the corresponding freeform optical surface. At its current state, the approach is applicable to Gaussian input irradiances, single-sided freeform optics and for the paraxial case. An example freeform optical surface for laser material processing is shown and discussed demonstrating the performance and the limitations of the presented approach.

DDRE-24. ACQUIRED RESISTANCE TO TARGETED INHIBITORS IN EGFR-DRIVEN GLIOBLASTOMA: IDENTIFICATION OF DUAL KINASE TARGETS

Glioblastoma (GBM) is a devastating primary brain tumor with 5-year survival < 5%. CDKN2A deletion (~60%) and EGFR amplification (55–60%) mutations frequently co-occur in these tumors. EGFR is an attractive therapeutic target due to its mutational frequency and availability of multiple brain-penetrant tyrosine kinase inhibitors (TKI). Several EGFR TKI have failed clinically, due in part to acquired resistance. To mechanistically examine this type of resistance, we used genetically engineered mouse astrocytes harboring Cdkn2a deletion and EGFRvIII, a common (35%) activating mutation. Resistant cells were generated via chronic exposure to gefitinib or erlotinib, either in vitro or in vivo. Resistance to these first-generation EGFR TKI conferred cross resistance (up to 36-fold ΔIC50) to a panel of second- and third-generation TKI relative to sensitive parental lines. Moreover, integrated RNA sequencing (RNA-seq) and chemical proteomics (multiplexed inhibitor beads and mass spectrometry (MIB-MS)) showed that the kinase transcriptome and proteome were rewired in resistant cells: 113 of ~300 detected kinases were differentially expressed (p< 0.05). We then used these techniques to examine acute (≤ 48 h) kinome changes in both sensitive and resistant cells upon treatment with a CNS-penetrant, second-generation EGFR TKI, afatinib. Whereas exposure of treatment-naïve, sensitive cells to afatinib significantly rewired the kinome (120 differentially expressed kinases), the response of resistant cells to drug re-challenge was significantly blunted (13 differentially expressed kinases). A subset of expressed kinases (35 of 263) dynamically responded to afatinib in both sensitive and resistant cells. Overall, upregulated kinases include those implicated in the biology of gliomas (Bmx, Fgfr2) and of other cancers (Pdgfrb, Mapk3/4, Ddr1/2, Pdk2). These kinases thus represent putative druggable targets for dual inhibition therapy. Integrated kinome profiling using MIB-MS and RNA-seq in GBM models with defined mutational profiles provides a powerful framework to identify novel therapeutic targets that could significantly alter current treatment paradigms.

CSIG-16. SEXUAL DIMPORHISM IN IRON ACQUISITION IN GLIOBLASTOMA

Glioblastoma (GBM) is the most aggressive brain cancer. Sex differences in incidence and clinical outcomes have been reported, however, our knowledge of contributing mechanisms is limited. Iron acquisition is key to robust tumor growth. Upregulation of Transferrin (Tf, iron transport protein)/Transferrin receptor (TfR) is found in multiple different cancers. We have identified H-ferritin (FTH1) as involved in iron transport and explore its uptake in GBM in this study. We interrogated iron uptake in a syngeneic orthotopic mouse model (GL261 cells) using male and female mice. After the tumors were established, radioactive 125I labeled Tf and FTH1 proteins were injected retro-orbitally in the mice. After 24 hours, tumors were removed, homogenized and analyzed for Tf and FTH1 uptake. There was a significant difference in Tf uptake into the tumor versus matched non-tumor tissue in both males and females and the uptake in the tumors was 1.5-fold higher in males than females. There was no significant difference in FTH1 uptake between male and female tumors nor between tumor and matched non-tumor brain tissue. Binding analyses were performed on homogenized samples of human male and female GBM tissue samples using 125I labeled Tf and FTH1. Tumors from males had increased binding of both proteins compared to tumors from females. We next queried the TCGA database and found in females, high TfR expression was associated with poor survival but not in males. TCGA database revealed a robust expression of Tim1, a putative receptor for FTH1, but its expression did not relate to survival. In summary, this study demonstrates FTH1 binding to GBMs and sexual dimorphism in iron acquisition via Tf and survival.

TAMI-17. RELATIONSHIP BETWEEN IRON METABOLISM, IMMUNE CELL INFILTRATION AND SEX-BASED SURVIVAL DIFFERENCES IN GLIOMAS

Iron plays a central role in cellular metabolism, both in normal cellular functioning and in tumorigenesis. Recent evidence has shown sex-based survival differences in glioblastoma (GBM) may be related to differential expression of metabolism genes. We previously reported the iron regulating gene, HFE, was shown to have a sex-based survival impact in both low-grade gliomas and GBM. We additionally found that females with low HFE expressing tumors have significantly higher survival than males in GBM. To evaluate the relationship between iron gene expression and sex-based survival differences in GBM, we analyzed TCGA GBM gene expression and clinical data. We first analyzed the impact of iron genes on sex-based survival. In addition to HFE, FTL, TFRC, TF, and SLC39A8 (ZIP8), also showed sex-based survival differences. We then compared correlations of HFE and other iron genes to identify whether male and female GBMs differ in iron regulation and metabolism. HFE expression is significantly positively correlated with HMOX1, SLC25A28, SLC11A2, FTH1, HAMP, and TFR2 only in females. Alternatively, HFE expression is negatively correlated with ACO2 (mitochondrial aconitase) in males and ACO1 (cytoplasmic aconitase) in females. We noted that the expression of certain iron genes was highly associated with immune cell infiltration based on sex. TFR2, LRP1, and XIST expression were negatively correlated with low immune cell infiltration in females, but not males. Alternatively, in males, SLC11A2, ACO2, FOXO1, HIF1a, and HAMP genes were negatively correlated with immune infiltration. This suggests that differences in iron regulation between males and females may be contributing to differences in immune function and subsequent survival in GBM. These data suggest that the iron signature of a tumor reflects and possibly drives the metabolic and immune landscape of the tumor microenvironment thereby directly impacting survival differences between male and female GBMs.

BIOM-24. IDENTIFYING VULNERABILITY SIGNATURES THROUGH MACHINE LEARNING IN AN UMBRELLA TRIAL FOR GLIOBLASTOMA

Glioblastoma is characterized by intra- and inter-tumoral heterogeneity. An umbrella trial tests multiple treatment arms depending on corresponding biomarker signatures. A contingency of an umbrella trial is a suite of preferably orthogonal molecular biomarkers to classify patients into the likely-most-beneficial arm. Assigning thresholds of molecular signatures to classify a patient as a “most-likely responder” for one specific treatment arm is a crucial task. Gene Set Variation Analysis (GSVA) of specimens from a GBM clinical trial of methoxyamine associated differential enrichment in DNA repair pathways activities with patient response. The 44 DNA-repair related pathways confound confident “high” enrichment of responder, as well as obscuring to what degree each feature contributes to the likelihood of a patient’s response. Here, we utilized semi-supervised machine learning, Entropy-Regularized Logistic Regression (ERLR) to predict classification. By first training all available data using semi-supervised algorithms we transformed unclassified TCGA GBM samples with highest certainty of predicted response into a self-labeled dataset. In this case, we developed the predictive model which has a larger sample size and potential better performance. Our umbrella trial design currently includes three treatment arms for GBM patients: arsenic trioxide, methoxyamine, and pevonedistat. Each treatment arm manifests its own signature developed by the above (or similar) machine learning pipeline based on selected gene mutation status and whole transcriptome data. By expansion to three, independent treatment arms within a single umbrella trial, a “mock” stratification of TCGA GBM patients binned 56% of all cases into a “high likelihood of response“ arm. Predicted vulnerability using genomic data from preclinical PDX models placed 4 out of 6 models into a “high likelihood of response” regimen. Our utilization of multiple vulnerability signatures in an umbrella trial demonstrates how a precision medicine model can support an efficient clinical trial for heterogeneous diseases such as GBM.

CSIG-10. GENOTYPE – KINOME GUIDED DEVELOPMENT OF PRECISION EGFR-TARGETED THERAPEUTICS FOR GLIOBLASTOMA

Glioblastoma (GBM) is an aggressive primary brain tumor with poor survival and limited treatment options. However, it is an attractive candidate for precision therapeutic approaches due to the frequency of amplification and/or activating mutations in the epidermal growth factor receptor (EGFR) gene and the availability of several brain penetrant second- and third-generation EGFR tyrosine kinase inhibitors (TKI). We used comprehensive molecular profiling of a panel of genetically engineered mouse astrocyte models to examine whether mutational profiles, particularly EGFR and PTEN status, could be used to identify kinases upregulated in specific mutational backgrounds. Using RNA-seq and multiplex inhibitor bead/mass spectrometry (MIB-MS) to analyze the kinase transcriptomes and proteomes, respectively, we have identified several potential targets for combination therapy. Overexpression of wild type EGFR in immortalized, Cdkn2a-/- astrocytes resulted in mild rewiring of the GBM kinome. Only 5 kinases aside from EGFR itself were overexpressed on either the transcript or protein levels. One overexpressed kinase, Hck, has been shown to be involved in cell survival, proliferation, adhesion, and migration. In contrast, overexpression of EGFRvIII, a constitutively active, extracellular domain truncation mutant of EGFR, resulted in significant alteration of the GBM kinome – 81 kinases showed differential expression, with 27 upregulated. One potentially attractive target among these was Cdk6, a drug-targetable, prognostically significant cyclin-dependent kinase implicated in proliferation, migration, and invasion. Finally, overexpression of EGFRvIII in cells lacking Pten dysregulated 46 kinases, including 15 upregulated. One particularly interesting target in these cells was Ddr2, a tyrosine kinase involved in migration, invasion, and extracellular matrix remodeling. We conclude that Hck, Cdk6, and Ddr2 represent attractive targets for therapeutic intervention in their relevant genetic contexts. These findings also suggest that molecular diagnostics for EGFR and PTEN status may be useful in guiding development of rational, EGFR TKI-centric drug combinations.

TAMI-43. IMPACT OF SEX AND RADIATION ON IRON TRAFFICKING IN BONE MARROW DERIVED MACROPHAGES

The tumor microenvironment in glioblastoma provides cancer cells with favorable conditions to proliferate and invade surrounding tissues. Macrophages comprise a large portion of the glioblastoma tumor microenvironment (TME) both in terms of volume and function. These cells have been reported to influence tumor progression by modulating immune responses, remodeling extracellular matrix, and providing nutrients to cancer cells among numerous other functions. Radiation therapy forms one of the pillars of glioblastoma management along with surgical resection and chemotherapy. Here we investigated the effects of radiation on macrophage iron metabolism. Using mouse bone-marrow-derived macrophages (BMDMs) we performed in-vitro 59Fe radiotracer assays to study how radiation exposure modified iron trafficking in these cells. We found that low dose radiation at 0.25, 0.5, or 2 Gy from a 60Co source stimulated iron release from the BMDMs with maximal release occurring at 0.5 Gy. Moreover, we observed that iron release was dependent on the amount of serum present in culture media with cells cultured in 20% fetal bovine serum (FBS) showing reduced iron release profiles compared to those cultured in 10% or 1% FBS. Since glioblastoma patients exhibit sexually dimorphic survival outcomes, we investigated whether these radiation-induced responses occurred in a sexually dimorphic pattern. At radiation doses of 0.25 Gy we observed that male macrophages tended to release more iron than female macrophages despite no differences in iron uptake between the sexes – raising the question as to whether differential iron trafficking in response to treatment contributes to the poorer survival outcomes observed in males. Our data suggest that delineating how supporting cells such as macrophages respond to glioblastoma treatment regimens may provide insights into addressing mechanisms of treatment resistance and further our understanding of the sexual dimorphism observed in patient outcomes.

Abstract 6681: Inhibition of LSD1 using the reversible inhibitor SP2577 promotes Interferon dependent anti-tumor response in adrenocortical carcinoma

Purpose: Adrenocortical carcinoma (ACC) is a rare and aggressive disease with poor prognosis. The first line of treatment is surgery followed by chemotherapy (Mitotane), though these therapies still result in a dismal 15-44% survival rate after 5 years. Currently, there is no effective second line of treatment. Efforts have been made to treat ACC with immune therapy agents such as Avelumab, a PD-L1 antibody, but a recent study showed that there was only an overall response rate of 6% in a phase 1 clinical trial, further exhibiting a need for the development of novel therapeutics. Further research suggests that the current first line ACC treatments may benefit from the addition of immune therapies that induce an Interferon Beta (IFN-β) response in the tumor microenvironment. Previous studies have shown that treatment with IFN-β induces cell death and sensitizes human ACC lines to Mitotane, implying that combination of therapies that induce an IFN-β response may improve the anti-tumor response to Mitotane.

Recently, inhibition of Lysine-Specific histone Demethylase 1 (LSD-1) was shown to stimulate an IFN-dependent anti-tumor response driven by the expression of Endogenous Retroviral Elements (ERVs) in a variety of cancer types. In addition, PD-L1 expression has also been shown to increase after LSD-1 inhibition. Yang, et al. reported that SP-2509, a preclinical analog of SP-2577 (Seclidemstat, currently in clinical phase I trials, Salarius Pharmaceuticals, Houston, TX) was able to induce the expression and release of interferons and cytokines in triple negative breast cancer models. In this study we investigate the ability of SP-2577 to promote the expression of IFN-β in human ACC cell lines, resulting in an anti-tumor immune response.

Experimental Design: qRT-PCR was performed on ACC cell lines H295R and SW-13 at 48 and 72 hours post treatment to verify the ability of SP-2577 to promote IFN-β expression. ACC cells were co-cultured with peripheral blood mononuclear cells (PBMCs) in a 3D platform to evaluate lymphocyte infiltration following SP-2577 treatment. Interferons, chemokines, and cytokines were quantified in the culture supernatant after treatment with SP-2577 using the Meso Scale Discovery platform.

Results: qRT-PCR showed that SP-2577 promotes ERV-mediated immune response in ACC cell lines via the expression and release of interferons, chemokines, and cytokines. Additionally, SP-2577 stimulates lymphocyte infiltration in an ACC derived 3D-spheroid platform further resulting in a cytotoxic effect.

Conclusion: Our data suggests that the LSD-1 inhibitor SP-2577 can be used as a single agent in ACC patients to promote immune response and can potentially sensitize the tumor to first line of therapy, Mitotane.

This study was supported by funding from the TGen Foundation.

Citation Format: Sunil Sharma, Raffaella Soldi, Alexis Weston, Samuel Sampson, Sherin Daniel Ampanattu, Scott Celinski, Haiyong Han, Mohan Kaadige, Michael Berens, Jeffrey Trent. Inhibition of LSD1 using the reversible inhibitor SP2577 promotes Interferon dependent anti-tumor response in adrenocortical carcinoma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 6681.

Abstract 2007: Development of a clinical assay for predicting glioblastoma (GBM) patients most likely to respond to arsenic trioxide (ATO)

Glioblastoma (GBM) is the most frequently reported primary malignant brain tumor (29.6%). The prognosis for patients who develop GBM is bleak, with average survival after diagnosis of 12-16 months. Although conventional treatment with surgery, irradiation, and temozolomide postpones tumor progression and extends patients survival, these tumors universally recur and unrelentingly result in patient death. Personalized therapies against molecular targets that drive the growth of the bulk of primary tumors have so far been unsuccessful in clinical trials, due to lack of biomarker driven approaches. Thus, there is significant unmet need to begin biomarker driven precision medicine trials for treatment of GBM. Arsenic trioxide (ATO) is an inorganic compound that induces apoptosis via multiple pathways. Arsenic trioxide (TRISENOX®) is approved by the FDA for patients with acute promyelocytic leukemia (APL). Pre-clinical studies in brain tumors suggest that ATO is synergistic with radiation therapy (RT) and may enhance effects of radiation. In an earlier Phase II clinical trial (NCT00275067) using intravenous ATO and temozolomide in combination with radiation therapy for patients with newly diagnosed malignant gliomas, a subset of patients demonstrated notable benefit (Progression free survival (avg. = 638 days) and overall survival (avg. = 967 days)). Comparing RNAseq data from preclinical models and specimen from the Phase II clinical trial, the responder group could be confidently distinguished from the non-responder cohort leading to gene signatures of differential ATO sensitivity. Applying a Relative Expression Ordering (REO) Analysis framework, we pinpointed a probability-based roster of 28 top scoring pairs (TSP) as the classifier by which to identify patients with a higher likelihood to benefit from including ATO in combination with TMZ and radiation. This method is completely independent of platform on which data is collected and can be used for analysis of individual, newly-enrolled, n = 1 patients. We are advancing a protocol using the above gene classifier as enrollment criteria for an Adaptive clinical trial testing an oral formulation of ATO for newly diagnosed IDH1 WT Primary GBM patients; the trial will test whether patients whose tumors with ATO Classifier show 6-month PFS benefit by addition of ATO to Standard-of-Care. The trial will validate and refine the comprehensive biomarker panel that could identify most likely GBM responders to ATO and TMZ treatment in combination with radiation. Supported by a grant from the Baylor Scott & White Foundation.

Citation Format: Sen Peng, Jinghua Gu, Xuan Wang, Sanhita Rath, Jacob Cardenas, Nicholas Schork, George Snipes, Harshil Dhruv, Karen Fink, Michael Berens. Development of a clinical assay for predicting glioblastoma (GBM) patients most likely to respond to arsenic trioxide (ATO) [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2007.

Abstract 1434: Loss of PTEN confers resistance to neddylation inhibition through TOP2A in glioblastoma

As a disease that has not seen lasting advances in its standard of care for decades, glioblastoma (GBM) calls for novel therapeutic approaches, and given GBM's heterogeneous and aggressive nature, precision medicine stands out as an appealing treatment modality for this tragic disease. Neddylation is a post-translational modification process that regulates protein function and turnover. MLN4924, or Pevonedistat, inhibits the E1 ligase Nedd8 activating enzyme (NAE) and, in doing so, the E1, E2, and E3 neddylation cascade. In its most researched role, NEDD-8 is conjugated to the cullin subunit of cullin-RING ligases (CRLs), a family of E3 ligases that facilitate the ubiquitination of different substrates. Thus, inhibition of neddylation decreases ubiquitination and degradation of particular substrates: e.g., p21, p27, wee-1, and other cell cycle regulatory and pro-apoptotic proteins. We and others have shown that neddylation stands out a promising therapeutic approach to cancer and GBM, given the vulnerability to MLN4924 in subset of different in vitro and in vivo models. In this work, we establish the selective vulnerability to MLN4924 in GBM and explore the biological mechanisms that underlie this differential response. Using the Cancer Cell Line Encyclopedia (CCLE) and the Genomics of Drug Sensitivity in Cancer (GDSC) databases, we found that mutations and deletions in phosphatase and tensin homolog (PTEN) are associated with resistance to MLN4924 in both low-grade gliomas and GBM. Our established models validated these findings, as our sensitive models – GB1 (IC50= 0.28 μM) and LN18 (IC50 = 0.19 μM), – expressed PTEN, while our resistant models – M059K (IC50 = 5.5 μM) and SNU1105 (IC50 = 20.9 μM) – did not. Knocking down PTEN with siRNA in our sensitive models increased their IC50 values three- to four-fold. This relationship was further validated in multiple isogenic systems. We sought to uncover whether this PTEN-status dependent resistance was contingent on its well-studied lipid phosphatase activity and PI3K/AKT signaling. While the resistant models had higher baseline levels of phosphorylated AKT (p-AKT), we found that MLN4924 treatment did not consistently alter p-AKT levels in our sensitive and/or resistant models. We also used a PTEN null isogenic glioma stem cell system (GSC23), in which wild-type PTEN and lipid phosphatase dead (G129A) PTEN was re-expressed, and found that both PTEN models sensitized GSC23 to MLN4924 treatment equally. While independent of the PI3K/AKT axis, we did find that there were significant increases in topoisomerase II alpha (TOP2A) levels in the post-treatment lysate of our resistant (PTEN-null) models. Targeting this increase in TOP2A expression, we found strong synergy with two different TOP2 poisons – etoposide and doxorubicin – and MLN4924 in a combination DDR. For the first time, we show that PTEN status not only serves as a novel biomarker for MLN4924 response but also reveals a vulnerability to TOP2A inhibitors when used in combination with MLN4924.

Citation Format: Brett Taylor, Shayesteh Ferdosi, Nanyun Tang, Rita Bybee, George Reid, Lauren Hartman, Darren Finlay, Kristiina Vuori, Matthew Lee, Sen Peng, Frank Furnari, Harshil Dhruv, Michael Berens. Loss of PTEN confers resistance to neddylation inhibition through TOP2A in glioblastoma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1434.

Abstract 857: Discovering molecular biomarkers through machine learning to assign patients to one of multiple arms for most-likely-to-benefit in an umbrella trial for glioblastoma

Glioblastoma Multiforme (GBM) is the most common primary brain tumor, characterized by intra- and inter-tumoral heterogeneity and relentless, lethal progression. Molecular data from next-generation sequencing subclassifies the disease into discrete subtypes, setting the stage for biomarker-driven clinical trials. An umbrella trial tests various treatment arms depending on various biomarker signatures that preferentially assign each new patient into a specific treatment arm. A contingency of an umbrella trial is a suite of distinct, orthogonal molecular biomarkers to classify patients into the likely-most-beneficial arm of the different separate treatment arms. Assigning thresholds to molecular signatures in order to classify a patient as a likely-responder within a single treatment arm is a crucial task. For example, in GBM patients treated with methoxyamine (covalently binds to apurinic/apyrimidinic (AP) DNA damage sites), Gene Set Variation Analysis (GSVA) identifies the differential enrichment in several DNA repair pathways of individuals. The 44 DNA-repair related pathways (or features) make it difficult to define “high” enrichment, as well as which, and to what degree, each feature contributes to the likelihood of a patient's response. Here, we utilize two semi-supervised algorithms, the Transductive Support Vector Machine (TSVM) and Entropy-Regularized Logistic Regression (ERLR). By first training these algorithms in a semi-supervised method using both patient as well as TCGA samples we can then transform the output of GSVA on unlabeled TCGA samples into a labeled data set based on those with the highest certainty of their predicted response. By employing semi-supervised algorithms to self-label the response of GBM Cases in TCGA, a predictive model can be developed which has a larger sample number, thereby reducing uncertainty.

Our umbrella trial design currently includes three treatment arms for GBM patients: arsenic trioxide, methoxyamine, and pevonedistat. Each treatment arm contains its own biomarker developed by the above machine learning pipeline based on selected gene mutation status in addition to whole transcriptome data. By conducting a “mock” stratification of GBM patients in TCGA, we identify likely patient responders in a single arm (as low as 9%) to upwards of 45% by expansion to three, independent treatment arms within a single trial. Our utilization of multiple biomarker signatures in an umbrella trial demonstrates how a precision medicine model can be effective in heterogeneous diseases such as GBM.

Citation Format: Matthew E. Lee, Sen Peng, Brett Taylor, Shay Ferdosi, Harshil Dhruv, Manmeet Ahluwalia, Karen Fink, Michael Berens. Discovering molecular biomarkers through machine learning to assign patients to one of multiple arms for most-likely-to-benefit in an umbrella trial for glioblastoma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 857.

EPID-01. SEX DIFFERENCE IN EXPRESSION OF IRON-RELATED GENES AND SURVIVAL IN GLIOBLASTOMA PATIENTS

Sex impacts the clinical outcome of glioblastoma (GBM) patients. Iron is a key metabolic driver in glioma biology impacting both the neoplastic cells directly and tumor progression through the immune system. We previously reported the expression of HFE (iron homeostatic) gene has a sexually dimorphic impact on survival in GBM patients. To further interrogate the relationship of iron and sexual dimorphism in GBMs, we expanded our analysis to include the ferrome (profile of genes and proteins involved in iron regulation). These genes were identified using IronChip and then interrogated for sexually dimorphic expression and survival in primary GBM patients using The Cancer Genome Atlas GBM database. Biological pathway analysis indicated that the Toll receptor signaling and immune system signaling pathways are most strongly correlated to sex difference and GBM patient survival. Toll like receptor 4 (TLR4) expression in the bottom quartile is associated with a modest survival benefit for female GBM patients compared to male, but in the top quartile of expression of this gene the survival for males is dramatically greater than that for females (18 months versus 3.6 months). Males expressing top quartile of Dual specificity phosphatases (DUSP1) also known as MKP1 survive 24.2 months whereas females survive 5.8 months. There is no sex effect of DUSP1 in the bottom quartile of expression. These data suggest that top quartile expression of TLR4/DUSP1 are both a benefit for males (extending their survival) and a detriment to females (shortening their survival). Both TLR4 and DUSP1 are richly expressed in macrophages. The data suggest that macrophage infiltration into the tumor could be particularly negatively impactful in females; a finding consistent with our previous reports for the HFE gene. In summary, iron related gene expression is linked to sexual dimorphism in survival and may provide insights on sex specific tumor biology.

ACTR-43. GENOMIC ANALYSIS OF RESPONDERS OF PHASE II TRIAL OF TEMOZOLOMIDE AND TRC-102 (BASE EXCISION REPAIR INHIBITOR) IN BEVACIZUMAB-NAÏVE GLIOBLASTOMA AT FIRST RECURRENCE

BACKGROUND

Temozolomide forms O6-methylguanine (O6mG), 7-methylguanine (N7mG), and 3- methyladenine (N3mA) DNA adducts. The O6mG DNA adduct is repaired by MGMT. N7mG and N3mA DNA adducts are removed by the base excision repair (BER) pathway, initiated by N- methylpurine DNA glycosolase (MPG). TRC-102 is a BER inhibitor that binds to the apurinic site created through the action of MPG.

METHODS

A phase II study of adult glioblastoma in first recurrence was performed in the Adult Brain Tumor Consortium with temozolomide, 150 mg/ m2 and TRC-102, 150 mg (1–5/ 28 days). Primary objective included radiographic response rate. Secondary objectives included safety and PFS-6. Exploratory objectives included tumor expression of N-methylpurine DNA glycosylase (MPG). The study tested hypothesis that combination therapy will achieve 30% RR. To understand the context of vulnerability to TRC102 we performed RNA sequencing on treatment naïve tissue from 7 patients.

RESULTS

Nineteen patients were enrolled in first stage. Median age was 60 years (range: 48–76), 53% females, median KPS was 80 (range: 70–90). Median cycles of treatment was 2 (range: 1–12). No responses were observed. Median OS was 11.0 months (95% CI: 8–18 months), median PFS was 2.0 months (95% CI: 1.8–3.6 months). PFS-6 rate was 10.5 % (2/19). The combination was safe. MPG staining was negative in six, 1+ in five and 2+ in three patients. PFS of 11 + months in two patients (exceptional responders) was associated with MPG expression. Preliminary analysis on RNA sequencing revealed significant enrichment for DNA Damage Response pathways (MsigDB), chromosomal instability gene signature (CIN70 and CIN25), and proliferative gene signature (PCNA25) in these 2 patients.

CONCLUSIONS

TRC 102 with temozolomide has acceptable safety but did not meet the primary endpoint of response. Gene signature of MsigDB, CIN70, CIN25 and PCNA25 was seen in exceptional responders and biomarker driven study is planned.

TMIC-31. IMPACT OF IRON ON MACROPHAGE IMMUNE PHENOTYPE IN THE GLIOBLASTOMA TUMOR MICROENVIRONMENT

The tumor microenvironment (TME) in glioblastoma presents a significant hurdle to effective immunotherapies as it consists of an immunosuppressive niche that results in inhibition of anti-tumor immunity. Macrophages comprise a large portion of the glioblastoma TME as they are prolific secretors of immunosuppressive cytokines and can comprise of up to 30% of tumor volume. In addition to their crucial role in immune function, macrophages are important players in iron regulation owing to their ability to efficiently sequester and release iron. We studied the impact of iron status in macrophage anti-tumor immune functions and phenotypic plasticity using RAW264.7 macrophages and bone-marrow-derived macrophages (BMDMs). Using RAW264.7 cells, we found that the iron status of macrophages impacts their immune function by modulating expression of the co-stimulation membrane proteins CD80 and CD86 as well as the mannose receptor CD206. Interestingly, the iron-mediated immunomodulation was dependent on the formulation of iron with nanoparticle formulations such as ferumoxytol (Fe-NP) upregulating CD80, CD86, and decreasing CD206 while ferric ammonium citrate (FAC) downregulated CD80, CD86 and upregulated CD206. We analyzed expression of inflammatory cytokines in BMDMs with multiplex cytokine analysis and found that both FAC and Fe-NP increased expression of the chemotactic signals CXCL10, CCL1, CCL3, and CCL4. Additionally, we found that iron status impacts the ability of macrophages to repolarize from an inflammatory, immune-activating phenotype into an immunosuppressive phenotype upon exposure to glioblastoma tumor-conditioned media. We used qRT-PCR to examine gene expression of the tumor immunity-related genes TNFa, IL1B, NOS2, and IL10 and found that iron loaded RAW264.7 macrophages stimulated with lipopolysaccharide resisted repolarization into an immunosuppressive phenotype 24 hours after exposure to glioblastoma tumor-conditioned media. Our results suggest that understanding the link between iron status and immune function in the tumor microenvironment may be an important step in improving therapies against glioblastoma.

EPID-03. HISTOLOGY-SPECIFIC BRAIN TUMOR INCIDENCE AND SURVIVAL VARIES BY SEX

BACKGROUND

Significant sex differences exist in cancer, and males have higher incidence and lower survival compared to females for most cancers. No large-scale studies have systematically examined sex differences in incidence and survival across all primary brain histologies. We performed a comprehensive investigation of the differences in incidence and survival in patients diagnosed with primary malignant brain and other CNS tumors by specific histologies.

METHODS

Age-adjusted incidence rate ratios (IRR) and 95% confidence intervals (945% CI) were generated from the United States Cancer Statistics (USCS) Public Use Database. Data from the Surveillance, Epidemiology, and End-Results (SEER) program were used to calculate overall survival. Data was restricted to patients with histologically or radiologically confirmed, primary malignant tumors diagnosed between 2001 and 2015. Histological groupings were categorized based on the Central Brain Tumor Registry of the United States (CBTRUS). Cox proportional hazards models were used to calculate hazard ratios (HR) adjusted for age for males as compared to females.

RESULTS

Males exhibited higher incidence than females in all brain histologies except meningioma (IRR=0.83; 95% CI 0.73–0.93) and other neuroepithelial tumors (i.e. polar spongioblastomas and astroblastomas) (IRR=0.48; 95% CI 0.26–0.88). Males experienced better overall survival in germ cell tumors, cysts and heterotopias (HR=0.68; 95% CI 0.49–0.94) compared to females, but were observed to have lower survival in all other histologies. Survival was lowest for males among patients with nerve sheath tumors (HR=2.32; 95% CI 1.31–4.12) and other neoplasms related to the meninges (e.g. chondrosarcomas and chordomas) (HR=2.23; 95% CI 1.24–3.99). Survival in meningioma patients was significantly higher in females (HR=1.49; 95% CI 1.25–1.77). Patients with glioblastomas, had a slightly worse survival outcome in males (HR=1.02, 95% CI 1.00–1.05).

CONCLUSION

Understanding the role of sex differences is critical for addressing sex based inequalities and needs to be taken into account in clinical paradigms.

EXTH-39. BENCH TO BEDSIDE NEURO-ONCOLOGY: ADVOCATING FOR A CLINICALLY RELEVANT STRATEGY

INTRODUCTION

Therapeutic successes in neuro-oncology have lagged dramatically behind exciting laboratory discoveries, and successful translation of promising laboratory findings into clinical practice is rare. We hypothesize that one important reason for this discordance is the use of very different paradigms for designing laboratory and clinical trials, and that utilizing clinically relevant procedures could improve laboratory study impact.

METHODS

We identified all pre-clinical neuro-oncology therapeutic trials published in four high-impact journals between 11/2018 and 4/2019, and assigned a level of evidence (LOE) to each study using the American Academy of Neurology evidence classification system. We also extracted details about statistical techniques, funding, and institutional setting.

RESULTS

Of the 26 articles identified, 85% had a LOE of IV (highest-I, lowest-IV) and 15% were class III. Factors contributing to the low LOE included lack of randomization (65%) or allocation concealment (100%), absence of masked assessment (96%), and no comparison of treatment and control groups with respect to prognostically relevant characteristics (92%). Effect size and confidence interval reporting, power calculation, correction for multiple hypothesis testing, and multivariate analysis were absent in 92%, 92%, 100%, 88%, and 92% of studies respectively. The “reverse” analysis (starting with successful human trials and analyzing existing pre-clinical support studies) showed significantly more high quality laboratory studies supporting “successful” compared to “unsuccessful” trials (p=0.04).

CONCLUSIONS

When rigorous human clinical trials criteria are applied to therapeutic laboratory studies, laboratory study LOE is uniformly poor. For laboratory research intended to inform clinical trial design and improve clinical outcomes, this situation dooms most basic science research to fail in the clinical arena. The steps required to solve this problem are challenging but addressable, and costs associated with the solution are dwarfed by the expense of early phase human trials which are destine to failure.

TMIC-02. JUNCTIONAL ADHESION MOLECULE-A (JAM-A) DEFICIENCY DRIVES SEX-SPECIFIC DIFFERENCES IN GLIOBLASTOMA PROGRESSION VIA DIFFERENTIAL MICROGLIA RESPONSES IN THE TUMOR MICROENVIRONMENT

Despite the male preponderance for developing glioblastoma (GBM) and better survival outcomes in females, current treatment paradigms do not account for biological sex as a biological or clinical variable. Sex-specific molecular alterations that drive tumor cell growth and therapy response have been documented, however, sex-specific extrinsic differences in the tumor microenvironment have yet to be identified. Based on well-established sex-specific gene signatures and functional differences in microglia, we interrogated influences of male and female microglia in driving GBM growth. Specifically, manipulation of JAM-A expression, a tight junction protein on microglia, was exploited as a paradigm for determining effects on in vivo syngeneic GBM mouse models. Male and female JAM-A KO mice that received orthotopic injection of syngeneic GBM cells presented differential overall survival distinct from their wildtype counterparts. Wild-type male mice phenocopied human males, presenting shorter overall survival than females, this trend was reversed in JAM-A KO mice. Compared to the other genotypes, female JAM-A KO mice presented the greatest number of phagocytic, tumor-promoting, activated microglia. RNA-sequencing of tumors from JAM-A KO and WT mice revealed that female JAM-A KO mice had increased expression of Ifi202b (interferon activated gene 202b), a member of the Activity-regulated Inhibitor of Death (AID) gene family that contributes to mitochondrial resistance to cellular stress. Ifi202b has a role in sex-specific inflammatory diseases, which is consistent with our observation. Female KO microglia had enhanced Ifi202b expression, along with the secretion of Ifi202b associated cytokines, including interleukin-6. Treatment of wild-type female microglia with a JAM-A function blocking antibody demonstrated an increase in Ifi202b levels, confirming direct regulation of Ifi202b expression by neutralizing JAM-A. While cell intrinsic, sex-specific differences have been reported in GBM, our findings demonstrate that differences in the GBM tumor microenvironment also drive sexually dimorphic tumor growth.

DRES-13. DUAL KINASE INHIBITION TO COMBAT EGFR-INHIBITOR RESISTANCE IN GLIOBLASTOMA

Glioblastoma (GBM) is an aggressive primary brain tumor with a poor survival rate. One of the most common molecular alterations seen in GBM is amplification and/or mutation of the Epidermal Growth Factor Receptor (EGFR), which has made it an attractive therapeutic target. However, several EGFR tyrosine kinase inhibitors have been tested clinically in GBM with minimal success. One reason for this lack of efficacy could be due to acute, adaptive resistance via alternative pathway activation. To investigate this mechanism of tumor resistance, we used RNA-seq and multiplex inhibitor bead/mass spectrometry (MIB-MS) to analyze the transcriptomes and kinomes of genetically engineered murine astrocytes with common GBM genotypes. We have previously shown that 38% of the expressed kinome varied among a panel of diverse nGEM astrocytes harboring Cdkn2a deletion (C) plus Pten deletion (CP), wild-type human EGFR (CE) or EGFRvIII (CEv3) overexpression or both EGFRvIII overexpression and Pten deletion (CEv3P). Although CE have a similar transcriptional profile to C cells at baseline, when treated with the EGFR inhibitor afatinib, CE respond more similarly to CEv3 cells. When cells containing endogenous murine EGFR (C and CP) are treated with afatinib, fewer than 0.5% of kinases showed differential expression. In cells with EGFR overexpression alone, more than 6% of kinases were differentially expressed upon afatinib treatment, including Ntrk3, Fgfr2 and 3, Lyn, Bmx, Epha2 and 5, Fn3k, a kinase involved in fructosamine processing, and Nrbp2, a kinase involved in regulation of apoptosis. This effect was blunted in cells lacking Pten in addition to having EGFRvIII (CEv3P), resulting in less than 2% of kinases being differentially expressed. The only kinase upregulated in all three EGFR-overexpressing cell types was Coq8a, which is involved in electron transport and response to DNA damage. Given this overlap in response, Coq8a could be a potential dual treatment target for GBM.

EXTH-43. DYNAMIC MULTI-OMICS ANALYSIS OF MLN4924-TREATED GLIOBLASTOMA CELLS REVEALS ABSENCE OF PTEN AS A CANDIDATE DRIVER OF NEDDYLATION INHIBITION RESISTANCE

Glioblastoma multiforme (GBM) is the most lethal primary brain tumor and calls for novel therapeutic development. The neddylation pathway may be a therapeutic target, as global over activation of neddylation has been found in GBM patients and has correlated with shorter patient survival. Analogous to ubiquitination in reaction scheme and enzyme classes used, neddylation is a post-translational modification that is essential to many protein regulation and biological processes. Although non-cullin NEDD8 substrates have been investigated in recent years, the most well-characterized substrates of neddylation are the cullin subunits of Cullin-RING ligases (CRLs), which act as a scaffold for substrates to be ubiquitinated and degraded. The neddylation inhibitor MLN4924 targets NEDD8 Activating Enzyme (NAE), an upstream activator of neddylation, and induces cell cycle arrest, apoptosis and senescence in cancer cells. In this work, we investigated the context of vulnerability to Pevonedistat (MLN4924) in GBM. Efficacy of MLN4924 in glioma cell models was evaluated by measuring cell viability (CellTiterGlo®), colony formation efficiency, and cell cycle progression (flow cytometry with propidium iodide staining). GB1 (IC50= 0.28 μM), LN18 (IC50 = 0.19 μM), and GBM43 (IC50= 0.45 μM) were established as sensitive and M059K (IC50 = 5.5 μM), SNU1105 (IC50 = 20.9 μM), and GBM39 (IC50= 10.3 μM) as non-sensitive cell lines based on IC50 values. To discover genomic and/or proteomic markers of differential response, we collected RNA and protein from sensitive and non-sensitive cell lines after 0, 2, 8 and 24 hour treatment with MLN4924 for RNA sequencing and mass spectroscopy analyses. Multi-OMICS data point towards the absence of functional PTEN as a driver of MLN4924 resistance. Uncovering the mechanism underlying GBM’s vulnerability to MLN4924 will expand the knowledge of neddylation’s role in cancer and may arm physicians with an understanding of whether a GBM patient would respond to MLN4924 treatment or not.

TMIC-53. IMPACT OF HFE EXPRESSION AND SEX ON THE TUMOR IMMUNE MICROENVIRONMENT IN GLIOBLASTOMA

Glioblastoma (GBM) is the most common primary malignant brain tumor in adults. Despite aggressive treatment, the median survival for patients with GBM remains approximately 1 year. Recent evidence demonstrates GBM is a sexually dimorphic disease, and that females have greater overall survival. We previously showed expression levels of HFE, an iron-regulating gene, significantly impacts survival in GBM. Moreover, this effect is sex-specific: females with low HFE expressing tumors have significantly longer survival than males with similar HFE expression. Moreover, HFE expression and survival appeared to correlate with markers of immune activity within the tumors. To further explore the impact of HFE expression on the immune infiltration of GBM, we utilized available TCGA GBM data for analysis by EstimateMe and DeconvuluteMe. These platforms utilize expression data to estimate the magnitude and cellular composition of immune cell infiltrates in tumors, respectively. We sought to correlate these computational measures with HFE expression levels and survival. We confirmed that immune and stromal cell infiltration of tumors is negatively correlated to survival. Consistent with our findings that high HFE levels have a negative impact on survival, high HFE expressing tumors possessed higher levels of both immune and stromal cell infiltration. Additionally, sex of the patient, but not MGMT methylation status, was significantly correlated with immune and stromal infiltration, with males displaying higher levels of infiltration. High HFE tumors were composed of greater numbers of all immune cell types than low HFE tumors. In both HFE groups, monocytes and macrophages comprised the greatest fraction of cells. At low HFE and positive MGMT methylation (where we previously noted the greatest survival differences), we find females have lower monocyte and M2 macrophage markers compared to males with similar features. Collectively, these data suggest that an HFE-immune cell infiltrate axis contributes to sex differences in GBM survival.

CBMT-23. NON-CANONICAL FUNCTIONS OF TERT IN GLIOBLASTOMA

Telomerase is an enzyme with a catalytic subunit, telomerase reverse transcriptase (TERT), that is in charge of telomere elongation in the nucleus. Promotor region of TERT is commonly mutated across cancers, especially in glioblastoma (GBM) with over 80% frequency. In the absence of any effective molecular targeting therapy for GBM, elucidating oncogenic signaling of TERT could open new avenues in GBM treatment. Canonically, mutations of TERT, which result in TERT upregulation, maintain telomere length in the nucleus and promote indefinite proliferation of cancer cells. However, a non-canonical function of TERT in the mitochondria has recently been suggested. We screened GBM cell models against a novel small molecule inhibitor (RG1534, Reglagene Inc.) that interferes with the functionality of a mutated hTERT promoter. RG1534 selectively suppresses glioma cell viability without affecting non-transformed normal human astrocytes. More interestingly, RG1534 treatment leads to rapid apoptosis induction in glioma cell lines that does not correlate with the time course of the telomere shortening effect. We further validated this rapid apoptosis behavior in glioma cell lines using siRNA and CRISPR/Cas-9 mediated hTERT knockdown. We also measured the protein expression of TERT in subcellular fractions of glioma cell lines and demonstrated the presence of higher TERT expression in mitochondrial extract compared to the nucleus. Finally, using MitoSOX dye we assessed ROS generation in glioma cells in response to an oxidant with or without TERT expresssion. In summary, our results demonstrate that non-canonical functions of TERT may play critical role in glioma pathobiology and require more detail investigation.

COMP-06. PROBING GLIOBLASTOMA AND ITS MICROENVIRONMENT USING SINGLE-NUCLEUS AND SINGLE-CELL SEQUENCING

Single-cell (scSeq) and single-nucleus sequencing (snSeq) are powerful tools to investigate cancer genomics at single cell resolution. Multiple studies have recently illuminated intratumoral heterogeneity in glioblastoma, however, the majority focused on molecular complexity of tumor cells, without taking into account unexplored host cell types that contribute to the microenvironment around tumor. To address the glioblastoma microenvironment composition and potential tumor-host interactions, we performed deep coverage sequencing of freshly resected primary GBM patient tissue without implementing any tumor enrichment strategies. The sequencing resulted in 902 cells and 1186 nuclei, respectively, passing quality control and with low mitochondrial gene percentage. We customized reference transcriptome by listing gene transcript loci as exons to take into account immature RNA, which greatly improved the alignment rate for single-nucleus data. We applied Cell Ranger pipelines (Version 3.0.2) and Seurat package (Version 2.3.1) and discovered 10 clusters in both scSEQ and snSEQ. Pathway analysis of each cluster signature in scSeq data along with known GBM microenvironment cell signatures revealed glioma tumor population along with surrounding microglia/macrophages, astrocytes, pericytes, oligodendrocytes, T cells and endothelial cells. The analysis of snSEQ was able to capture the majority of cell types from patient tissues (tumor and microenvironment cells), but interestingly presented different cell type composition in microenvironment cell types such as microglia/marcophages. Integrating single-cell and single-nucleus transcriptomic data using canonical correlation analysis facilitated a comparison of snSEQ and scSEQ, contrasting depiction for certain cell types (e.g. NKX6-2 gene in Oligodendrocytes). Differential analysis of pathways between tumor and microenvironment cells unveiled potentially rewired pathways such as double strand break repair pathway. Our results demonstrate the cellular diversity of brain tumor microenvironment and lay a foundation to further investigate the individual tumor and host cell transcriptomes that are influenced not only by their cell identity but also by their interaction with surrounding microenvironment.

Abstract 3757: Probing glioblastoma and its microenvironment at single cell resolution

Single-cell sequencing (scSeq) is a powerful tool to investigate cancer genomics at single cell resolution. Multiple studies have recently illuminated intratumoral heterogeneity in glioblastoma, however, the majority focused on molecular complexity of tumor cells, without taking into account unexplored host cell types that contribute to the microenvironment around GBM tumor. To address the glioblastoma microenvironment composition and potential tumor-host interactions, we performed deep coverage (176k average reads per cell) scSeq of freshly resected primary GBM patient tissue without implementing any tumor cell enrichment strategies. scSeq libraries for 902 cells were prepared using 10X Gemcode platform and sequenced on Illumina NextSeq 500. This run was of high quality with 2,663 median genes per cell and low mitochondrial gene percentage (median < 5%). We used Cell Ranger analysis pipelines and Seurat packages to classify individual cells into 10 clusters and visualize them using t-SNE two-dimensional projections. We then identified the signature gene set for each cluster, relative to all other cells. Pathway analysis of each cluster signature along with known GBM microenvironment cell signatures revealed glioma tumor population along with surrounding microglia/marcophages, astrocytes, pericytes, oligodendrocytes, T cells and endothelial cells. Cell type markers identified by single cell transcriptomics were validated by IHC analysis. Microenvironmental composition and single cell signature will be confirmed through single nuclei sequencing of preserved (Frozen) tumor sample. Our results demonstrate the cellular diversity of brain tumor microenvironment and lay a foundation to further investigate the individual tumor and host cell transcriptomes that are influenced not only by their cell identity but also by their interaction with surrounding microenvironment.

Citation Format: Sen Peng, Sanhita Rath, Saumya Bollam, Jenny Eschbacher, Shwetal Mehta, Nader Sanai, Michael Berens, Seungchan Kim, Harshil Dhruv. Probing glioblastoma and its microenvironment at single cell resolution [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3757.

Abstract 2189: Evaluating efficacy of repurposed drugs in treatment of glioblastoma

Glioblastoma (GB) is the most aggressive adult brain tumor with a devastating median survival time of about fourteen months post-surgery and standard of care therapy with radiation and temozolomide. The low incidence of GB is a dis-incentive to develop novel therapies. To overcome that obstacle, we investigated the efficacy of repurposing four FDA approved drugs known to cross the BBB, minocycline, propranolol, chlorpromazine, and metformin, to inhibit signaling and metabolism in GB cells. Minocycline is a tetracycline class broad spectrum antibiotic commonly used to treat severe acne and other skin infections. Propranolol is a beta blocker type heart medication primarily used to treat high blood pressure and irregular heartbeat. Chlorpromazine is a phenothiazine antipsychotic usually used for schizophrenia. Metformin is the most widely used first-line oral treatment for type-2 diabetes. Based on a literature survey, minocycline is expected to prevent the phosphorylation of STAT3, a transcription factor downstream of EGFR; propranolol is expected to disrupt EGFR trafficking; chlorpromazine is expected to target the mTOR/AKT pathway; metformin is believed to exploit vulnerabilities in cancer cell metabolism. Efficacy of minocycline in inhibiting EGFR-driven STAT3 activation was investigated using western blot analysis. Our results demonstrate that Minocycline effectively inhibits activation of EGFR-driven STAT3 in U373 glioma cells at 100μM. The ability of chlorpromazine to inhibit the mTOR/AKT pathway was similarly tested via western blot, which showed inhibition of phosphorylated Akt and S6 at 10μM. Efficacy of propranolol in perturbing EGFR trafficking was evaluated using flow cytometry and immunofluorescence, which depicted altered membrane-associated EGFR abundance. Finally, concentration-dependent inhibition of colony formation was tested for all four drugs. Propranolol and minocycline showed potential stimulatory effects at 10μM, but all drugs inhibited cell growth at 50μM and higher. Efficacy of these drugs in treatment of GB is being further evaluated using in vitro neurosphere cultures from patients identified as having the cellular vulnerabilities potentially targeted by these drugs. Successful completion of this project will lead to in vivo efficacy testing of these four drugs in orthotopic GB PDX models.

Citation Format: Tristan Neal, Nanyun Tang, George Reid, Sara Byron, Harshil Dhruv, Michael Berens. Evaluating efficacy of repurposed drugs in treatment of glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2189.

Abstract 2745: Tumor microenvironment and host genetics impact glioma progression in a Collaborative Cross-based orthotopic allograft model

Gliomas are diffusely invasive brain tumors with fatal outcomes and few effective treatments. Precision medicine focuses on targeting the genetics of individual tumors, but not host genetics, despite studies that have linked germline polymorphisms with glioma risk. Accordingly, glioma survival studies in mice utilize genetically variable tumors on identical host genetic backgrounds, which fails to differentiate between cancer cell-autonomous (CCA) and tumor microenvironment (TME) effects on glioma progression and host survival. The Collaborative Cross (CC) is a panel of genetically diverse mouse strains derived from both wild- and traditional inbred laboratory strains that facilitates high-resolution genetic mapping in models of complex disease. Here, we implement a novel platform to discover genetic modifiers of both CCA and TME phenotypes using genetically defined orthotopic murine allograft gliomas and CC hosts. We stereotactically injected Nf1;Trp53-/-oligodendrocyte progenitor-derived mouse tumor cells into syngeneic C57BL/6 control mice and 14 different CC strains. Seven strains survived significantly longer than controls (P<0.05), suggesting slower tumor growth (Gs, growth slow). The remaining 7 strains survived similarly to controls, suggesting fast growth (Gf, growth fast). Variable tumor growth in CC mice suggests that genetic background influences molecular processes in the TME that inhibit or potentiate tumor growth, respectively. To identify candidate genes, we performed RNA sequencing on 36 tumors from 3 Gf strains, 4 Gs strains, and controls. 134 genes were differentially expressed among Gf, Gs, and control tumors (P<0.05). Hierarchical clustering on these genes revealed that Gs strains clustered separately from Gf and controls. Gene ontology analysis using GOrilla showed 30 enriched processes, (FDR q<0.001), all of which were involved in immune responses or extracellular matrix biology. These results suggest that Gs strains activate immune and TME processes that slow tumor growth. Quantitative trait locus (QTL) analyses of host genetics and tumor data are pending and will facilitate identification of genetic variants that influence TME effects on tumor progression.

Citation Format: Kasey Skinner, Martin Ferris, Ryan Bash, Abigail Shelton, Erin Smithberger, Steve Angus, Brian Golitz, Noah Sciaky, Jeremy Simon, Jason Stein, Glenn Matsushima, Quinn Ostrom, Lindsay Stetson, Jill Barnholtz-Sloan, Harshil Dhruv, Michael Berens, Fernando Pardo Manuel de Villena, C. Ryan Miller. Tumor microenvironment and host genetics impact glioma progression in a Collaborative Cross-based orthotopic allograft model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2745.

Abstract 5226: Non-canonical functions of TERT in glioblastoma pathobiology

Mutations in the promotor region of TERT are the most common non-coding mutations across cancers, especially in glioblastoma (GBM), which has a TERT mutation frequency rate over 80%. In the absence of any effective molecular targeting therapy for GBM, elucidating oncogenic signaling of TERT could open new avenues in GBM treatment. Canonically, mutations of TERT, which result in TERT upregulation, maintain telomere length in the nucleus and promote indefinite proliferation of cancer cells. However, a non-canonical function of TERT in the mitochondria has recently been suggested. We screened GBM cell models against a novel small molecule inhibitor (RG1534, Reglagene Inc.) that interferes with the functionality of a mutated hTERT promoter. RG1534 selectively suppresses glioma cell viability without affecting non-transformed normal human astrocytes. More interestingly, RG1534 treatment leads to rapid apoptosis induction in glioma cell lines that does not correlate with the time course of the telomere shortening effect. We further validated this rapid apoptosis behavior in glioma cell lines using siRNA and CRISPR/Cas-9 mediated hTERT knockdown. We also measured the protein expression of TERT in subcellular fractions of glioma cell lines and demonstrated the presence of higher TERT expression in mitochondrial extract compared to the nuclear extract. Finally, using MitoSOX dye we demonstrated significant increase of ROS generation in glioma cells treated with RG1534 as compared to control. In summary, our results demonstrate that non-canonical functions of TERT may play critical role in glioma pathobiology and require more detail investigation.

Citation Format: Shayesteh R. Ferdosi, Saumya Bollam, Sen Peng, Brett Taylor, Vijay Gokhale, Laurence Hurley, Michael Berens, Harshil Dhruv. Non-canonical functions of TERT in glioblastoma pathobiology [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 5226.

Abstract 248: Dynamic multi-OMICS analysis of glioblastoma cells reveals context of vulnerability to neddylation inhibition by pevonedistat

As the most lethal primary brain tumor, glioblastoma multiforme (GBM) calls for novel therapeutic development. Global over activation of neddylation (a post-translational modification) has recently been found in GBM patients and has correlated with shorter patient survival. Significant accumulation of neddylation in recurrent GBM tissues indicates its importance in tumorigenesis and tumor progression. Analogous to the ubiquitination pathway, neddylation is essential to many protein regulation and biological processes. Although most well-characterized substrates of neddylation are the cullin subunits of Cullin-RING ligases (CRLs), non-cullin NEDD8 substrates have been investigated in recent years. Neddylation and subsequent degradation of PARC, p53, MDM2 and EGFR exemplify the broader functional role of neddylation. The neddylation inhibitor MLN4924 targets NEDD8 Activating Enzyme (NAE), an upstream activator of neddylation, and, as a result, induces cell cycle arrest, apoptosis and senescence in cancer cells. In this work, we investigated the context of vulnerability to Pevonedistat (MLN4924) in GBM by comparing the dynamic response of sensitive and non-sensitive cells using transcriptomics and proteomics profiling, using long-established and patient derived glioma cell lines. Efficacy of MLN4924 in glioma cell models was evaluated by measuring cell viability (CellTiterGlo®), colony formation efficiency, and cell cycle progression (flow cytometry with propidium iodide staining). GB1 (IC50= 0.28 μM), LN18 (IC50 = 0.19 μM), and GBM43 (IC50= 0.45 μM) were established as sensitive and M059K (IC50= 5.5 μM), SNU1105 (IC50 = 20.9 μM), and GBM39 (IC50= 10.3 μM) as non-sensitive cell lines based on IC50 values. Western blot analysis of known cell cycle regulatory pathways and DNA damage response pathway did not show significant dynamic differences between sensitive and non-sensitive glioma cell models. To discover genomic and/or proteomic markers of differential response we collected RNA and protein for LN18 (sensitive) and SNU1105 (Non-sensitive) cells after 0, 2, 8 and 24 h treatment with MLN4924 at 100 nM and 500 nM concentration for transcriptomics and proteomics analysis. RNA sequencing was utilized for dynamic transcriptomic analysis. Cell lysates were processed using bottom-up proteomics workflow and the data was acquired on a Thermo Scientific Orbitrap Fusion Lumos Tribrid mass spectrometer. Proteins were identified by querying spectral data against canonical and RNA-Seq predicted proteins and differential analysis was carried out to identify candidate determinats of vulnerability. An understanding of determinants of vulnerability to MLN4924 will expand knowledge of neddylation’s role in cancer and may point to signatures of GBM patients most likely to respond to this targeted intervention.

Citation Format: Shayesteh R. Ferdosi, Brett Taylor, Nanyun Tang, Rita Bybee, Sen Peng, Victoria David-Dirgo, Krystine Garcia-Mansfield, Ritin Sharma, Patrick Pirrotte, Michael Berens, Harshil Dhruv. Dynamic multi-OMICS analysis of glioblastoma cells reveals context of vulnerability to neddylation inhibition by pevonedistat [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 248.