Brain Tumor Subtypes Research Articles

Vision transformer-based weakly supervised histopathological image analysis of primary brain tumors

Diagnosis of primary brain tumors relies heavily on histopathology. Although various computational pathology methods have been developed for automated diagnosis of primary brain tumors, they usually require neuropathologists' annotation of region of interests or selection of image patches on whole-slide images (WSI). We developed an end-to-end Vision Transformer (ViT)- based deep learning architecture for brain tumor WSI analysis, yielding a highly interpretable deep-learning model, ViT-WSI. Based on the principle of weakly supervised machine learning, ViT-WSI accomplishes the task of major primary brain tumor type and subtype classification. Using a systematic gradient-based attribution analysis procedure, ViT-WSI can discover diagnostic histopathological features for primary brain tumors. Furthermore, we demonstrated that ViT-WSI has high predictive power of inferring the status of three diagnostic glioma molecular markers, IDH1 mutation, p53 mutation, and MGMT methylation, directly from H&E-stained histopathological images, with patient level AUC scores of 0.960, 0.874, and 0.845, respectively.

NIMG-102. RAPNO-DEFINED SEGMENTATION AND VOLUMETRIC ASSESSMENT OF PEDIATRIC BRAIN TUMORS ON MULTI-PARAMETRIC MRI SCANS USING DEEP LEARNING; A ROBUST TOOL WITH POTENTIAL APPLICATION IN TUMOR RESPONSE ASSESSMENT

Abstract Volumetric measurements of whole tumor and its components on MRI scans, facilitated by automatic segmentation tools, are essential to reduce inter-observer variability in monitoring tumor progression and response assessment for pediatric brain tumors. Here, we present a fully automatic segmentation model based on deep learning that reliably delineates the tumor components recommended by the Response Assessment in Pediatric Neuro-Oncology (RAPNO) working group for evaluation of treatment response. Multi-parametric MRI (mpMRI) scans (T1-pre, T1-post, T2, and T2-FLAIR), acquired on multiple MRI scanners with different field strengths and vendors, for a cohort of 218 pediatric patients with a variety of histologically confirmed brain tumor subtypes were collected. The mpMRI scans were co-registered and manually segmented by experienced neuroradiologists in consensus to identify the tumor subregions including the enhancing tumor (ET), non-enhancing tumor (NET), cystic components (CC), and peritumoral edema (ED) regions. A convolutional neural network model based on DeepMedic architecture was trained using mpMRI scans as the inputs for segmentation of the whole tumor and subregions. The trained model showed excellent performance in segmentation of the whole tumor, as suggested by median dice of 0.90/0.85 for validation (n = 44)/independent test (n = 22) sets. ET and non-enhancing components (union of NET, CC, and ED) were segmented with median dice scores of 0.78/0.84 and 0.76/0.74 for validation/test sets, respectively. The automated and manual segmentations demonstrated strong agreement in estimating VASARI (Visually AcceSAble Rembrandt Images) MRI features with Pearson’s correlation coefficient R > 0.75 (p < 0.0001) for ET, NET, CC, and ED components. Our proposed automated segmentation method developed based on MRI scans acquired with different protocols, equipment, and from a variety of brain tumor subtypes, shows potential application for reliable and generalizable volumetric measurements which can be used for treatment response assessment in clinical trials.

MODL-23. MULTISCALE MODELS OF GLIOMA PROGRESSION: INSIGHTS ABOUT VASCULATURE AND THERAPY EFFECTS

Abstract Gliomas are the most prevalent, aggressive, and invasive subtype of primary brain tumors, characterized by fast cell growth, strong invasion capability, and well-developed tumor vasculature. Although the research advances and the clinical trials testing the efficacy of novel combined therapies have allowed significant progress in the comprehension and treatment of gliomas, these tumors are characterized by a poor prognosis, and recurrence remains the main cause of mortality. The growth and migration of tumor cells inside the brain is a highly complex phenomenon, influenced by a multitude of intrinsic and extrinsic factors at different spatial and temporal scales. The high complexity of this invasion process remains a challenge to face and several important questions are still unanswered. In this context, mathematical models emerge as powerful tools to face these challenges. Here, we propose a multiscale approach for the description of glioma progression under the influence of vascularization. Precisely, we firstly focus on the effects of (hypoxia-driven) acidity and phenotypic heterogeneity on tumor progression. Then, we extend the setting to include vascular endothelial growth factor evolution and compare various therapeutic approaches. Starting from a microscopic description of the interactions between specific cell membrane receptors and the extracellular microenvironment, we define two different coupled systems for endothelial and tumor cell evolution. The analysis of these systems allows us to assess tumor growth and spread on the basis of DTI and glioma patient data. In particular, we perform several data-driven numerical tests aimed at assessing the role of vasculature and phenotypic heterogeneity on tumor progression and comparing the effect of different combinations of radio-, chemo-, and anti-angiogenic therapy on it.

TMET-37. SODIUM-DEPENDENT GLUCOSE TRANSPORTER 2 (SGLT2) IS OVEREXPRESSED IN THE MAJORITY OF ATRTS AND A SUBSET OF PEDIATRIC PATIENTS WITH HIGH-GRADE GLIOMA; A POTENTIAL IMAGING AND THERAPEUTIC TARGET

Abstract Sodium glucose cotransporters (SGLT) are a class of glucose transporters independent of the conventional glucose transporters (GLUT). Like the GLUT, SGLT2 plays an important role in glucose uptake in many solid cancers. Importantly, SGLT2 inhibitors are a new class of anti-diabetic drugs and have recently been investigated as a potential anti-cancer therapy in a variety of solid cancers. In fact, a recent study in adult high-grade gliomas showed that an SGLT-specific PET imaging probe ([18F] Me-4FDG) is sensitive for visualization of high-grade glioma (HGG) with superior tumor-normal brain uptake relative to 18F-FDG-PET. We aimed to investigate the expression of SGLT2 within pediatric brain tumor subtypes using RNA sequencing data from PedcBioPortal and immunohistochemistry of tissue microarrays (TMAs). Using the RNA sequencing data from 603 patients with pediatric brain tumors, we found that HGGs have the highest SGLT2 expression (Z score=0.13± 1.03) and were significantly higher compared to low-grade gliomas (Z score= -0.12, p=0.049), medulloblastomas (Z score=-0.28, p=0.001), and ependymomas (Z score=-0.30, p=0.002). Atypical teratoid rhabdoid tumors (ATRT) also had significantly higher Z score (0.10±0.67) compared to medulloblastomas (p=0.038), and ependymomas (p=0.049). There was no significant difference between HGGs and ATRTs in SGLT2 expression (p=0.899). Analysis of 4 TMAs including HGGs, medulloblastomas, ependymomas, and ATRTs after SGLT2 immunohistochemistry demonstrated strong membranous staining in a small subset and moderate-weak staining in the majority of HGGs. The majority of ATRTs were positive. The majority of medulloblastomas and ependymomas were negative, noting that the membranous nature of the staining may be a limiting factor in evaluation of tumors with less cytoplasm. In conclusion, our work demonstrates that SGLT2 has heterogeneous expression in pediatric brain tumors and is overexpressed in the majority of ATRTs, and a subset of pediatric HGGs, and may represent a compelling paradigm that integrates metabolic imaging and therapy of these tumors.

TMET-29. THE GLUTAMINE TRANSPORTER SLC1A5 IS ASSOCIATED WITH REWIRING OF AMINO ACID, LIPID, AND IMMUNE PATHWAYS AND TUMOR PROGNOSIS IN PEDIATRIC BRAIN CANCERS

Abstract Glutamine transporters play an important role in supporting increased tumor nutritional demands, often through overexpression of the solute carrier (SLC) family of membrane transporters. We aimed to understand the relationship of SLC transporter expression with pediatric brain tumor subtypes, lipid metabolism and their potential prognostic significance using data from the Pediatric Brain Tumor Atlas (PBTA). Using the expression of amino acid transporter genes, we found that elevated expression of glutamine transporters (SLC1A5, SLC7A5, SLC7A11, SLC38A5, SLC38A3) predicted shorter progression-free survival (PFS) in low-grade gliomas (LGGs) and poorer overall survival in pediatric ependymomas, high-grade gliomas (HGGs), and medulloblastomas. We focused specifically on SLC1A5 given the availability of imaging probes (18 F-Fluoroglutamine and 18F-Fluciclovine) for the corresponding amino acid transporter (ASCT2). Kaplan-Meier analysis found that higher expression of SLC1A5 was associated with shorter OS in ependymoma and medulloblastoma (p = 0.032 and p = 9.8e-4) and shorter PFS in LGG (p = 0.022). We found a universally higher expression of ATP citrate lyase (ACLY) and Acetyl-CoA carboxylase 1 (ACC) relative to normal brain tissue across all histologies. BRAF driven low grade gliomas and SHH-driven medulloblastomas showed differential expression of ACLY compared to other histologies. In addition, Acyl-CoA synthetase short chain family member 1 and 2 (ACSS1 and 2) were universally downregulated in pediatric gliomas, indicating that pediatric CNS tumors seem to rely on ACLY over ACSS for lipid synthesis. Gene set analysis showed higher expression and network rewiring of amino acid, lipid, and immune pathways in SLC1A5-high expressing clusters. SLC1A5 correlated negatively with pathways associated with lipid metabolic breakdown/degradation and correlated positively with pathways associated with lipid biosynthesis. In conclusion, our work demonstrates that glutamine transporters, particularly SLC1A5, represent compelling targets in pediatric brain tumors and can be exploited with theranostic approaches and amino-acid PET imaging.

TMET-36. ACID CERAMIDASE INHIBITION EXPLOITS SPHINGOLIPID VULNERABILITIES IN IDH MUTANT GLIOMAS

Abstract The presence of the IDH mutation in gliomas is a major classifier of brain tumor subtypes and has several important implications for cancer growth. Our recent work uncovered that IDH-mutant tumors are susceptible to increased apoptosis via alterations of the sphingolipid pathway due to their excess production of pro-apoptotic ceramides over pro-proliferative sphingosine 1-phosphate (S1P). To that end, we proposed that this rheostat can be modulated to induce cell death in IDHmut tumors by targeting acid ceramidase, a critical sphingolipid enzyme in gliomas. We hypothesize that pharmacological inhibition of acid ceramidase will increase ceramide levels and therefore induce apoptosis in IDHmutgliomas. Using a preliminary drug screen, we have identified a group of haloacetate C2-ceramide derivatives known as SOBRACs that potently inhibit acid ceramidase. We selected five candidate compounds from this family and assessed the effectiveness of each drug in 3 I IDHmut (BT142, TS603, & U251mut) and 3 IDHmut (GSC923, GSC827, U251wt) patient-derived glioma cell lines, as well as non-immortalized normal human astrocytes, using the CCK8 cell viability assay. Our results indicate that the SOBRAC drugs are nearly 10 times more potent in IDH-mutant tumors compared to IDHmut cell lines. Additionally, the SOBRAC drugs are more effective than other known acid ceramidase inhibitors, making them attractive as potential novel therapeutics. To date, azide-SOBRAC is the most potent drug in the family, with EC50 value of 300 nM in BT142 cells (IDHmutmut

Metabolic biomarkers of radiotherapy response in plasma and tissue of an IDH1 mutant astrocytoma mouse model.

Astrocytomas are the most common subtype of brain tumors and no curative treatment exist. Longitudinal assessment of patients, usually via Magnetic Resonance Imaging (MRI), is crucial since tumor progression may occur earlier than clinical progression. MRI usually provides a means for monitoring the disease, but it only informs about the structural changes of the tumor, while molecular changes can occur as a treatment response without any MRI-visible change. Radiotherapy (RT) is routinely performed following surgery as part of the standard of care in astrocytomas, that can also include chemotherapy involving temozolomide. Monitoring the response to RT is a key factor for the management of patients. Herein, we provide plasma and tissue metabolic biomarkers of treatment response in a mouse model of astrocytoma that was subjected to radiotherapy. Plasma metabolic profiles acquired over time by Liquid Chromatography Mass Spectrometry (LC/MS) were subjected to multivariate empirical Bayes time-series analysis (MEBA) and Receiver Operating Characteristic (ROC) assessment including Random Forest as the classification strategy. These analyses revealed a variation of the plasma metabolome in those mice that underwent radiotherapy compared to controls; specifically, fumarate was the best discriminatory feature. Additionally, Nuclear Magnetic Resonance (NMR)-based 13C-tracing experiments were performed at end-point utilizing [U-13C]-Glutamine to investigate its fate in the tumor and contralateral tissues. Irradiated mice displayed lower levels of glycolytic metabolites (e.g. phosphoenolpyruvate) in tumor tissue, and a higher flux of glutamine towards succinate was observed in the radiation cohort. The plasma biomarkers provided herein could be validated in the clinic, thereby improving the assessment of brain tumor patients throughout radiotherapy. Moreover, the metabolic rewiring associated to radiotherapy in tumor tissue could lead to potential metabolic imaging approaches for monitoring treatment using blood draws.

Bayesian dynamic profiling and optimization of important ranked energy from gray level co-occurrence (GLCM) features for empirical analysis of brain MRI

Accurate classification of brain tumor subtypes is important for prognosis and treatment. Researchers are developing tools based on static and dynamic feature extraction and applying machine learning and deep learning. However, static feature requires further analysis to compute the relevance, strength, and types of association. Recently Bayesian inference approach gains attraction for deeper analysis of static (hand-crafted) features to unfold hidden dynamics and relationships among features. We computed the gray level co-occurrence (GLCM) features from brain tumor meningioma and pituitary MRIs and then ranked based on entropy methods. The highly ranked Energy feature was chosen as our target variable for further empirical analysis of dynamic profiling and optimization to unfold the nonlinear intrinsic dynamics of GLCM features extracted from brain MRIs. The proposed method further unfolds the dynamics and to detailed analysis of computed features based on GLCM features for better understanding of the hidden dynamics for proper diagnosis and prognosis of tumor types leading to brain stroke.

Multi-Magnification Image Search in Digital Pathology.

This paper investigates the effect of magnification on content-based image search in digital pathology archives and proposes to use multi-magnification image representation. Image search in large archives of digital pathology slides provides researchers and medical professionals with an opportunity to match records of current and past patients and learn from evidently diagnosed and treated cases. When working with microscopes, pathologists switch between different magnification levels while examining tissue specimens to find and evaluate various morphological features. Inspired by the conventional pathology workflow, we have investigated several magnification levels in digital pathology and their combinations to minimize the gap between AI-enabled image search methods and clinical settings. The proposed searching framework does not rely on any regional annotation and potentially applies to millions of unlabelled (raw) whole slide images. This paper suggests two approaches for combining magnification levels and compares their performance. The first approach obtains a single-vector deep feature representation for a digital slide, whereas the second approach works with a multi-vector deep feature representation. We report the search results of 20×, 10×, and 5× magnifications and their combinations on a subset of The Cancer Genome Atlas (TCGA) repository. The experiments verify that cell-level information at the highest magnification is essential for searching for diagnostic purposes. In contrast, low-magnification information may improve this assessment depending on the tumor type. Our multi-magnification approach achieved up to 11% F1-score improvement in searching among the urinary tract and brain tumor subtypes compared to the single-magnification image search.

Sox9 directs divergent epigenomic states in brain tumor subtypes

Epigenetic dysregulation is a universal feature of cancer that results in altered patterns of gene expression that drive malignancy. Brain tumors exhibit subtype-specific epigenetic alterations; however, the molecular mechanisms responsible for these diverse epigenetic states remain unclear. Here, we show that the developmental transcription factor Sox9 differentially regulates epigenomic states in high-grade glioma (HGG) and ependymoma (EPN). Using our autochthonous mouse models, we found that Sox9 suppresses HGG growth and expands associated H3K27ac states, while promoting ZFTA-RELA (ZRFUS) EPN growth and diminishing H3K27ac states. These contrasting roles for Sox9 correspond with protein interactions with histone deacetylating complexes in HGG and an association with the ZRFUS oncofusion in EPN. Mechanistic studies revealed extensive Sox9 and ZRFUS promoter co-occupancy, indicating functional synergy in promoting EPN tumorigenesis. Together, our studies demonstrate how epigenomic states are differentially regulated in distinct subtypes of brain tumors, while revealing divergent roles for Sox9 in HGG and EPN tumorigenesis.

Abstract 6049: Identifying shared features of low-H3K27me3 pediatric brain tumors

Abstract Brain tumors have the worse survival rates of all childhood malignancies, and a better understanding of their biology is needed for therapeutic development. Two especially devastating subtypes of pediatric brain tumors—diffuse midline gliomas, H3K27M-mutant (DMGs) and posterior fossa ependymomas, group A (PFAs) are both characterized by a hallmark of global reduction of histone 3 lysine 27 trimethylation, a mark associated with transcriptional repression. DMGs primarily harbor mutations of histone 3 (H3K27M) that inhibit the function of the H3 methyltransferase enhancer of zeste homolog 2 (EZH2)-containing polycomb repressive complex (PRC2). Most PFAs overexpress EZH inhibitory protein (EZHIP), resulting in a similar effect on PRC2 activity. Intriguingly, a small subset of DMGs found to be lacking H3K27M mutations express EZHIP, and a fraction of histological PFAs lacking EZHIP expression harbor H3K27M mutations. Given the highly similar putative driving alterations and epigenetic features in these two tumor types, we undertook a systematic evaluation of the clinical and molecular features of each tumor class to elucidate additional shared features and potential targetable vulnerabilities. We performed detailed analyses of genomic aberrations, gene expression, and epigenomic landscapes to identify key similarities and differences in tumor biology. From preliminary studies, our findings demonstrate similar H3K27me3 landscapes with convergence of residual tri-methylated loci between PFA and DMG tumors. We additionally found shared recurrent copy number gain of the long arm of chromosome 1 (1q) and identified several 1q genes whose expression correlated with survival differences in cohorts of both tumors. Together, these findings better define the commonalities and differences of these highly aggressive, low-H3K27me3 pediatric brain tumors and will provide a framework for understanding which therapeutic strategies may translate from one tumor to the other. Citation Format: Matthew Pun, Drew Pratt, Sriram Venneti. Identifying shared features of low-H3K27me3 pediatric 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 6049.

Abstract 1430: N-cadherin is a driver of adaptive radioresistance in malignant brain tumors

Abstract Many subtypes of brain tumors are highly malignant and resistant to chemo- and radio- therapy. Tumor cells can shift their phenotype in response to treatments, the so-called adaptive resistance. Adaptive resistance mechanisms in malignant brain tumors are still poorly understood, and effective treatments have not yet been developed. To unveil such mechanisms, we have developed unique new experimental models to identify the adaptive resistance mechanisms to fractionated radiation in malignant brain tumors. We performed repeated irradiation (2-5Gy every 3-4 days, 3-6 weeks) on 6 human Glioma stem cells (GSCs), 2 mouse GSCs and 4 medulloblastomas (MB) cells in vitro and examined how tumor cells adapt to repeated irradiation. Brain tumor cells demonstrated dynamic adaptation to fractionated irradiation. They rapidly altered cell proliferation, intercellular adhesion, and stemness and acquired strong radioresistance. To identify genes responsible for radio-resistance, we performed RNA-seq analysis and CRISPR library screening using primary and radioresistant cells. We found that N-cadherin was upregulated in the majority of radioresistant GSCs. Stably transfecting N-cadherin in parental GSC rendered them radioresistant, reduced their proliferation, and increased their stemness and intercellular adhesive properties. Conversely, radioresistant GSCs lost their acquired phenotypes upon CRISPR/Cas9-mediated knockout of N-cadherin. Mechanistically, elevated N-cadherin expression resulted in the accumulation of b-catenin at the cell surface, which suppressed Wnt/b-catenin proliferative signaling, and reduced neural differentiation. Moreover, N-cadherin increased Clusterin secretion, which protected GSCs against apoptosis after radiation treatment. We also demonstrated that N-cadherin upregulation was induced by radiation-induced IGF1 secretion, which caused an EMT-like phenotype change in GSCs. The N-cadherin-mediated radioresistance phenotype could be reverted with picropodophyllin (PPP), a clinically applicable blood-brain-barrier permeable IGF1 receptor inhibitor. Adjuvant PPP combined with irradiation significantly extended the survival of orthotopically xenografted mice versus irradiation-only or drug-alone controls, supporting clinical translation. In conclusion, our data indicate that IGF1R inhibition can block the N-cadherin-mediated resistance pathway. Our study deepens our understanding of adaptive resistance during repeated irradiation in GBM, and validates the IGF1/N-cadherin/b-catenin/Clusterin signaling axis as a novel target for radio-sensitization, which has direct therapeutic applicability. These findings also confirmed that our radioresistant models effectively identify new adaptive resistance mechanisms in malignant brain tumors.

INSP-10. Minimal residual disease monitoring for pediatric brain tumors using ctDNA whole genome sequencing: opportunities and challenges

Abstract Liquid biopsy offers a noninvasive approach to monitor cancer burden during therapy and surveillance. However, in pediatric brain cancers, liquid biopsy methods from the blood have been unsuccessful due to a low tumor burden and low number of mutations in coding regions. In contrast with targeted panels, whole genome sequencing (WGS)-derived patient specific mutational signature from a matched tumor-normal WGS can provide a personalized, highly specific approach to detect mutations in circulating cell free tumor DNA (ctDNA) and provide blood-based monitoring in pediatric patients with high sensitivity. Furthermore, it can be performed on lower amount of peripheral blood since WGS requires less depth compared to targeted ctDNA panels. We have profiled a diverse cohort of brain tumors including medulloblastomas, ependymomas, low- and high-grade gliomas. Using WGS of matched tumor-normal and plasma samples, we could derive a personalized mutational pattern and used an AI-based error suppression model for quantification and ultra-sensitive detection of ctDNA in plasma samples. A patient-specific personalized genome-wide compendium of somatic mutations could be established across all tumor types and ctDNA tested at the time of diagnosis, during the therapy or surveillance period. An AI-based error suppression model is implemented to filter out the noise in the cell free DNA (cfDNA) while the personalized mutational signature was used to detect the ctDNA in the cfDNA and to amplify the somatic signal contained in it. The ctDNA Tumor Fraction (TF) is compared to the clinical status and MR-based imaging. All subtypes of pediatric brain tumors contain sufficient number of mutations to derive personalized signatures and corelate with the clinical status. Patient-specific WGS tumor signature in ctDNA from blood can be used for sensitive monitoring of children with brain tumors. However, correlation between ctDNA levels and therapeutic response need to be established for various subtypes of brain tumors.

MODL-29. Molecular Landscape of a comprehensive panel of pediatric brain cancer Patient-derived orthotopic xenograft (PDOX) models inform unique targets for drug responsiveness

Abstract Brain tumor is a leading cause of cancer related death in children. In addition to replicating histopathology, animal models faithfully replicating genetic/epigenetic abnormalities, molecular subtypes and broad inter-tumoral heterogeneities are needed. Through direct implantation of patient surgical or autopsied tumor tissues into matching locations in the brains of SCID mice, we developed a panel of 150 PDOX mouse models. Here, we report the analysis of 74 of the 150 PDOX models, 45 matching patient tissues and 60 non-tumorigenic samples to a well-annotated reference cohort of 2,801 methylation profiles of primary brain tumors. Our data showed that the lack of tumorigenicity was neither correlated with molecular subtypes nor predicted by low cell viabilities of the patient samples. Methylation profiling identified PDOX models representing nearly a full spectrum of molecular subtypes of pediatric brain tumors including GBM, medulloblastoma, ependymoma and ATRT. Direct comparison with the original patient tumors confirmed the replication of molecular subtypes. ONCOplot [FB1] analysis of PDOX models derived from matching pairs of primary and recurrent tumors (n=8) revealed close clustering with the patient tumors. Investigation of metastatic properties was performed in 13 MB models by harvesting and sub-transplanting matching PDOX primary tumors in the cerebella and metastatic tumors in the spinal cords. To confirm the potential and power of PDOX models in preclinical drug testing, we applied fractionated radiation (2 Gy/day x 5 days) and optimized multi-agent combinatory chemotherapies in MB models of the four major subgroups. High-throughput combination drug screening with ~ 8,000 drugs in PDOX-derived GBM cell lines and primary cultures of MB PDOX cells identified a library of ~ 3,500 drugs that were active in pediatric brain tumors. In summary, this study provides detailed information on molecular subclassification of a uniquely large cohort of PDOX models to serve as essential tools for brain tumor research.

PATH-04. Array-based global DNA Methylation profiling of mouse brain tumors allows comparison to human tumors

The classification of human brain tumors by global DNA methylation profiling has become an essential part of modern integrated neuropathological diagnostics. It has proven to reliably identify known and novel brain tumor (sub-)types that are biologically and clinically distinct. Therefore, this technique has critically improved diagnostic accuracy and risk stratification of brain tumor patients. Although indispensable for the understanding of tumor biology and for preclinical drug trials, the comparison of genetically engineered mouse models to human brain tumors is still difficult. The assessment of tumor morphology only provides an approximate picture, and transcriptomic data from human brain tumors are sparse and suffer from platform-related technical incomparability. Here, we show array-based DNA methylation profiling of well-established murine brain tumors, such as Wnt and Shh medulloblastoma, YAP and RELA ependymoma, ETMR, and AT/RT. Similar to human brain tumors, unbiased clustering methods revealed distinct methylation profiles and mean methylation levels for mouse brain tumors types. Analyses were possible for fresh-frozen as well as for paraffin-embedded tissue, and copy number alterations could be inferred from methylation profiles. Most importantly, first results suggest that inter-species comparisons allow the classification of brain tumors from known or novel mouse models based on the constantly growing spectrum of human brain tumor types and subtypes with hundreds of thousands of available datasets. As an example, upon DNA methylation profiling, cerebellar tumors arising in Math1-cre::SmoM2Fl/+ mice display the highest similarity to human SHH medulloblastoma when compared to multiple human brain tumor entities including WNT and Non-WNT/Non-SHH medulloblastoma. These results suggest that global DNA methylation profiling may add another very important level of information to the characterization of genetically engineered mouse models.

TBIO-11. The glutamine transporter and candidate diagnostic and therapeutic target SLC1A5 is associated with subtype-specific metabolic phenotypes and tumor prognosis in pediatric brain cancers

Abstract Glutamine transporters play an important role in supporting increased tumor nutritional demands relative to non-cancerous cells, often through overexpression of the solute carrier (SLC) family of membrane transporters. Preclinical studies in adult cancers demonstrate that targeting glutamine addiction via SLC1A5 inhibition results in growth-inhibitory and tumoricidal effects. Given their relatively higher expression in cancer versus normal brain tissue, SLC transporters represent compelling targets for molecularly-targeted radiation and application of available prognostic amino-acid PET imaging probes. However, the role of SLC transporters in pediatric brain cancers has yet to be investigated. We aimed to understand the relationship of SLC transporter expression with pediatric brain tumor subtypes and their potential prognostic significance using data from the Pediatric Brain Tumor Atlas (PBTA). Using the expression of amino acid transporter genes in ensemble survival models (Reactome: R-HSA-352230), we found that elevated expression of glutamine transporters (SLC1A5, SLC7A5, SLC7A11, SLC38A5, SLC38A3) predicted shorter progression-free survival (PFS) in low-grade gliomas (LGGs) and poorer overall survival in pediatric ependymomas, high-grade gliomas (HGGs), and medulloblastomas. We focused specifically on SLC1A5 given the availability of imaging probes (18 F-Fluoroglutamine and 18F-Fluciclovine) for the corresponding amino acid transporter (ASCT2). Through transcriptome-based consensus clustering, we found that supratentorial, RELA fusion-positive ependymomas and sonic hedgehog-activated medulloblastomas were over-represented among clusters expressing higher levels of SLC1A5 (p = 3.38e-7 and p = 2.18-26, respectively). Kaplan-Meier analysis found that higher expression of SLC1A5 was associated with shorter OS in ependymoma and medulloblastoma (p = 9.8e-4 and p = 0.032) and shorter PFS in LGG (p = 0.022). Gene set analysis showed higher expression and network rewiring of amino acid, lipid, and immune pathways in SLC1A5-high expressing clusters. Our work demonstrates that glutamine transporters, particularly SLC1A5, represent compelling targets in pediatric brain cancers that warrant further investigation for molecularly-targeted treatment and amino-acid PET imaging.

Dematin inhibits glioblastoma malignancy through RhoA-mediated CDKs downregulation and cytoskeleton remodeling

Glioblastoma multiforme (GBM) is well known as a highly aggressive brain tumor subtype. Here, we show that overexpression (OE) of dematin actin-binding protein (DMTN) inhibits GBM proliferation and invasion by affecting cell cycle regulation and actin remodeling, respectively. RT–qPCR, western blotting, and immunohistochemical (IHC) staining demonstrated a significant reduction in DMTN expression in gliomas, especially in high-grade gliomas (HGG) compared with normal brains, which correlates with worse survival in HGG patients. Functional studies revealed inhibitory effects of DMTN on tumor proliferation and migratory capacities. The attenuation in tumor proliferative ability upon DMTN OE was accompanied by RhoA suppression and CDK1, CDK2, CDK4, and cyclin D1 downregulation, while RhoA rescue restored the proliferative phenotype. Meanwhile, overexpression of DMTN produced profoundly disorganized stress fibers, which led to impaired tumor invasion. Furthermore, DMTN overexpression produced substantial suppression of tumor growth upon subcutaneous and intracranial implantation in mice, and this was accompanied by significantly reduced vinculin expression and Ki67 positivity. Taken together, these findings demonstrate the role of DMTN in regulating GBM cell proliferation, actin cytoskeleton, and cell morphology and identify DMTN as a vital tumor suppressor in GBM progression.

Infra-tentorial brain tumor subtypes in children and adults\u2014surgical outcome in an ethnic population with a single regional tertiary center

BackgroundTo analyze clinically and radiologically the surgical outcome like residual disease, progression of disease, recurrence, disabilities, event-free survival (EFS), and mortality of different infra-tentorial tumor subtypes in children and adults of a strictly non-migratory and ethnic population.MethodsThe 410 histologically proved, out of 589, infra-tentorial brain tumor patients were analyzed clinically and by the imaging post-surgically in a single tertiary center for an ethnic region. In this analytico-observational study, retrospectively postoperative records of 589 infra-tentorial brain tumors from November 1998 to December 2018 (20 years) were retrieved, scrutinized, and compiled. The post-operative clinic-radiological records of 410 patients with proved histopathological examination results were included. Statistical law of variance was applied where-ever necessary.ResultsThe 63.2% of the all 410 operated infra-tentorial brain tumors were males while females predominated in meningiomas and pineoblastomas. About 31.7% infra-tentorial tumors were children (below 18 years). About 54.1% cases were histologically malignant. The residual tumors comprised 40.2% and symptoms of disease-progression occurred in 10.9%. The tumor recurrence occurred in 14.3% while 6.0% patients developed severe disability. The overall mortality was 11.4% but 18.9% in malignant tumors. The event-free survival (EFS) for all the patients was 66.0%, patients with malignancies had 47.7% and benign group had 87.7%.ConclusionThe study, surgical outcome of infra-tentorial brain tumor subtypes in children and adults (approx. 1/3rd of patients being children), conducted in a tertiary center at a remote land-locked location with non-migratory ethnic population as its catchment area, has a significant epidemiological value for the community and the region.

Promising Chemotherapy for Malignant Pediatric Brain Tumor in Recent Biological Insights.

Brain tumors are the most widespread malignancies in children around the world. Chemotherapy plays a critical role in the treatment of these tumors. Although the current chemotherapy process has a remarkable outcome for a certain subtype of brain tumor, improving patient survival is still a major challenge. Further intensive treatment with conventional non-specific chemotherapy could cause additional adverse reactions without significant advancement in survival. Recently, patient derived brain tumor, xenograft, and whole genome analysis using deep sequencing technology has made a significant contribution to our understanding of cancer treatment. This realization has changed the focus to new agents, targeting the molecular pathways that are critical to tumor survival or proliferation. Thus, many novel drugs targeting epigenetic regulators or tyrosine kinase have been developed. These selective drugs may have less toxicity in normal cells and are expected to be more effective than non-specific chemotherapeutics. This review will summarize the latest novel targets and corresponding candidate drugs, which are promising chemotherapy for brain tumors according to the biological insights.

State-of-the-Art Explainability Methods with Focus on Visual Analytics Showcased by Glioma Classification

This study aims to reflect on a list of libraries providing decision support to AI models. The goal is to assist in finding suitable libraries that support visual explainability and interpretability of the output of their AI model. Especially in sensitive application areas, such as medicine, this is crucial for understanding the decision-making process and for a safe application. Therefore, we use a glioma classification model’s reasoning as an underlying case. We present a comparison of 11 identified Python libraries that provide an addition to the better known SHAP and LIME libraries for visualizing explainability. The libraries are selected based on certain attributes, such as being implemented in Python, supporting visual analysis, thorough documentation, and active maintenance. We showcase and compare four libraries for global interpretations (ELI5, Dalex, InterpretML, and SHAP) and three libraries for local interpretations (Lime, Dalex, and InterpretML). As use case, we process a combination of openly available data sets on glioma for the task of studying feature importance when classifying the grade II, III, and IV brain tumor subtypes glioblastoma multiforme (GBM), anaplastic astrocytoma (AASTR), and oligodendroglioma (ODG), out of 1276 samples and 252 attributes. The exemplified model confirms known variations and studying local explainability contributes to revealing less known variations as putative biomarkers. The full comparison spreadsheet and implementation examples can be found in the appendix.