Glioma Subtypes Research Articles

PATH-12. CONTEMPORARY MOLECULAR LANDSCAPE, SURVIVAL, AND PROGNOSTIC FACTORS IN OVER 4,000 GLIOMAS

Abstract INTRODUCTION Molecular glioma classification and advances in treatment modalities have expanded therapeutic options for patients. We sought to capture the genomic landscape, prognostic features, and contemporary outcomes of gliomas. METHODS We assembled patients from TCGA (n=770) and multi-institutional datasets (n=3,546) with somatic mutations and copy number/structural variants. All gliomas were classified using molecular definitions (WHO 2021/cIMPACT-NOW). We identified correlations between genes through permutations testing and prognostic features through multivariate cox models with backwards elimination. RESULTS We analyzed 4,316 patients (median age 52yrs, 41% female), including 2,155 glioblastoma, 1,199 IDH1/2-mutant astrocytoma, 531 oligodendroglioma, and 343 other IDH1/2-wiltype gliomas. Molecular classification shifted 27.2% of gliomas from their original histopathologic designation. Canonical alterations categorized glioma subtypes, showing unique correlations with clinical features and mutual exclusivity of alterations in tumorigenic pathways. 2,433 patients had clinical outcomes: 1,296 glioblastomas, 584 IDH1/2-mutant astrocytoma, and 327 oligodendrogliomas. Patients from contemporary institutional cohorts had improved overall survival compared to TCGA patients: glioblastoma (non-TCGA:17.9mo,TCGA:15.0mo,p< 0.0001), IDH1/2-mutant astrocytoma (non-TCGA:180mo,TCGA: 87.9mo,p=0.0002), and oligodendroglioma (non-TCGA: not reached,TCGA:135mo,p< 0.0001). In contemporary cohorts, median survival in grade 2-3 IDH1/2-mutant astrocytoma (180.3mo) was double that of grade 4 (75.4mo, p< 0.0001). In GBM and oligodendrogliomas, patients >65yrs had worse outcomes than patients 20-64yrs (p< 0.001). In glioblastoma, age (HR:1.03), CDKN2A/B homozygous deletion (HR:1.22), and NF1 alterations (HR:1.31) were negatively prognostic. Chemotherapy (HR:0.64), MGMT-methylation (HR:0.44), 19q-gain (HR:0.70), and non-TCGA cohort (HR:0.54,HR:0.34) were positively prognostic. In IDH1/2-mutant astrocytoma, CDKN2A/B homozygous deletion (HR:2.75), CCND2 alteration (HR:2.89), EGFR amplification (HR:2.93), PDGFRA alteration (HR:2.87), 9p-loss (HR:3.09), and 10q-loss (HR:1.99) were negatively prognostic; white race was the only prognostically positive feature (HR:0.35). CONCLUSIONS Overall survival in gliomas has improved over time. We corroborated canonical features and identified new prognostic features that are subtype-specific. The unique molecular profiles across glioma subtypes provide insight into biological drivers that influence heterogenous clinical outcomes.

NIMG-59. RADIOMIC FEATURE CLUSTER ANALYSIS OF IDH-MUTANT GLIOMA SUBTYPES

Abstract INTRODUCTION IDH-mutant glioma subtypes share similarities on MRI and histopathology but have distinct differences in survival based on their molecular subtypes. Radiomic features provide novel analysis of glioma imaging phenotypes that may not appear in clinical evaluations. We evaluated whether clustering of radiomic features of IDH-mutant gliomas could identify different subtypes based on the 2021 WHO classification. METHODS We extracted up to 5,763 radiomic features per case from pretreatment FLAIR and T1 post-contrast tumor segmentations of 206 patients with IDH-mutant gliomas WHO grade 2-4 using a standardized Pyradiomics 3.1.0 pipeline. Segmentations were performed using a semiautomatic workflow with a UNETR algorithm trained on the BRaTS 2021 dataset. To visualize high-dimensional feature space, we evaluated the following unsupervised clustering algorithms to differentiate tumor subtypes: PCA, t-SNE, UMAP, and PHATE. We performed redundancy reduction by calculating a Pearson correlation-matrix, and removed highly correlated features above a threshold of 0.7. Significant features were selected using the L1-norm regularization (LASSO). RESULTS The dataset of 206 patients included 19 grade 4, 28 grade 3, and 36 grade 2 astrocytomas as well as 23 grade 3 and 58 grade 2 oligodendrogliomas. The UNETR algorithm performed with a 0.85 Dice score on the segmentations. Clustering of the extracted 2D radiomic features from FLAIR and T1postgad using PCA, tSNE, UMAP, and PHATE did not identify distinct classes. To further analyze the two tumor entities, we applied XGBOOST, RandomForest, and KNN classifiers and found them to have AUCs < 0.5. CONCLUSIONS We showed that the spectrum of imaging features of IDH-mutant gliomas has a significant overlap on FLAIR and T1 post-contrast imaging. We propose that standard anatomical imaging does not provide sufficient information on these two sequences. To further identify specific subtypes of IDH-mutant gliomas, additional sequences, such as T2, ADC, SWI, and T1, can provide additional information.

TMET-06. METABOLOGENOMIC CHARACTERIZATION UNCOVERS HETEROGENEITY AMONGST IDH MUTANT

Abstract INTRODUCTION Mutations in isocitrate dehydrogenase (IDH) enzymes are recognised to drive the molecular footprint of diffuse gliomas through the accumulation of oncometabolite R-2-hydroxyglutarate which drives the widespread restructuring of the DNA methylome; however, beyond R-2-hydroxyglutarate , a comprehensive metabologenomic characterization of IDH-mutant gliomas has yet to be performed. We aimed to characterize the global metabolic profile of IDH-mut and IDH-wt gliomas, identify clinically relevant subgroups based on distinct metabolic differences, and utilize matched epigenetic and transcriptional data to characterize these distinct groups. Method: Glioma samples from a cohort of 154 patients underwent multiplatform molecular analysis, including metabolomic studies, genome-wide DNA methylation profiling and bulk RNA sequencing. A integrative analysis was performed, including hierarchical clustering and principal component analysis of identified metabolites, differentially methylated probes, copy number alteration and bulk mRNA data. Survival analysis as well as multivariable hazard ratios for clinical variables were calculated by fitting Cox Proportional Hazards Models. RESULTS We discovered a group of IDH-mutant gliomas whose metabolic profile highly resembled IDH wildtype tumors. Notably, these IDH-mutant gliomas with dysregulated metabolism were distinguished from their IDH-mutant counterparts by significantly shorter overall survival (median OS 118.9 months vs 173.6 months, p=0.048). The IDH-mutant tumors with dysregulated metabolism harbored distinct epigenetic alterations that converged to drive proliferative and stem-like transcriptional profiles. The prognostic relevance of dysregulated metabolism complements, but was not wholly explained by canonically recognized prognostic classifications in IDH-mutant gliomas including 1p/19q codeletion, glioma CpG Island Hypermethylator (GCIMP) status and CDKN2A homozygous deletion. CONCLUSION Utilizing a cross-platform analysis we have uncovered a novel subtyping of IDH-mutant gliomas with dysregulated cellular metabolism with similar survival to IDH-wildtype tumors. The metabolic profile provides unique information on glioma phenotypes, which may facilitate a more comprehensive understanding of glioma biology, and provide a window to target novel dependencies.

EXTH-37. FUNCTIONAL GENOMICS IDENTIFIES THERAPEUTIC VULNERABILITIES OF GLIOMA SUBTYPES AND YAP/TAZ-ENRICHED TUMORS

Abstract Sequencing efforts have identified key oncogenic drivers and transcriptomic subtypes of high-grade gliomas (HGGs), but this knowledge has not resulted in individualized therapy or improved patient outcomes. Here, we used a functional genomics approach combining integrated genomic/transcriptomic analyses with patient-derived 3D micro-cancers (mCancers) to rapidly identify individualized therapeutic vulnerabilities and determine predictive response biomarkers. 30 primary/recurrent HGGs, including 20 glioblastomas, 5 astrocytomas, and 5 oligodendrogliomas, were included for drug testing. mCancers recapitulated the structural and molecular alterations, cellularity, and heterogeneity, of parental patient tumors. A total of 119 single-agent drugs and 183 drug combinations were tested, while a pre-selected drug panel comprising 25 glioma-relevant treatment strategies was tested across all cases. In general, mCancer efficacy corresponded well to historical responses to previously tested agents (EGFR, HDAC inhibitors, etc.). Combined, strong efficacy (i.e. >70% growth inhibition at Cmax) of single treatments targeting deregulated receptor tyrosine kinase (RTK), p53/apoptosis, cell cycle, DNA damage repair, IDH, or telomere maintainance pathways, was observed in 14 out of 30 cases. Surprisingly, transcriptional signatures, but not specific genetic alterations, were found to stratify drug responses. For example, mCancers enriched for the most aggressive mesenchymal (MES) and glycolytic/plurimetabolic (GPM) transcriptional subtypes exhibited stronger response to the BET inhibitor JQ1 (and other clinical BET inhibitors) and the DNA-PKcs/mTOR dual inhibitor CC-115. This therapeutic vulnerability was associated with enrichment of YAP/TAZ signaling, and the glioblastoma stem cell G-STEM signature. Notably, individual combination strategies were effective across tumors, with strong responses ranging from 38% to 77% of all tumors tested. Overall, this study underscores the value of functional genomics in identifying novel vulnerabilities of HGGs and associated response biomarkers. It also highlights the need for further clinical validation of several novel therapies and especially drug combination strategies with strong efficacy in the ex vivo setting.

TMIC-60. CROSS-SPECIES ANALYSIS OF HUMAN AND NEW MOUSE MODELS OF GBM SUBTYPES IDENTIFIES VISTA AS A PROMISING IMMUNOTHERAPY TARGET

Abstract Immunotherapy is a promising treatment modality for GBM; however, clinical trials thus far have failed to provide significant clinical benefit to most patients. GBM is one of the “immune cold” tumors characterized by poor infiltration of T cells and the abundance of immune-suppressive myeloid cells that block T cell infiltration and/or activation. A major bottleneck to developing more effective immunotherapies is the paucity of faithful preclinical models to discover and validate new treatments. Here, we report multi-dimensional analyses of six different transplantable mouse glioma models in the C57BL6/J background that represent the three human GBM molecular subtypes. We performed whole exome sequencing, STR fingerprinting, immune phenotyping, and single-cell RNA-sequencing of each model and confirmed that the new models exhibit highly heterogeneous and immune suppressive TME, similar to human GBM. In addition, we performed cross-species comparisons of glioma and immune cell subtypes between human and mouse GBMs at the single-cell level and demonstrate strong concordances in molecular phenotypes and cellular heterogeneity. These result in qualitative and quantitative differences in the cell: cell communication among different stromal cells and glioma cells in each model, and we propose that these interactions shape the local niche and immune suppressive neighborhoods. Notably, in both human tissues and mouse models, PD1/PDL1 expression is low or sporadic, and, unlike the GL261 model, the new mouse models are resistant to anti-PD1 and CTLA4 treatment in vivo. Finally, we discovered that VISTA is a widely expressed immune checkpoint in both human and mouse GBM and demonstrate the efficacy of anti-VISTA and anti-PD1 combination therapy in the most immune-suppressive mouse model. In summary, we demonstrate the utility of these immune-competent mouse models and nominate anti-VISTA+PD1 treatment as a promising novel combination therapy for GBM.

PATH-23. CLINICOPATHOLOGIC ANALYSIS OF NOVEL METHYLATION CLUSTERS OF IDH-WILDTYPE DIFFUSE GLIOMAS

Abstract Methylation profiling of brain tumors has revealed novel clusters whose biological and clinical significance is unknown. These include three entities termed by the 12.5 version of the Heidelberg methylation classifier as adult-type diffuse high grade glioma, IDH-wildtype, subtype B (HGG_B), subtype E (HGG_E), and subtype F (HGG_F). We have identified 75 cases belonging to these groups: HGG_B (25 cases, 32% female), HGG_E (40 cases, 42% female), and HGG_F (18 cases, 33% female). Although originally characterized as “adult-type,” we find that 70% of HGG_E cases are pediatric, with a median age at diagnosis of 11 years. HGG_B cases are predominantly seen in young adults, and HGG_F in older adults. The histology varies among methylation classes, with HGG_B cases demonstrating high-grade, predominantly astrocytic, morphology, while HGG_E cases demonstrate high-grade glial histology with distinct nuclear features, and HGG_F cases show highly infiltrative but lower grade glial morphology. Patient outcome data obtained to date shows a collective median survival exceeding 5 years, suggesting re-consideration of the proposed grading for these tumors. MGMT promoter methylation differs across the HGG_B (0%), HGG_E (12%), and HGG_F (0%) subtypes. Mutational analyses show enrichment of TERT promoter mutations in the HGG_F group, while HGG_E samples are enriched for tumors with POLE mutations and elevated tumor mutation burden. HGG_E tumors show epigenetic similarity to pediatric high-grade diffuse gliomas (pedHGG_RTK1A); we therefore propose that HGG_E be considered a pediatric-type HGG. HGG_B tumors show demographic and genomic similarities to high-grade astrocytomas with piloid features, including a subset of cases with CDKN2A/2B loss and ATRX mutation. HGG_F separate into two groups dictated by brain location, and despite the presence of TERT promoter mutations, appear distinct from GBMs. Overall, our work represents an initial effort to understand the morphologic, genomic and clinical characteristics of these previously uncharacterized subtypes of gliomas.

EXTH-28. ISOCITRATE DEHYDROGENASE MUTATIONS PREDICT SENSITIVITY TO HISTONE DEACETYLASE INHIBITION AND ENHANCED HISTONE ACETYLATION DYNAMICS ONLY IN THE CONTEXT OF GLIOMA

Abstract Isocitrate dehydrogenase 1/2 mutations (IDHmut) define a subtype of glioma with profound epigenetic dysregulation through genomic hypermethylation. IDHmut gliomas are incurable and fatal, demonstrating new treatments are needed. These mutations are common in other solid tumors, including chondrosarcoma (CS) and intrahepatic cholangiocarcinoma (ICC). Histone deacetylase (HDAC) enzymes compose a class of epigenetic drug targets utilized clinically to treat cancer, but their potential against IDHmut gliomas has not been extensively studied. We previously showed that IDHmut glioma cultures exhibit pronounced sensitivity to the effects mediated by the FDA-approved HDAC inhibitor (HDACi) panobinostat. Here, we integrate proliferation and cytoxicity data to show that this effect is unique to gliomas. In the context of histone H3 acetylation (H3KAc), panobinostat increased H3KAc, on average, by 7.6-fold and 30.5-fold in IDHwt and IDHmut glioma cultures, respectively (p< 0.001). This difference in HDACi-mediated H3KAc based on IDH status was not significant for CS and ICC. IDHmut gliomas are more sensitive two other FDA-approved HDACis: belinostat and vorinostat. RNA-Seq analysis showed that panobinostat more greatly affected gene expression in IDHmut glioma cells, particularly in the context of gene downregulation. Similarly, H3KAc ChIP-Seq showed a decrease in promoter histone acetylation moreso in IDHmut glioma cultures treated with panobinostat: there is an average loss of 1,236 and 14,759 H3KAc peaks at gene promoters in IDHwt and IDHmut cultures, respectively. This suggests that IDHmut gliomas exhibit enhanced histone acetylation dynamics in response to HDACi. Furthermore, using a PDX glioma model, we show that only mice engrafted with IDHmut glioma exhibit a significant survival increase in response to the HDACi belinostat: median survival was extended by 14 days in IDHmut glioma-bearing mice (p= 0.01), whereas belinostat extended median survival by 4 days in IDHwt glioma-bearing mice (p= 0.11). These results suggest that HDACi may be an effective in treating IDHmut gliomas.

Context-dependent regulation of Notch signaling in glial development and tumorigenesis.

In the mammalian brain, Notch signaling maintains the cortical stem cell pool and regulates the glial cell fate choice and differentiation. However, the function of Notch in regulating glial development and its involvement in tumorigenesis have not been well understood. Here, we show that Notch inactivation by genetic deletion of Rbpj in stem cells decreases astrocytes but increases oligodendrocytes with altered internal states. Inhibiting Notch in glial progenitors does not affect cell generation but instead accelerates the growth of Notch-deprived oligodendrocyte progenitor cells (OPCs) and OPC-related glioma. We also identified a cross-talk between oligodendrocytes and astrocytes, with premyelinating oligodendrocytes secreting BMP4, which is repressed by Notch, to up-regulate GFAP expression in adjacent astrocytes. Moreover, Notch inactivation in stem cells causes a glioma subtype shift from astroglia-associated to OPC-correlated patterns and vice versa. Our study reveals Notch's context-dependent function, promoting astrocytes and astroglia-associated glioma in stem cells and repressing OPCs and related glioma in glial progenitors.

MAPK signaling pathway-based glioma subtypes, machine-learning risk model, and key hub proteins identification

An early diagnosis and precise prognosis are critical for the treatment of glioma. The mitogen‑activated protein kinase (MAPK) signaling pathway potentially affects glioma, but the exploration of the clinical values of the pathway remains lacking. We accessed data from TCGA, GTEx, CGGA, etc. Up-regulated MAPK signaling pathway genes in glioma were identified and used to cluster the glioma subtypes using consensus clustering. The subtype differences in survival, cancer stemness, and the immune microenvironment were analyzed. A prognostic model was trained with the identified genes using the LASSO method and was validated with three external cohorts. The correlations between the risk model and cancer-associated signatures in cancer were analyzed. Key hub genes of the gene set were identified by hub gene analysis and survival analysis. 47% of the MAPK signaling pathway genes were overexpressed in glioma. Subtypes based on these genes were distinguished in survival, cancer stemness, and the immune microenvironment. A risk model was calculated with high confidence in the prediction of overall survival and was correlated with multiple cancer-associated signatures. 12 hub genes were identified and 8 of them were associated with survival. The MAPK signaling pathway was overexpressed in glioma with prognostic value.

Variant allelic frequencies of driver mutations can identify gliomas with potentially false-negative MGMT promoter methylation results

MGMT promoter methylation testing is required for prognosis and predicting temozolomide response in gliomas. Accurate results depend on sufficient tumor cellularity, but histologic estimates of cellularity are subjective. We sought to determine whether driver mutation variant allelic frequency (VAF) could serve as a more objective metric for cellularity and identify possible false-negative MGMT samples. Among 691 adult-type diffuse gliomas, MGMT promoter methylation was assessed by pyrosequencing (N = 445) or DNA methylation array (N = 246); VAFs of TERT and IDH driver mutations were assessed by next generation sequencing. MGMT results were analyzed in relation to VAF. By pyrosequencing, 56% of all gliomas with driver mutation VAF ≥ 0.325 had MGMT promoter methylation, versus only 37% with VAF < 0.325 (p < 0.0001). The mean MGMT promoter pyrosequencing score was 19.3% for samples with VAF VAF ≥ 0.325, versus 12.7% for samples with VAF < 0.325 (p < 0.0001). Optimal VAF cutoffs differed among glioma subtypes (IDH wildtype glioblastoma: 0.12–0.18, IDH mutant astrocytoma: ~0.33, IDH mutant and 1p/19q co-deleted oligodendroglioma: 0.3–0.4). Methylation array was more sensitive for MGMT promoter methylation at lower VAFs than pyrosequencing. Microscopic examination tended to overestimate tumor cellularity when VAF was low. Re-testing low-VAF cases with methylation array and droplet digital PCR (ddPCR) confirmed that a subset of them had originally been false-negative. We conclude that driver mutation VAF is a useful quality assurance metric when evaluating MGMT promoter methylation tests, as it can help identify possible false-negative cases.

Tumor heterogeneity and tumor-microglia interactions in primary and recurrent IDH1-mutant gliomas

The isocitrate dehydrogenase (IDH) gene is recurrently mutated in adult diffuse gliomas. IDH-mutant gliomas are categorized into oligodendrogliomas and astrocytomas, each with unique pathological features. Here, we use single-nucleus RNA and ATAC sequencing to compare the molecular heterogeneity of these glioma subtypes. In addition to astrocyte-like, oligodendrocyte progenitor-like, and cycling tumor subpopulations, a tumor population enriched for ribosomal genes and translation elongation factors is primarily present in oligodendrogliomas. Longitudinal analysis of astrocytomas indicates that the proportion of tumor subpopulations remains stable in recurrent tumors. Analysis of tumor-associated microglia/macrophages (TAMs) reveals significant differences between oligodendrogliomas, with astrocytomas harboring inflammatory TAMs expressing phosphorylated STAT1, as confirmed by immunohistochemistry. Furthermore, inferred receptor-ligand interactions between tumor subpopulations and TAMs may contribute to TAM state diversity. Overall, our study sheds light on distinct tumor populations, TAM heterogeneity, TAM-tumor interactions in IDH-mutant glioma subtypes, and the relative stability of tumor subpopulations in recurrent astrocytomas.

SUZ12 inhibition attenuates cell proliferation of glioblastoma via post-translational regulation of CDKN1B.

Human gliomas are aggressive brain tumors characterized by uncontrolled cell proliferation. Differential expression of Polycomb repressive complex 2 (PRC2) has been reported in various subtypes of glioma. However, the role of PRC2 in uncontrolled growth in glioma and its underlying molecular mechanisms remain to be elucidated. We aimed to investigate the functional role of PRC2 in human glioblastoma cell growth by silencing SUZ12, the non-catalytic core component of PRC2. Knockdown of SUZ12 was achieved by infecting T98G cells with lentivirus carrying sequences specifically targeting SUZ12 (shSUZ12). Gene expression was examined by quantitative PCR and western analysis. The impact of shSUZ12 on cell growth was assessed using a cell proliferation assay. Cell cycle distribution was analyzed by flow cytometry, and protein stability was evaluated in cycloheximide-treated cells. Subcellular localization was examined through immunofluorescence staining and biochemical cytoplasmic-nuclear fractionation. Gene expression analysis was also performed on human specimens from normal brain and glioblastoma patients. SUZ12 knockdown (SUZ12 KD) led to widespread decrease in the PRC2-specific histone mark, accompanied by a slowdown of cell proliferation through G1 arrest. In SUZ12 KD cells, the degradation of CDKN1B protein was reduced, resulting from alterations in the MYC-SKP2-CDKN1B axis. Furthermore, nuclear localization of CDKN1B was enhanced in SUZ12 KD cells. Analysis of human glioblastoma samples yielded increased expression of EZH2 and MYC along with reduced CDKN1B compared to normal human brain tissue. Our findings suggest a novel role for SUZ12 in cell proliferation through post-translational regulation of CDKN1B in glioblastoma.

Exploring the mechanism of 18F-fluorodopa uptake in recurrent high-grade gliomas: A comprehensive histomolecular-positron emission tomography analysis.

Dihydroxy-6-[18F]fluoro-L-phenylalanine (18F-FDOPA) positron emission tomography (PET) is a valuable tool for managing high-grade gliomas (HGGs), but there is a lack of literature on its relationship with glioma subtypes since the 2021 reclassification of brain tumors. There is also debate surrounding the mechanism of 18F-FDOPA uptake, particularly after chemoradiation therapy. This study aimed to investigate the correlation between 18F-FDOPA uptake and histomolecular characteristics, particularly L-amino acid transporter 1 (LAT1) expression, in recurrent gliomas, and examine their impact on survival in HGGs. Thirty-nine patients with recurrent HGGs (14 isocitrate dehydrogenase [IDH]-mutant, 25 IDH-wildtype) who underwent a brain 18F-FDOPA PET/computed tomography (CT) or PET/magnetic resonance imaging (MRI) followed by surgical resection of the 18F-FDOPA-avid lesion within 6 months, were retrospectively reviewed. PET results were compared with histological examination and for SCL7A5/LAT1 immunostaining. The study also examined the relationship between PET parameters, LAT1 expression, and survival outcomes. Astrocytoma IDH-mutant G4 had higher 18F-FDOPA uptake than glioblastoma IDH-wildtype G4 (maximum tumor-to-normal brain ratio [TBRmax] 5 [3.4-9] vs. 3.8 [2.8-5.9], p = 0.02). IDH-mutant gliomas had higher LAT1 expression than IDH-wildtype gliomas (100 [14-273] vs. 15.5 [0-137], p < 0.05) as well as higher TBRmax (5 [2.4-9] vs. 3.8 [2.8-6], p < 0.05). In survival analysis, LAT1 score >100 was a predictor for longer progression-free survival in IDH-mutant HGGs. To our knowledge, our study is the first to suggest a link between LAT1 expression and IDH mutation status. We showed that higher TBRmax was associated with higher LAT1 expression and IDH mutation status. Further studies are needed to better understand the mechanisms underlying amino acid PET tracers uptake, especially in the post-radiation and chemotherapy settings.

Increased confidence of radiomics facilitating pretherapeutic differentiation of BRAF-altered pediatric low-grade glioma.

Currently, the BRAF status of pediatric low-grade glioma (pLGG) patients is determined through a biopsy. We established a nomogram to predict BRAF status non-invasively using clinical and radiomic factors. Additionally, we assessed an advanced thresholding method to provide only high-confidence predictions for the molecular subtype. Finally, we tested whether radiomic features provide additional predictive information for this classification task, beyond that which is embedded in the location of the tumor. Random forest (RF) models were trained on radiomic and clinical features both separately and together, to evaluate the utility of each feature set. Instead of using the traditional single threshold technique to convert the model outputs to class predictions, we implemented a double threshold mechanism that accounted for uncertainty. Additionally, a linear model was trained and depicted graphically as a nomogram. The combined RF (AUC: 0.925) outperformed the RFs trained on radiomic (AUC: 0.863) or clinical (AUC: 0.889) features alone. The linear model had a comparable AUC (0.916), despite its lower complexity. Traditional thresholding produced an accuracy of 84.5%, while the double threshold approach yielded 92.2% accuracy on the 80.7% of patients with the highest confidence predictions. Models that included radiomic features outperformed, underscoring their importance for the prediction of BRAF status. A linear model performed similarly to RF but with the added benefit that it can be visualized as a nomogram, improving the explainability of the model. The double threshold technique was able to identify uncertain predictions, enhancing the clinical utility of the model. Radiomic features and tumor location are both predictive of BRAF status in pLGG patients. We show that they contain complementary information and depict the optimal model as a nomogram, which can be used as a non-invasive alternative to biopsy. • Radiomic features provide additional predictive information for the determination of the molecular subtype of pediatric low-grade gliomas patients, beyond what is embedded in the location of the tumor, which has an established relationship with genetic status. • An advanced thresholding method can help to distinguish cases where machine learning models have a high chance of being (in)correct, improving the utility of these models. • A simple linear model performs similarly to a more powerful random forest model at classifying the molecular subtype of pediatric low-grade gliomas but has the added benefit that it can be converted into a nomogram, which may facilitate clinical implementation by improving the explainability of the model.

Identifying predictors of glioma evolution from longitudinal sequencing.

Clonal evolution drives cancer progression and therapeutic resistance. Recent studies have revealed divergent longitudinal trajectories in gliomas, but early molecular features steering posttreatment cancer evolution remain unclear. Here, we collected sequencing and clinical data of initial-recurrent tumor pairs from 544 adult diffuse gliomas and performed multivariate analysis to identify early molecular predictors of tumor evolution in three diffuse glioma subtypes. We found that CDKN2A deletion at initial diagnosis preceded tumor necrosis and microvascular proliferation that occur at later stages of IDH-mutant glioma. Ki67 expression at diagnosis was positively correlated with acquiring hypermutation at recurrence in the IDH-wild-type glioma. In all glioma subtypes, MYC gain or MYC-target activation at diagnosis was associated with treatment-induced hypermutation at recurrence. To predict glioma evolution, we constructed CELLO2 (Cancer EvoLution for LOngitudinal data version 2), a machine learning model integrating features at diagnosis to forecast hypermutation and progression after treatment. CELLO2 successfully stratified patients into subgroups with distinct prognoses and identified a high-risk patient group featured by MYC gain with worse post-progression survival, from the low-grade IDH-mutant-noncodel subtype. We then performed chronic temozolomide-induction experiments in glioma cell lines and isogenic patient-derived gliomaspheres and demonstrated that MYC drives temozolomide resistance by promoting hypermutation. Mechanistically, we demonstrated that, by binding to open chromatin and transcriptionally active genomic regions, c-MYC increases the vulnerability of key mismatch repair genes to treatment-induced mutagenesis, thus triggering hypermutation. This study reveals early predictors of cancer evolution under therapy and provides a resource for precision oncology targeting cancer dynamics in diffuse gliomas.

Integrated molecular and multiparametric MRI mapping of high-grade glioma identifies regional biologic signatures

Sampling restrictions have hindered the comprehensive study of invasive non-enhancing (NE) high-grade glioma (HGG) cell populations driving tumor progression. Here, we present an integrated multi-omic analysis of spatially matched molecular and multi-parametric magnetic resonance imaging (MRI) profiling across 313 multi-regional tumor biopsies, including 111 from the NE, across 68 HGG patients. Whole exome and RNA sequencing uncover unique genomic alterations to unresectable invasive NE tumor, including subclonal events, which inform genomic models predictive of geographic evolution. Infiltrative NE tumor is alternatively enriched with tumor cells exhibiting neuronal or glycolytic/plurimetabolic cellular states, two principal transcriptomic pathway-based glioma subtypes, which respectively demonstrate abundant private mutations or enrichment in immune cell signatures. These NE phenotypes are non-invasively identified through normalized K2 imaging signatures, which discern cell size heterogeneity on dynamic susceptibility contrast (DSC)-MRI. NE tumor populations predicted to display increased cellular proliferation by mean diffusivity (MD) MRI metrics are uniquely associated with EGFR amplification and CDKN2A homozygous deletion. The biophysical mapping of infiltrative HGG potentially enables the clinical recognition of tumor subpopulations with aggressive molecular signatures driving tumor progression, thereby informing precision medicine targeting.

Concordance analysis of cerebrospinal fluid with the tumor tissue for integrated diagnosis in gliomas based on next-generation sequencing.

Purpose: The driver mutations of gliomas have been identified in cerebrospinal fluid (CSF). Here we compared the concordance between CSF and tumor tissue for integrated diagnosis in gliomas using next-generation sequencing (NGS) to evaluate the feasibility of CSF detection in gliomas. Patients and methods: 27 paired CSF/tumor tissues of glioma patients were sequenced by a customized gene panel based on NGS. All CSF samples were collected through lumbar puncture before surgery. Integrated diagnosis was made by analysis of histology and tumor DNA molecular pathology according to the 2021 WHO classification of the central nervous system tumors. Results: A total of 24 patients had detectable circulating tumor DNA (ctDNA) and 22 had at least one somatic mutation or chromosome alteration in CSF. The ctDNA levels varied significantly across different ages, Ki-67 index, magnetic resonance imaging signal and glioma subtypes (p < 0.05). The concordance between integrated ctDNA diagnosis and the final diagnosis came up to 91.6% (Kappa, 0.800). We reclassified the clinical diagnosis of 3 patients based on the results of CSF ctDNA sequencing, and 4 patients were reassessed depending on tumor DNA. Interestingly, a rare IDH1 R132C was identified in CSF ctDNA, but not in the corresponding tumor sample. Conclusion: This study demonstrates a high concordance between integrated ctDNA diagnosis and the final diagnosis of gliomas, highlighting the practicability of NGS based detection of mutations of CSF in assisting integrated diagnosis of gliomas, especially glioblastoma.

Disclosing transcriptomics network-based signatures of glioma heterogeneity using sparse methods

Gliomas are primary malignant brain tumors with poor survival and high resistance to available treatments. Improving the molecular understanding of glioma and disclosing novel biomarkers of tumor development and progression could help to find novel targeted therapies for this type of cancer. Public databases such as The Cancer Genome Atlas (TCGA) provide an invaluable source of molecular information on cancer tissues. Machine learning tools show promise in dealing with the high dimension of omics data and extracting relevant information from it. In this work, network inference and clustering methods, namely Joint Graphical lasso and Robust Sparse K-means Clustering, were applied to RNA-sequencing data from TCGA glioma patients to identify shared and distinct gene networks among different types of glioma (glioblastoma, astrocytoma, and oligodendroglioma) and disclose new patient groups and the relevant genes behind groups’ separation. The results obtained suggest that astrocytoma and oligodendroglioma have more similarities compared with glioblastoma, highlighting the molecular differences between glioblastoma and the others glioma subtypes. After a comprehensive literature search on the relevant genes pointed our from our analysis, we identified potential candidates for biomarkers of glioma. Further molecular validation of these genes is encouraged to understand their potential role in diagnosis and in the design of novel therapies.

Distinct aneuploid evolution of astrocytoma and glioblastoma during recurrence

Astrocytoma and glioblastoma (GB) are reclassified subtypes of adult diffuse gliomas based on distinct isocitrate dehydrogenase (IDH) mutation in the fifth edition of the WHO Classification of Tumors of the Central Nervous System. The recurrence of gliomas is a common and inevitable challenge, and analyzing the distinct genomic alterations in astrocytoma and GB could provide insights into their progression. This study conducted a longitudinal investigation, utilizing whole-exome sequencing, on 65 paired primary/recurrent gliomas. It examined chromosome arm aneuploidies, copy number variations (CNVs) of cancer-related genes and pathway enrichments during the relapse. The veracity of these findings was verified through the integration of our data with multiple public resources and by corroborative immunohistochemistry (IHC). The results revealed a greater prevalence of aneuploidy changes and acquired CNVs in recurrent lower grade astrocytoma than in relapsed grade 4 astrocytoma and GB. Larger aneuploidy changes were predictive of an unfavorable prognosis in lower grade astrocytoma (P < 0.05). Further, patients with acquired gains of 1q, 6p or loss of 13q at recurrence had a shorter overall survival in lower grade astrocytoma (P < 0.05); however, these prognostic effects were confined in grade 4 astrocytoma and GB. Moreover, acquired gains of 12 genes (including VEGFA) on 6p during relapse were associated with unfavorable prognosis for lower grade astrocytoma patients. Notably, elevated VEGFA expression during recurrence corresponded to poorer survival, validated through IHC and CGGA data. To summarize, these findings offer valuable insights into the progression of gliomas and have implications for guiding therapeutic approaches during recurrence.

A H3K27M-targeted vaccine in adults with diffuse midline glioma.

Substitution of lysine 27 to methionine in histone H3 (H3K27M) defines an aggressive subtype of diffuse glioma. Previous studies have shown that a H3K27M-specific long peptide vaccine (H3K27M-vac) induces mutation-specific immune responses that control H3K27M+ tumors in major histocompatibility complex-humanized mice. Here we describe a first-in-human treatment with H3K27M-vac of eight adult patients with progressive H3K27M+ diffuse midline glioma on a compassionate use basis. Five patients received H3K27M-vac combined with anti-PD-1 treatment based on physician's discretion. Repeat vaccinations with H3K27M-vac were safe and induced CD4+ T cell-dominated, mutation-specific immune responses in five of eight patients across multiple human leukocyte antigen types. Median progression-free survival after vaccination was 6.2 months and median overall survival was 12.8 months. One patient with a strong mutation-specific T cell response after H3K27M-vac showed pseudoprogression followed by sustained complete remission for >31 months. Our data demonstrate safety and immunogenicity of H3K27M-vac in patients with progressive H3K27M+ diffuse midline glioma.