Subtypes Of Diffuse Glioma Research Articles

Erratum to Clinical management and survival outcomes of patients with different molecular subtypes of diffuse gliomas in China (2011-2017): a multicenter retrospective study from CGGA.

Erratum to Clinical management and survival outcomes of patients with different molecular subtypes of diffuse gliomas in China (2011-2017): a multicenter retrospective study from CGGA.

Molecular subtyping of diffuse gliomas using magnetic resonance imaging: comparison and correlation between radiomics and deep learning.

The molecular subtyping of diffuse gliomas is important. The aim of this study was to establish predictive models based on preoperative multiparametric MRI. A total of 1016 diffuse glioma patients were retrospectively collected from Beijing Tiantan Hospital. Patients were randomly divided into the training (n = 780) and validation (n = 236) sets. According to the 2016 WHO classification, diffuse gliomas can be classified into four binary classification tasks (tasks I-IV). Predictive models based on radiomics and deep convolutional neural network (DCNN) were developed respectively, and their performances were compared with receiver operating characteristic (ROC) curves. Additionally, the radiomics and DCNN features were visualized and compared with the t-distributed stochastic neighbor embedding technique and Spearman's correlation test. In the training set, areas under the curves (AUCs) of the DCNN models (ranging from 0.99 to 1.00) outperformed the radiomics models in all tasks, and the accuracies of the DCNN models (ranging from 0.90 to 0.94) outperformed the radiomics models in tasks I, II, and III. In the independent validation set, the accuracies of the DCNN models outperformed the radiomics models in all tasks (0.74-0.83), and the AUCs of the DCNN models (0.85-0.89) outperformed the radiomics models in tasks I, II, and III. DCNN features demonstrated more superior discriminative capability than the radiomics features in feature visualization analysis, and their general correlations were weak. Both the radiomics and DCNN models could preoperatively predict the molecular subtypes of diffuse gliomas, and the latter performed better in most circumstances. • The molecular subtypes of diffuse gliomas could be predicted with MRI. • Deep learning features tend to outperform radiomics features in large cohorts. • The correlation between the radiomics features and DCNN features was low.

Deep cross-view co-regularized representation learning for glioma subtype identification

The new subtypes of diffuse gliomas are recognized by the World Health Organization (WHO) on the basis of genotypes, e.g., isocitrate dehydrogenase and chromosome arms 1p/19q, in addition to the histologic phenotype. Glioma subtype identification can provide valid guidances for both risk-benefit assessment and clinical decision. The feature representations of gliomas in magnetic resonance imaging (MRI) have been prevalent for revealing underlying subtype status. However, since gliomas are highly heterogeneous tumors with quite variable imaging phenotypes, learning discriminative feature representations in MRI for gliomas remains challenging. In this paper, we propose a deep cross-view co-regularized representation learning framework for glioma subtype identification, in which view representation learning and multiple constraints are integrated into a unified paradigm. Specifically, we first learn latent view-specific representations based on cross-view images generated from MRI via a bi-directional mapping connecting original imaging space and latent space, and view-correlated regularizer and output-consistent regularizer in the latent space are employed to explore view correlation and derive view consistency, respectively. We further learn view-sharable representations which can explore complementary information of multiple views by projecting the view-specific representations into a holistically shared space and enhancing via adversary learning strategy. Finally, the view-specific and view-sharable representations are incorporated for identifying glioma subtype. Experimental results on multi-site datasets demonstrate the proposed method outperforms several state-of-the-art methods in detection of glioma subtype status.

Classification of Diffuse Glioma Subtype from Clinical-Grade Pathological Images Using Deep Transfer Learning.

Diffuse gliomas are the most common primary brain tumors and they vary considerably in their morphology, location, genetic alterations, and response to therapy. In 2016, the World Health Organization (WHO) provided new guidelines for making an integrated diagnosis that incorporates both morphologic and molecular features to diffuse gliomas. In this study, we demonstrate how deep learning approaches can be used for an automatic classification of glioma subtypes and grading using whole-slide images that were obtained from routine clinical practice. A deep transfer learning method using the ResNet50V2 model was trained to classify subtypes and grades of diffuse gliomas according to the WHO’s new 2016 classification. The balanced accuracy of the diffuse glioma subtype classification model with majority voting was 0.8727. These results highlight an emerging role of deep learning in the future practice of pathologic diagnosis.

Three Immune-Associated Subtypes of Diffuse Glioma Differ in Immune Infiltration, Immune Checkpoint Molecules, and Prognosis.

Diffuse glioma is one of the most prevalent malignancies of the brain, with high heterogeneity of tumor-infiltrating immune cells. However, immune-associated subtypes of diffuse glioma have not been determined, nor has the effect of different immune-associated subtypes on disease prognosis and immune infiltration of diffuse glioma patients. We retrieved the expression profiles of immune-related genes from The Cancer Genome Atlas (TCGA) (n = 672) and GSE16011 (n = 268) cohorts and used them to identify subtypes of diffuse glioma via Consensus Cluster Plus analysis. We used the limma, clusterProfiler, ESTIMATE, and survival packages of R for differential analysis, functional enrichment, immune and stromal score evaluation respectively in three subtypes, and performed log-rank tests in immune subtypes of diffuse glioma. The immune-associated features of diffuse glioma in the two cohorts were characterized via bioinformatic analyses of the mRNA expression data of immune-related genes. Three subtypes (C1–3) of diffuse glioma were identified from TCGA data, and were verified using the GSE16011 cohort. We then evaluated their immune characteristics and clinical features. Our mRNA profiling analyses indicated that the different subtypes of diffuse glioma presented differential expression profile of specific genes and signal pathways in the TCGA cohort. Patients with subtype C1, who were mostly diagnosed with grade IV glioma, had poorer outcomes than patients with subtype C2 or C3. Subtype C1 was characterized by a higher degree of immune cell infiltration as estimated by GSVA, and more frequent wildtype IDH1. By contrast, subtype C3 included more grade II and IDH1-mutated glioma, and was associated with more infiltration of CD4+T cells. Most subtype C2 had the features between subtypes C1 and C3. Meanwhile, immune checkpoints and their ligand molecules, including PD1/(PD-L1/PDL2), CTLA4/(CD80/CD86), and B7H3/TLT2, were significantly upregulated in subtype C1 and downregulated in subtype C3. In addition, patients with subtype C1 exhibited more frequent gene mutations. Univariate and multivariate Cox regression analyses revealed that diffuse glioma subtype was an effective, independent, and better prognostic factor. Therefore, we established a novel immune-related classification of diffuse glioma, which provides potential immunotherapy targets for diffuse glioma.

EPCO-08. TUMOR-IMMUNE INTERACTIONS ARE DYNAMIC AND INFLUENCE THE EVOLUTIONARY TRAJECTORY OF ADULT DIFFUSE GLIOMA

Abstract Diffuse glioma is an aggressive brain cancer that is characterized by a poor prognosis and a universal resistance to therapy. The evolutionary processes behind this resistance remain unclear. Previous studies by the Glioma Longitudinal Analysis (GLASS) Consortium have indicated that therapy-induced selective pressures shape the genetic evolution of glioma in a stochastic manner. However, single cell studies have revealed that malignant glioma cells are highly plastic, and capable of changing their cell state in response to diverse challenges in their microenvironment. The tumor immune response has been implicated as a major driver of these malignant cell state transitions, and is known to be affected by the administration of therapy, but the extent to which tumor genetics, therapy, and the different components of the immune response interact to influence a glioma’s evolutionary trajectory are poorly understood. To further investigate these factors, we collected DNA and RNA sequencing data on pre- and post-treatment tumor pairs from over 150 glioma patients that have received chemotherapy, radiotherapy, and/or immune checkpoint blockade agents. By integrating mutation, copy number, and in silico deconvolution analyses of bulk transcriptome data across the three molecular subtypes of diffuse glioma, we show that longitudinal increases in chromosomal instability and gene fusions associate with decreased immune infiltrate and altered cell states at recurrence. We additionally find that specific molecular alterations and malignant cell states associate with unique inflammatory and immunosuppressive programs in tumor-associated macrophages and microglia. Lastly, we show that the abundance of T cells in the tumor microenvironment does not associate with changes in neoantigen depletion and the acquisition of antigen presentation machinery defects, suggesting minimal immunoediting activity over time. Collectively, our results indicate that the administration of therapy can alter the dynamics of tumor-immune interactions in glioma, resulting in new steady-states at recurrence that can be subsequently targeted.

Adult diffuse midline gliomas: Clinical, radiological, and genetic characteristics

Diffuse midline gliomas (DMGs) are a diffuse glioma subtype arising from midline brain structures. It is predominantly a disease of childhood; however, it can also occur in adults. Adult DMG has not been previously well described. The aim of this study was to define the characteristics of adult DMG. We described and analyzed the clinical, radiological, and genetic alterations of 9 adult DMGs and compared them with those of 257 non-midline adult high-grade IDH-WT gliomas. The median age of all patients was 38-years old (23–68-years). Most common symptoms were headache, motor/sensory deficit, ataxia, cranial nerve deficit, and confusion. Tumor locations were brainstem (44.5%), thalamus (22.2%), pineal region (22.2%), spinal cord (22.2%), and cerebellum (11.1%). Six-patients (66.7%) were H3 K27M-WT and three (33.3%) were H3 K27M-mutant. In addition to H3 K27M mutations, TP53 gene (55.5%), CDKN2A/B and TERTp (33.3%), PDGFRA (33.3%), PIK3CA, PTEN, KDR, NF1, and MYC (22.2%) were the most frequently mutated genes. Neither IDH1/IDH2 nor EGFR alterations were present. Compared to non-midline high-grade glioma, adult DMG patients were younger (38 vs 61 years, p < 0.001) and lacked EGFR-alterations (0/9 vs 123/257, p = 0.004). The median survival of DMG and non-midline high-grade gliomas was 19 and 18 months respectively (p = 0.964). Our data support that adult DMGs have different oncogenic drivers compared to non-midline high-grade gliomas. Regardless of H3 K27M mutation status, neither of the nine adult DMG cases demonstrated IDH1/IDH2 or EGFR alterations. Larger multi-institutional studies are needed to further characterize the biology of this rare type of diffuse glioma in adults.

Molecular Subtyping of Diffuse Gliomas Using Magnetic Resonance Imaging: Comparison and Correlation Between Radiomics and Deep Learning

The 2016 WHO classification of diffuse gliomas is significant. The aim of this study was to establish comprehensive predictive models based on preoperative multi-parametric MRI. A total of 1,016 diffuse glioma patients were retrospectively collected from Beijing Tiantan Hospital, China. The patients were randomly divided into a training set (n = 780) and an internal validation set (n = 236). According to the 2016 WHO classification, the diffuse gliomas can be categorized according to four binary classification tasks (tasks I—IV). Predictive models based on radiomics and deep convolutional neural networks (DCNN) were developed respectively, and their performances were compared using receiver operating characteristic (ROC) curves. Additionally, the radiomics and DCNN features were visualized and compared based on t-distributed stochastic neighbor embedding technique and Spearman’s rank-correlation test. In the training set, areas under the curves (AUCs) of the DCNN models (ranging from 0.99—1.00) outperformed those of the radiomics models in all conducted tasks. In the independent validation set, the AUCs of the DCNN models outperformed those of the radiomics models in tasks I, II, and III. DCNN features demonstrated superior discriminative capability compared to radiomics features based on feature visualization analysis, and their general correlations were weak. In conclusion, both the radiomics and DCNN models could preoperatively predict the molecular subtypes of diffuse gliomas, but the latter models performed better in most circumstances. Funding Statement: This study was supported by the National Natural Science Foundation of China (No. 81601452). Declaration of Interests: No conflict of interest to declare. Ethics Approval Statement: Ethical approval for this retrospective study was received from the institutional review board of the Beijing Tiantan Hospital.

Longitudinal molecular trajectories of diffuse glioma in adults.

The evolutionary processes that drive universal therapeutic resistance in adult patients with diffuse glioma remain unclear1,2. Here, we analyzed temporally separated DNA sequencing data and matched clinical annotation from 222 patients with glioma. Through mutational and copy number analyses across the three major subtypes of diffuse glioma, we observed that driver genes detected at initial disease were retained at recurrence, while there was little evidence of recurrence-specific gene alterations. Treatment with alkylating-agents resulted in a hypermutator phenotype at different rates across glioma subtypes, and hypermutation was not associated with differences in survival. Acquired aneuploidy was frequently detected in recurrent gliomas characterized by presence of an IDH mutation but without 1p/19q codeletion and further converged with acquired cell cycle alterations and poor outcomes. We show that the clonal architecture of each tumor remains similar over time and that absence of clonal selection was associated with increased survival. Finally, we did not observe differences in immunoediting levels between initial and recurrent glioma. Our results collectively argue that the strongest selective pressures occur early during glioma development and that current therapies shape this evolution in a largely stochastic manner.

Rapid intraoperative molecular genetic classification of gliomas using Raman spectroscopy.

BackgroundThe molecular genetic classification of gliomas, particularly the identification of isocitrate dehydrogenase (IDH) mutations, is critical for clinical and surgical decision-making. Raman spectroscopy probes the unique molecular vibrations of a sample to accurately characterize its molecular composition. No sample processing is required allowing for rapid analysis of tissue. The aim of this study was to evaluate the ability of Raman spectroscopy to rapidly identify the common molecular genetic subtypes of diffuse glioma in the neurosurgical setting using fresh biopsy tissue. In addition, classification models were built using cryosections, formalin-fixed paraffin-embedded (FFPE) sections and LN-18 (IDH-mutated and wild-type parental cell) glioma cell lines.MethodsFresh tissue, straight from neurosurgical theatres, underwent Raman analysis and classification into astrocytoma, IDH-wild-type; astrocytoma, IDH-mutant; or oligodendroglioma. The genetic subtype was confirmed on a parallel section using immunohistochemistry and targeted genetic sequencing.ResultsFresh tissue samples from 62 patients were collected (36 astrocytoma, IDH-wild-type; 21 astrocytoma, IDH-mutated; 5 oligodendroglioma). A principal component analysis fed linear discriminant analysis classification model demonstrated 79%–94% sensitivity and 90%–100% specificity for predicting the 3 glioma genetic subtypes. For the prediction of IDH mutation alone, the model gave 91% sensitivity and 95% specificity. Seventy-nine cryosections, 120 FFPE samples, and LN18 cells were also successfully classified. Meantime for Raman data collection was 9.5 min in the fresh tissue samples, with the process from intraoperative biopsy to genetic classification taking under 15 min.ConclusionThese data demonstrate that Raman spectroscopy can be used for the rapid, intraoperative, classification of gliomas into common genetic subtypes.

Gliomatosis cerebri

A 21-year-old immunocompetent woman in apparent good health and without a significant past medical history presented with progressive headaches and horizontal diplopia. She was alert, oriented, and the remainder of the neurologic exam was unremarkable. Based on her brain MRI, what is the diagnosis?a)Acute viral encephalitisb)Multiple sclerosisc)Progressive multifocal leukoencephalopathyd)Gliomatosis cerebrie)Primary CNS lymphoma Stereotactic biopsy demonstrated WHO grade III astrocytoma, IDH mutant, ATRX lost. Her extensive involvement evidenced by imaging suggests a gliomatosis cerebri growth pattern. Gliomatosis cerebri refers to an extensive and invasive growth pattern that involves at least three contiguous lobes and typically resembles a low-grade lesion radiologically (e.g., non-contrast enhancing), but is associated with a poor prognosis [[1]Georgakis M.K. Spinos D. Pourtsidis A. Psyrri A. Panourias I.G. Sgouros S. et al.Incidence and survival of gliomatosis cerebri: a population-based cancer registration study.J Neurooncol. 2018; 138: 341-349Crossref PubMed Scopus (15) Google Scholar]. Until recently, it was classified separately within gliomas, but molecular studies did not warrant separation [[2]Cavenee W.K. Louis D.N. Ohgaki H. Wiestler O.D. International Agency for Research on Cancer. WHO classification of tumours of the central nervous system.Revised 4th ed. International Agency For Research On Cancer, Lyon2016Google Scholar]. It can be present in any diffuse glioma subtype, but it is especially common in anaplastic astrocytoma. Gliomatosis cerebri is generally categorized in two groups based on having a focal tumor mass: type 1 (no focal mass), and type 2 (with a focal mass). Treatment is challenging due to the extensiveness and irresectability of the lesion. Radiation therapy is especially restricted due to the extensive involvement, which often requires dosage adjusting, decreased length of therapy, and have a high likelihood of causing neurotoxicities [[3]Morales La Madrid A. Ranjan S. Warren K.E. Gliomatosis cerebri: a consensus summary report from the Second International Gliomatosis cerebri Group Meeting, June 22–23, 2017, Bethesda, USA.J Neuro-oncol. 2018; Crossref Scopus (9) Google Scholar]. Lower doses of ionizing radiation may also reduce the effectiveness of chemotherapy, as the length of concurrent radiation can be reduced. Our patient being younger, received a full dose of radiation (59.4 Gy) with concurrent temozolomide for 6 weeks, followed by maintenance temozolomide for one year. She tolerated treatment well, but unfortunately progressed soon afterwards and passed away after initiating bevacizumab. Gliomatosis cerebri remains a therapeutic challenge that warrants further investigation (Fig. 1). Dr. Garcia: Concept, design, drafting, revising, and critical revision of content. Dr. Villano: Concept, design, drafting, revising, and critical revision of content. The authors have no disclosures.

Machine Learning-Based Radiomics for Molecular Subtyping of Gliomas.

Purpose: The new classification announced by the World Health Organization in 2016 recognized five molecular subtypes of diffuse gliomas based on isocitrate dehydrogenase (IDH) and 1p/19q genotypes in addition to histologic phenotypes. We aim to determine whether clinical MRI can stratify these molecular subtypes to benefit the diagnosis and monitoring of gliomas.Experimental Design: The data from 456 subjects with gliomas were obtained from The Cancer Imaging Archive. Overall, 214 subjects, including 106 cases of glioblastomas and 108 cases of lower grade gliomas with preoperative MRI, survival data, histology, IDH, and 1p/19q status were included. We proposed a three-level machine-learning model based on multimodal MR radiomics to classify glioma subtypes. An independent dataset with 70 glioma subjects was further collected to verify the model performance.Results: The IDH and 1p/19q status of gliomas can be classified by radiomics and machine-learning approaches, with areas under ROC curves between 0.922 and 0.975 and accuracies between 87.7% and 96.1% estimated on the training dataset. The test on the validation dataset showed a comparable model performance with that on the training dataset, suggesting the efficacy of the trained classifiers. The classification of 5 molecular subtypes solely based on the MR phenotypes achieved an 81.8% accuracy, and a higher accuracy of 89.2% could be achieved if the histology diagnosis is available.Conclusions: The MR radiomics-based method provides a reliable alternative to determine the histology and molecular subtypes of gliomas. Clin Cancer Res; 24(18); 4429-36. ©2018 AACR.

The genetic landscape of gliomas arising after therapeutic radiation.

Radiotherapy improves survival for common childhood cancers such as medulloblastoma, leukemia, and germ cell tumors. Unfortunately, long-term survivors suffer sequelae that can include secondary neoplasia. Gliomas are common secondary neoplasms after cranial or craniospinal radiation, most often manifesting as high-grade astrocytomas with poor clinical outcomes. Here, we performed genetic profiling on a cohort of 12 gliomas arising after therapeutic radiation to determine their molecular pathogenesis and assess for differences in genomic signature compared to their spontaneous counterparts. We identified a high frequency of TP53 mutations, CDK4 amplification or CDKN2A homozygous deletion, and amplifications or rearrangements involving receptor tyrosine kinase and Ras-Raf-MAP kinase pathway genes including PDGFRA, MET, BRAF, and RRAS2. Notably, all tumors lacked alterations in IDH1, IDH2, H3F3A, HIST1H3B, HIST1H3C, TERT (including promoter region), and PTEN, which genetically define the major subtypes of diffuse gliomas in children and adults. All gliomas in this cohort had very low somatic mutation burden (less than three somatic single nucleotide variants or small indels per Mb). The ten high-grade gliomas demonstrated markedly aneuploid genomes, with significantly increased quantity of intrachromosomal copy number breakpoints and focal amplifications/homozygous deletions compared to spontaneous high-grade gliomas, likely as a result of DNA double-strand breaks induced by gamma radiation. Together, these findings demonstrate a distinct molecular pathogenesis of secondary gliomas arising after radiation therapy and identify a genomic signature that may aid in differentiating these tumors from their spontaneous counterparts.

Epitope mapping of an anti-alpha thalassemia/mental retardation syndrome X-linked monoclonal antibody AMab-6

The alpha-thalassemia/mental-retardation-syndrome-X-linked (ATRX) gene is located on the q arm of the X chromosome. ATRX gene mutations were first discovered in pancreatic neuroendocrine tumors, and subsequently in other cancer subtypes, including gliomas. Molecular subgrouping of gliomas has been more important than conventional histological classifications. Mutations in the isocitrate dehydrogenase (IDH), telomerase reverse transcriptase (TERT) promoter, and ATRX and the codeletion of chromosomes 1p/19q are used as biomarkers for diagnosing the subtypes of diffuse gliomas. We recently developed a sensitive monoclonal antibody (mAb) AMab-6 against ATRX by immunizing mice with recombinant human ATRX. AMab-6 can help to detect ATRX mutations via Western blotting and immunohistochemical analyses. In this study, we characterized the binding epitope of AMab-6 using enzyme-linked immunosorbent assay (ELISA), Western blotting, and immunohistochemical analysis, and found that Gln2368 of ATRX is critical for AMab-6 binding to ATRX. Our findings could be applied to the production of more functional anti-ATRX mAbs.

Novel internal regulators and candidate miRNAs within miR-379/miR-656 miRNA cluster can alter cellular phenotype of human glioblastoma

Clustered miRNAs can affect functioning of downstream pathways due to possible coordinated function. We observed 78–88% of the miR-379/miR-656 cluster (C14MC) miRNAs were downregulated in three sub-types of diffuse gliomas, which was also corroborated with analysis from The Cancer Genome Atlas (TCGA) datasets. The miRNA expression levels decreased with increasing tumor grade, indicating this downregulation as an early event in gliomagenesis. Higher expression of the C14MC miRNAs significantly improved glioblastioma prognosis (Pearson’s r = 0.62; p < 3.08e-22). ENCODE meta-data analysis, followed by reporter assays validated existence of two novel internal regulators within C14MC. CRISPR activation of the most efficient internal regulator specifically induced members of the downstream miRNA sub-cluster and apoptosis in glioblastoma cells. Luciferase assays validated novel targets for miR-134 and miR-485-5p, two miRNAs from C14MC with the most number of target genes relevant for glioma. Overexpression of miR-134 and miR-485-5p in human glioblastoma cells suppressed invasion and proliferation, respectively. Furthermore, apoptosis was induced by both miRs, individually and in combination. The results emphasize the tumor suppressive role of C14MC in diffuse gliomas, and identifies two specific miRNAs with potential therapeutic value and towards better disease management and therapy.

Contrast enhancement predicting survival in integrated molecular subtypes of diffuse glioma: an observational cohort study

To assess the predictive value of magnetic resonance imaging (MRI) gadolinium enhancement as a prognostic factor in the 2016 World Health Organization Classification of Tumors of the Central Nervous System integrated glioma groups. Four-hundred fifty patients with histopathologically confirmed glioma were retrospectively assessed between 07/1997 and 06/2014 using gadolinium enhancement, survival, and relevant prognostic molecular data [isocitrate dehydrogenase (IDH); alpha-thalassemia/mental retardation syndrome X-linked (ATRX); chromosome 1p/19q loss of heterozygosity; and O6-methylguanine DNA methyltransferase (MGMT)]. The Kaplan-Meier method was used to assess univariate survival data. A multivariate Cox proportional hazards model was performed on significant results from the univariate analysis. There were significant differences in survival between patient age (p < 0.0001), WHO glioma grades (p < 0.0001), and integrated molecular profiles (p < 0.0001). Patients with IDH1/2 mutation, loss of ATRX expression, and methylated MGMT promoter showed significantly better survival than those with the IDHwild-type (p < 0.0001), retained ATRX expression (p < 0.0001), and unmethylated MGMT promoter (p = 0.019). Survival was significantly better in patients without gadolinium enhancement (p = 0.009) who were in the IDHwild-type glioma and glioma with retained ATRX expression groups (p = 0.018 and 0.030, respectively). In univariate analysis, the presence of gadolinium enhancement on preoperative MRI scans is an unfavorable factor for survival. Regarding the molecular subgroups, gadolinium enhancement is an unfavorable prognostic factor in gliomas with IDHwild-type and those with ATRX retention. However, in multivariate analysis only patient age, IDH1/2 mutation status, MGMT promoter methylation status, and WHO grade IV are relevant for predicting survival.

Perfusion and diffusion MRI signatures in histologic and genetic subtypes of WHO grade II–III diffuse gliomas

The value of perfusion and diffusion-weighted MRI in differentiating histological subtypes according to the 2007 WHO glioma classification scheme (i.e. astrocytoma vs. oligodendroglioma) and genetic subtypes according to the 2016 WHO reclassification (e.g. 1p/19q co-deletion and IDH1 mutation status) in WHO grade II and III diffuse gliomas remains controversial. In the current study, we describe unique perfusion and diffusion MR signatures between histological and genetic glioma subtypes. Sixty-five patients with 2007 histological designations (astrocytomas and oligodendrogliomas), 1p/19q status (+ = intact/- = co-deleted), and IDH1 mutation status (MUT/WT) were included in this study. In all patients, median relative cerebral blood volume (rCBV) and apparent diffusion coefficient (ADC) were estimated within T2 hyperintense lesions. Bootstrap hypothesis testing was used to compare subpopulations of gliomas, separated by WHO grade and 2007 or 2016 glioma classification schemes. A multivariable logistic regression model was also used to differentiate between 1p19q+ and 1p19q- WHO II-III gliomas. Neither rCBV nor ADC differed significantly between histological subtypes of pure astrocytomas and pure oligodendrogliomas. ADC was significantly different between molecular subtypes (p = 0.0016), particularly between IDHWT and IDHMUT/1p19q+ (p = 0.0013). IDHMUT/1p19q+ grade III gliomas had higher median ADC; IDHWT grade III gliomas had higher rCBV with lower ADC; and IDHMUT/1p19q- had intermediate rCBV and ADC values, similar to their grade II counterparts. A multivariable logistic regression model was able to differentiate between IDHWT and IDHMUT WHO II and III gliomas with an AUC of 0.84 (p < 0.0001, 74% sensitivity, 79% specificity). Within IDHMUT WHO II-III gliomas, a separate multivariable logistic regression model was able to differentiate between 1p19q+ and 1p19q- WHO II-III gliomas with an AUC of 0.80 (p = 0.0015, 64% sensitivity, 82% specificity). ADC better differentiated between genetic subtypes of gliomas according to the 2016 WHO guidelines compared to the classification scheme outlined in the 2007 WHO guidelines based on histological features of the tissue. Results suggest a combination of rCBV, ADC, T2 hyperintense volume, and presence of contrast enhancement together may aid in non-invasively identifying genetic subtypes of diffuse gliomas.

DNA methylation signatures for 2016 WHO classification subtypes of diffuse gliomas

BackgroundGlioma is the most common of all primary brain tumors with poor prognosis and high mortality. The 2016 World Health Organization classification of the tumors of central nervous system uses molecular parameters in addition to histology to redefine many tumor entities. The new classification scheme divides diffuse gliomas into low-grade glioma (LGG) and glioblastoma (GBM) as per histology. LGGs are further divided into isocitrate dehydrogenase (IDH) wild type or mutant, which is further classified into either oligodendroglioma that harbors 1p/19q codeletion or diffuse astrocytoma that has an intact 1p/19q loci but enriched for ATRX loss and TP53 mutation. GBMs are divided into IDH wild type that corresponds to primary or de novo GBMs and IDH mutant that corresponds to secondary or progressive GBMs. To make the 2016 WHO subtypes of diffuse gliomas more robust, we carried out Prediction Analysis of Microarrays (PAM) to develop DNA methylation signatures for these subtypes.ResultsIn this study, we applied PAM on a training set of diffuse gliomas derived from The Cancer Genome Atlas (TCGA) and identified DNA methylation signatures to classify LGG IDH wild type from LGG IDH mutant, LGG IDH mutant with 1p/19q codeletion from LGG IDH mutant with intact 1p/19q loci and GBM IDH wild type from GBM IDH mutant with an accuracy of 99–100%. The signatures were validated using the test set of diffuse glioma samples derived from TCGA with an accuracy of 96 to 99%. In addition, we also carried out additional validation of all three signatures using independent LGG and GBM cohorts. Further, the methylation signatures identified a fraction of samples as discordant, which were found to have molecular and clinical features typical of the subtype as identified by methylation signatures.ConclusionsThus, we identified methylation signatures that classified different subtypes of diffuse glioma accurately and propose that these signatures could complement 2016 WHO classification scheme of diffuse glioma.

Multigene signature for predicting prognosis of patients with 1p19q co-deletion diffuse glioma.

Co-deletion of 1p and 19q marks a diffuse glioma subtype associated with relatively favorable overall survival; however, heterogeneous clinical outcomes are observed within this category. We assembled gene expression profiles and sample annotation of 374 glioma patients carrying the 1p/19q co-deletion. We predicted 1p/19q status using gene expression when annotation was missing. A first cohort was randomly split into training (n = 170) and a validation dataset (n = 163). A second validation set consisted of 41 expression profiles. An elastic-net penalized Cox proportional hazards model was applied to build a classifier model through cross-validation within the training dataset. The selected 35-gene signature was used to identify high-risk and low-risk groups in the validation set, which showed significantly different overall survival (P = .00058, log-rank test). For time-to-death events, the high-risk group predicted by the gene signature yielded a hazard ratio of 1.78 (95% confidence interval, 1.02-3.11). The signature was also significantly associated with clinical outcome in the The Cancer Genome Atlas (CGA) IDH-mutant 1p/19q wild-type and IDH-wild-type glioma cohorts. Pathway analysis suggested that high risk was associated with increased acetylation activity and inflammatory response. Tumor purity was found to be significantly decreased in high-risk IDH-mutant with 1p/19q co-deletion gliomas and IDH-wild-type glioblastomas but not in IDH-wild-type lower grade or IDH-mutant, non-co-deleted gliomas. We identified a 35-gene signature that identifies high-risk and low-risk categories of 1p/19q positive glioma patients. We have demonstrated heterogeneity amongst a relatively new glioma subtype and provided a stepping stone towards risk stratification.

A mesenchymal/stem cell predictor of survival of patients with malignant gliomas.

2049 Background: Identification of the molecular subtypes of diffuse gliomas will guide therapy and risk stratification. Attempts to dissect the biologic diversity of these tumors using array data have been limited to small cohort sizes or to a single histologic subtype. We proposed to overcome these limitations by unifying data from multiple series. Methods: Public and unpublished Affymetrix arrays data were compiled and samples with raw intensity files and clinical annotation were included. Unsupervised clustering was performed to identify major subgroups. Data from 517 tumors were randomly divided into training and validation sets for predictive modeling. Candidate models were evaluated by cross-validation. An independent verification cohort (n = 328) was used for further validation. Stem cell array data were obtained from published sources. Uni/multivariate correlations and survival analyses were performed to published radiation-associated and oncogene-pathway signatures. Results: Two primary molecular subtypes of gliomas, each with distinct survival outcomes, were observed by unsupervised analyses. Genes in the subtype whose expression correlated with poorer survival were associated mesenchymal ontology and characterized the majority of, in contrast to the lower grade tumors. The signature correlated with a glioma stem cells (GSC) profile. Predictive modeling identified a robust prognostic model in both training and primary validation sets (accuracy > 85%) and was validated on an independent data set, including cases from the Cancer Genome Atlas. The mesenchymal signature was correlated with radiation and ras and src oncogene pathway signatures (p < 0.0001) and was an independent predictors of survival. Conclusions: Using the largest reported unified glioma expression profiling dataset we identified a mesenchymal signature associated with poor survival and which correlated to a GSC-specific signature. We used this dataset to develop a robust prognostic marker of patient survival which outperforms existing clinico-pathologic factors, such as age and grade, and which was validated on an independent dataset. Based on its radiation/oncogene association, this predictor may be useful in guiding treatment. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Castle Biosciences, Inc.