Linda B. C. Bralten

Intrinsic Gene Expression Profiles of Gliomas Are a Better Predictor of Survival than Histology

Gliomas are the most common primary brain tumors with heterogeneous morphology and variable prognosis. Treatment decisions in patients rely mainly on histologic classification and clinical parameters. However, differences between histologic subclasses and grades are subtle, and classifying gliomas is subject to a large interobserver variability. To improve current classification standards, we have performed gene expression profiling on a large cohort of glioma samples of all histologic subtypes and grades. We identified seven distinct molecular subgroups that correlate with survival. These include two favorable prognostic subgroups (median survival, >4.7 years), two with intermediate prognosis (median survival, 1-4 years), two with poor prognosis (median survival, <1 year), and one control group. The intrinsic molecular subtypes of glioma are different from histologic subgroups and correlate better to patient survival. The prognostic value of molecular subgroups was validated on five independent sample cohorts (The Cancer Genome Atlas, Repository for Molecular Brain Neoplasia Data, GSE12907, GSE4271, and Li and colleagues). The power of intrinsic subtyping is shown by its ability to identify a subset of prognostically favorable tumors within an external data set that contains only histologically confirmed glioblastomas (GBM). Specific genetic changes (epidermal growth factor receptor amplification, IDH1 mutation, and 1p/19q loss of heterozygosity) segregate in distinct molecular subgroups. We identified a subgroup with molecular features associated with secondary GBM, suggesting that different genetic changes drive gene expression profiles. Finally, we assessed response to treatment in molecular subgroups. Our data provide compelling evidence that expression profiling is a more accurate and objective method to classify gliomas than histologic classification. Molecular classification therefore may aid diagnosis and can guide clinical decision making.

Isocitrate dehydrogenase-1 mutations: a fundamentally new understanding of diffuse glioma?

The discovery of somatic mutations in the gene encoding isocitrate dehydrogenase-1 (IDH1) in glioblastomas was remarkable because the enzyme was not previously identified with any known oncogenic pathway. IDH1 is mutated in up to 75% of grade II and grade III diffuse gliomas. Apart from acute myeloid leukaemia, other tumour types do not carry IDH1 mutations. Mutations in a homologous gene, IDH2, have also been identified, although they are much rarer. Although TP53 mutations and 1p/19q codeletions are mutually exclusive in gliomas, in both of these genotypes IDH1 mutations are common. IDH1 and IDH2 mutations are early events in the development of gliomas. Moreover, IDH1 and IDH2 mutations are a major prognostic marker for overall and progression-free survival in grade II-IV gliomas. Mutated IDH1 has an altered catalytic activity that results in the accumulation of 2-hydroxyglutarate. Molecularly, IDH1 and IDH2 mutations are heterozygous, affect only a single codon, and rarely occur together. Because IDH1 does not belong to a traditional oncogenic pathway and is specifically and commonly mutated in gliomas, the altered enzymatic activity of IDH1 may provide a fundamentally new understanding of diffuse glioma.

IDH1 R132H decreases proliferation of glioma cell lines in vitro and in vivo

A high percentage of grade II and III gliomas have mutations in the gene encoding isocitrate dehydrogenase (IDH1). This mutation is always a heterozygous point mutation that affects the amino acid arginine at position 132 and results in loss of its native enzymatic activity and gain of alternative enzymatic activity (producing D‐2‐hydroxyglutarate). The objective of this study was to investigate the cellular effects of R132H mutations in IDH1.

Segregation of Non-p.R132H Mutations in IDH1 in Distinct Molecular Subtypes of Glioma

Mutations in the gene encoding the isocitrate dehydrogenase 1 gene (IDH1) occur at a high frequency (up to 80%) in many different subtypes of glioma. In this study, we have screened for IDH1 mutations in a cohort of 496 gliomas. IDH1 mutations were most frequently observed in low grade gliomas with c.395G>A (p.R132H) representing >90% of all IDH1 mutations. Interestingly, non‐p.R132H mutations segregate in distinct histological and molecular subtypes of glioma. Histologically, they occur sporadically in classic oligodendrogliomas and at significantly higher frequency in other grade II and III gliomas. Genetically, non‐p.R132H mutations occur in tumors with TP53 mutation, are virtually absent in tumors with loss of heterozygosity on 1p and 19q and accumulate in distinct (gene‐expression profiling based) intrinsic molecular subtypes. The IDH1 mutation type does not affect patient survival. Our results were validated on an independent sample cohort, indicating that the IDH1 mutation spectrum may aid glioma subtype classification. Functional differences between p.R132H and non‐p.R132H mutated IDH1 may explain the segregation in distinct glioma subtypes.

Integrated genomic profiling identifies candidate genes implicated in glioma-genesis and a novel LEO1-SLC12A1 fusion gene.

We performed genotyping and exon‐level expression profiling on 21 glioblastomas (GBMs) and 19 oligodendrogliomas (ODs) to identify genes involved in glioma initiation and/or progression. Low‐copy number amplifications (2.5 < n < 7) and high‐copy number amplifications (n > 7) were more frequently observed in GBMs; ODs generally have more heterozygous deletions per tumor. Four high‐copy amplicons were identified in more than one sample and resulted in overexpression of the known oncogenes EGFR, MDM2, and CDK4. In the fourth amplicon, RBBP5, a member of the RB pathway, may act as a novel oncogene in GBMs. Not all hCNAs contain known genes, which may suggest that other transcriptional and/or regulatory elements are the target for amplification. Regions with most frequent allelic loss, both in ODs and GBMs, resulted in a reduced expression of known tumor suppressor genes. We identified a homozygous deletion spanning the Pragmin gene in one sample, but direct sequencing of all coding exons in 20 other glioma samples failed to detect additional genetic changes. Finally, we screened for fusion genes by identifying aberrant 5′‐3′ expression of genes that lie over regions of a copy number change. A fusion gene between exon 11 of LEO1 and exon 10 of SLC12A1 was identified. Our data show that integrated genomic profiling can identify genes involved in tumor initiation, and/or progression and can be used as an approach to identify novel fusion genes.

The CASPR2 cell adhesion molecule functions as a tumor suppressor gene in glioma

Genomic translocations have been implicated in cancer. In this study, we performed a screen for genetic translocations in gliomas based on exon-level expression profiles. We identified a translocation in the contactin-associated protein-like 2 (CASPR2) gene, encoding a cell adhesion molecule. CASPR2 mRNA was fused to an expressed sequence tag that likely is part of the nuclear receptor coactivator 1 gene. Despite high mRNA expression levels, no CASPR2 fusion protein was detected. In a set of 25 glioblastomas and 22 oligodendrogliomas, mutation analysis identified two additional samples with genetic alterations in the CASPR2 gene and all three identified genetic alterations are likely to reduce CASPR2 protein expression levels. Methylation of the CASPR2 gene was also observed in gliomas and glioma cell lines. CASPR2-overexpressing cells showed decreased proliferation rates, likely because of an increase in apoptosis. Moreover, high CASPR2 mRNA expression level is positively correlated with survival and is an independent prognostic factor. These results indicate that CASPR2 acts as a tumor suppressor gene in glioma.

Exon expression arrays as a tool to identify new cancer genes.

Background Identification of genes that are causally implicated in oncogenesis is a major goal in cancer research. An estimated 10–20% of cancer-related gene mutations result in skipping of one or more exons in the encoded transcripts. Here we report on a strategy to screen in a global fashion for such exon-skipping events using PAttern based Correlation (PAC). The PAC algorithm has been used previously to identify differentially expressed splice variants between two predefined subgroups. As genetic changes in cancer are sample specific, we tested the ability of PAC to identify aberrantly expressed exons in single samples. Principal Findings As a proof-of-principle, we tested the PAC strategy on human cancer samples of which the complete coding sequence of eight cancer genes had been screened for mutations. PAC detected all seven exon-skipping mutants among 12 cancer cell lines. PAC also identified exon-skipping mutants in clinical cancer specimens although detection was compromised due to heterogeneous (wild-type) transcript expression. PAC reduced the number of candidate genes/exons for subsequent mutational analysis by two to three orders of magnitude and had a substantial true positive rate. Importantly, of 112 randomly selected outlier exons, sequence analysis identified two novel exon skipping events, two novel base changes and 21 previously reported base changes (SNPs). Conclusions The ability of PAC to enrich for mutated transcripts and to identify known and novel genetic changes confirms its suitability as a strategy to identify candidate cancer genes.

Absence of common somatic alterations in genes on 1p and 19q in oligodendrogliomas

A common and histologically well defined subtype of glioma are the oligodendroglial brain tumors. Approximately 70% of all oligodendrogliomas have a combined loss of the entire 1p and 19q chromosomal arms. This remarkably high frequency suggests that the remaining arms harbor yet to be identified tumor suppressor genes. Identification of these causal genetic changes in oligodendrogliomas is important because they form direct targets for treatment. In this study we therefore performed targeted resequencing of all exons, microRNAs, splice sites and promoter regions residing on 1p and 19q on 7 oligodendrogliomas and 4 matched controls. Only one missense mutation was identified in a single sample in the ARHGEF16 gene. This mutation lies within- and disrupts the conserved PDZ binding domain. No similar ARHGEF16 mutations or deletions were found in a larger set of oligodendrogliomas. The absence of common somatic changes within genes located on 1p and 19q in three out of four samples indicates that no additional “second hit” is required to drive oncogenic transformation on either chromosomal arm.

Genetic Alterations in Glioma

Gliomas are the most common type of primary brain tumor and have a dismal prognosis. Understanding the genetic alterations that drive glioma formation and progression may help improve patient prognosis by identification of novel treatment targets. Recently, two major studies have performed in-depth mutation analysis of glioblastomas (the most common and aggressive subtype of glioma). This systematic approach revealed three major pathways that are affected in glioblastomas: The receptor tyrosine kinase signaling pathway, the TP53 pathway and the pRB pathway. Apart from frequent mutations in the IDH1/2 gene, much less is known about the causal genetic changes of grade II and III (anaplastic) gliomas. Exceptions include TP53 mutations and fusion genes involving the BRAF gene in astrocytic and pilocytic glioma subtypes, respectively. In this review, we provide an update on all common events involved in the initiation and/or progression across the different subtypes of glioma and provide future directions for research into the genetic changes.

Abstract 3932: Tetraploid gliomas share molecular features with pilocytic astrocytomas

Proceedings: AACR 102nd Annual Meeting 2011‐‐ Apr 2‐6, 2011; Orlando, FL We have performed expression profiling on 276 glioma samples of all histological subtypes, which resulted in the identification of seven distinct molecular subgroups. Interestingly, pilocytic astrocytomas (PAs) (n=6; adults) were assigned to one specific molecular cluster, together with four other, more malignant, gliomas. All the non-PAs were histologically diagnosed as higher grade gliomas with pilocytic features. Interestingly, there was a dramatic difference between survival of PAs and gliomas of other histological subtypes in this molecular cluster (>10.6 years vs. 3.4 (avg.) years; p = 0.03). Validation with an external dataset containing only PAs ([GSE12907][1]) showed that PAs are virtually always assigned to this molecular cluster, confirming the stability of the cluster. However, similar to our dataset, a subset of samples of both the REMBRANDT (8%) and TCGA (1%) datasets was also assigned to this molecular cluster. To further explore the differences between PAs and non-PAs in this molecular cluster, we performed genotyping using SNP 6.0 chip arrays. As reported previously, all PAs have only one larger genetic aberration; a focal amplification on locus 7q34, which is indicative for the presence of the tandem duplication KIAA1549-BRAF. One of the four samples of other histology also had this identical genetic aberration as PAs. The other (3/4) non-PA gliomas showed more genetic aberrations than the PAs. All patients harboring the KIAA1549-BRAF duplication were still alive (“survivors”) at the moment of writing this abstract (survival 10.6-19.6 years), whereas the remaining patients (“non-survivors”) all died within 0.44-2.7 years. High copy EGFR amplification was seen in none of the survivors but all of the other tumors. None of the samples in this cluster showed an IDH1-132H mutation. Closer inspection of the SNP arrays indicated that all non-survivors are tetraploid, whilst tumors of all survivors are near diploid (except for 3n on 7q34). The ploidy of all samples is currently validated using Fluorescence In Situ Hybridization (FISH). Polyploidy was not observed in any of the other molecular clusters. Validation with the REMBRANDT and the TCGA datasets showed that non-PAs assigned to this molecular cluster had a poor survival, similar to the non-PAs in our dataset. Interestingly, tetraploidy and EGFR amplification were also seen in the GBM samples from the TCGA that were assigned to this cluster. Gliomas from other molecular subtypes did not show tetraploidy on SNP chip data. In conclusion, we have discovered and validated a glioma subtype that shares molecular (RNA expression profile) and histological features with PAs. In spite of these similarities (and in contrast to the PAs), such tumors have a relatively poor prognosis. They are characterized by EGFR amplification and a near tetraploid cytogenetic profile. Identification of this specific subtype may have important therapeutic consequences. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3932. doi:10.1158/1538-7445.AM2011-3932 [1]: /lookup/external-ref?link_type=NCBIGEO&access_num=GSE12907&atom=%2Fcanres%2F71%2F8_Supplement%2F3932.atom