Glioblastoma Genetics: In Rapid Flux

Abstract

Glioblastoma is the most common and most lethal primary brain tumor. While small progress has been made in treating this cancer in recent years, glioblastoma remains largely resistant to all existing therapies. It has been hoped that dissection of the genetics of this cancer would lead to more targeted and effective treatments, and new advances may finally be bringing this closer to fruition. Within the last few years, high-throughput efforts such as The Cancer Genome Atlas and a massive sequencing project have yielded novel insights and classifications of this dreaded cancer. The likely impact on care delivery in the clinic may only be a few years away. The rapid and exciting pace of advances in glioblastoma genetics has prompted this up-to-date review.

Introduction

Glioblastoma multiforme (GBM) is the most common and lethal primary brain tumor. While not as prevalent as some of the epithelial carcinomas, GBM is resistant to current therapy and is nearly universally fatal. Current treatment includes surgery, radiation, and the chemotherapy drug temozolomide, and at disease recurrence these are often followed with repeat surgery or treatment with the antiangiogenic drug bevacizumab plus/minus chemotherapy. Even with the most sophisticated treatment, median survival for patients with GBM is approximately 15 months from diagnosis (Stupp et al., 2005). This has appropriately led to intense efforts to better understand the underpinnings of this malignancy. While many important genetic lesions in glioblastoma have been known for years or even decades, new high-throughput technologies have allowed dramatic new advances to be made in a fastpaced fashion. We recently summarized the current understanding of GBM genetics in a review article last year (Purow et al., 2009), but several exciting reports since then already necessitate a fresh look at this important subject. In this review we will provide an updated view of the genetics of glioblastoma, with particular attention to the growing implications for prognosis, classification, and treatment.

“Classic” GBM Genetics

In the last two to three decades some general genetic distinctions were made between primary and secondary GBMs, which occur de novo or arise from low-grade gliomas, respectively. Among the classic features of primary GBMs is amplification or activating mutations of the epidermal growth factor receptor (EGFR) or both (Ekstrand et al., 1991; Libermann et al., 1985; Nishikawa et al., 1994). The altered EGFR gene or its expression acts as an oncogene in numerous cancer types. Expectations were high for inhibitors of this receptor tyrosine kinase (RTK) when they reached the clinic, but they proved rarely effective in clinical trials in patients with GBM (Haas-Kogan et al., 2005; Prados et al., 2006). Amplification of the MDM2 oncogene is also present in a smaller but significant percentage of primary GBMs, resulting in inhibition of the p53 tumor suppressor (He et al., 1994; Reifenberger et al., 1993). Secondary GBMs frequently inactivate p53 as well, but through a more direct mechanism-mutation of the gene itself (Watanabe et al., 1996). Less commonly, secondary GBMs also display amplification of the oncogenic RTK platelet-derived growth factor receptor (PDGFR), potentially in conjunction with over-expression of the ligand PDGF to yield an autocrine loop (Fleming et al., 1992; Saxena et al., 1999). Several genetic lesions were noted to be present in both primary and secondary GBMs, including frequent deletion or mutation of the PTEN tumor suppressor that leads to up-regulation of the powerful Akt oncogenic pathway (Li et al., 1997; Liu et al., 1997). Another common genetic lesion found often across both primary and secondary GBMs is homozygous deletion of the CDKN2A gene (Schmidt et al., 1994), which encodes the distinct tumor suppressors p16INK4A and p14ARF. p16INK4A inhibits the activity of the cell cycle, while p14ARF inhibits MDM2 and thus increases p53 expression (Kamijo et al., 1997; Zhang et al., 1998).

Familial Syndromes

While most GBMs arise without a clear etiology, in some cases they occur at higher incidence in families with identified genetic syndromes. In most of these cases, the connections to known genetic lesions and pathways linked to GBM are clear. These include Neurofibromatosis 1 due to mutations in the NF1 gene (Guillamo et al., 2003), Li-Fraumeni syndrome 1 due to mutations in the p53 gene (Watanabe, et al., 1996; Zhou et al., 1999), melanoma-astrocytoma syndrome due to lesions in CDKN2A (Bahuau et al., 1998), and hereditary nonpolyposis colorectal cancer syndrome (HNPCC, also known as Lynch syndrome) due to mutations in DNA mismatch repair genes such as MSH2 and MSH6 (Watson et al., 2008). It seems quite possible, even likely, that there remain to be discovered other genetic syndromes predisposing to GBM. More recently, new discoveries are emerging of single-nucleotide polymorphisms (SNPs), some relatively common, that increase the risk of GBM. These notably include SNPs in DNA repair genes (Bethke et al., 2008; Liu et al., 2007; Liu et al., 2008; Wrensch et al., 2005), metabolic genes (Bethke et al., 2008; De Roos et al., 2003; Ezer et al., 2002; Lai et al., 2005), and inflammation-related genes (Scheurer et al., 2008; Schwartzbaum et al., 2005), though none of these have a well-established role at this time and remain to be confirmed by other reports.

Output from Recent Efforts

While incremental discoveries in GBM genetics were made in the last 10–15 years through traditional methods, it required large multi-group collaborations using the most modern high-throughput methods to begin rapidly extending our understanding toward a more comprehensive picture. Foremost among these efforts has been The Cancer Genome Atlas (TCGA), a national effort funded by the NIH to thoroughly profile three cancer types using multiple modalities. These have included studies of gene and microRNA expression with microarrays, copy number analysis with array CGH (comparative genomic hybridization), and sequencing of over 600 genes in dozens to hundreds of banked high-quality tumor specimens. GBM was one of the cancer types initially investigated, and important results began to be published in the fall of 2008 (McLendon et al., 2008). This initial TCGA publication revealed some notable findings. p53 mutation was found more frequently than expected in primary GBMs, and in about 25% of GBMs overall. Mutations in the neurofibromatosis 1 gene (NF1), a tumor suppressor gene that suppresses the oncogenic Ras pathway, had been thought to be rare in GBM but were actually noted in 18% of them. A few oncogenes were found to be amplified at a higher-than expected rate, including ErbB2, c-Met, and PIK3R1 (a component of PI3 kinase, which activates the Akt oncogene). In an overall analysis of these initial TCGA results, nearly all GBMs were found to have lesions in three general pathways: activating lesions in oncogenic receptor tyrosine kinase pathways, activating lesions in the cell cycle pathway, and inhibitory lesions in the p53 tumor suppressor pathway.

At the time the first TCGA findings were reported in GBM, a smaller multi-group collaboration spearheaded by investigators at The Johns Hopkins University published findings from unbiased sequencing of most genes in 22 GBMs (Parsons et al., 2008). A truly striking result emerged from this study-a very frequent point mutation in secondary GBMs in the metabolic gene isocitrate dehydrogenase 1 (IDH1), which never before was linked to cancer. Several important publications have followed, showing that a majority of low-grade gliomas and secondary GBMs have this IDH1 point mutation, and a few that do not instead have an equivalent mutation in the related IDH2 gene (Hartmann et al., 2009; Yan et al., 2009). IDH1 mutations have been associated with better prognosis (Dubbink et al., 2009; Nobusawa et al., 2009; Sanson et al., 2009; Weller et al., 2009; Wick et al., 2009). Intense efforts are underway to understand the function of this mutation. While it was initially thought that it might operate through reduced synthesis of the metabolite alpha-ketoglutarate (Zhao et al., 2009), a dramatic new report demonstrates that the mutant IDH1 produces an alternative metabolite 2-hydroxyglutarate (2-HG) (Dang et al., 2009). This breakthrough begs the question of whether treatments that reduce 2-HG production will be effective against gliomas with IDH1 mutation, and the hunt is already on to find such agents.

Molecular Classification of GBM

Large-scale profiling and sequencing results have also been used in attempts to divide GBM into molecular subgroups that could shed light on their mechanisms and sensitivities to certain therapies. A seminal report a few years ago used gene expression profiling to divide GBM into three types: proneural, mesenchymal, and proliferative (Phillips et al., 2006). This was expanded upon in a new publication that utilizes TCGA data to derive four subsets of GBMs: classic, proneural, neural, and mesenchymal (Verhaak et al., 2010). Importantly, frequencies of common mutations were also associated with the various subtypes, potentially illuminating their genetic origins. The classic subtype is associated with EGFR amplification (chromosome 7) and/or mutation, PTEN loss (chromosome 10), and CDKN2A deletion. The proneural subtype demonstrates frequent IDH1 mutations and lesions in the PDGFRA gene. The neural subtype is marked by expression of neuronal genes and lacks a distinctive genetic profile. The mesenchymal subtype is distinguished by the most frequent occurrence of hemizygous NF1 deletions. This updated classification of GBM subtypes, with the inclusion of mutation frequencies, offers new insights into the genetic heterogeneity of GBM; while not yet “personalized medicine,” it may still provide leads in matching the right treatment to each patient’s tumor.

More Undiscovered Frontiers

While it may seem that the struggle to fully understand GBM genetics is nearing an endpoint, there are new frontiers that have barely been explored and others that may lie undiscovered. For example, recent studies are still identifying important genes at genetic “hot spots,” regions of the genome in GBM patients where genetic amplifications or deletions have been found. One new report identifies a new tumor suppressor with suppressor activity of EGFR, the ANXA7 gene, in a region of chromosome 10 long known to be frequently deleted in GBMs (Yadav et al., 2009).

The relatively recent discovery of microRNAs has opened up a major new front in the war on GBM and other cancers. MicroRNAs are small hairpin RNAs encoded in our genome that regulate, sometimes potently, the expression of many of our genes. These microRNAs can act as oncogenes or tumor suppressors, and a growing body of work is demonstrating key roles for them in GBM (Chan et al., 2005; Ciafre et al., 2005; Gillies et al., 2007; Godlewski et al., 2008; Kefas et al., 2008; Kefas et al., 2009; Li et al., 2009; Silber et al., 2008; Wurdinger et al., 2008). New findings on microRNAs in GBM are expected from further analysis of the TCGA data.

The study of epigenetics in GBM is another important frontier at a relatively early stage, and there are major discoveries yet to be made. Powerful input from this field has come during the last several years with the discovery of frequent methylation in GBM of the promoter of the MGMT gene (Gonzalez-Gomez et al., 2003), which performs DNA repair and can fix the DNA damage from alkylating chemotherapy drugs such as temozolomide. Found in about 30% of GBMs, MGMT promoter methylation appears to sensitize this subset to the effects of temozolomide and confers improved prognosis (Hegi et al., 2005).

There is a desperate need for further understanding of how specific genetic lesions confer sensitivity or resistance to established or developing therapies. One striking example of this was the discovery that mutations in the DNA repair gene MSH6 are occurring in GBMs treated with temozolomide (Cahill et al., 2007; Yip et al., 2009). This finding was confirmed and extended in the initial TCGA publication, which identified, in temozolomide-treated GBMs, mutations not only in MSH6 but also in other DNA repair genes in the MMR (DNA mismatch repair) pathway (McLendon, et al., 2008). These MMR-mutated cancers became even more genetically unstable than typical GBMs, with a “hypermutator” phenotype that has numerous point mutations. Efforts are ongoing to counter or exploit these therapy-resistant mutations with new therapies or combinations.

Conclusion

The rapid pace of genetic discovery in GBM seems only to be accelerating, and while the velocity of the number of published reports of new gene mutations and copy number changes seems likely to taper off in the coming years, discoveries in the fields of microRNAs and epigenetics will likely compensate for this. In addition, there are massive amounts of data from projects such as TCGA that are still being digested, and no doubt important findings will continue to emerge. We are finally beginning to understand the roots of this complex and challenging cancer, and with the boost from the newest genetic findings we seem on the verge of developing targeted and more effective therapies.