Gene Markers in Brain Tumors – What the Epileptologist Should Know

Summary

Gene markers or biomarkers can be used for diagnostic or prognostic purposes for all different types of complex disease, including brain tumors. Prognostic markers can be useful to not only explain differences in overall survival but also for differences in response to treatment and for development of targeted therapies. Multiple genes with specific types of alterations have now been identified that are associated with improved response to chemotherapy and radiotherapy, such as o⁶-methylguanine methyltranferase (MGMT) or loss of chromosomes 1p and/or 19q. Other alterations have been identified that are associated with improved overall survival, such as mutations in isocitrate dehydrogenase 1 (IDH1) and/or isocitrate dehydrogenase 2 (IDH2) or having the glioma CpG island DNA methylator phenotype (G-CIMP). There are many biomarkers that may have relevance in brain tumor associated epilepsy that does not respond to treatment. Given the rapidly changing landscape of high throughput “omics” technologies, there is significant potential for gaining further knowledge via integration of multiple different types of high genome wide data. This knowledge can be translated into improved therapies and clinical outcomes for brain tumor patients.

Introduction

A biomarker is a biochemical or genetic feature that can be assessed in a biospecimen in order to indicate a particular diagnosis, prognosis, or response to treatment. Biomarkers can be used for diagnostic, prognostic or predictive purposes for many complex diseases, including gliomas, the most common malignant brain tumor in adults (WHO grades II, III or IV) (Olar and Aldape, 2012). Prognostic biomarkers are those used to assess differences in overall disease course (i.e. overall survival time) or time to recurrence, while predictive biomarkers are those that can be used to assess the likelihood of response to a particular treatment (de Groot et al., 2011, Fischer and Aldape, 2010). Predictive biomarkers may also suggest new drug targets for disease. Several biomarkers have been described in brain tumors that may be relevant in epilepsy (as shown in Table 1), especially for long-term epilepsy associated tumors (LEAT).

Table 1.

Summary of most clinically relevant biomarkers in brain tumors

Biomarker	Associated Brain Tumor Type(s)	Frequency	Effect on Prognosis	Relevant References
Methylation of O6-methylguanine methyltranferase (MGMT)	Glioblastoma (GBM) (Grade IV)	20–40% of GBMs	Hypermethylation increases responsiveness to alkylating chemotherapeutic agents, by silencing DNA repair mechanism. Enhances the effectiveness of temozolomide chemotherapy and radiation	(Brandes et al., 2009, Everhard et al., 2009, Hegi et al., 2005, Kreth et al., 2011)
Loss of chromosomes 1p and/or 19q	Anaplastic Oligodendroglioma (Grade III)	50–70% of grade III oligodendrogliomas	Increase responsiveness to temozolomide chemotherapy. Significantly increased survival and progression-free survival after treatment with temozolomide.	(Cairncross et al., 1998)
Mutation in Isocitrate dehydrogenase 1 (IDH1) and/or Isocitrate dehydrogenase 2 (IDH2)	Oligodendrogliomas and oligoastrocytomas (Grade II and III)	91% of grade II oligodendrogliomas, 94% of grade III oligodendrogliomas, 79% of grade II oligoastrocytomas, and 91% of grade III oligoastrocytomas 0% of GBMS	Increases overall survival, one-year overall survival, and recurrence-free survival.	(Capper et al., 2011, Pollack et al., 2011, Yan et al., 2009)
Gene expression based subtypes	GBM (Grade IV)	~ 25% Proneural ~ 17% Neural ~26% Classical ~29% Mesenchymal	Proneural subtype was originally found to be associated with improved survival, but in a recent update analysis this survival advantage is no longer present	(Noushmehr et al., 2010, Brennan et al., 2013)
Glioma-CpG island DNA methylator phenotype (G-CIMP)	Adult low grade gliomas (LGG) (Grade II and III), and secondary GBM (grade IV, derived from LGG)	~ 10% of GBMs	Increased overall survival. Strongly associated with IDH1/IDH2 mutations.	(Noushmehr et al., 2010, Brennan et al., 2013, Yan et al., 2009)

Biomarkers in Brain Tumors

Isocitrate dehydrogenase 1 (IDH1) and Isocitrate dehydrogenase 2 (IDH2) mutations

IDH1 mutations are common in infiltrating lower grade (WHO II and III) and secondary high grade (WHO IV) glioblastomas, where the most common mutation in this gene is R132H (Yan et al., 2009). Primary glioblastomas arise with no prior glioma diagnosis while secondary glioblastomas represent malignant transformation in pre-existing lower grade gliomas. A review of adult cases immunohistochemically stained for IDH1-R132H found that 91% of grade II oligodendrogliomas, 94% of grade III oligodendrogliomas, 79% of grade II oligoastrocytomas, and 91% of grade III oligoastrocytomas were positive for this mutation. Neurocytomas, meningiomas, primary GBM with oligodendroglial component, and pilocytic astrocytomas with oligodendroglioma-like differentiation stained negative, as did all pediatric gliomas (Capper et al., 2011). Other studies have reported similar rates of mutation in adult glioma brain tumors. In a PCR based analyses of pediatric malignant gliomas, IDH1 mutations were found in 16.3% of tumors and IDH2 mutations were found in none (Pollack et al., 2011).

These mutations are useful for diagnosis of low grade malignant adult gliomas, but are of little utility in other adult tumors or pediatric gliomas. Mutations in IDH1 and IDH2 have been associated with improved overall survival (Yan et al., 2009). In adolescents that had ID1H mutation, one year overall and recurrence-free survival was significantly improved (p=0.035, and p=0.03 respectively) (Pollack et al., 2011)

Gene expression based subtypes of glioblastoma and the Glioma-CpG island DNA methylator phenotype (G-CIMP)

The Cancer Genome Atlas (TCGA) is a National Cancer Institute funded effort to fully molecularly characterize multiple different cancer types including glioblastoma (WHO grade IV; GBM) and lower grade gliomas (WHO grade II and III). Using an unsupervised clustering approach with genome wide gene expression array data, 4 clusters of GBMs were determined, denoting 4 gene expression based subtypes of GBMs: proneural, neural, classical and mesenchymal (Verhaak et al., 2010). The proneural subtype was originally found to be associated with improved survival, but in a recent update analysis of the TCGA GBM data this survival advantage is no longer present (Brennan et al., 2013).

Another analysis of the TCGA GBM data using DNA methylation array data, revealed a tightly-clustered DNA methylation subtype that comprised 8.8% of all samples. Tumors positive for the G-CIMP phenotype most often clustered into the proneural tumor subtype (Noushmehr et al., 2010). Compared to non G-CIMP proneural tumors, patients with G-CIMP tumors were significantly younger at time of diagnosis (p < 0.0001), and showed significantly longer survival (p=0.0165) after adjusting for age, recurrence status, and primary versus secondary GBM status

Further analyses conducted in low grade gliomas and GBMs suggest that the G-CIMP phenotype is associated with secondary rather than primary GBMs. Positive G-CIMP status is tightly associated with IDH1 mutation in low grade glioma. G-CIMP was also a significant independent predictor of survival in low grade glioma, after adjusting for age and tumor grade (Noushmehr et al., 2010).

However, the most recently updated analysis of the TCGA data does not show a survival advantage by gene expression subtype class (Brennan et al., 2013), but does continue to show an overall survival advantage for those that are G-CIMP positive and have IDH1 mutations (who are also more likely to be proneural) (Verhaak et al., 2010, Noushmehr et al., 2010, Yan et al., 2009), although this is a small proportion of all GBM patients (~10%).

O⁶-methylguanine methyltranferase (MGMT) Methylation in glioma

Many of the common chemotherapeutic agents used for brain tumors are alkylating agents. The effectiveness of these agents is compromised by direct DNA repair via the MGMT DNA repair gene. When this gene is silenced by promoter methylation, it cannot repair damaged DNA, hence chemotherapy and radiation treatment is significantly more effective and survival is improved. Hypermethylation has been found in a significant number of glioblastoma (GBM) patients (20–40%). Patients with MGMT methylation had a significant survival benefit from concurrent temozolomide (TMZ) chemotherapy and radiotherapy as compared to radiotherapy alone, while those who did not have methylation of the MGMT promoter had a statistically insignificant gain in survival from the addition of chemotherapy (Hegi et al., 2005). However, clinical utility of MGMT promoter methylation determinations is confounded by the controversy regarding how best to measure it (Suri et al., 2011), the absence of a direct one-to-one correspondence between MGMT promoter methylation and response to TMZ (Brandes et al., 2009), and a growing appreciation that methylation-independent pathways of MGMT expression regulation are in operation (Kreth et al., 2011, Everhard et al., 2009).

Loss of chromosomes 1p and/or 19q in glioma

In patients with grade III anaplastic oligodendroglioma coincident loss of chromosomal arms 1p and 19q is a commonly occurring phenotype, seen in 50–70% of tumors (Cairncross et al., 1998). Loss of 1p has been significantly associated with response to TMZ chemotherapy, with all tumors expressing this phenotype responding positively to chemotherapy in one analysis (p < 0.001). In patients whose tumors lost both 1p and 19q, there was also a significantly increased likelihood of response to chemotherapy (p < 0.001), and a significantly longer progression-free survival after chemotherapy. Loss of either 1p alone or both 1p and 19q was also a significant predictor of improved survival, even after adjustment for other clinical features (Cairncross et al., 1998).

Brain Tumors and Epilepsy

Any type of brain tumor can potentially cause seizures, and some may cause long-term drug-resistant epilepsy (LEATs). The tumors most frequently reported as causing this outcome are circumscribed and lower grade infiltrating gliomas (WHO I and III) and glioneuronal neoplasms, especially dysembryoplastic neuroepithelial tumors and gangliogliomas. In some tumor types, such as astrocytoma, presentation with chronic epilepsy predicts better prognosis (Thom et al., 2012). Presenting with seizure as first symptom is significantly associated with IDH1 or IDH2 mutation in lower grade glioma (Stockhammer et al., 2012). In addition, BRAF V600E mutations have been found in gangliogliomas, pilocytic astrocytomas and pleomorphic xanthoastrocytomas, but not in diffuse glioma (as reviewed in (Thom et al., 2012)). Assessment of this mutation, along with IDH1 mutation, could be useful to distinguish between diffuse glioma and PXAs.

There is a strong association between specific histological types of brain tumors and chronic seizures, and some studies have suggested that this may be due to the cellular composition and neurochemical profile of these tumor types. Studies of glioneuronal tumors have suggested that high neuronal density may be associated with epileptiform discharge patterns. Gangliogiomas have been reported to have increased expression of sodium-potassium chloride co-transporter expression and decreased expression of potassium-chloride co-transporter. Decreased expression of glial glutamate transporters may also contribute to epileptogenesis. Tumors may also disturb intercellular communication, damage surrounding tissue, and effect enzymatic changes that may result in epilepsy (Thom et al., 2012).

Biomarkers in epilepsy

There are currently no reliable biomarkers for epilepsy, though potential biomarkers are currently being investigated and multiple markers are used as part of standard diagnosis of brain tumors of different histological types that have implications in epilepsy (Sulman et al., 2009, Kaye and Laws, 2001, Engel, 2011). Biomarkers are needed to predict both epileptogenesis and epileptogenicity.

Currently, pharmacologic treatment for existing epilepsy is determined by acceptability of side effects, drug interactions, and patient tolerance for dosing interval. Development of biomarkers could assist physicians in determining appropriate antiepileptic drugs for each patient. Biomarkers could also be beneficial in predicting patients with drug-resistant epilepsy who would benefit most from surgical intervention, and allow these patients to receive surgical therapy earlier and prevent irreversible morbidity from recurrent seizures. As there are numerous types of seizures and epileptic conditions, there may be different biomarkers for each. Being able to predict epileptic conditions prior to onset, or shortly after onset would allow physicians the opportunity to prevent progression of disease.

Current areas of research into genetic biomarkers for epilepsy include identification of cellular changes causing neuronal death and synaptic reorganization, potential measures of gene expression alterations via PET or peripheral blood tests, and exploration of inflammatory changes (Engel, 2011).

Looking Forward

Given the rapidly changing landscape of high throughput “omics” technologies, significant further knowledge is to be gained via integration of multiple different types of high genome wide data in order to translate this knowledge into improved therapies and clinical outcomes for brain tumor and epilepsy patients.