Molecular markers of glioma: an update on recent progress and perspectives

Abstract

Background

Significant progress has been made in the molecular diagnostic subtyping of brain tumors especially gliomas. Designing effective tailored therapy remains the cornerstone for delving into the molecular heterogeneity and classification of gliomas. More homogenous tumor populations may lead to more uniform tumor responses in particular molecular constellation. Recent decade has seen a surge of molecular markers of glioma which hold a promise and potential of being strong prognostic, predictive, and diagnostic markers. They are also extremely critical for the stratification of current clinical trails.

Method

Review of the pertinent literature regarding the molecular markers of glioma was performed. Methods of detection of these markers and their clinical relevance are also discussed.

Results and conclusions

This review provides an update on progress and perspectives of these newest set of biomarkers which can also supplement and refine histological classification and serves as important prognostic and predictive markers; particularly relevant in this aspect are O⁶-methylguanine-DNA methyltransferase promoter methylation, IDH1 mutations, and codeletion of 1p/19q. BRAF fusion/mutations and EGFR amplification provide important clues diagnostically.

Introduction

Gliomas are the most frequent primary brain tumors and include a variety of histologic types. Experimental work on both animal models and primary gliomas has suggested the likely possibility that they arise from neoplastically transformed neural stem or progenitor cells. However, most glioma classifications over decades have relied on the morphological similarities of the tumor cells with non-neoplastic glial cells and thus are classified as astrocytic, oligodendroglial, mixed oligo–astrocytic, or ependymal tumors. These are distinguished based on the morphological criteria and graded on a scale of I–IV [astrocytic (grade I–IV); oligodendrogliomas (grade II–III); mixed tumors (grade II–II)] with increasing malignancy, according to the World Health Organization (WHO) classification of primary brain tumors (Louis et al. 2007). The WHO classification reflects the malignancy of the tumor and serves as a criterion to estimate the prognosis of patients. So far, histological evaluation remains the gold standard for glioma diagnosis and for grading the tumors. Clinical experiences derived from the prospective randomized clinical trials have shown that the histomorphological criteria alone might not be sufficient to predict the clinical outcome. Moreover, lately integrated genomic studies and exome sequencing have revealed the existence of multiple distinct molecular subtypes within histologically similar looking tumors (Verhaak et al. 2010). Gliomas even with identical histopathological features differ considerably regarding clinical course or response to therapy. The recent decade has seen the emergence of new molecular markers which are proving to be useful as prognostic, predictive, and some also as diagnostic markers. To date, only three biomarkers—codeletion of chromosomes 1p/19q, O⁶-methylguanine-DNA methyltransferase (MGMT) promoter methylation, and mutations in isocitrate dehydrogenase (IDH) IDH1/2 genes—have been identified as potent prognostic factors in gliomas. A more detailed molecular understanding of these markers is thus crucial to improve upon glioma classification, to better predict outcome, to better stratify patients in clinical trials, and to finally provide specific tailored therapy to individual tumor types or patients. This review aims at analyzing the markers in glioma research that are heralding a new era of integrating the molecular advances with histological assessment of these tumors. They hold an immense potential as prognostic, predictive, and diagnostic markers.

O⁶-methylguanine-DNA methyltransferase

The gene encoding the O⁶-methylguanine-DNA methyltransferase (MGMT) has garnered exciting interest as a promising molecular marker in neuro-oncology ever since the association between its promoter methylation status and response to alkylating drug was documented by Esteller et al. almost a decade ago (Esteller et al. 2000). MGMT promoter hypermethylation has been identified in a wide variety of human cancers, including lung, head and neck, pancreatic, renal, bladder carcinomas, and also in lymphoma, leukemia, and melanoma (Esteller et al. 1999). The frequency of MGMT promoter hypermethylation in gliomas varies substantially ranging from 35 to 75 % in glioblastoma multiforme (GBM) (Brandes et al. 2008; Dunn et al. 2009; Esteller et al. 2000; Hegi et al. 2004, 2005; Zawlik et al. 2009).

The MGMT gene located on chromosome 10q26 has five exons and a CpG-rich island of 763 bp with 98 CpG sites encompassing the first exon and large parts of the promoter (Nakagawachi et al. 2003). A minimal promoter and an enhancer region are located within the CpG island (Nakagawachi et al. 2003). Most of the CpG sites within the island are unmethylated in the normal tissue. Methylation of the CpG islands in the promoter region leads to alteration in the chromatin structure and reduced affinity of binding of transcription factors resulting in silencing of MGMT gene expression (Nakagawachi et al. 2003). MGMT is a DNA repair protein, which rescues the DNA damage induced by alkylating agents by removing the alkyl groups from O⁶-guanine in DNA to a cysteine residue at its own position 145. The alkylated MGMT is eventually degraded and needs to be resynthesized de novo. The alkylating drug, Temozolomide (TMZ) used in gliomas, adds the alkyl groups to the O⁶ position of guanine and O⁴ of thymine, thereby causing DNA breakage and triggering apoptosis (Hegi et al. 2005). Thus, the tumor cells expressing MGMT are resistant to alkylating agents, while those that have silenced expression of gene owing to hypermethylated phenotype are chemosensitive (Hegi et al. 2005).

The landmark study by Esteller in 2000 first established the beneficial prognostic role of MGMT promoter methylation with chemoresponsiveness of gliomas to alkylating drugs (Esteller et al. 2000). Later, a prospective multicentric study conducted through the collaboration of the European organization for research and treatment for cancer (EORTC) and the National Cancer institute of Canada (NCIC) (EORTC/NCIC trial 26981/22981) showed that the addition of TMZ to radiotherapy for newly diagnosed GBM results in a clinically meaningful and statistically significant survival benefit with minimal toxicity (Stupp et al. 2005). A translational study carried out parallel to the EORTC/NCIC trial revealed a strong correlation of the MGMT methylation status with TMZ treatment effect and outcome (Hegi et al. 2005). Subsequently, many clinical studies over the years have established the prognostic and predictive role of MGMT methylation status to treatment with alkylating agents (Gorlia et al. 2008). Predictive power of MGMT promoter methylation status in elderly glioma patients has been documented in recent studies (Wick et al. 2012; Reifenberger et al. 2012). Results of randomized phase 3 trial by German Neuro-oncology working group (NOA-08) have revealed better event-free survival for newly diagnosed elderly glioma patients with methylator phenotype on TMZ alone than those under radiotherapy alone (Wick et al. 2012). Results of German Glioma Network have yielded comparable results (Reifenberger et al. 2012). Thereby, MGMT promoter methylation status is a relevant predictive biomarker indicative of patients who might be under-treated with radiotherapy alone. Furthermore, the risk of cognitive side effects of cranial irradiation in elderly patients could also be significantly reduced. The role of MGMT promoter methylation in adult recurrent glioblastoma has however not been found to be significantly influential in improving the progression-free survival (PFS) (Brandes et al. 2009; Sadones et al. 2009). The same for the recurrent anaplastic oligodendrogliomas (AO) and oligoastrocytomas has however shown superior survival in study by Sadones et al. (2009).

Similar to the adult glioblastoma, studies on pediatric glioblastoma (Brandes et al. 2009; Donson et al. 2007; Pollack et al. 2006) have documented favorable prognostic and predictor role of MGMT methylation status with overall survival and TMZ sensitivity. Two studies involving a large cohort of pediatric high-grade glioma by Pollack et al. (109 cases of GBM and anaplastic astrocytoma [AA]) and Cohen et al. (107 cases of AA, GBM, and gliosarcoma) have documented a significant difference in overall survival in association with MGMT methylation status and also significant difference in chemoresponsiveness to TMZ (Pollack et al. 2006; Cohen et al. 2011).

Methods of MGMT analysis

Several methods are available for the assessment of MGMT methylation status at DNA, RNA, and protein levels.

1. The DNA-based assays include methylation-specific polymerase chain reaction (MS-PCR), real-time MS-PCR, methylation-specific pyrosequencing, and methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA). MS-PCR is a highly sensitive and most commonly employed bisulfite-based method for analyzing the MGMT promoter methylation status and has been used in a majority of the clinical trials till date (Brandes et al. 2008, 2009; Donson et al. 2007; Esteller et al. 2000; Hegi et al. 2004, 2005; Sadones et al. 2009). The frequency of methylation status varies from 35 to 68 % (Brandes et al. 2008; Hegi et al. 2004). This is primarily due to the inherent inhomogeneous methylation pattern of CpG sites in the promoter region. Each method focuses on only a few CpG sites with the assumption that it reflects the methylation of all CpG sites, thereby leading to disparity in the reported frequency (von Deimling et al. 2011).

2. The MGMT protein can be assessed by immunohistochemistry (IHC) using commercially available anti-MGMT antibodies (Cohen et al. 2011; Pollack et al. 2006). However, the assessment of MGMT methylation status using IHC has limitations and remains controversial. Assessment of MGMT methylation status by IHC has failed to correlate with disease outcome (Preusser et al. 2008) which has been because of (1) inter-observer variability; (2) heterogeneity of glioma sample (at the protein level, MGMT protein expression is heterogenous within the tumor and within small regions of tumor suggesting molecular heterogeneity); (3) contamination with non-neoplastic cells (lymphocytes, microglia, endothelial cells, and astrocytes) express MGMT leading to fallacious results (Felsberg et al. 2009); and (4) differences in tumor immunoreactivity (Preusser et al. 2008). Owing to these limitations, IHC has limited applicability.

3. The detection of MGMT mRNA expression can also be utilized; however, it needs to be determined on fresh tumor tissue and has the same drawback of being diluted by non-neoplastic MGMT-producing cells (von Deimling et al. 2011).

Implications of MGMT methylation status

A body of the literature supports the hypothesis that the level of tumor MGMT correlates with survival and with alkylating drug sensitivity in gliomas. Currently, various methods are being employed for its assessment leading to the generation of divergent results from multiple laboratories for the same tumor tested. As there is no consensus regarding the optimal method for MGMT testing, and the relative cutoff values, it is difficult to compare the test results. A high throughput, robust, clinically validated test needs to be standardized and developed in order to put the results of its methylation phenotype to a therapeutic level.

Lack of alternative treatment options for patients with unmethylated phenotype and small number of documented responsive cases of GBM with unmethylated phenotype puts the results of methylation status on difficult grounds as a sole parameter for weighing treatment options. Furthermore, as the prognostic assessments in GBM are typically not of major use (i.e., implying modest differences in survival), MGMT testing is not typically undertaken for prognostic purposes. MGMT evaluation is currently essential only for clinical trials (as a selection criterion for study inclusion), so that adequate comparisons with currently available therapeutic options can be derived.

BRAF fusion and mutations

BRAF is a member of RAS/RAF/MEK/ERK protein kinase (mitogen-activated protein kinase, MAPK) pathway. This pathway by regulating the activity of several transcription factors plays a pivotal role in cell proliferation, cell survival, differentiation, and apoptosis (von Deimling et al. 2011). The highest frequency of BRAF mutation among the human cancers is seen in malignant melanoma (Davies et al. 2002). Among the non-CNS tumors, oncogenic activation of BRAF has been documented in papillary carcinoma of thyroid, congenital melanocytic nevi, carcinoma of biliary tract, ovarian, and colon carcinoma (Davies et al. 2002; Tannapfel et al. 2003). Among the CNS tumors, pilocytic astrocytoma (PA) has been documented to show the highest frequency of BRAF mutations or tandem duplication and the high specificity of this fusion in PA in comparison with other CNS tumors has diagnostic utility (Schindler et al. 2011). BRAF alterations implicated in primary CNS tumors can be in form of oncogenic fusion genes (KIAA1549:BRAF) or as missense mutation of V600E type (Schindler et al. 2011). In PA, genomic alterations on chromosome 7q34 create fusion between a gene of unknown function, KIAA1549, and the BRAF gene. Consequent to these fusions, a 2-Mbp region between the two genes is duplicated in tandem such that the 5′ end of the KIAA1549 gene becomes fused with the 3′ end of BRAF (Forshew et al. 2009; Tian et al. 2011). Five known exonic fusion variants occurring due to tandem duplication are–KIAA1549:BRAF exon 16-exon 9, exon 15-exon 9, exon 16-exon 11, exon 18-exon 10, and exon 19-exon 9 (Jeuken and Wesseling 2010). In all cases, the fusion leads to loss of the BRAF N-terminal autoregulatory domain and subsequent activation of B-Raf kinase domain. This constitutive activation of BRAF leads to oncogene-induced senescence (OIS) in slowly growing benign tumors. It is likely that this process of OIS plays a major role in restricting PA to its relatively slow growth pattern and generally more benign behavior compared with high-grade astrocytomas. The tandem duplication between KIAA1549 and BRAF occurs in 70 % of these tumors (Jones et al. 2008, 2009). In about 2 % of PA, there occurs an alternative mechanism of MAPK pathway activation, by tandem duplication at 3p25 leading to oncogenic fusion between SRGAP3 and RAF1 (Forshew et al. 2009; Jones et al. 2009).

The distribution of KIAA1549:BRAF fusion in PA spans across all ages and all locations; however, some studies have revealed that frequency of BRAF fusion/tandem duplication is more in cerebellar tumors in comparison with extra-cerebellar location which are more strongly associated with BRAF V600E mutation (Forshew et al. 2009; Horbinski et al. 2010a, b; Tannapfel et al. 2003). Also, on a multivariate logistic regression analysis, studies by Hasselblatt et al. (2011) documented the frequency of fusion transcripts to be significantly lower in adult patients with PA in comparison with the pediatric age group. The fusion is a characteristic event in PA.

Missense mutations of V600E type in BRAF gene constitute an alternate mechanism of MAP/ERK pathway activation. This mutation is caused by single-nucleotide exchange (T to A transversion) at codon 600 which results in replacement of valine by glutamic acid (V600E) (Tian et al. 2011; Wan et al. 2004). Up to 10 % of pediatric PAs contain mutation, and this appears to be more frequent in extra-cerebellar tumors (Forshew et al. 2009; Horbinski et al. 2012). In addition, V600E is common in other low-grade pediatric gliomas including 80 % of pleomorphic xanthoastrocytoma and 25 % of ganglioglioma (Horbinski et al. 2012), which makes it a valuable diagnostic marker for these rare tumor entities.

Methods of detection of BRAF-KIAA1549 fusion gene and point mutation

Detection of fusion gene can be accomplished by FISH analysis or specialized RT-PCR performed on fresh-frozen or FFPE tissue. FISH analysis requires the detection of a fusion signal of two fluorescently labelled probes. The KIAA1549:BRAF gene fusion is defined in the cases showing nuclei with a single fusion red–green or yellow signal in addition to the normal pair of split red and green signals. At least 100 non-overlapping, intact nuclei are scored and BRAF fusion is typically detected in 20–50 % of nuclei in positive cases. RT-PCR is a gel-based method, performed on fresh-frozen or FFPE tissue, for the detection of the fusion gene and is easy to standardize and quantify (Riemenschneider et al. 2010).

Pyrosequencing is the method used for the detection of BRAF point mutation. It is a DNA sequencing technology based on the principle of “sequencing by synthesis,” that is, real-time detection of DNA synthesis monitored by bioluminescence (Setty et al. 2011).

Clinical relevance of BRAF fusion and mutations

(1) The prognostic relevance of KIAA1549:BRAF fusion is still unclear, as few studies have found no difference in survival between tumors with and without BRAF duplication/fusion (Forshew et al. 2009; Horbinski et al. 2010a, b; Jacob et al. 2009). Nonetheless, the frequency and specificity of this tandem duplication serves as a diagnostic marker for PA, and combined with IDH1 mutational analysis, it aids in its distinction from diffuse astrocytoma. (2) It holds a promise of a potential therapeutic target for pharmacological inhibitors of BRAF and MAPK pathway, particularly in tumors which are not fully surgically resectable (Table 1). Chemo/radiotherapy protocols applied to this group of patients again run the risk of severe side effects. Targeted therapy using pathway inhibitors is being recognized as potential novel treatment approaches. The results from the currently ongoing preliminary phase I/II clinical trials testing small molecule kinase inhibitors targeting the MAPK pathways including MEK inhibitors, RAF inhibitors, and mTOR inhibitors in patients with or without neurofibromatosis type 1 (NF1) will prove to be greatly beneficial for developing larger clinical trials and will take the treatment strategies for PA to the next level (Jones et al. 2012).

Table 1.

Biomarkers most relevant to the molecular diagnostics of gliomas

MGMT promoter methylation	Prognostic in anaplastic glioma patients treated with radio- and/or alkylating chemotherapy
	Predictive for response to alkylating chemotherapy in both adult and pediatric glioblastoma. Role in recurrent GBM is not found to be significantly influential in improving PFS
BRAF duplication/mutations	Diagnostic marker for PA
	Prognostic significance within PA patients yet to determined
	Holds a potential of being a therapeutic target with BRAF and MAPK pathway inhibitors in surgically unresectable tumors
1p/19q codeletion	Associated with better prognosis in oligodendroglioma patients receiving adjuvant radio-and/or chemotherapy
	Not predictive for response to a particular type of therapy
	Useful diagnostic marker for oligodendroglioma and OA
IDH1/IDH2 mutations	Major and independent prognostic factor for diffuse WHO grade II and III gliomas and secondary GBM
	Diagnostic marker for these tumors
	Predictive significance remains to be determined
EGFR and PTEN alterations	Both used as diagnostic markers for small cell glioblastomas
	EGFR important therapeutic applicability
	PTEN holds a prognostic potential yet to be explored

Codeletion of chromosome arms 1p and 19q

Loss of 1p and 19q is a hallmark or genetic signature of oligodendroglial tumors, wherein usually both 1p and 19q are completely lost due to an unbalanced centromeric translocation t(1;19)(q10;p10). The 1p-19q derivative is lost, while the 1q-19p derivative is retained during cell replication (Cairncross and Jenkins 2008). This codeletion pattern is an early genetic event and is tightly linked to the oligodendroglial lineage and to the presence of its classic histologic features of perinuclear halo and “chicken-wire” vascular pattern. This codeletion of chromosomal arms is detected in up to 80 % of oligodendrogliomas and approximately 60 % of AO, 30–50 % of oligoastrocytomas, 20–30 % of anaplastic oligoastrocytomas, and 10 % of diffuse astrocytic gliomas, including GBM (Riemenschneider et al. 2010). Interestingly, this combination of loss of heterozygosity (LOH) is almost never found in any non-glial malignancy. Histological evidence of an oligodendroglioma or at least an oligodendroglial component in a mixed glioma is always coupled with an IDH1/2 mutation (Labussiere et al. 2010). Although 1p/19q codeletion is an early event, it is not the earliest tumorigenic event. IDH1/2 mutations are thought to precede both the codeletion step in oligodendroglioma and TP53 mutation in astrocytoma (Horbinski et al. 2012; Ichimura et al. 2009).

Cairncross et al. in 1998 for the first time reported that this codeletion pattern predicts better response to chemotherapy and longer survival in AO patients (Cairncross et al. 1998). Subsequently, few studies including prospective randomized phase III trails have confirmed this codeletion pattern as a powerful prognostic and predictive marker in WHO grade III oligodendrogliomas (Cairncross et al. 1998; van den Bent et al. 2006; Wick et al. 2009). Moreover, these studies have documented that the prognostic power is independent of the type of adjuvant therapy, that is, radiotherapy, chemotherapy, or combined radio/chemotherapy. As a result, this codeletion pattern has become a useful diagnostic, prognostic, and predictive marker in neuro-oncology practice. Although this is associated with a more favorable prognosis in patients receiving adjuvant treatment, it needs to be emphasized that this marker is of limited utility for directing a choice regarding specific tailored therapy (chemotherapy versus radiotherapy). Also, the prognostic utility is further compromised in the presence of other prognostically unfavorable genetic alterations (i.e., 9p or 10q loss) (Trost et al. 2007). Additionally, the prognostic significance is dependent on the type of 1p losses, for instance, oligodendroglial tumors with terminal or interstitial 1p losses are associated with shorter patient survival in comparison with tumors with combined complete 1p/19q losses (Idbaih et al. 2005; Nikiforova and Hamilton 2011). These partial 1p deletions are more prevalent in astrocytic tumors in comparison with oligodendroglial tumors and are associated with worse prognosis (Ichimura et al. 2009).

Interestingly, long-term follow-up results of the EORTC 26951 (randomized phase III study) call for devising new standard of care for patients with AO. The results have shown improvement in PFS and overall survival in AO patients with 1p/19q codeletion on the addition of PCV (procarbazine, lomustine, and vincristine) to radiation. Also, the improved trend was noted in overall survival in patients with MGMT methylation and IDH mutations (van den Bent et al. 2012).

Methods for the detection of codeletion

Assessment of the codeletion in oligodendrogliomas is most frequently accomplished by fluorescent in situ hybridization (FISH), PCR-based LOH analysis, multiplex ligation-dependent probe amplification, and array comparative genomic hybridization (Nikiforova and Hamilton 2011). Of all the methods, FISH has proven to be robust, easy to implement, and cost-effective (Horbinski et al. 2011). It can be performed either on isolated nuclei or on tissue sections. It uses dual-color fluorescently labelled DNA probes to provide a targeted approach for detecting chromosomal abnormalities in interphase nuclei. Target probes hybridize to subtelomeric 1p36 and 19q13.3 in combination with control probes on 1q and 19p, respectively. The FISH approach does not require corresponding reference normal tissue or autologous blood samples as control and maintains preservation of tissue architecture. For evaluation, the signal ratio is assessed for 100–200 adjacent, non-overlapping interphase nuclei. The number of nuclei exhibiting a balance, imbalance, or deletion is summed up, and the results are expressed as percentage (Horbinski et al. 2011; Nikiforova and Hamilton 2011).

Microarray gene expression demonstrated that 1p/19q codeletion glioma has a proneural gene expression profile, which has been associated with a good prognosis in high-grade glioma (Phillips et al. 2006). One of the most differentially expressed of these proneural genes is INA, which encodes α-internexin (INA), a neurofilament-interacting protein. INA expression is a surrogate marker for 1p/19q codeletion with specificity of 86 % and a sensitivity of 96 % (Ducray et al. 2009). Moreover, in grade III gliomas, similar to 1p/19q codeletion, positive INA expression has been documented with a better progression-free survival and overall survival (Ducray et al. 2009).

IDH1 and IDH2 mutations

Mutations in IDH1 were identified in 2008 as a chance finding from the exome-wide deep sequencing performed on the glioblastomas (Parsons et al. 2008). Interestingly, it was noted that IDH1 mutations were mostly associated with young age, a secondary glioblastoma pattern, and a higher survival (Parsons et al. 2008). Subsequent studies identified IDH1 in 70–80 % of grade II and III astrocytomas, oligodendrogliomas and oligoastrocytomas, and secondary glioblastomas (Hartmann et al. 2009; Ichimura et al. 2009; von Deimling et al. 2011; Yan et al. 2009). This suggests a possible unifying role of these mutations in early gliomagenesis and an existing close relationship between these tumors. In contrast, IDH1 mutations are rare in primary glioblastomas and are completely absent in PA (Yan et al. 2009). Additionally, mutations in gene encoding a related enzyme, IDH2, were detected in a small group of gliomas, mostly in oligodendroglial tumors (Yan et al. 2009). Intriguingly, these mutations coexist with 1p/19q codeletion in oligodendrogliomas and with TP53 mutations in astrocytoma but are mutually exclusive with EGFR amplification and loss of chromosome 10 (Yan et al. 2009; Balss et al. 2008; Ichimura et al. 2009) (Fig. 1). IDH1 mutation is also associated with methylation of MGMT promoter (Sanson et al. 2009).

Fig. 1.

Most frequent molecular pathways and common genetic alterations implicated in astrocytic, oligodendroglial, and oligoastrocytic neoplasms. TP 53 tumor protein p53 gene, A astrocytoma, OA oligoastrocytoma, O oligodendroglioma, AA anaplastic astrocytoma, AOA anaplastic oligoastrocytoma, AO anaplastic oligodendroglioma, PA pilocytic astrocytoma (modified from Ichimura et al. 2009)

Three distinct IDH enzymes exist in the mammalian cells (IDH1, IDH2, and IDH3) that catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate (α-KG), a key component in the citric acid cycle (Reitman and Yan 2010). IDH1 and IDH2 genes encode for homodimeric enzymes that utilize NADP+ as a cofactor to generate α-KG and NADPH in a reversible reaction (Thompson 2009; Zhao et al. 2009). Both α-KG and released NADPH are known cell defenders against oxidative damage (Lee et al. 2002). IDH1 is found in the cytoplasm and peroxisomes, while IDH2 (and IDH3) is found in the mitochondria (Geisbrecht and Gould 1999). IDH3 is distinct from IDH1 and IDH2; it is a heterotrimeric enzyme complex with multiple known allosteric modifiers and utilizes NAD+ to produce α-KG in an irreversible manner (Reitman and Yan 2010).

The most common mutation type for the IDH1 gene is restricted to a single residue of the gene and is a CGT → CAT change, leading to an Arginine¹³² → Histidine substitution (Parsons et al. 2008). Subsequently, five other mutations (Arg¹³² to serine, cytosine, glycine, valine, and leucine) were identified (Balss et al. 2008; Yan et al. 2009). Some patients harbor a mutation in the IDH2 gene at amino acid residue (Arg¹⁷² to lysine, methionine, and glycine) (Hartmann et al. 2009; Yan et al. 2009).

Different hypotheses have been proposed to explain the mechanisms by which mutations in IDH1 and IDH2 mediate oncogenesis: as “loss of function” of IDH enzymes leading to reduced production of α-KG and NADPH, and “gain of function” leading to increased production of d-2-hydroxyglutarate (D2HG). The reduced generation of α-KG and NADPH makes the cell more susceptible to oxidative damage and results in reduced degradation of hypoxia inducible factor (HIF-1α) (Lee et al. 2002; Gupta et al. 2011), which promotes tumor growth and angiogenesis. Thus, oxidative damage is significantly increased with reduced IDH function. “Gain of function” leads to increased production of d-2HG as the IDH1 mutations in glioma result in neomorphic enzyme activity. The α-KG is a co-substrate for over 60 dioxygenase enzymes, including many histone lysine methyltransferases and TET family of 5-methylcytosine hydroxylases (Chowdhury et al. 2011; Xu et al. 2011). The mutant IDH1 converts α-KG to d-2HG. d-2HG inhibits variety of dioxygenases including PHD (propyl hydroxylase), TET2, and histone demethylases which might trigger aberrant angiogenesis, DNA methylation, or gene expression, respectively. Mutant IDH1 also reduces the cellular level of glutathione by depleting NADPH and rendering the cells vulnerable to oxidative DNA damage (Reitman and Yan 2010). This may promote further genetic changes, for instance TP53 mutations or t(1;19) translocation, leading to the development of either astrocytomas or oligodendrogliomas.

The recent finding of IDH1 mutations being tightly lined with the CpG island methylator phenotype across all glioma tumor grades further raises the possibility that IDH1 mutations may predispose to large-scale epigenetic disruptions in diffuse gliomas (Noushmehr et al. 2010). The mechanism of action of mutant IDH is still being explored and a clear understanding of this would help in bridging the diverse preneoplastic processes such as oxidative damage and epigenetic regulation.

Methods for the detection of IDH mutations

Currently available methods investigate either the altered structure of protein or the sequence of gene.

Immunohistochemistry

Currently, two monoclonal antibodies, H09 and IMab-1, have been developed against the common mutation, IDH1 R132H (Kato et al. 2009), and another one against R132S (Kaneko et al. 2011). The commercially available clone H09 has been used for routine immunohistochemistry on FFPE tissue. The tumor cells with mutant IDH1 R132H show cytoplasmic immunopositivity, whereas the cells with wild-type IDH1 are not immunoreactive (Gupta et al. 2011). The sensitivity and specificity of clone H09 antibody for detecting IDH1 R132H have been reported to be 100 % (Capper et al. 2010). Moreover, immunohistochemistry has an advantage of identifying single infiltrating cells in sections with large proportions of non-neoplastic brain parenchyma, reactive astrocytosis, or admixed with inflammation. The antibody is stated to be highly specific for IDH1 R132H; however, the other rare substitutions in IDH1 (R132C, R132S, R132L, R132G) and IDH2 are not detected (Capper et al. 2010; von Deimling et al. 2011).

Direct sequencing, pyrosequencing, and derived cleaved amplified polymorphic sequence (dCAPS) analyses are the methods for detecting the gene sequence (von Deimling et al. 2011). Sanger sequencing is an automated relatively inexpensive process considered as the “gold standard” for the detection of mutations within the amplified region. The sensitivity of this method has been reported to be as 20 % of mutant sequences in a wild-type background (Horbinski et al. 2010a, b). However, in the presence of heterozygous mutations in brain tumor biopsy specimens, detection of mutant protein gets problematic because of the contaminating non-tumor tissue.

The major advantages of pyrosequencing analysis are its quantitative nature and high sensitivity which is able to detect 5 or 7 % mutant alleles in a wild-type background (Felsberg et al. 2010). This is particularly useful for diffusely infiltrating low-grade gliomas with low proportion of glioma cells within the normal brain parenchyma.

dCAPS is PCR-based restriction fragment length polymorphism assays which detect the presence of mutations in codon 132 of IDH1 gene which determines the generation of restriction endonuclease sites (von Deimling et al. 2011). Digestion with appropriate endonucleases yields DNA fragments of different sizes which are separated on agarose gels. Different primers are used for the detection of both wild-type IDH1 and other known R132 mutants.

Another approach is the melting curve analysis performed on real-time PCR products allowing the detection of IDH1 and IDH2 mutations using difference in the melting temperatures between wild-type and mutant sequences. It is a rapid and sensitive analysis easily accessible in the diagnostic laboratory (Horbinski et al. 2010a, b).

Clinical implications of IDH1 mutations

Various studies have documented that the presence of IDH1/2 mutations predicts significantly longer survival for patients with glioblastomas, anaplastic astrocytomas oligodendroglial tumors (Hartmann et al. 2010; Ichimura et al. 2009; Parsons et al. 2008; Yan et al. 2009; van den Bent et al. 2010). There is however no difference in response to therapy between them; thus, presence of these mutations is a prognostic but not a predictive marker for these tumors (van den Bent et al. 2010).

IDH1 mutation also has diagnostic potential; for instance, IDH1 mutational status may help to differentiate grade II gliomas from PA or pleomorphic xanthoastrocytomas, or secondary glioblastomas from primary glioblastomas (Gupta et al. 2011). This is considerably facilitated by the commercial availability of two monoclonal antibodies specifically targeting against IDH1 R132H mutation (Kato et al. 2009). Also, the presence of IDH mutation may be useful to distinguish oligodendrogliomas from its morphological mimics—central neurocytomas, clear cell ependymomas, and dysembryoplastic neuroepithelial tumors—which are all negative with IDH1R132H (Capper et al. 2011). Moreover, immunohistochemical method is very useful in detecting residual glioma cells in tumors with extensive post-therapy changes and also in differential diagnosis of low-grade diffuse glioma with reactive changes.

Stratification of glioblastomas according to their IDH status might become mandatory in future clinical studies, because glioblastomas with IDH mutations are likely to be a distinct entity (Nikiforova and Hamilton 2011).

EGFR and PTEN alterations

The epidermal growth factor receptor (EGFR) affects cell proliferation and growth through the activation of downstream effector molecules in MARK and PI3 K-Akt pathways. Its gene (EGFR) is located at 7p12 is the most frequently amplified and over-expressed gene in primary glioblastoma, affecting approximately 40 % of these tumors (Louis et al. 2007; Yip et al. 2008). EGFR rearrangements are also frequent, the most common variant being EGFRvIII characterized by an in-frame deletion of 267 amino acids in the extracellular domain of EGFR gene (exons 2–7) that results in a constitutively active truncated receptor lacking the ligand-binding domain (Sugawa et al. 1990). EGFRvIII rearrangements are identified in about half of the EGFR-amplified glioblastomas and in 20–30 % of primary glioblastoma (Gan et al. 2009). Detection of EGFR amplification and rearrangements is indicative of high-grade glioma, specifically the more common primary or de novo subtype. It is also extremely helpful in the differential diagnosis of high-grade oligodendroglial neoplasms with small cell variant of GBM. EGFR amplification is seen in approximately 70 % of small cell GBMs usually in association with PTEN or chromosome 10q deletions (Horbinski et al. 2011). Besides being used as a diagnostic marker, it also provides prognostic information (Louis et al. 2007); however, its role as a latter is still unclear. EGFRvIII over-expression and EGFR amplification have been promoted as indicators of poor survival in GBM in some studies (Shinojima et al. 2003); however, other studies have failed to show an effect of EGFR amplification on patient survival (Huncharek and Kupelnick 2000).

Detection of EGFR aberrations may also serve as a potential therapeutic target. Initial studies failed to document the therapeutic benefits of anti-EGFR tyrosine kinase inhibitors in patients with glioblastoma (Halatsch et al. 2006). Subsequent studies however suggest that EGFR amplification or EGFRvIII expression can predict responsiveness to tyrosine kinase inhibitors, especially when PTEN expression is preserved (Mellinghoff et al. 2005). The EGFRvIII mutant may also serve as an attractive target for immunotherapy (Li and Wong 2008). Recent studies reported that the anti-EGFRvIII peptide vaccine when added to radiation and chemotherapy has resulted in increased overall survival and progression-free survival in patients with glioblastoma (Heimberger and Sampson 2009). The predictive role of EGFR/EGFRvIII abnormalities in treatment response remains to be elucidated yet in future clinical trials.

EGFR amplification is detected by FISH as double-minutes, small fragments of extra-chromosomal DNA (Horbinski et al. 2011). Other techniques such as real-time PCR can be used to identify and quantify EGFR amplification. Multiplex ligation-dependent probe amplification (MLPA) analysis provides a simultaneous and semiquantitative copy number analysis (Masui et al. 2012). Immunohistochemistry is also employed to detect the tumor with EGFRvIII rearrangements by using antibody specific for EGFRvIIII (Masui et al. 2012). Also, a real-time RT-PCR assay has been developed for the quantification of EGFRvIII expression in FFPE samples.

Phosphatase and tensin homolog (PTEN) is a tumor suppressor gene located on the long arm of chromosome 10 at 10q23. The intracellular protein which this genes encodes inhibits phosphatidylinositol-3,4,5-triphosphate signalling which in turn inhibits the Akt pathway, which are critical pro-oncogenic pathways in gliomas. Around 15–40 % of primary GBMs contain PTEN mutations and up to 80 % of all GBMs have LOH on 10q, usually including 10q23, the site of PTEN localization (Ohgaki and Kleihues 2009). The small cell glioblastoma phenotype frequently shows 10q loss together with EGFR amplification and PTEN mutations, making it a useful diagnostic marker (Louis et al. 2007). It also has the potential prognostic role, for PTEN deletions have been found to be independently associated with shorter survival in pediatric GBMs (Horbinski et al. 2011). The same however does not hold good for adults GBMs (Horbinski et al. 2011).

Biomarkers are an emerging field for neuro-oncology and hold a great potential for significantly complementing and refining the histology and WHO classification, thereby providing valuable leads in patient management (Table 1).

1. MGMT promoter methylation is a strong predictor factor of response to radiochemotherapy regimens in glioblastomas. Robust and clinically validated techniques for analyzing MGMT status would impart it greater applicability and move it a step further beyond clinical trials.

2. BRAF fusion and mutations serve as important diagnostic marker for PA and hold a promise of being potential therapeutic target.

3. Codeletion of 1p/19q is a useful diagnostic, prognostic, and predictive marker for oligodendroglioma and oligoastrocytoma.

4. IDH1/IDH2 mutations represent a major and independent prognostic factor. Since it is mostly restricted to gliomas, it has a significant diagnostic potential. Its predictive significance is yet to be determined.

5. BRAF fusion and IDH1 mutation analyses promise to be very helpful for classifying and grading gliomas.

6. EGFR and PTEN are used as diagnostic markers for small cell glioblastomas. EGFR has potentially important therapeutic applicability and PTEN holds a prognostic potential yet to be explored and validated.

Techniques of the burgeoning field of molecular diagnostics are being increasingly employed in tumor classification and grading. The advances in our understanding of molecular mechanisms of cancer and the molecular diagnostic techniques have significantly changed the practice of oncologic pathology. Biomarkers are standing at the crossroads, heralding an era and bridging the gaps for formulating and designing specific tailored therapy. Besides having a diagnostic potential, these are proving to be essential in monitoring disease progression, prognosis, and in predicting response to therapy. Neuropathologist and neuro-oncologists thus have been positioned at a challenging role of subsequent application of the information derived from these assays in formulating more individualized therapies as they become available in future.