IDH catalyzes the oxidative decarboxylation of isocitrate to α-ketoglutarate (α-KG), producing NADH or NADPH in the process. α-KG is a substrate for multiple dioxygenase enzymes that control histone modification, and it is important in the regulation of glutamate production and in the cellular response to oxidative and energetic stress. IDH mutation results in aberrant production and accumulation of D-2-hydroxyglutarate (D-2-HG), resulting in changes in cell energetics and the methylome that predispose cells to transformation.
IDH mutation is thought to be an early if not the initial event in the development of low-grade astrocytomas and oligodendrogliomas. In support of this hypothesis, IDH mutation is found in secondary glioblastomas but not in primary glioblastomas. IDH mutation does not occur in pilocytic astrocytomas or in other World Health Organization (WHO) grade I lesions.
Turkalp Z, Karamchandani J, Das S. IDH Mutation in GliomaNew Insights and Promises for the Future. JAMA Neurol. 2014;71(10):1319-1325. doi:10.1001/jamaneurol.2014.1205
Over the past 4 years, our understanding of gliomagenesis and the practice of neuro-oncology have been radically changed by the discovery of mutations involving the isocitrate dehydrogenase (IDH) enzymes. IDH mutation has been found to be an inciting event in gliomagenesis and to have a profound effect on the molecular and genetic route of oncogenic progression and on clinical outcome.
To review the role of IDH enzymes in normal physiology and describe aberrations in the IDH pathway that are associated with gliomagenesis, to review recent work examining the effect of IDH-targeted therapy in cancers harboring IDH mutation, and to determine how this work has expanded our understanding of the role of IDH in the development and progression of glioma.
A systematic review of the literature dating from 2008, when IDH mutation was discovered to be clinically significant in glioma, to 2013 was performed using the PubMed database. The following search terms were used: IDH, IDH1, IDH2, and isocitrate dehydrogenase, in conjunction with glioma or leukemia. The search was limited to articles published in English. Further hand searching was performed using a review of the pertinent references from the identified publications. All identified original articles were investigated for content and critiqued by Z.T. and S.D.
IDH mutation is an early event in gliomagenesis and has significant implications for glioma progression and tumor behavior. Early evidence suggests that IDH may be a therapeutic target in IDH-mutant gliomas.
Conclusions and Relevance
IDH mutation is a central and defining event in the development and progression of glioma and may be a key target for future therapies for these types of neoplasms.
Over the past 4 years, our understanding of gliomagenesis and the practice of neuro-oncology have been radically changed by the discovery of mutations involving the isocitrate dehydrogenase (IDH) enzymes. Mutations in the IDH1 and IDH2 genes have been postulated to be the initiating event in the development of many gliomas,1 and their presence dictates a particular path for oncogenic progression—and a favorable clinical behavior—in these cancers.2,3
To paraphrase Yang et al,4 oncogenic mutations involving the IDH1 and IDH2 genes share 4 distinct biochemical features. First, IDH1 and IDH2 mutations are predominantly somatic.5 Second, oncogenic IDH mutations in situ are universally heterozygous.6,7 Third, nearly all IDH mutations involve a single amino acid substitution affecting a residue in the enzyme active site: the arginine residue at codon 132 in IDH1 (Table 1) and the arginine residue at codon 172 or codon 140 in IDH2.9 Finally, mutations occur in a mutually exclusive manner in most cases, indicating a common underlying biochemical mechanism and common physiologic consequences.
In addition to these biochemical traits, mutations targeting the IDH1 and IDH2 genes exhibit 3 distinct clinical features. First, they occur in a restricted spectrum of tumors. In gliomas, they occur frequently in grade II and grade III oligodendrogliomas and astrocytomas, and in secondary glioblastomas, but not in primary glioblastomas (Table 2). Similarly, they are frequently found in cytogenetically normal subtypes, but not other subtypes, of acute myelogenous leukemia (AML). Second, IDH1/2 mutations occur at an early stage of tumorigenesis and, in fact, are thought to be the initiating mutation in IDH-mutant low-grade gliomas.1 Finally, in glioma,3 AML,13 and intrahepatic cholangiocarcinoma,13 an IDH1 or IDH2 mutation is associated with a better prognosis. Interestingly, IDH mutations have also been found to play a role in the development of benign cartilaginous tumors, about one-third of which progress to malignant chondrosarcomas.14
Here, we have performed a comprehensive review of the IDH literature to allow us to describe the role of IDH in normal physiology and to speak to aberrations in the IDH pathway that are associated with gliomagenesis. The IDH pathway is summarized in Figure 1. Our review will also report on recent basic work examining strategies to target IDH that may become relevant to clinical practice.
A systematic review of the literature dating from 2008, when IDH mutation was discovered to be clinically significant in glioma, to 2013 was performed using the PubMed database. The following search terms were used: IDH, IDH1, IDH2, and isocitrate dehydrogenase, in conjunction with glioma or leukemia. The search was limited to articles published in English. Further hand searching was performed using a review of the pertinent references from the identified publications. All identified original articles were investigated for content and critiqued by Z. T. and S. D.
Using whole-genome sequencing, Parson and colleagues15 first identified stereotypic somatic mutations at codon 132 of the IDH1 gene in a small proportion of patients with glioblastoma. Their finding was corroborated by the Cancer Genome Atlas Research Network,16 which described IDH1 as defining a clinically relevant subgroup of glioblastoma. IDH1 mutation has been found to be a frequent, likely early event in gliomagenesis, occurring in 70% to 80% of low-grade gliomas. Furthermore, mutation in the IDH genes predicted an alternative pathway of glioblastoma genesis: IDH-mutant tumors were more likely to harbor mutations in TP53 or to reveal a loss of chromosome 1p or 19q, and were less likely to have alterations in PTEN, EGFR, CDKN2A, or CDKN2B.3 As importantly, IDH mutation is associated with better outcomes in high-grade glioma. The median overall survival was 31 months for patients with an IDH-mutant glioblastoma and 15 months for patients with wild-type IDH1, whereas the median overall survival was 65 months for patients with an anaplastic astrocytoma harboring an IDH mutation and 20 months for those without mutations.2,17- 20 Interestingly, IDH mutations were not found in pilocytic astrocytoma, suggesting that these tumors are biologically unrelated to infiltrative gliomas. A summary of IDH and other mutations in glioma is provided in Figure 2.
IDH mutation has also been found to be prognostic of longer survival for patients with low-grade glioma.21- 24 These events are most common in cases of oligodendroglioma (94%), and less so in cases of astrocytoma (72%) or mixed tumors (83%).21IDH mutation in low-grade glioma has a significant positive effect on overall survival (hazard ratio, 0.64), independent of histologic phenotype, and usually predicts the presence of MGMT promoter methylation (in 84% of IDH-mutant low-grade tumors). IDH mutation also appears to impart the benefit of progression-free survival (with a mean [SD] 3-year Kaplan-Meier estimated progression-free survival of 73.9 [4.5] years for patients with an IDH-mutant tumor vs 61.4 [6.9] years for patients with wild-type IDH).22 Conversely, the absence of IDH mutation in low-grade glioma has been found to be predictive of a briefer latency to malignant transformation and a shorter overall survival.23
Houillier and colleagues20 have postulated that the survival benefit garnered by IDH mutation for patients with low-grade glioma is indicative of the effect of IDH mutation on the response to chemotherapy, rather than of a divergent biological behavior. IDH mutation does indeed appear to predict chemosensitivity in both low-grade20 and secondary high-grade25,26 gliomas. Regardless of the mechanism involved, the presence of IDH mutation is clearly associated with improved survival for patients with glioma. In the next sections, we will further describe the biology of the IDH genes and review recent work explicating its role in disease.
IDH1, IDH2, and IDH3 catalyze the reversible decarboxylation of isocitrate to α-ketoglutarate (α-KG, also known as 2-oxyglutarate [2OG]). IDH1 and IDH2 use NADP+ as a cofactor, producing NADPH in the process, whereas IDH3 uses NAD+ as a cofactor and so produces NADH. IDH1 is present in the cytosol and peroxisomes,27,28 whereas IDH2 and IDH3 are located in mitochondria. Under physiologic conditions, isocitrate and α-KG levels are balanced in a manner that reflects the cellular energy state.27- 29
Recent work suggests that cells in a hypoxic environment rely almost exclusively on glutamine-derived α-KG for lipid synthesis; this shift of metabolism away from the Krebs cycle and glucose consumption to glutamine reduction is also characteristic of tumors, including gliomas.30 In glia, IDH also regulates aspects of glutamine and glutamate metabolism, as well as the synthesis of N-acetylated amino acids.31 The IDH enzymes further appear to play a crucial role in cellular protection and response to oxidative and energetic stress. NADPH plays a vital role in the regeneration of the antioxidant glutathione.32 In addition, α-KG itself functions as an antioxidant.33 In a recent study of HT-22 neurons,34 energetic challenge induced by substituting galactose for glucose resulted in increased expression of the IDH enzymes and cessation of proliferation, whereas induction of oxidative stress via glutathione depletion resulted in no alteration to IDH expression. Taken together, these findings indicate that moderate oxidative stress favors the production of NADPH and α-KG via IDH, but that IDH activity may be unaltered or even curtailed during periods of severe oxidative stress to minimize further production of oxygen-free radicals.
The effect of IDH mutation appears to be driven by both a loss of its normal catalytic function and a gain of function. IDH mutation decreases its binding affinity for isocitrate, while increasing its affinity for NADPH. This change abrogates the “forward” catalytic activity of IDH (ie, the conversion of isocitrate to α-KG), while limiting the “reverse” catalytic activity to a partial reaction in which α-KG is reduced but not carboxylated. As a result, cells harboring an IDH mutation experience a relative depletion of α-KG and NADPH and dysfunction of cell processes requiring these substrates.35 Just as, if not perhaps more, important, IDH1 mutation results in the aberrant production of 2-hydroxyglutarate (2-HG).36 Production of 2-HG by mutant IDH1 appears to depend on the presence of the wild-type enzyme, suggesting a mechanism underlying the uniform heterozygosity of an IDH mutation in order to achieve its metabolic effect successfully.37 Mutant IDH1 catalyzes the formation of the 2-HG enantiomer, D-2-HG, while tumor-derived IDH2 mutations produce an alternative 2-HG enantiomer, R-2-HG.38- 40 It is unclear whether this structural difference is of functional consequence. D-2-HG and R-2-HG have both been postulated to act as an oncometabolite in glioma and leukemia cells.41
IDH mutation is thought to be an early if not the initial event in tumorigenesis in IDH-mutant gliomas and leukemias.42,43IDH mutation appears to result in a cell state permissive of transformation. Mutation results in a profound change in the cell methylome; epigenetic changes are thought to be primary drivers of oncogenic evolution in some cancers and perhaps in gliomas.44 In addition, IDH mutation results in a block in cell differentiation and promotion of cell proliferation, 2 other frequent harbingers of carcinogenesis. We will review these hypotheses further in the next sections.
A recent study45 has shown that the mutant 2-HG–producing subunit of the IDH enzyme does not affect the normal reactions carried out by the wild-type α-KG–producing subunit. However, 2-HG exerts a profound effect on the function of α-KG–related processes, particularly affecting the dioxygenases, a family of enzymes involved in methylation of histones and DNA.38,46 2-HG binds to the dioxygenase catalytic core in a nearly identical orientation as α-KG and thereby inhibits binding of the normal substrate. 2-HG accumulation particularly affects the histone lysine demethylases (KDMs) and the TET family of DNA hydroxylases47- 49; changes in the methylome that arise from inhibition of these 2 enzyme families are thought to underlie IDH-directed transformation. In support of this hypothesis, mutations of TET2 have been found in about 22% of AML cases but are mutually exclusive with IDH mutation,39,50 whereas TET2 promoter hypermethylation has been found in low-grade gliomas lacking IDH mutations.7 In addition, 2-HG accumulation has been shown to perturb collagen maturation and basement membrane function and thus may serve as a facilitator of further oncogenic change.51
Expression of IDH1 R132H or IDH2 R172K mutants in immortalized astrocytes results in a significant increase in the repressive trimethylation of H3K9 (H3K9me3) and H3K27 (H3K27me3).52 The resulting methylation patterns mirror those seen in the hypermethylation state and found in some low-grade gliomas and proneural glioblastomas, called the glioma-CpG island methylator phenotype (G-CIMP).53 This concordance extends to astrocytes expressing mutant IDH1 and low-grade gliomas with G-CIMP. Furthermore, whole-genome sequencing of a low-grade glioma cohort set used to generate the methylome data showed that none of the G-CIMP–negative tumors possessed IDH mutation. These findings show that IDH1 mutation results in dramatic, widespread changes in histone methylation and gene expression, and that IDH mutation is likely the driver of the methylation changes that result in the G-CIMP phenotype in gliomas. Acquisition of a hypermethylation phenotype is also seen in endochondromas from patients with Ollier disease and Maffucci syndrome, nonfamilial diseases characterized by the development of multiple benign cartilaginous tumors.14,54 Interestingly, although mutant IDH is expressed in both normal and tumor cells in these patients, tumor development appears to be associated with higher expression levels of the mutant, suggesting that the risk of tumor development and of acquisition of the hypermethylator phenotype is related to cellular levels of 2-HG.
Interestingly, the expression of mutant IDH1 in immortalized astrocytes directed these cells toward a stem cell–like phenotype, characterized by decreased expression of the astrocyte marker, GFAP, and enhanced expression of the neural stem cell marker, nestin.52 Similarly, primary neurospheres infected with an IDH1 R132H mutant retrovirus failed to express markers of differentiation, despite exposure to retinoic acid and other culture conditions that induce astrocytic and neuronal differentiation in control cells. This inhibitory effect on cell differentiation was accompanied by a progressive remodeling of the methylome over successive passages,53 which could be reproduced by short interfering RNA–mediated inhibition of the H3K9-specific Jumonji-C histone demethylase, KDM4C (also known as JMJD2C).52 The acquisition of a progenitor phenotype is thought to be a necessary if not initiating step in gliomagenesis.55 It is unclear, however, if the phenotype change in IDH-mutant cells is solely due to changes in methylation; for example, the inhibition of IDH in glioma and leukemia cells, in vitro, appears to drive cell differentiation without having an effect on the cell methylome.
Proteasomal degradation of hypoxia-inducible factor 1α (HIF-1α) is mediated by its polyubiquitylation by the HIF prolyl 4-hydroxylases, EGLN1, EGLN2, and EGLN3. Koinunen et al56 found that 2-HG accumulation enhances EGLN activity and thus leads to decreased levels of HIF. Furthermore, the introduction of mutant IDH1 into human astrocytes results in a decrease in HIF-1α, parallel to the finding of decreased HIF-1α activity in proneural (IDH1-mutant) glioblastomas. The knockdown of HIF-1α or EGLN1 promotes the proliferation and colony formation of human astrocytes and suggests that EGLN might be a good target for glioma therapy.
Under normal physiological conditions, HIF-1α plays a central role in mediating mitochondrial oxygen consumption and limiting the production of reactive oxygen species.57 Loss of HIF-1α could expose a cell to increased risk of reactive oxygen species–mediated DNA damage and mutation.58 Conversely, HIF-1α activation in glioblastoma has been shown to enhance tumor cell proliferation and angiogenesis.59 It is interesting to postulate that the effect of IDH1 mutation on HIF-1α might be oncogenic by affording an initial proliferative advantage and an increased risk of mutational burden to IDH1-mutant cells, but advantageous to outcome by conferring a less aggressive phenotype to the resulting cancer.
Zheng and colleagues60 developed a series of mutant IDH inhibitors that have little effect on the wild-type enzyme, based on their finding that isocitrate binding to the ligand binding site in wild-type IDH results in a conformational change in the enzyme that enables its catalytic activity, while, in the mutant enzyme, isocitrate binds to an alternative binding site that does not facilitate the necessary conformational change. The 2 candidate inhibitors were found to form hydrogen bonds and electrostatic interactions with a stronger affinity to the alternative isocitrate binding site than that of isocitrate, resulting in the stabilization of the mutant enzyme in its inactive, open conformational state.
Rohle and colleagues61 have developed a small-molecule IDH1 (R132H) mutation–specific inhibitor (AGI-5198) and studied its effects both in vitro and in vivo. Treatment of human glioma cells with AGI-5198 reversed the differentiation block associated with IDH mutation. Interestingly, reduction in tumor volume in murine xenografts following treatment with AGI-5198 occurred despite only partial inhibition of mutant IDH1, a reduction in (but not elimination of) 2-HG, and a lack of a change in CIMP status. Upregulation of differentiation genes and downstream inhibition of histone demethylation required complete inhibition of mutant IDH1. Histopathological analysis of the tumors showed no changes in the concentration of cleaved caspase-3, indicating that the reduction in tumor volume was due to the inhibition of tumor proliferation rather than an increase in cell death.
To be effective, the molecules used in the treatment of IDH-mutant gliomas will need to be capable of penetrating the blood-brain barrier and reaching therapeutic levels within the tumor. The work of Rohle and colleagues61 highlights the functional heterogeneity of the IDH mutation effectors, including the downstream effects of 2-HG, CIMP, and histone modifications; further investigation will be needed to clarify the role of each of these factors in tumorigenesis, the effects on the vulnerability of the tumor to other treatments such as chemotherapy, and their impact on other crucial aspects of cell metabolism. Furthermore, it is unclear whether the effects observed in studies using an oligodendroglioma cell line can be generalized to astrocytoma and glioblastoma cell lines; it has been postulated that the role of IDH may vary among these tumors.62 An IDH-mutant inhibitor probe (ML309) capable of reducing 2-HG levels in a glioblastoma multiforme cell line has also been developed.63
Wang and colleagues64 examined AGI-6780, a urea sulfonamide inhibitor of IDH2 (R140Q), in leukemia. AGI-6780 is a tight-binding allosteric noncompetitive inhibitor of isocitrate and an uncompetitive inhibitor of the NADPH cofactor. Binding of the inhibitor to the mutant enzyme holds the enzyme in an open conformational state, leaving it unable to perform catalysis. In IDH2 (R140Q) AML cells in vitro, AGI-6780 causes a dose-dependent reduction in 2-HG and cell differentiation. Surprisingly, however, and despite its effect on cell “stemness,” treatment with the IDH2 inhibitor in AML cells resulted in an increase in cell proliferation. These findings highlight the contextual specificity of IDH function in oncogenesis and speak to the need to investigate its role in each particular disease.
DNA methyltransferases (DNMTs) are a family of enzymes that catalyze the addition of a methyl group to DNA. The use of DNMT inhibitors in IDH-mutated cancers is motivated by the hypothesis that the oncogenic effects of IDH mutation are driven by its effects on the methylome. Two groups65,66 have recently studied the DNMT inhibitors decitabine and 5-azacytidine in IDH-mutant glioma cell models.
Turcan and colleagues65 observed that the administration of decitabine to an IDH1 R132H mutant anaplastic astrocytoma cell line (TS603), while not affecting 2-HG levels, efficiently induced upregulation of differentiation genes, an effect associated with a change in methylation markers leading to reexpression of Polycomb-controlled genes. Remarkably, the differentiation effect was maintained even after drug therapy was stopped. Of note, treatment with the IDH-mutant enzyme inhibitor AGI-5198 had no further benefit in cells pretreated with decitabine, implying that IDH inhibitors may be redundant when coupled with DNMT inhibitor therapy.
A concurrent study by Borodovsky and colleagues66 examined the DNMT inhibitor 5-azacytidine in an anaplastic astrocytoma xenograft (JHH-273). This inhibitor was able to reverse the G-CIMP hypermethylation state in a dose-dependent manner. As with decitabine, long-term low-dose treatment slowed tumor growth (determined by Ki-67 staining) even after the treatment was stopped, with a similar upregulation of differentiation genes.
Because decitabine and 5-azacytidine are already approved by the US Food and Drug Administration and can cross the blood-brain barrier effectively, they represent exciting new prospective drugs for clinical trial. Furthermore, their effects appear to be translatable to other IDH-mutated tumors such as AML and chondrosarcoma. Further study will be needed to determine the generalizability of these preliminary findings.
In addition to the IDH enzymes and DNMT, there may be other entities in the IDH effector scheme for which inhibitors can be developed that may result in a cumulative clinical benefit. Potential targets include the α-KG–dependent dioxygenases,67 such as TET2,49,50,68 and components of the glutamine/glutamate pathways.69 For example, suppression of BCAT1, an amino acid transferase involved in catabolism of branch-chain amino acids and mutated in glioma in a mutually exclusive fashion with IDH1 and IDH2, resulted in reduced invasiveness and proliferation in vitro and in a glioblastoma multiforme xenograft model.70
Recent studies have broadened our still preliminary understanding of the consequences of a specific point mutation of a single gene for glioma biology. IDH mutation in glioma is an early event, which results in widespread changes to the methylome, the generation of the oncometabolite, 2-HG, and, ultimately, tumorigenesis.
The promise of IDH-directed therapy must be tempered by an appreciation that our understanding of their role in glioma biology is premature. For example, although IDH mutation is sufficient to induce cellular transformation in vitro, it does appear capable of driving transformation in vivo.51,71 The conditions necessary for IDH mutation to be permissive in the physiologic milieu are unclear. Similarly, although dysregulation of the cell redox state has been shown to contribute to transformation in other types of cancer, it is not clear whether NADPH and reactive oxygen species are in fact abnormal in IDH-mutant cancers. Finally, it must be remembered that the studies targeting IDH have been performed using chemical and small-molecule inhibitors rather than short interfering RNA–mediated knockdown and, therefore, cannot exclude the possibility that off-target effects might account for the identified effects. Therefore, the extent of the redundancy of IDH inhibitors when coupled with DNMT inhibitors is also presently unclear and requires further research.
Interestingly, the expression of mutant IDH1 R123H in established IDH wild-type glioma cell lines resulted in a marked decrease in the proliferation of these cells in vitro and increased latency of fatal tumor formation following intracranial xenotransplantation.72 This finding highlights a paradox of IDH biology: that a mutation that should appear to reduce cellular fitness could play a formative role in tumorigenesis. It also requires that the development and application of IDH-directed therapies be considered in a nuanced and complex manner.
At least 2 possible hypotheses can be entertained to explain the causative role of IDH and 2- HG in transformation. First, IDH mutation and 2-HG accumulation may be necessary both to promote transformation and to drive further mutation even after transformation has occurred. Conversely, the process of oncogenic progression may at some point become independent of the initial IDH mutation. Two corollaries result from these divergent hypotheses. In the first case, IDH inhibition would be an effective means of treating an IDH-mutant lesion because a reversal of the effect of IDH on the methylome or a decrease in 2-HG levels would relieve these cells of a persistent driving force for their progression. On the other hand, if the effects of IDH1 are important for initiation but not maintenance or progression of the transformed state, then inhibition of IDH in a mature IDH-mutant cancer would likely be ineffective.
The results of early basic work investigating IDH and DNMT inhibitors suggest an intermediate conjecture situated somewhere between these 2 hypotheses. Future studies will be required to pinpoint the exact role of IDH mutation in glioma initiation and progression and to optimize the targeting of this pathway in cancer therapy.
Accepted for Publication: April 15, 2014.
Corresponding Author: Sunit Das, MD, PhD, Division of Neurosurgery, St Michael’s Hospital, University of Toronto, 30 Bond St, Toronto, ON M5B 1W8, Canada (firstname.lastname@example.org).
Published Online: August 25, 2014. doi:10.1001/jamaneurol.2014.1205.
Author Contributions: Dr Das had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Turkalp, Das.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: Karamchandani, Das.
Study supervision: Das
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Das is supported by awards from the Toronto-based philanthropic organization b.r.a.i.n.child and the Canadian Cancer Society Research Institute.
Role of the Sponsor: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.