Gray Areas in the Gray Matter: IDH1/2 Mutations in Glioma

Abstract

Since the first discovery of isocitrate dehydrogenase (IDH) mutations in cancer, considerable progress has been made in our understanding of their contribution to cancer development. For glioma, this has helped to identify two diagnostic groups of tumors (oligodendroglioma and astrocytoma IDHmt) with distinct clinical characteristics and that are now diagnosed by the presence of the IDH mutations. The metabolic changes occurring as the consequence of the altered substrate affinity of the mutant IDH protein results in a cascade of intracellular changes, also inducing a relative sensitivity to chemotherapy and radiotherapy compared with IDHwt tumors. Pharmacologic blockade of the mutant enzyme with first-in-class inhibitors has been efficacious for the treatment of IDH-mutant acute myeloid leukemia (AML) and is currently being evaluated in phase III trials for IDH-mutant glioma (INDIGO) and cholangiocarcinoma (ClarIDHy). It seems likely that acquired resistance to mutant IDH inhibitors will eventually emerge, and combination therapies to augment the antitumor activity of mutant IDH inhibitors have already been initiated. Approaches to exploit, rather than inhibit, the unique metabolism of IDH-mutant cancer cells have emerged from laboratory studies and are now also being tested in the clinic. Results of these clinical trials are eagerly awaited and will likely provide new key insights and direction of the treatment of IDH-mutant human cancer.

OVERVIEW OF GLIOMA-ASSOCIATED IDH1/2 MUTATIONS (MARTIN J. VAN DEN BENT)

In 2008, a genome-wide sequencing study observed unknown mutations in the gene encoding isocitrate dehydrogenase (IDH1) in 18 (12%) of 149 glioblastoma samples.¹ Remarkably, these mutations occurred in young patients with glioblastomas that had progressed from low-grade gliomas to what was known as secondary glioblastoma. Moreover, patients with glioblastoma with IDH1 mutations had a much better survival compared with patients with glioblastomas without IDH1 mutations. Shortly afterward, a series of studies showed that these mutations occurred in up to 80% of low-grade gliomas, and, in some IDH1wt glial tumors, mutations in the IDH2 gene were found that carried a similar prognostic significance.^2–4 Early clinical studies suggested the occurrence of IDH mutations was an early event in gliomagenesis, occurring before the development of a 1p/19q codeletion.⁵ With more clinical data emerging, it became clear that virtually all 1p/19q codeleted tumors have an IDH1/2 mutation and that nearly all these tumors have MGMT promoter methylation. Other mutations were found associated with IDH1/2 mutations, in particular, mutations in the TP53 and ATRX genes in tumors without 1p/19q codeletion and mutations in the TERT promoter region in IDH1/2-mutated tumors with 1p/19q codeletion. Clinical studies showed that IDH1/2mt gliomas had a better outcome compared with histologically similar tumors without IDH1/2 mutations.^6,7 Further studies showed that a classification of glial tumors based on their molecular profile resulted in a much stronger prognostication compared with classic histology.^8–10 This resulted in the revised 2016 World Health Organization (WHO) classification of brain tumors in which diffuse glioma are now classified according to their IDH1/2 mutational and 1p/19q status.¹¹ In the aftermath of that fundamental change, ongoing discussions (cIMPACT-NOW) have further moved the classification by renaming low-grade astrocytoma with mutations observed in classic glioblastoma (gain of 7 combined with loss of 10, and/or EGFR amplification, and/or TERT promoter mutations only) by “low-grade astrocytoma with molecular features of glioblastoma.”¹² These tumors indeed carry a prognosis that is as dismal as glioblastoma, and the question is whether they should not be labeled as such.¹³ A further cIMPACT report on diffuse glioma entity proposes to revise the name of “glioblastoma, IDHmt” to “astrocytoma grade 4, IDHmt,” and adds additional molecular criteria to that grade 4 diagnosis (i.e., the presence of homozygous deletion of CDKN2A).^14–16 The rationale for this modification is the similarity at the molecular level between these grade 4 tumors and IDH1/2-mutated grade 2 and 3 astrocytoma (e.g., MGMT promoter methylation, sensitivity to radiotherapy) and the overall better outcome in this group. Other molecular indicators of poor prognosis have been proposed, but larger validation cohorts are needed to clarify the role of CDK4 amplification, increased copy number alterations, and absence of CpG island hypermethylation.^14,17,18 Even in the molecular era, classic clinical features beyond histology still remain of prognostic relevance. Age of the patient, tumor size, and the presence of contrast enhancement have been identified as indicators of worse outcome.^19–21

For treatment-related factors, residual tumor after surgery impacts outcome, in particular, grade 2 and 3 IDHmt glioma, with much better survival in completely resected patients.^22,23 All studies on adding chemotherapy after radiotherapy have shown improved outcome in patients with oligodendroglioma or astrocytoma IDHmt. The first reports were inconsistent to the question of whether it was the 1p/19q status that was associated with improved outcome after adjuvant chemotherapy in grade 2 and 3 glioma, the IDH mutations that are invariably present in 1p/19q codeleted tumors, or MGMT promoter methylation that is usually present in IDHmt tumors.^7,24,25 More recently, the benefit to chemotherapy in IDHmt astrocytoma lacking 1p/19q codeletion was confirmed in the CATNON study, which was a study on anaplastic astrocytoma without 1p/19q codeletion.²⁶ A similar benefit in grade 2 astrocytoma IDHmt was observed in a post hoc study on a subset of patients from the pivotal PCV (procarbazine, lomustine, vincristine) in low-grade glioma trial.^27,28 Tumor grade (grade 2 vs. 3) has some impact but less so than in the pre-WHO 2016 era.^14,29

The early observation on the strong association between outcome and IDH mutations initiated further studies into the different clinical characteristics of these tumors. Functional studies resulted in the understanding that IDH-mutated tumors represent a completely different entity on the molecular level. The function of all isocitratedehydrogenases is to catalyze the oxidative decarboxylation of isocitrate into α-ketoglutarate. During that enzymatic reaction, nicotinamide adenine dinucleotide phosphate (NADP+) functions for IDH1 and IDH2 as the electron acceptor. The IDH1 gene is located in the cytoplasm and peroxisomes and IDH2 in the mitochondria. The mutant IDH protein has an altered substrate affinity of the enzyme and catalyzes the conversion of α-ketoglutarate into 2-hydrogylutarate (2HG), during which process NADP+ is reduced to nicotinamide adenine dinucleotide phosphate hydrogen. Because of this, the intracellular concentration of 2HG increases and the intracellular concentration of α-ketoglutarate decreases. Subsequent studies found that 2HG acts as an oncometabolite by altering many cellular functions resulting in genome-wide CpG island hypermethylation, increased double-stand DNA breaks, decrease in NADP, and loss of function because of depletion of α-ketoglutarate–dependent enzymes.^30–36 From these data, two opposite treatment strategies have been developed: reduce the amount of intratumoral 2HG by directly inhibiting the function of mutant IDH enzyme, or do not reduce the amount of 2HG and instead exploit the the cellular consequences of 2HG accumulation, in particular, the impaired DNA repair mechanisms that may underly the increased sensitivity of IDH-mutant gliomas to radiotherapy and alkylating chemotherapy.

INHIBITING THE FUNCTION OF THE MUTANT IDH1/2 ENZYME (INGO K. MELLINGHOFF)

The discovery of somatic IDH mutations in glioblastoma and lower-grade gliomas jettisoned the mutant metabolic enzyme to the top of the list of new drug targets for diffuse glioma.^1–3 Enthusiasm for direct inhibition of the mutant enzyme was based on several observations. First, cancer-associated IDH mutations cluster in key arginine residues within the enzyme’s active site (R132 of IDH1 and R140 or R172 of IDH2), raising the possibility of developing mutantselective inhibitors with a wider therapeutic window. Second, mutant IDH1 (mIDH1) appeared to play a prominent role in the pathogenesis of lower-grade glioma given the extraordinarily high prevalence of IDH mutation in this disease (70%–80%), the distinct DNA hypermethylation pattern associated with IDH mutations,^31,34 and the persistence of IDH-mutant tumor cell clones throughout the disease course.^37,38 Third, 2HG, the direct product of the mutant enzyme, was sufficient to phenocopy the cancer-promoting effects of the mutant IDH enzyme in a reversible fashion,^39,40 providing rationale for maximal 2HG inhibition as biologically relevant and readily quantifiable pharmacologic strategy. Last, inhibition of the mutant IDH enzyme showed antitumor activity in experimental models of glioma and leukemia.^41,42

Nonetheless, the paucity of experimental models that faithfully recapitulate the genetic alterations, tumor microenvironment, and growth patterns of IDH-mutant low-grade glioma has made it challenging to establish the role of mutant IDH in established tumors. In vitro studies using IDH1-mutant cancer cell lines, which often harbor genetically more complex genomes, suggest that several factors might relieve mutant IDH cancer cells from their dependency on the mutant enzyme for survival.⁴³

The clinical development of inhibitors of mIDH proceeded most expeditiously in patients with AML, another cancer type found to harbor mutations in IDH1 and, more commonly, IDH2.^44,45 Enasidenib, the first-in-class inhibitor of mIDH2, produced clinical responses in approximately 40% of patients with advanced mIDH2 AML.^46,47 Ivosidenib, the first-in-class inhibitor of the mutant IDH1 enzyme, similarly induced clinical and molecular remissions in patients with advanced mIDH1 AML.⁴⁸ Both drugs were recently approved in the United States for the treatment of mIDH AML.

The role of mutant IDH in glioma and other solid tumors is currently under investigation. A phase I study with ivosidenib in subjects with advanced solid tumors, including glioma, with an IDH1 mutation (ClinicalTrials.gov identifier: NCT02073994) was initiated in March 2014 across 12 study sites in the United States and one in France and included 66 with glioma. Twelve of 66 (18.2%) patients had glioblastoma; the remainder had lower-grade glioma. The median number of prior systemic therapies was two (range, one to six), and 49 of 66 patients had received prior radiotherapy. No dose-limiting toxicities were reported, and the maximum tolerated dose was not reached. A dose of 500 mg once daily was selected for expansion based on the pharmacokinetic/pharmacodynamic data from all solid tumor cohorts and preliminary clinical activity observed in the dose-escalation phase. As of the data cutoff, patients with nonenhancing gliomas had a median treatment duration of 18.4 months (range, 1.4–47.2 months), and 15 patients (22.7%) remained on treatment. An exploratory analysis showed a reduction in tumor volume growth rates (i.e., compared with pretreatment growth rates) and tumor shrinkage in several patients.⁴⁹ Interestingly, tumor shrinkage was primarily noted in tumors that did not enhance on MRI after contrast injection, despite the fact that these tumors presumably have an intact blood–brain barrier. A subsequent clinical trial (ClinicalTrials.gov identifier: NCT03343197) showed greater than 90% inhibition of the mutant IDH enzyme by ivosidenib in on-treatment biopsies from patients with nonenhancing gliomas.⁵⁰ Both studies have completed enrollment. Similar to the experience in glioma, ivosidenib was well tolerated and showed antitumor activity in patients with IDH1-mutant advanced cholangiocarcinoma.^51,52

Vorasidenib (AG-881) is a first-in-class dual inhibitor of mutant IDH1 and 2, which was designed for enhanced brain penetrance. Vorasidenib was well tolerated and showed preliminary evidence of antitumor activity in a now completed phase I trial in patients with nonenhancing glioma (ClinicalTrials.gov identifier: NCT02481154).⁵³ In the above-mentioned perioperative trial (ClinicalTrials.gov identifier: NCT03343197), vorasidenib resulted in consistent and dose-dependent 2HG suppression with greater than 90% reduction at 50 mg daily compared with untreated controls. Preliminary efficacy data show objective tumor responses (approximately 30%) and durable disease control with postoperative treatment. Clinical responses were also reported, with the mutant IDH1 inhibitor DS-1001b responses in a phase I study in Japan.⁵⁴

The collective data from the early clinical trials with mIDH inhibitors in glioma suggest that lower-grade gliomas that do not enhance on MRI after contrast injection (so-called nonenhancing) may be susceptible to the antitumor activity of mIDH inhibitors. Because contrast enhancement in low-grade gliomas is often associated with transformation to a higher tumor grade and acquisition of additional genetic alterations, it seems plausible that the role of mIDH for tumor maintenance is greatest at the earliest disease stage and before additional radiation and chemotherapy.⁵⁵ This finding is reminiscent of the enhanced responsiveness of early-stage chronic myelogenous leukemia to targeted BCR-ABL kinase inhibition.⁵⁶ This conclusion led to the design of a randomized, placebo-controlled phase III trial of AG-881 in participants with residual or recurrent grade 2 glioma with an IDH1 or IDH2 mutation (INDIGO; ClinicalTrials.gov identifier: NCT04164901). This trial is currently open to accrual and will enroll patients who have not yet received radiation or chemotherapy and are under radiographic surveillance (watch and wait) following their initial surgery.

EXPLOITING ONCOMETABOLITE-INDUCED DEFECTS IN IDH1/2-MUTANT GLIOMAS (RANJIT S. BINDRA)

As mentioned previously, there are more than 70 α-ketoglutarate (αKG)-dependent dioxygenases in the cell, which perform a diverse range of functions including epigenetic regulation, DNA repair, and metabolism. Mutant IDH1/2-induced 2HG typically exceeds a concentration of 1 to 2 mM in cells, and thus it is likely that nearly all of these dioxygenases are inhibited, albeit to varying degrees, by such high levels of oncometabolite production.⁵⁷ Inhibition of even a single αKG-dependent dioxygenase, or subset of these proteins, can induce profound changes in the cell. For example, the TET family of 5 methylcytosine hydroxylases are αKG-dependent dioxygenases and thus are inhibited by 2HG.⁵⁸ Loss of these proteins promotes a CpG island methylator phenotype, which leads to silencing of numerous loci in the genome.^31,59 Based on these findings, an alternative therapeutic strategy has been proposed to exploit vulnerabilities associated with oncometabolite production rather than suppressing mutant IDH1/2 neomorphic enzymatic activity. This type of approach overlaps with the concept of synthetic lethality, in which targeting one of two related pathways does not affect viability, but targeting both pathways simultaneously leads to cell death.⁶⁰ Synthetic lethality gained momentum after the publication of two seminal papers in the early 2000s, which reported that mutations in the homologous recombination (HR) genes, BRCA1 and BRCA2, confer exquisite sensitivity to PARP inhibitors.^61,62 As multiple PARP inhibitors are now U.S. Food and Drug Administration approved for the treatment of BRCA1/2-mutant cancers, the concept of synthetic lethal tumor targeting is now well accepted as a potential efficacious strategy in oncology. In the following, we discuss four therapeutic classes of drugs that have been proposed for use to exploit oncometabolite-induced vulnerabilities associated with IDH1/2-mutant gliomas.

PARP Inhibitors

In 2017, the Bindra and Glazer Laboratories were the first to report that mutant IDH1/2 induces an HR defect that confers PARP inhibitor sensitivity.³³ Mechanistically, it was found that 2HG-induced suppression of two histone demethylases, KDM4A and KDM4B, induced an HR defect similar to that associated with BRCA1/2 mutations. Soon after this initial discovery, several other groups reported a similar synthetic lethal interaction between PARP inhibition and mutant IDH1/2 across a wide range of tumor types.^63–65 Of note, the phenomenon of oncometabolite-induced HR suppression was further extended to tumor-associated mutations in other tricarboxylic acid cycle genes, such as fumarate and succinate.⁶⁶ This novel approach to target IDH1/2-mutant cancers with PARP inhibitors is now being tested in the clinic for glioma (ClinicalTrials.gov identifiers: NCT03749187 and NCT03914742), as well as many other tumor types, including chondrosarcoma and cholangiocarcinoma (ClinicalTrials.gov identifiers: NCT02576444 and NCT03212274). These trials collectively are testing the efficacy of two PARP inhibitors: olaparib and BGB290. Both of these drugs are notable for high potency, which likely can be attributed to their action as PARP-trapping PARP inhibitors.⁶⁷ Relevant to glioma, BGB290 appears to demonstrate relatively high levels of blood–brain barrier penetration in animal models.⁶⁸ Olaparib appears to penetrate at least T1-postcontrast-enhancing areas of disease in glioma, based on the results of the OPARATIC trial (ClinicalTrials.gov identifier: NCT01390571).⁶⁹ This has prompted its evaluation in two additional glioblastoma trials, PARADIGM-1 and −2, which do not select for patients based on IDH1/2 mutation status.⁷⁰ Another, albeit less potent, PARP inhibitor, veliparib, is also being tested in a cohort of pediatric patients with newly diagnosed glioma with IDH1/2 mutations (ClinicalTrials.gov identifier: NCT03581292).

Alkylating Agents

It has also been shown that mutant IDH1/2-induced 2HG impairs alkylation damage repair via inhibition of the ALKBH family of αKG-dependent dioxygenases, ALKBH2 and ALKBH3, leading to enhanced sensitivity to multiple alkylating agents, including temozolomide.^71,72 It should be noted that IDH1/2 mutations frequently co-occur with MGMT promoter silencing, which is linked to the phenomenon of 2HG-induced CpG island methylator phenotype via TET inhibition discussed previously. MGMT loss also confers alkylator sensitivity, but this pathway is distinct from ALKBH, and thus loss of both pathways in IDH1/2-mutant gliomas likely results in sensitivity to this class of drugs, which is at least additive. Collectively, these findings reveal two pathways via which oncometabolites induce alkylator sensitivity: (1) direct inhibition of ALKBH2/3 by 2HG and (2) MGMT silencing via 2HG-induced TET inhibition and CpG island methylator phenotype. This also suggests that mutant IDH1/2 inhibitors should not be given concurrently with alkylators, given the potential for antagonistic interactions, at least based on the effects of 2HG on ALKBH2/3 function.

BCL2 Family Inhibitors

A synthetic lethal interaction between mutant IDH1/2 and members of the BCL-2 family was first reported in AML tumor models³ but has since been shown in IDH1/2-mutant glioma.^73,74 BCL2 inhibitors have been extensively tested in the clinic, and one drug is now U.S. Food and Drug Administration approved for chronic lymphocytic leukemia (venetoclax).⁷⁵ As such, this strategy can be readily and feasibly tested in the clinic for IDH1/2-mutant gliomas in the future.

NAMPT Inhibitors

Tateishi and colleagues also reported that IDH1/2-mutant gliomas are sensitive to a class of drugs that target nicotinamide adenine dinucleotide (NAD) metabolism, called nicotinamide phosphoribosyltransferase (NAMPT) inhibitors.⁴³ Mechanistically, they demonstrated that IDH1/2 mutations silence a key gene in NAD metabolism, nicotinate phosphoribosyltransferase (NAPRT), which leads to decreased levels of NAD+ and consequent NAMPT inhibitor sensitivity. In a subsequent study, the authors demonstrated that temozolomide and NAMPT inhibitor combinations were particularly active in IDH1/2-mutant glioma models.⁷⁶ Multiple potent NAMPT inhibitors have been developed and were tested previously in numerous clinical trials, and there is currently one drug actively being tested in patients: KPT-9274 (ClinicalTrials.gov identifier: NCT02702492).⁷⁷ As such, there is great potential to test these drugs in IDH1/2-mutant glioma in future clinical trials.

CONCLUSIONS AND FUTURE DIRECTIONS

Since the first discovery of IDH mutations in cancer, considerable progress has been made in our understanding of their contribution to cancer development. For glioma, this has helped to identify two diagnostic groups of tumors (oligodendroglioma and astrocytoma IDHmt) with distinct clinical characteristics. Pharmacologic blockade of the mutant enzyme with first-in-class inhibitors has been efficacious for the treatment of IDH-mutant AML and is currently being evaluated in phase III trials for IDH-mutant glioma (INDIGO) and cholangiocarcinoma (ClarIDHy). It seems likely that acquired resistance to mutant IDH inhibitors will eventually emerge, and combination therapies to augment the antitumor activity of mutant IDH inhibitors have already been initiated. This will require a much deeper understanding of which of the many cellular and molecular effects associated with intratumoral 2HG accumulation in glioma are reversible following inhibition of the mutant enzyme. Approaches to exploit, rather than inhibit, the unique metabolism of IDH-mutant cancer cells have emerged from laboratory studies and are now also being tested in the clinic. Results of these clinical trials are eagerly awaited and will likely provide key new insights and direction on the treatment of IDH-mutant human cancer.

PRACTICAL APPLICATIONS.

• Grade 4 tumors with IDH mutations will now be labeled astrocytoma, IDHmt grade 4.

• IDH-mutated tumors share a metabolic background: the accumulation of 2HG and a reduction of α-ketoglutarate.

• 2HG is an oncometabolite and is responsible for many of the early oncogenic events in astrocytoma and oligodendroglioma.

• Drugs inhibiting the mutant IDH protein are effective in AML and cholangiocarcinoma and are currently being tested in glioma.

• Other metabolic alterations in IDH mutations present other possible avenues for treatment, such as PARP inhibitors.