A Closer Look at IDH Mutations

If you have a diagnosis of a grade II or III glioma (which includes astrocytomas, oligodendrogliomas and mixed oligoastrocytomas), I highly recommend asking your oncologist to have your tumour tissue tested for the IDH1 mutation. IDH2 mutations may also occur in gliomas, but far less frequently.

The IDH1 gene encodes an enzyme called isocitrate dehydrogenase 1. A mutation in this gene was discovered in a small percentage of glioblastoma samples in 2008 [1], and has since been found in a majority of grade II and III gliomas. While a small percentage of primary glioblastomas show mutations of IDH1, it is hypothesized that these cases are in reality secondary glioblastomas, likely having evolved from lower grade gliomas which were undetected at earlier stages in tumour development. On the other hand, grade II and III gliomas without mutant IDH could be considered “pre-glioblastomas” [2], having a poorer prognosis than IDH mutant tumours of the same grade.

It appears that IDH mutant gliomas are a distinct disease entity, having a different genetic and epigenetic basis as compared to IDH wild-type (unmutated) gliomas. Grade 1 pilocytic astrocytomas are in a different category altogether, stemming from genetic aberrations other than IDH mutations.

IDH mutations tend to occur in younger brain tumour cases, most commonly between the ages of 20 and 40, with a median age at diagnosis in the 30s. The mutation is also associated with tumours of the frontal lobe as approximately 70% of IDH-mutated gliomas are located there [3].

Figure 3 Lai 2011 IDH1-mut age and location

Figure 3A shows most common locations in the brain for IDH1-mutant and wild-type glioblastomas. 3B shows that IDH1-mutant glioblastomas tend to cluster in the frontal lobe near the rostral extension of the lateral ventricle. Figure 3C shows the age distribution of IDH1-mutant anaplastic astrocytomas and glioblastomas. From “Evidence for Sequenced Molecular Evolution of IDH1 Mutant Glioblastoma From a Distinct Cell of Origin” Lai et al. 2011.

IDH Mutation Frequencies in grade II and III gliomas

The largest single analysis of IDH mutations in lower grade gliomas was a European study of 1,010 diffuse glioma samples (WHO grades II and III) led by researchers in Heidelberg, Germany (Christian Hartmann, Andreas von Deimling et al.) [4]. From this large study we have a good overview of the frequencies of both IDH1 and IDH2 mutations in lower grade gliomas.

IDH1 mutations occurred in:

  • 165 of 227 (72.7%) grade II astrocytomas
  • 146 of 228 (64%) grade III astrocytomas
  • 105 of 128 (82%) grade II oligodendrogliomas
  • 121 of 174 (69.5%) grade III oligodendrogliomas
  • 62 of 76 (81.6%) grade II oligoastrocytomas
  • 117 of 177 (66.1%) grade III oligoastrocytomas
  • 716 of 1010 (70.9%) total of all 6 subgroups.

IDH2 mutations occurred in:

  • 2 of 227 (0.9%) of grade II astrocytomas
  • 2 of 228 (0.9%) of grade III astrocytomas
  • 6 of 128 (4.7%) of grade II oligodendrogliomas
  • 9 of 174 (5.3%) of grade III oligodendrogliomas
  • 1 of 76 (1.3%) of grade II oligoastrocytomas
  • 11 of 177 (6.2%) of grade III oligoastrocytomas
  • 31 of 1010 (3.1%) total of all 6 subgroups

IDH Mutant Pathobiology

Normally, the IDH1 enzyme catalyzes the conversion of isocitrate to alpha-ketoglutarate. Mutant IDH1 acquires an alternative activity, catalyzing the conversion of alpha-ketoglutarate to 2-hydroxyglutarate (2-HG). 2-HG is considered to be an “oncometabolite” which accumulates to high levels and competitively inhibits alpha-ketoglutarate dependent enzymes, such as histone demethylases and TET2, which mediates DNA demethylation. The outcome is an abnormal DNA hypermethylation profile in these cells resulting in a tumour category known as G-CIMP or Glioma CpG Island Methylator Phenotype [5].

Figure 1 Mellai IDH1

The normal IDH1 enzyme converts isocitrate into alpha-ketoglutarate. The mutant IDH1 enzyme converts alpha-ketoglutarate into 2-hydroxyglutarate (2-HG). From “The Distribution and Significance of IDH Mutations in Gliomas” by Marta Mellai et al. published in “Evolution of the Molecular Biology of Brain Tumors and the Therapeutic Implications”

CpG refers to a DNA sequence where a cytosine molecule is followed by a guanine (cytosine and guanine are two of the four bases which compose the DNA code). A CpG island is a region of DNA in which C-G sequences (cytosine followed by guanine) occur at high frequency. CpG islands frequently occur in the promotor region of genes, which are involved in the initiation of gene transcription (or activation). Methylation of a gene promoter results in the silencing, or de-activation of a gene. Therefore a CIMP tumour is a tumour with widespread gene silencing.

CpG island methylation UCSF school of medicine

Image courtesy UCSF School of Medicine

Some of the most commonly silenced genes in G-CIMP tumours include genes involved in the retinoid pathway, which is important for normal cell differentiation; tumour suppressor genes such as PTEN and RASSF1A; and DNA repair genes such as MGMT. The silencing of MGMT however, renders the tumour more vulnerable to alkylating chemotherapy such as temozolomide.

Virtually all IDH1 mutant gliomas are MGMT methylated

An important study [11] published in 2011 by researchers at University of Cambridge showed that virtually all IDH1/2 mutant gliomas (127 of 129, 98%) also had significant levels of methylation of the MGMT gene, specifically at 16 CpG sites which are most critical in determining MGMT gene transcription, or lack thereof. An average level of methylation of 10% for these 16 sites was the cutoff dividing tumours classed as methylated versus unmethylated. This study differs from other studies in that MGMT methylation was determined by pyrosequencing, which was found to be a more sensitive method than the conventional methylation specific PCR method. Furthermore, in the astrocytoma grade II, oligodendroglioma grade II, anaplastic oligodendroglioma, and anaplastic oligoastrocytoma groups, all but two tumours classed as MGMT methylated were also IDH1/2 mutant, demonstrating a nearly complete overlap of IDH mutation and MGMT methylation in those groups.

This correlation between IDH mutation and MGMT methylation is surely one factor in the more positive prognosis of IDH mutant tumours versus IDH wild type, as silenced MGMT renders a tumour more vulnerable to conventional alkylating chemotherapies such as temozolomide.

Different patterns of recurrence and prevalence of glioma stem cell markers in anaplastic gliomas according to IDH status

A retrospective study [25] published online in Neuro-Oncology (June 23, 2014) by Japanese researchers showed that anaplastic gliomas tended to have different patterns of recurrence depending on the IDH mutational status. 86 anaplastic glioma (astrocytoma, oligodendroglioma, and oligoastrocytoma) patients were included in the study. 63 of the patients had been treated with radiation and ACNU (nimustine) chemotherapy.

Recurrence patterns

Of 40 patients with tumour recurrence, 18 had IDH-mutant and 22 had IDH wild-type tumours. 15 of 18 (83%) IDH-mutant tumours recurred locally. In contrast, only 10 of 22 (45%) IDH wild-type tumours recurred locally (at, adjacent to, or contiguous with the original tumour site / resection cavity). 17 of 18 (94%) IDH-mutant tumours recurred within the radiation field, while only 13 of 22 (59%) IDH wild-type tumours recurred “infield”. These differences in pattern of recurrence between IDH-mutant and IDH wild-type anaplastic gliomas were statistically significant (p=.022 and p=.013).

IDH mutation and better response to treatment

Improved survival in IDH-mutant grade III astrocytoma and glioblastoma with maximal surgical resection

A study [18] published online in Neuro-Oncology in December 2013 demonstrated that IDH-mutant malignant astrocytomas are more amenable to complete surgical resection of enhancing tumour (93% complete resections in the IDH-mutant group versus 67% in the IDH wild-type group). In contrast with IDH wild-type grade III and IV astrocytomas, which didn’t show significant benefit from further resection of non-enhancing tumour, the IDH-mutant grade III and IV astrocytoma group showed a survival benefit with maximal resection of both enhancing and non-enhancing tumour.

The IDH-mutant group consisted of 113 patients (27 glioblastoma [24%] and 86 anaplastic astrocytoma [76%]), while the IDH wild-type group consisted of 222 patients (180 glioblastoma [81%] and 42 anaplastic astrocytoma [19%]). Only 40% of the IDH-mutant tumours showed pre-operative contrast enhancement on MRI, while 87% of the IDH wild-type tumours showed pre-op contrast enhancement. Only 3 of 113 (3%) of the IDH-mutant tumours showed post-operative contrast enhancement, meaning that nearly all of the surgeries in the IDH-mutant group involved complete resection of the enhancing tumour volume. 29% of the IDH wild-type tumours showed post-operative contrast enhancement.

By Kaplan-Meier estimate, the median survival in the IDH-mutant group was 163.4 months (13.6 years) with a median follow-up of 4 years for the surviving patients. As already noted, 60% of these patients were without pre-operative contrast enhancement on MRI, and 97% were without post-operative contrast enhancement. For the 45 IDH-mutant patients with pre-operative contrast enhancement on MRI, median survival by Kaplan-Meier estimate was 118.7 months (9.9 years). As already noted, nearly all patients in the IDH-mutant group were without post-operative contrast enhancement (ie. had total resection of the enhancing tumour).

Median survival was not different in the IDH wild-type group receiving total resection of enhancing and nonenhancing tumour versus total resection of enhancing tumour only. In contrast, in the IDH-mutant group median survival was significantly improved with maximal resection of both enhancing and nonenhancing tumour mass. For IDH-mutant tumours with over 5 cubic cm of residual nonenhancing plus enhancing tumour, median survival was 9.75 years. Median volume of residual tumour (enhancing plus nonenhancing) for the entire IDH-mutant group was 16.5 cubic cm. For IDH-mutant tumours with total resection of enhancing and nonenhancing tumour (residual volume less than 5 cubic cm), median survival was not reached at the time of analysis.

IDH1-mutant anaplastic astrocytoma and glioblastoma long-term survival with complete resection of enhancing and non-enhancing tumour. Figure 2, Beiko et al. 2013.

This study shows a benefit for maximal surgical resection (enhancing and non-enhancing tumor) of IDH-mutant anaplastic astrocytomas and glioblastomas. Whether or not this is achievable depends to a large extent on the precise location and degree of infiltration of the tumour, the surgical technique used and the skill of the surgeon.

Increased response to concurrent radiotherapy and temozolomide in IDH-mutant glioblastoma

Another study [17] published online in Neuro-Oncology in December 2013 compared the response to conventional radiation with concurrent temozolomide treatment in 10 newly diganosed IDH-mutant glioblastoma patients versus 29 glioblastoma patients with IDH wild-type. Both contrast-enhanced T1 and T2/FLAIR MRI images were compared between the groups. All 10 IDH-mutant patients had tumour shrinkage, with decreased contrast enhancement on MRI following therapy, compared with only 13 of 29 (45%) of the IDH wild-type patients having the same effect. The average weekly change in contrast-enhanced T1 and T2/FLAIR images was -3.6% and +0.6% for the IDH-mutant group versus +0.8% and +5.2% for the IDH wild-type group. Median progression-free survival following treatment was 72.1 months (6 years) in the IDH-mutant group versus 8.1 months in the IDH wild-type group.

Figure 2 Tran 2013 IDH1 increased reponse to radiochemotherapy

Response to radio-chemotherapy in IDH1-mutant or IDH1 wild-type glioblastoma patients as measured by MRI (T2/FLAIR or contrast-enhanced). Figure 2, Tran et al. 2013.

IDH mutation predicts response to temozolomide in newly diagnosed grade II glioma

A retrospective analysis [12] published in 2010 reviewed the records of 271 grade II glioma patients. 84 of these patients received adjuvant temozolomide therapy prior to any other treatment other than surgery. In this group of 84: median age was 39; median KPS was 90; 66% were oligodendrogliomas, 21% were oligoastrocytoma and 13% were astrocytoma; 24% had a gross total resection, 27% a partial resection, and 49% had biopsy only.

  • Patients with combined IDH mutation and 1p/19q codeletion had the best response to TMZ with a response rate of 80%, and a response plus stable disease rate of 93%. Median progression-free survival was 37.9 months for this subgroup.
  • Patients with IDH mutation but no 1p/19q codeletion had the next best response to TMZ, with a response rate of 33% and a response plus stable disease rate of 92%. Median progression-free survival was 32.9 months for this subgroup.
  • Patients with neither IDH mutation nor 1p/19q codeletion had the poorest response to TMZ, with a response rate of 16%, and a response plus stable disease rate of 41%. Median progression-free survival was 18.7 months in this subgroup.

These results demonstrate that in this population of low grade gliomas, predominantly oligodendroglioma and oligoastrocytoma, both 1p/19q codeletion and IDH mutation predicted a better response to TMZ. Patients with IDH mutation but no 1p/19q codeletion also had better response to TMZ than patients with neither alteration. Therefore, both IDH mutation and codeletion of chromosomes 1p and 19q may be independent predictors of positive response to TMZ in grade II gliomas. Interestingly, in the subgroup of 171 patients who received no adjuvant therapy from the time of surgery to the time of disease progression, progression-free survival was not different between IDH-mutated and IDH non-mutated cases. This shows that in grade II glioma, the improved survival outcome seen in IDH-mutated cases may be related to increased response to adjuvant therapies such as TMZ or radiation, rather than IDH mutation being an independent prognostic factor. At least three other retrospective studies agree with this finding, showing that in grade II gliomas, IDH1 mutation has either little prognostic value [13], or has a survival value that comes into play only later on, possibly being dependent on better response to conventional treatments [14, 15].

Reduced angiogenesis in IDH-mutant grade 2/3 gliomas

In November 2015, an important study appeared in Scientific Reports, demonstrating significant differences in the biology of IDH-mutant lower grade gliomas relative to IDH non-mutant lower grade gliomas [27]. Diffuse or anaplastic gliomas of WHO grades 2 and 3 were included in this study. Data on mRNA expression levels in 288 low grade glioma samples was downloaded from The Cancer Genome Atlas. When these samples were stratified by IDH status, IDH1 or IDH2 mutant tumors were found to have significantly inhibited (relative to IDH non-mutant tumors) expression of various upstream regulators of angiogenesis, including HIF1A, VEGF, PDGF, and angiopoietin-2 (ANGPT2). Downstream biological function analysis also revealed significant decreases in the following categories in IDH-mutant versus IDH non-mutant lower grade gliomas: development of blood vessel, migration of endothelial cells, vasculogenesis, movement of endothelial cells, angiogenesis, neovascularization, vascularization, and development of endothelial cells.

Relative cerebral blood volume (rCBV) imaging, “a robust and clinically meaningful estimate of tumor angiogenesis”, for 73 treatment-naive lower grade glioma samples from the University of Heidelberg was next analyzed. IDH1 or IDH2-mutant lower grade gliomas were found to have significantly lower rCBV compared with IDH non-mutant lower grade gliomas. rCBV was also found to be a relatively useful predictor of IDH status: a one unit increase in rCBV corresponded to a 69% decrease in the odds of an IDH1/2 mutation, and rCBV correctly predicted IDH mutation status in 88% of patients.

This study confirms previous reports that HIF1A (hypoxia-inducible factor 1-alpha) expression is reduced in IDH-mutant gliomas, rather than increased as was originally theorized. Furthermore, expression of transcriptional targets of HIF1A, including many critical pro-angiogenic genes, are also reduced in IDH-mutant lower grade gliomas in comparison to IDH non-mutant lower grade gliomas. To confirm this data, relative cerebral blood volume, an estimate of tumor angiogenesis, is reduced in IDH-mutant tumors relative to IDH non-mutant tumors of WHO grades 2 and 3. A reasonable conclusion is that anti-angiogenic strategies are less likely to succeed in grade 2 and 3 astrocytomas and oligodendrogliomas mutant for IDH1 or IDH2.

Novel Therapies Targeting Mutant IDH1

As mentioned before, the IDH1 mutation was first discovered in gliomas in 2008 during genomic analysis of glioblastoma. Five years later, in 2013, the mutation is being targeted in preclinical studies by one experimental drug and two epigenetic drugs already FDA approved (in 2004 and 2006) for myelodysplastic syndrome.

IDH1 R132H vaccines

When you hear “mutant IDH1” what is usually meant is the specific mutation IDH1 R132H. This is the IDH1 mutation occuring in the overwhelming majority of people with an IDH1-mutant glioma, though other variations are possible. In a parallel development, researchers in Heidelberg, Germany and at Duke University in the USA are independently working on IDH1 vaccines, to stimulate an immune response to specifically target the mutation and eradicate any cell carrying the mutation. Trials are being planned by both the Heidelberg researchers and the Duke researchers. An abstract for a session which will be presented by Michael Platten (Heidelberg) at the annual meeting of the American Academy of Neurology in Philadelphia on April 29 2014, reveals that the vaccine worked to control IDH1-mutant tumours in immunocompetent mice. A human trial for recurrent, IDH1-mutated grade II glioma is scheduled to begin in October 2014 at Duke University in North Carolina, and another trial is expected to begin early 2015 in Heidelberg Germany.

A research paper [26] was published online in Nature on June 25, 2014 by scientists based in Heidelberg, Germany (Schumacher, von Deimling, Wick, Platten et al.), detailing the response of IDH1 R132H tumour-bearing mice to the IDH1 R132H peptide vaccine.

MHC (major histocompatibility complex) class I and II are cell surface protein complexes which present peptides to immune cells, such as T-cells. This study utilized transgenic mice with humanized MHC expression, and devoid of mouse MHC. When these transgenic mice were vaccinated with a peptide containing the IDH1 R132H mutation, a mutation-specific CD4+ T-cell and antibody response was induced. In the efficacy study, carcinogen-induced mouse sarcoma cells were engineered to express the IDH1 R132H mutation, and were then injected subcutaneously into the flanks of the mice. On day 6, when tumours were detectable, mice were divided into two groups and either vaccinated or sham-treated. Booster vaccine was given on day 13. As shown in figure 3e of the study, tumours in the sham-treated mice grew steadily until the experiment was terminated at day 28. In contrast, tumours of the IDH1 R132H peptide vaccinated mice scarcely grew at all until day 22, at which time tumours started to grow at a very slow rate. At day 28, tumours of the sham-treated mice had reached about 90 square mm in volume, while tumours of the vaccinated mice were less than 30 square mm in volume. This difference would have certainly been even larger had the experiment been continued for a longer duration. The immune response in vaccinated mice was dependent on CD4+ T cells, as well as CD19+ B cells, as depletion of either of these populations abrogated the efficacy of the vaccine.

Figure 3 Schumacher 2014 smaller

Figure 3b shows the effects of preventative pre-treatment with IDH1 R132H peptide vaccine in mice later implanted with subcutaneous IDH1-mutant sarcomas. Figure 3e shows the effects of IDH1 R132H peptide vaccine given after establishment of subcutaneous IDH1-mutant tumours. Red line indicates size of tumours with vaccine and black line indicates sham treatment. From Schumacher et al. 2014.

In summary, vaccination of humanized-MHC mice with an IDH1 R132H-containing peptide vaccine largely controlled mutant IDH1 tumours over a 28 day period. This tumour control was dependent on CD4+ T-cells as well as CD19+ B cells. The safety and immunogenicity of IDH1 R132H-containing peptide vaccine in glioma patients will be tested in the RESIST trial at Duke University beginning in October 2014, and soon after that in a separate trial in Heidelberg, Germany.

March 18, 2015
IDH1 R132H vaccine tested in intracranial murine GL261 glioma model. In this study, the mouse glioma cell line GL261 was genetically modified to carry the IDH1 R132H mutation. Treatment with IDH1 R132 peptide vaccine prolonged median survival slightly, and “cured” 25% of the mice (out to 100 days) while all untreated control mice were dead by around day 30.

Agios Pharmaceuticals and AG-120

The first published report of a drug specifically targeting IDH-mutant glioma cells appeared in the May 2013 edition of Science [6]. In this study, an experimental agent called AGI-5198 was shown to dose-dependently inhibit the IDH mutant associated oncometabolite, 2-hydroxyglutarate (2-HG). Further, when mice implanted with endogenous IDH1 mutated glioma cells were administered oral doses of AGI-5198, their subcutaneous tumours were reduced by 50-60% compared to control mice after two weeks of daily treatment. No toxicity was observed.

In a November 2013 press release, Agios Pharmaceuticals, the manufacturer of AGI-5198, announced that “in early 2014, Agios anticipates submitting an IND (Investigational New Drug) and initiating Phase 1 clinical trials for AG-120 in patients with advanced solid and hematological malignancies that carry an IDH1 mutation.”

November 8, 2015
First data from phase 1 dose escalation study of AG-120 (mutant IDH1 inhibitor) for IDH1-mutant solid tumors was presented today. The press release may be found here, and sign up to listen to the webcast and view the slides here.

62 patients were included in the present report, with a data cutoff of September 3, 2015. There were 20 glioma patients (9 low grade and 11 grade 3/4). Investigators found that AG-120 is able to lower the levels of 2-hydroxyglutarate (2-HG) in the brain. 2-HG is the oncometabolite produced by the enzymatic activity of mutant IDH1. The dose level of 500 mg twice daily was chosen for expansion cohorts.

AG-120 was safe, with no dose limiting toxicity and the maximum tolerated dose was not reached. Most frequent adverse effects in the entire population were low-grade nausea (26%), low-grade diarrhea (16%), vomiting (16%), high or low-grade anemia (15%), QT prolongation (15%), low-grade fatigue (13%), headache (11%), peripheral edema (11%), abdominal pain (10%), and ascites (10%).

Looking at the graph from page 32 of the presentation (see below), six of nine (66.7%) grade 2 gliomas responded with disease stabilization or better, with only three not responding to treatment. One of these six had a strong response with large reduction of tumor size, and of these six, five are still on treatment in the trial, at approximately 14 – 30 weeks from enrollment. Four of eleven (36%) grade 3/4 gliomas responded to treatment with disease stabilization. Two of these four are still on treatment in the trial, at approximately 11 and 20 weeks from enrollment, respectively. Of the 16 high or low grade glioma patients starting treatment at least six months prior to the data cutoff date (September 3), four (25%) were without disease progression at the six month (26 week) mark.


Also of note, the Agios drug AG-881, mutant IDH1/2 inhibitor, was said in this presentation to have superior central nervous system penetration in comparison with AG-120. A phase 1 trial of AG-881 for IDH1 or IDH2 mutant advanced solid tumors began in June 2015, and is recruiting at six locations in the United States.

November 19, 2017
AG-120, a first-in-class mutant IDH1 inhibitor in patients with recurrent or progressive IDH1 mutant glioma: updated results from the phase 1 non-enhancing glioma population.

Summary by SW who attended the presentation and took photos of the slides.

At the 2017 Society for Neuro-Oncology (SNO) conference in San Francisco, Ingo Mellinghoff of Memorial Sloan Kettering Cancer Center presented results of a phase 1 trial of AG-120 (ivosidenib), a mutant IDH1 inhibitor, for IDH1-mutant cancers. The presentation specifically focused on a subset of patients in the phase 1 trial: those with non-enhancing (no contrast enhancement on MRI images) IDH1-mutant gliomas. This analysis included 11 patients in the dose escalation phase, and an additional 24 patients from the dose expansion phase, a total of 35 patients. The primary study objective was to evaluate safety and tolerability of AG-120, and determine the maximum-tolerated dose and/or the recommended phase 2 dose.

28 out of 35 patients (80%) of patients in this analysis were treated with 500 mg of AG-120 daily. The majority (24/35, or 69%) of patients in this non-enhancing glioma cohort were WHO grade 2 gliomas. An additional 23% were WHO grade 3. Only one grade IV glioma (3%) was included.

Most patients in this analysis had been previously treated with either radiation (57%) or chemotherapy (69%) and the median number of prior systemic therapies was 2.

AG-120 was well tolerated, and the maximum tolerated dose was not reached. The majority of adverse events were low grade, and only 20% of patients experienced a grade 3 or higher adverse event.

Pharmacodynamic analysis of tumor tissue in two patients revealed that AG-120 treatment strongly suppressed 2-hydroxyglutarate levels in the tumors. 2-hydroxyglutarate is the oncometabolite produced by the mutant IDH1 enzyme.

By far the most common response to AG-120 treatment in this cohort was stable disease, which was achieved in 83% of patients. Only two patients (5.7%) achieved a minor response, including one grade 2 and one grade 3 glioma. Only four out of the 35 patients (11%) had progressive disease with neither stabilization or response.

More important is the duration of stable disease without progression. Median duration of AG-120 treatment for all 35 patients is 16 months. For the grade 2 gliomas, representing nearly 70% of the population of this study, median progression-free survival has not yet been reached, and looks to be at least 19 months at the time of the analysis (data cutoff May 12 2017).

When volumetric growth rates pre- and post AG-120 treatment were calculated by imaging studies for the 24 patients in the dose expansion group, the mean percentage change in tumor volume per six months was found to be 24% prior to treatment, and 11% after AG-120 treatment. In the 1p/19q intact (that is, the astrocytoma) subgroup of 15 patients, before and after AG-120 growth rates per six months were 38% and 14%. This confirms that the primary effect of AG-120 in this group of non-enhancing (mostly) lower grade gliomas is to significantly slow tumor progression, which was deemed to be a disease stabilization in the majority of cases.

A different drug by Agios Pharmaceuticals, called AG-881, is a dual inhibitor of mutant IDH1 and IDH2, is more brain penetrant than AG-120, and is also being studied in clinical trials for IDH mutant gliomas.

Decitabine (Dacogen) and 5-azacytidine (Vidaza)

These two drugs, structurally similar inhibitors of DNA methyltransferase, were approved (in 2006 and 2004, respectively) for treatment of myelodysplastic syndrome. They are also under investigation for epigenetic therapy in various solid cancers. A pair of research papers published on September 16, 2013 in the online journal Oncotarget, detail their promising use in IDH1 mutant gliomas. These papers are perhaps the most exciting papers yet published for IDH1 mutant specific therapy, as these drugs are already FDA approved, and showed impressive efficacy in these preclinical studies with mice [7, 8].

In the study with decitabine, IDH1 mutant cells derived from a grade III oligodendroglioma patient were pre-treated for seven days with a low concentration of the drug. The cells were then injected into the flanks of immunodeficient mice. The treated cells formed tumours which barely grew at all, and were over 90% smaller than tumours in the control group over 50 days after cell implantation. The main limitation of this study is that the IDH1 mutant cells were pre-treated with the drug ex vivo and therefore there was no demonstration of the drug’s ability to cross the blood-brain barrier and accumulate to therapeutic levels [7].

The 5-azacytidine study, from the laboratory of Gregory Riggins at Johns Hopkins University, is historic, in that it marks the first in vivo model of a patient derived IDH mutant anaplastic astrocytoma. In this study, cells were injected into the flanks of athymic mice. Five days after implantation, mice were injected intraperitoneally with 5-azacytidine on a 5 day on, 2 day off schedule. In order to continue the experiment past the point that tumours in the control group reached maximum allowable size, tumours were harvested and re-passaged in three consecutive groups of mice (Cycle 1-3), with the experiment lasting a total of 20 weeks. At seven weeks, the end of cycle 1, treated tumours were still growing, but at a slower rate than control tumours. During cycle 2 (another 7 weeks), the treated tumours had ceased to grow significantly. Finally, during cycle 3 (the remaining 6 weeks), the treated tumours began to shrink. The control tumours grew rapidly during all three cycles. In a press release describing the results of this study [9], the researchers emphasized the impressive results of this trial, saying that tumour regression is rarely observed in preclinical trials. Riggins stated that the drug “worked amazingly well” in the mice though he cautions that this drug may or may not work as well in humans. Due to these positive preclinical results, a clinical trial is expected to be initiated as soon as the regulatory system allows.

The nature of malignant progression of IDH-mutant gliomas

In what is one of the most important studies of IDH1-mutant gliomas yet published, a research team based at Yale University set out to show in great detail the full genetic and epigenetic basis for the malignant progression of lower grade IDH1-mutant gliomas to a higher grade [28].

To do this, the team first sequenced the entire exome (all protein-coding regions of the genome) of 41 IDH1-mutant grade II and III glioma samples, in addition to matched recurrences from the same patients, at the time of progression to a higher grade of malignancy. Thus, 82 samples in total were subjected to whole exome sequencing. 28 of the 41 tumors were considered as astrocytic, and 11 tumors were oligodendroglial, using molecular criteria, rather than histological analysis. As discussed below, the remaining two tumors appeared as astrocytomas at first diagnosis, but had morphed into oligodendrogliomas at the time of malignant progression. 17 of the patients underwent radiation therapy between surgeries, while only four of the patients underwent chemotherapy. Three of the recurrent tumors were considered to be hypermutated and had acquired mutation in mismatch repair genes, most notably MSH6. One of the hypermutated tumors was listed as having had no chemotherapy between surgeries, which is odd, as the hypermutation phenomenon has been typically associated with temozolomide chemotherapy (in the absence of effective mismatch repair due to MSH6 mutations or other defects, TMZ becomes a powerful mutagen).

Mutations aquired at malignant progression

Pooling together the data from both astrocytic and oligodendroglial tumors, the most frequently mutated genes newly occurring in (>10%) of the tumors at the time of malignant progression were:

  • ATM (10%), a DNA damage response gene, newly mutated in four of the progressive astrocytomas.
  • CIC (10%), commonly mutated in oligodendrogliomas, was found newly mutated in two of the progressive oligodendrogliomas, as well as two of the progressive astrocytomas.
  • MTOR (10%), a kinase more commonly associated with glioblastoma pathology, was found newly mutated in two of the progressive oligodendrogliomas and two of the progressive astrocytomas.

Histones are a component of the chromatin, and have been described as the “spools” around which strands of DNA are wound. Changes in histones influence whether genes are expressed or stay silent. Mutations in the genes for histone-altering enzymes and readers of histone marks were relatively common in the progressive gliomas, with 24% having a new mutation in a histone “eraser” (for example, a histone deacetylase, or HDAC). 20% had a new mutation in a histone “writer”, and 17% had new mutations in a histone “reader”.

Copy number

Gene copy number analysis showed that deletions and amplifications of critical cancer-associated genes were far more common than protein-altering mutations in these samples. For example, at the time of malignant progression, 61% of the progressive gliomas had deletions of the tumor suppressor gene CDKN2A, a cyclin-dependent kinase inhibitor (versus only 2% with mutations in the gene).


Further investigations looked at differential gene expression and methylation levels of gene promoters, rounding out this highly detailed analysis of the basis for malignant progression in IDH1-mutant glioma.

Pathway analysis showed that cell cycle regulation was the category most highly associated with malignant progression, with amplifications of cyclin D1 (5%), cyclin D2 (17%), cyclin-dependent kinase 4 (10%), and cyclin-dependent kinase 6 (7%); deletions of cyclin-dependent kinase inhibitor 1C (29%); deletion (61%) and mutation (2%) of cyclin-dependent kinase inhibitor 2A; deletion of cyclin-dependent kinase inhibitors 2B (54%), 1B (17%), and 2C (12%); and deletion (20%) or mutation (7%) of the retinoblastoma 1 (RB1) tumor suppressor gene.

The second most strongly associated pathway in malignant progression of IDH1-mutant glioma was found to be MYC signaling. MYC is a transcription factor responsible for many aspects of cell proliferation. At progression, gliomas had MYC amplification in 22% of cases. FBXW7 and MAX are both inhibitors of MYC signaling. There were deletions (29%) and mutation (2%) in FBXW7 and deletions (24%) in MAX in the progressive glioma samples.

Somewhat more relevant to our current arsenal of oncology drugs, receptor tyrosine kinase (RTK)-RAS-PI3K signaling was also significantly associated with IDH1-mutant glioma progression. Within this category, there was amplification of EGFR (12%), PDGFRA (12%), and KRAS (12%); amplification (10%) or mutation (2%) in MET; mutation of PIK3R1 (7%) and PIK3CA (2%), the regulatory and catalytic subunits of the PI3 kinase; deletion (34%) or mutation (2%) in the PTEN tumor suppressor gene, which counteracts PI3K signaling; mutation of MTOR (10%); and deletion (15%) or mutation (2%) in NF2, another tumor suppressor.

Bai et al. Figure 6

Figure 6c, from “Integrated genomic characterization of IDH1-mutant glioma malignant progression”, Hanwen Bai et al., 2015. Published in Nature Genetics.

Exceptional cases

As mentioned above, two of the 41 tumors studied began as molecular astrocytomas, with the matched progressive tumors from the same patients appearing as molecular oligodendrogliomas. One case was a 48-year old grade II astrocytoma patient. The initial tumor had mutations in IDH1 (R132H), a frameshift mutation in ATRX, and a missense TP53 mutation (R141C), all very typical for IDH1-mutant astrocytoma. Additionally, the initial tumor carried a missense MTOR mutation. The tumor was treated with radiation. Just over 4 years later, the woman had a second surgery for tumor recurrence. This second tumor was classed as a grade III, progressed from the previous grade II. Unexpectedly, the new tumor had lost both the TP53 and ATRX mutations, which define the astrocytoma lineage in IDH1-mutant tumors, and had gained a FUBP1 mutation, common in the oligodendroglioma lineage. We may speculate that the second tumor evolved from a very early clone, which had not yet acquired the astrocytoma-defining mutations in TP53 and ATRX, but instead followed the oligodendroglial evolutionary pathway. However, we are not told of the 1p/19q codeletion status in the published version of this study, which would confirm that the second tumor followed an evolutionary pathway towards oligodendroglioma in contrast with the original astrocytoma.

In summary, this new study takes our understanding of the genetic and epigenetic basis for malignant progression of IDH1-mutant gliomas to a new, more comprehensive level, and demonstrates the importance of loss of cell cycle regulation (eg. deletion of CDKN2A), and activation of MYC signaling, among other pathways. Drugs currently under development, though not yet approved, may in the near future be used to target these specific alterations, including cyclin-dependent kinase inhibitors, and as seen in this study, BET inhibitors.

Is the loss of mutant IDH1 function a form of therapy-resistance in recurrent tumours?

It is well-appreciated that IDH-mutant glioma patients typically have prolonged survival compared to their IDH-wild type (non-mutated) counterparts of the same WHO grade. This prolonged survival is certainly at least partially due to a decreased ability of the tumour cells to detoxify in the face of cytotoxic therapies such as radiation and chemotherapy. This begs the question: could an evolutionary loss of the mutant IDH1 gene, or a loss of its functioning, be a form of therapy-resistance in recurrent tumours which were initially positive for IDH1 mutation and high level 2-HG production?

A fascinating study [20] published by a group of researchers from Germany and the Netherlands indicates that this may be the case, in at least a minority of recurrent IDH1-mutant glioma cases. In this study, which was concerned with genetic changes from primary to recurrent tumours and genetic heterogeneity within the same tumour, paired tumour samples including both primary and recurrent tumour from 51 patients were analyzed. A majority of these were positive for IDH1 mutation at diagnosis. Three of 38 patients were positive for IDH1 mutation in the primary tumour but negative in the recurrent sample. Another five paired (primary plus recurrent) samples from initially IDH1-mutant cases were micro-dissected. One of these five cases was negative for IDH1 mutation at recurrence. In total, 4 of 43 (9%) of all the paired IDH1-mutant cases lost evidence of the mutant allele (copy) in at least that part of the recurrent tumour subjected to direct DNA sequencing.

IDH mutations in cancer are always heterozygous, meaning that they occur in only one of the two copies of the IDH1 or IDH2 gene within each cell nucleus. Several studies have revealed that high production of 2-hydroxyglutarate (2-HG) by the mutant IDH1 enzyme requires the co-operation of the IDH1 wild-type enzyme [21]. The mutant and unmutated forms of the enzyme bind together, forming heterodimers. Presumably, alpha-ketoglutarate produced by the normal IDH1 enzyme is cycled into 2-HG production by the mutant IDH1 enzyme.

In one of the earliest studies of IDH mutations in cancer, 445 central nervous system tumours (mostly gliomas) were sequenced for IDH1 and IDH2 mutations [22]. 11 of 13 secondary glioblastomas were positive for mutant IDH1. A later study [21] tells us that 2 of these 11 secondary glioblastoma cases had lost the wild-type (unmutated) allele of IDH1, impairing the ability of these tumours to produce 2-HG. Consequently, 2-HG levels were reduced by nearly 90% in these secondary glioblastomas compared to the matched anaplastic astrocytoma tumours that the two secondary glioblastomas evolved from.

To sum up thus far, these studies have told us that the mutant IDH1 allele may be undetected in about 10% of tumour recurrences from initially IDH1-mutant tumours, while 2 of 11 (18%) secondary glioblastomas had lost the wild-type allele, though other estimates of this latter phenomenon are as low as 5%.

In addition to the actual deletion of one copy of the IDH1 gene (either the mutated or non-mutated copy) in recurrent tumours, there is a phenomenon known as monoallelic gene expression (MAE), in which both copies of a gene are retained, but only one copy is actively expressed [23]. In a cohort of 67 IDH1-mutated glioma patients (42 low grade, 25 high grade), 10 of 67 (15%) exhibited monoallelic expression of the normal allele (ie loss of mutant IDH1 expression). 2 of 67 (3%) exhibited monoallelic expression of the mutant allele (loss of expression of normal IDH1). In the low grade glioma cohort, monoallelic expression of the normal allele was associated with statistically significant reduced survival (p=0.04).

Summing all of this information, to the potentially 15% (or more) of cases of IDH1-mutant glioma which lose one copy (either the mutant or wild-type copy) at post-treatment recurrence, we can add another ~18% of cases (low grade plus high grade) which retain both copies, but express only one copy (usually the wild-type/normal copy).

As explained above, the mutant and the normal IDH1 enzymes co-operate to produce high levels of 2-HG, which leads to tumour development, but which also makes these tumours more vulnerable to therapies by consuming NADPH, thus limiting the ability of these cells to neutralize reactive oxygen species. One likely explanation of the observed loss of mutant IDH1, or loss of mutant IDH1 ability to produce 2-HG (via loss of wild-type IDH1 as a co-factor), is that cells without mutant IDH1 activity, being more resistant to therapy, may emerge post-treatment in cases which were originally entirely populated with cells positive for IDH1 mutation.

A review of the effects of IDH mutations published in May 2014 is in accordance with these views [24]. These authors end on a cautionary note: “Patients should be monitored carefully in near-future clinical trials, because IDH1/2 mutant-specific inhibitors may counteract the putative survival-prolonging effects of IDH1/2 mutations and result in a decrease, rather than an increase in survival. We speculate that such inhibitors may only be used during rest periods between or after conventional ROS-inducing treatments such as IR {ionizing radiation}.”

  1. An integrated genomic analysis of human glioblastoma multiforme. Parsons et al. 2008.
  2. IDH mutation and neuroglial developmental features define clinically distinct subclasses of lower grade diffuse astrocytic glioma. Gorovets et al. 2012.
  3. Evidence for sequenced molecular evolution of IDH1 mutant glioblastoma from a distinct cell of origin. Lai et al. 2011.
  4. Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1,010 diffuse gliomas. Hartmann et al. 2009.
  5. IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype. Turcan et al. 2012.
  6. An inhibitor of mutant IDH1 delays growth and promotes differentiation of glioma cells. Rohle et al. 2013.
  7. Efficient induction of differentiation and growth inhibition in IDH1 mutant glioma cells by the DNMT inhibitor decitabine. Turcan et al. 2013.
  8. 5-azacytidine reduces methylation, promotes differentiation and induces tumor regression in a patient-derived IDH1 mutant glioma xenograft. Borodovsky et al. 2013.
  9. Johns Hopkins researchers erase human brain tumor cells in mice. September 24, 2013.
  10. Disulfiram is a DNA demethylating agent and inhibits prostate cancer cell growth. Lin et al. 2011.
  11. MGMT CpG island is invariably methylated in adult astrocytic and oligodendroglial tumors with IDH1 or IDH2 mutations. Mulholland et al. 2012.
    READ SOURCE DOCUMENT (abstract only. email me for PDF)
  12. IDH1 or IDH2 mutations predict longer survival and response to temozolomide in low-grade gliomas. Houillier et al. 2010.
    READ ABSTRACT Email me for a PDF copy
  13. No prognostic value of IDH1 mutations in a series of 100 WHO grade II astrocytomas. Ahmadi et al. 2012.
    READ ABSTRACT (Email me for a PDF copy)
  14. Molecular markers in low-grade gliomas: predictive or prognostic? Hartmann et al. 2011.
  15. IDH1 mutations in grade II astrocytomas are associated with unfavorable progression-free survival and prolonged postrecurrence survival. Thon et al. 2011.
  16. Overexpression of isocitrate dehydrogenase mutant proteins renders glioma cells more sensitive to radiation. Li et al. 2013.
  17. Increased sensitivity to radiochemotherapy in IDH1 mutant glioblastoma as demonstrated by serial quantitative MR volumetry. Tran et al. 2013.
    READ ABSTRACT (Email me for a PDF copy)
  18. IDH1 mutant malignant astrocytomas are more amenable to surgical resection and have a survival benefit associated with maximal surgical resection. Beiko et al. 2014.
    READ ABSTRACT (Email me for a PDF copy)
  19. Targetable signaling pathway mutations are associated with malignant phenotype in IDH-mutant gliomas. Wakimoto et al. 2014.
  20. Clonal Analysis in Recurrent Astrocytic, Oligoastrocytic and Oligodendroglial Tumors Implicates IDH1- Mutation as Common Tumor Initiating Event. Lass et al. 2012.
  21. Disruption of Wild-Type IDH1 Suppresses D-2-Hydroxyglutarate Production in IDH1-Mutated Gliomas. Jin et al. 2013.
  22. IDH1 and IDH2 Mutations in Gliomas. Yan et al. 2009.
  23. Monoallelic Expression Determines Oncogenic Progression and Outcome in Benign and Malignant Brain Tumors. Walker et al. 2012.
  24. The driver and passenger effects of isocitrate dehydrogenase1 and 2 mutations in oncogenesis and survival prolongation. Molenaar et al. 2014.
    READ ABSTRACT Email me for a PDF copy
  25. Malignant clinical features of anaplastic gliomas without IDH mutation. Shibahara et al. 2014.
    READ ABSTRACT Email me for a PDF copy
  26. A vaccine targeting mutant IDH1 induces antitumour immunity. Schumacher et al. 2014.
    READ ABSTRACT Email me for a PDF copy
  27. IDH mutation status is associated with a distinct hypoxia/angiogenesis transcriptome signature which is non-invasively predictable with rCBV imaging in human glioma. Kickingereder 2015.
  28. Integrated genomic characterization of IDH1-mutant glioma malignant progression. Bai et al. 2015.
    READ ABSTRACT Email me for a PDF copy.