Exploring Strategies for IDH1 Mutated Gliomas

IDH1, and the far less common IDH2 mutations, were discovered in glioma only in 2008. The first two therapeutic preclinical studies using patient-derived IDH1-mutated glioma cells were published in May and September 2013 (see A Closer Look at IDH Mutations). There have been no prospective trials for glioma stratified by IDH1 status. Only a couple prospective clinical trials have included IDH1 mutations in their final analysis. IDH1-targeted therapy in terms of clinical usage is for the most part, unexplored. My intention here is to explore potentially useful options.

PARP inhibitors

A new study [50] by researchers at Yale University took a novel approach to the treatment of IDH-mutant tumors. Instead of trying to block the activity of mutant IDH enzymes and thereby shut down the production of the 2-hydroxyglutarate metabolite, the researchers decided to find ways to exploit the weaknesses of IDH-mutant cells. In three different cell lines engineered to contain IDH1 mutations, the IDH1 mutant cell lines had greatly reduced capacity to repair DNA double strand breaks following ionizing radiation, compared to their isogenic IDH1 wild-type counterparts. The IDH1-mutant cell lines also showed evidence of increased DNA damage and double strand breaks even in the absence of irradiation, similar in magnitude to cells lacking functional BRCA2 (a gene involved in double strand break repair implicated in some breast and ovarian cancers). Further investigation showed that the increased DNA damage and double strand breaks in the IDH1-mutant cell lines was due to deficiency in homologous recombination, a mechanism critical to double strand break repair.

Similar to BRCA2-mutant ovarian and breast cancer cells, the IDH1 mutant cell lines were also very sensitive to the PARP inhibitors olaparib and rucaparib, with greatly increased cell killing compared to their IDH wild-type isogenic counterparts.

Further investigation revealed the mechanism involved in reducing double strand break repair capacity in the IDH1-mutant cell lines. The researchers found that the oncometabolite produced by mutant IDH1 enzymes, 2-hydroxyglutarate, inhibits the alpha-ketoglutarate dependent enzymes KDM4A and KDM4B, which are lysine demethylases that “play roles in the orchestration of DSB repair”. Notably, treatment of the IDH1-mutant cell lines with AGI-5198, a mutant IDH inhibitor similar to the Agios mutant IDH inhibitors currently in clinical trial, reversed the sensitivity of the cells to PARP inhibitors.

Confirming the results found in the engineered IDH1-mutant cell lines, primary patient derived IDH1-mutant samples were also found to have reduced double strand break repair and increased sensitivity to PARP inhibitor treatment, compared to IDH wild-type glioma samples. Adding 2-hydroxyglutarate to the IDH wild-type samples increased their sensitivity to PARP inhibition.

Mouse studies showed that olaparib treatment significantly inhibited the growth of subcutaneous HCT116 tumors engineered to contain the IDH1 mutation, while no such tumor inhibition was observed in HCT116 IDH wild-type tumors.

Olaparib suppresses growth of IDH-mut xenografts, but not IDH1 wild-type
Subcutaneous HCT116 IDH1 mutant xenografts in mice were strongly suppressed by olaparib treatment. The HCT116 IDH1 wild-type tumors were resistant to olaparib treatment.

This study is very significant in showing a reduced ability to repair DNA double strand breaks and increased sensitivity to approved PARP inhibitors, such as olaparib, for cells with IDH mutations. The researchers drew the inevitable comparison to BRCA-mutant ovarian cancers, which are also impaired in double strand break repair and sensitive to PARP inhibitors, reflected in the use of the term “BRCAness” for the phenotype also seen in the IDH1-mutant cells. However, no intracranial mouse model was used in this study, and the ability of approved PARP inhibitors such as olaparib and rucaparib to penetrate across the blood-tumor barrier into gliomas in sufficient therapeutic quantities will need to be determined by clinical trials. Clinical trials testing PARP inhibitors such as olaparib or the not-yet-approved PARP inhibitor veliparib for glioma should pay special attention to the IDH1 status of the participants.

As a sidenote, IDH-mutant cells also appear to have reduced baseline levels of NAD+ and are highly vulnerable to strategies targeting enzymes (such as NAMPT) involved in NAD+ salvage (see the subsection below concerning NAMPT inhibitors). As PARP is one of the enzymes that consumes NAD+, and therefore depletes NAD+ levels even further, one might hypothesize that the strategy of targeting IDH1-mutant cells with PARP inhibitors might counter the strategy of targeting IDH1-mutant cells through NAD+ depletion. Alternatively, these two strategies may co-operate with each other to severely depress proliferation of IDH-mutant cells. Combining these different strategies remains to be tested in the lab.

Oxidative Stress

There is good reason to believe that IDH1 mutant tumour cells are particularly sensitive to oxidative stress caused by free radicals (or reactive oxygen species, ROS). The unmutated IDH1 enzyme catalyzes the conversion of isocitrate to alpha-ketoglutarate. This is a redox (oxidation-reduction) reaction which converts oxidized NADP+ (nicotinamide adenine dinucleotide phosphate) to reduced NADPH. Alternatively, the mutated IDH1 enzyme catalyzes the conversion of alpha-ketoglutarate to the “oncometabolite” 2-hydroxyglutarate (or 2-HG), and in the process converts NADPH to NADP+, a reaction opposite to that which occurs with unmutated IDH1 [1].

NADPH is required by both glutathione reductase and thioredoxin reductase in the generation of reduced glutathione (GSH) and reduced thioredoxin, which form the basis of two crucial cellular antioxidant systems. In fact, reduced glutathione is considered the main antioxidant in mammalian cells [2].

One study found that IDH enzymes are the major producers of NADPH in the human brain, which is not the case in rodents [3]. Furthermore, IDH1 mutant glioblastoma samples had 38% less NADPH producing capacity in comparison with IDH unmutated samples. In a separate study, IDH-mutant glioma samples contained significantly less glutathione than IDH1-wild type samples [4].

The implication of these studies, showing less NADPH and consequently, less glutathione in IDH1-mutant tumour samples, is that these cells should have less capacity to neutralize reactive oxygen species, making them more vulnerable to therapies inducing oxidative stress, such as radiation therapy.

I hypothesize that the success of photodynamic therapy seen in the retrospective study at the Royal Melbourne Hospital is due to the increased vulnerability to oxidative stress generated by this therapy in the patients with IDH1 mutant gliomas. There is some experimental proof for this hypothesis: two studies published by Korean researchers in 2003 and 2007, prior to the discovery of IDH mutations in glioma, showed that normal IDH1 (called cytosolic isocitrate dehydrogenase in these studies) provided protection against damage from photodynamically-induced singlet oxygen [19]. In contrast, cells with reduced expression of normal IDH1 were vulnerable to this treatment. It is highly likely that mutant-IDH1 cells, with their impaired production of the reducing agent NADPH and their decreased levels of glutathione (antioxidant) would be especially sensitive to photodynamic therapy, as suggested by these early studies.

IDH1 “mutabolism”

In April 2014, two studies were published which improve our understanding of the metabolic particularities of IDH-mutant cells. Before I describe these studies, let’s review.

IDH1 is a gene which encodes the enzyme isocitrate dehydrogenase 1. The IDH1 enzyme resides outside the mitochondria, while its counterpart – IDH2 – resides within mitochondria. In addition, IDH3 (structurally unrelated to IDH1 and IDH2) resides in the mitochondria and is integral to the citric acid cyle (Krebs cycle, TCA cycle). IDH1 and IDH2 enzymes catalyze two reactions, a forward reaction and a reverse reaction:

  • the decarboxylation of isocitrate (a 6-carbon molecule) to alpha-ketoglutarate (a 5-carbon molecule)
  • the carboxylation of alpha-ketoglutarate (5-carbon) to isocitrate (6-carbon)

Mutant IDH enzymes (IDH1 or IDH2) are unable to catalyze either of these two normal reactions, but instead convert alpha-ketoglutarate into the “oncometabolite” 2-hydroxyglutarate (2-HG), leading to profound epigenetic deregulation and tumorigenesis. IDH mutations are “heterozygous”, meaning that in an IDH-mutant cell, there exists one wild-type (unmutated) copy of the gene, and one mutant copy. Consequently, even IDH-mutant cells retain some normal IDH functioning, albeit at a reduced level.

One of the major outcomes of altered IDH1 functioning in terms of cellular metabolism, is an increased requirement for alpha-ketoglutarate, due to its decreased production and increased consumption by the mutant IDH1 reaction. There are at least two major ways that an IDH1-mutant cell could compensate for this increased need for alpha-ketoglutarate:

  • import glutamine or glutamate into the cell and convert it to alpha-ketoglutarate via glutamate dehydrogenase enzymes [16].
  • import citrate or isocitrate into the cytosol (cellular fluid) from the mitochondria. Isocitrate could then be converted to alpha-ketoglutarate by the wild-type IDH1 enzyme (which coexists with the mutant copy).
  • Studies have found that alpha-ketoglutarate levels do not differ significantly in IDH-mutant versus IDH wild-type cells, due to some combination of compensatory measures to feed the 100-fold increased production of 2-hydroxyglutarate from alpha-ketoglutarate in the mutant cells.

    The particular deficiencies and requirements of mutant IDH cells suggest several distinct ways to target these cells.

    IDH1-mutant cells are highly vulnerable to NAD+ depletion via pharmacological inhibition of NAMPT

    One of the breakthrough studies showing a path to exploitation of the special vulnerabilities of IDH-mutant cells was a paper [48] published in December 2015 in Cancer Cell by a group from Massachusetts General Hospital (Tateishi, Chi, Cahill et al). In this study, the researchers discovered that application of a mutant IDH1 inhibitor to IDH1 mutant glioma cells not only decreased production of the oncometabolite 2-HG, but also perturbed the levels of NADH, citrate, and glycerol-3-phosphate by 50% or more. Further observations showed that application of the mutant IDH inhibitor increased levels of NAD+ in IDH1 mutant cell lines. Application of two different NAMPT inhbitors to endogenous IDH1 mutant cells proved highly effective in compromising cell viability, with low nanomolar potency. IDH non-mutant cell lines were completely resistant to such treatment, even at a 10 micromolar concentration. NAMPT is the rate-limiting enzyme of one of the NAD+ salvage pathways.

    NAD+ was found to be significantly lower in IDH mutant cells compared to IDH non-mutant cells at baseline. A likely explanation for this was found to be the very low levels of NAPRT in the IDH mutant cells. NAPRT is the rate-limiting enzyme of an alternative NAD+ salvage pathway. These low levels of NAPRT in IDH mutant cells could be due in part to the hypermethylation of the NAPRT promoter in the IDH mutant cells, leading to gene silencing. Thus, the baseline deactivation of one NAD+ salvage pathway (via NAPRT) in IDH mutant cells creates a special reliance on a second NAD+ salvage pathway (via NAMPT), as well as a special vulnerability to NAMPT inhibition and consequent NAD+ depletion.

    Pharmacological NAD+ depletion through targeted NAMPT inhibition was found to disrupt the mitochondrial citric acid cycle (also known as the tricarboxylic acid cycle or Krebs cycle) by limiting the activity of crucial NAD+ dependent enzymes, such as alpha-ketoglutarate dehydrogenase and malate dehydrogenase. Furthermore, NAD+ depletion led to phosphorylation and activation of the nutrient sensor complex AMPK, which in turn led to the initiation of autophagy and suppression of mTOR pathway activity. In short, NAD+ depletion through NAMPT inhibition led to an acute metabolic crisis and activation of autophagy specifically in IDH mutant cell lines, while IDH wild-type cells were completely resistant to these effects. Incidentally, a second study by this same group published in 2016 identified a similar vulnerability to NAMPT inhibitors in MYC amplified cancer cell lines [49]. In these cells, NAD+ depletion was found to limit the activity of another NAD+ dependent enzyme, GAPDH, and disrupt the overactive glycolytic metabolism of these cells.

    The most exciting part of the study came when the NAMPT inhibitor GMX-1778 was fed orally to mice bearing orthotopic IDH1-mutant gliomas. This treatment almost completely blocked NAD+ levels in the tumors at 24 hours, and prolonged survival in the mice. At a point in time when all the mice in the untreated control group had died, all the mice treated with GMX-1778 were still alive (all the treated mice died later, but the prolongation of survival was highly significant). The mice showed no signs of toxicity from the treatment.


    This study represents the most convincing preclinical experiment for an IDH1-mutant specific therapy yet seen. Unfortunately, previous human clinical trials testing the NAMPT inhibitor GMX1777 were discontinued, though there is currently a clinical trial of a novel NAMPT inhibitor called KPT-927 recruiting in the USA and Canada for advanced solid malignancies or non-Hodgkins lymphoma (NCT02702492).

    Inhibiting the mitochondrial electron transport chain

    The following strategy was put forward in an April 2014 study published in Cancer Research by a group of researchers based in Cambridge Massachusetts, and San Diego, California [18]. In a normal cell, under conditions of low oxygen (hypoxia) or mitochondrial dysfunction, the cell will compensate by increasing the reductive (versus oxidative) metabolism of glutamine-derived alpha-ketoglutarate, in the reverse reaction catalyzed by IDH enzymes. This reaction provides citrate and citrate-derived acetyl coenzyme A for manufacturing lipids and requires functional IDH activity. Experimentally, mutant IDH1 cells (but not mutant IDH2) were unable to properly adapt to hypoxia or disruptions in the mitochondrial electron transport chain, due to their limited capacity for the reductive metabolism of glutamate-derived alpha-ketoglutarate.

    IDH1-mutant cells (but not mutant IDH2) were also selectively sensitive to inhibition by electron transport chain (also called respiratory chain) complex 1 inhibitors. In the diagram, a moderate concentration (10 micromolar) of phenformin (a biguanide related to metformin) reduced the proliferative ability of an IDH1 mutant (IDH1 R132C) fibrosarcoma cell by about 50%.

    Phenformin, formerly used in type-2 diabetes, was taken off the market in the USA in 1978 due to its capacity to induce lactic acidosis. According to wikipedia, “As of 2008, phenformin was still legally available in Italy, Brazil, Uruguay, China, Poland, Greece and Portugal”. Metformin is considered the safer of the two drugs, but is also a significantly less potent inhibitor of respiratory complex 1 (probably one of the main reasons it is safer).

    Inhibiting Glutamate dehydrogenase

    The following strategy was outlined in an April 2014 study in Biochimica et Biophysica Acta by a group led by researchers from Radboud University Medical Centre in the Netherlands [16]. Glutamine pathways have been suggested as potential targets for IDH-mutant cells at least as far back as 2010, when researchers from Johns Hopkins University published a study called Inhibition of Glutaminase Preferentially Slows Growth of Glioma Cells with Mutant IDH1 [17]. However, their efforts were unable to discover clinically relevant inhibitors that would slow IDH1-mutant tumour growth in experimental models, and this group is now focusing on DNA hypomethylating agents as a more promising strategy. The recent paper [16] references some evidence that lower grade gliomas (the majority of which are IDH1-mutated) have increased expression of glutamate transporters and are addicted to sucking up glutamate from the extracellular environment. Inhibitors of glutamate dehydrogenase are therefore suggested as a potentially valid mutant-IDH therapy.

    Dependence of mutant IDH1 glioma on hominoid glutamate dehydrogenase 2 (GLUD2)

    September 17, 2014
    A study [41] just published by Genentech researchers (South San Francisco) establishes the special dependence of IDH1 (R132H) glioma cells on the glutamate dehydrogenase 2 (GLUD2) enzyme, which is only found in humans and ape species.

    Utilizing a genetically-engineered mouse model of glioma, the researchers found that viral transfection of mutant IDH1 (R132H) significantly decreased growth of these cultures compared to IDH1 wild-type controls. Growth of the mutant cultures was fully restored when transfected with wild-type IDH1.

    Hypothesizing that human IDH1-mutant gliomas upregulate expression of wild-type IDH1 to compensate for any deficits due to mutant IDH1, they compared mRNA levels of IDH1, IDH2, and IDH3 in IDH1-mutant gliomas versus histologically matched IDH1 wild-type gliomas. No differences were found. However, this analysis revealed that GLUD1 and GLUD2 mRNA was increased in the IDH1-mutant gliomas versus the wild-type samples. This finding was confirmed by examination of The Cancer Genome Atlas database.

    A commercially available IDH1-mutant human glioma cell line, BT142, derived from an oligoastrocytoma grade III patient, was grafted into mouse brains. This cell line has lost the wild-type copy of IDH1, and expresses only the mutant allele of IDH1. The mice were then treated with short hairpin RNA (shRNA) targeted to inhibit GLUD1/GLUD2. In the GLUD1/2-inhibited mice, tumour volume was significantly decreased and levels of 2-HG (the oncometabolite produced by mutant IDH), alpha-ketoglutarate (precursor to 2-HG), glucose, and citrate were all decreased compared with the control mice without inhibition of GLUD1/2.

    To distinguish between the effects of GLUD1 and GLUD2 in vitro, the genetically engineered murine glioma cells were also engineered to express either GLUD1 or GLUD2. IDH1mut/GLUD1 cells proliferated just as slowly as IDH1mut cells without GLUD1, showing that GLUD1 gave no advantage to these cells. In contrast, IDH1mut/GLUD2 cells increased proliferation to the same level as IDH1 wild-type cells, showing a significant growth advantage with the addition of GLUD2. To test the effects of glutamate on the various cell lines, 20 micromolar of glutamate was added to the cultures. This exogenous addition of glutamate increased proliferation of all cell lines, regardless of IDH1 status.

    To investigate metabolic pathways, carbon-14 labelled glutamine and glucose were added to the cell cultures. IDH1 knockout or mutant cells had reduced incorporation of these molecules into lipids. The addition of GLUD1 made no significant difference, while the addition of GLUD2 significantly increased the incorporation of labelled glutamine and glucose into lipids. Carbon-14 labelled glutamate was also incorporated into lipids, though at a low level.

    Finally, to supplement the mouse experiments using inhibition of GLUD1/GLUD2 (described above), the genetically engineered mouse model was used, combined with viral transfection of the GLUD2 gene in either IDH1 mutant or wild-type glioma-bearing mice. While GLUD2 had no effect on the survival of IDH1 wild-type glioma-bearing mice, the survival of GLUD2/IDH1 mutant mice was reduced to the same timeframe as the IDH1 wild-type mice. This shows that the inhibitory effect of mutant IDH1 on cell proliferation was abolished when accompanied by increased levels of GLUD2. In vitro studies on these same cell lines revealed that GLUD2/IDH1mut cells had increased flux of alpha-ketoglutarate into the mitochondrial citric acid cycle versus IDH1mut cells without increased GLUD2 expression.

    The authors conclude with the following observations. The glutamate dehydrogenase 2 (GLUD2) enzyme, which is highly expressed mainly in the brain [42] and only found in humans and apes, likely allows mutant IDH1 cells in the brain to compensate for metabolic deficits by allowing the increased production of alpha-ketoglutarate to supply the mitochondrial citric acid cycle as well as lipid synthesis. This hypothesis is supported by evidence from Maffuci syndrome and Ollier disease. Most patients with these conditions display an early (postzygotic) occurence of the IDH1 R132H mutation, yet the only cancer these patients are at increased risk for is glioma. This suggests that mutant IDH1 cells may require the GLUD2 enzyme (found mainly in the brain) for proliferation. The authors also note that the GLUD2 gene evolved in parallel with the expansion of the prefrontal cortex in hominids (humans and apes) and that the GLUD2 enzyme is optimized for glutamate metabolism in the brain. Glutamate is a vital neurotransmitter in the brain, and it is likely that the extracellular glutamate in the brain from neurotransmitter release aids in the proliferation of IDH1-mutant glioma.

    Summary: While exogenous glutamate promoted proliferation of glioma cell lines in vitro, regardless of IDH1 status, the GLUD2 enzyme was specifically required by IDH1 R132H mutant glioma cells for maximum proliferation, while IDH1 wild-type glioma cells had no such requirement. Though a previous study [16] suggested chloroquine as a potential therapy for IDH-mutant glioma, this drug has been shown to preferentially inhibit GLUD1, rather than GLUD2 [43]. The search for a clinically relevant inhibitor of brain-specific GLUD2 continues.

    What is the primary carbon source for 2-HG? Extracellular glutamate versus glutamine import

    In the April, 2014 study described above (van Lith et al. Reference 16), the authors ask a very pertinent question: “why tumor cells would embark on glutamine import for glutamate production when they are actually bathing in glutamate: glutamate is an important neurotransmitter and is present in synaptic clefts but also in the white matter, a location where diffuse infiltrative glioma growth is often found. There is circumstantial evidence that IDH-mutated tumor cells import glutamate from their microenvironment.” They are basing their hypothesis around the idea that IDH-mutant glioma cells, located in the glutamate-rich brain matter, have all the substrate for 2-HG production they need in the form of glutamate, and may not require increased import of extracellular glutamine from the bloodstream. The “circumstantial evidence” they refer to is the high expression of excitatory amino acid transporter 2 (EAAT2), a glutamate importer, in low-grade gliomas, but low expression in glioblastomas, which instead export glutamate into the exctracellular environment [34].

    While several in vitro studies [35] found glutamine to be the carbon source for 2-HG production in IDH-mutant cells, extracellular glutamate might be more relevant in the in vivo setting, at least for brain tumours. This hypothesis is consistent with a recent finding based on a fluoroglutamine PET scan trial for glioma patients. This first-in-human study, which reported preliminary data in March 2014, found that five low-grade glioma patients had no elevated uptake of glutamine into their tumours, while the one patient with a higher grade glioma who had elevated glucose uptake (viewed on an FDG-PET scan) also had elevated glutamine uptake on the fluoroglutamine PET scan.

    It is tempting to adjust dietary intake according to the known or apparent metabolic needs of one’s tumour. Several questions remain to be answered. Do IDH-mutant tumours indeed have elevated glutamate import via transporters such as EAAT2? If so, does dietary intake of glutamate and glutamine have any impact on brain levels of glutamate? Glutamate is an important neurotransmitter and its levels in the nervous system may be tightly controlled internally by homoeostatic mechanisms, despite varying dietary intake of glutamate and glutamine. The authors of the referenced study [16] propose a search for clinically relevant glutamate dehydrogenase inhibitors, which would not alter glutamate levels, but might inhibit the conversion of glutamate to alpha-ketoglutarate, the direct precursor to 2-HG. It remains to be determined which of several enzymes are the most relevant alpha-ketoglutarate producers in proliferating cells. Glutamate oxaloacetate transaminase (GOT)and glutamate pyruvate transaminase (GPT) are additional candidates.

    Dichloroacetate (DCA) inhibits patient-derived IDH1-mutant neurosphere proliferation

    In June of 2015, an early online access article [45] was published in the journal Cancer Research entitled IDH1 Mutation Induces Reprogramming of Pyruvate Metabolism. The bulk of the study focused on metabolic differences between two genetically engineered IDH-mutant cell lines derived from normal human astrocytes, and from the U87 glioblastoma cell line.

    More importantly, this study additionally looked at two patient-derived IDH1-mutant cell lines developed at the University of Calgary, called BT54 and BT142. BT54 was derived from an IDH1-mutated, 1p/19q codeleted anaplastic oligodendroglioma. BT142 was also derived from an IDH1-mutant anaplastic oligodendroglioma, and has since lost the wild-type (non-mutant) copy of the IDH1 gene, with a resulting decrease in 2-hydroxyglutarate production in this cell line (high output of 2-HG seems to require co-operation between both the mutant and wild-type IDH1 enzyme).

    These cell lines were exposed to a 10 mM (millimolar) concentration of DCA for 12 days (BT54) or 7 days (BT142). As expected, DCA increased pyruvate dehydrogenase activity in both cell lines. Significantly, exposure to DCA also inhibited proliferation of the IDH1-mutant neurosphere cultures, by 50% in BT54 and by 37% in BT142.

    This is one of the very first studies testing therapeutic drug response in patient-derived IDH1-mutant cell lines.

    Reversing hypersuccinylation

    November 28, 2015
    A research team at Fudan University, China, published a study [46] in the journal Molecular Cell, demonstrating a novel mechanism for the tumorigenic effects of IDH mutations. In this study, the oncometabolite 2-HG, produced in large quantities by mutant IDH enzymes, was found to competitively inhibit the activity of the citric acid cycle enzyme succinate dehydrogenase (SDH), leading to the buildup of succinate and succinyl-coenzyme A, and the hypersuccinylation of lysine. Lysine is an amino acid component of proteins. IDH1-mutated patient glioma samples were found to have significantly higher levels of succinylated lysine than IDH1 wild-type glioma samples, with the most intense succinylation being localized in mitochondria.

    The effects of this mitochondrial hypersuccinylation were then examined. Overexpression of IDH1 R132H mutation in cells led to increased succinylation at succinylation sites in mitochondrial enzymes pyruvate dehydrogenase (PDHA1), succinate dehydrogenase (SDHB) and cytochrome c oxidase. The latter two enzymes are components of the electron transfer chain. Succinylation of these enzymes led to their decreased activity, causing impaired oxidative mitochondrial metabolism. Importantly, IDH1 R132H overexpression induced the accumulation of BCL-2, an anti-apoptotic protein, in the mitochondrial membrane. Reversing hypersuccinylation by genetic manipulations also reversed the apoptosis resistance of the cells.

    The researchers then hypothesized that glycine may be able to reverse the hypersuccinylation found in 1DH1 mutant cells, as the condensation of glycine and succinyl-coenzyme A would send the resulting 5-aminolevulinic acid into the heme biosynthesis pathway, leaving fewer succinyl groups available to succinylate proteins. This was indeed found to be the case in vitro. Next, mice were subcutaneously xenografted with HT1080 fibrosarcoma cells, which harbor a naturally-occuring IDH1 R132C mutation. The mice were fed a chow supplemented with 5% glycine. Impressively, tumor weights in the glycine supplemented mice were 67% smaller than the tumors from mice fed the control diet, and succinylation levels were also lower, providing proof of principle of the glycine-supplemented diet.

    Note, however, that a glycine concentration of 100 mM was used in vitro, far higher than the ~1 mM that is achievable in serum, even with very high glycine doses. The study did not exclude that tumor growth inhibition may have been primarily due to anti-angiogenic effects, as seen in melanoma mouse studies. Additionally, glycine does not cross the blood-brain barrier effectively, as explained in a neurochemistry textbook [47].

    In contrast, polar molecules, such as glycine and catecholamines, enter the brain only slowly, thereby isolating the brain from neurotransmitters in the plasma…Small neutral amino acids, such as alanine, glycine, proline and γ-aminobutyric acid (GABA), are markedly restricted in their entry into the brain. These amino acids are synthesized by the brain, and several are putative neurotransmitters.”


    “In general, the higher the oil—water partition coefficient, the greater the brain uptake.” This figure shows the relatively slow brain uptake of glycine. From “Brain Neurochemistry: Molecular, Cellular and Medical Aspects. 6th edition”. Chapter 32

    Inhibiting the Hedgehog pathway with vitamin D3

    The Hedgehog signalling pathway is a cellular pathway which regulates cell growth during embryogenesis, and is also known to be active in various cancers, especially basal cell carcinoma (a type of skin cancer). Hedgehog pathway components include Sonic Hedgehog (ligand), Patched and Smoothened (two transmembrane receptors) and GLI transcription factors. The Patched receptor suppresses Smoothened. However, when Sonic Hedgehog binds to Patched, this negative regulation on Smoothened is relieved, allowing Smoothened to activate GLI transcription factors. Hedgehog pathway activation can lead to increased angiogenesis (via angiopoietins), cell proliferation (via cyclins) and cell survival (via anti-apoptotic genes). The Hedgehog pathway is also associated with glioma stem cells in several studies.

    In January 2013, a study was published linking Hedgehog pathway activity to IDH-mutant glioma [36]. The investigators noticed that the prevalence of IDH mutations within the various glioma grades was similar to the prevalence of Hedgehog activity within these grades, with both IDH mutations and Hedgehog activity most prevalent in lower grade gliomas and least prevalent in primary glioblastoma. More precisely, Patched transcript levels were highest in grade II gliomas, lowest in glioblastoma.

    Next, the Hedgehog (Hh) pathway was found to be active in all 12 cell cultures derived from IDH-mutated gliomas, while only a minority of IDH wild-type glioblastoma samples were postive for Hh activity. Levels of the GLI transcription factor, a downstream component of the Hh pathway, were enriched within the CD133-positive cells (CD133 is a marker of glioma stem cells) in an IDH1-mutant glioblastoma xenograft (grown in mice) tumour, and in one IDH1-mutant anaplastic astrocytoma specimen. There was no such correlation found in the two IDH wild-type glioblastoma tumours tested, which were positive for CD133, but had minimal expression of GLI. This study [36] first established a linkage between IDH-mutant gliomas and Hedgehog pathway activity, though lower grade tumours without mutant IDH could also be positive for Hh activity. Only a small minority of IDH wild-type glioblastomas were positive for Hedgehog pathway components.

    A second study [37], published in the Journal of Neuro-Oncology in May 2014, further investigated Hh signalling in IDH-mutant gliomas. Twelve IDH1 R132H (mutated) tumour samples were tested by double immunofluorescence staining for both IDH1 R132H and the Sonic Hedgehog (SHH) ligand. SHH was detected in a subset of IDH1-mutant cells in 11 of 12 specimens. Two grade II oligodendroglioma and two grade II astrocytoma specimens were scrutinized in greater detail. The percentage of cells co-expressing IDH1 R132H and SHH ranged from 0.5% to 16.1%, while the percentage of cells co-expressing IDH1 R132H and Patched (the receptor for SHH) ranged from 9% to 19.4%. SHH and Patched, representing the ligand and receptor of upstream Hedgehog signalling, tended to be found in neighboring, separate cells. In the four samples, the percentage of IDH1-mutant cells positive for SHH or Patched averaged around 20%.

    How might this information translate into clinical practice for IDH-mutated glioma patients? While one Hedgehog inhibitor (Vismodegib) has been approved by the FDA for use in basal cell carcinoma in 2012, Hh inhibitors for use in glioma is still in the early stages of testing (phase I trials of LDE225, or sonidegib/erismodegib, are active and recruiting).

    A more immediately practicable and highly appealing Hh pathway inhibitor is common vitamin D3 found in certain foods or manufactured in the body upon exposure to sunlight. Preclinical support for this function of vitamin D3 (cholecalciferol) comes from a mouse model of clear cell renal cell (kidney) carcinoma with known Hedgehog pathway activity [38].

    In this study, mice were divided into several different groups and were either injected intraperitoneally with vitamin D3 or fed high amounts of the vitamin in their diet, either before or after implantation of the renal carcinoma cells. The vitamin D3-injected mice showed a 90% reduction in GLI (downstream component of the Hh pathway), and dietary supplemented mice showed a 60% reduction. This degree of Hh pathway inhibition corresponded to a 25% reduction in tumour volume (compared to controls) in the dietary supplemented mice after 12 weeks. The mice injected with vitamin D3 harboured tumours which barely grew at all, and there was a complete tumour regression in one of these.

    Mouse dosages of drugs cannot be directly translated into human dosages based on body weight. Conversion factors based on body surface area are commonly used, though even these are rough approximations. Fortunately, the blood levels of circulating vitamin D3 metabolites were also measured in the mice at baseline and after 12 weeks of vitamin D3 supplementation. Calcidiol (25-hydroxyvitamin D3) is the vitamin D metabolite most commonly tested in humans as an indicator of vitamin D status. The baseline calcidiol levels in the mice averaged 120 nmol/l (Americans may be more familiar with vitamin D3 levels expressed as nanograms per millilitre (ng/ml). Convert units here). After 12 weeks of high-dose dietary vitamin D3 supplementation, blood calcidiol increased 2.2 fold (120×2.2= 264 nmol/l) and levels around 300 nmol/l are shown in figure 6A. Injected vitamin D3 led to a 3-fold increase to 360-400 nmol/l. Significantly, no vitamin D receptor protein was detected in tumours in the experiment, showing that the observed benefits of vitamin D supplementation in this study were independent of vitamin D receptor (VDR). Vitamin D inhibits the Hedgehog pathway by targeting Smoothened, the transmembrane receptor which activates GLI, discussed above.

    To put the above figures into perspective, one study found that human supplementation with 10,000 IU vitamin D3 per day for 2 months led to serum calcidiol (25-hydroxyvitamin D3) levels of 120 nmol/l. Published cases of vitamin D3 toxicity have involved daily doses of vitamin D3 above 40,000 IU. It must be noted that the blood levels of 25-hydroxyvitamin D3 achieved by the mice in these studies are very high and may not be safe for humans. Anyone comtemplating high-dose vitamin D3 supplementation should also have their blood monitored for signs of hypercalcemia [39].

    DNA demethylation

    The term epigenetics broadly refers to the regulation of gene expression by means other than changes in the DNA code itself. As epigenetic changes are reversible, they provide an attractive therapeutic option in the treatment of cancer. One of the major categories of epigenetics is the status of DNA methylation, especially methylation of gene promoter regions, which influences whether a gene may be actively transcribed, or instead silenced. A methyl group is simply a carbon atom with three hydrogen atoms attached. DNA methylation patterns are maintained by DNA methyltransferase enzymes, which catalyzes methylation, or the attachment of a methyl group, onto a specific location on cytosine within DNA. As explained on the Closer Look at IDH Mutations page, IDH1 or IDH2 mutations cause competitive inhibition of certain epigenetic regulatory molecules, leading to the G-CIMP tumour type, characterized by widespread silencing of various differentiation and tumour suppressive genes via methylation of the gene promoter regions. A strategy which has been used preclinically to target IDH1 mutated gliomas in mice, is to apply approved DNA methyltransferase (DNMT) inhibitors such as decitabine and 5-azacytidine to reverse cancer-promoting hypermethylation of gene promoters, restoring cell differentiation and tumour suppressing functions.

    Reviving the retinoic acid pathway

    In a study published by UCLA researchers in 2012 [13], virtually all IDH1-mutant tumours were found to have epigenetically silenced RBP1 genes, and consequent decrease in the cellular retinol binding protein (CRBP1) levels, leading the investigators to hypothesize that IDH1-mutant glioma may be especially sensitive to retinoic acid based therapeutics such as Accutane (13-cis retinoic acid). This hypothesis, first put forward in the 2012 study, is being actively investigated. The UCLA researchers are currently sequencing tumour samples from various biorepositories of over 850 glioma patients who were treated with Accutane. The plan is to sequence all these Accutane treated tumours for IDH1 mutation, and to determine a possible correlation between increased survival and Accutane use for IDH1-mutant glioma. At least one of the investigators in this study has reported that no correlation was found between use of Accutane and improved survival for IDH1-mutant glioma patients. However, as one recent trial indicated that Accutane may actually be antagonistic with temozolomide, results of studies such as this might best be stratified according to whether Accutane was given together with temozolomide, or only as a maintenance therapy after standard-of-care treatments.

    Case Study – Ben Williams and the successful treatment of an IDH mutant glioblastoma


    Timeline information compiled from Ben’s book Surviving Terminal Cancer, and from the summary of his story at virtualtrials.com.

    • March 31, 1995. At the age of 50, Ben underwent a subtotal resection of a large (180 cubic centimetre) glioma of the right parietal cortex, and was given an initial diagnosis of anaplastic astrocytoma, which was later upgraded to glioblastoma after a more thorough inspection of the resected tumour tissue. Extensive residual tumour remained post-surgery.
    • Radiation therapy consisted of the standard 55-60 Gy to the tumour area plus 2 cm beyond the tumour boundary.
    • The first MRI post-radiation showed neither shrinkage nor growth of the tumour.
    • June 1995. Two weeks prior to his first round of chemotherapy Ben began taking oral high-dose tamoxifen at a dose of 220mg daily. Tamoxifen treatment was continued until March 1998. Side effects of tamoxifen included blood clots which he treated with Aspirin and long walks.
    • July 1995. First round of intravenous BCNU (carmustine) chemotherapy combined with 600mg per day of verapamil taken during the week surrounding BCNU chemo. The verapamil was intended to block the drug extrusion pump mechanism at the blood-brain barrier, and therefore increase the penetration of BCNU past this barrier.
    • The first post-chemotherapy MRI showed a moderate degree of tumour shrinkage.
    • Between the first and second round of chemotherapy, Ben began taking oral Accutane (13-cis retinoic acid) at a dose of 160mg per day on a two week on/ one week off schedule. Accutane was not taken on the days of chemotherapy. Accutane treatment continued until December 1995.
    • Also around this time he added melatonin at 15mg per evening and the immune-boosting mushroom supplement polysaccharide Krestin (PSK) at a dose of 3 grams per day. He continues taking 10mg of melatonin to this day (2014).
    • August 1995. Second chemotherapy cycle, this time consisting of oral procarbazine, oral lomustine (CCNU) and intravenous vincristine. This regimen is known as PCV. Verapamil was again taken to improve the brain uptake of the chemotherapy during the week surrounding oral lomustine treatment.
    • Second post-chemotherapy MRI showed an “enormous” reduction in the residual tumour.
    • Third chemotherapy cycle, PCV.
    • Added oral gamma linolenic acid (GLA) at a dose of 2-2.5 gram GLA daily, consisting of 10 capsules of borage seed oil.
    • Late November. Third post-chemotherapy MRI again showed substantial shrinkage of the residual tumour.
    • Early December. Fourth cycle of chemotherapy consisting of BCNU. Ben decided to switch back to BCNU due to stomach pain caused by procarbazine and neuropathy caused by vincristine.
    • January 1996. Fourth post-chemotherapy MRI. No evidence of residual tumour, first “clean” MRI.
    • Fifth cycle of chemotherapy again consisted of BCNU, followed by another clean MRI.
    • The sixth and last cycle of chemotherapy consisted of PCV with a half-dose of vincristine to increase its tolerability. This was again followed by another clean MRI.
    • Many clean MRIs followed, though Ben continued daily high-dose tamoxifen treatment until March 1998.

    Ben Williams has contributed enormously to the brain tumour community for nearly 20 years, sharing his wisdom in the form of essays, an annual updated review of glioma treatment options, a book describing his own ordeal with glioblastoma, and very importantly, showing by example that an intelligently designed strategy utilizing repurposed drugs and non-prescription supplements in addition to conventional treatments may in fact allow long-term survival despite a grave diagnosis.

    Ben has always considered it likely that his tumour was a secondary glioblastoma (one that evolves from a lower grade tumour) and his tumour was initially classified as an anaplastic astrocytoma (grade III), later being upgraded to glioblastoma (grade IV) after a more thorough sampling following surgery.

    This status of secondary glioblastoma has recently been confirmed – his paraffin-embedded tumour sample was found positive for the IDH1 mutation (as well as MGMT promoter methylation), a clear indication of evolution from a lower grade tumour. This new information has special significance for any IDH1-mutated astrocytoma patient, as what worked in one case could potentially work in another case. This logic is confounded however, by the fact that it is impossible to attribute Ben’s long-term survival to any one item or combination of items from the menu of drugs and treatments he exposed himself to. It is nevertheless possible and tempting to perform some educated guesswork.


    Though we have some information on the median survival of IDH-mutant glioblastoma, there is currently little information on the prevalence of very long-term (15-20 year) survivorship for this patient grouping. One relevant study published online in Neuro-Oncology in December 2013 describes the variable effects of surgery for IDH-mutant and IDH wild-type anaplastic astrocytoma and glioblastoma patients [20]. This study shows a Kaplan-Meier survival curve extending out 200 months (17 years). The IDH-mutant anaplastic astrocytoma and IDH-mutant glioblastoma subgroup with complete resection of enhancing tumour or no pre-operative contrast enhancement (as seen on an MRI), but with some residual post-operative non-enhancing tumour had about 35% surviving at about 200 months (17 years) according to the Kaplan-Meier estimate in Figure 2. The majority (76%) of the IDH-mutant patients included in this study were anaplastic astrocytoma (grade III), and virtually all of these had no post-operative contrast enhancing tumour. This study also demonstrates a clear benefit of maximal surgical resection of both enhancing and non-enhancing tumour for the IDH-mutant subgroup. Though we don’t know the percentage of 15-20 year survivors of IDH-mutant glioblastoma with incomplete surgical resection, it is not likely to be high.

    Ben’s case is clearly exceptional, especially in light of the extensive residual tumour which remained post-surgery, and the absence of any tumour recurrence, implying a total or near-total eradication. While conventional treatments for glioma (ie surgery, radiation and cytotoxic chemotherapy) certainly extend lives, they are not life-savers for the majority. An evaluation of Ben Williams self-treatment protocol in light of what is currently known about IDH1-mutant glioma is therefore warranted.

    In Ben’s case, radiation therapy appeared to stabilize the tumour, while the first sign of tumour shrinkage occured following the first round of BCNU plus verapamil and daily high-dose tamoxifen.


    Tamoxifen, an estrogen receptor blocking drug commonly used as a breast cancer therapy, has also been tested in several single-arm trials for malignant glioma. None of these trials showed sufficiently convincing efficacy to warrant a larger phase III trial. As these trials were all conducted prior to the 2008 discovery of IDH mutations in glioma, it would be interesting to test the efficacy of tamoxifen in this patient subgroup.

    In a small trial [21] published in 1999, tamoxifen was tested as a monotherapy for 24 heavily pretreated anaplastic astrocytoma patients under the age of 50. Nine of these patients had frontal lobe tumours and were under age 40. Though we don’t know their IDH status, we can make an educated guess that the majority of these frontal lobe tumours were IDH-mutant. Of these nine, two (22%) had a partial tumour shrinkage (lasting 15 and 24 months), while six (67%) had disease stabilization for a median of 12 months. Only one of these nine patients had no response or disease stabilization with tamoxifen monotherapy. For these young, heavily pretreated anaplastic astrocytoma patients, single-agent tamoxifen was helpful for the majority, but was not a cure, and the longest response to tamoxifen lasted 24 months.

    The mechanism of action of high-dose tamoxifen in glioma is purported to be protein kinase C inhibition, though tamoxifen is also cited in the literature as an inhibitor of the mitochondrial electron transport chain. Recent evidence [18] indicates that IDH1-mutant cells are particularly sensitive to electron transport chain complex 1 inhibition. Tamoxifen should be tested, at the very least in vitro, as an inhibitor of mutant IDH1 glioma cells.

    Accutane (13-cis retinoic acid)

    Ben included Accutane in his self-treatment regimen for approximately half a year, from July to December 1995, until just before his first clean MRI. He refrained from Accutane during the days of chemotherapy, as he believed it could interfere.

    One recent phase II trial [22] tested combined Accutane and the standard of care drug temozolomide (TMZ) prior to radiation therapy in a population of anaplastic glioma patients. This combination regimen was not found to be a significant improvement when compared with the TMZ arm of the phase III NOA-04 trial. IDH mutational status was not included in the survival analysis of the phase II trial.

    As noted elsewhere, researchers at UCLA hypothesized that since IDH-mutant glioma tends to have a silencing of the retinol binding protein 1 gene, these patients may show a particular benefit to therapy with retinoids. These researchers are now sequencing hundreds of tumour samples for the IDH mutations, and comparing outcomes of those with the mutation who were exposed to Accutane versus those who weren’t. Preliminary data showed a longer survival in the Accutane group, though the final analysis hasn’t been published to date.

    Melatonin and PSK

    Long after Ben had acheived a clean MRI and had ceased taking repurposed medications such as tamoxifen and Accutane, he continued daily intake of supplements such as melatonin and the immune-boosting mushroom-derived medication PSK. He took 15mg of melatonin every evening for years and to this day continues taking 10mg every night. He also replaced PSK with Maitake D-fraction, and later reishi, on a regimen of once every 3 days. These immune-modulating agents could have contributed to the eradication of any remaining tumour cells following his prior treatments. Melatonin also has a documented cytostatic effect on cancer growth.

    Additionally, Ben was taking multiple nutritional and nutraceutical supplements during and after his conventional treatments, including gamma linolenic acid (from borage seed oil), genistein (from soy), flax seed oil, selenium, and green tea extract. His supplementation was based on published research showing efficacy against cancer of various types. As little testing of these agents has been performed in animal models of glioma, or in patients, there is little way of knowing which of these compounds or combinations synergized with the other treatments he underwent. Gamma linolenic acid was tested in glioma patients and was found to have some effect, though the mode of administration was intratumoral injection, rather than oral intake [23]. It is possible that oral gamma linolenic acid might be taken up by brain tumor cells leading to their apoptosis (cell death) by free radical and lipid peroxide damage, as occurs in glioma cells in vitro. Any of these supplements could have had additional degrees of activity, and it is also possible that the outcome of long-term survival for Ben could have been the result of all of the conventional treatments, drugs, and nutritional supplements combined, rather than any one or few of them particularly. With regard to supplementation, I would encourage patients to emulate the spirit of Ben’s approach – experimentation based on the best, most current available evidence, rather than a precise mimicking of his protocol. On the other hand, some of the drugs in his self-treatment regimen, such as tamoxifen, have had an observed anti-tumour efficacy in glioma clinical trials, and this might warrant a more carefully exact approach.

    A case report of Ben’s story from a more medical perspective is in preparation for publication, and this may bring us closer to understanding his exceptional case. As understanding of IDH-mutant biology and glioma biology in general advances in the years to come, more answers will undoubtedly be uncovered.

    1. Oncogenic isocitrate dehydrogenase mutations: mechanisms, models, and clinical opportunities. Cairns R, and Mak T. 2013.
    2. IDH1 gene mutations: a new paradigm in glioma prognosis and therapy? Labussiere et al. 2010.
    3. The prognostic IDH1(R132) mutation is associated with reduced NADP+-dependent IDH activity in glioblastoma. Bleeker et al. 2010.
    4. Non-invasive detection of 2-hydroxyglutarate and other metabolites in IDH1 mutant glioma patients using magnetic resonance spectroscopy. Pope et al. 2012.
    5. Cytotoxic effect of disulfiram/copper on human glioblastoma cell lines and ALDH-positive cancer-stem-like cells. Liu et al. 2012.
    6. A conceptually new treatment approach for relapsed glioblastoma: Coordinated undermining of survival paths with nine repurposed drugs (CUSP9) by the International Initiative for Accelerated Improvement of Glioblastoma Care. Kast et al. 2013.
    7. Selective enhancement of cellular oxidative stress by chloroquine: implications for the treatment of glioblastoma multiforme. Toler et al. 2006.
    8. Disulfiram is a DNA demethylating agent and inhibits prostate cancer cell growth. Lin et al. 2011.
    9. How could a drug used to treat alcoholism also be effective against glioblastoma? Wang et al. 2013.
    10. Molecular determinants of retinoic acid sensitivity in pancreatic cancer. Gupta et al. 2012.
    11. Differential retinoic acid signaling in tumors of long- and short-term glioblastoma survivors. Barbus et al. 2011.
    12. Aberrant expression of retinoic acid signaling molecules influences patient survival in astrocytic gliomas. Campos et al. 2011.
    13. Identification of retinol binding protein 1 promoter hypermethylation in isocitrate dehydrogenase 1 and 2 mutant gliomas. Chou et al. 2012.
    14. Epigenetically mediated downregulation of the differentiation-promoting chaperon protein CRABP2 in astrocytic gliomas. Campos et al. 2012.
    15. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Noushmehr et al. 2010.
      READ SOURCE DOCUMENT (see Table 2)
    16. Glutamate as chemotactic fuel for diffuse glioma cells; are they glutamate suckers? van Lith et al. 2014.
      READ ABSTRACT Email me for a PDF copy
    17. Inhibition of Glutaminase Preferentially Slows Growth of Glioma Cells with Mutant IDH1. Seltzer et al. 2010.
    18. IDH1 mutations alter citric acid cycle metabolism and increase dependence on oxidative mitochondrial metabolism. Grassian et al. 2014.
      READ ABSTRACT Email me for a PDF copy.
    19. Regulation of singlet oxygen-induced apoptosis by cytosolic NADP+-dependent isocitrate dehydrogenase. Kim et al. 2007.
      READ ABSTRACT Email me for PDF
    20. 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
    21. Salvage chemotherapy with tamoxifen for recurrent anaplastic astrocytomas. Chamberlain et al. 1999.
    22. Temozolomide and 13-cis retinoic acid in patients with anaplastic gliomas: a prospective single-arm monocentric phase-II study (RNOP-05). Grauer et al. 2011.
      READ ABSTRACT Email me for a PDF copy
    23. Gamma-linolenic acid therapy of human glioma-a review of in vitro, in vivo, and clinical studies. Das et al. 2007.
      READ ABSTRACT Email me for a PDF copy
    24. cMYC expression in infiltrating gliomas: associations with IDH1 mutations, clinicopathologic features and outcome. Odia et al. 2013.
      READ ABSTRACT Email me for a PDF copy
    25. Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction. Wise et al. 2008.
    26. Phenylbutyrate-induced glutamine depletion in humans: effect on leucine metabolism. Darmaun et al. 1998.
    27. Phenylacetic acid as a potential therapeutic agent for the treatment of human cancer. Neish WJ, 1971.
    28. Inhibitory effects of phenylbutyrate on the proliferation, morphology, migration and invasiveness of malignant glioma cells. Engelhard et al. 1998.
      READ ABSTRACT Email me for a PDF copy
    29. Phase II Study of Phenylacetate in Patients With Recurrent Malignant Glioma: A North American Brain Tumor Consortium Report. Chang et al. 1999.
    30. Oral sodium phenylbutyrate in patients with recurrent malignant gliomas: A dose escalation and pharmacologic study. Phuphanich et al. 2005.
    31. Complete response of a recurrent, multicentric malignant glioma in a patient treated with phenylbutyrate. Baker et al. 2002.
      READ ABSTRACT Email me for a PDF copy
    32. What Dr. Stanislaw Burzynski doesn’t want you to know about antineoplastons. By Dr. David Gorski (Orac), December 12, 2011.
    33. Cotreatment with dichloroacetate and omeprazole exhibits a synergistic antiproliferative effect on malignant tumors. Ishiguro et al. 2012.
    34. The excitatory amino acid transporter-2 induces apoptosis and decreases glioma growth in vitro and in vivo. de Groot et al. 2005.
    35. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Dang et al. 2009.
    36. Identification of Hedgehog pathway responsive glioblastomas by isocitrate dehydrogenase mutation. Valadez et al. 2013.
      READ ABSTRACT Email me for a PDF copy
    37. Expression of Hedgehog ligand and signal transduction components in mutually distinct isocitrate dehydrogenase mutant glioma cells supports a role for paracrine signaling. Abiria et al. 2014.
      READ ABSTRACT Email me for a PDF copy
    38. Vitamin D3 triggers antitumor activity through targeting hedgehog signaling in human renal cell carcinoma. Dormoy et al. 2012.
    39. Vitamin D supplementation, 25-hydroxyvitamin D concentrations, and safety. Vieth, 1999.
    40. Targetable Signaling Pathway Mutations Are Associated with Malignant Phenotype in IDH-Mutant Gliomas. Wakimoto et al. 2014.
      READ ABSTRACT Email me for a PDF copy
    41. Hominoid-specific enzyme GLUD2 promotes growth of IDH1 R132H glioma. Chen et al. 2014.
    42. The human GLUD2 glutamate dehydrogenase: localization and functional aspects. Zaganas et al. 2009.
    43. Inhibitory Properties of Nerve-Specific Human Glutamate Dehydrogenase Isozyme by Chloroquine. Choi et al. 2007.
    44. Comparison of the anti‑cancer effect of Disulfiram and 5‑Aza‑CdR on pancreatic cancer cell line PANC‑1. Dastjerdi et al. 2014.
    45. IDH1 mutation induces reprogramming of pyruvate metabolism. Izquierdo-Garcia et al. 2015.
      READ ABSTRACT Email me for a PDF copy.
    46. NADP(+)-IDH Mutations Promote Hypersuccinylation that Impairs Mitochondria Respiration and Induces Apoptosis Resistance. Li et al. 2015.
    47. Basic Neurochemistry: Molecular, Cellular and Medical Aspects. 6th edition. Chapter 32, Blood-Brain Barrier. Laterra et al. 1999.
    48. Extreme Vulnerability of IDH1 Mutant Cancers to NAD+ Depletion. Tateishi et al. 2015.
      READ ABSTRACT Email me for a PDF copy
    49. Myc-Driven Glycolysis Is a Therapeutic Target in Glioblastoma. Tateishi et al. 2016.
      READ ABSTRACT Email me for a PDF copy
    50. 2-Hydroxyglutarate produced by neomorphic IDH mutations suppresses homologous recombination and induces PARP inhibitor sensitivity. Sulkowski et al. 2017.