The excitatory milieu in glioma: Mechanisms and therapeutic avenues

Abstract

Gliomas are the most common primary malignant brain tumors. Their electrobiologic properties drive disease development, and in select tumors, aberrant neurosignaling is situated at the crux of gliomagenesis and glioma-related epilepsy. Tumor microtubes and the neuronal-glioma synapse are defined components of the glioma circuitry. The nidus of cortical hyperexcitability—the peri-glioma—undergoes severe alterations during disease progression and is influenced by genetic mutations, anomalous synaptic remodeling, inflammatory changes, and an imbalance in neurotransmitters. Such pathologic mechanisms have been exploited for anticancer and anti-seizure value wherein a subset remains to be explored. In this Review, we discuss the hyperexcitable conditions within the glioma microenvironment and candidate therapies for seizure and tumor control.

Key Points.

• Peri-glioma neuronal hyperexcitability and the progression of malignant gliaparticipate in a positive feedback loop.

• Upstream genetic mutations induce aberrant synaptic remodeling in the gliomamicroenvironment to facilitate excitatory transmission.

• Therapies targeted against the excitatory glioma network may simultaneously treat seizures and glioma progression.

Gliomas are the most common primary malignant tumors of the central nervous system (CNS).¹ The electrical integration of gliomas plays a fundamental role in their oncogenesis and epileptogenesis. Weblike membrane protrusions, tumor microtubes (TMs), interconnect individual glioma cells into a functional syncytium, whereas the neuronal-glioma synapse promotes bidirectional communication between neurons and glioma cells. Synaptic input to gliomas extends beyond the tumor microenvironment (TME) with neuronal signaling originating from as far as the contralateral hemisphere.^2,3 To date, known routes of direct neuron-to-glioma neurotransmission occur via glutamatergic^4,5 and cholinergic^2,6 pathways. A subset of cells comprising gliomas exhibit spiking behavior reminiscent of GABAergic neurons.⁷ The glioma circuitry is heavily dependent on mutations remodeling the synaptic constituency toward an excitatory milieu. It is possible there are epigenetic drivers for this process as well.⁸ In consequence, an imbalance ensues with excitatory glutamatergic signals predominating over GABAergic inhibition. Tumor-associated microglia and macrophages (TAMs), known to infiltrate glioma-lesioned brain in abundance, may have underappreciated nonimmune functions in the context of peri-glioma excitability. Due to rapid tumor growth in higher-grade neoplasms, metabolic and mechanical insults can further exacerbate excitatory signaling. These multimodal pathogenic processes can culminate to evoke electrical outbursts, resulting in glioma-related epilepsy (GRE). The incidence of GRE depends on many variables: lower-grade lesions have the highest incidence with tumor location and molecular signature comprising additional risk factors.

Beyond eliciting GRE, enhanced electrical activities of peri-glioma neurons propel mechanisms of tumor progression. Potentiating peri-glioma neuronal excitability modulates the paracrinology of the TME-releasing pro-oncogenic molecules such as brain-derived neurotropic factor (BDNF) and neuroligin-3. The high density of peri-glioma excitatory synapses and alterations in membrane transporters lead to the buildup of glutamate in the TME. In turn, glutamate may drive malignancy by binding both ionotropic and metabotropic receptors to trigger growth-signaling cascades. The impact of γ-aminobutyric acid (GABA) on tumor growth is variable as expression of GABA receptors on the glioma cell surface decreases with higher WHO grades.

In this Review, the role of neuronal-glioma synaptic and paracrine communication and glioma cell-glioma cell microtube-facilitated communication will be discussed. We will incorporate an overview of intracellular signaling cascades, many familiar to our understanding of gliomagenesis and central to GRE. Next, how peri-glioma hyperexcitability can serve as a potential target for managing both GRE and glioma progression will be covered. Finally, future directions for therapeutic investigations at the junction of GRE and gliomagenesis will be highlighted.

Neuronal-Glioma Communication

Neurons and Gliomas Have a Bidirectional Relationship

A description of the structure and function of synapses between malignant glia and neurons has introduced a new field of oncologic mechanisms in gliomas. Bidirectional electrochemical interaction between these entities is proposed to fuel a positive feedback loop of glioma progression and neuronal excitability (ie, a pro-seizure state). Direct integration of glioma into the neurocircuitry is mediated by synapses between presynaptic neurons and postsynaptic glioma cells via α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)^4,5 and muscarinic^2,6 receptors. Several factors promote this synaptogenesis and foster a hyperactive state including NLGN3, BDNF, thrombospondins (TSPs), glypican 3 (GPC3), and semaphorin 4F. In addition to the glioma cell surface, glutamatergic neuronal-glioma synapses are positioned on TMs creating a network between tumor cells via gap junctions, as well as a method for neurons to communicate with tumor cells.⁹ The depolarization of glioma cells is furthermore thought to be mediated by nonsynaptic K⁺ currents generated by hyperactive peri-glioma neurons.¹⁰

How Is the Neuronal-Glioma Synapse Created and Established?

Neuronal activity results in the release of at least 2 paracrine substances: NLGN3 and BDNF. While they act via distinct mechanisms, their release leads to creation and strengthening of neuronal-glioma synapses in addition to glioma proliferation. NLGN3 promotes expression of genes encoding AMPA receptor (AMPAR) subunits as well as tropomyosin-related kinase B (TrkB) while BDNF-TrkB signaling promotes the movement of translated AMPAR subunits to the postsynaptic membrane. Both thereby promote malignant neuronal plasticity and connectivity.¹¹

Paracrine Signaling Molecules: NLGN3.—

NLGN3 is a postsynaptic cell adhesion molecule located at excitatory synapses. Neuronal activity stimulates release of NLGN3 which has been established as a key player in glioma proliferation and overall survival.¹² NLGN3 secretion stimulates the phosphoinositide 3-kinase/mammalian target of rapamycin (PI3K/mTOR) pathway in glioma cells via a feedforward mechanism. NLGN3 cleavage into the extracellular matrix (ECM) is mediated by ADAM protease.¹² Inhibiting ADAM protease has prevented the release of NLGN3 and curbed tumor growth in pediatric glioma-xenografted mice.¹³ However, inhibition of ADAM10 in mice with nontumoral temporal lobe epilepsy has been shown to increase neuroinflammation and induce seizures.¹⁴ It remains uncertain whether targeting the ADAM10/NLGN3 pathway could offer dual therapeutic benefits for both seizures and tumors. An ADAM10/17 inhibitor is currently being investigated for pediatric patients with high-grade gliomas (HGGs) and results are pending (NTC04295759).

The use of mTOR inhibitors has been promising as they may modulate the signaling pathways that regulate NLGN3 expression and function. These medications are further discussed below.

Paracrine Signaling Molecules: BDNF.—

BDNF, similar to NLGN3, is a paracrine molecule released in response to neuronal activity. It is a growth factor crucial in facilitating neuronal growth during early life and promoting synaptic transmission/neuronal plasticity in adulthood.¹⁵ Endogenous BDNF signaling via its corresponding TrkB receptor is thought to promote epileptogenesis in normal brain.¹⁶ BDNF is associated with glioma proliferation but to a lesser degree than NLGN3.¹² In pediatric glioma-xenografted mice, BDNF signaling increased neuronal-glioma synapses and AMPARs on the glioma membrane.¹⁷ These network alterations were accompanied by enhanced glutamate signaling and tumor progression. Antagonizing BDNF activity has shown to block glioma proliferation and have a positive impact on survival in preclinical studies.¹⁷ It is important to note BDNF expression may be modulated by 5-HT_2A agonists.¹⁸ While inhibition of BDNF/TrkB signaling may suppress gliomagenesis and epileptogenesis, no trials have aimed to target this mechanism.

TSPs Assemble the Glioma Circuitry.—

TSPs are synaptogenic proteins released by nonmalignant astrocytes.¹⁹ The formation of neuronal-glioma synapses and epilepsy-related network bursts correspond to TSP1 expression.²⁰ TSP2 levels in glioma tissue have furthermore been correlated with the incidence of epileptiform discharges.²¹ The ɑ2δ1/Rac1 pathway is believed to mediate TSP-induced synaptogenesis.²¹

Targeting α2δ1/Rac1 Pathway: Gabapentin.

Gabapentin, an ɑ2δ antagonist, has mitigated excitatory synaptogenesis, peritumoral hyperexcitability, and proliferation of glioma preclinically.²⁰ Proliferation has only been restricted in the context of glioma cells in co-culture with neurons. The neuronal ɑ2δ1 receptor seems to be a critical factor involved in tumor growth, and such data provide additional support in targeting peri-glioma neurons to control malignancy. In a small cohort, gabapentin as an add-on agent has improved refractory GRE.²² Additionally, extended survival as well as low serum TSP1 concentrations in glioma patients are observed with gabapentin treatment in a retrospective dataset.²³ Currently, gabapentin in combination with sulfasalazine, memantine, and chemoradiotherapy is undergoing evaluation for survival in glioma.²⁴

Kir4.1 Channelopathy Drives Glioma Growth and Excitability

Ion dysregulation creates a hyperexcitable peritumoral microenvironment. Under normal circumstances, an astrocytic K⁺ channel, inward-rectifying potassium channel (Kir4.1), maintains K⁺ and glutamate homeostasis to prevent excessive neuronal firing.²⁵ Reduction of Kir4.1 drives epileptogenesis as fewer Kir4.1 channels lead to higher levels of extracellular K⁺ and glutamate at the synaptic cleft. Human glioma-to-mouse functional genomic analysis identified Kir4.1 to co-localize with the membrane protein immunoglobulin superfamily member 3 (IGSF3).²⁶ IGSF3 interrupted Kir4.1 buffering capacity and had severe ramifications in glioma-xenografted mice resulting in epileptiform discharges and tumor progression. Low levels of Kir4.1 have also been implicated to evoke GRE in human glioblastoma tissue.²⁷

Potassium dysregulation is an uncommon therapeutic target for anti-seizure medications (ASMs). There are only 2 medications’ primary mechanism being related to potassium channels (ie, ezogabine, retigabine) which have not been studied in GRE.

TMs: Functional Syncytium

In the background of neuronal-glioma synapses is a highly integrated network of glioma cells connected via TMs which create a functional syncytium. The TM network is implicated in glioma invasion, proliferation, and resistance to chemoradiation.²⁸

Targeting Gap Junction Formation: Meclofenamate

Meclofenamate is a fenamate nonsteroidal anti-inflammatory drug. Multiple biologic functions have been attributed to fenamates, one of which is blocking the formation of gap junctions.²⁹ In xenografted gliomas, meclofenamate significantly reduced cell proliferation.^4,5 It has also enhanced temozolomide (TMZ)-induced cell death.³⁰ Preclinical models suggest gap junction blockers such as meclofenamate have anticonvulsant effects.³¹ Currently, meclofenamate-TMZ dual therapy is undergoing evaluation in a trial for progressive MGMT promoter-methylated glioblastoma.³²

Signaling Cascades

PIK3CA/mTOR Signaling Cascade

In addition to upstream modulators promoting synaptogenesis and hyperexcitability, dysregulation of PI3K/Akt/mTOR signaling contributes to similar pathophysiological outcomes. PI3Ks are heterodimeric lipid kinases that control metabolism, proliferation, and survival. The PI3K/Akt/mTOR pathway is often mutated in glioblastoma, and dysfunction in PIK3CA, the catalytic subunit of PI3K-α, is tumorigenic.³³ PIK3CA mutations are correlated with heightened cortical excitability and refractory GRE.^34,35 In a preclinical model, PIK3CA mutations enhanced gliomagenesis and peri-glioma synaptogenesis with different genetic variants evoking hyperexcitability seemingly via discrete pathomechanisms.³³ One member of the glypican family, GPC3, may promote peri-glioma hyperexcitability by remodeling synaptic circuits toward excitatory, glutamatergic transmission.

Mutations of genes in PI3K/Akt pathway have been associated with epileptogenic brain malformation including focal cortical dysplasia and hemimegalencephaly.³⁶ While targeted PI3K inhibitors (eg, alpelisib, duvelisib, umbralisib) have been commercially available and Akt inhibitors (eg, capvisertib, ipatasertib) have been investigated in clinical trials for oncologic purposes, data for humans with lesional epilepsy are limited.³⁷ Alpelisib is currently being investigated in trials for PIK3CA-related overgrowth spectrum which may include seizures in the clinical spectrum.

In the PI3K signaling cascade, mTOR functions dually as an upstream repressor/downstream effector.³⁸ Data support the role of mTOR signaling in glioblastoma invasion and poor prognosis.³⁹ Neuronal morphology, plasticity, and signaling are influenced by the mTOR pathway.⁴⁰ As such, mutations in the mTOR cascade drive epileptogenic conditions in gliomas.⁴¹ In patients with preoperative GRE, amplified peri-glioma electrophysiologic readings were accompanied by mTOR pathway activation.⁴² As discussed earlier, the PI3K/mTOR pathway is an intermediary in the feedforward production of NLGN3.¹²

Targeting mTOR Signaling Cascade: mTOR Inhibitors

Temsirolimus was the first mTOR inhibitor tested in a glioma trial twenty years ago.⁴³ Since then, mTOR inhibitors have been evaluated in various combinatorial therapies with limited success.^44,45 Metabolic shifts increasing levels of glutamate after mTOR inhibitor therapy have been attributed to glioblastoma resistance.⁴⁶ Nonetheless, blocking mTOR signaling is effective in mitigating seizures. As an integral component in management of tuberous sclerosis complex (TSC), mTOR inhibitors address disease features of drug-resistant epilepsy and growth of subependymal giant-cell astrocytoma.⁴⁷

Inhibiting mTOR activity may also have potent modulation on the hyperexcitable glioma network. mTOR blockade has shown to restore GABAergic interneuron signaling⁴⁸ and reverse pathologic remodeling of peri-glioma neurons.⁴⁹ Sirolimus has furthermore decreased D-2-hydroxyglutarate (D2HG) production and reduced bursting activity in an isocitrate dehydrogenase (IDH)-mutant glioma cell model.⁴¹ In refractory GRE, the peri-glioma is fraught with discordant neurotransmission which impedes the therapeutic effects of commonly prescribed ASMs. Under these circumstances, incorporating mTOR inhibitors may revert neuronal signaling back to near-physiologic conditions and enhance ASM efficacy.

Glutamate Signaling

Glutamate Signaling in Epileptogenesis and Oncogenesis

Excitatory-inhibitory neurotransmitter imbalance is a major pathophysiologic tenet of nonlesional epilepsy. Glutamate is crucial in glioma biology given its dual role in epileptogenesis and oncogenesis (Figure 1).⁵⁰ The increased concentration of excitatory synapses at the peri-glioma accompanied by pathologic changes of glutamate cell surface transporters leads to glutamate accumulation in the ECM and sets the stage for glutamate excitotoxicity. After prolonged exposure to glutamate, peri-glioma neurons either undergo neurotoxic death or if they survive, electrical bursting.

Figure 1.

Peri-glioma cortex. Glutamate accumulates to excessive concentrations in the tumor milieu. EAAT2 downregulation results in decreased clearance of glutamate from the ECM, whereas system xc- upregulation extrudes glutamate from glioma cells. Decreased expression of NKCC and increased expression of KCC2 contribute to paradoxical GABA excitation of peri-glioma neurons. The neuronal-glioma synapse signals glioma growth, invasion, and calcium transients throughout tumor microtubes. Potassium uptake derived from peritumoral neuronal hyperexcitability may trigger calcium transients throughout tumor microtubes.

To maintain low extracellular glutamate, proper functioning levels of excitatory amino acid transporter-1 and -2 (EAAT1, EAAT2) on astrocytes are necessary.⁵¹ Peri-glioma astrocytes have reduced glutamate uptake, and a subset has displayed a depolarized resting membrane potential.⁵² Dysfunction of EAAT transporters and/or limited astrocytic arborization in glioma may cause an imbalance of the glutamate pool. Increased expression of the cystine-glutamate antiporter, system xc-, also contributes to the accumulation of glutamate.^53,54 System xc- may elicit neuronal excitability via unrestricted extrusion of intra-glioma glutamate to the TME.⁵³ xCT, the functional subunit of system xc-, is furthermore upregulated in GRE.⁵⁵

Targeting System xc- as a Therapeutic Target: Sulfasalazine.—

Sulfasalazine, an aminosalicylate, gained attention as a treatment for glioma because of its ability to block system xc-. Seizures and tumor growth are reduced in xenografted glioma models with sulfasalazine.⁵⁶ The first trial of sulfasalazine was terminated for futility at interim analysis. In a subsequent trial, sulfasalazine combined with chemoradiotherapy for newly diagnosed glioblastoma provided an element of seizure control without a correspondent survival benefit.⁵⁷

Glutamate Pathway as a Therapeutic Target: AMPAR Antagonists.—

Abrogating communication between glioma and adjacent neurons is an attractive therapeutic approach to stunt glioma progression. Two antagonists of the ionotropic glutamate AMPAR—talampanel and perampanel—are candidates to block neuronal-glioma synaptic connectivity.

As an adjunct, perampanel has been effective in controlling refractory GRE.⁵⁸ Glioblastoma proliferation and extracellular glutamate are both reduced in vitro after perampanel administration.⁵⁹ It disrupts neuronal activity-mediated glioma cell proliferation by blocking the glutamatergic neuronal-glioma synapse.^2,4,5 In contrast, perampanel combined with radiotherapy has not shown to impede tumor growth.⁶⁰ A phase 1/2 trial evaluated perampanel in terms of its effect on survival and peri-glioma hyperexcitability in HGG.⁶¹ The study was terminated early for futility with potential confounding variables.

Talampanel has been tested in both newly diagnosed⁶² and recurrent⁶³ glioblastoma. It may confer a survival benefit in combination with radiotherapy but was not beneficial as a sole agent. In addition, the short half-life of talampanel makes it less than ideal for GRE compared to perampanel.⁶⁴

Glutamate Signaling as a Therapeutic Target: NMDAR Antagonists.—

Activation of ionotropic glutamate N-methyl-D-aspartate receptor (NMDAR) has been implicated in the invasion, survival, and epileptogenicity of gliomas.⁶⁵

Riluzole, typically used in amyotrophic lateral sclerosis (ALS), has increased glioma sensitivity to mTOR inhibitors⁶⁶ and radiotherapy.⁶⁷ In astrocytes, it has enhanced expression and function of EAAT2.⁶⁸ Similarly, riluzole may alleviate elevated glutamate levels in the TME. Glioblastoma cell DNA repair mechanisms seem to depend on NMDAR activity.⁶⁹ Riluzole via NMDAR blockade increased cytotoxicity and DNA damage biomarkers in mice with xenografted glioma, somewhat at odds with its neuronal cell-preserving function in ALS.⁷⁰

The therapeutic potential of riluzole is restricted by its poor bioavailability, extensive cytochrome P450 metabolism, and food interactions.⁷¹ A tripeptide prodrug of riluzole, troriluzole, was developed to circumvent these limitations. In a preclinical model, troriluzole reduced peri-glioma glutamate levels and altered microenvironmental T-cell recruitment, improving survival.⁷² This survival effect becomes more pronounced in conjunction with immunotherapies⁷² which is currently a combinatorial regimen being investigated for melanoma brain metastasis (NCT04899921). A clinical trial has also been set forth to investigate troriluzole for recurrent glioblastoma. The study plans to assess its impact on electrocorticographic activity as a measure of peritumoral hyperexcitability (NCT06552260).

Memantine has demonstrated clinical neuroprotective activity in studies of patients with brain metastases receiving whole-brain radiation therapy.⁷³ Similar to riluzole, it has disrupted DNA repair pathways and in consequence, glioma cells become sensitized to radiotherapy.⁶⁹ The anti-glioma activities of memantine have inspired the synthesis of memantine derivatives to enhance its tumoricidal efficacy by chemically linking it to various histone deacetylase inhibitors.⁷⁴

Glutamate Signaling as a Therapeutic Target: mGluR Antagonists/Agonists.—

In contrast to ionotropic AMPAR/NMDAR channels, metabotropic glutamate receptors (mGluRs) are G-protein-coupled and modulate downstream cellular messengers. The 8 mGluRs are divided into 3 groups predicated on their amino acid sequence, G-protein cascade, and pharmacology.⁷⁵ These include group 1 (mGluR1, 5), group 2 (mGluR2, 3), and group 3 (mGluR4, 6, 7, 8). The mRNA expression of all mGluRs except for mGluR6 is upregulated at the leading edge of invasion in glioblastoma.⁷⁶

Group 1⁷⁷ and 2⁷⁸ mGluRs are implicated in driving glioma proliferation. mGluR1, expressed in several glioma cell lines, is noncanonically connected to PI3K/Akt/mTOR activation.⁷⁷ Antagonism of mGluR1 diminished PI3K/Akt/mTOR pathway molecules from their active forms and suppressed glioma growth.⁷⁷ mGluR3 activity has maintained an undifferentiated/pro-proliferative state in neural⁷⁹ and glioma⁸⁰ stem cells while its antagonism chemosensitized glioblastoma to TMZ.⁸¹ In contrast, agonizing group 3 mGluRs may suppress glioma cell proliferation.^82,83 Furthermore, expression profile of glutamate transporters is regulated by mGluRs. In human astrocytes, mGluR3 increased expression of EAAT1 and EAAT2 whereas mGluR5 decreased levels of both these transporters.⁸⁴

Microglial Involvement in Glutamate Handling and Signaling

Microglia are macrophage analogs of the CNS that maintain homeostasis through orchestration of actions beyond immunity/inflammation. The principal immune cell in glioblastoma, TAMs, comprise approximately 25% of all tumor cells.⁸⁵ Even though gliomas have an immunosuppressive TME, microglia amass in progressive stages of epileptogenicity and invasion.⁸⁶ In addition, microglial infiltration into medically refractive epileptic lesions is phenotypically pro-inflammatory.⁸⁷ As glutamate builds in gliomas, compensatory mechanisms may be triggered. Glutamate-guided chemotaxis is 1 mode of microglial recruitment to a site of local injury.⁸⁸ Major glutamate transporters on microglia may modulate glutamate in the TME.⁸⁹ xCT is upregulated in human TAMs co-cultured with glioblastoma cells.⁹⁰ In contrast, EAAT1 and EAAT2 are downregulated in isolated human TAMs, and these expressive alterations may oppose high levels of glutamate at the peri-glioma zone. In non-glioma models, microglia are suggested to contribute to the excitotoxic conditions of glutamate. As examples, activated microglia alter astrocyte physiology by reducing their capacity for glutamate uptake⁹¹ while signaling glutamate release.⁹¹ Furthermore, an important nonimmune process of microglia is synaptogenesis facilitated by glutamate signaling.⁹² Neurons, particularly those with high activity, recruit microglia. In return, microglia secrete soluble factors (BDNF, etc.) to form/strengthen local synaptic connections while pruning excess synapses.⁹³

Targeting Microglial Function.—

A number of approaches have been utilized to target the microglial component of infiltrating gliomas.⁹⁴ To our knowledge, none have looked at directly targeting GRE. In non-glioma seizure models where seizures are associated with a shift toward pro-inflammatory (M1) macrophages, inhibition of inflammasome activity has decreased neuronal loss and improved seizure control.^95,96 How this would translate into the setting of GRE is unclear.

Monitoring Therapeutic Response: Glutamate Quantification

Advanced MRI techniques such as chemical exchange saturation transfer (CEST), a technique that harnesses selective radiofrequency pulses to saturate the water-exchangeable protons of the target compound, could further elucidate metabolic alterations in gliomas with and without epilepsy.⁹⁷ This method is excellent for assessing the perturbation of glutamate homeostasis.⁹⁸ Furthermore, peritumoral glutamate-weighted CEST (GluCEST) signal correlates with recent seizures and drug-resistant epilepsy further implicating the peritumoral brain regions in epileptogenesis.⁹⁹ If glutamate-targeting approaches prove successful, one could imagine utilizing GluCEST in response assessment.

GABAergic Signaling

GABAergic Signaling in Oncogenesis and Epileptogenesis

Malfunctions in inhibitory interneuron networking are suggested to be a prerequisite for hyperexcitable conditions in glioma.¹⁰⁰ Such alterations have challenged the premise that GABA is an excitatory rather than inhibitory neurotransmitter for peri-glioma neurons. In the TME, GABAergic interneurons are significantly decreased relative to control samples.¹⁰⁰ Successful GABAergic neurotransmission relies on 2 transporters: K⁺-Cl^– cotransporter (KCC2) and Na⁺-K⁺-Cl⁻ cotransporter (NKCC1). Given the opposite Cl⁻ flux through these channels, an imbalance in their expression/function interferes with neuroinhibition. Extrusion of intracellular Cl⁻ via KCC2 specifically drives fast GABA_A synaptic inhibition. In the TME, there is decreased expression of KCC2 and increased expression of NKCC1.¹⁰¹ After knocking down KCC2, seizures were evoked spontaneously in a glioma model. In this same model, bumetanide, an NKCC1 antagonist, mitigated seizures.¹⁰² In part, alterations in KCC2 and NKCC1 compromise GABAergic signaling and consequently, bolster glutamatergic excitation in GRE.

Ineffective inhibitory neurosignaling in glioma may influence patient survival. Higher GABA_A receptor (GABA_AR) subunit expression is associated with better prognosis.¹⁰³ However, GABAergic transmission may differentially impact glioma progression. The loss of GABA_AR expression in HGG such as glioblastoma is 1 important factor for this discrepency.¹⁰⁴ In an optogenetic study, GABAergic interneuron activity reduced glioma proliferation.¹⁰⁵ Selectively agonizing ion currents through GABA_AR slowed glioma development and increased survival of glioma-implanted mice.¹⁰⁶ Such data suggest GABA_AR signaling may interrupt growth-signaling cascades stimulated by peri-glioma excitatory currents. In contrast, GABA_AR stimulation has facilitated disease progression in pediatric low-grade glioma (LGG)-xenografted mice which may be attributed to paradoxical GABAergic excitation mediating tumor growth.¹⁰⁷ In concordance, pediatric HGG was unresponsive to GABAergic signaling. This corroborates previous evidence of more malignant tumors demonstrating minimal/nonexistent GABA_AR expression. The potential for GABA to fuel glioma growth as an energic substrate may also play a role in tumor progression.¹⁰⁸

Targeting Aberrant GABA Signaling

To our knowledge, no published clinical trials have evaluated GABA antagonism for the treatment of gliomas. While pregabalin has been studied in a phase 2 trial in such a setting, its mechanism of action is that of a voltage-gated Ca²⁺ channel blocker despite its structural similarity to GABA.¹⁰⁹

Targeting Aberrant GABA Signaling: Vigabatrin.—

Vigabatrin (γ-vinyl-GABA) enhances neuroinhibitory transmission via inhibition of GABA transaminase. Vigabatrin not only inhibits GABA, but it may also dampen mTOR hyperactivation associated with TSC.¹¹⁰ Early treatment of vigabatrin at first detection of epileptiform discharges on electroencephalogram decreases incidence and prolongs onset of infantile spasms in patients with TSC. There is currently opposing evidence whether preventative initiation of vigabatrin improves cognition or seizures in these patients.^111,112 In addition, vigabatrin suppresses¹¹³ the activity of the KCa3.1 channel implicated in glioma invasion.¹¹⁴ No studies have yet investigated the role of vigabatrin in GRE.

Targeting Aberrant GABA Signaling: Clobazam.—

Clobazam, a benzodiazepine used in refractory epilepsy, facilitates neuronal hyperpolarization by targeting the GABA_AR. As an adjunct, clobazam led to a 94% response rate in a cohort of intractable GRE whereby 30% had 6-month seizure freedom.¹¹⁵ Given that clobazam targets the dysfunctional inhibitory network in the peritumoral cortex, it is a plausible anticonvulsant for refractory cases of GRE that have a high degree of neuronal integration.

Targeting Aberrant GABA Signaling: MicroRNA Molecules.—

MicroRNAs (miRNAs) are small noncoding RNA molecules that regulate gene expression by binding to mRNA. Dysregulated miRNA expression is a driver of tumor proliferation and invasion in glioblastoma.¹¹⁶ Elevated levels of miR-155 have suppressed GABAAR-α1 (GABRA1) expression, impairing GABA-mediated inhibition and fostering glioma cell proliferation.¹¹⁷ Additional miRNAs that disrupt GABAergic signaling include miR-10a-5p targeting GABRB2 and miR-34a-5p targeting GABRA3.¹¹⁸ However, miRNAs may act as tumor-suppressing agents. For example, reduced miRNA-139-5p levels correlate with decreased GABRA1 expression and poorer prognosis in glioma patients. This inverse relationship with survival suggests miRNA-139-5p may exert an antitumor effect via GABARs.¹¹⁹ Although more investigations are required to elucidate the role of miRNAs in modulating components of neuroinhibition, targeting their activities may hold power to diminish peritumoral hyperexcitability associated with gliomagenesis and GRE.

Matrix Metalloproteinases

Matrix metalloproteinases (MMPs) secreted from microglia/glioma cells facilitate excitatory neurotransmission. In the TME, MMPs degrade ECM proteins paving paths for invasion. They are known to promote synaptic plasticity.¹²⁰ MMPs help establish glutamatergic transmission by trafficking the NMDAR to¹²¹ and immobilizing the AMPAR from¹²² the cell surface. MMPs can damage the blood-brain barrier causing leakage of pro-epileptogenic elements into the parenchyma.¹²³ MPP proteolysis of perineuronal nets (PNNs) furthermore compromises peri-glioma inhibitory signaling.¹²⁴ Neuronal membrane capacitance is modulated by PNNs, and their degradation has decreased firing rates of GABAergic interneurons. The expression of MMPs has been found to increase during postictal inflammation and as a consequence, epileptogenic tissue may strengthen excitatory neurotransmission with repeated episodes of seizure activity over time.¹²⁵

Targeting ECM Proteins: Tissue Inhibitors of Metalloproteinases

To our knowledge, no published studies have targeted MMP to address GRE. Studies evaluating this approach in nonlesional epilepsy are limited. With this in mind, tissue inhibitors of metalloproteinases have improved seizure control in preclinical models.¹²⁶ MMP inhibition has been studied more extensively for the treatment of gliomas. A phase 2 trial of the MMP inhibitor marimastat had limited efficacy in recurrent glioblastoma with some difficulties in tolerating the agent.

Pathogenic Mutations

Isocitrate Dehydrogenase

IDH1 and IDH2 are enzymes catalyzing the oxidative decarboxylation of isocitrate to α-ketoglutarate. Neomorphic mutations result in a NADPH-dependent reduction of isocitrate to an oncometabolite: D2HG. Diffuse infiltrating gliomas harboring IDH1/2 mutations (IDH^mut) include both astrocytoma and oligodendroglioma. Multiple studies have described IDH^mut as a preoperative risk factor for seizures in both LGGs and HGGs.¹²⁷ Compared to the wildtype isoform, IDH^mut gliomas have marked elevation of D2HG.¹²⁸ Concurrently, pronounced depletions in glutamate levels are seen in association with IDH^mut as the glutamine/glutamate substrate pool is utilized instead of isocitrate to produce α-ketoglutarate.¹²⁹

While the relationship between D2HG and seizure frequency is clear, the mechanistic role of this oncometabolite in seizures is not fully elucidated (Figure 2).¹³⁰ Initially, D2HG was believed to behave as a glutamate-mimetic because it shares a carbon skeleton with glutamate but differs by 1 functional group. Given the role of glutamate as the principal excitatory neurotransmitter, one hypothesis was D2HG tips neuronal network electrochemical gradients toward a hyperexcitable state.¹³¹ In an in vitro study, a prolongation in synchronized bursts was recorded after exogenous D2HG was added to cultured rodent cortical neurons. Selective NMDAR antagonism led to electrical blockade. Another preclinical study suggested that D2HG-driven spike activity occurs via a metabolomic shift rather than direct agonism of the NMDAR.⁴¹ Using neuronal-glial cultures and electrographically sorted human cortical tissue from patients with IDH^mut glioma, D2HG induced neuronal spiking and promoted a unique metabolic profile and upregulation of mTOR signaling. Another group has suggested neuronal excitation is mediated by D2HG activation of mTOR signaling.⁴² However, a follow-up study demonstrated that D2HG does not activate glutamate receptors nor impede glutamate uptake, and that D2HG downregulates rather than activates the mTOR pathway.¹³² Some of these discrepancies in the literature may be due to differences in experimental methods used as the utilization of exogenous D2HG does not take the glutamate-depleted state of IDH^mut tumors into consideration.^129,133 Consequently, one currently prevailing hypothesis is that non-neoplastic glia in the TME are necessary to elicit D2HG-induced hyperactivity.¹³² Further studies including those utilizing in vivo metabolic imaging techniques could elucidate the mechanistic details by which D2HG creates a hyperexcitable environment.

Figure 2.

Proposed mechanisms of D2HG-induced seizures. D2HG has both up- and downregulated the mTOR pathway in different preclinical GRE models. Direct activation of the NMDAR is also postulated to elicit neuronal hyperexcitability. IDH and mTOR inhibition may decrease seizure frequency in certain epileptogenic gliomas.

Figure 3.

Single-voxel MR spectroscopy in an IDH^mut diffuse astrocytoma. 3T single-voxel ¹H MR spectrum (echo time 97 ms) of a patient with a pretreatment IDH1-R132H^mut grade 2 astrocytoma who presented with secondary generalized seizures. The presence of the D2HG peak as well as the biphasic lactate peak are pathological. MRS detects the sum of both D- and L-2HG (courtesy: Georg Oeltzschner, PhD).

IDH as a Therapeutic Target: IDH Inhibitors.—

IDH inhibitors can potentially deplete >90% of intratumoral D2HG.¹³⁴ They have primarily been evaluated for their oncologic role rather than their antiepileptic properties. Results of the phase 3 trial of vorasidenib for IDH^mut glioma revealed improved seizure control compared to placebo.^135,136 In an in vivo IDH^mut glioma model, daily administration of vorasidenib decreased epileptiform spikes independent of any effects on tumor volume.¹³² Given the potential dual application in tumor and seizure control, IDH inhibitors may prove to be a reliable mode of precision medicine for IDH^mut gliomas. Given their strong association to IDH^mut gliomas, seizures have been proposed to be a secondary outcome in determining the efficacy of treatment regimens in future clinical trials.¹³⁷

Monitoring Therapeutic Response: Imaging for IDH^mut Gliomas.—

The newly reinforced need for early imaging biomarkers is highlighted by the limited utility of standard MRI changes in predicting seizure-related outcomes. In one such study, metabolic volume reductions on amino-acid PET predicted seizure control after treatment in LGG.¹³⁸ Magnetic resonance spectroscopy has also shown pharmacodynamic response with a >70% D2HG level reduction after a week of IDH inhibitor therapy (Figure 2). However, discordance with conventional MRI sequences was noted with short-term net growth on FLAIR.¹³⁹ Another study evaluating the impact of frontline ivosidenib on the growth trajectories of grade 2-3 IDH^mut gliomas found a volumetric response rate of 75%. The only patient with intractable seizures in this cohort had a temporary reduction in tumor growth rate without successful control of seizures.¹⁴⁰ Therefore, additional studies may be needed to help identify early predictive biomarkers of both tumor and seizure responses.

B-Raf Proto-Oncogene

B-Raf proto-oncogene (BRAF) is an integral regulator in the mitogenic response of the MAP kinase/ERK signaling cascade. ERK is a pluripotent transcription factor activating numerous cellular activities. A missense mutation (BRAF^V600E) is detected in various glial neoplasms.¹⁴¹ In glioneuronal tumors (GNTs), BRAF^V600E is associated with worse postoperative seizure outcomes.¹⁴² The feature of BRAF^V600E giving rise to a hyperexcitable neuronal phenotype is described in preclinical studies.^143,144 During brain development, integration of BRAF^V600E into murine neural progenitor cells forms epileptogenic tumors resembling gangliogliomas.¹⁴³ These models overexpressed RE1-silencing transcription factor, a transcriptional repressor implicated in epilepsy, which is also detected at elevated levels in pediatric GNTs/LGGs with BRAF mutations.¹⁴⁴

BRAF as a Therapeutic Target: BRAF Inhibitors.—

BRAF inhibitors represent an advancement in targeted therapy for brain tumors. The combination of dabrafenib (BRAF inhibitor) and trametinib (MEK inhibitor) are components of first-line treatment for BRAF^V600E pediatric glioma.¹⁴⁵ The pan-RAF inhibitor tovorafenib is used in pediatric patients with relapsed/refractory LGGs. Furthermore, BRAF and MEK inhibitor combinatorial therapy is used in unresectable and/or metastatic solid tumors with BRAF^V600E mutation.¹⁴⁶ In a preclinical ganglioglioma model, intraventricular infusion of vemurafenib in vivo decreased seizures.¹⁴⁴ While this may be a clinically impractical approach, it lends support for this pathway as a viable target in GRE.

Conclusions and Future Perspectives

While the clinical association between gliomas and seizures has long been known, we are entering an era in which pathophysiologic mechanisms for epileptogenesis are becoming better understood, and more importantly, neuronal-glioma interactions are being elucidated. These complex multifaceted processes are intertwined in a manner which will likely take years to unravel. Physical, electrical, and chemical forces interacting to drive gliomagenesis and epileptogenesis may prove to reveal promising targets to stem both processes.

For over a decade, the role of ASMs as antitumorigenic agents with a correspondent survival benefit in glioma patients has been suggested. Preclinical and clinical studies are supportive, but despite promising data, the evidence is still conflicting and currently inadequate to support the use of ASMs in all patients with gliomas and as a routine component of oncologic management.¹⁴⁷ Molecular studies and the continued development of preclinical GRE models are advancing steps toward identifying glioma subtypes and mutations which will most benefit from mitigating cortical hyperexcitability.

Understanding the etiologic basis of hyperexcitability in individual gliomas can guide the assessment of the effectiveness of the various therapies discussed in this Review. There are at least 3 overarching mechanisms leading to hyperexcitability in gliomas: genetic mutations, pathophysiology involved in gliomagenesis and the associated TME, and the consequences of treatment regimens (eg, radiation, chemotherapy). The predominant factors likely vary between active, proliferating tumor and more quiescent disease. Our understanding is limited by the difficulty of studying these distinct phases in preclinical models.

Initial genetic testing will provide a relatively straightforward path to determine epileptogenic mutations such as mTOR, IDH, and BRAF for the indication of precision therapeutics. As discussed, multiple aberrant processes accompanying gliomagenesis produce epileptogenic tissue including the reorganization of the synaptic constituency, inflammation, and excitation-inhibition imbalance. Uncertainty remains which of these factors most contributes to peritumoral hyperexcitability. Given that several of these processes are likely linked to one another, targeting a central modulator may prove to be sufficient in managing certain manifestations of GRE. A barrier in advancing studies to target peri-glioma disarray is the void in current diagnostic tools to detect/monitor these epileptogenic drivers related to gliomagenesis. Initiatives can be pursued to address this gap. As examples, the incorporation of GluCEST can help assess the efficacy of glutamate inhibitors while candidates for future trials targeting hyperexcitable modulators can be identified with novel biomarkers as recently highlighted with TSP1.²³ By clarifying the origin of hyperexcitability in glioma patients, we will open the door for more personalized treatment regimens for both seizure and tumor management.

Uncertainty still exists around the best timing to initiate precision therapeutics. Such recommendations are challenging as seizure response is seldom a measured outcome in clinical trials. Its inclusion in recent IDH inhibitor trials represents an encouraging shift.^134,135 Given shared mechanisms, there is growing clarity that future trials should incorporate seizure frequency.¹⁴⁷ Currently, ASMs are not utilized as a preventive strategy for epilepsy in tumor cases. This recommendation is based on studies using ASMs other than those covered in this Review. Further investigation into potential therapeutic benefits of early administration of these medications is warranted. There is furthermore no consensus regarding first-line ASMs for an initial seizure or seizure recurrence. By obtaining data that systematically document seizure response and tumor response, we will be better positioned to formulate recommendations tailored to specific genetic signatures of gliomas.

In parallel with our growing understanding in gliomas, substantial opportunities exist for analogous exploration in brain metastases. Certain cancers metastasizing to the brain form indirect¹⁴⁸ and direct¹⁴⁹ synapses and display functional crosstalk with neurons¹⁵⁰ and astrocytes. Glutamate appears to be a key regulator in the malignant potential of brain metastases via activation of ionotropic and metabotropic^151,152 receptors as with gliomas. As such, cutting off communication of signals such as glutamate may become a viable treatment option of CNS neoplasms of different cellular origins. The inhibition of upstream effectors coordinating the excitatory synaptic connectivity of these tumors is more attractive therapeutic targets as glutamatergic input can be completely stunted while avoiding the potent adverse effects of certain AMPAR/NMDAR antagonists. The learnings from this field of research, in turn, have the potential for a broader impact in the large population of patients with brain metastases.

A number of therapeutics reviewed have shown futility in glioma trials. These agents, although ineffective in controlling malignancy, could be repurposed to manage seizures, particularly downregulatory agents of PI3K/Akt/mTOR pathway. It is uncertain, however, if there is adequate enthusiasm in the field for such repurposing of said compounds. An understanding of previously investigated therapeutics can help inform future directions. The most attractive future targets in epileptogenic gliomas will be those that simultaneously interrupt both GRE and tumor progression.

The intersection of GRE and gliomagenesis and progression will remain fertile ground for investigation. A multifaceted understanding of glioma-neuron-astrocyte-immune cell interactions across space and time, a difficult scenario to model, has the potential to allow for important therapeutic advances for patients with gliomas and other CNS tumors. This understanding may reveal therapeutic roles for existing pharmacologic agents and may highlight unique, new therapeutic approaches. Cross-pollination and closer collaboration between oncology, epilepsy, and other neuroscience disciplines may facilitate bringing this to fruition more quickly.

Funding

B.F.K. has no funding to report. J.W.T. has no funding to report. A.B.H. and R.V.L. receive support from NCI P50CA221747. R.V.L. has received funding from BrainUp. D.O.K. receives funding through Bristol-Meyer Squibb/AACR Winn CDA scholarship. M.C.T. has no funding to report. D.J.P. has no funding to report. K.D. has no funding to report. C.H. has no funding to report. A.B.H. receives research funding from Alnylam.

Conflict of Interest

B.F.K. has no conflicts of interest. J.W.T. has no conflicts of interest. R.V.L. has received research support (drug only) from BMS, honoraria for serving on Advisory Boards for AstraZeneca, Cardinal Health, Curio, Merck, Novartis, Novocure, Servier, and Telix, honoraria for consulting for Novartis and Servier, honoraria for serving on the Speakers’ Bureau for Merck, Novocure, and Servier, and honoraria for editing for EBSCO, Elsevier, Medlink Neurology, and Oxford University Press. D.O.K. has no conflicts of interest. M.C.T. has no conflicts of interest. D.J.P. has no conflicts of interest. K.D. has no conflicts of interest. C.H. has no conflicts of interest. A.B.H. serves on the advisory board of Caris Life Sciences and the WCG Oncology Advisory Board; owns stock in Caris Life Sciences; has received consulting fees from BlueRock Therapeutics, Istari Oncology, and Novocure; been provided in-kind support for research from Moleculin, Takeda, ImmunoGenesis, and Carthera.

Authorship

B.F.K., R.V.L., and J.W.T. conceptualized the content and structure of the review. B.F.K. wrote the initial draft of the manuscript. B.F.K., K.D., D.O.K., C.H., D.J.P., M.C.T., A.B.H., R.V.L., and J.W.T. extensively revised the manuscript for intellectual content. All authors approved the final draft.