Finding a better drug for epilepsy: Antiinflammatory targets


Address correspondence to Stefanie Dedeurwaerdere, PhD, Department of Translational Neuroscience, CDE T4.20, Faculty of Medicine and Health Sciences, University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Belgium. E-mail:


This monograph summarizes one of the sessions of the XI Workshop on Neurobiology of Epilepsy (WONOEP), and provides a critical review of the current state of the field. Speakers and discussants focused on several broad topics: (1) the coexistence of inflammatory processes encompassing several distinct signal-transduction pathways with the epileptogenic process; (2) evidence for the contribution of specific inflammatory molecules and processes to the onset and progression of epilepsy, as well as to epilepsy-related morbidities including depression; (3) the complexity and intricate cross-talk of the pathways involved in inflammation, and the discrete, often opposite roles of a given mediator in neurons versus other cell types. These complexities highlight the challenges confronting the field as it aims to define inflammatory molecules as promising targets for epilepsy prevention and treatment.

The Workshop on Neurobiology of Epilepsy (WONOEP), a biennial meeting preceding the international epilepsy conferences, has been created by the International League Against Epilepsy (ILAE) Commission on Neurobiology to deal with important areas of basic research in epilepsy. The topic of the 2011 XI workshop was “Finding a better drug for epilepsy.” This critical appraisal is a synthesis of the presentations and questions raised during the discussion of the panel session on antiinflammatory targets. Through this mechanism, this review provides a critical assessment of the field, and the remaining challenges.

The unmet clinical need remains unacceptably high for patients with refractory epilepsy. A key goal in the therapy of epilepsy is to improve treatments for controlling spontaneous recurrent seizures in these patients. Another major challenge is to develop disease-modifying drugs endowed with antiepileptogenic properties. Such drugs should prevent the onset of the disease after an epileptogenic injury, or arrest its progression once the disease has already developed. These actions would represent a crucial added value to currently available, mainly symptomatic therapies. In this context, the putative use of antiinflammatory drugs in epilepsy has been invoked as a promising therapeutic strategy on the basis of the accumulating experimental and clinical evidence linking brain inflammation to epilepsy in a causal and reciprocal relationship (Fabene et al., 2010; Vezzani et al., 2010, 2011b). Indeed, proof-of-concept studies evaluating compounds with clinical potential, which are directed at inflammatory targets, are emerging (reviewed in Vezzani et al., 2010, 2011b).

The presence of various inflammatory mediators in experimental and human epileptic tissue and serum has been well described (Fabene et al., 2008; Friedman et al., 2009; Dube et al., 2010; Vezzani et al., 2011b). In addition, epileptogenic insults including trauma, ischemia/hypoxia, fever, and infection, as well as recurrent seizures, have been shown to cause rapid onset inflammatory processes in the brain regions affected by the epileptogenic event. Inflammatory processes are found during the period preceding the onset of frank epilepsy (spontaneous seizures) in experimental models, raising the potential of their role in this pathologic process (Dube et al., 2010; Friedman & Kaufer, 2011; Vezzani et al., 2011b). Several specific inflammatory mediators can contribute to neuronal network hyperexcitability and decrease seizure threshold. Other actions of inflammatory molecules augment cell loss and influence blood–brain barrier (BBB) permeability (Fabene et al., 2008; Friedman et al., 2009; Kim et al., 2009; Friedman, 2011; Vezzani et al., 2011b), thereby contributing to seizures and to the epileptogenic process (Marchi et al., 2007; (Friedman et al., 2009; Friedman, 2011; Pitkanen & Lukasiuk, 2011). The mechanisms of function of inflammatory mediators induced in the brain after an epileptogenic challenge include both activation of posttranslational pathways in neurons, which affect their excitability threshold, as well as the more widely recognized transcriptional activation of genes in endothelial cells, glia, and neurons (Vezzani et al., 2011b).

An additional aspect of the role of inflammation in epileptogenesis stems from the growing literature on persisting effects of early life inflammation (Koh et al., 1999; Choi et al., 2009; Riazi et al., 2010). These may include reduced seizure threshold and vulnerability to seizure-induced cell injury, which is accompanied by enhanced brain inflammation. Therefore, early life inflammation might provide the first “hit” in a two-hit hypothesis of epileptogenesis (Koh et al., 1999).

Genetic tools have also enriched our understanding of inflammation and epilepsy. The use of transgenic mice that overexpress or lack cytokines or cyclooxygenase 2 (COX-2) in astrocytes or in neurons, conditionally lack CD11b (thus suppressing microglia activation), or lack specific endothelial cells adhesion molecules, enabling dissection of the role of inflammation in the developmental and adult aspects of network excitability. Whereas such tools provide important information about the role of specific mediators on specific cell types, the confounders of potential developmental and compensatory changes intrinsic to these models remain (Vezzani et al., 2011a).

The evolution of epilepsy itself offers an analogous complexity: we must consider that the beneficial effects obtained by blockade of specific inflammatory processes in the context of established chronic epilepsy may not necessarily influence the dynamic and evolving process of epileptogenesis. It is notable that inflammation is a broad spectrum of molecules and processes and is salubrious or detrimental in specific contexts. These include the extent and duration of injury; the types of tissue, cellular sources, and cellular targets of inflammatory mediators; and the type of effector molecules released. Within the complexity of this dynamic system, the same molecules playing a detrimental role in chronic spontaneous seizures recurrence may instead have restorative and repair properties in the immediate or proximal phases of epileptogenesis.

As we consider therapeutics, the discovery of biomarkers or surrogate markers is crucial to define populations at risk and to monitor therapeutic efficacy. Imaging techniques, such as magnetic resonance imaging (MRI) and positron emission tomography (PET), offer promise for monitoring and quantify brain inflammation across preclinical and clinical studies. Such sensitive, noninvasive, and repeatable techniques are instrumental for selecting suitable patients who could best benefit from antiinflammatory treatments, and provide a powerful tool for diagnostic, prognostic, and therapeutic purposes (Fabene & Sbarbati, 2004; Dedeurwaerdere et al., 2007; Dube et al., 2010; Vezzani & Friedman, 2011).

In summary, over the past decade, the challenge has been to accumulate supportive information for a potential role of inflammation in epilepsy. This challenge is now behind us. As mentioned in the reviews cited earlier, there is a plethora of evidence in human tissue and animal models for brain inflammation in epilepsy of various types. The challenge that is facing the field with respect to “antiinflammatory” treatment at this point is significantly more daunting: We need to sort out the primary from the secondary phenomena and we need to discern causal contributions to epileptogenesis from “innocent bystanders” and epiphenomena. We need to tease out the complexities of (1) the inflammatory network of molecules and (2) multiple brain cell types and connectivity, and create a precise and useful framework of effectors and their targets. These advancements are required to fully exploit the tremendous potential of antiinflammatory strategies in epileptogenesis.

Sources and Routes for Inflammation in Epileptogenesis

Inflammation in general is usually induced in response to a noxious stimuli (such as infection or injury), and its purpose is to defend the host against pathogenic threats. In addition, in developed countries, misguided immune responses and autoimmune processes are a significant source of inflammation. In the central nervous system (CNS), any tissue injury will induce activation of the innate immune response, which may be followed by activation of adaptive immunity, and can give rise to a gradual and slowly progressing change in the function of the surviving local network. This long-lasting change will be functionally expressed as a reduced threshold of the circuit to generate apparently spontaneous, hypersynchronous, mostly excitatory, and neuronal activity (i.e., seizures) (Prince & Tseng, 1993). Recent studies indicate that even a mild injury to endothelial cells, which will increase vessel permeability within the brain tissue (that is, dysfunction of the BBB), is sufficient to induce a rapid (within few hours) astroglial response and significant immune response (Friedman et al., 2009). Indeed, BBB dysfunction is a common finding in many epileptogenic conditions, including status epilepticus and traumatic and ischemic brain injuries. An association was reported between vascular alterations and hyperemia in certain brain regions and apoptotic cell death (Fabene et al., 2003). Furthermore, a positive correlation was described between the extent of BBB opening and the number of spontaneous seizures evoked in rats by electrically induced status epilepticus (van Vliet et al., 2007). Direct evidence for the involvement of BBB dysfunction in inflammation and epileptogenesis has recently been established by a series of animal studies in which a long-lasting opening of the BBB in the rat neocortex was shown to result in the delayed appearance of epileptiform activity (for reviews see (Shlosberg et al., 2010). The mechanisms underlying the gradual development of network hypersynchronicity under the BBB-deprived brain are not entirely clear; however, experimental data support a role for an inflammatory process induced by serum proteins that normally are not found in the adult CNS (e.g., albumin), via transforming growth factor β (TGF-β) signaling. The rapid activation of astrocytes and microglia following the extravasation of serum proteins into the brain, implicates BBB as a key regulatory element of the brain innate immune system and the communication between intrinsic brain cells and peripheral immunocompetent cells (Marchi et al., 2011; Vezzani & Friedman, 2011).

Potential Molecular Targets: Cell Adhesion Molecules

A role for neuronal cell recognition molecules in epileptogenesis, specifically of neural cell adhesion molecule (NCAM), polysialylated NCAM (PAS-NCAM), extracellular matrix glycoprotein tenascin-R (TN-R), cadherins, and reelin has been postulated. Indeed, in the epilepsy-like disorder (EL) mice, an established model with idiopathic complex partial seizures that secondarily generalize through the hippocampus (Murashima et al., 2002), levels of PSA-NCAM, cadherin, TN-R, and reelin were significantly increased during early developmental stages (3–7 weeks), and then decreased at 10 weeks and remained very low thereafter. The down-regulation in these proteins was observed before the development of frequent seizures, and contrasted with the unchanged expression of NCAM. The precise ways in which changes in cell recognition molecules in the brain of EL mice interact with inflammation remain obscure (Downer et al., 2010; Wang & Neumann, 2010).

Nonneuronal adhesion molecules, such as intercellular adhesion molecule (ICAM), vascular cell adhesion molecule (VCAM) and E- and P-selectins have also been demonstrated to play a key role in BBB leakage and in the onset of epileptogenesis induced in mice by pilocarpine (Fabene et al., 2008). In particular, α4 integrin and its ligand VCAM-1 contribute to pathogenic events required for epileptogenesis developing in mice after pilocarpine-induced status epilepticus (Fabene et al., 2008). Furthermore, genetically modified mice deficient in PSGL-1 (Selplg −/−) or in α1-3-fucosyltranferases (FucTs) FucT-VII and FucT-IV, which are enzymes required to generate functional selectin-binding carbohydrates, have been reported to be resistant to pilocarpine-induced status epilepticus, or to have a reduction in the subsequent number of spontaneous recurrent seizures (Fabene et al., 2008). It is still not yet resolved if these findings are confined to the pilocarpine model or can be generalized (Zattoni et al., 2011).

Inflammation in Epilepsy Comorbidities

Brain inflammation might also contribute to cognitive defects and depression that commonly are associated with temporal lobe epilepsy (TLE) (Mensah et al., 2006; Dube et al., 2009; Hecimovic et al., 2011). Epilepsy-associated depression is severe and promotes the rate of suicide (Hecimovic et al., 2011). Therefore, specific inflammatory pathways relevant to TLE may also contribute to the evolvement of epilepsy-associated mood and cognitive disorders. Depression-like impairments in the pilocarpine status epilepticus model were associated with reduced serotonin output from raphe nucleus and the up-regulation of presynaptic serotonin 1-A (5-HT1A) receptors. The selective serotonin reuptake inhibitor (SSRI) fluoxetine exerted no antidepressant effects, whereas treatment with the interleukin (IL)-1 receptor antagonist (IL-1ra) led to the reversal or improvement of performance in tasks considered indicative of depression-like emotions in rodents. Combined administration of fluoxetine and IL-1ra completely abolished all hallmarks of epilepsy-associated depression. SSRI-resistance in epilepsy-associated depression may result from excessive activation of the IL-1β–mediated signaling; consequently, the use of IL-1β blockers together with SSRIs may represent an effective therapeutic approach for SSRI-resistant epilepsy-associated depression (Mazarati et al., 2010; Pineda et al., 2012).

High-mobility group box-1 (HMGB1) is released by cell injury or activation and stimulates toll-like receptor 4 (TLR4) and receptor for advanced glycation end products (RAGE) (Mazarati et al., 2011; Vezzani et al., 2011b), which have both been implicated in seizure precipitation and recurrence as well as in memory impairments. Recombinant HMGB1 disrupted object memory encoding in wild-type, TLR4 knockout, and RAGE knockout animals. Neither TLR4 knockout nor RAGE knockout mice exhibited memory deficits. Blockade of TLR4 in RAGE knockout mice prevented the detrimental effect of HMGB1 on memory. These data suggest that seizure-induced release of HMGB1 may impair nonspatial memory by acting at both TLR4 and RAGE (Mazarati et al., 2011).

Together these studies support a role for specific proinflammatory pathways in depressive-like behaviors and some cognitive functions.

Biomarkers of Inflammation in Epilepsy: A Role for Neuroimaging

Imaging (MRI, PET, ictal single-photon emission computed tomography [SPECT]) is widely used for evaluation of patients with epilepsy, including refractory partial epilepsy (Goffin et al., 2008). In addition, neuroimaging offers a valuable tool for longitudinal investigations of the development and progression of brain abnormalities accompanying several epilepsy syndromes (Dedeurwaerdere et al., 2007). Imaging protocols providing biomarkers for brain inflammation will enhance our understanding of the role of inflammation during epileptogenesis. In addition, once antiinflammatory treatments become available, imaging of brain inflammation would greatly improve evaluation of these novel therapeutic targets in a mechanistic noninvasive way.

Imaging techniques for investigation of the BBB integrity and monocyte infiltration including MRI fluid-attenuated inversion recovery (FLAIR) sequence, gadolinium, and gadofluorine contrast MRI, and superparamagnetic labeling have been reviewed previously and are not discussed further here (Oude Engberink et al., 2008; Stoll & Bendszus, 2010). So far, the most successful way to visualize microglia activation has been by means of translocator protein (TSPO)–specific PET tracers (Winkeler et al., 2010), which might represent a sensitive imaging approach to monitor brain inflammation in epilepsy. TSPO, also known as peripheral benzodiazepine receptor (PBR) is localized in microglial cells in low amounts under normal healthy conditions. However, when microglia is activated, TSPO is expressed highly and hence acts as a biomarker for brain inflammation (Banati, 2002). TSPO binding has been detected in surgically resected brain tissue (Johnson et al., 1992; Kumlien et al., 1992; Sauvageau et al., 2002), and in vivo by means of PET imaging in patients with TLE (Hirvonen et al., 2012). Dedeurwaerdere and colleagues performed a small animal PET study in the kainic acid–induced status epilepticus model using [18F]-PBR111, a radioligand with high specificity for TSPO. They found increased [18F]-PBR111 binding during early epileptogenesis in brain regions involved in seizure generation. These in vivo results were validated and confirmed by postmortem immunohistochemistry of activated microglia (OX-42) and TSPO autoradiography. Future efforts should identify novel PET ligands targeting specific pathways in the inflammatory cascade.


Inflammation: roles in epileptogenesis and disease progression

As apparent from the introduction, as well as from the presentations and discussion, there is an increasingly clear role for inflammatory molecules in the generation of epilepsy. Mechanisms exist for initiation and perpetuation of inflammation in the brain, including the local activation of glial cells by the epileptogenic insult (e.g., trauma, fever, and infection) or by the recurrence of seizures, and docking and trans–blood vessel invasion of leukocytes into the brain either due to brain parenchymal or to peripheral inflammation, or both, which may contribute to BBB opening (Librizzi et al., 2007; Fabene et al., 2008; Marchi et al., 2011; Vezzani et al., 2011b; Zattoni et al., 2011; Librizzi et al., 2012). Diverse types of inflammatory mediators are synthesized and released in epileptogenic conditions (Gorter et al., 2006); specific molecules such as IL-1β and tumor-necrosis factor α (TNF-α) not only alter neuronal excitability and promote seizures, but appear also to be involved in the plasticity of the network, which seems to be essential in the epileptogenic process itself.

There is also an emerging role for inflammatory molecules in disease progression: molecules released from cells injured or activated during an initial insult or during epileptic seizures contribute to inflammatory cascades. An eloquent example is the molecule HMGB-1, released during a variety of cell insults, which activates toll-like receptors (TLRs). Finally, there is evidence for a contribution of inflammatory processes in the comorbidities that often accompany epilepsy. A salient one is depression, that arises in close to 50% of individuals with refractory epilepsy (Jackson & Turkington, 2005), and comprises specific defects in the function of select neuronal populations. In view of the striking resemblance of cytokine-induced “sickness behavior” and depression, the discovery of a role for inflammatory cytokines in depression is not surprising.

Therefore, in considering the potential utility of antiinflammatory drugs, four therapeutic goals arise: prevention of epilepsy, management of chronic seizures, management of comorbidities, and prevention of epilepsy progression.

Pathway-specific versus global approaches to inflammation targeting in epilepsy

The presentations and the discussion focused on several distinct molecular and cellular inflammatory pathways. BBB induced activation of TGF-β signaling, TLRs, and the agents that activate them, and the downstream consequences: IL-1β-mediated signaling, leukocytes and adhesion molecules, and the mTOR-pathway. The latter is discussed in a separate monograph and will not be considered further here.

Evidence in support of contributions of several of these molecular pathways to epilepsy is strong, which raises the possibility that a therapeutic approach might aim to block all inflammation. However, two conceptual problems exist with such an approach. First, the initial trigger or insult likely sets in motion also immune agents and processes that are antiinflammatory (e.g., IL-ra, IL-10, inhibitors of complement cascade, and insulin growth factor). Blocking all of the immune signaling would depress these important “endogenous antiinflammatory” agents. In addition, inflammation might contribute to repair and processes that serve to minimize or protect from the major changes in neuronal circuits that promote the emergence of spontaneous seizures. Such a salubrious effect of inflammation is considered to take place in multiple sclerosis, and potentially in other chronic neurodegenerative disorders and spinal cord injury (Schwartz & Shechter, 2010). Therefore, a more prudent approach should involve the targeting of a single inflammatory cascade, potentially at several signaling points. An example is blocking IL-1β synthesis via interleukin-1β-converting enzyme (ICE)/caspase 1 inhibitors, as well as the interaction of the cytokine with its receptor using IL-1ra, or downstream signaling activation (Vezzani et al., 2010). The use of such innovative drugs for chronic epilepsy treatment needs to be carefully evaluated with respect to side effects and efficient brain delivery as reviewed by (Vezzani et al., 2010). Whereas the above approach might appear logical, it neglects to consider the strong evidence of cross-talk among distinct inflammatory signaling cascades and pathways. Interference with IL-1β might augment “compensatory” changes and potentially worsen the overall outcome. A further confounder, which is only beginning to emerge, is discussed in subsequent text.

Cell-specific consequences of antiinflammatory approaches

As we consider a potential role for antiinflammatory strategies in preventing or treating epilepsy, the remarkable complexity of the brain substrate where epilepsy arises must be recognized. Numerous specific cell types, including neurons of various subcategories and different glial cells express and respond to inflammatory mediators in distinctly different ways. Therefore, a given molecule might be beneficial for a population of cells and be detrimental for another. This was elegantly shown recently for the consequences of deleting neuronal versus astrocytic cyclooxygenase 2 (COX-2), a prostaglandin synthetic enzyme (Serrano et al., 2011). Similarly, inflammation was recently found to influence neurogenesis differentially in the subventricular zone and subgranular zone of the hippocampus of immature rats (Covey et al., 2011). Therefore, indomethacin, a COX-2 inhibitor, decreased medially situated subventricular zone cells and enhanced proliferation in lateral subventricular zone and hippocampal subgranular zone. The agent also differentially influenced IL-6 expression and microglia migration.

Whereas a role for leukocyte–endothelium interaction in epileptogenesis is suggested, the complete spectrum of the effects of peripheral immune cells is still under active study.

Hence, when addressing means to manipulate selectively distinct inflammatory pathways, these diverse actions should be considered, and the complex total sum of these approaches on epileptogenesis, seizure generation, and mechanisms of neuronal repair should guide the therapeutic potential of antiinflammatory drug candidates.


Dr. Dedeurwaerdere received support from Fonds voor Wetenschappelijk Onderzoek (FWO), Bijzonder Onderzoeksfonds (BOF) Universiteit Antwerpen and Dr. Baram’s work was supported by National Institutes of Health (NIH) grants NS35439 and NS28912.


None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.