Asla Pitkänen MD, PhD
Professor of Neurobiology, A.I. Virtanen Institute, University of Kuopio, Kuopio, Finland
GENERAL PRINCIPLES OF EPILEPTOGENESIS
Epileptogenesis refers to the dynamic processes underlying the appearance and natural history of epilepsy (1). It is part of the “epileptic process” that can be divided into three phases: initial brain damaging insult → latency phase or epileptogenesis (no seizures) → appearance of spontaneous seizures (newly diagnosed epilepsy) (Figure 15.1). Various brain insults, including head trauma, stroke, encephalitis, or status epilepticus (SE) can cause neuronal damage and can initiate epileptogenesis (2). After a latency period that may last from weeks to months to years, spontaneous recurrent seizures begin, and the diagnosis of epilepsy is made. Finally, as experimental and human data show, neurobiologic changes as well as clinical symptoms can continue to be altered even after the spontaneous seizures have become recurrent.
In humans, epileptogenesis and the progressive aspects of epilepsy are best understood in patients with seizure onset in the temporal lobe. Moreover, most of the animal models of epileptogenesis mimic the generation of symptomatic temporal lobe epilepsy (TLE) in mature brain. Finally, experimental and clinical attempts to prevent epileptogenesis have focused on preventing symptomatic epilepsy with focal onset of seizures. Therefore, this chapter reviews the currently available strategies aimed at preventing epileptogenesis triggered by brain damage and the development of symptomatic TLE.
NEUROBIOLOGIC BASIS OF EPILEPTOGENESIS
Understanding neurobiologic changes underlying epileptogenesis at the molecular level is key to designing rational therapies to prevent epileptogenesis in patients who are at risk. The dearth of human samples emphasizes the importance of using adequate animal models in such studies. Kindling is perhaps the most often used model of epileptogenesis (3).Kindling refers to a phenomenon in which an initially subconvulsive stimulus eventually evokes seizures when it is administered repeatedly. Unlike in the genesis of symptomatic TLE in humans, however, in kindling, the following pertain: (a) the electrical stimulus is applied to a
P.170
structurally intact brain; (b) the “epileptogenic phase” (development of kindling) includes evoked seizures that result in neuronal damage, axonal plasticity, and memory impairment; and finally, (c) the development of spontaneous seizures is rare. With these caveats in mind, kindling may actually model the progression of cryptogenic TLE better than the epileptogenesis preceding the occurrence of spontaneous seizures.
FIGURE 15.1. The epileptic process leading to the development of human symptomatic temporal lobe epilepsy includes three phases: brain injury (trauma, stroke, infection, status epilepticus) → latency period (epileptogenesis) → spontaneous seizures (epilepsy). Available data suggest that several parallel pathologic processes occur during epileptogenesis. These processes include neuronal damage (acute and delayed), plasticity (axonal and dendritic), gliosis, neurogenesis, and molecular reorganization. Some data suggest that remodeling of neuronal circuits continues after the spontaneous seizures begin (see text). |
Experimental available data are from models in which epileptogenesis results in the occurrence of spontaneous seizures triggered by inducing SE either chemically, with kainic acid, pilocarpine, or lithium-pilocarpine, or electrically, by stimulating the amygdala, hippocampus, perforant pathway, or angular bundle. Data obtained from these models indicate that several neurobiologic events can progress in a parallel and serial manner during epileptogenesis. These events include neuronal damage, gliosis, axonal and dendritic plasticity, changes in the extracellular matrix, and molecular reorganization (Figure 15.1). The following discussion summarizes the major aspects of the reorganization of neuronal circuits occurring during epileptogenesis (4).
Neuronal Loss
In experimental models, SE that lasts for 30 to 40 minutes is long enough to induce neuronal loss and epileptogenesis. In addition to acute necrotic neuronal damage in various brain areas, SE induces delayed programmed cell death that can continue for several days or even weeks. Based on experimental data, activation of caspases by both intrinsic (mitochondrial origin) and extrinsic (death receptor-mediated) pathways is involved in programmed cell death.
Histologic findings, magnetic resonance (MR) imaging volumetry data, and MR spectroscopy data indicate that SE may also damage various regions of the human brain, including the hippocampus, amygdala, medial temporal cortex, striatum, thalamus, and cerebellum. An elevation in serum neuron-specific enolase (a marker of brain injury) after convulsive or nonconvulsive SE is another indicator favoring the idea that SE may cause structural damage to the human brain. In primates, an 82-minute duration of SE is enough to induce histologic damage, SE lasting 1.5 hours is sufficient to elevate serum neuron specific enolase, and SE lasting 45 to 72 minutes is enough to reduce hippocampal volume. Further, serial MR volumetry imaging studies of the hippocampus demonstrated that, after prolonged focal febrile seizures as well as after SE associated with encephalitis, the progression of hippocampal volume loss can continue for several months or years. Whether a direct relationship exists between the hippocampal damage and the development of TLE remains to be shown. It has been demonstrated, however, that the risk of later epileptogenesis is higher in individuals with SE associated with structural damage than in subjects without such damage (5). Taken together, these data provide a testable hypothesis that neuroprotective treatment started during or after SE and targeted to alleviate delayed or programmed cell death will prevent epileptogenesis.
Axonal Plasticity
The best understood form of axonal plasticity in TLE is mossy fiber sprouting. Mossy fibers are granule cell axons that normally innervate hilar cells and the apical dendrites of CA3 pyramidal cells. Probably because of the death of their normal target neurons in the hilus and CA3 during the epileptic process, mossy fibers sprout and innervate postsynaptic targets in abnormal locations, including the granule cell dendrites in the inner molecular layer of the dentate gyrus and basal dendrites of CA3 pyramidal cells in the hippocampus proper. Through these contacts, granule cells form excitatory circuitries with adjacent granule cells in the epileptic brain.
The contribution of mossy fiber sprouting to the circuitry that generates spontaneous seizures as an end result of epileptogenesis has been questioned by several investigators (4). For example, axonal remodeling also occurs in the CA1 field of the hippocampus proper in experimental animals, as well as in the CA1 and the entorhinal cortex in humans. Otherwise, prevention of mossy fiber sprouting by cycloheximide, a protein synthesis inhibitor, does not prevent the development of epilepsy despite the prevention of mossy fiber sprouting. Further, the density of sprouting is not associated with the latency to the appearance of the spontaneous seizures or their frequency. In addition to the sprouting of excitatory axons, there are reports that the inhibitory axons may also sprout.
Assuming that the sprouting of excitatory axons and the formation of new synapses with other excitatory neurons are responsible for decreasing the seizure threshold and the development of spontaneous seizures, prevention or guidance of axonal sprouting forms an appealing target for the design of new antiepileptogenic compounds. In fact, previous attempts to prevent or delay epileptogenesis in kindling or spontaneous seizure models have revealed an association between the delay in epileptogenesis and the reduction in mossy fiber sprouting (see later). Whether a functional causality exists between the two phenomena remains to be shown. There is, however, also the possibility that manipulation of the naturally occurring plastic response compromises normal recovery (6).
Dendritic Plasticity
In addition to neuronal output regions (axons), neuronal input regions (dendrites) undergo morphologic plasticity in epilepsy. These include changes in spine number and morphology as well as in dendritic branching. According to observations in the rat pilocarpine model, plastic changes in spine morphology and density are dynamic (7). The density
P.171
of spines in granule cells decreased by 95% within 2 weeks of the onset of SE. Some recovery occurs by the time spontaneous seizures appear.
Neurogenesis
The infragranular region in the adult dentate gyrus contains progenitor cells that may differentiate into neurons after single seizures or SE in rat models of TLE (8). Some of the newly formed neurons in the dentate gyrus have the immunochemical phenotype of granule cells, and they project to the CA3 subfield of the hippocampus. In contrast, investigators have described that some newly formed neurons along the hilar-CA3 border have the immunohistochemical phenotype and intrinsic electrophysiologic properties of granule cells, but they are synchronized with spontaneous rhythmic bursts of CA3 pyramidal cells of the hippocampus proper, and consequently they may contribute to the abnormal hyperexcitability in the epileptic hippocampus (9). Whether manipulation of the rate of neurogenesis modifies epileptogenesis remains to be investigated.
Gliosis
After neuronal damage by SE, the number and morphology of astrocytes expressing glial fibrillary acidic protein increase dramatically and chronically (10). Regulation of the extracellular microenvironment, in particular, buffering of extracellular potassium ion increases and clearance of glutamate by astrocytes, is critical for the control of neuronal excitability. Astrocytes can regulate synaptogenesis and neurite outgrowth as well as long-distance signaling through glial networks. The association of these astrocytic functions with epileptogenesis remains to be explored.
Activation of microglia occurs after various brain insults, such as SE or stroke, which are associated with epilepsy later in life. The proposed functions of microglial cells include the release of cytokines, proteases, reactive oxygen species, and nitrogen intermediates. Therefore, these cells are proposed to have a significant role in cell death processes after brain injury that can precipitate epileptogenesis. Manipulation of microglial function provides another route by which the severity of overall neuronal damage can be affected after brain-damaging insults such as stroke or SE that are associated with an enhanced risk of epilepsy later in life.
Molecular Reorganization
In situ hybridization, immunohistochemical tests, and high-output molecular screening techniques have demonstrated changes in the expression of many genes after SE and during epileptogenesis. For example, Nedivi and colleagues (11) estimated that 500 to 1,000 genes become expressed after kainate-induced SE. More recently, we analyzed hippocampal tissue in rats using complementary DNA arrays containing more than 5,000 genes. Tissue samples were collected 2 weeks after the onset of SE in animals that were monitored with videoelectroencephalography to confirm the induction of the epileptogenic phase. Our data indicate upregulation of 88 genes and downregulation of 32 genes. Of these, 27 have been previously described to be expressed in brain, seven have been shown to have role in epilepsy or seizures, and 16 are expressed in other pathologic conditions of the brain (12). It remains to be seen whether these data will guide us to an understanding of the molecular basis of previously described neurobiologic changes, such as neuronal damage, plasticity, gliosis, or neurogenesis occurring during epileptogenesis.
EFFECT OF CURRENTLY AVAILABLE ANTIEPILEPTIC DRUGS ON EPILEPTOGENESIS
Neuronal loss appears to be a common factor that precedes the appearance of spontaneous seizures after SE both in experimental models and humans. Therefore, we hypothesize that a compound would be antiepileptogenic in the following circumstances: (a) if it alleviates neuronal damage caused by brain insults that are associated with an increased risk of epilepsy later in life, even when its administration starts during or after the beginning of the insult (e.g., during ischemia or SE); (b) if the compound delays or suppresses the development of kindling; and finally, (c) if the compound prevents (or at least delays) the development of epilepsy in models, in which spontaneous seizures develop after brain damage (e.g., SE). Table 15.1 summarizes the neuroprotective and antiepileptogenic effects of currently available antiepileptic drugs (AEDs).
Neuroprotection after Brain Damaging Insults
Neuroprotective effects of most of the currently used AEDs have been tested in ischemia models. Both sodium channel blockers (carbamazepine, phenytoin, lamotrigine) and compounds enhancing γ-aminobutyric acid (GABA)ergic neurotransmission (clonazepam, tiagabine, topiramate, vigabatrin) alleviate ischemia-induced neuronal damage in rats when treatment is started during ischemia or soon after the beginning of reperfusion (Table 15.1). Surprisingly, fewer data are available regarding the neuroprotective effects of compounds used to treat SE or of AEDs on SE-induced damage (Table 15.1). Thus far, most of the compounds enhancing GABAergic transmission have a mild neuroprotective effect. In addition, treatment afterward with lamotrigine and valproate alleviates SE-induced neuronal loss.
P.172
TABLE 15.1. NEUROPROTECTIVE AND ANTIEPILEPTOGENIC EFFECTS OF DRUGS USED TO TREAT STATUS EPILEPTICUS OR SEIZURES IN HUMANS |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Delay of Kindling
Studies investigating the effects of AEDs on the development of kindling indicate that most of the sodium channel blockers do not delay the development of kindling, whereas all compounds enhancing GABAergic transmission do have such an effect (Table 15.1). However, when stimulation is reapplied after the drug washes out, the rats typically become kindled. This finding makes it difficult to assess whether AEDs have any direct effects on neurobiologic alterations leading to epileptogenesis or whether the delay of kindling is just associated with the suppression of afterdischarges. Levetiracetam, a newer compound with unknown mechanisms, also delays the development of kindling (13). Unlike any other compound, levetiracetam, used during the induction of kindling, shortens the duration of afterdischarges from 62 to 41 seconds and the duration of seizures from 53 to 34 seconds after kindling has been established following a washout period (13). Whether this finding predicts a disease-modifying effect of levetiracetam in experimental models with spontaneous seizures and in humans remains to be investigated.
Prevention of the Development of Spontaneous Seizures
Next one can hypothesize that compounds that alleviate neuronal damage caused by epileptogenic brain insults and also delay the development of kindling are the best candidate AEDs for the prevention of epileptogenesis and the development of spontaneous seizures after SE. These compounds
P.173
include clonazepam, phenobarbital, valproate, lamotrigine, tiagabine, topiramate, and vigabatrin (Table 15.1). So far, phenobarbital, valproate, and vigabatrin have been tested. In one study, vigabatrin treatment (75 mg/kg/day) that was started 2 days after SE induced by amygdala stimulation in adult rats and was continued for 10 weeks had no clear antiepileptogenic effects. Moreover, seizure frequency and duration did not differ from those noted in vehicle-treated rats, a finding suggesting that vigabatrin administration during epileptogenesis has no disease-modifying effects (14). André, Marescaux, and Nehlig and coworkers did not demonstrate any antiepileptogenic effects of vigabatrin treatment (250 mg/kg/day) that was started after pilocarpine administration in the lithium-pilocarpine model (15). Bolanos et al. (16) investigated the antiepileptogenic effect of phenobarbital (70 mg/kg/day) and valproate (1200 mg/kg/day) in a kainic acid model in 35-day-old rats. Phenobarbital-treated rats developed spontaneous seizures like the vehicle-treated animals. The valproate-treated group, however, had no spontaneous seizures. In this study, only behaviorally generalized seizures were recorded, a practice that compromised the interpretation of the data because partial or subclinical electrographic seizures may have remained undetected.
Clinical trials aimed at preventing epileptogenesis in humans have tested the efficacy of prophylactic treatment with carbamazepine, phenytoin, and valproate in patients with head trauma (17) or valproate in patients with newly diagnosed tumors (18). As these studies demonstrate, there have been no beneficial effects. It is uncertain whether the lack of an effect in humans could have been predicted by preclinical animal studies because the compounds studied in spontaneous seizure models do not include, for example, phenytoin. Moreover, in experimental studies, epileptogenesis was induced by SE, whereas in human studies, the antiepileptogenic effect was assessed after head trauma.
AEDs have been designed to prevent the initiation and spread of spontaneous seizures; that is, they are expected to work on the neuronal circuits that have already undergone modifications, rather than on the networks in which the modifications preceding seizure occurrence are still ongoing. The difference in the molecular mechanisms of epileptogenesis and ictogenesis (seizure initiation) is probably a critical factor underlying the lack of an association between antiepileptic efficacy and antiepileptogenic effects.
OTHER APPROACHES USED TO PREVENT EPILEPTOGENESIS OR TO MODIFY THE SEVERITY OF DEVELOPING EPILEPSY
In addition to AEDs, several other treatment strategies have been attempted to prevent epileptogenesis in experimental models. In a kindling model, pretreatment with the N-methyl-D-aspartic acid (NMDA) antagonist MK-801 (19), enhancement of noradrenergic transmission with intraperitoneal injection of epinephrine bitartrate (20), pretreatment with intraperitoneal injection of the immunosuppressant calcineurin inhibitors cyclosporine or FK506 (21), intrahippocampal injection of brain-derived neurotrophic factor (22) or blockade of receptor function with TrkB receptor bodies (23), intracerebroventricular infusion of nerve growth factor antibodies (24), and intrahippocampal administration of peptides blocking the nerve growth factor receptor (25) prevent or delay kindling and associated plastic changes in the mossy fiber pathway.
Contrary to the idea of using compounds that suppress neuronal activity (e.g., AEDs), several studies indicate that the stimulation of neuronal activity using sensory stimuli, environmental enrichment, or electrically generated stimuli reduces neuronal damage and epileptogenesis in spontaneous seizure models and also delays the development of kindling (26). Finally, experiments using gene therapy to increase the expression of proteins restoring energy metabolism or blocking programmed cell death (27), as well as vaccination against NMDA-type glutamate receptors (28) to protect brain from stroke or SE-induced injury, provide novel ideas to be tested in experimental antiepileptogenesis trials.
Even though no treatments convincingly prevent or reduce the risk of epileptogenesis after brain injury, some treatments do have disease-modifying effects. For example, a ketogenic diet (29) and intraventricular administration of basic fibroblast growth factor (30), started during or after kainate-induced SE, reduced the frequency and duration of spontaneous seizures, even though these regimens did not prevent the development of epilepsy in all animals. Moreover, mice chronically treated with an L-type calcium channel blocker (nicarpidine) starting at the time of pilocarpineinduced SE had milder cognitive deterioration than did vehicle-treated controls (31).
The conspicuous heterogeneity of approaches that have been tested to prevent epileptogenesis in experimental models is indicative of the lack of understanding of the underlying mechanisms of the disease. Breakthroughs in our understanding of the molecular mechanisms of epileptogenesis are necessary for rational systematic attempts to prevent epilepsy after brain injury in the future.
DO NEUROBIOLOGIC ALTERATIONS UNDERLYING EPILEPTOGENESIS CONTINUE AFTER THE BEGINNING OF SPONTANEOUS SEIZURES?
Data from experimental and human studies suggest that the different categories of neurobiologic alterations occurring during SE-induced epileptogenesis can be induced by brief seizures that typically last for less than 2 minutes (4). Using stereologic cell counting methods, silver staining techniques,
P.174
markers of apoptosis, or immunohistochemical staining of subpopulations of neurons, several laboratories have reported that even a few brief seizures can induce neuronal damage in the amygdala and the hilus of the dentate gyrus in the kindling model of TLE in rat. The severity of damage correlates with the number of seizures the animal has experienced. In humans, histologic analyses, MR imaging volumetry, and MR spectroscopy, as well as serum neuron-specific enolase studies, suggest that recurrent seizures induce progressive damage. Further, programmed cell death contributes to seizure-induced damage in rats and probably also in humans (32). In all these studies, only a subpopulation of patients with drug-refractory seizures had progressive damage. Whether this is related to the genotype of the patient or to other factors remains to be explored.
In addition to neuronal damage, the density of mossy fiber sprouting increases according to the number of seizures in the kindling model. Similarly, data provide evidence that sprouting may be an ongoing phenomenon in the human dentate gyrus (33) and the entorhinal cortex (4). Whether ongoing axonal remodeling is related to the progressive neuronal loss or seizure-induced stimulation of axonal plasticity remains to be determined. Brief kindled seizures in adult rats may induce differentiation of progenitor cells into neurons in the infragranular region of the dentate gyrus. One sees increased astrocytosis and activated microglia in samples taken from kindled animals or in patients operated on for drug-refractory TLE. Finally, seizures induce alterations in the expression of certain molecular markers.
Taken together, extensive evidence favors the idea that molecular, cellular, and network reorganization continues after the diagnosis of epilepsy, particularly in patients who are not free of seizures. More data are, however, needed to link each of these alterations with the progression in the frequency or type of seizures, as well as with the progression of other symptoms, such as memory impairment in individual patients. If there is a connection, therapeutic prevention of such modifications in patients who are not seizure free would require consideration.
Another question is whether there is any evidence that ongoing treatments have any neuroprotective or disease-modifying effects in patients who continue to have seizures. Previous studies of the neuroprotective effects of AEDs indicate that pretreatment with phenobarbital, tiagabine, lamotrigine, and vigabatrin alleviates SE-induced neuronal damage, even though these drugs do not completely suppress seizure activity (Table 15.1). Conversely, compounds delaying the development of kindling could also have a disease-modifying effect if they would shorten the seizure duration (Table 15.1). As my colleagues and I have demonstrated, carbamazepine and valproate shorten the duration of spontaneous seizures in an amygdala stimulation model in rat (unpublished). Whether long-term use will provide neuroprotective or seizure-shortening effects that will translate into a functional benefit for drug-refractory patients with epilepsy remains to be studied.
FUTURE PERSPECTIVES
Prevention of epileptogenesis is a major challenge for specialists in epilepsy. More data are needed to fill in the lacunae of our current knowledge about the natural course of epileptogenesis in humans and about markers that can be used to identify those who are at highest risk. Development of new experimental models that better mimic human epileptogenesis after experimental stroke, head trauma, or encephalitis or meningitis will provide tools to answer certain questions, such as the following: How similar are the molecular mechanisms of epileptogenesis after various brain insults? Preclinical testing using these models may also better predict the antiepileptogenic efficacy of new treatments in seizures of different origins. Appreciation that epileptogenesis and ictogenesis have very different neurobiologic bases will undoubtedly guide researchers toward the discovery of compounds that more selectively target epileptogenic, rather than ictogenic, mechanisms. Considering that different molecular pathways are activated in parallel during epileptogenesis, it remains to be seen whether future antiepileptogenic treatment will consist of monotherapy or polytherapy.
Neuroprotection appears to be a critical component related to epileptogenesis. Whether alleviation of neuronal loss after brain-damaging insults will result in a delay or prevention of epileptogenesis remains a testable hypothesis. Otherwise, continuous remodeling of neuronal circuitries in established epilepsy, at least in some drug-refractory patients, will challenge the future treatment of epilepsy to include not only the antiepileptic effect but also features such as neuroprotection.
REFERENCES
P.175
P.176