Therapeutic Approaches to Epileptogenesis – Hope on the Horizon

Summary

Prevention of epileptogenesis is an unmet need in medicine. During the past three years, however, several preclinical studies have demonstrated remarkable favorable effects of novel treatments on genetic and acquired epileptogenesis. These include the use of immunosuppressants and treatments that modify cellular adhesion, proliferation and/or plasticity. Also, the use of antiepileptic drugs in rats with genetic epilepsy or proconvulsants in acquired epilepsy models have provided somewhat unexpected favorable effects. This review summarizes these studies, and introduces some caveats when interpreting the data. In particular, the effect of genetic background, the severity of epileptogenic insult, the method and duration of seizure monitoring, and size of animal population are discussed. Also, a novel scheme for defining epileptogenesis-related terms is presented.

Introduction

According to the World Health Organization, around 50 million people worldwide have epilepsy and the estimated proportion of the general population with active epilepsy, meaning continuing seizures or the need for treatment, is between 4 and 10 per 1,000 people at a given time http://www.who.int/mediacentre/factsheets/fs999/en/index.html). The current treatment of epilepsy is based on the suppression of symptoms (i.e., seizures) by antiepileptic drugs (AEDs). We have no treatments that would favorably modify the disease process in patients with a known increased risk of epilepsy due to genetic predisposition or a history of acquired brain injury, or in patients with a progressive course of epilepsy (Pitkänen and Kubova, 2006; Temkin, 2009). The urgent need to solve this problem has been acknowledged both in Europe and the USA, where the development of treatments to prevent epileptogenesis has been listed as a major objective for epilepsy research (Baulac and Pitkänen, 2009; Kelley et al., 2009).

The development of an acquired epileptic disorder implies abnormal neuronal reorganization occurring over a long period of time following a specific cerebral insult (Engel, 1989). This includes neurodegeneration, neurogenesis, axonal sprouting, axonal injury, glial cell activation, invasion of peripheral inflammatory cells, vascular damage and angiogenesis, changes in the extracellular matrix, and alterations in the molecular structure of cellular components like ligand and receptor-gated ion channels (for review, see Pitkänen and Lukasiuk, 2009). Many of these changes can be found at the time when seizure threshold is lowered and no spontaneous seizures occur. They continue to progress beyond the time of the first unprovoked seizure and even long after epilepsy diagnosis (Pitkänen and Sutula, 2002). From the neurobiological point of view, epileptogenic network reorganization has a starting point, that is, the moment of an insult such as traumatic brain injury (TBI). However, it is difficult to define an endpoint for epileptogenesis, at least relative to the occurrence of seizures.

In experimental models the type, severity, and distribution of pathologic changes vary depending of the epileptogenic insult (Pitkänen et al., 2007: Pitkänen and Lukasiuk, 2009). So does the time window for the development of these alterations before the occurrence of the first unprovoked seizure (Pitkänen et al., 2007). For example, in status epilepticus (SE) models most of the animals develop epilepsy within 2–4 wk, whereas after TBI or stroke the proportion of animals developing epilepsy is lower, and the latency period is longer.

In humans, the details about the course of epileptogenesis after various types of brain insults are poorly known. The data available suggests that most of the patients develop epilepsy within 2 years after after SE, TBI, or stroke demonstrating some similarity between etiologies (Hesdorfer et al., 1998; Bladin et al., 2000; Englander et al., 2003). Even though in vivo imaging has expanded our view about the severity and extent of brain damage in the human epileptic brain, the pathologic analysis of tissue is typically made from the hippocampus in patients with drug-refractory seizures, and the data rarely specifies the findings in different etiologies (see Mathern et al., 1996). Consequently, our understanding of the spectrum and temporal course of epileptogenic circuitry reorganization in different brain areas after various epileptogenic brain insults in humans is far from complete, and the formulation of a unifying hypothesis for the maturation of the epileptogenic network after various brain injuries has been difficult.

Based on the data available, so far, it is unlikely that there is only one type of epileptogenesis. Figure 1 summarizes the possible outcomes after epileptogenic brain insults. The epileptic process can include various stages including lowering of seizure threshold, appearance of seizures, worsening or progression of epilepsy, and in rare cases, spontaneous remission or even cure (Sillanpää et al., 1998; Kwan and Sander, 2004; Shorvon and Luciano, 2007). In addition to seizures, co-morbidities can occur, and they can be progressive (Hermann et al., 2008).

Figure 1.

A schematic drawing showing the different outcomes after epileptogenic brain insults. The blue boxes on the right indicate the natural outcome. Green boxes show the possible treatment effects.

Epileptogenic Process - Terminology

Epilepsy is a multifaceted diagnosis which is reflected in the complexity of terminology. According to the ILAE definition, elements for defining epilepsy include a history of at least one seizure, enduring alteration in the brain that increases the likelihood of future seizures as well associated neurobiological, cognitive, psychological, and social disturbances (Fisher et al., 2005). This definition includes both the epileptic disease and epilepsy syndrome. Epileptic disease refers to a pathologic condition with a single specific, well-defined etiology (Engel Jr, 2001). Epilepsy syndrome was defined as a complex of signs and symptoms that define a unique epilepsy condition (Engel Jr, 2001).

The terminology describing the neurobiological processes that occur before and after epilepsy diagnosis is lacking universally accepted definitions, which complicates the interpretation and comparison of data from different preclinical and clinical studies. For example, in the literature cited in Table 2 the term "epileptogenesis" or "latency period" is often used as an operational term to refer to a time period between the initial insult and the occurrence of the first unprovoked seizure. This definition does not acknowledge the data showing that neurobiological processes leading to epilepsy can continue to progress even after epilepsy diagnosis and contribute to the progression of epilepsy (Pitkänen and Sutula, 2002; Williams et al., 2009). The proposed new definition for epileptogenesis includes both the development of the condition as well as its progression after the condition is established (Table 1).

Table 2.

Effect of treatments o epileptogenesis, co-morbidities, and pathologic changes.

Compound	Model	DISEASE or SYNDROME MODIFICATION
		ANTI-EPILEPTOGENESIS							COMORBIDITY- MODIFICATION		REVERSAL OF PATHOLOGY1
		Prevention		Seizure-modification				Cure
		% of animals that develop epilepsy ⇩	delay in onset	frequency ⇩	duration ⇩	milder seizure type	prevention of progressive increase in seizure frequency	% of animals that become seizure free ⇧	learning and memory	mood and behavior
Rapamycin	Tsc1GFAP CKO mouse [1]	yes	yes	yes	yes	yes	yes	n.d.	yes (Tsc2+/− mice) [2]	n.d	yes
	Pten CKO mouse [3]	no	n.d.	no	yes	yes	yes	n.d.		yes reduced anxiety and improved social behavior	yes
	Pten CKO mouse [4]	n.d.	n.d.	n.d.	yes	n.d.	n.d.	n.d.	n.d.	n.d.	yes
	KA-induced SE [5]	n.d.	n.d.	yes	n.d.	n.d.	yes	n.d.	n.d.	nd.	yes
FK506 (Tacrolimus®)	Electrical stimulation-induced SE [6]	no	no	no	n.d.	no	n.d.	n.d.	n.d.	nd.	n.d.
Celecoxib	Li-pilocarpine –induced SE [7]	no	n.d.	yes	yes	n.d.	n.d.	n.d.	n.d.	nd.	yes
Parecoxib	Pilocarpine-induced SE [8]	no	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	nd.	n.d.
SC58236	Electrical stimulation-induced SE [90]	no	no	no	no	n.d.	n.d.	n.d.	n.d.	nd.	no
α4 integrin specific mAb	Pilocarpine-induced SE in mice [10]	n.d.	no	yes	no	n.d.	n.d.	n.d.	n.d.	yes preservation of exploratory behavior	yes
Erythropoietin	Li-pilocarpine-induced SE in rats [11]	no	n.d.	yes	yes	n.d.	n.d.	n.d.	n.d.	nd.	yes
FGF-2 and BDNF gene therapy	Pilocarpine-induced SE in rats [12]	n.d.	n.d.	yes	n.d.	yes	n.d.	n.d.	n.d.	nd.	yes
Levetiracetam	Spontaneously epileptic rats [13]	n.d.	n.d.	yes	yes	n.d.	n.d.	n.d.	n.d.	nd.	n.d.
Ethosuximide	WAG/Rij rats with spontaneous absence seizures [14]	n.d.	n.d.	yes	no	n.d.	n.d.	n.d.	n.d.	nd.	yes
Atipamezole	Electrical stimulation-induced SE [15]	no	n.d.	yes	no	no	n.d.	n.d.	n.d.	nd.	yes
SR141716A	Lateral FPI-induced TBI [16]	yes (?) Reduction in seizure threshold ⇩	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	nd.	n.d.
	KA-induced SE [17]	no	n.d.	no	n.d.	n.d.	n.d.	n.d.	n.d.	nd.	n.d.

¹Partial or complete reversal of pathology.

Abbreviations: BDNF, brain-derived neurotrophic factor; CKO, conditional knock-out; FGF, fibroblast growth factor; FPI, fluid-percussion injury; KA, kainic acid; n.d., no data; SE, status epilepticus; TBI, traumatic brain injury. References in brackets: 1. Zeng et al. (2008), 2. Ehninger et al. (2008), 3. Zhou et al. (2009), 4. Ljungberg et al. (2009), 5.Zeng et al. (2009), 6. Lukasiuk et al. (2008), 7.Jung et al. (2006), 8. Bankstahl et al. (2009), 9. Holtman et al. (2009), 10. Fabene et al. (2008), 11. Chu et al. (2008), 12. Paradiso et al. (2009), 13. Yan et al. (2005), 14. Blumenfeld et al. (2008), 15. Pitkänen et al. (2004), 16. Echegoyen et al. (2009), 17. Pouliot et al. (2009).

Table 1.

The current definitions of epilepsy, epileptic disease, and epilepsy syndrome. Proposals for defining epileptogenesis, disease or syndrome modification, and antiepileptogenesis are presented. Note that the anti-epileptogenic effect is judged based on clinical outcome, not on pathology. Please see also Figure 2.

Terminology
Epilepsy: A disorder of the brain characterized by an enduring predisposition to generate epileptic seizures and by the neurobiological, cognitive, psychological, and social consequences of this condition. The definition of epilepsy requires the occurrence of at least one epileptic seizure (Fisher et al., 2005)
Epilepsy disease: A pathologic condition with a single specific, well-defined etiology (Engel Jr, 2001)
Epilepsy syndrome: A complex of signs and symptoms that define a unique epilepsy condition (Engel Jr, 2001).
Epileptogenesis: The development and extension of tissue capable of generating spontaneous seizures, including
Development of an epileptic condition Progression after the condition is established
Disease or Syndrome Modification: A process that alters the development or progression of a “disease”, in this case epilepsy (either epileptic disease or epilepsy syndrome). Disease or syndrome modifying interventions may be antiepileptogenic. They could also modify comorbidity by reducing or preventing deleterious nonepileptic functional changes in the brain. They can also modify the pathological changes underlying epileptogenesis or co-orbidities.
antiepileptogenesis co-morbidity modification reversal of pathology (related to either one)
Antiepileptogenesis: A process that counteracts the effects of epileptogenesis, including
prevention seizure modification cure
Prevention: Complete prevention prevents the development of epilepsy. Partial prevention can delay the development of epilepsy or reduce its severity. Antiepileptogenesis can also prevent or reduce the progression of epilepsy after it has already been established.
Seizure modification. Seizures occur but they may be fewer in frequency, shorter, or of milder seizure type.
Cure: The complete reversal of epilepsy.

Proposal for New Terminology for Assessing the Effects of Novel Treatments

As the term epilepsy includes both epileptic disease and epilepsy syndrome, we propose to use the term disease or syndrome modification to imply that the therapy has some effect on the epileptic process. This would include both "anti-epileptogenesis" and "co-morbidity modification", in which co-morbidity includes all non-epileptic disturbances that make up the disease or syndrome (Table 1, Figure 2).

Figure 2.

Partitioning of disease or syndrome modification into antiepileptogenesis, co-morbidity modification, and reversal of pathology. Antiepileptogenesis can be further divided into prevention, seizure modification, and cure. See the definitions of each term in Table 1. At the bottom of the page there are examples of the indicators for seizure modification and co-morbidity modification.

The next question is, how should we define "antiepileptogenesis"?. In studies summarized in Table 2, the term "anti-epileptogenic" has been used to refer to the ability of the treatment to delay the onset of epilepsy, to completely prevent the appearance of seizures, or to lower the percentage of subjects developing epilepsy as compared to the vehicle group. Antiepileptogenesis can also refer to conditions, in which epilepsy develops despite treatment but its appearance is delayed or the epilepsy is milder (lower seizure frequency, shorter seizure duration, milder behavioral seizure type). Sometimes this has been considered as "disease-modification".

"Anti" is a Greek term and generally means "against, opposite or opposing, and contrary". For example, anti-febrile refers to compounds that are effective against fever. Thus, they can lower the fever but do not necessarily result in normal body temperature. Anti-hypertensive drugs lower the blood pressure but do not necessarily result in a normotensive condition. Antibiotics can kill 100% of one type of bacteria but have only partial or no effect on others. If we adopt the analogy, the treatment would be considered anti-epileptogenic even if it does not suppress the percentage of subjects developing epilepsy but just results in epilepsy that is milder. Thus, the current term "disease-modification" would become included into the term "anti-epileptogenic" (see Table 1, Figure 2). The analogy to antibacterial compounds would imply that the treatment can be considered "anti-epileptogenic" even if it works only in one type of epileptogenesis but not in others. In addition, there is a possibility that the "antiepileptogenic treatment" can modify co-morbidities.

Based on this reasoning, antiepileptogenesis can be divided into epilepsy prevention, seizure modification, and cure, and each one of these could have specific indicators (Table 1). Thus, antiepileptogenesis expands over the entire epileptic process including the time period not only before but also after epilepsy diagnosis, which reflects the continuation of epileptogenic neurobiological processes after the occurrence of the first seizures. Even though the outcome measures used to define antiepileptogenic effects would be clinical, alleviation or reversal of the pathology underlying either epileptogenesis or co-morbidities (or both) would provide additional proof to support the disease and syndrome modification.

Do Molecular Profiling Data Show the Roadmap to Antiepileptogenesis?

As the traditional approaches of using AEDs have shown their ineffectiveness in preventing epileptogenesis, novel approaches are needed (Pitkänen and Kubova, 2006; Temkin, 2009). One may ask, have the large scale molecular profiling studies provided any novel targets for antiepileptogenic strategies?

The first large scale molecular profiling studies addressing the molecular changes during epileptogenesis were published around 8 years ago (for review, see Lukasiuk and Pitkänen, 2006). These studies indicated that genes involved in cell death and survival, adhesion, neuronal plasticity, inflammation, or immune response are altered in SE models already at the acute post-injury phase. Analysis of post-injury molecular changes in TBI and ischemic stroke has revealed many similarities in the pattern of gene expression as compared to those after SE (Li et al., 2004; Mitsios et al., 2007). More recently, genes that regulate the transcription of other genes such as repressor element-1 silencing transcription factor (REST) have received attention, and it remains to be seen whether regulation of these "master switches" could beneficially modulate epileptogenesis over different etiologies without major adverse effects (Palm et al., 1998; Garriga-Canut et al., 2006; Spencer et al., 2006).

The strategy for the present literature review was based on the hypothesis that "normalizing the components of circuitry reorganization implicated by functions of genes which are most prominently altered during the post-insult phase would favorably modify the epileptogenic process". Consequently, the recently published treatment approaches were categorized into those that modulate immune response, inflammation, cellular adhesion, or cell proliferation and plasticity. Then it was asked whether the treatments showed any "disease or syndrome modification" when administered during epileptogenesis. Other approaches, including the effects of AEDs on development-related genetic epilepsies, and the unexpected results recently reported by using proconvulsants are also summarized.

Table 2 lists the outcome measures that were used to compare different treatments. The selection of outcome measures was based on the proposed terminology for "disease or syndrome modification" (Figure 2). Thus, to assess antiepileptogenic effect, we noted the effect of the treatment on prevention of epilepsy [effect on seizure threshold, duration of latency from the insult to the occurrence of unprovoked seizure(s), percentage of animals that develop epilepsy], seizure modification (seizure frequency, duration, and behavioral severity, progression of epilepsy), and whether remission of seizures or even cure was obtained. We separately list the effects on co-morbidities (e.g., learning and memory, emotional behavior, social behavior), and reversal of pathology as indicators of disease or syndrome modification.

As most of the studies have been conducted in SE models, we also pay attention to the method of induction of SE (electrical stimulation, pilocarpine, kainate) as recent studies suggest that, e.g., inflammation could play different role in different preparations (Marchi et al., 2009; Vezzani, 2009). Particular attention is paid on the administration scheme of the treatment relative to the initiation and duration of SE, and on the quantification of SE. As shown in previous studies, alleviation of SE can significantly modify the epileptogenic process, and therefore, exclusion of the direct effect of the treatment on the insult itself (i.e., duration of SE) is critical for the data interpretation. Finally, the duration and method of seizure detection (visual observation, video monitoring, video-EEG monitoring) was considered when justifying the data presented. The effects of various treatments on epileptogenesis are summarized in Table 2.

Immunosuppressants

Rapamycin

Rapamycin was originally discovered as a fungal agent in the soil, and is currently approved for clinical use as an immunosuppressant. Rapamycin inhibits the 290 kDa serine threonine protein kinase mTOR. mTOR responds to different extracellular and intracellular stimuli, including nutrient and energy status, growth factors, and stress, resulting in activation of cellular growth, proliferation, metabolism, and survival (for review, see Wong, 2010). In normal conditions, mTOR activity is inhibited by proteins hamartin and tuberin that are encoded by TISC1 and TISC2 genes. In tuberous sclerosis (TIC), an inactivating mutation in either TSC1 (chromosome 9q34) encoding hamartin or TSC2 (chromosome 16p13.3) encoding tuberin results in increased mTOR activity and therefore activation of its downstream pathways, which ultimately lead to increased cell growth and tumor formation (Napolioni et al., 2009). Epilepsy occurs in 80–90% of TSC patients and seizures often originate within or around hamartomous lesions or tubers, but the specific mechanisms of ictogenesis are unknown (Holmes and Staftsrom, 2007; Wong, 2008). Co-morbidities include subependymal giant cell astrocytomas, learning disabilities, and autism (Holmes and Staftsrom, 2007; Wong, 2008).

By taking advantage of the understanding of how to control the abnormality in signaling pathways leading to TSC, Zeng et al. (2008) investigated whether rapamycin treatment in a mouse model of TSC would prevent epileptogenesis. The Tsc1^GFAP CKO (conditional knock-out) mouse has astrocyte specific hamartin loss (Uhlmann et al., 2002). It results in increased brain weight, astroglial proliferation, hippocampal pyramidal cell dispersion, epilepsy that typically begins at 4 wk of age, and increased mortality. The first experiment investigated whether the suppression of pathologic changes dependent on the mTOR pathway by administration of rapamycin would prevent the development of epilepsy, and reverse the presumed epileptogenic pathologies in the hippocampus. Rapamycin treatment (3 mg/d) was started at P14 (about 2 wk before the onset of spontaneous seizures) and continued for 15 wk. The occurrence of seizures was monitored weekly with continuous video-EEG for 48 h/wk. Seizure frequency (seizures/48 h), frequency of interictal spikes (per min), histological analysis of tissue, body weight, and survival were used as outcome measures. Data analysis showed that in the vehicle group, the mean seizure frequency increased from 2 seizures/48 h at 5 wk of age to 14 seizures/48 h at 9 wk of age. Rapamycin treatment completely prevented the development of seizures in eight Tsc1^GFAP CKO mice included in the study. Also, the interictal spike frequency was similarly suppressed. It should be noted that epileptiform activity was assessed during rapamycin treatment. Even though unlikely, based on previous electrophysiological studies (Daoud et al., 2007; Ruegg et al., 2007), the direct effect of rapamycin on ictogenesis cannot be completely excluded.

Biochemical analysis showed that phosphorylation of S6 protein, one of the downstream effectors of mTOR activity, was reduced and levels of an astrocyte specific glutamate transporter, Glt1, were increased to normal. Histological analysis indicated that astrogliosis and pyramidal cell dispersion were reduced, brain weight was decreased, and survival of mice was increased in the rapamycin group as compared to the vehicle group. Thus, rapamycin treatment that was started before the onset of epilepsy suppressed both the development of ictal and interictal epileptiform activity and also partially reversed the biochemical and pathological changes believed to underlie epileptogenesis. The authors also did an experiment in which rapamycin treatment was started at 6 wk of age when mice were already expressing seizures. As compared to the treatment started before the onset of seizures, suppression of the mean number of seizures and interictal activity as well as reversal of the histopathological changes was less complete. Importantly, if rapamycin was discontinued, epilepsy reappeared.

Recently Zhou et al. (2009) investigated the effect of rapamycin on Pten (a phosphatase and tensin homolog deleted on chromosome 10) CKO mice, in which Pten loss is limited to postmitotic neurons in the cortex and hippocampus. Mice develop macrocephaly (enlargement of cortex and hippocampus) by 2 months of age as well as epilepsy, increased anxiety, and impaired social behavior reminiscent of human autism, which are apparent by 3 months of age. Interestingly, PTEN is a major negative regulator of the phosphatidylinositol-3 kinase (PI3k)/AKT pathway that contributes to phosphorylation of TSC2, and consequently, suppression of mTOR pathway. As a result, hypertrophic neurons in Pten CKO mice have increased mTOR activity (Ljungberg et al., 2009). In the study by Zhou et al. (2009), rapamycin (10 mg/kg, 4–6 wk treatment) was started either at the presymptomatic phase (5–6 wk) or at the symptomatic phase (10–12 wk). The treatment started at the presymptomatic phase largely reversed the histological abnormalities in the hippocampus and the cortex as well as abnormalities in anxiety and social behavior. The effect was less complete when rapamycin was started at the symptomatic phase. To assess the effect of rapamycin on epileptogenesis, EEG/EMG was recorded continuously for 3 d/wk over a period of 5 wk while mice received rapamycin treatment. Rapamycin did not affect the development of epilepsy. However, during the 3^rd and 4^th week of recording, the duration of seizures (seizures lasting >2 sec were counted) was reduced in the rapamycin group. There was also a tendency towards reduced seizure frequency. In another study, Ljungberg and colleagues (2009) administered rapamycin (10 mg/kg, i.p.) to Pten mutant mice during the 4^th–5^th wk of life. Video-EEG recordings after discontinuation of rapamycin (6^th wk) indicated that rapamycin treated mice spent only about 3% of the time showing repetitive spikes, repetitive spike runs, or seizures as compared to 38% of the time for the vehicle group. The effect was still present at 3 wk after discontinuation of treatment when the rapamycin group still spent only 3% of the time in epileptiform activity as compared to 47% for the vehicle group.

Rapamycin appears to suppress epileptogenesis and reverse presumed epileptogenic pathology in genetic epilepsies, in which the mTOR pathway is activated. A question arises, is the mTOR pathway involved in the aftermath of acquired epileptogenic etiologies, and would rapamycin treatment prevent acquired epilepsy? Previous studies show that pilocarpine or kainate-induced SE as well as lateral fluid-percussion induced traumatic brain injury (TBI) in rats can trigger induction of the mTOR pathway lasting for a few hours to several days (Chen et al., 2007; Buckmaster et al., 2009; Zeng et al., 2009) whereas cerebral stroke or ischemia suppresses it (Koh et al., 2008; Pastor et al., 2009) suggesting some etiology specificity. More recently, Zeng and colleagues (2009) investigated the effect of rapamycin on kainate-induced epileptogenesis in adult rats. Rapamycin was administered, 24 h after intraperitoneal kainate administration, at a dose of 6 mg/kg (i.p.) for 6 d and then every other day for up to 6 wk. Rats were continuously video-EEG monitored for the first wk post-SE and then for 48 h/wk. Rapamycin treatment prevented the progressive increase in seizure frequency and reduced the mean seizure frequency at 6 wk post-kainate from 6.0 to 0.7 seizures per 48 h. Whether the effect had remained after discontinuation of rapamycin remains to be studied. Mossy fiber sprouting was reduced but no effect was seen on neurodegeneration or neurogenesis. Thus, when assessed during medication, rapamycin reduced seizure frequency and had a partial effect on axonal plasticity (mossy fiber sprouting) typical of temporal lobe epilepsy. Interestingly, Buckmaster and colleagues (2009) recently showed that even though rapamycin reduced mossy fiber sprouting after pilocarpine-induced SE, sprouting was restored after discontinuation of rapamycin. Moreover, already established mossy fiber sprouting could not be reversed by rapamycin. In weight-drop induced TBI in Sabra mice, a single injection of rapamycin improved motor recovery when started at 4 h post-TBI at a dose of 0.5 or 1.0 mg/kg (Erlich et al., 2007). The treatment also reduced neurodegeneration and microglial/macrophage activation when assessed 3 d post-TBI. The effect on epileptogenesis was not assessed. However, this study suggest that rapamycin had no compromising effect on motor recovery.

There are some caveats in these elegant studies reporting positive effects of rapamycin on epileptiform interictal activity, seizure frequency, and pathology associated with epilepsy. Firstly, analyses of seizure occurrence have been largely made during rapamycin treatment, and consequently, the anti-ictogenic effect of rapamycin cannot be fully excluded. Also, the monitoring period was 48 h per wk which may predispose the data obtained to bias because of the cyclic pattern of spontaneous seizures (Goffin et al., 2008). The number of animals in the study groups has been relatively small which compromises the statistical power of any subgroup analysis. Finally, even though the effect of rapamycin treatment seems to continue after the discontinuation of treatment, the data available suggest that the effect is reversible. Further studies are warranted to optimize the administration protocol to balance the antiepileptogenic effects with adverse effects.

Taken together, data from the studies of rapamycin in genetic epilepsies with increased mTOR pathway signaling are promising and suggest that disease and syndrome modification can occur in mouse models of tuberous sclerosis and cortical dysplasia. Controlled preclinical studies with long-term video-EEG follow-up are needed to demonstrate the effects in acquired epilepsy models. It remains to be seen whether the experimental data translate to the clinic as some preliminary observations suggest (Muncy et al., 2009).

FK506

FK506 or Tacrolimus^® is an immunosuppressant that blocks the activation of T-cell response. It binds to the intracellular FK506-binding protein that is one of the immunophilins. The drug-immunophilin complex inhibits the activity of calcineurin, and consequently, the dephosphorylation of the nuclear factors of activated T cells (NFATs). Subsequently, T cells can no longer respond to activation by antigen-presenting cells, and no functional cytokine response occurs (for review, see Weischer et al., 2007). Reduction in microglial activation and cytokine expression has been proposed to underlie FK506-mediated neuroprotection in ischemia (Zawadzka and Kaminska, 2005).

Lukasiuk and colleagues (2008) investigated the effect of FK506 on SE-induced epileptogenesis that was triggered by electrical stimulation of the amygdala in adult rats. FK506 (0.5 mg/kg/d) was started 24 h post-SE and continued for 2 wk. Rats were video-EEG monitored 24/7 during the drug-treatment and for 4 wk thereafter. No positive effects were found on the percentage of animals that developed epilepsy, latency to the first seizure, seizure frequency, or seizure type.

Inflammation

Non-steroidal anti-inflammatory drugs (NSAIDs)

The idea to use NSAIDs in the prevention of epileptogenesis is based on their ability to inhibit cyclo-oxygenase 2 (COX-2). COX-2 inhibition reduces activation of prostanoid pathways (prostaglandin E₂ production). Consequently, microglial activation and leukocyte infiltration are reduced, resulting in suppressed cytokine release and oxidative stress, and reduced neurodegeneration (Weischer et al., 2007).

Celecoxib

The first NSAID tested in epileptogenesis models was celecobix. Jung et al. (2006) induced SE with Li-pilocarpine in adult rats, and started celecoxib (20 mg/kg, p.o.) 1 d after SE and continued the treatment for 42 d. During 28–42 d post-SE, rats were video-monitored. Only stage 4–5 seizures (i.e., secondarily generalized seizures) on Racine's scale (Racine, 1972) were scored. Further, monitoring was done while rats were still on medication. Eight of the 9 (89%) vehicle-treated rats developed epilepsy while 5 of 9 rats (56%) in the celecoxib group developed epilepsy (p>0.05). However, celecoxib treatment decreased the seizure frequency from 1.9 to 0.6 seizures/d, and reduced seizure duration from 14.9 sec to 7.1 sec (both p<0.01). In addition, celecoxib reduced hippocampal neurodegeneration and microglial activation and inhibited the generation of ectopic granule cells in the hilus and new glia in the CA1. The caveats of the study relate to the assessment of seizure activity during medication, lack of video-EEG monitoring, and no effort was made to monitor partial seizures.

Parecoxib

Parecoxib belongs to the second generation of selective COX-2 inhibitors with enhanced biochemical cyclo-oxygenase (COX)-2 selectivity, better analgesic efficacy, and reduced gastrointestinal toxicity as compared to celecoxib (Tacconelli et al., 2004). Bankstahl et al. (2009) conducted an experiment, in which they administered parecoxib 10 mg/kg (i.p., b.i.d.) for 18 d after pilocarpine-induced SE. Several weeks after SE, rats were video-EEG monitored to detect the occurrence of spontaneous seizures. No reduction in the occurrence of epilepsy was found in the parecoxib group as compared to controls.

SC58236

Gorter's group investigated the effect of a selective COX-2 inhibitor, SC58236 on SE-induced epileptogenesis that was triggered by electrical stimulation of the angular bundle (Holtman et al., 2009). SE was allowed to continue for 4 h, after which it was interrupted with isoflurane anesthesia. At this time point, SC58236 was started at 10 mg/kg/d (p.o., one injection/d) and continued for 7 d. Animals were continuously video-EEG monitored during the treatment and for up to 35 d post-SE. SC58236 treatment did not delay the latency to the occurrence of spontaneous seizures or the percentage of rats that developed epilepsy. If anything, there was a tendency to increased seizure frequency in the SC58236 group as compared to vehicle treated rats (7.0 vs. 3.8 seizures/last monitoring wk). Seizure duration was similar in both groups. Also, SC58236 had no effect on the severity of neurodegeneration, mossy fiber sprouting, or microglial activation even though it was shown to reduce the levels of prostaglandin E₂.

Prevention of leukocyte-endothelial cell adhesion with an α₄ integrin specific monoclonal antibody

Fabene and colleagues (2008) expanded the previous observation by Marchi et al. (2007) showing that administration of pilocarpine causes peripheral inflammation. In particular, it was shown that after pilocarpine-induced SE, integrin α4β1 and selectin ligand PSGL-1 are the mediators of leukocyte adhesion to endothelial cells in cerebral blood vessels. This was proposed to result in increased leukocyte extravasation, cerebral inflammatory response, leakage of blood-brain-barrier (BBB), impaired K⁺ buffering, and epileptogenesis. They hypothesized that by preventing leukocyte adhesion by using α4 integrin specific monoclonal antibody (α4 mAb) after SE would prevent epileptogenesis. To address this question they induced SE in C57BL/6 mice (Harlan). Two hours after pilocarpine injection, mice were treated with diazepam (3 mg/kg, i.p.). Administration of α4 mAb was started at 1 h after SE (i.p.), and continued every other day for 20 d. Mice were video or video-EEG monitored during 5–20 d post-SE. In the α4 mAb group, the latency to the appearance of spontaneous seizures was similar to that in the vehicle group. Also, the duration of seizures was not altered. However, the seizure frequency as assessed during α4 mAb therapy was reduced from about 0.8 to 0.2 seizures per day. According to authors, α4 mAb did not have any direct effect on the duration of SE, which was assessed by comparing the power spectrum in 60-sec epochs sampled at 90 min after SE onset. Mice treated with α4 mAb had less severe blood-brain-barrier damage at 18–24 h and reduced neurodegeneration when assessed 30 d post-SE. Also, their exploratory behavior was better preserved than in the vehicle group.

Cell proliferation and plasticity

Erythropoietin

Erythropoietin has multiple beneficial effects on the brain via activation of neurotrophic, anti-apoptotic, anti-oxidant, and anti-inflammatory signaling (see Siren et al., 2009). Chu and co-workers (2008) induced SE in rats with Li-pilocarpine (multiple dosing of pilocarpine 30 mg/kg) in adult rats. Animals received diazepam (10 mg/kg, i.p.) at 1 h after the beginning of SE. Erythropoietin treatment (5000 IU/kg/d, i.p.) was started "immediately after SE cessation" and continued for 7 d. Video-monitoring of seizures was performed at 35–42 d post-SE. In the erythropoietin group, 67% (6 of 9) rats developed epilepsy that was not different from that in the vehicle group (88%). However, the seizure frequency and duration as assessed by video monitoring were reduced in the erythropoietin group as compared to vehicle treated animals. Also, the histological outcome was improved in the erythropoietin group, including reductions in BBB damage, neurodegeneration, microglial activation, development of ectopic granule cells in the hilus, and gliosis. Caveats of the present study relate to lack of quantification of SE and to a likely underestimation of the number of seizures in a relatively short-lasting video-monitoring.

FGF-2 and BDNF gene duotherapy

The hypothesis that limiting tissue damage and enhancing repair by neurotrophins alleviates epileptogenesis was tested by Paradiso et al. (2009). They triggered SE with pilocarpine in adult rats and interrupted in at 2 h by administering 10 mg/kg (i.p.) of diazepam. Four days after SE, rats received a unilateral injection of vector expressing fibroblast growth factor-2 (FGF-2) and brain derived neurotrophic factor (BDNF). Animals were continuously video-EEG monitored for 20 d (i.e., during increased neurotrophin expression). The data show that FGF-2 and BDNF therapy reduced both the frequency and severity of spontaneous seizures. It should be noted that the authors controlled the possible effect of FGF-2 and BDNF therapy on ictogenesis in a separate group of animals, and the effect was found to be negligible. Histological analysis of tissue demonstrated normalized pattern of neurogenesis as well as preserved dendritic inhibition of granule cells by somatostatin neurons.

New use of old AEDs

Recent experimental data suggest that administration of AEDs before the appearance of genetic epilepsy can modify the disease process. So far, we have data from two compounds, levetiracetam and ethosuximide. Even though the mechanisms are unknown, some data from the ethosuximide trial suggest that chronic AED use at early age could reverse some ion channel abnormalities related to epilepsy in these animals.

Levetiracetam

Spontaneously epileptic rats (SER: zi/zi, tm/tm double mutant) develop air-puff induced tonic convulsions approximately at the age of 8 wk and absence seizures by about 12 wk. Yan et al. (2005) administered levetiracetam (80 mg/kg/d, i.p.) during weeks 5–9, that is during the time when the seizures have not yet started. They observed the air-puff induced tonic seizures 1–2 times per wk between weeks 5–13, and found reduced seizure frequency and duration in the levetiracetam group as compared to vehicle treated rats. Also, the number and duration of electrographically recorded absence seizures on weeks 12–13 was reduced in the levetiracetam treated rats.

Ethosuximide

WAG/Rij rats develop absence seizures approximately at the age of 3 months. To one group of rats, Blumenfeld et al. (2008) started the administration of ethosuximide (300 mg/kg/d) at postnatal day (P) 21 and continued it for 5 months. In another group, the rats received the drug from P21 to the age of 8 months. Animals were EEG monitored during the period of 5–8 months. As expected, rats on ethosuximide therapy had a very low seizure number. Importantly, also the rats in which ethosuximide was discontinued at the age of 5 months had reduced seizure frequency when assessed at the age of 5–8 months. The duration of remaining seizures was not altered as compared to the vehicle group. Unfortunately, it was not mentioned whether epilepsy had been completely prevented in any of the rats. The authors show that abnormalities in the Na_v1.1 and Na_v1.6 as well as in HCN1 channels as assessed by immunohistochemistry were normalized in the ethosuximide group.

Proconvulsants

A large number of preclinical and clinical studies have demonstrated that drugs designed to prevent epileptic seizures, i.e., AEDs, do not prevent epileptogenesis (Pitkänen and Kubova, 2006; Temkin, 2009). Recent data have provided surprising evidence that administration of proconvulsant drugs could have favorable effects on disease/syndrome modification after epileptogenic brain insults, including SE and TBI (Pitkänen et al., 2006; Echegoyen et al., 2009). Even though the compounds appear to have different mechanisms of actions, atipamezole being α₂-noradrenergic antagonis and SR141716A cannabinoid receptor 1 antagonist, it remains to be studied whether there is convergence in the molecular mechanisms or cellular location of the effects of these compounds. Also, whether these findings can be extrapolated to other recovery-enhancing treatments.

Atipamezole

Atipamezole is a selective α₂-adrenergic antagonist, with a high α₂/α₁ adrenoceptor selectivity ratio and does not display differential affinity for α₂-adrenoceptor subtypes. Administration of atipamezole enhances the kainate-induced seizures, indicating its proconvulsant activity (Halonen et al., 1995). Atipamezole rapidly penetrates the BBB and increases the release of noradrenaline. One of the net effects of α₂ blockage is the increased responsiveness of locus ceruleus noadrenergic neurons to stimulation (for review, see Pertovaara et al., 2005). Considering that α₂-receptors are located both in noradrenergic terminals (autoreceptors) as well as in dopaminergic and serotonergic terminals, the effects of atipamezole are likely not to be confined to the noradrenergic system. Jolkkonen and co-workers showed that ATI treatment increased motor recovery after medial cerebral artery stroke in rats as assessed by a limb-placing test (Puurunen et al., 2001), which is in line with NAergic hypothesis of recovery, stating that increased efferent output from the locus ceruleus (LC) will enhance recovery (Feeney et al., 1997). The effect was associated with normalization of [¹⁴C]deoxyglucose uptake (Barbelivien et al., 2002). These observations initiated our interest to investigate the effects of ATI on the outcome of SE.

We induced SE with electrical stimulation of the amygdala, and discontinued the SE activity by administering diazepam at 3 h after initiation of SE (Pitkänen et al., 2006). One week later, we started atipamezole treatment with subcutaneous osmotic minipumps for 9 wk (100 µg/kg/h). The first (24/7) video-EEG monitoring of spontaneous seizures was done during the 2 last wk of the treatment period. The second 2-wk video-EEG monitoring was done after discontinuation of the treatment. Atipamezole treatment had no effect on the percentage of rats that developed epilepsy. However, the seizure frequency was reduced from about 8.4 to 0.7 seizures/d (p<0.01). The duration of seizures was prolonged from 59 sec to 73 sec (p<0.05). Atipamezole-treated rats had milder hippocampal neurodegeneration and less intense mossy fiber sprouting than the vehicle-treated animals. This study was the first one to demonstrate that SE-induced epileptogenesis can be modified by pharmacotherapy.

Rimonabant

More recently, the Soltez group investigated the effect of a cannabinoid 1 receptor antagonist, SR141716A (Rimonabant^®) on epileptogenesis (Echegoyen et al., 2009). SR141716A is also a proconvulsant as it has been demonstrated to lower the threshold for kainate-induced seizures (Chen et al, 2007). They induced epileptogenesis by lateral fluid-percussion induced TBI, and administered SR141716A (1 mg/kg or 10 mg/kg, i.p.) as a single injection at 2 min post-injury. The threshold for kainate-induced seizures was assessed at 6 wk post-TBI. The reduction in the latency to kainate-induced seizures was prevented by SR141716A. Also, the total time spent in seizures after kainate administration was reduced in the SR141716A group as compared to the vehicle group. Importantly, no positive effect was found if SR141716A was administered 20 min post-TBI. A recent preliminary report suggests that if SR141716A (10 mg/kg, i.p.) is given after the first electrographic seizure during the kainate-induced SE, it had no effect on the percentage of rats that developed epilepsy or seizure frequency when assessed during the fist 10 wk post-SE (Pouliot et al., 2009). It is important to note that the post-TBI epileptiform activity in rats is not comparable to that after kainate injection (Pitkänen et al., 2009). It remains to be seen whether innate seizure activity in SE models compromises the effect of SR141716A.

How to Verify Antiepileptogenic Effect?

A critical challenge for any antiepileptogenesis study is: how to minimize the possibility of obtaining false negative data and obtaining meaningful true positive data? Minimizing false negative data lowers the likelihood of missing a favorable effect, which may just be present only in a subgroup of animals. Reducing the risk of false positive data requires vigorous analysis of both the study design and data analyses employed, and even re-investigation of the treatment in the same or another laboratory, or in another animal model. Defining significant positive data is also a challenge: which indicators of disease/syndrome modification obtained in the experimental laboratory should drive further studies in clinic?

Genetic Background– a Friend or Foe in Epileptogenesis Studies?

Candidate antiepileptogenic therapies have been tested in genetic models as well as in acquired models, in which epileptogenesis has been triggered by chemically or electrically induced SE or by TBI (see Table 1). Both rats and mice from different strains and vendors have been used.

Schauwecker (2002) pioneered the work demonstrating that various mouse strains have different sensitivities to kainate-induced neurodegeneration and epileptogenesis. Since that work, the genetic background of both experimental and control animals has become an issue. Further, it is important to consider not only the substrain but also the subline of mice used (Muller et al., 2009). Studies on SE models show that the B6 background is resistant to kainate or pilocarpine -induced epileptogenesis (Schauwecker, 2002; Mohajeri et al., 2004; Yang et al., 2005). Recent data suggest that also the risk of post-traumatic epileptogenesis can depend on the mouse genetic background, the CD1 mice being more likely to post-traumatic epilepsy (PTE) than B6 mice suggesting etiology-independent effect of genetic background on epileptogenesis (Bolkvadze et al., 2008; Hunt et al., 2009). Variability in the sensitivity to chemoconvulsants has also been demonstrated between the rat strains (Xu et al., 2004).

Even though the control of genetic background adds the cost and labor in epileptogenesis studies, it can also offer benefits. Understanding the effect of genetic background on epileptogenesis may offer an opportunity to screen the susceptibility genes for antiepileptogenesis (Kong et al., 2008). Also, choosing a genetic background that associates with a high percentage of animals developing epilepsy within an acceptable time frame, can be used to speed up epileptogenesis, for example after TBI, in which it is otherwise quite slow. Finally, the introduction of an epileptogenesis risk gene to a resistant background can offer a tool to investigate its mechanism on enhanced epileptogenesis.

Assessment of the severity of the initial epileptogenic insult

Studies from several laboratories have shown that the severity/duration of SE as well as the severity of TBI associates with the risk of epilepsy both in experimental models and humans (Pitkänen et al., 2009; Lowenstein, 2009). Therefore, verification of the similarity of epileptogenic insult between treatment groups is a major factor for data interpretation as the following examples in SE and TBI models show.

Status epilepticus

Most of the studies investigating the antiepileptogenic effects of treatments are done by using either chemically or electrically induced SE, assuming that SE is similar within and between the study groups (Table 1). We recently explored this question by re-analyzing the data from a study that was designed to investigate the effects of three conventional AEDs on epileptogenesis (Pitkänen A, Nissinen J; unpublished). We induced SE under video-EEG control in adult rats by electrical stimulation of the amygdala (Nissinen et al., 2000), and allowed the SE to continue for 4 h. Then, the animals were treated with diazepam (20 mg/kg, i.p.) to titrate the risk of epileptogenesis to about 50% of animals based on our previous observations (Pitkänen et al., 2005). This would allow us to detect both the reduction as well as increase in the rate of epileptogenesis. For example, as we had 47 rats in the vehicle group and 18 in the treatment group, it was calculated that we could have detected the antiepileptogenic effect if no more than 5/18 (28%) of the rats had developed epilepsy.

Administration of AEDs [carbamazepine (CBZ, 40mg/kg, t.i.d.), valproate (VPA, 200 mg/kg, t.i.d.), or levetiracetam (LEV, 100 mg/kg, t.i.d.)] was started at 24 h after the initiation of SE and continued for 7 d. The first 2-wk (24/7) video-EEG monitoring to detect spontaneous seizures was performed at weeks 11–12 post-SE, and the second at weeks 15–16 post-SE. The data showed that 62% (29 of 47) of rats in the vehicle group, 53% (9/17) in the CBZ group, 88% (14/16) in the VPA group, and 89% (16/18) in the LEV group (p=0.039 compared to the vehicle group) developed epilepsy during the follow-up period. Thus, the initial data analysis assumed that the severity of the initial insult (i.e., SE) was comparable in all groups as the initiation of AEDs was postponed well after the decay of SE activity (about 12 h, Pitkänen et al., 2005), and suggested that LEV increased epileptogenesis. For confirmation, we analyzed the number of spikes during the first 24 h after SE induction that we previously found to well describe the severity of SE in individual animals (Pitkänen et al., 2005). As shown in Figure 3, there was a remarkable variability in the number of spikes/24 h in all treatment groups. For example, in the vehicle group the spike number varied from 15,571 to 116,874 (mean 56,986 ± 20,122). The variation-% [(SD/mean) × 100%] was 35% in the vehicle, 41% in the CBZ, 23% in the VPA, and 38% in the LEV group. In the two treatment groups with the highest percentage of epilepsy (VPA and LEV), also the spike numbers during the first 24 h were the highest. Next we assessed the relationship between the spike number and epileptogensis in the vehicle group (n=47) by using logistic regression analysis. This indicated that every 10,000 spikes increases the risk of epilepsy by 1.5-fold. When the spike number was taken as a covariate in the analysis of epileptogenesis rate in different treatment groups, data showed that none of the treatments had any effect on epileptogenesis under the conditions used.

Figure 3.

Variability in the severity of status epilepticus (SE). (A) A dot-plot demonstrating the variability in the severity of SE that was induced by electrical stimulation of the amygdala. The severity of SE (y-axis) in individual animals randomized to different treatments is shown as the number of epileptiform spikes per 24 h after the initiation of electrographic SE. Note the remarkable variability within each treatment group (Veh, vehicle; CBZ, carbamazepine; VPA, valproate; LEV, levetiracetam). (B) A bar graph showing the percentage of rats that developed epilepsy (number of animals is in parenthesis). When the severity of SE (number of spikes/24 h) was taken as a covariate, there was no difference between the treatment groups.

The second caveat when analyzing data from SE models relates to the question whether the treatment modifies the "initial insult" (duration of SE) or whether it has a "true antiepileptogenic effect" by modulating the molecular events leading to epilepsy. We have shown that by reducing the spiking activity during SE, we can reduce the risk of epileptogenesis. In that experiment SE was triggered by electrical stimulation of the amygdala and it lasts for about 12 h. In the vehicle group 94% of rats developed epilepsy. Only 42% of rats in the group that was treated with DZP (20 mg/kg, i.p.) at 2 h and 71% treated at 3 h developed epilepsy (Pitkänen et al., 2005).

Typically in the SE models used epilepsy develops in >80% of animals. For practical purposes this is convenient because it reduces the number of animals needed to gain statistical power. However, this may also introduce a bias into the study. For example, unfavorable effects of treatments on epileptogenesis may remain undetected because almost all animals in the vehicle group develop epilepsy, and thus, there is little room to show worsening of the condition. Also, brain damage triggered by SE may be too severe for any treatment to repair. In many of the recent studies, the duration of SE has been reduced by administration of anticonvulsants (e.g., diazepan, phenobarbital). However, even in these cases it is important to electrographically verify the effect of treatment on the severity and duration of SE in each animal, and control its interactions with the antiepileptogenic therapy studied.

Finally, data are accumulating indicating that all SE models are not the same. For example, in the pilocarpine model, peripheral inflammation and lymphocyte-endothelial cell adhesion play a role (Marchi et al., 2007; Fabene et al., 2008; Marchi et al., 2009) suggesting that pilocarpine might present an epileptogenic insult whereby SE is associated with systemic inflammation (see also Vezzani, 2009). Whether peripheral inflammation plays a role in other chemically induced SE models or in those triggered by electrical stimulation of the brain remains to be studied.

Traumatic brain injury

Models of post-traumatic epilepsy for antiepileptogenesis studies are now available (for review, see Kharatishvili and Pitkänen, 2009). Like in humans, the risk of epileptogenesis in mice and rats depends on the severity of injury (Kharatishvili and Pitkänen, 2009). Post-traumatic epilepsy develops only in a subpopulation of mice and rats. Latency period seems to be longer and seizure frequency lower than that in SE models (Pitkänen et al., 2009). Therefore, labor intensity of the follow-up of epileptogenesis after TBI is significantly greater than that after SE. To minimize the labor in preclinical antiepileptogenesis trials after TBI, it would be important to assess the severity of brain injury at the early post-TBI phase, and select the animals with the highest risk of epileptogenesis into the study population.

Traditionally, the impact severity (expressed as atm) or the severity of cognitive and/or motor impairment during the first days post-TBI have been considered as markers of injury severity. To investigate whether MRI would provide a useful tool for assessing injury severity, we recently injured rats with moderate (2.05 atm, range 1.99–2.14) or severe (3.12 atm, range 3.06–3.20) lateral fluid-percussion TBI. We performed both the behavioral and MRI assessment at the early phase, and then, compared the data to the lesion size analyzed histologically (Kharatishvili et al., 2009). As the data show, the injury severity as determined by injury pressure did not give a reliable assessment of histologically assessed injury severity as there was a significant overlap between the moderate (cortical lesion size varied from 0.9–12.1 mm³) and severe (7.3–22.5 mm³) groups. Similarly, there was an overlap in motor (neuroscore) and cognitive (Morris water-maze) assessments. The best correlation was found between the quantitative T2 MRI at 3 d post-TBI and later histological damage. These data suggest that the use of MRI evaluation of injury severity at the early post-TBI phase might help to optimize the study groups for studies on antiepileptogenesis after TBI.

Seizure Monitoring – How, How Long, and How Many Animals?

In general, it is relatively easy to demonstrate that an animal has epilepsy, that is, it expresses spontaneous seizures. However, showing that the treatment prevented or suppressed epileptogenesis is more difficult because of variable length of the latency period, clustering of seizures, and variable seizure frequencies. Moreover, these may vary between the animals within the model as well as between the models. Also, we have to face the critical question: do we expect the treatment to work in every animal or should we look the possible effects only in a subpopulation of animals. Therefore, decisions on how to monitor (observation, video, video-EEG monitoring), for how long to monitor (several short epochs vs. continuous monitoring), and how many animals should be included in the study are all critically important for data interpretation.

Continuous long-term video-EEG monitoring studies have shown that the latency period from SE to the appearance of spontaneous seizures can vary significantly between the electrical and chemical models, being 3–5 wk in electrical models, <1 wk in the pilocarpine model, and 1–2 wk in the kainate model (Nissinen et al., 2000; Mazaratti et al., 2006; Goffin et al., 2007; Williams et al., 2009). Even though continuous (24/7) studies in TBI models are not available, observations suggest that in general the latency after TBI to epilepsy is longer than that in SE models (Kharatishvili et al., 2006; Hunt et al., 2009). Moreover, even within the model, the duration of latency can vary, for example, being 6–85 d in the SE model induced by amygdala stimulation to 5–17 d in the pilocarpine model (Nissinen et al., 2000; Goffin et al., 2007). Therefore, information about the characteristics of the model and interanimal variability is important for determining the duration of monitoring, as well as for calculation of the number of animals needed to discern whether the treatment, for example, delays the onset of epilepsy.

Another interesting aspect, particularly in SE models, is the gradual increase in seizure frequency during the first two months post-SE that was first reported in the amygdala stimulation model (Nissinen et al., 2000) and later shown also in the kainate model (Williams et al., 2009). This information is important for the timing of the monitoring because monitoring at the time when seizure frequency is still increasing might provide incorrect information about the treatment efficacy. On the other hand, the effect of treatments on the increasing part of seizure frequency curve, i.e., prevention of progression in the increase in seizure frequency, could perhaps be used as a marker of seizure-modifying effect.

Long-term monitoring of seizures has revealed clustering of seizures in both chemical and electrical SE models (Goffin et al. 2007; Pitkänen and Nissinen, unpublished; Williams et al., 2009) as well as in TBI models (Kharatishvili et al., 2006). For example, in the pilocarpine model the seizure clusters cycle at 5–8 d periods (Goffin et al., 2007). In a population of animals, it also appears that clusters are animal dependent. In practice this indicates a need for a monitoring time that covers the entire cycle (rather several) to reduce the sampling bias in seizure detection.

Also the seizure type and duration vary depending on the model used, as well as between the animals within the model. For example, in the amygdala stimulation SE model, the majority (80%) of all seizures were partial when assessed starting 11 wk after SE, that is, after stabilization of seizure frequency (Nissinen et al., 2000). Interestingly, partial seizures were the dominant seizure type in animals with frequent seizures. In the lateral fluid-percussion -induced PTE model, 78% of seizures were secondarily generalized, and thus more easily detectable (Kharatishvili et al., 2006). Monitoring the partial seizures, which sometimes may only be electrographic, presents a special challenge. Therefore, the assessment of seizures based on visual observation or video monitoring only may introduce a great bias for data analysis as the number of partial seizures may be greatly underestimated.

Finally, all of the above measurements require vigorous statistical analyses. There are two caveats for an analysis. First, can the negative or positive result occurring only in a subpopulation of animals be reliably detected based on the number of animals included in the study. Second, even if there is a reliable statistical difference between the treatment and placebo group, is the statistically significant result clinically significant. Further, is that parameter strong enough for proceeding from the preclinical to clinical trial. Performing power analysis before initiating the trial is valuable. This, however, requires knowledge of the characteristics of the model, in particular, about the interanimal variability. Unfortunately, achieving enough statistical power often requires a significantly higher number of animals per experiment than we normally use in preclinical studies (<15 animals per group). Currently a typical analysis includes the comparison of group means and largely ignores the likely outcome, that is, the drug works only in a subpopulation of subjects. Let's make a practical example. We plan to do a preclinical anti-epileptogenesis trial, to which we include 30 rats in the placebo group. From these, we expect 80% of rats (24 of 30) to develop epilepsy (typical to many trials using SE model). If we expect the treatment to reduce the percentage of rats that develop epilepsy to 50% of that in the vehicle group (i.e., 40% of rats on therapy develop epilepsy), we can ask how many rats we need in the treatment group. The answer is 25 animals if the power is 80%, and 63 rats if we wish to achieve 95% power (2-way khi-square, p<0.05). This suggests that preclinical trials with sufficient power require animal numbers that can be difficult to handle by one laboratory. One solution could be to conduct controlled preclinical multicenter trials. This would require standardization of the model, animal strain, vendor, and other practicalities. Whether the benefits of such studies would outweigh the practical, methodological, and financial challenges needs to be discussed. The question is, should we even think that the treatment works in every animal, or rather, should we try to identify those who respond, and why?

Seizure Threshold as an Outcome Measure

One could hypothesize that the appearance of the first seizure is preceded by (gradual) lowering of seizure threshold. This has been shown in the kindling model, in which the brain, however, is normal in the beginning (Pinel et al., 1976; Sutula, 2004). In acquired epilepsy models, in which the epileptogenic process is initiated in the injured brain, we do not have much data available confirming the hypothesis. Recent data show that seizure threshold in PTE models is lowered, even though no spontaneous seizures have been monitored (Kharatishvili et al., 2006; 2008; Bolkvadze et al., 2008). Moreover, rats with spontaneous seizures have the lowest seizure threshold (Kharatishvili et al., 2008). Thus, a question arises whether prevention of lowering the seizure threshold by therapy can be considered as a meaningful outcome measure in antiepileptogenesis studies.

Conclusions

We have a wide range of clinically relevant models available for studies on epileptogenesis. Further, the experimental data reported during recent years have elicited a great promise that modulating epileptogenesis is possible by multiple approaches. However, before translating the preclinical data to the clinic, a number of questions remain to be answered. For example, is the prevention of the lowering of seizure threshold a valid outcome measure in models, in which only a low proportion of animals develop spontaneous seizures? Is it enough to demonstrate that the treatment is effective in one model, or should it be done in different etiologies? What are the outcome measures in preclinical studies with the strongest implications to move to the clinic and eventually, labeling a compound as antiepileptogenic? What kind of adverse events can be tolerated during antiepileptogenic therapy, and for how long? Finally, are the markers chosen to measure treatment effects sensitive enough to highlight the full therapeutic potential of treatments and to avoid false negative results? Apparently we still face many challenges in tackling epileptogenesis. Overall though, clinically relevant models and the first glimpses of promising therapeutic approaches are available. There is hope on the horizon.