The Relevance of Kindling for Human Epilepsy


Address correspondence and reprint requests to Edward Bertram, Department of Neurology, University of Virginia, P.O. Box 800394, Charlottesville, VA 22908-0394, U.S.A. E-mail:


Summary:  Kindling is one of the most widely used models of seizures and epilepsy, and it has been used in its more than three decade history to provide many key insights into seizures and epilepsy. It remains a mainstay of epilepsy related research, but the question remains how the results from kindling experiments further our understanding of the underlying neurobiology of human epilepsy. In this article we compare the basic features of kindling and human epilepsy, especially human limbic or temporal lobe epilepsy. In this review we focus on a limited number of topics that may show areas in which kindling has been often cited as a tool for better understanding of human epilepsy. These areas include the underlying circuits, the importance of seizure spontaneity, the associated neuropathology, the contribution of genetics, seizure susceptibility, and the underlying pathophysiology of epilepsy. In the course of this article we will show that there are many features that kindling can teach us by direct comparison or implication about human temporal epilepsy. We will also see that not all findings associated with kindling may be applicable to the human condition. Ultimately we wish to encourage critical thinking about kindling and the similarities that it shares and does not share with the human epilepsy so the results from studies using this model are applied rationally to further our insights the mechanisms of human epilepsy.

This first paragraph is a disclaimer. One should be wary of articles and authors that start with disclaimers. What follows is a very personal view of kindling and what it can teach us about human epilepsy. The article starts philosophically from the perspective of human epilepsy and its neurobiology, based on the idea that we use models of human disorders to understand the mechanisms of the disorder that we cannot study directly in people. The logic then follows that to understand the human condition, the model ideally must have identical features. If the features are not identical, then, before one draws conclusions from the model, one must be very specific about how exactly the two states match and how they differ. Otherwise one risks the potential of drawing conclusions that may be inappropriate for the clinical situation. It is in this light that this article is written. It is highly selective in the issues discussed, and no doubt somewhat biased in its presentation. But it is based on some thought of what are some of the issues that are important in understanding human epilepsy, and then using those issues to compare reality to the model. The reader will no doubt notice the limited topics as well as the relatively sparse referencing, especially for a subject that has been researched as much as this one. There are no doubts some readers who will object significantly to some of the conclusions that are drawn, and “intense” discussions may ensue. Those objections are a good thing, because it is the ferment of scientific disagreement that often rescues us from our complacent acceptance of dogma and results in greater progress.

Kindling has been used as a model for human temporal lobe epilepsy and for epileptogenesis (the process of developing epilepsy) by many investigators for over 30 years, and the results have been frequently interpreted in terms of their implications for the human condition. In this review we will examine the phenomenon of kindling from a neurobiological perspective, what it is in the strictest sense, the kinds of studies that have been performed using the model and what narrow interpretations of the associated neurobiology might suggest for the human condition. We will also examine some aspects of human epilepsy especially as it may relate to the kindling phenomenon. At the end of this review we hope to determine where the model and the human condition have logical parallels and where they diverge. Through this exercise we would like to stimulate discussion that will allow for the appropriate use and interpretation of kindling as it relates to human epilepsy. The paper should not be viewed as a criticism of the model and the many neurobiological discoveries and observations that have come from it, but as an examination of it as a model of epilepsy in light of the growth of our knowledge and understanding of epilepsy and kindling over the years. The article should also not be viewed as a complete review of the area. Readers who desire reviews with a greater overview or more in-depth discussions are referred to some of a number of excellent reviews (McNamara et al., 1980; McNamara, 1994; Coulter et al., 2002; McIntyre et al., 2002; Morimoto et al., 2004)


Kindling is the process by which repeatedly induced seizures result in an increasing seizure duration and enhanced behavioral involvement of those induced seizures until a plateau is reached. Although a variety of approaches have been used over the years, kindling is usually carried out by focal electrical stimulation in the brain. The first stimulation produces a seizure that is quite short in duration and has only minimal behavioral accompaniment. With repeated stimulation, however, the durations lengthen and the behavior intensifies until these characteristics reach a plateau (Goddard et al., 1969; Racine, 1972). Two key features of kindling have to be kept in mind: there has to be a seizure for kindling to occur and the seizure induced is not spontaneous. The kindling phenomenon is long lasting so that once a plateau is achieved and maintained, there is little fall off in the kindled response even after periods with no stimulation (Goddard et al., 1969; Hiyoshi and Wada, 1992). The behavioral progression in the kindling process suggests that it starts with a limited number of neural circuits and subsequently recruits additional circuits as the behavioral component of the seizure evolves to convulsions. The increasing duration of the seizures suggests that the existing resistance of the circuits to seizure activity is also weakened. It is clear that kindling results in changes in the ability of the brain to limit seizures.

Over the years kindling has become a tool to examine the effects of multiple seizures on the brain and whether these effects can be modified. There are a number of reports that kindling can lead to the development of spontaneous seizures after enough stimulations, although this finding is not universal across laboratories or species. Investigators have examined the effect of multiple manipulations (pharmacological, counterstimulation, and lesioning) on the progression of kindling or on fully kindled seizures and have then drawn parallels between these observations and the potential applicability to certain aspects of the human condition, such as the prevention of epilepsy or the suppression of seizures. Predisposition to developing epilepsy has been equated with whether a strain kindles slowly or quickly. Kindling has also been used as a tool to examine the potential consequences of seizures on behavior, other types of seizures, and on neuroanatomy and neurophysiology. But are all of these observations relevant to human epilepsy? In this review we will examine some of these important observations in kindling and compare them to human epilepsy to see what observations are truly relevant and which may be important neurobiological findings that are more relevant to other processes.


Perhaps the most distinguishing feature of human epilepsy is the great diversity of causes and physiologies that are seen. Channelopathies, malformations, tumors, and brain injuries are among the many causes. EEG recordings of seizures have a similar diversity: spike and wave, multiple spike and wave, tonic and clonic patterns. Kindling has been most closely linked with mesial temporal lobe or limbic epilepsy for several reasons. First, and most importantly, the two (for kindling, when the stimulating electrode is placed in the limbic system) have initial seizure activity in the same brain regions such as the hippocampus and amygdala. Second, the seizure activity on EEG is similar, with early high frequency activity followed by a progression to clonic bursts before a terminal suppression of the EEG. For these reasons we will predominantly focus on the parallels between limbic kindling and human limbic epilepsy.

There are, however, several significant differences as well, the most important of which is seizure spontaneity. Although, following enough stimuli, the animal may begin to have spontaneous seizures; the finding is not universal for all kindled animals. Further, the kindling process relies on induced seizures. The second difference is the associated anatomical alterations in the limbic structures. There is a well-defined, albeit variable, pattern of neuronal loss and gliosis in the hippocampus, amygdala, and entorhinal cortex as well as other regions in limbic epilepsy that is not seen in kindling (Bertram et al., 1990; Bertram and Lothman, 1993; Mathern et al., 1997). Although there are a number of well-described anatomic changes associated with kindling (glial activation, decreased neuronal density, some mossy fiber sprouting), the changes are much less pronounced and widespread than seen in epilepsy (Cavazos and Sutula, 1990; Cavazos et al., 1994; Mathern et al., 1997; Adams et al. 1998). There are also reports of changes in the physiology of neurons in these regions in rats with limbic epilepsy that are not seen in kindled animals (Mangan et al., 2000; Bertram et al., 2001), although changes have been reported in kindling as well (Mody, 1999). These observations suggest that there are fundamental differences in the two conditions. Still there are parallels (circuits, responsivity to drugs) that encourage a comparison of the similarities.


Seizures beget seizures has been the saying, but is that always true? This concept has been discussed for more than a century, and it has come to mean that each seizure experienced increases the risk for a subsequent seizure. Certainly the clinical observation has been that each seizure experienced increases the probability for a subsequent one, but there are two possible interpretations with regard to the possible mechanisms. The first is that recurrent seizures indicate an underlying process that causes seizures, and that the seizures are just a symptom of the underlying process that is slowly progressing. The second interpretation is that each seizure lowers the threshold for a subsequent seizure, that is, the seizures themselves create the conditions for the next seizure. Although this second interpretation has been often cited as a process of epileptogenesis and thus kindling, it really does not fit the kindling phenomenon, which is not spontaneous. In kindling, the threshold for inducing the seizure can drop over some stimuli until a floor is reached, but the stimulus remains necessary for each seizure. The reports that the kindling can result in spontaneous seizures if enough stimuli are given suggest that kindling can cause epilepsy, but the observation is not universal to all species and laboratories, and the process still relies on many induced seizures to happen.

Against this concept that epilepsy is a progressive disorder that perpetuates its existence with each new seizure is the observation that there are many epilepsy syndromes that are self-limited, independent of the number of seizures an individual may experience over the months or years that the seizures are active. This occurrence is seen most often in many of the childhood epilepsies such as the absence syndromes or the benign partial epilepsies such as Rolandic seizures (Loiseau et al., 1983). In these two situations, in spite of numerous seizures, the condition usually comes to an end. The natural history of chronic limbic epilepsy (CLE) is not completely known but much evidence suggests that it has an overall less benign course (Glaser, 1987). It has often been considered a permanent condition that is frequently resistant to medications, but the reality is that no one really knows what the natural history is for all patients with the syndrome, as only those who prove intractable to medications ever receive the full evaluation that documents the underlying cause of the seizures, and truly only for that small number of patients that undergo surgery to control the seizures. It is possible that many of the patients with CLE have seizures that are readily controlled and in fact may resolve with time.

Patient histories do suggest that a kindling like process takes place (also in rats with spontaneous limbic epilepsy following an episode of limbic status epilepticus) (Cavalheiro et al., 1991; Bertram and Cornett, 1994; Gorter et al., 2001). Many patients report a prolonged history of “feelings” that later get recognized as part of a seizure when the patients develop more obvious complex partial seizures with alterations in consciousness or awareness (Glaser, 1987). Similarly patients that have had stable partial complex seizures for many years seek further medical help when they start to have more involved motor manifestations that lead to injuries. Although it is impossible to be certain that these events represent a clinical/behavioral progression of seizures arising from the same focus, the resolution of the symptoms following surgery in these patients supports the hypothesis. Why does this observation support the idea that a kindling process occurs in humans? Because over time, with repeated spontaneous events, the seizures become longer and clinically more complicated, similar to the prolongation of the afterdischarge and increasing behavioral seizure score in animals undergoing kindling. It does not, however, tell us how spontaneous seizures develop from a nonepileptic state or how each seizure starts. This issue gets to the heart of the definition of “epileptogenesis” (see Fig. 1). Is it the process of becoming epileptic, a series of events that leads to the first spontaneous, nonprovoked seizure from a previous nonepileptic state, or is epileptogenesis the process by which seizures slowly increase in duration and recruit additional brain circuits with a subsequent more complex/severe clinical seizure? If it is the latter, then kindling is an excellent model of epileptogenesis. If the former, then kindling is probably not an appropriate model because spontaneous seizures, if they do eventually arise, do so only after many induced seizures, something that would be quite unusual in people. The closest parallel to repeated induced seizures is electroconvulsive therapy (ECT) used to treat severe depression. In this group of patients the incidence of epilepsy is extremely low, and there is a report of a patient who underwent over 1,200 separate ECT treatments and never developed spontaneous seizures (Lippman et al., 1985). Although reports exist of patients who have had ECT who have unprovoked seizures, many of these reports are complicated by the medications the patients may have received that are known to cause seizures (Logothetis, 1967; Dallos and Heathfield, 1969). Still, kindling to spontaneous seizures could be circuit specific, and ECT could be stimulating seizures in circuits less susceptible to kindling to spontaneous events, an issue that will be difficult to resolve in people.

Figure 1.

The term epileptogenesis has had variable meanings, and kindling has often been referred to as a model of epileptogenesis. The process of kindling is compared to two of the most common concepts of epileptogenesis as applied to epilepsy. For epileptogenesis (A) describes in broad strokes the primary definition of epileptogenesis: the series of events that take place in the latent period from the appearance of the causative abnormality (genetic, maldevelopment, injury) and the first spontaneous seizure. In other words, the process of becoming epileptic. (B) outlines the process of seizure strengthening of the seizures once the first spontaneous seizure has occurred, a process of seizure enhancement and spread. Kindling, as used by the majority of investigators has more parallels with B than A.


What kindling research has shown is that the limbic circuits, with some variability among the stimulus sites, are highly susceptible to seizure induction and spread as it is relatively easy to induce seizures in these regions, and seizures rapidly spread to other regions/brain circuits. It is possibly in the circuitry of limbic epilepsy that kindling has the greatest parallels with the human condition as, with a few exceptions, the wiring of the brains of the two species is quite similar, at least in terms of the general organization of major pathways among different regions. The circuitry of seizures can be thought of as having two broad categories: (a) the initiating circuit(s) and (b) the pathways of seizure spread. The first is the minimal region in which a seizure can start and be sustained. The second represents the process by which additional brain circuits are recruited as the seizure continues and spreads. Kindling is useful for learning about circuits in that seizures can be induced by a highly focal stimulation, and the pattern of initiation and spread throughout the brain can be readily examined.

When the observations made during the kindling studies are interpreted with regard to the known anatomic pathways and connections, hypotheses can be tested about the flow of seizure activity from the site of stimulation as well as about the role different structures play. Two hypotheses for the seizure initiation circuitry have evolved as a result of the known anatomy. One hypothesis has involved the trisynaptic circuit of the temporal lobe involving the entorhinal cortex, the dentate gyrus and Ammon's horn of the hippocampus. In this hypothesized circuit the seizure starts and reverberates in the loop until it breaks out and spreads to the rest of the brain by recruiting adjacent regions of cortex. This proposed circuitry has designated the dentate gyrus as a “gatekeeper” in which it resists recruitment into the seizure, but eventually the resistance is broken down, allowing the seizure to proceed and spread (Lothman et al., 1991). The anatomic substrate to this proposed circuit exists, and a number of authors have reported a “dentate seizure resistance.” In the slice preparation a number of investigators have shown that induced epileptiform activity can reverberate around the circuit and this regional interaction can be interrupted by physically cutting the pathway (Walther et al., 1986; Chesnut and Swann, 1988; Dreier and Heinemann, 1991; Swann et al., 1993). However, these hypotheses come into conflict with other observations. This presumed circuit for seizure initiation is based on activity reverberating around the trisynaptic pathway from the entorhinal cortex, through Ammon's horn and back to the entorhinal cortex. This scenario implies that seizures require an intact circuit and that any interruption in the circuit will prevent the onset of seizures. However, experience with surgery for human epilepsy has shown that the removal of one or more of these regions, an intervention that interrupts the trisynaptic circuit at one or more points, sometimes has no impact on seizure occurrence. Although this observation could have several interpretations, it does raise the possibility that the critical circuitry for seizure initiation may be different than the proposed hippocampal trisynaptic pathway, or at least that this circuit may be only one of several local circuits that initiate seizures independently of one another. This possibility that the hippocampal trisynaptic pathway may not be critical also lies in the observation that other sites that lie outside this circuit may be easier to kindle from (Goddard et al., 1969).

Another hypothesized circuitry of limbic epilepsy/seizures is based on the existence of multiple independent sites that have connections to a common central site that can organize and spread the seizures. The major limbic sites associated with a good kindling predisposition have reciprocal connections with the midline thalamic nuclei, including the medial dorsal and reuniens, which in turn have projections to many neocortical sites. The existence of these thalamolimbic circuits could fit the observations associated with kindling of the different sites (amygdala, piriform cortex, claustrum among others) (Goddard et al., 1969; Piredda and Gale, 1985; Zhang et al., 2001), all of which have only weak connections, if any, to the trisynaptic hippocampal pathways. The spread of seizures from these “other” sites as the seizures intensify behaviorally have had several hypotheses. The first suggests that the seizures spread to the neocortex through the perirhinal cortex, by spread to physically adjacent sites (Kelly and McIntyre, 1996; McIntyre and Kelly, 2000). The other hypothesized pattern for spread involves the midline nuclei of the thalamus which have multiple reciprocal connections to the sites of stimuli as well as to a number of sites in the neocortex (Bertram et al., 2001).

Kindling offers the tools to evaluate the initiating circuitry as it allows a precise selection of the site of stimulation and seizure initiation. Although using kindling to examine epilepsy circuitry may be criticized because of the lack of seizure spontaneity seen in epilepsy, it does allow for the exploration of potential seizure circuits and the role different regions may play in the beginnings and spread of a seizure. Several key points have been made over the years. First there are a number of sites that allow for a more rapid kindling than direct stimulation of the hippocampus (Goddard et al., 1969). Second, stimulation of several subcortical sites such as the claustrum and the medial dorsal nucleus offer very rapid evolution to kindled motor seizures, suggesting that these sites may be key to the process of seizure generalization (Zhang et al., 2001; Bertram and Williamson, 2005). Finally, kindling seizures in the hippocampus or amygdala results in rapid involvement of the medial dorsal nucleus, suggesting strong connections between the regions (Bertram et al., 2001). Overall, the findings strongly suggest that there are multiple potential independent circuits for initiating seizures that may overlap one another.


Because kindling is viewed as a chronic animal model of limbic epilepsy and, at least by one of its definitions, as a model for epileptogenesis, there has been active research examining the underlying changes in anatomy and physiology that accompany kindling. Although some of the animals stimulated under particular protocols, if kindled long enough, eventually develop spontaneous seizures, many do not. For this reason, the study of tissue from kindled animals, especially those animals that do not develop spontaneous seizures, may be viewed as an examination of the consequences of a number of seizures rather than changes that lead to chronic epilepsy. This issue was not really recognized until the appearance of rodent models of chronic limbic epilepsy that followed an episode of status epilepticus. In comparison to kindled animals, this group of rats had many of the features that more closely paralleled human CLE (spontaneous seizures with similar EEG patterns, neuronal loss in the limbic system, gliosis). There have been few studies comparing kindled and epileptic animals for changes in anatomy and physiology, but those that have, have revealed that there are significant differences. Anatomically, the rats with CLE have evidence for significant (although variable from rat to rat) neuronal loss throughout the limbic system and related subcortical sites such as the midline thalamic nuclei (Bertram et al., 1990; Cavalheiro et al., 1991; Bertram and Lothman, 1993; Du et al., 1993a; Mathern et al., 1997; Bertram and Scott, 2000; Schwarcz et al., 2002). In addition, there is significant sprouting of the mossy fibers in the dentate hilus as well as widespread gliosis and glial activation (Tauck and Nadler, 1985; Sutula et al., 1989; Cavazos et al., 1991; Mathern et al., 1997). Neuronal loss as a consequence of kindling is a controversial topic, with some studies suggesting that kindled seizures do result in neuronal loss, whereas others do not (Cavazos and Sutula, 1990; Cavazos et al., 1994; Mathern et al., 1997). However, when the tissue from the epileptic animals is compared to the kindled animals, the epileptic rats have a far more severe and widespread neuronal loss that is more comparable to the human condition. In addition, the mossy fiber sprouting in the two models is also significantly different with the epileptic rats showing a far greater intensity of Timm's staining. Both groups show an activation of glial cells, an observation that suggests that at least some of the changes seen in glia in the two models could be the consequence of seizures (Steward et al., 1991; Steward et al., 1992; Du et al., 1993b; Adams et al., 1998) (Table 1).

Table 1. Influence of factors on the rate of kindling and the devlopment of spontaneous seizures
FactorSpontaneous seizuresKindling rate
Rat strainUnknownYes
Presence of other seizure type/predispositionYesYes
 Effect varies with type
Stimulation parametersUnknownYes
Stimulation siteUnknownYes

Differences have also been demonstrated between kindled and epileptic rats in studies examining the physiological changes at the system or cellular level. At the systems level, when examining evoked potentials in such regions as the dentate gyrus or CA1, it is clear that changes have occurred in two groups in relation to controls, however the differences between the kindled and epileptic animals are also quite different (Stringer and Lothman, 1988; Lothman et al., 1990; Mangan et al., 2000; Bertram et al., 2001). When examining neuronal physiology at the cellular level, these differences also become quite clear. The neurons from epileptic animals have a more “hyperexcitatory” response, often with prolonged EPSPs, shortened IPSPs, and multiple action potentials produced from a single stimulation (Rempe et al., 1997; Mangan et al., 2000; Bertram et al., 2001). Changes have been reported in receptors and channels in the epileptic animals. The kindled animals, although changes have been described, often differ little from control rats (Mathern et al., 1997). These observations taken together strongly suggest that there are a number of features of epilepsy that neurons and neuronal circuits from kindled animals do not have (Table 2).

Table 2. Comparison of limbic kindling and limbic epilepsy
  1.  *Although categorized similarly, most reports indicate that the alterations in kindling and limbic epilepsy are quite different.

Spontaneous seizuresAfter lesion inductionFollowing many stimulated seizures in some strains
Kindling-like processYesYes
PathologyExtensive neuronal lossControversial, from little to significant
Mossy fiber sproutingModerate to heavySlight to moderate
Glial activationYesYes
Synaptic reorganizationLikely extensiveModerate mossy fiber sprouting
Limbic seizure initiationDiffuse/mutifocalDefined focus


The ability to kindle some animals to the point of spontaneous seizures raises a number of interesting and important questions about how epilepsy arises in humans, a true question of epileptogenesis. However, there are several key points to remember. First, with the exception of repeated ECT, recurrent induced seizures in a normal brain almost never occur in people. Second, there is growing evidence that susceptibility to seizures and kindling in any animal may be linked to a specific seizure type and circuit, as there is growing evidence that tendency to develop one type of seizure does not translate to a tendency to develop other types. In fact there is evidence that susceptibility to some types may reduce the tendency to develop other types of seizures (Table 1).

The literature regarding the development of spontaneous seizures through kindling is quite varied, but, when the issue of spontaneous seizures is raised, the literature likely has a disproportionate share of positive experiences, as the reports of failure to achieve spontaneous seizures are less likely to see the light of day, or at least publication. Reports of spontaneous seizures following kindling have involved rats, cats and various primate species. In addition there is the further factor of age at the beginning of the kindling process and the kindling protocol itself (daily, multiple times per day, multiple stimulations on alternate days). In most reports it is clear that many stimuli most be given before the spontaneous seizures appear. The first reports of the phenomenon in rats indicated that over 300 stimulations were usually required in adult rats before the first spontaneous seizures were seen (Pinel and Rovner, 1978). Wada (Wada and Osawa, 1976; Wada et al., 1978) has reported variable results in different primate species, mostly baboons (Papio sp.) and macaques (Macaca sp.). In general the baboons kindled more readily, and, during the period when they were undergoing stimulation, convulsive and nonconvulsive spontaneous seizures were seen in some, but not all, animals. The interpretation of these findings is complicated by the observation that many of these baboons were drawn from a population with photosensitivity and observed convulsions in individuals in the absence of any kindling. In contrast the macaques showed a far greater resistance to the kindling process, never becoming “fully kindled” with convulsive seizures even after several hundred stimuli, and spontaneous seizures were also not observed. These findings suggest that a variety of factors contribute to how far seizures will spread in the brain and whether spontaneous seizures will ever develop.

Kindling in cats has suggested that the development of spontaneous seizures is linked to the age of the animal at the time when kindling is initiated. Shouse and her colleagues (1990, 1992) have demonstrated that when prepubescent cats (2–5 months old) are given daily kindling stimuli about half of the animals begin to have spontaneous seizures. This observation is significantly different from kindling in adult cats, in which less than 10% develop spontaneous seizures, and then only after many more stimuli. Whether this observation of an age relation to the development of spontaneous seizures is true in other species is less clear. In one study examining the progress of kindling in rat pups, several rats were observed by chance to have spontaneous seizures, but there was no systematic evaluation for this outcome (Haas et al., 1992). That adult rats can develop spontaneous seizures after many stimulations has been reported by several authors (Pinel and Rovner, 1978; Michalakis et al. 1998; Sayin et al., 2003), but it has not been a universal observation by all laboratories (by informal polling). Our own experience is that spontaneous seizures in adult kindled rats is a very infrequent phenomenon even after many hundreds of stimulated seizures.

Why is there such a difference in the reported (and nonreported) experiences across laboratories? There is no definitive answer, but some hypotheses can be derived from the experiences with other species in which there are clear genetic (species) influences on the development of spontaneous seizures, suggesting that there may be strain variations. In addition the protocols that were used to stimulate were varied. In the reports of the animals with spontaneous seizures the stimuli were given daily or several times daily, while we give multiple stimuli (3–8) in a day, but stimulate no more than three days per week with a day between stimulation sessions and only rarely see spontaneous seizures. There are reports that suggest that stimulation frequency can influence the rate and nature of kindling, and perhaps it influences the development of spontaneous seizures as well (Lothman and Williamson, 1993). Only a lengthy study comparing strains and stimulus protocols will be able to answer that question.

What is the relevance of kindling to spontaneous seizures to human epilepsy? The short answer is that its unknown. In people the seizures are already spontaneous by the time they become clinically apparent and, as mentioned earlier, with the exception of ECT, people tend not to be exposed to repeated seizure inducing stimuli. Is it possible that the repeated seizures from one focus induce an independent site of seizure onset, a so-called secondary focus? Although there are reports in the literature about such foci, they are impossible to differentiate from two areas of abnormality in the brain that matured at different rates, with one expressing spontaneous seizures some time before the second site (Morrell, 1989). That individuals have more than one focus for seizure onset is well documented, especially when malformations and trauma are involved. Individuals with two seizure types arising from separate areas of the brain have been recorded in epilepsy monitoring units. Much of the clinical support for kindling of a second site has come through observing the differences in the localization of interictal discharges over a period of several years in which the earlier EEGs recorded spikes from one primary focus. In subsequent EEGs spikes have been seen in homologous contralateral regions, first synchronized with the primary spike and then occurring independently. The appearance and gradual independence of the interictal spikes were equated with the development of another focus. The relation of spikes to the focus has been in discussion for many years, but the general consensus is that they are not always linked to the seizure focus, but can appear at sites distant to the site of seizure onset (Goldensohn, 1984). In short, there is no strong evidence for the induction of secondary sites in people through a kindling like process.

The issue of the relevance of kindling and consequent spontaneous seizures over time to the development of epilepsy in humans is further complicated by the observations from these animal studies that the success of kindling overall and the possible development of spontaneous seizures as a result is heavily influenced by the species and the age of the animal at the time of kindling. Clearly, induced seizures can eventually result in spontaneous seizures in some species under certain circumstances. Is it possible that the “seizures beget seizures” play a role in patients that begin with the spontaneous seizures of epilepsy? Perhaps over time, the repetitive seizures do alter the threshold for a seizure to occur, so that the seizures start more easily. However, there are patients with the benign epilepsies whose seizures stop regardless of the numbers of seizures that they have experienced (Loiseau et al., 1983). Similarly there are patients with stable seizure patterns over a lifetime. It still may be possible that repetitive seizures over time induce an independent focus, but if so, it will be in a subset of patients with a predisposition in a particular circuit, the seizures will have a particular physiology that “kindles” and the seizures will happen in the circuits that are more likely to support a kindling phenomenon. Given the complexities of the causes, physiologies and genetics associated with human epilepsy it will likely be extremely difficult to prove this scenario unequivocally. However, if one considers kindling by its original definition of progressive increases in seizure duration and severity, there is ample anecdotal evidence (there will never be direct evidence) that some people with some types of partial epilepsy do undergo a slow intensification of their spontaneous seizures, suggesting that seizure activity, under the right circumstances, does kindle stronger seizures.

Kindling antagonism is another phenomenon that has been observed in kindling experiments from very early in the history of the model that is almost a counterpoint to kindling to spontaneous seizures. Briefly kindling antagonism is seen when kindling from two alternating sites. It results in the development of fully developed kindled seizures from one site but retards and at times suppresses the development of kindled seizures from the other (Burchfiel et al., 1986). Electrographic seizures occur at both sites, but the seizures only recruit other circuits to full expression from one site. Since its first descriptions over 20 years ago this phenomenon has been examined for the potential underlying mechanisms as the observation is almost counterintuitive, that seizures from one site would suppress seizures from another. Some factors have become clear: norepinephrine can play a role in modulating antagonism, that antagonism is protocol and site specific and that it may not occur in all species (Applegate et al., 1986; Teskey et al., 1999; Kogure et al., 2000; Matsuda et al., 2003). Does the observation have relevance for human epilepsy? It is not clear, but there are two situations in which a kindling antagonism like phenomenon may play a role. In epilepsy surgery there has been an observation that sometimes a whole new type of seizure appears after the original seizures were stopped by the removal of the one focus. Several hypotheses have been raised about the basis for this occurrence. One is that the seizure focus remains, at least in part, but is now following an alternative route of spread. A second possibility is that there was a second focus that was suppressed by the activity of the primary and now that the primary has been removed, the antagonism is no longer present and the second site is free to express itself. Proof for these hypotheses will be difficult, but they do give us ideas about the modulation of seizure foci.

Another area in which the mechanisms of kindling antagonism may play a role is in the various brain stimulation techniques (vagal nerve stimulation, deep brain stimulation) that are used to suppress spontaneous seizures. The stimulation protocols are quite different from traditional kindling antagonism methods, in that seizures are not elicited by the stimulation and that the stimuli are given multiple times a day (often every 5 min). The mechanisms behind the success of these techniques is unclear, but the techniques can be effective in significantly reducing seizure frequency in some patients. Perhaps the as yet not understood mechanisms of kindling antagonism may offer an explanation for the as yet not understood mechanisms behind the success of brain stimulation in suppressing epileptic seizures.


Early in the course of kindling history it was noted that some sites were more likely to kindle or kindle more quickly than others (Goddard et al., 1969). This observation led to the obvious conclusion that some brain circuits were more prone to support and sustain seizure activity than others. It was then observed that some animals kindled from a particular site more quickly than other conspecific brethren, and that this trait could be bred for (fast and slow kindlers) (Racine et al., 1999). The ability to create separate strains of fast and slow kindling animals coupled with the results from kindling in various primate species indicated that the predisposition to kindling was under significant genetic influence. Population studies in humans have suggested something similar. Although many forms of epilepsy appear sporadic, with only one member of the family affected, there are many extended families in which epilepsy appears to “run in the family.” A clear inheritance pattern in these families is not obvious, but the number of family members with a seizure disorder exceeds what would be expected from a purely random occurrence. As a result there has been much discussion about seizure susceptibility genes or polygenic inheritance of certain forms of epilepsy. Can kindling and such strains such as the fast and slow kindlers help sort out the potential genetics of seizure susceptibility? Perhaps, in some cases, but there are a number of recent observations that suggest that the issue may be far more complex and that susceptibility may be linked to specific types of epilepsy.

Over the years there have been a number of rat strains that have been reported to have spontaneous seizures or a genetic predisposition to reflex seizures, usually induced by loud noises (genetically epilepsy prone rats: GEPRs) (Coffey et al. 1996). The spontaneous seizures that have been found in other animals (such as the GAERS rat) (Onat et al., 2005) are most closely linked to the human absence syndromes with generalized spike wave patterns. The GEPR, kindling and absence models each have a seizure pattern that involves specific circuits and mechanisms. For these reasons, they offer the opportunity to study the question of whether predisposition to one form of epilepsy facilitates the development of other types of seizures in other circuits. The studies to date suggest that the answer is sometimes yes and sometimes no depending on the seizure type. In the tested absence model the susceptibility to kindling is significantly reduced with an inability to achieve a traditional fully kindled motor seizures (Onat et al., 2005). The GEPR rats, on the other hand, which may share some brain circuits with limbic kindled seizures, kindle more rapidly than standard rats, and there is further evidence that the experience of audiogenic seizures further hastens the kindling process (Coffey et al., 1996). These observations suggest that seizure susceptibility may be seizure type, or in some cases a limited number of seizure types, specific and not a universal predisposition to any type of epilepsy.

The fast and slow kindling strains can be kindled, but the number of stimulations required to achieve a similar level of kindling is significantly greater in the slow strain than the fast. Although the phenomenon was first seen with traditional kindling of the amygdala, the strain differences are maintained for other sites of kindling, although the differences in kindling rates between the two strains is not uniform across all sites. The implications of the fast and slow kindling strains within the kindling model are similar: the propensity of a particular circuit or series of interconnected circuits to support seizures is under some degree of genetic control. The question obviously arises: Do the differences in these strains apply only to the kindling phenomenon or to other forms of seizures as well? Clear differences have been seen when rapid kindling (stimuli every five minutes over three hours) as well as when status epilepticus has been induced by kainic acid and several GABAergic antagonists. The kainic acid, although given systemically, still induces much of the seizure activity through the limbic system. The GABAergic antagonists may work more globally, although the apparent difference in the distribution of different receptor subtypes could alter the relative sensitivity of one strain or a particular region in one strain to the effects of a particular drug, making the limbic system in one strain more sensitive to a particular drug than in the other (Gilby et al., 2005). The observation in these strains confirms the hypothesis that the tendency to support seizures in a particular circuit is under some degree of genetic control, and may not transfer to other circuits or seizure types.


Kindling has provided us with a wealth of information about seizure propensity, the effects of repeated seizures in the brain, the roles of different structures in regulating seizure duration and spread, the relative susceptibility of different regions to support seizures as well as the complex genetics of seizure proneness. In addition kindling has given us insight into factors that may enhance or diminish the excitability of the limbic circuits (endogenous seizure control mechanisms). It has also been very useful for screening potential new treatments for partial epilepsy. These are extremely useful observations and applications. But, is kindling a model of chronic temporal lobe epilepsy? There will be, no doubt, great debate on this subject, but because the seizures are induced in normal brain as opposed to arising spontaneously in abnormal, at times grossly abnormal, brain, there are many things that kindling cannot tell us. These issues include what leads to the occurrence of the first spontaneous seizure in the face of the pathological changes, as opposed to how many induced seizures are necessary for a spontaneous seizure to occur after kindling, something that is likely very rare in human epilepsy.

How should we use kindling to enhance our understanding of seizures and epilepsy? Because kindling allows for the precise timing of seizure onset, the intensity of the inducing current and the site of seizure induction, it provides an ideal opportunity for identifying key points in the circuits for seizure control, the interactions of different regions in supporting seizures and controlling their spread as well as how to identify regions most likely to induce seizures. The model can be used to identify the consequences of repeated seizures on the brain, but several caveats are appropriate. First, the effect of any seizure may be circuit specific, so that stimulation in one site, may not predict the consequences for other sites. In addition, because kindling is carried out in normal brains, the consequences of seizures in normal animals may not be the same as the effects of seizures in the grossly pathological brains of epileptic animals.

Endogenous predisposition to kindling and the potential development of epilepsy are also useful goals for kindling, but there is strong evidence that predisposition to one type of epilepsy does not necessarily impart predisposition to other types. The real benefit to the technique is that kindling can provide a number of important neurobiological observations, many of which are quite relevant to epilepsy. The caveat for all investigators is to evaluate the results of the model realistically with careful consideration of the model in light of the epileptic condition that they are trying to model.