A variety of morphologic changes can cause recurrent seizures. They range from the most obvious, such as some of the major developmental anomalies (see Chapter 5) or postasphyxial changes (see Chapter 6), to minor dysgenetic lesions such as the gray matter heterotopias (see Chapter 5). Although morphologic alterations would not be expected to be found in the primary epilepsies, sometimes they are (49). Mutations in the gene filamin A, that codes for a protein with a role in actin cross-linking and membrane stabilization have been reported to be responsible for the aberrance in migration that results in periventricular heterotopia (50,51). A number of authors have called attention to minor developmental anomalies in the molecular layer of the cerebral cortex and in the cerebellar cortex, some of which are clearly the result of disturbed cell migration (52). Malformations, notably gray matter heterotopias, cryptic tubers, or angiomas arising within the temporal lobe, can cause recurrent seizures. Such lesions also can be found in other areas of the brain (53,54). The genetic basis of some of the dramatic cerebral malformations associated with severe epilepsies of early childhood, such as the double cortex syndrome or band heterotopia, are also beginning to be understood (55,56) (see Chapter 5).

Altered neuronal migration that results in granule cell disorganization in the dentate gyrus has been seen in tissue resected from patients with temporal lobe epilepsy (57). Although initially this was thought to be a congenital lesion, provocative data from Parent demonstrates that, even in mature animals, status epilepticus can result in neurogenesis in the dentate gyrus, and that the nascent granule cells may migrate aberrantly. The data suggest that aberrant
synapse formation by these cells could contribute to abnormal excitability (58,59,60).

The role of infectious processes in the pathogenesis of epilepsy has received relatively little attention since Aguilar and Rasmussen established that some epileptic patients with focal seizures, slowly progressive intellectual deterioration, and cerebral atrophy demonstrate a pathologic picture consistent with a viral encephalitis (61). Attempts at viral isolation have been unsuccessful in these cases. Nevertheless, it is likely that not only focal epilepsy, but also other forms of seizure disorders, particularly disorders beginning in early childhood in previously healthy children (such as epileptogenic encephalopathy or progressive facial hemiatrophy), are caused by a smoldering viral disease within the brain (see Chapter 7).

Among the lesions considered to be secondary to recurrent seizures are those that result from the physical trauma that often attends seizures, and those that result from hypoxia, vascular alterations, or the action of the excitatory neurotransmitters.

Meldrum and colleagues (62) explored the possibility that the seizure itself, rather than systemic changes, was responsible for brain damage. They showed that brain damage occurred in the absence of systemic abnormalities in paralyzed, ventilated, adolescent baboons that were subjected to prolonged, bicuculline-induced seizures. Although the neurochemical changes attending cell death owing to prolonged seizures are similar to those seen in ischemia and hypoglycemia, significant differences in the time course and anatomic distribution of brain damage occur (see Table 17.1).

Under clinical conditions, damage results from a combination of the increased metabolic demands that accompany excessive neuronal activity and the reduced circulation and substrate supply induced by the combination of hyperthermia, hypoglycemia, hypotension, and hypoxia. Cell death under these conditions occurs through a process that resembles cell death in asphyxia, namely through the release of excitotoxins that increase intracellular calcium in the course of prolonged seizures. A more extensive discussion of this process can be found in Chapter 6.

Several areas of the brain, especially the hippocampus, appear to be particularly vulnerable to recurrent and prolonged seizures. Anatomic manifestations of cell damage to the hippocampus include loss of interneurons in the hilus, pyramidal cell loss within the Sommer’s sector (prosubiculum and subfield CA1 of Ammon’s horn), and subfield CA3, with consequential glial scarring and atrophy (63,64). Using Golgi techniques to study the hippocampus and dentate nucleus, Scheibel and associates have observed loss of dendritic spines and deformation of the dendritic shaft (65). This selective hippocampal vulnerability has been postulated to result from a high density of excitatory receptors on nerve cells in Sommer’s sector (66). Other factors also could be operative. Using in situ hybridization techniques, Sommer and coworkers showed unique developmental patterns in the mRNA expression of the Glu R-1, -2, and -3 glutamate receptor subunits in CA1, CA3, and the dentate gyrus. Differences in receptor structure could result in differences in receptor function and differences with maturation in resistance of the hippocampus to epileptic damage (67). The protective role of calbindin, a calcium-binding protein, from glutamate-induced neurotoxicity also could account for the selective nerve cell loss (68).

Within the gray matter of the cerebral hemispheres, neuronal cell loss is most likely to occur in laminae 3 and 4, where the thalamocortical afferents terminate. Damage also occurs in the pars reticularis of the substantia nigra, globus pallidus, and thalamus. In animal models, the substantia nigra has been demonstrated to play an
important role in the propagation of seizures, and damage to this structure could conceivably contribute to increased propensity for seizures (69). The caudate nucleus appears to be spared (70).

There is controversy as to the effect of seizures on the immature brain and whether brief but repetitive seizures can induce brain damage. One of us (R.S.) has reviewed the arguments for both sides in this controversial issue (71,72). Some types of experimental seizures fail to produce histologic lesions (71). This observation gave rise to the argument that the immature brain is not vulnerable to seizure-induced damage. This argument runs contrary to the observations on surgically resected tissue from epileptic children that show structural alterations that can be attributed to seizures (73,74). More recent work has shown that the effect of seizures on the developing brain is age- and model-specific, and also that both prolonged and brief and repetitive seizures are associated with the induction of neuronal apoptosis in specific cell populations in human and experimental animals (75,76,77,78).

Brief but recurrent seizures induced by pentylenetetrazol have been shown to contribute to morphologic and functional alterations in neonatal rat pups (79,80,81). The question whether brief but recurrent seizures also have a similar potential to induce brain damage cannot be answered with assurance in humans. In terms of clinical practice, patients with recurrent seizures are invariably treated with antiepileptic drugs, which on their own can affect development (82); the natural history of the condition without treatment is, therefore, impossible to study. Autopsy material does not permit an easy distinction of the pathology that caused the frequent seizures from the effect of the seizures themselves.

The studies of Shewmon and Erwin are more directly applicable to the clinical problem. These workers have demonstrated that interictal spikes, when followed by prominent inhibitory after potentials, can transiently disrupt cortical function. Thus, frequent interictal spikes could interfere with modality-specific learning (83). Presumably, recurrent electrical discharges could influence activity-dependent plasticity of the developing brain.

The damage seen in the hippocampus obtained by surgical resection in chronic temporal lobe epilepsy differs from that seen in postmortem specimens after status epilepticus (84,85). In contrast to the selective damage seen after status epilepticus, the hippocampus of subjects with chronic temporal lobe epilepsy shows more widespread damage throughout the CA1, CA2, and CA3 subfields as well as the dentate granule cells (84). Hippocampal cell loss ranges from mild and random to almost complete. Although initially the changes have a bilateral distribution, with time, one hemisphere becomes more affected (86). This pathologic abnormality has been designated as mesial temporal sclerosis (MTS) or hippocampal sclerosis. It was seen in 47% of resected temporal lobes in the series of Falconer and associates (Table 14.6) (87), and in 64% of a more recent series compiled by Engel and associates (88). Cell loss and gliosis in the amygdala also has been observed and can occur in the absence of significant hippocampal changes (89). In more severe cases, nerve cell loss and gliosis involves not only the entire hippocampus, but also the uncus, amygdala, and adjacent cortex (Fig. 14.1). Atrophic changes in the cerebellum or the thalamus are not uncommon.

The cause or causes of MTS are still unresolved. MTS has been seen as early as 1 year of age, and combined morphologic and electrophysiologic studies suggest that the seizure focus is generated in part when abnormal, recurrent, monosynaptic excitatory synapses are formed after damage to normal intrahippocampal synapses (90). A detailed analysis of the epileptogenic potential of the lesions produced by intrahippocampal or systemic kainic acid or by ischemia led Franck to conclude that hippocampal sclerosis and seizures are both symptoms of an underlying pathology, and that although MTS may be produced by seizures, the development of epilepsy as a syndrome does not depend on cell loss or plasticity in the hippocampus (91). Neuroimaging studies performed within 48 hours of a prolonged febrile convulsion indicate the presence
of hippocampal edema that resolves within a few days, and that in some instances is replaced by hippocampal atrophy (92,92a). These studies do not resolve the questions of whether a pre-existing hippocampal abnormality predisposed to the prolonged febrile convulsion and whether the anatomic features of MTS may become incorporated into, and sustain, an epileptic focus. In this regard, the identification of an interleukin gene polymorphism identified with both prolonged febrile convulsions and temporal lobe epilepsy with hippocampal sclerosis is quite interesting (92b).

Vascular, metabolic, genetic, and immunologic factors, acting singly or in concert, can be responsible for MTS (93,94). Epidemiologic studies have been used to ascertain risk factors for this condition. Febrile convulsions were seen in 20% of subjects, as compared with 2% of controls. The majority experienced at least one complicated febrile seizure. An increased incidence of head trauma and neonatal convulsions also could be documented. Additionally, there was a significant association with maternal seizures (95). In a significant population of patients, the brain, after a lifetime of recurrent epileptic attacks, shows neither gross nor microscopic abnormalities. This observation reflects the current limitation of morphologic studies in furthering our understanding of the epilepsies.

An epileptic neuron has, among other characteristics, an increased electric excitability and the ability to sustain an autonomous paroxysmal discharge that can be influenced from the outside by synaptic activity. Intracellular recordings within an epileptic focus reveal that during the time when an interictal discharge is recorded on the scalp EEG, a compact population of neurons displays a stereotyped abnormality called paroxysmal depolarization shift (PDS) in intracellular recordings. A PDS is characterized by a sudden, large, and sustained (approximately 30 mV for 70 to 150 msec) depolarization that is synchronized in many neurons. Multiple, high-frequency action potentials are superimposed on the PDS. The PDS and the interictal spike on scalp EEG, which represents that synchronized PDS of a local population of neurons can occur spontaneously or can be triggered by afferent stimuli. The PDS is followed by a hyperpolarization of 10 to 20 mV below the resting potential that lasts 700 msec or longer. During this period, the focus is refractory to afferent stimulation.

The large and lasting depolarization that characterizes a PDS is attributed to the triggering of voltage-gated calcium channels by the incoming action potential. This depolarizing calcium conductance is mediated by a subtype of excitatory amino acid receptor, which is characterized by its high affinity to N-methyl-D-aspartate (NMDA). The calcium channel is regulated by magnesium through a voltage-dependent block that can be removed by the initial sodium influx triggered by the action potential. The rise in intracellular calcium in turn triggers the opening of a specific type of potassium channel that initiates the hyperpolarization phase.

Ictogenesis is the spread of localized epileptic discharges to induce a clinical seizure during which thousands of neurons fire synchronously for prolonged periods. Several experimental systems have been used to study how localized discharges are able to spread. In a nonepileptic brain, an area of neuronal hyperpolarization, the inhibitory surround, surrounds the region of synchronous paroxysmal discharges. This inhibitory surround limits the duration of the interictal discharge, determines its frequency, and prevents its progression into a full-blown seizure. Neurons can become hyperpolarized by several processes that can differ from one set of neurons to another.

The mechanisms responsible for this transition from interictal to ictal period probably involve nonsynaptic processes such as electrical field effects (ephaptic interactions) and electrotonic coupling via gap junctions (100). Changes
in the extracellular environment, such as K⁺ and Ca²⁺ concentrations, also can affect the excitability of neuronal populations (101,102), suggesting that astrocytes also may play important roles in this process.

The observation that the normal limbic system can become epileptogenic by repeated stimulation, a process termed kindling, has provided a model for the study of the development of complex partial epilepsy (103,104). Kindling refers to a process by which brief trains of subconvulsive electrical stimuli are repeatedly delivered at appropriate intervals to a susceptible area of the brain. Initially, these stimuli produce after-discharges, which become progressively more prolonged until they give rise to limbic and clonic motor seizures. When stimulations are continued for even longer periods, spontaneous seizures appear. Once established, the effects of kindling are permanent. Kindling also can be achieved by chemical stimulation of the cortex.

During kindling, the mossy fiber pathway (efferent from the granule cells of the dentate gyrus) undergoes reorganization of its synaptic connections (105). The resulting recurrent excitatory connections have been implicated in the progressive development of hypersynchronous discharge. Such synaptic reorganization associated with loss of pyramidal cells in the CA3 subfield has been demonstrated in human epileptic tissue (106,107,108). Sloviter has suggested that the recurrent mossy fiber terminals include synapses on inhibitory interneurons (109).

Using his perforant path stimulation model of status epilepticus, Sloviter has studied the development of mossy fiber sprouting, chronic epilepsy, and time-dependent alterations in dentate inhibition. He suggested that the recurrent mossy fiber terminals include synapses on inhibitory interneurons (109,110). In his conceptualization, the hilar basket cells [inhibitory, g-aminobutyric acid (GABA)-ergic] are deafferented by the loss of another group of cells, the mossy cells (excitatory, glutamatergic), which normally receive mossy fiber input, and drive the inhibitory basket cells (Fig. 14.2); hence the term dormant basket cell hypothesis for this concept. Mossy fiber sprouting compensates for the loss of drive to the basket cells. GABAergic cells, indeed, appear to be preserved in human epileptic tissue (111) and also in animal models (110), thus supporting this concept.

Studies have compared the expression of excitatory amino acid receptor subunits and glutamic acid dehydrogenase (GAD) (presynaptic marker for GABA terminals) to the extent of mossy fiber sprouting in tissue from patients who underwent surgery. Patients’ granule cell KA2 and GluR5 mRNA levels were increased in association with aberrant fascia dentata mossy fiber sprouting; however, increased glutamic acid dehydrogenase immunoreactivity also was present in such tissue (112,113).

The preceding discussions pertain to the structural (network) plasticity that may be associated with focal epileptogenesis. There also is evidence for functional plasticity of synapses. Excitatory synapses show robust enhancement when they undergo repetitive high-frequency activation (114). This includes facilitation during the course of sustained stimulation and long-term potentiation that lasts hours to days after such a burst of activity in excitatory synapses. In contrast, similar repetitive high-frequency driving diminishes the efficacy of inhibitory (GABAergic) synapses (115,116).

Evidence has emerged to suggest that the epileptogenic process also may involve cellular plasticity in addition to synaptic and network plasticity. Lasting changes in the subunit composition of hyperpolarization-induced, cyclic nucleotide-activated cation channels (HCN), which regulate cellular excitability, have been noted after hyperthermic convulsions in rodents (117). HCN channel plasticity also has been demonstrated in resected hippocampal tissue from humans (118). Alterations in channels that mediate low-threshold calcium currents (T-calcium channels) also has been demonstrated in both experimental and resected human hippocampi (119).

Three factors determine whether a focal seizure will become generalized. The first is the excitability of the epileptic neurons, the second is the ease with which an electric discharge can be propagated from the focus, and the third is the threshold of the brainstem centers for disseminating an electric discharge. The last is believed to reflect in part a genetic predisposition, and in part the frequency with which the brainstem centers are activated by the primary epileptic focus.

Current concepts propose that a secondarily generalized tonic-clonic seizure results from the axonal propagation of the cortical ictal discharge to the contralateral cortex, and to subcortical structures via intrahemispheric and interhemispheric association pathways. The substantia nigra, in particular, appears to be involved, at least in the control of experimental seizures (120).

Once neuronal excitation derived from the epileptic cortical focus spreads to involve the brainstem, particularly the midbrain and the pontine reticular formation, a generalized seizure develops almost instantly. These areas are responsible for the dissemination of epileptic potentials (121,122). With the subcortical neurons involved by the epileptic discharge, a positive excitatory feedback circuit is established between the cortex and the subcortical neurons, inducing discharges at a rate of 10 to 40 Hz. This circuit is responsible for the tonic phase of the focal motor seizure. As inhibitory neurons are recruited, a negative inhibitory feedback circuit develops, which periodically interrupts the excitatory activity and produces the clonic phase of the seizure. When the negative feedback wins ascendancy, the seizure subsides, leaving the neuronal membrane in a far greater hyperpolarized state than before the onset of the seizure.

Considerable evidence suggests that postictal (Todd’s) paralysis, a common sequel to a focal seizure, is caused by persistence of the active inhibitory state, rather than by metabolic exhaustion of epileptic neurons (123).

The various areas of the cortex differ in their potential for secondary generalization. A number of areas including the temporal, frontal, and prefrontal cortex have particularly strong corticofugal projections to the centrencephalic system, and focal lesions within them readily induce a generalized seizure discharge (124,125). By contrast, the potential for secondary generalization is low in the motor strip. Additionally, small cortical lesions are more likely to induce a focally restricted seizure, whereas multiple or diffuse cortical lesions are more likely to result in a generalized seizure.

The pathophysiology of the primary generalized epilepsies is less well understood, and much of the experimental data is based on animal models that might not be applicable to the human epilepsies. In the 1940s, Penfield introduced the concept of centrencephalic epilepsy. This idea was that a generalized spike and wave discharge, such as occurs in absence epilepsy, originates in the rostral brainstem structures and the diencephalon, with the thalamus being responsible for the sudden generalized cortical discharge (126).

Gloor and Fariello postulate that in the primary generalized epilepsies, the cortex is in a diffusely hyperexcitable state, perhaps as a result of the discharge of a group of excitatory neurons. As a result, an epileptic discharge can be triggered by excitation of the brainstem and the midline thalamic reticular system induced by thalamocortical input (127). Engel suggests that the firing of a small group of excitatory neurons stimulates a set of inhibitory neurons that have connections throughout the cortex. A second burst of properly timed excitatory impulses then produces a synchronized discharge over a wide area of the cortex (128). Thus, the spike-wave discharges arise from the rhythmic, reverberatory interactions between interconnected thalamic and cortical neurons.

The oscillations in the thalamocortical circuits rely in part on the intrinsic membrane properties of the involved thalamic neurons. These neurons undergo slow, calcium-dependent depolarizations, attributed to the so-called T-channel (129). These low-threshold calcium currents provide the pacemaker quality to these cells that forms the basis of the thalamocortical reverberations. These spike-wave paroxysms are abolished by substances such as trimethadione or ethosuximide, which have a specific effect of abolishing calcium currents associated with the T-channel (129).

Clinically, such a discharge produces grand mal or absence epilepsy. In the latter condition, quick excitation of a neuronal inhibitory system prevents a prolonged clinical seizure and the development of the tonic-clonic components. These seizures can occur without known cerebral injury or disease, and, as indicated previously, they have a high rate of genetic transmission.

Views on the propagation of cortical electrical discharges have been considerably modified in the last few years, and the importance of the cortex in the production of both primary and secondarily generalized seizures has become increasingly evident from implanted or surface electrodes and from ictal and interictal PET scanning. These techniques also have shown the marked discrepancies between the scalp EEG and the actual seizure focus in approximately one-third of patients with refractory seizures (130,131).

The most common and most clearly delineated of the idiopathic focal epilepsies is midtemporal epilepsy (sylvian epilepsy, rolandic epilepsy, benign childhood epilepsy with centrotemporal spikes). In the experience of Loiseau and coworkers, it represented 10.7% of children seen in a specialized private practice (231). A high incidence of similar seizures occurs in first-degree relatives, and an autosomal dominant form of rolandic epilepsy has been reported (see Table 14.5) (25,232). In approximately 75% of children, this type of a seizure commences between ages 5 and 10 years. Seizures are infrequent, in the majority of cases occurring less than three or four times a year, and are generally brief. Most characteristically, attacks commence with a somatosensory aura, usually referred to the tongue, cheek, or gums and less often to the abdomen. As a result of motor interference, speech is arrested, the child salivates, and tonic or tonic-clonic movements involve the face. Consciousness is preserved in some 60% of children. Because approximately three-fourths of the attacks occur during sleep, a good patient history is difficult to obtain. The interictal EEG shows midtemporal spikes, probably a reflection of discharges arising from the rolandic cortex (233,234). In about 30% of cases, spikes are only seen during sleep (235). The spikes in this syndrome display a horizontal dipole with maximum negativity in the centrotemporal regions and positivity over the superior frontal area (236). In contrast to other seizures having a temporal spike focus, the prognosis in this condition is excellent (237); most children respond well to anticonvulsant therapy. In the experience of Lombroso, almost 50% were seizure free within 3 years of their first attack, and in more than 30%, the EEG reverted to normal as well (233). Beaussart and Faou share this optimism. In their series of 334 cases, no patient had seizures after age 13 years, even when anticonvulsants had been withdrawn (238).

Benign epilepsy with an occipital focus (239) is somewhat less common, accounting for 6% of all children with seizures (240). The age of onset is variable; ranging between 15 months and 17 years, with the peak age being 5 years (239,240). Seizures are infrequent and tend to mainly occur at night. Attacks during the day are initiated by hemianopia, phosphenes (white or colored luminous spots), or visual hallucinations. Amaurosis or nonvisual symptoms follow. The latter include unilateral convulsions, automatisms, or generalized tonic-clonic seizures. Aphasia is not unusual. Ictal deviation and vomiting may be seen, and in about one-half of children, there is postictal migraine (241). The interictal EEG most commonly demonstrates nearly continuous unilateral or bilateral high-voltage spike-wave discharges from the occipital or posterior temporal regions. This is generally suppressed by opening of the eye. The prognosis is excellent with or without anticonvulsant therapy, and all children in Beaumanoir’s series were seizure free with a normal EEG by 9 to 13 years of age (239). The experience of Panayiotopoulos is similar, with 89% of children gaining remission (242).

European epileptologists also distinguish a focal epilepsy with midtemporal spikes and affective symptoms. Seizures commence with a sensation of fear or terror and proceed with chewing and swallowing movements and speech arrest. Consciousness is depressed. Unlike complex partial seizures, seizures cease with therapy, and no evidence for a focal lesion can be found on clinical examination or by imaging studies (228).

Based on the different clinical manifestations of the epileptic attacks, at least four seizure types can be distinguished: atonic (astatic or akinetic) seizures, brief tonic seizures, atypical petit mal seizures, and myoclonic seizures (245).

Pseudo-Lennox syndrome is a condition that was first described by Aicardi and Chevrie under the term atypical benign partial epilepsy of childhood. It is marked by generalized minor seizures, such as atonic seizures, atypical absences, and myoclonic seizures, and focal sharp slow waves and spikes with generalization during sleep. Seizure onset peaked at about three years of age (255). In contrast to LGS, prognosis in terms of seizure remission is good. In the series of Hahn and coworkers, 84% of children with this condition ultimately went into remission (256). However, over 50% of children were attending schools for the mentally handicapped.

Myoclonic seizures also are seen as a nonspecific symptom in several forms of viral encephalitis, metabolic disturbances such as uremia, and progressive cerebral degenerative diseases such as the various lysosomal storage diseases, some of the leukodystrophies, Menkes disease, and Unverricht-Lundberg disease (see Chapter 3).

Several types of nonepileptic myoclonus have been recognized. Cortical reflex myoclonus is believed to result from hyperexcitability of a small region of the sensory portion of the sensorimotor cortex. The muscular contractions are irregular and are precipitated or aggravated by sensory stimuli, most commonly by light, noise, or tapping on the face or chest. Postural changes intensify the contractions. The movements disappear in sleep. The EEG is clearly abnormal, and a paroxysm precedes both the spontaneous and the reflex-induced myoclonic jerks. Giant somatosensory-evoked potentials are characteristic for this type of myoclonus (257).

In reticular reflex myoclonus, the myoclonic jerks affect the entire body, with the flexor muscles being more involved than the extensors. The seizures are believed to be the result of hyperexcitability of the caudal brainstem reticular formation, possibly the nucleus reticularis gigantocellularis (248). An EEG spike usually follows the first electromyographic evidence of myoclonus. Another nonepileptic form of myoclonus is considered to be of spinal cord or brainstem origin. The muscular contractions are rhythmic and are unaffected by sensory stimuli. They persist during sleep. The EEG can be normal or abnormal, and response to anticonvulsant medication is poor.

Nonepileptic myoclonus also is encountered in the context of involuntary movements in some of the disorders of the extrapyramidal system. It is seen in normal individuals while falling asleep. Rarely, frequent but nonprogressive myoclonic movements occur unaccompanied by other types of epileptic attacks or intellectual subnormality. This condition is termed paramyoclonus multiplex or essential myoclonus. It is probably identical to myoclonic dystonia (DYT 11) and is considered with the various dystonias covered in Chapter 3 (258). Closely related to this condition is the exaggerated startle response (hyperekplexia), conditions also considered in Chapter 3. In hyperekplexia, patients exhibit momentary muscular stiffness and loss of voluntary postural control without loss of consciousness.
Other features of this condition include transient hypertonia during infancy, nocturnal jerking of the legs, and insecure gait. No other epileptic phenomena occur, although the EEG can have paroxysmal features. Minor, quantitatively less severe forms of this condition also have been observed. The condition should be distinguished from startle epilepsy, a type of reflex epilepsy (259).

A condition that has a similar ominous prognosis as infantile spasms, and also is marked by frequent minor generalized seizures is Ohtahara syndrome (282) Onset of seizures is earlier than in infantile spasms and is usually prior to three months. The main seizure type is tonic spasms with or without clustering. The EEG is characteristic. It consists of high voltage bursts interspersed with a nearly flat pattern. The condition results from a variety of brain malformations, notably focal cortical dysplasias. It is more resistant to treatment than infantile spasms, and the prognosis for mental function is poor. In many instances, the tonic spasms evolve into infantile spasms, and like infantile spasms into LGS (283).

The relationship between treatment regimens and outcome is controversial, and the recommendations prepared by the American Academy of Pediatrics and the Child Neurology Society are too conservative to clarify the situation. These bodies concluded that adrenocorticotropic hormone (ACTH) is “probably an effective agent” for the short-time treatment of infantile spasms. They also concluded that there was insufficient evidence to state that successful treatment of infantile spasms improves the long-term prognosis (284). The Israeli group under Lerman and Kivity, however, found that optimal results are obtained when treatment is initiated within a month after the onset of seizures and when high doses are given for a prolonged period (259,285), and as already noted, their results with early treatment of infants with cryptogenic infantile spasms were uniformly excellent in terms of mental development and seizure control. The series of Glaze and coworkers did not have a sufficient number of cryptogenic cases to determine the benefit of early therapy (286). The more recent series of Askalan and coworkers also is too small and included infants who already had shown developmental regression before treatment (286a). Certainly, early ACTH therapy does not alter the uniformly poor outcome of symptomatic infantile spasms (286).
In this respect, it is of interest that the relapse after ACTH treatment is much higher in patients with tuberose sclerosis than in other forms of symptomatic infantile spasms (280). Even when treatment does not improve mental functioning, it can have a beneficial effect on seizure control and on the EEG in over one-half of the cases.

Many regimens for ACTH treatment have been proposed. As a rule, at higher ACTH doses, there is a greater proportion of remissions but also a higher incidence of major adverse reactions. One convenient treatment schedule follows: ACTH is given intramuscularly in a daily dose of 80 to 120 U. This dose is maintained until spasms and hypsarrhythmia disappear. If a response is noted, the dose is changed to 40 to 80 U every other day and then gradually reduced to half that dose. This is maintained for approximately 3 more months. Thereafter, the dosage is reduced further by 10 U at 2-month intervals until the drug is completely withdrawn or the minimal dosage for seizure control is reached. One of us (JHM) has recommended a starting dose of 40 U, which is increased at weekly increments to 60 U and 80 U if there has not been any response. A prospective study has indicated that there is no difference in the effectiveness of high-dose and low-dose regimens with respect to seizure control and normalization of the EEG (287). Ito and coworkers have suggested a starting dose of 20 U of synthetic ACTH for infants over 1 year of age and 10 U for infants under 1 year of age (288). On this dosage, excellent seizure control was seen in 76% of patients, and normalization of EEG was seen in 38%. Outcome in terms of mental function was poor, however, with only 6% of subjects being normal.

Side reactions to ACTH include hypertension, which can become apparent at a dosage of 80 U per day, gastroenteritis, sepsis, osteoporosis, and unexplained CNS hemorrhage. The incidence of these complications varies considerably and probably reflects the different treatment schedules. In some series, it can be as high as 37% (289), whereas others have not shown any serious morbidity (290). One interesting concomitant to ACTH therapy is an enlargement of the cerebral ventricular system, which is disclosed by serial neuroimaging studies. This enlargement, which is occasionally permanent (perhaps as a result of abnormalities predating the onset of infantile spasms), is believed to reflect loss of water, owing to altered CSF absorption (291).

The mechanism by which ACTH controls infantile spasms is a matter of some debate, and Snead has reviewed the various hypotheses (292). Baram and colleagues have shown that in infantile spasms, CSF levels of ACTH and cortisol are reduced and have postulated that there is an increased synthesis of corticotropin-releasing hormone (CRH), which in infant rats has been shown to induce seizures originating from the amygdala. They suggest that ACTH desensitizes the CRH receptors and thus induces a negative feedback on CRH release (293). ACTH also could modulate second messenger systems or increase the expression of several genes and thus accelerate myelination, in this manner shortening a vulnerable, hyperexcitable period (294). In that respect, it is significant that several studies, as well as our own experience, indicate that ACTH is a potent anticonvulsant for Lennox-Gastaut syndrome (290,295).

Of the other medications recommended for the treatment of infantile spasms, vigabatrin appears to be the most effective. In an uncontrolled study, Mitchell and Shah started patients on 125 to 250 mg per day, with gradual increments to a target dose of 100 mg/kg per day or any lower effective dose (296). In this unselected series, 60% responded with cessation of spasms and resolution of hypsarrhythmia. Appleton recommends starting at 25 to 50 mg/kg per day and increasing the dosage to a maximum of 80 to 120 mg/kg per day. In their group of children, making up 81% of symptomatic infantile spasms, 81% experienced total cessation of spasms. On follow-up 2 years later, 67% had remained seizure free (297). Chiron and coworkers have used vigabatrin with considerable success in the treatment of infantile spasms associated with tuberose sclerosis (298). In a retrospective study by Cossette and colleagues, vigabatrin was found to be as effective as ACTH in terms of controlling seizures and offering a good developmental outcome. Side effects were less frequent in vigabatrin-treated children than in those who received ACTH (299). Their results are at variance with those of Lux and colleagues, who found that synthetic adrenocorticotropic hormone and prednisolone were more likely to induce cessation of spasms than vigabatrin given in minimal doses of 100 mg/kg per day (299a). The high incidence of visual field defects seen on even low doses of vigabatrin has raised concerns with respect to the safety of the medication. On that account, Riikonen believes that the benefits of vigabatrin do not outweigh the risks of irreversible visual changes (300). In spite of these drawbacks, vigabatrin has become the first-line drug in other clinics for the treatment of infantile spasms, particularly when these occur in a setting of tuberose sclerosis (301,302).

Oral steroids are probably not as effective for infantile spasms as ACTH. In a prospective, randomized study, high-dose ACTH (150 U/m² per day) was more effective than prednisone by both clinical and EEG criteria (303). Prats and coworkers have suggested treating infantile spasms with extremely high doses of sodium valproate (100 to 300 mg/kg per day). According to their protocol, the anticonvulsant is given for 21 days or until the EEG resolves. Then, the medication is reduced to 25 to 50 mg/kg per day (304). Although their results require confirmation, the outcome of children treated with valproate appears to be as good or better than those treated with ACTH both in terms of seizure control and normal intellectual development.

Kossoff and coworkers have used the ketogenic diet to treat infantile spasms and found the response rate to be significant, with 46% being more than 90% improved and
100% of subjects being more than 50% improved in terms of EEGs and parent reports (305).

Many other treatment regimens have been suggested. In Japan, large doses of pyridoxine are preferred by the majority of institutions, followed by the combination of pyridoxine and valproate, and valproate monotherapy (306). Combined treatment with vigabatrin and topiramate has been used in Italy (307). Zonisamide has been found to be effective in one-fourth to one-third of patients with infantile spasms (308,309,310).

As ominous as are infantile spasms in terms of ultimate mental development, Lombroso and Fejerman were able to distinguish a benign form (311). In this condition, myoclonic attacks are accompanied by normal intellectual development and a normal EEG, in contrast to infantile spasms in which the EEG is abnormal at the onset of seizures or becomes abnormal within a few months, as is the case in young infants. In the benign form of the condition, myoclonus ceases before 2 years of age, and no other seizures supervene.

Attacks characterized by shuddering and resembling myoclonic seizures can occur in monosodium glutamate poisoning and also can be noted as a prelude to familial (essential) tremor (see Chapter 3).

A syndrome of infantile spasms, mental retardation, chorioretinitis, colobomas of the retina, and agenesis of the corpus callosum limited to female subjects was first described by Aicardi and coworkers (Aicardi syndrome) (see Chapter 4) (312,313,314).

Paroxysmal attacks of abdominal pain can occur as an aura for a major motor attack or can be the only manifestation of a seizure. The abdominal pain is usually periumbilical, radiating to the epigastrium. In the majority of cases, it lasts 5 to 10 minutes, but it can persist for 24 to 36 hours. It is usually associated with disturbed awareness (321).

The pediatrician often sees a child who has recurrent bouts of paroxysmal abdominal pain accompanied by vomiting and in whom the usual gastrointestinal evaluation result has been normal. Only a small proportion of these children has abdominal epilepsy. A more common cause of this complaint is childhood migraine. A history of pain and vomiting is common in small children who subsequently develop the usual clinical picture of migraine. The diagnosis of abdominal epilepsy rests on the presence of other epileptic manifestations, usually complex partial seizures, an abnormal EEG pattern during or between attacks, and, less convincingly, a favorable response to anticonvulsants, usually carbamazepine, valproate, or phenytoin (322).

Epilepsia partialis continua, a variant of a simple partial (focal) seizure, is characterized by clonic movements, usually localized to the face or upper extremities, that persist over long periods either continuously or with only brief interruptions. Consciousness is not impaired, but postictal weakness is usually evident. We have seen several children with this condition who had an underlying chronic encephalitis of the kind described by Aguilar and Rasmussen (61) and Rasmussen and McCann (323). Rasmussen syndrome was the most common cause for epilepsia partialis continua in the series of Thomas and coworkers (324). This condition, and its relationship to immune disorders and a
variety of infectious agents, and its treatment is considered in Chapter 8.

Various space-occupying lesions, including tumor, abscess, and cerebral cysticercosis, are less frequent causes for epilepsia partialis continua. Epilepsia partialis continua also has been encountered in various metabolic disorders, notably nonketotic hyperglycinemia (see Chapter 1).

Several types of nonconvulsive status epilepticus (NCSE) have been recognized: a generalized form, also termed absence status or petit mal status, and a partial type, termed complex partial status epilepticus; NCSE in patients with learning difficulties—including status epilepticus during sleep, atypical absence status epilepticus, and tonic status epilepticus—and NCSE in coma (325,326). NCSE is frequently seen in patients who are critically ill, such as in end-stage renal disease, or who have other major medical problems and can arise from treatment with cephalosporins, cefepine, isfosfamide, or tiagabine or upon acute withdrawal of diazepines, such as lorazepam.

Absence status is the most common form of nonconvulsive status epilepticus. It is characterized by a prolonged state of clouded mental activity usually accompanied by an EEG picture of atypical, slow spike-wave complexes and polyspike discharges. Untreated, the condition can last from several hours to as long as 2 years (327). Absence status almost always occurs in children with pre-existing organic cerebral lesions, a history of intellectual retardation, and a variety of other seizure forms (Fig. 14.6) (328). Approximately one-fourth of children with Lennox-Gastaut syndrome enter absence status one or more times during their lifetime (251).

Farran and coworkers have suggested that absence status is seen in patients with cerebellar anomalies and a consequent defect in the inhibitory system mediated by the Purkinje cells (329). The PET scan obtained during an attack fails to confirm this supposition in that it demonstrates only diffuse hypometabolism of fluorodeoxyglucose (330).

Complex partial status epilepticus is characterized by altered mentation, with impaired or absent language function, and a variable degree of responsiveness. The EEG during an attack shows various abnormalities, ranging from focal temporal spike and wave discharges to rhythmic slowing.

Benzodiazepines (lorazepam or diazepam) or valproate are the best drugs for treatment of both forms of nonconvulsive status, followed by institution or resumption
of long-term anticonvulsant therapy (331). There is no conclusive evidence that prolonged NCSE can induce brain damage, and Kaplan cautions against overtreatment of the condition (332).

Landau-Kleffner syndrome, which has attracted considerable attention in recent years, is marked by an acquired aphasia in children who have had normal language and motor development (333). Symptoms develop between 4 and 7 years of age. A variety of seizures accompany this condition in approximately 80% of cases. They can be rare and can disappear completely before adolescence. The EEG abnormalities are diagnostic. They include bilateral independent temporal or temporoparietal spikes, or spike and wave discharges, and in a large proportion of children, a continuous spike and wave discharge during as much as 90% of the sleeping state (334). Neuroimaging study results are normal. The aphasia frequently fluctuates in severity, but as a rule, the younger the age when symptoms start, the worse the prognosis in terms of language function (335). Other aspects of higher cortical function are preserved. Although in the initial report the condition was seen in two brothers, the etiology remains totally unknown.

Valproate, ethosuximide, and the benzodiazepines can improve the condition; phenobarbital and carbamazepine are generally ineffective. ACTH or corticosteroids also can be partially effective (336). Other therapeutic approaches include the use of intravenous immunoglobulins, which can be effective in some patients (337), and multiple subpial transections (338). As a rule, the EEG abnormalities regress with time, leaving the child with a severe receptive and expressive aphasia.

Landau-Kleffner syndrome has much in common with a condition in which subclinical electrical status epilepticus occurs during slow-wave sleep; continuous spike-wave discharges during slow-wave sleep (ESES) (339,340). This is a relatively rare seizure disorder. It is marked by partial or generalized seizures that are usually nocturnal, associated with reduction in language function, reduced attention span, and a variety of behavior disorders (341,342). The characteristic EEG pattern appears with onset of sleep and consists of spike-wave discharges, usually at 1.5 to 2 Hz, that persist throughout slow-wave sleep. During REM sleep, the frequency of spike and wave discharges is the same as during the waking state.

Treatment of the EEG abnormalities is not successful, although a number of anticonvulsants and corticosteroids have been used (341). In distinction from Landau-Kleffner syndrome, where there is an isolated disturbance of language function, other cognitive functions also are affected in ESES. The clinical course and response to anticonvulsants is similar for the two conditions, and many workers consider ESES to be a variant of the Landau-Kleffner syndrome or to represent a clinical continuum (334,340).

Rarer seizure forms are summarized in Table 14.8 (343,344,345,346,347,348,349,350,351). The most likely of these to be encountered are the reflex seizures. This term denotes seizures that are repeatedly initiated by clearly defined stimuli. Reflex epilepsies account for some 5% of all epilepsies, with visually induced reflex (photogenic) epilepsy being the most common form, constituting 53% of Forster’s series (352). Visual reflex epilepsy includes television epilepsy and video game epilepsy as well as most cases of epilepsy in which the child induces his or her own seizures by looking at a light source and rapidly moving the hand back and forth in front of the eyes (“fanning”) (353,354). Other patients, not included in the group of reflex epileptic patients, can have both photogenic and nonphotogenic seizures or, even more
commonly, can show only a photoconvulsive response on the EEG (355).

The mechanism for reflex seizures is unclear. The most attractive theory is that the attacks are caused by hyperexcitability of the sensory cortex, perhaps combined with a lack of cortical inhibition for a certain type of afferent input.

Treatment consists of avoiding the specific stimulus that induces seizures and administering anticonvulsants, preferably valproate or clonazepam. It is of note that the frequency of the TV or video game screen is important in provoking a seizure, with a 100-Hz screen being significantly safer than a 50-Hz screen. In addition, the distance of the child from the screen also is important, in that 1 m is safer than 50 cm (354).