Child Neurology
7th Edition

Numerous studies suggest that the genetic susceptibility to seizures is normally distributed in the general population, and that there is a threshold above which the condition becomes clinically evident.

An interaction between one or more genes and various nongenetic events operates in several conditions accompanied by seizures. These include head trauma, brain tumors, and congenital hemiplegia (9,10). Genetic factors appear to be most significant in patients with the various primary epilepsies (11). The various nonprogressive hereditary epilepsies are summarized in Table 14.5 (12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30). In one study, Lennox and Lennox found a 70% concordance for monozygotic twins and 5.6% concordance for dizygotic twins for epilepsies without organic brain lesions (31). Metrakos and Metrakos found a 12% incidence of seizures among parents and siblings of children with absence seizures; 45% of siblings had an abnormal EEG. They proposed that this EEG abnormality is an expression of an autosomal dominant gene with nearly complete penetrance during childhood and low penetrance in infancy and adult life (32). Gerken and Doose, interpreting data derived from their clinic, concluded that it was unlikely that a single autosomal dominant gene was responsible for the 3-Hz spike and wave trait and suggested a polygenic inheritance with neurophysiologic and genetic heterogeneity (33). Indeed, at least three genes for absence epilepsy have been mapped at this point in time (see Table 14.5) (34,35).

In families with centrotemporal spikes or sharp-wave discharges and rolandic seizures, EEG abnormalities are transmitted in a dominant manner with age-dependent penetrance (11). Only 12% of relatives with EEG abnormalities, however, develop clinically apparent seizures. This type of seizure has been mapped to chromosome 10q22-q24, with the defective gene being LGI1 (36). A significant genetic predisposition also occurs in juvenile myoclonic epilepsy, in photosensitive seizures, and in the various other primary generalized epilepsies. In seizures

with secondary generalization, the genetic factors, although demonstrable through controlled twin studies, are not as striking as in the primary generalized epilepsies. However, even in absence epilepsy, in which the genetic factor is most prominent, the overall risk of developing seizures is only 8% for siblings of affected subjects and 2% in as yet unaffected siblings older than 6 years of age (37). For offspring of subjects with absence seizures, the risk for EEG abnormalities is 64% and for seizures is 6.7% (38). When promazine is used to activate the EEG, 73.5% of 7- to 14-year-old siblings of subjects with idiopathic absence seizures develop an abnormal EEG (39).

Several other genes responsible for epilepsy have been mapped and cloned. The first epilepsy gene to be cloned was one of three genes responsible for autosomal dominant nocturnal frontal lobe epilepsy (24,40). It has been mapped to chromosome 20q13.2-13.3 and encodes the nicotinic acetylcholine receptor alpha-4 subunit (CHRNA 4). The same authors later reported a different mutation in the same gene for a different pedigree with this syndrome (41). Two other genes for this condition have been mapped to chromosome 15q24 and chromosome 1 (24). The discovery of mutations in this gene was perplexing to many because this receptor has not been considered to be involved in the modulation of neuronal excitability relevant to seizure disorders.

The finding that benign familial neonatal convulsions are attributable to mutations of voltage-gated potassium channels, KCNQ2 (12,13,42,43) and KCNQ3 (44), is more in tune with our understanding of the mechanisms of excitability. Altered K⁺-channel function could impair neuronal repolarization and thus contribute toward increased excitability. The extremely transient nature of this disorder suggests that compensatory changes probably take place in other genes controlling excitatory or inhibitory ion channels.

The relationship between the genotype and the phenotypic expression of the gene disorder is complex and is complicated by phenotypic convergence—that is, two different genetic mutations can induce the same clinical picture. Thus, afebrile seizures during the first year of life can result not only from mutations in the sodium channel gene, SCN2A1, but also from mutations in the GABA receptor gene, GABRG2. Conversely, there is phenotypic divergence, and different mutations of the same calcium channel gene CACNA1A are associated with familial hemiplegic migraine, episodic ataxia, and epilepsy (45). Mutations in LGI1 can be associated with a variety of phenotypes: partial epilepsy with auditory features, mesial temporal lobe epilepsy, temporal lobe epilepsy with febrile seizures, and temporal lobe epilepsy with developmental delay (36). Mutations in SCN1A are associated with a clinical continuum, including severe myoclonic epilepsy of infancy, generalized epilepsy with febrile seizures plus, and intractable childhood epilepsy with tonic-clonic seizures as well milder forms such as classical “febrile seizures”(46). Likewise, mutations in the chloride channel gene CLCN2 can be found in childhood absence epilepsy, juvenile myoclonic epilepsy, or epilepsy with grand mal upon awakening (46a). The reasons that underlie this clinical diversity are unclear. It is likely that additional genes contribute and modify the phenotypic expression. Indeed, Durner and colleagues found statistical support for a major susceptibility gene for idiopathic generalized epilepsy and different modifying genes (47).

Seizures can occur in patients with almost any pathologic process that affects the brain. Two types of abnormalities are seen: those that are responsible for recurrent seizures, and those that are the consequence of recurrent seizures.

Gowers stated more than 100 years ago that seizures beget seizures (48). The question whether lesions produce seizures or seizures produce lesions has been extensively investigated.

It should be stated at the outset that modulation of transmitter effects, of voltage-gated channels, and of cell electrical properties involves processes that occur continually during normal brain function. This plasticity is the basis of the cortex to learn from experience. It seems that the same plastic mechanisms are involved in epileptogenicity. One extreme result of such plasticity is the hyperexcitability and hypersynchrony that characterize epileptiform activities. The risk of epileptiform activity is the price that has to be paid for a nervous system that is so adaptive (97).

Each clinical form of epilepsy is generated by a different set of mechanisms. In general, there is greater understanding presently of generators of focal epileptiform activity than generalized epileptiform activities. Cellular aspects of epileptogenesis are reviewed by DeLorenzo (98) and Velísek and Moshé (99).

From a neurophysiologic point of view, an epileptic seizure has been defined as an alteration of central nervous system (CNS) function resulting from spontaneous electrical discharge in a diseased neuronal population of cortical gray matter or the brainstem.

Epileptogenesis requires a set of epileptogenic neurons, the presence of disinhibition, and circuitry to permit synchronization.

Excitatory neurotransmitters can play a role in the development of seizure discharges. Glutamate and aspartate are the major excitatory neurotransmitters found in the
mammalian brain (132). Focal application of glutamate to hippocampal slices induces a calcium ion current and depolarizes the neuron. Another excitatory neurotransmitter system, the cholinergic system, has been successfully manipulated to produce experimental limbic seizures (133). Although the development of antiepileptic drugs based on their action at excitatory amino acid receptors is an active area of current research, presently available antimuscarinic agents are not likely to be useful as anticonvulsants because of their widespread central and peripheral sites of action.

Nicotinic receptors have not been thought to be important in the mechanisms underlying seizures. The recent finding of an association between autosomal dominant nocturnal frontal lobe epilepsy and mutations in the gene for a nicotinic cholinergic receptor subunit (CHRNA4) was thus surprising to investigators (24).

GABA has been shown to have inhibitory postsynaptic activity and is one of the principal inhibitory neurotransmitters in the mammalian brain. The role of inhibitory GABA-releasing neurons in producing hyperpolarization has been well established, and temporary disinhibition might predispose a normal neuronal population to epileptiform activity. The GABA receptor has been found in all areas of the brain (134). This receptor is coupled to the chloride channel (chloride ionophore), so that GABA binding to its receptor results in a rapid opening of the chloride channel, with an ensuing increase in the postsynaptic membrane conductance to chloride. Increased chloride ion permeability stabilizes the cell near its resting membrane potential and reduces its response to excitatory inputs. Modulation of the GABA receptor-chloride ionophore complex mediates the actions of benzodiazepines and barbiturates as well as the convulsant effect of picrotoxin and its analogues (135). Interactions at the GABA receptor-chloride ionophore complex also underlie some of the mechanisms of action of felbamate and topiramate. These substances all have receptor sites on the GABA receptor complex. The receptor protein has been isolated and purified, and the genes coding for its five subunits have been cloned (136,137). The physiology and pharmacology of the receptor assembly depend on the subunit composition (137,138,139). No genetic diseases resulting from a defect in these genes have been demonstrated as yet.

Convulsions can be induced by substances that block the biosynthesis of GABA. Allylglycine, an inhibitor of glutamic acid decarboxylase, the enzyme promoting conversion of glutamic acid to GABA, is a potent epileptogenic compound (94). Inhibition of GABA binding to its receptor by bicuculline and inhibition of the postsynaptic GABA-chloride conductance responses by picrotoxin also induces convulsions. When GABAergic inhibition is progressively blocked with picrotoxin, stimulation of one neuron excites more and more neurons until neuronal behavior begins to resemble a seizure. Conversely, a number of compounds that elevate nerve-terminal GABA concentrations are potential anticonvulsants (140). These include such inhibitors of GABA transaminase as γ-vinyl-GABA (vigabatrin) and valproate. It is unclear, however, whether the anticonvulsant effects of valproate are indeed related to its elevation of brain GABA because this effect is seen only at supratherapeutic levels.

Magnetic resonance (MR) spectroscopy has provided evidence that gabapentin and topiramate also measurably increase brain concentrations of GABA even though they do not inhibit GABA transaminase (141,142). Extracellular (presumably also synaptic) concentration of GABA is increased by tiagabine, a nipecotic acid derivative (143).

Pyridoxine functions as a coenzyme for glutamic acid decarboxylase. Consequently, an increased glutamic acid to GABA ratio is expected in pyridoxine deficiency, a state marked by prolonged seizures. In pyridoxine dependency, an autosomal recessive disorder also characterized by seizures, the GABA content of brain is reduced, and brain glutamic acid concentration is elevated (144). Even though skin fibroblasts show reduced activity of pyridoxal-dependent glutamic acid decarboxylase, no abnormality in the two genes coding for glutamic acid decarboxylase, the biosynthetic enzyme for GABA, has been documented (145).

This relatively simple model for the epileptogenicity of insufficient inhibition and excessive excitation may, however, not hold true. Strong inhibition can predispose to hypersynchronization, and in petit mal seizures, hyperfunction of GABAergic inhibitory pathways is evident (146). Brainstem and diencephalic influences, which normally induce slow-wave sleep, also increase cortical neuronal synchronization and raise the potential for epileptogenicity. In some of the experimental epilepsies, notably the reflex epilepsy of Mongolian gerbils, the number of GABAergic neurons is increased (147).

A role for other inhibitory neurotransmitters, notably the opioid peptides, in the production of interictal inhibition, and a role for postictal depression have been suggested from various animal models (148). At low dosages, morphine and other opioids have anticonvulsant activity that is reversed by naloxone, whereas at higher dosages, they act as convulsants, producing a petit mal–like seizure disorder. The relevance of these observations to human petit mal is uncertain, although positron emission tomography (PET) has revealed increased opiate receptor binding in the temporal cortex in patients with complex partial seizures (149).

Adenosine is another potent inhibitor of cortical neurons, acting primarily by depressing spontaneous neuronal firing or synaptic transmission through its inhibition of the presynaptic release of excitatory neurotransmitters. In a rat model of limbic status epilepticus, an
adenosine A1 receptor agonist antagonized the development of status (150). Evidence from in vivo microdialysis experiences, performed intraoperatively on humans, suggests that adenosine may be a potential mediator of seizure arrest and postictal refractoriness (151). The clinical use of adenosine and of adenosine analogues has not been investigated extensively (152), and there is evidence that systemically administered adenosine analogues do not cross the blood–brain barrier (153). Moreover, the widespread effects of adenosine in the brain may not permit the development of a highly specific anticonvulsant with minimal CNS side effects.

Adrenergic neurotransmitters are believed to play a significant role in the regulation of cortical excitability; consequently, a number of laboratories have searched for abnormalities in this system. Brain norepinephrine levels are low in several species of epileptic animals, and a decrease in brain norepinephrine concentration increases their seizure susceptibility. Conversely, an increase in brain norepinephrine levels decreases seizure severity (154). The lesioning of noradrenergic projections from the locus ceruleus lowers the threshold for forebrain and brainstem seizures (155). The importance of the role of brain stem catecholaminergic projections also is suggested by experiments in cats invoving the vagus nerve stimulator (VNS). The efficacy of VNS was lost when the locus coeruleus was lesioned by 6-hydroxydopamine (156).

The serotonergic system also influences the expression of seizures. In view of the recent popularity of highly selective serotonin-uptake antagonists as antidepressants, it is of interest that fluoxetine (Prozac) appears to have anticonvulsant effects (157,158,159). Experiments have suggested that the antiepileptic action of fluoxetine on CA1 neurons is caused by an enhancement of endogenous serotonin that in turn seems mediated by 5-hydroxytryptamine-1A (5-HT1_A) receptor (160,161). Interestingly, in a genetic rat model of absence epilepsy, a 5-HT1_A agonist increased spike waves in a dose-dependent manner (162). Genetically altered mice lacking the gene for 5-HT2_C receptor seem to exhibit spontaneous seizures, while the wild type background could be made to mimic such behavior by the administration of a specific 5-HT2_C antagonist (163).

During a brief seizure, the brain undergoes several biochemical alterations. Cerebral oxygen and glucose consumption increase strikingly, but with maintenance of adequate ventilation, the increase in cerebral blood flow is sufficient to meet the increased metabolic requirements of the brain.

These studies, derived from experimental animals, have been confirmed in the epileptic human by PET. Autoradiography using labeled 2-deoxyglucose as a substrate provides excellent information on the local cerebral glucose metabolism. Using this technique, cerebral metabolism is generally found to be increased in the area of the epileptic focus during a seizure. In some instances, the area of increased metabolic activity extends to adjacent tissue and into areas of neuronal projection (164). In absence seizures, a marked and diffuse increase in cerebral metabolic rate occurs (165). The hypermetabolism probably reflects the enhanced excitatory and inhibitory neuronal activity during a seizure.

By contrast, the interictal metabolic picture reveals areas of hypometabolism (128,166). These observations, important to the localization of epileptic foci, are discussed in a subsequent section of this chapter.

When a generalized seizure lasts 30 minutes or longer, it is usually accompanied by apnea. Apnea induces hypoxia and carbon dioxide retention. As a result of the energy demands of the convulsing muscles, the subject becomes hypoxic and hyperpyrexic (167). Oxygen tension within the brain decreases, a shift toward anaerobic metabolism occurs, and lactic acid accumulates (168). MR spectroscopy performed on rabbits subjected to status epilepticus corroborates these data. Phosphocreatine levels decrease, inorganic phosphorus levels increase, lactate increases, and intracellular pH decreases within 30 minutes of the onset of seizures (169). The increase in cerebral lactate after prolonged seizures results from activation of phosphofructokinase, which is a primary regulatory enzyme for cerebral glycolysis (169).

A generalized tonic-clonic seizure, occurring as a manifestation of primary generalized epilepsy, occurring as a secondary generalization of a partial epilepsy, or alternating with other seizure forms is the most common epileptic manifestation of childhood (see Table 14.3) (170). These conditions do not represent a homogeneous group but are seen in a variety of clinical settings.

The seizure can be primary generalized or secondarily generalized. A primary generalized seizure can occur without warning, whereas a secondarily generalized seizure can be preceded by an aura. Occasionally, the child might be irritable or might manifest unusual behavior for several hours before the seizure. In localizing the epileptic focus, the aura offers the most important clinical clue, more reliable at times than the EEG. The examining physician should always try to elicit its history. The most common epileptic aura is a sensation of dizziness or an unusual feeling of ascending abdominal discomfort. These sensations have been attributed to a discharge in the area of visceral sensory representation but offer less evidence for the site of the epileptic focus than do focal sensory symptoms (171).

In the classic form of an attack, the aura, if present, can be followed by rolling up of the eyes and loss of consciousness. A generalized tonic contraction of the entire body musculature occurs, and the child can utter a piercing, peculiar cry, after which he or she becomes apneic and cyanotic. With the onset of the clonic phase of the convulsion, the trunk and extremities undergo rhythmic contraction and relaxation. As the attack ends, the rate of clonic movements slows, and finally, the movements cease abruptly. The duration of a seizure varies from a few seconds to half an hour or more. In the series of Shinnar and coworkers, new-onset seizures lasted longer than 5 minutes in 50% of children. In 29% of children, they lasted longer than 10 minutes; in 16%, they lasted longer than 20 minutes; and in 12%, they lasted longer than 30 minutes. The authors pointed out that the longer a seizure lasts, the less likely it will stop spontaneously (172). This is commensurate with our experience.

A series of attacks at intervals too brief to allow the child to regain consciousness between attacks is known as status epilepticus. Because status epilepticus is one of the few neurologic conditions requiring emergency treatment, it will be referred to again in the section on treatment.

After the seizure, the child can remain semiconscious and then confused for several hours. When examined soon after an attack, he or she is poorly coordinated, with mild impairment of fine movements. Truncal ataxia, increased deep tendon reflexes, clonus, and extensor plantar responses can be present. Occasionally, the child appears blind and speechless. Postictally, he or she may vomit or complain of severe headache.

The major motor attack has numerous variations. Occasionally, particularly when drug therapy has been partly effective in controlling a secondarily generalized seizure, a typical aura occurs but is not followed by a seizure. In other patients, either the tonic or the clonic phase is too brief to be noted. Attacks can occur at any time of the day or night, although their frequency is somewhat greater shortly before or after the child falls asleep or awakens. Approximately one-fourth of patients experience nocturnal seizures; the remainder experience diurnal or mixed seizures. Generally, patients who for 1 year or more have experienced seizures only during sleep are unlikely to have attacks at other times of the day. In the experience of D’Allessandro and colleagues, who studied both pediatric and adult patients, the risk of a daytime seizure during a six-year follow-up was 13% (173). In some girls, seizures occur a few days before or shortly after their menstrual period.

A generalized tonic-clonic seizure or other epileptic manifestations can be precipitated by infection and fever, fatigue, emotional disturbances, hyperventilation and alkalosis, and drugs.

Fever can induce a seizure not only in children experiencing febrile seizures, but also in patients who previously had recurrent epileptic attacks unassociated with fever. In some children, dehydration and ketosis accompanying an acute infectious illness decrease seizure frequency.

In a few children, excessive fatigue or lack of sleep appears to precede a seizure but probably does not represent an important precipitating factor. Considerable clinical evidence indicates that epileptic children have fewer seizures when engaged in regular strenuous physical activity. Although fatigue seems to have little effect on the EEG, sleep deprivation can activate the EEG of epileptic patients and can precipitate seizures (174).

Parents often believe that emotional disturbances precipitate seizures in epileptic children. No evidence supports this idea, however, and there is no justification for parents’ failing to set limits to the behavior of their epileptic child.

Hyperventilation and alkalosis induce absence attacks in approximately 90% of patients subject to them but are less effective in precipitating other seizure forms.

A variety of drugs can induce single seizures or status epilepticus. In the experience of Messing and coworkers, isoniazid and psychotropic medications accounted for approximately one-half of cases (175). Other drugs implicated in bringing on seizures include cocaine (176), theophylline, and penicillin. Isoniazid (INH) antagonizes glutamic acid decarboxylase by binding with its cofactor, pyridoxal phosphate, thus interfering with the biosynthesis of GABA. Demonstrable decreases in brain GABA levels result from INH administration. An increasing number of patients has presented with status epilepticus and encephalopathy attributable to INH overdoses (176,177). In these cases, immediate administration of pyridoxal (vitamin B₆) is crucial. Theophylline acts as an adenosine antagonist, whereas penicillin acts as a GABA antagonist and interferes with benzodiazepine binding in the brain (152). For INH and theophylline, the convulsive effects tend to be dose related (175). Various anticonvulsants, notably phenytoin, when given in toxic amounts,
increase seizure frequency. Trimethadione, which was formerly used in the treatment of absence epilepsy, occasionally precipitated a grand mal seizure within 1 to 2 days of first being administered. The sudden withdrawal of anticonvulsant medication, particularly barbiturates and benzodiazepines, is the most common cause for status epilepticus. Finally, adolescent patients should be warned against excessive alcohol consumption.

Spontaneous variations in the frequency of attacks are common and must be taken into account when judging the effectiveness of anticonvulsant medication. In the experience of van Donselaar and colleagues, who studied untreated tonic-clonic seizures in children younger than 16 years of age, 42% showed a decelerating pattern, and ultimately became seizure free even in the absence of anticonvulsant treatment. An accelerating pattern of seizure frequency was seen in only 20% of children with four or more untreated seizures (178). Some children have seizures during early childhood, then remain asymptomatic until puberty, when their attacks recur for 4 to 7 years.

Typical absence seizures are most commonly identified with absence (petit mal) epilepsy. In its pure form, an absence (petit mal) attack was defined by Gowers as a “transient loss of consciousness without conspicuous convulsions” (48). Absence seizures are relatively uncommon; they comprise between 2 and 11 percent of seizure types in all ages (179).

The onset of seizures is usually abrupt, and the child suddenly develops an estimated 20 or more attacks each day. The characteristic attack is a brief arrest of consciousness, usually lasting 5 to 10 seconds, appearing without warning or aura. There can be a slight loss of body tone, causing the child to drop objects from his or her hand, but this loss is rarely profound enough to induce a fall. Minor movements occur in approximately 70% of patients; these are usually lip-smacking or twitching of the eyelids or face, often at the three-per-second frequency of the EEG abnormality (Fig. 14.3). Urinary incontinence is rare (180). The seizure terminates abruptly, and the patient is often unaware of the lapse of consciousness. The occurrence and duration of a lapse can be determined by reciting to the child a series of numbers during the attack and asking the child to repeat them when consciousness appears to have returned. Attacks first appear during childhood; in 64% of instances, they begin between 5 and 9 years of age. They are more common in girls; Dalby’s series had 99 female and 62 male subjects (181).

When attacks are frequent, the child’s intellectual processes are slowed and, often, the first indication of the presence of absence seizures is a deterioration in schoolwork and behavior (182). Responsiveness to auditory stimuli is impaired in the majority of patients in 0.1 to 0.4 seconds after the onset of a generalized spike-wave
paroxysm, but it recovers in 2 to 3 seconds after the cessation of the attack. Extrapolations from spike-wave discharges associated with focal epilepsies have implicated the slow-wave component of the spike-wave complex in the interrupted cognitive function (83). Even brief and clinically undetectable attacks can impair intellectual performance.

Some absence attacks are more complex, and their appearance can be difficult to differentiate from complex partial seizures. They involve brief behavioral automatisms or, less commonly, prolonged symmetric myoclonic movements of the head or extremities (myoclonic petit mal). In the series of Sato and coworkers, some 40% of patients with typical absence seizures also experienced a grand mal attack, either before or after the onset of their absence attack (183). This figure is probably high and perhaps reflects the patterns of referral to a university center.

On a clinical basis, absence attacks must be distinguished from brief complex partial seizures and daydreaming. Complex partial seizures are often preceded by an aura and followed by postictal depression. Additionally, brief complex partial seizures tend to be less frequent and clustered and rarely will occur more than twice a day. Routine EEG studies in most instances clarify the situation. In a child with typical absence attacks, the EEG demonstrates a 3-Hz spike-wave discharge, which occasionally slows as the seizure progresses. The interictal EEG is normal in approximately one-half of patients, and the background EEG frequency is generally age appropriate (184). Approximately 10% of children with clinical and EEG features of typical absence attacks have a focal onset to their EEG seizures and demonstrate focal cortical lesions (185). Attacks of daydreaming also can occur several times a day and can last for several seconds to minutes. Automatisms are absent, there is no postictal depression, and attacks are not induced by hyperventilation or photic stimulation. The EEG is normal.

Typical absence seizures also should be differentiated from atypical absence attacks, which are associated with the Lennox-Gastaut syndrome (atypical petit mal, petit mal variant). This relatively common seizure type is associated with a variety of CNS insults, and the majority of affected children have a significant developmental delay. Unlike typical absence attacks, the frequency of atypical attacks is cyclic, in that seizure-free periods alternate with days or weeks of a high seizure frequency. The EEG shows complexes occurring at 1.5 to 2.5 Hz or multiple spike and wave discharges. Diffuse slowing of background activity was seen in 85% of children in the series of Holmes and coworkers (184).

Absence attacks are a common prelude to juvenile absence epilepsy and start about one to nine years earlier. Its characteristic features consist of typical absence attacks, seen in approximately one-third of patients and commencing later than those of typical childhood absence epilepsy. Brief myoclonic jerks that generally occur on awakening usually develop around 15 years of age (i.e., several years after the appearance of absence attacks), and tonic-clonic seizures follow some months thereafter. Attacks tend to occur much less frequently than they do in absence seizures, usually once a day or less (185). Intelligence is preserved. The interictal EEG in juvenile absence epilepsy demonstrates discharges not only at 3 Hz, but also at 4 to 6 Hz (186,187,188,189). The background activity is usually normal. In the series of Appleton and coworkers, photosensitivity was seen in 90% of patients (190).

Experimental studies designed to clarify the mechanism for absence seizures have already been reviewed (127,128,191).

Complex partial seizures have been defined as seizures that arise from a limited area of one cerebral hemisphere and produce a period of impaired consciousness that varies from mild to profound. Although these seizures are most characteristically associated with lesions of the temporal lobe and have been called temporal lobe seizures, they also can be associated with lesions of the frontal (192) or occipital lobes (193). As the terminology indicates, they are focal seizures.

The epilepsies characterized by this seizure type are heterogeneous. Pathologic alterations seen within the surgically resected temporal lobe have been summarized in Table 14.6 (194).

The most common abnormality is MTS (Ammon’s horn sclerosis or hippocampal sclerosis). In the series of Harvey and colleagues, it was seen in 21% of children aged 15 years or younger with new-onset temporal lobe epilepsy (195). The association of this characteristic lesion with both epileptogenicity and seizure-related damage has been discussed in the Neuropathologic Factors section of this chapter. Less commonly, 13% of cases in the series of Harvey and colleagues (195), a variety of tumors in the epileptogenic cortex can occur. These include hamartomas, which occasionally undergo malignant transformation, small gliomatous nodules, hemangiomas, and lesions suggestive of tuberous sclerosis (196,197). In early childhood, however, tumors are the most common etiology for complex partial seizures. Thus, in the series of Wyllie and colleagues, comprising children younger than 12 years of age who underwent temporal lobectomy, tumors were seen in 64%, as compared with MTS, seen in 29%. It is of note that 75% of children with MTS in the series of Wyllie and coworkers had a history of previous febrile convulsions (198). When bilateral MTS develops early in life, usually after prolonged seizures or status epilepticus, the clinical picture is marked by loss of language, or failure of language
development, and impaired social and adaptive learning (199).

In a significant proportion of patients with complex partial seizures, focal abnormalities are found outside the hippocampus, in the limbic portion of the frontal lobes, in the lateral temporal lobe, and in nonlimbic areas outside the temporal lobe (200). In these instances, the ictal discharge probably spreads from the focus to involve the temporal lobe.

Even though seizures start before 10 years of age in approximately 75% of children (95), typical complex partial seizures are rarely seen until 10 years of age. Rather, children who later develop complex partial attacks can have an antecedent history of convulsive seizures associated with chewing, lip-smacking, or other oral automatisms (201). Additionally, they can experience a variety of behavior disorders, enuresis, nightmares, and sleepwalking (202).

Seizure manifestations in older children are presented in Table 14.7. The aura can consist of a variety of subjective phenomena. In children, Glaser and Dixon have found intense anxiety usually associated with visceral sensations to be the most common antecedent to complex partial seizures (203). Wyllie and colleagues have noted that a large proportion of children complain of a “funny” or bad taste (198). An epigastric “rising” sensation also is common. Olfactory hallucinations (uncinate fits) are usually described as unpleasant but unidentifiable odors. Their association with temporal lobe tumors is a matter of some dispute. In Daly’s series, almost 40% of 55 patients who experienced this aura were found to have neoplasms (204). Similarly, in the series of Acharya and colleagues, 73% of patients with such an aura harbored a tumor (205). By contrast, Howe and Gibson found a tumor incidence of only 8.1% in their series of 37 patients, which is comparable with the 9.1% overall incidence of gliomas in patients with temporal lobe seizures (206).

Hallucinatory experiences, most commonly a feeling of déjà vu, an adventitious sense of familiarity, and visual hallucinations, have been reported by children with complex partial seizures (207). According to Mullan and Penfield, they occur more frequently when the focus is in the nondominant temporal lobe (208).

Paroxysmal emotional states, particularly fear, are not rare in children and are commonly reported by parents. Rage reactions or temper tantrums are unusual auras of a complex partial seizure, and purposeful aggressive acts are uncommon in the course of seizure. A detailed history of the aura can assist in localizing the seizure focus, but does not help in lateralizing it. Whereas experiential auras, such as fear and déjà vu, almost invariably originate from the temporal lobes, notably the neocortex (200), cephalic auras, such as a sensation of dizziness or lightheadedness are of less localizing value and can emanate from either frontal or temporal areas. Viscerosensory auras were found to accompany a temporal lobe focus in 76% of subjects, and somatosensory auras accompanied a parieto-occipital focus in 62% (209).

On the basis of the initial seizure manifestations, Escueta and associates have divided complex partial seizures into two types (210). In the more common form, type I, the seizure originates from the temporal lobe. Patients briefly stop all activity after their aura. They stand still, stare, or turn pale. Shortly thereafter, minor motor acts are initiated. Prolonged postictal confusion is common in these patients. In type II, seizures originate from outside the temporal lobe, most commonly from the frontal lobe. In this type, automatisms initiate the attack. Commonly, these involve chewing and smacking movements, purposeless fumbling or patting of the hands, and picking at clothes. Postictal confusion is brief in this group of patients. Drop attacks as part of temporal lobe seizures are unusual in childhood (211).

The final part of the seizure generally involves more complex motor acts. The child might move about the
room, begin to undress, and occasionally utter stereotyped or nonsensical phrases. In the majority of cases, these automatisms usually do not last longer than 5 minutes, although reliable observers have recorded prolonged complex partial seizures. Complex partial status epilepticus (psychomotor status) is extremely rare, and to our knowledge, it has never initiated complex partial epilepsy (212,213). In children, the condition manifests by impaired consciousness with intermittent staring and wandering eye movements and intermittent automatisms, such as picking at clothes. At other times, it can result in a prolonged period of amnesia. The condition should be differentiated from absence status or hysterical amnesia. Depth electrode studies suggest that the seizure focus is more commonly within the frontal lobes (214). A variety of electrocardiographic abnormalities can accompany complex partial seizures. These can be responsible for some of the sudden deaths seen in inadequately treated epileptic patients (215).

Following the seizure, the patient experiences postictal confusion, drowsiness, or clouding of consciousness. When fully recovered, the child has complete amnesia for the entire attack.

As with major motor seizures, the frequency of complex partial seizures varies, but in contrast to typical absence attacks, more than one to two attacks a day is uncommon (203).

The possibility of complex partial seizures is often raised in a child with behavior problems who has an abnormal EEG result. Aird and Yamamoto examined this question and found that approximately one-half of children with behavior problems have an abnormal EEG result. Of all patients studied, 27% had EEG foci primarily involving the temporal lobe (216). Although some authors vigorously dispute this predilection to behavior disturbances (217), others support it and have found behavior disturbances to be three times as common in this as in the other types of epilepsy (218). Additionally, symptoms suggesting complex partial seizures are highly prevalent among violent juveniles (219). Several further studies have confirmed these observations. In a series published in 1980, the incidence of abnormal behavior was 36% (220). Schoenfeld and coworkers, however, found that children with complex partial seizures are not aggressive or delinquent. Rather they suffer from social withdrawal, various somatic complaints, anxiety, and depression (221). The cause of the seizures, their duration, and the presence of associated grand mal attacks do not seem to affect the likelihood of psychiatric disturbances; however, the frequency of seizures and early seizure onset do (221). Boys whose seizure focus is located in the dominant hemisphere and whose seizures commence between 5 and 10 years of age appear to be particularly vulnerable (218). In most instances, the psychiatric illness commences during adolescence; its manifestation before 12 years of age is unusual.

The basis for these psychiatric phenomena has undergone considerable speculation. The most attractive hypothesis states that recurrent complex partial seizures kindle a limbic dopamine system, whose paroxysmal activity does not produce seizures but does produce serious behavior disturbances (222). The role of limbic kindling in the genesis of psychiatric illness is outside the scope of this text. Adamec and Stark-Adamec and Weiss and Post review this subject (223,224).

Focal seizures are characterized by the development of localized motor or sensory symptoms without impairment of consciousness. In a large proportion of children, these seizures spread to other parts of the body, ultimately becoming generalized with loss of consciousness. On rare occasions, progression follows an orderly sequence, a phenomenon known as the jacksonian march. This type of a seizure is generally considered to be symptomatic of a structural cortical lesion.

Perhaps the most common focal attack observed in children is the versive seizure (225). These seizures begin in or spread to area 8, the frontal eye field, and the supplementary motor area or the mesial part of the premotor area. They are manifested most often by a turning of the eyes, or the eyes and head, away from the side of the focus. In some patients, the upper extremity on the side toward which the head turns is abducted and extended, and the fingers are clenched. Thus, the child appears to look at his closed fist.

The patient may be aware of this movement or may simultaneously lose consciousness. By means of combined telemetry and EEG monitoring, the cortical areas of discharge responsible for this form of seizure have been found to be either the contralateral temporal lobe or the contralateral frontal lobe, anterior to the rolandic gyrus. Patients whose seizure starts with versive movements and who retain awareness of them tend to have a frontal lobe seizure focus. Patients whose seizure starts with staring and automatisms tend to have a temporal lobe focus (210, 226,227). EEG changes, as recorded on scalp electrodes, are seen in only approximately one-third of patients during telemetry-verified attacks of a variety of simple partial seizures. Therefore, the presence of an unaltered EEG does not speak against the diagnosis (228).

Focal motor seizures are particularly common in hemiplegic children. The epileptic movements are usually clonic and begin in the hemiplegic hand, often heralded by localized sensory symptoms. The clonic movements spread over the entire affected side, ultimately becoming generalized in many cases. Postictal weakness (Todd’s paralysis) commonly follows this kind of seizure and can last for several hours or for a day or more. In the experience
of Kellinghaus and Kotagal, Todd’s paralysis was contralateral to the epileptiform focus in 93% of subjects (229).

Whereas adults have a high positive correlation between Todd’s paralysis and a structural cortical lesion, children do not, and often attacks of alternating left- and right-sided focal seizures with postictal paralyses are the initial events in a child with apparently idiopathic epilepsy. This observation probably reflects a smaller and less readily detectable focus in the pediatric population.

A heterogeneous group of conditions are characterized by focal (simple partial) seizures. In approximately one-half of patients, one can document an underlying structural lesion. These lesions include developmental malformations, notably gray matter heterotopias and localized pachygyria, gliosis resulting from asphyxial damage or perinatal and postnatal physical trauma, and a variety of space-occupying lesions. In the remainder of patients with simple partial seizures, no obvious etiology can be demonstrated. Several clinically distinct types of these idiopathic focal epilepsies have been recognized. They are characterized by the absence of anatomic or functional focal lesions and by a benign course with normal intellectual development and spontaneous cure. Aside from febrile seizures and a family history of seizures for approximately one-third of children, no obvious antecedents exist. The frequency of these conditions is difficult to ascertain. Roger and Bureau believe they constitute some 60% of partial epilepsies seen in school-aged children (230).

The epileptic syndromes considered in this section share an early onset, an EEG picture characterized by slow spike-wave forms at a frequency lower than the regular 3 Hz, and
poor prognosis, in terms of both seizure control and ultimate intellectual development. Livingston and associates have used the term minor motor seizures to designate the epilepsies described in this section (243). This term has been abandoned in recent years; not only is there nothing minor about these seizures, but the term is overly broad and encompasses a variety of seizures with limited expression, regardless of their etiology and EEG features.

The causes for Lennox-Gastaut syndrome (LGS) are multiple. In the experience of Lennox and Davis, compiled in 1950, 50% of patients had a history of perinatal cerebral injury, another 20% might have had an attack of encephalitis or meningitis, and 13% had major complications of gestation (244). Currently, complications of gestation appear to be the most significant cause, with perinatal asphyxia, infections, and genetic factors following in order of importance.

Infantile spasms most commonly develop between 3 and 8 months of age, with only 8% of cases first being encountered in infants older than 2 years of age. It occurs somewhat more frequently in male infants. Attacks are characterized by a series of sudden muscular contractions by which the head is flexed, the arms are extended, and the legs are drawn up. A cry or giggling can precede or follow the seizure, and the infant can flush or can turn pale or cyanotic.

Other clinical presentations of infantile spasms occur less commonly. They include head nodding and extensor spasms characterized by extension rather than flexion of arms, legs, and trunk. Rarely, the attacks are concluded by a brief clonic seizure.

Lightning attacks (Blitzkrämpfe) (260) are a variant involving a single, momentary, shocklike contraction of the entire body (261).

Clusters of seizures recur frequently, particularly on waking, and some children have 50 to 100 each day. In Jeavons and Bower’s series of 112 children, mental development was normal up to the onset of seizures in 52% and was definitely or probably delayed in the remainder (262). More recent series have had a higher incidence of delayed mental development before the onset of seizures. In 66% of patients, the EEG has the characteristics of hypsarrhythmia, namely diffuse dysrhythmia with high-voltage slow waves and multiple spike-wave discharges (Fig. 14.5) (262). This unique electrical pattern is seen in the early stages of the disorder, becoming apparent after 3 to 4 months of age. In some infants, it is most evident in non-REM sleep. During REM sleep and immediately after arousal from REM or non-REM sleep, the EEG can be normal for up to several minutes (263). The discharges tend to favor the posterior areas of the brain. This contrasts with the paroxysmal discharges in LGS, which are most evident anteriorly. The significance of this finding is not clear, particularly because many children with infantile spasms progress to LGS. To our knowledge, there has not been a systematic study of EEG maturation in infantile spasms. Variations in this pattern are common (modified hypsarrhythmia). They include hypsarrhythmia with a focus of abnormal discharge, hypsarrhythmia with increased interhemispheric synchronization, hypsarrhythmia with little spike or sharp-wave activity, and hypsarrhythmia with episodes of voltage attenuation, a variant termed suppression burst variant by Hrachovy and coworkers (263).

Hypsarrhythmia persists for some years, then is superseded by a variety of other paroxysmal abnormalities. These include focal discharges, most commonly arising from the temporal areas, multifocal spike discharges, and an atypical spike-wave pattern. The evolution of infantile spasms to LGS was commented upon earlier in the discussion of the latter syndrome.

Attacks of spasms are often associated with a discharge of multiple spike-wave or polyspikes. Sudden suppression of electric activity begins with an attack and persists briefly after its completion.

Although infantile spasms are considered as a generalized epilepsy, the neuroanatomic substrates responsible for them are poorly understood (264). A marked decrease in REM sleep has led to the proposal that the basic abnormality is at the pontile level in proximity to centers regulating sleep cycles (265). This hypothesis is supported by the finding on interictal PET scan of hypermetabolism of the brainstem. Additionally, hypermetabolism of the lenticular nucleus occurs, which suggested to Chugani and coworkers a neuronal circuitry involving cortex, lenticular nucleus, and brainstem in the generation of infantile spasms (266).

Infantile spasms are classified as cryptogenic or symptomatic. Cryptogenic spasms, a minority of the cases of infantile spasms [less than 15% in the series of Riikonen (267)], occur in infants with normal birth and development until the onset of seizures, in whom no obvious cause for the convulsions can be demonstrated. A variety of prenatal and perinatal insults are responsible for the majority
of cases in the symptomatic group. Abnormalities of gestation, manifested by a history of maternal infection and prematurity, are particularly common. Perinatal asphyxia or traumatic birth injuries are seen less often. Neonatal seizures are a common precursor, and symptomatic neonatal hypoglycemia is encountered in some 18% of infants (267). In the remainder of the infants, tuberous sclerosis is the major etiologic factor, with intracerebral calcifications detectable by CT scanning in up to 25%. Infantile spasms also are seen in neurofibromatosis, agenesis of the corpus callosum, and metabolic diseases such as phenylketonuria, maple syrup urine disease, and pyridoxine dependency (268). In a small group of infants, spasms begin shortly after pertussis immunization with the whole cell vaccine. Whether this association is fortuitous, as suggested by Bellman and coworkers (269), or whether pertussis vaccine indeed represents a rare cause for infantile spasms will probably never be resolved, and with the switch to acellular pertussis vaccine, this question has become moot (270).

As might be expected, a variety of neuropathologic findings have been reported. These include structural malformations of cortical gray matter. The most commonly observed abnormality observed in surgically resected specimens is a focal cortical dysplasia (271). Other alterations include hamartomatous proliferation of multipotential neuroectodermal cells (272), ulegyria, lissencephaly, unilateral megalencephaly, and pachygyria (273). Focal areas of cortical microdysgenesis are not readily detectable by magnetic resonance imaging (MRI) during the first few months of life but can be visualized indirectly in subsequent studies by a focal delay in maturation of subcortical white matter. This can only be visualized by repeated imaging studies (274). Tumors, cytomegalovirus infection (275) and spongy degeneration of gray and white matter are less common (271,276).

In most children, the outlook for normal intellectual development is poor, even though in approximately 50%, infantile spasms have ceased by 3 years of age or are replaced by the LGS or by major motor attacks (277). According to Jeavons and Bower, none of the patients whose development was already retarded when infantile spasms began had a subsequent normal intellectual development, whereas 29% of infants believed to have developed normally up to the onset of seizures (cryptogenic infantile spasms) were found to be neurologically intact and intellectually in the normal or low-normal range when their infantile spasms disappeared (262). More recent series, such as those of Riikonen (267) and Gaily and coworkers (278), present comparable results. The latter workers observed 67% of infants with cryptogenic infantile spasms to have normal intelligence when tested between the ages of 4 and 6 years. Specific cognitive deficits were found in 42% of children in this group. In the series of Kivity and coworkers, the outcome of children with cryptogenic infantile spasms treated with high doses of adrenocorticotropic hormone prior to mental deterioration was uniformly excellent (279). Riikonen found that in children whose infantile spasms are attributed to tuberose sclerosis, the long-term prognosis is worse than in other symptomatic or cryptogenic cases of infantile spasms (280). By contrast, the prognosis is relatively good when infantile spasms accompany neurofibromatosis (281).

Juvenile absence epilepsy and Juvenile myoclonic epilepsy (JME) are two clinically related conditions that are genetically heterogeneous, probably transmitted as a dominant trait. Several genes have been implicated in this condition. One of them encodes the GABA receptor (GABABR1) and has been mapped to chromosome 6p21.3 (Table 14.5) (315,316). Another gene (EFHC1) has been mapped to chromosome 6p12–p11 and encodes a protein with an EF-hand motif. EF hands are Ca(2+) binding motifs that are widely distributed throughout the entire animal kingdom (317).

Juvenile myoclonic epilepsy has a prevalence of 0.5 to 1.0 per 1,000 and represents some 4% of the primary generalized epilepsies (189).

Symptoms usually become apparent in adolescence, with age of onset between 12 and 18 years (189). Seizures are characterized by sudden myoclonic jerks of shoulders and arms that usually appear shortly after awakening. A majority of patients also experience grand mal attacks. As a rule, myoclonic jerks precede grand mal seizures. Up to one-third of patients also experience absence attacks, which tend to precede the onset of myoclonic jerks (189). Intelligence remains normal. The interictal EEG is characteristic, in that it demonstrates 4- to 6-Hz polyspike and wave complexes, with photosensitivity in approximately 30% (318). Valproic acid is effective in up to 90% of subjects but has to be continued for the remainder of the patient’s life because withdrawal of anticonvulsants results in seizure recurrence in some 90% of patients, even after many years of complete control (319). Alternatives to valproate therapy with newer anticonvulsant medications are being explored and are reviewed by one of us (R.S.) (320).

A number of epileptic conditions are manifested by unusual seizure forms. These are seen too rarely to warrant more than a brief description. The following types deserve mention.

Serum glucose and calcium levels are obtained for all infants with seizures and for older children whose histories raise the possibility of a metabolic disturbance (see Chapter 17). In neonates, an evaluation of renal function also is indicated. Serum electrolyte disturbances are rare in patients with recurrent seizures, but hyponatremia (sodium of less than 125 meq/l) was associated with seizures in 70% of infants under 6 months of age (356). In children with seizures owing to hypernatremia, the history
usually suggests a fluid imbalance. A toxicology study also is indicated.

The EEG can be useful in determining the type of seizure the patient experiences. It differentiates between absence and complex partial seizures, and between absence attacks and atypical petit mal. Occasionally, it can assist in establishing a diagnosis when the history is inadequate or not diagnostic. It also is the best predictor of the likelihood of recurrence. The EEG must be interpreted in conjunction with the patient’s history. A normal EEG, seen in approximately one-half of epileptic patients, does not exclude the diagnosis of epilepsy, nor does an abnormal EEG necessarily establish it. King and colleagues have found that it is easier to differentiate between a primary generalized and a secondarily generalized seizure disorder if the initial EEG is performed within 24 hours of the seizure. They also recommend that should the initial EEG be normal, a second, sleep-deprived EEG should be performed (369). Hirtz and colleagues also state that an EEG obtained within 24 hours of a seizure is more likely to show epileptiform abnormalities and can guide the physician as to whether neuroimaging is necessary (356).

Electric abnormalities can be seen during an overt attack as well as between seizures. Generalized tonic-clonic seizures arising from subcortical excitation can be recorded on the EEG, but the abnormalities are usually obscured by movement artifacts. One exception is the major seizure with a focal cortical origin. In this condition, random focal spiking or spike-wave discharges that had already been seen during the interseizure period increase in frequency and amplitude and spread both contiguously and via subcortical centers. Generalization can occur within a fraction of a second after the onset of the focal seizure; less commonly, it develops slowly over as long as 20 seconds. Following the seizure, electric activity is depressed. Recovery starts with slow waves, normal cortical activity first appearing contralateral to the epileptic focus, then ipsilateral, and finally over the focus itself.

Less commonly, the ictal tracing is marked by suppression of electrical activity, or low-voltage fast activity, an indication of disinhibition.

The EEG during an absence attack shows bilateral high-voltage synchronous alternating spike-wave complexes, most commonly 3 to 4 Hz (see Fig. 14.3). The discharge frequency can be increased by hyperventilation. As the discharge continues, the frequency of the spike-wave complexes tends to slow down.

In most patients with complex partial seizures, the EEG patterns accompanying the seizure are slow, rhythmic, and usually bilaterally synchronous, 4- to 6-Hz discharges, most prominent in the frontal and temporal areas but becoming generalized as the seizure progresses. At the conclusion of the attack, the EEG is featureless, and recovery of normal activity is delayed for many minutes.

The interseizure record in patients with major motor attacks can be normal or can demonstrate a number of nonspecific abnormalities, including disorganization, spike discharges, loss of a-rhythm or loss of the highest expected frequency activity for the child’s age, and, occasionally, continuous or intermittent focal slowing. This last finding, particularly when it is continuous, should always raise the possibility of an underlying tumor or other structural lesion and indicates the need for imaging studies.

In complex partial seizures, the interseizure tracing is normal or shows nonspecific abnormalities or spike foci, which usually arise from the anterior portions of the temporal lobes (Fig. 14.8). Often, the seizure focus is apparent only during the transition between wakefulness and sleep, and a combined wake and sleep tracing should always be obtained from children suspected of having complex partial seizures. In approximately 80% of children with this condition, three serial wake and sleep recordings uncover an abnormality. In a number of patients, secondary spike foci can be recorded from the opposite temporal lobe. It is rare to find focal temporal spike discharges in children younger than 8 years of age, even in those with a classic clinical picture of complex partial seizures.

Atypical spike-wave discharges (i.e., spike-wave discharges at a frequency less than 2.5 Hz) and polyspike-wave discharges are common in children with Lennox-Gastaut syndrome. They indicate a poor prognosi in terms of seizure control and normal intellectual function. The high positive correlation between hypsarrhythmia and infantile spasms has already been discussed.

With EEG maturation, the EEG in patients with seizure disorders can undergo significant changes. Focal spike activity tends to migrate from the occipital to the midtemporal and ultimately to the anterior temporal leads (370). In patients with 3-Hz generalized spike-wave discharges, an improvement in clinical status can be associated with the disappearance of the abnormal EEG pattern, which is often replaced by a wandering spike focus (371). In patients with focal seizure discharges, mirror spike activity can appear. In some, this second focus is suppressed by anticonvulsants; in others, it becomes the only active focus (371).

For further information on the clinical aspects of EEG, the reader is referred to one of the several standard EEG texts.

Lumbar puncture is not performed routinely in patients with seizure disorders. Because a significant proportion of infants with bacterial or tuberculous meningitis have a febrile seizure as their initial symptom (see Chapter 7), we believe a lumbar puncture is indicated in patients who have experienced their first febrile seizure and in all infants having their first seizure.

Although the CSF is by definition normal in the majority of patients with idiopathic epilepsy, minor abnormalities were found in the classic study of Lennox and Merritt (372). In 4% of their patients, pleocytosis was slight (5 to 10 cells/μL), and in 10%, protein content was increased (45 to 85 mg/dL). Both abnormalities were believed to be related to the presence of small areas of cerebral contusion that resulted from a fall attending the seizure. They had disappeared when the spinal tap was repeated several weeks later. More recent studies are consistent with these observations (373,374). Wong and coworkers, who examined the CSF of children within 24 hours of a seizure, found a normal range of 0 to 12 cells in infants under 4 months of age, and a range of 0 to 8 cells in infants over 4 months. The normal postictal protein was less than 100 mg/dl in infants less than 2 months, less than 60 mg/dl between 2 and 4 months, and less than 35 mg/dl in infants over 4 months of age (375). We suspect that abnormalities in cell count and protein level, when present, reflect the cerebral edema, disturbed autoregulation, and the breakdown of the blood–brain barrier that attend a prolonged seizure and that have been documented by MRI (376). In contrast to adults, seizure-induced neuronal injury is not marked by an increase in neuron-specific enolase; rather, increases reflect the presence of metabolic or genetic abnormalities (377).

MRI and CT scans are the procedures currently used in the initial evaluation of a child with a seizure disorder. The question as to whether an emergent imaging study should be performed as part of the evaluation of the child that presents with an afebrile seizure has been considered by several authors. In the experience of Maytal and coworkers, only 6% of children who presented with cryptogenic seizures had an abnormal CT. By contrast, the CT was abnormal in 60% of children whose seizures were considered to be symptomatic (378). Similar results were obtained by Sharma and coworkers (379). From studies such as these, it is evident that neuroimaging of children with the first afebrile seizure should be reserved for those who have conditions that place him or her at an increased risk for neuroimaging abnormalities. These include an abnormal neurologic examination, a history of malignancy, sickle cell disease, bleeding disorder, a closed head injury, or travel to an area endemic for cysticercosis (379). Infants under 33 months with focal seizures should undergo imaging as well. In this group, in the experience of Sharma and coworkers, CT abnormalities were seen in 29% (379). By contrast, only 2% of CTs were abnormal in children with nonfocal seizures and no predisposing conditions (379). Hirtz and colleagues found that the yield of abnormalities on CT scan is low in the presence of a normal EEG and a normal neurological examination, and that abnormalities when present do not influence treatment (356).

An imaging study is indicated for the child with recurrent seizures under several circumstances: in the presence of an abnormal neurologic examination, including dysmorphic features, or skin lesions suggestive of a phakomatosis (64%); in the presence of focal EEG abnormalities, particularly focal slowing (63%); with a history of neonatal seizures (100%); and with a history compatible with simple partial or complex partial seizures (52% and 30%, respectively). The numbers in parentheses represent percentage of positive CT scan results seen by Yang and coworkers in their series of seizure patients (380); the percentages of positive combined CT and MRI studies would undoubtedly have been higher.

Whether imaging studies also are indicated for children who by history and examination do not fulfill these criteria is a matter of individual judgment. Children whose neurologic examination and EEG are normal have a low yield of positive scan results (356), as do children with primary generalized seizures [5% and 8%, respectively, in the series of Yang and coworkers (380)]. In confirmation of this experience, King and coworkers found no MRI abnormalities in any of their patients with EEG-confirmed primary generalized seizures (369).

Several studies have compared CT scans and MRI for their respective advantages and disadvantages in evaluating a seizure patient. Although the two procedures occasionally provide complementary information, MRI is the preferred initial screening procedure. Unlike CT scanning, MRI detects congenital malformations of the cortex, gray matter heterotopias, and a large proportion of arteriovenous malformations, some of which can be angiographically occult. MRI also is an excellent screening test for detecting neoplasms, particularly when it is used with gadolinium enhancement. On the other hand, CT scans are more sensitive for small foci of calcification.

The joint use of EEG and imaging studies has been applied most successfully in the evaluation of patients with focal seizure disorders. MRI is effective in detecting underlying abnormalities in the cortical architecture of children with simple partial epilepsies (369,381) (Fig. 14.9). In young children with intractable partial epilepsies, malformations of the cortex are detectable on MRI. These abnormalities are most common in the central cortical area, particularly in the region of the operculum. The complexity of cortical gyration makes difficult the detection of subtle abnormalities, and when MRI is correlated with pathologic examination of brain tissue resected from epileptic seizure foci, only approximately one-half of the malformations were detected before surgery by MRI (382). MRI results are abnormal in most patients with severe mesial temporal sclerosis as well as in a large proportion of subjects with pathologically proven mild to moderate sclerosis (383,384). These abnormalities are best demonstrated with T2-weighted or fluid-attenuated inversion recovery (FLAIR) images. Additionally, MRI can detect small epileptogenic lesions in the temporal lobe, notably hamartomas and low-grade gliomas (385,386). When children with newly diagnosed epilepsy were subjected to both MRI
and CT, both studies were abnormal in 83%, and in the remainder, the CT was normal but the MRI was abnormal. The MRI also found a number of incidental abnormalities, such as Chiari 1 malformations, pineal cysts, and arachnoid cysts (387).

Transient abnormalities at the site of an actively discharging epileptic focus can be visualized by MRI, particularly on T2-weighted images (388). These abnormalities can reflect hyperemia or can be caused by a shift of water from the extracellular compartment into dendrites or glia (389). Similar changes also can be visualized on CT scans. The abnormalities take from 1 to 48 hours to develop and can resolve within 1 week or as late as 12 weeks (390). Additionally, some drugs, notably digoxin and heparin, can increase the signal intensity on MRI (391).

PET scanning provides three-dimensional functional images of the brain. Although in most centers PET scanning is still a research study, this procedure has become part of the diagnostic evaluation of a youngster with focal seizures or with intractable generalized seizures and assists in the surgical treatment of intractable epilepsies (392). Fluorodeoxyglucose, which measures glucose metabolism, is the most commonly used tracer. The interictal PET scan demonstrates focal areas of hypometabolism corresponding to the site of the epileptic focus as determined by EEG, and to focal changes on MRI (393) (Fig. 14.10). Juhasz and coworkers have noted, however, that the hypomtabolic area itself is not involved in epileptic activity, but rather arises from the surrounding cortex (394). During a seizure, this area of hypometabolism becomes hypermetabolic. When the seizure is generalized, such as in children with absence attacks, hypermetabolism is generalized (395). Focal areas of hypometabolism have been found in children with infantile spasms and in the Lennox-Gastaut syndrome (304). The relative value of localizing a seizure focus by means of EEG, MRI, and PET has been a matter of some debate. Most centers, including ours, consider PET to be more likely to detect a focal temporal abnormality than structural imaging studies, and we have seen a number of patients with PET abnormalities in whom MRI had been normal (396). The experiences of Won and colleagues with respect to the comparative ability of MRI, interictal PET, and ictal single photon emission CT (SPECT) show that PET was the best study to correctly localize the epileptic focus. SPECT and MRI were equally useful (397). In our experience, and that of Theodore and colleagues, the anatomic distribution of neuronal cell loss correlates poorly with the degree and the spatial extent of PET hypometabolism (398).

The advantages and drawbacks of SPECT in the localization of epileptic foci are considered in the Introductory chapter.

In the majority of instances, abnormal imaging studies do not influence treatment of the child with a seizure disorder. For the physician who must practice defensive medicine under the legal sword of Damocles, these logical guidelines are, however, of little help.

Cerebral angiography, requiring hospitalization, is hardly ever indicated, except to determine the vascular anatomy as part of the preoperative evaluation before removal of a seizure focus. In many cases, exceptionally vivid visualization of vascular structures can be achieved noninvasively with spiral (helical) CT (399).

In contrast to EEG, which records the electrical potentials, magnetoencephalography (MEG) records the magnetic fields generated by the brain. Although these fields are extremely small, the magnetic fields generated by epileptiform discharges can be localized three-dimensionally, particularly when the focus is in the more superficial areas of the cortex. At this time, the procedure is primarily a research tool, and because of its high cost and complexity of equipment, the procedure does not yet offer any significant advantages over other routine diagnostic studies (400,401). However, it has been the experience of most centers involved in the surgical treatment of epilepsy that the combination of EEG and MEG is superior to each in isolation in terms of localizing the site of a seizure discharge.

A number of tests can be used to localize an epileptogenic focus in patients whose seizures have remained intractable with standard medical therapy or in whom the epileptic nature of recurrent episodes has not been clarified. Several techniques have been used for long-term EEG monitoring. The most commonly employed is a combined EEG telemetry and video monitoring (402). For this procedure, the child is admitted to the hospital and is continuously monitored until enough ictal events have been recorded to determine the epileptic nature of the attacks, the epileptic seizure type, and the site of onset. Because the monitoring system is portable, the child is free to move around the ward and can engage in usual ward activities (403). Prolonged (8- to 10-hour) EEG recordings in the laboratory are somewhat simpler but require considerable cooperation by the patient. An ambulatory cassette recorder allows the child to remain at home or attend school, while the parent or teacher keeps a log of the activities. We have found the ambulatory technique useful to determine whether or not recurrent episodes are epileptic but of lesser value in the localization of the seizure focus.

Sphenoidal electrodes have been used to localize epileptic discharges generated by mesial temporal sclerosis in patients with complex partial seizures. Although electrodes are fairly easily placed in the cooperative adult, their use in children requires analgesia and sedation. Once in place, the electrodes are generally well tolerated. For most situations, anterior cheek electrodes provide adequate sensitivity for mesial temporal discharges. Other procedures used to localize an epileptic discharge and select patients for surgical ablation include thiopental activation intended to evoke focal attenuation of barbiturate-induced fast activity (404), the use of methohexital to distinguish between primary and secondary epileptic foci (405), and the pharmacologic ablation of a hemisphere with intracarotid amobarbital.

The activation of focal epileptiform discharges with pentylenetetrazol is no longer done in most institutions.

Metabolic screening studies and cytogenetics are performed when seizures are coupled with mental retardation of indiscernible cause or are associated with periods of prolonged impairment of consciousness. Generally, screening of plasma amino acids and other metabolites is combined with determination of urine organic acids, blood ammonia levels, and blood and CSF lactate and pyruvate levels. Several cytogenetic abnormalities are coupled with a seizure disorder. These are reviewed in Chapter 4.

The objective in the treatment of the epileptic patient is complete control of seizures, or at least a reduction in their frequency to the point at which they no longer interfere with physical and social well-being. Two aspects of therapy are discussed: who to treat and how to treat.

Although the value of fasting in the treatment of seizures has been recognized since biblical times, the ketogenic diet, which attempts to reproduce the ketosis and acidosis of starvation, was introduced only in 1921 (632). Despite the development of a variety of new and effective anticonvulsants in this country and abroad, it remains an attractive alternate means for the treatment of Lennox-Gastaut syndrome and other intractable seizures (633). The diet involves restricting protein and carbohydrate intake and supplying 80% to 90% of caloric intake through fats. The mechanism by which this regimen controls convulsions is still unknown. It is independent of respiratory or metabolic acidosis and of the accumulation of ketone bodies, and the anticonvulsant effects do not result from the direct effect of ketone bodies on voltage- and ligand-gated ion channels (634). Caloric restriction as well as the ketogenic diet reduce neuronal excitability, perhaps by inducing the activity of D-3-hydroxybutyrate dehydrogenase and allowing the brain to metabolize ketone bodies (635). It is still unclear if the benefits of the diet in seizure disorders stem primarily from the switch in cerebral metabolism involving the use of β-hydroxybutyrate instead of glucose as an energy substrate. In that respect, it is of interest that the Atkins diet also can reduce seizure frequency in focal and multifocal epilepsy (636). As well, there is accumulating experience suggesting that a low-glycemic diet may provide seizure control without a high degree of ketosis (637). A novel hypothesis has been advanced that lowering the rate of glycolysis decreases the production of cytosolic ATP, which inhibits current mediated by K⁺ channels that are sensitive to that nucleotide trophosphate. Increased outward K⁺ currents through these channels provides a hyperpolarizing mechanism. Research is under way to clarify this concept. The metabolic and endocrine aspects of this diet have been reviewed in detail by one of us (R.S.) (638).

The diet is most effective in children with minor motor seizures between 2 and 5 years of age. Older children do not respond as well because they fail to maintain an adequate degree of ketosis. We reserve the ketogenic diet for children who have been unable to tolerate anticonvulsant drugs because of multiple allergies. Details of the induction and maintenance of the diet are presented by Nordli and De Vivo (639). Medium-chain triglycerides as a substitute for a ketogenic diet have been advocated by Huttenlocher and associates (640). We have had relatively little experience with this regimen, but it appears to be as effective as the ketogenic diet and is better tolerated by some children. The need for initial fasting and fluid restriction has been questioned by Kim and coworkers, who find that the percentage of patients who become seizure-free for at least three months after institution of a non-fasting diet (34.1%) is no less than that obtained by using a fasting protocol (34.9%) (640a).

The increasing popularity of the diet in the 1990s prompted a multicenter study to evaluate its effectiveness (641). This study found that 10% of the patients became free of seizures on the diet. One-half the patients received considerable benefit from the diet, whereas the other
one-half had dropped out of the diet by the end of the year, either because of lack of efficacy or adverse effects. There have been concerns among clinicians as how to best screen patients for safety with this diet. To state that physicians should avoid placing a child with seizures on the diet if the child also demonstrates hypotonia and developmental delay, with metabolic acidosis and abnormalities of organic or amino acids, is too general to provide adequate guidance. A child with such a presentation could have pyruvate dehydrogenase deficiency, and hence may stand to benefit significantly from the ketogenic diet. On the other hand, a child with a primary or acquired carnitine deficiency or a defect in the β-oxidation of fatty acids sustains significant morbidity if fasted or placed on the diet. If a patient has undergone a muscle biopsy earlier in the workup for hypotonia and weakness, a biopsy showing a few ragged-red fibers may not be a contraindication to the diet, but the appearance of a lipid myopathy is. Increased serum pyruvate and lactate along with alaninuria may not be a problem, but increased dicarboxylic acid excretion points to a potential for trouble. We, therefore, suggest that in the presence of an abnormal pattern of urinary organic acids, the type of dicarboxylic aciduria should be determined before the child is started on a ketogenic diet. We should point out that dicarboxylic aciduria and carnitine deficiency can occur in a patient who is on both valproic acid and the ketogenic diet. In addition, children who present with a history of seizures and intermittent encephalopathy in the course of minor infections should be carefully evaluated for the nature of the metabolic abnormality before considering the ketogenic diet.

We agree with Demeritte and colleagues (642) that organic acid screening be done before and after the initiation of the ketogenic diet. However, we do not recommend that patients on the diet be supplemented routinely with carnitine. Rather, we agree with recommendations from a roundtable discussion on the role of L-carnitine supplementation in children with epilepsy that suggests that carnitine supplementation be reserved for children with demonstrable carnitine deficiency (643). Even though most centers offering the ketogenic diet have patients who also are on valproate, one study has hinted at the added risk for complications in such patients (644). Other complications include the development of renal calculi, osteoporosis, hypoglycemia, and, in about one-third of children, increased bruising with a prolonged bleeding time (645).

Phenobarbital can be administered in the presence of severe hepatic disease. Even though hepatic hydroxylation of the drug is impaired, with consequent doubling of its half-life, increased renal elimination of unchanged phenobarbital reduces the importance of hepatic metabolism.

Carbamazepine levels are altered in hepatic disease. Metabolism of the drug is reduced significantly, reducing its clearance. This effect is counteracted by reduced plasma binding and increased free carbamazepine. Generally, however, levels tend to increase and must be monitored carefully, keeping in mind that the free carbamazepine fraction is higher than normal.

No modification of clearance or bioavailability of carbamazepine has been observed in renal failure. In cardiac failure, carbamazepine absorption, which is normally erratic, is reduced. However, the drug is metabolized more slowly than usual, thus reducing its clearance. Carbamazepine can cause increased sodium and water retention, which may aggravate cardiac symptoms.

In cirrhosis and renal disease, the free fraction of valproic acid increases two- to threefold. The intrinsic metabolism of the drug is reduced, however, so that the actual clearance remains essentially normal.

Among the newer agents, gabapentin, topiramate, levetiracetam, zonisamide, oxcarbazepine, and lamotrigine offer the benefits of not having significant systemic toxicity, protein binding, or hepatic induction. Of those, oxcarbazepine and higher doses of topiramate have the potential to induce to some measure cytochrome P-450 C3A4 isoenzyme, while there is some chance of a rash with zonisamide, and even more so with lamotrigine. Thus, we prefer the use of levetiracetam, topiramate, or gabapentin in children who are transplant recipients maintained on immunosuppressants with brittle pharmacokinetic properties. In renal disease, it is important to recognize that the clearance of these three agents is increased with a concommitant decrease in their half-life, necessitating dose and dosage interval adjustments.

In summarizing medical therapy for seizure disorders, we would like to point out several common errors.

When epilepsy is intractable to medical treatment, surgical resection of the tissue responsible for the epilepsy should be considered. Epilepsy is considered medically intractable when a child continues to experience seizures despite adequate trials of three or more anticonvulsant medications, used either alone or in combination therapy (649). Ko and Holmes found symptomatic etiology, early age of onset, the presence of tonic seizures and/or simple partial seizures to be predictive of evolution to intractability (649). Diffuse slowing in the EEG and focal spike and wave activity were electrographic predictors. Kwan and Brodie (650) estimated that slightly greater than one-third of all the patients with new-onset epilepsy became medically intractable. Camfield and coworkers found the response of children to the first medication to be useful in predicting intractability (651). The fraction of their overall population that became intractable was only 8.4%, but the study defined seizure control without incorporating tolerability to the treatment. The number reflects the fact that many children present with seizure syndromes that may remit with time. However, when children present with complex partial seizures, they are much more likely to become intractable. Dlugos and colleagues found 37.5% of their pediatric patients with temporal lobe epilepsy to be refractory at 2 years after initiation of treatment (652). Only about one-third of medically refractory patients may be amenable to surgical therapy.

It is thought that at least 60% of all seizures are partial seizures arising from a single location in the cortex and that 30% of these partial seizures are intractable (653). Of these intractable partial seizure patients, many are not referred to epilepsy surgery centers. It is estimated that approximately 5,000 new patients annually in the United States may benefit from epilepsy surgery, but only one-third receive treatment (654). Much of this is related to a lack of education among primary physicians in referring patients for epilepsy surgery. This is especially true for pediatric epilepsy patients, as many pediatric neurologists are concerned about surgical risk in children and are reluctant to send patients to epilepsy surgery centers.

There are many benefits associated with epilepsy surgery, such as less morbidity and mortality due to intractable epilepsy and status epilepticus, less anticonvulsants used, improved quality of life, and improved cognition and development. It is common knowledge that the earlier surgery is performed, the better chance for developmental gains to occur (653). However, some studies recently have shown that the cognitive improvement after surgery may only be modest and is dependent on the etiology of the epilepsy (655). The complications for surgery include bleeding, risk for infection, cranial nerve dysfunction, increased intracranial pressure, and stroke. These complications are more prevalent for children due to the smaller body volume as well as extensive types of surgeries that are performed, such as hemispherectomy (656). However, iatrogenic residual deficits such as a hemianopsia or hemiplegia resulting from hemispherectomy may be indicated in cases where there is a catastrophic childhood illness in which the natural course of the disease would be worse than the resultant deficit. For example, a disease such as Rasmussen syndrome can eventually lead to the same deficit such as hemiparesis, but early surgery will prevent cognitive delay that occurs with the disease. Unlike in adults, some concerns regarding language lateralization do not apply to very young children due to plasticity in language development. Children younger than 10 years of age can transfer their language to the side contralateral to the resected side, if the resection involves the language dominant side.

For some patients, especially in temporal lobectomies, surgery is felt to be curative if the epileptogenic focus is localized to an area that is easily accessible and not involved in critical function (657). In fact, seizure freedom after surgery depends on whether or not the entire focus can be resected (658). If any area is left, there is a greater chance of seizure recurrence. However, there are many cases where the patient has diffuse areas of involvement that cannot be fully resected. This is especially true for mentally retarded children, considering that dysfunction involves both hemispheres. With these children, palliative surgery aids with decreased seizure frequency but provides no seizure freedom. For example, patients who have drop
attacks may benefit from corpus callosotomy, where the risk for head injury is lessened, although the other seizures will not stop.

Whether a surgery is for palliative reasons or curative reasons, medical intractability of seizures is important to establish (654). Many children are thought to be intractable, but on further surgical evaluation such as long-term videotelemetry, it is found that these children have nonepileptic events. Children who are surgical candidates must have partial-onset seizures. Medical intractability for partial seizures should include a substantial trial of anticonvulsant therapy. Some feel that this includes a trial of at least three anticonvulsants, one of which is a newer anticonvulsant such as felbamate, lamotrigine, topiramate, zonisamide, or levetiracetam. The seizures should be frequent, occurring at least monthly. Many feel that recurrent status epilepticus or regression in cognitive skills with less frequent seizures also should be a criterion, since the frequent seizures or frequent epileptiform discharges interfere with the functions of the healthy brain and therefore will cause cognitive decline to progress. Finally, if a patient has a catastrophic illness where the prognosis for cognition and seizure control is poor, then this would be strong criterion for surgery, regardless of whether the patient has failed three anticonvulsants. These catastrophic diseases include Rasmussen syndrome, Sturge-Weber syndrome, and hemimegalencephaly.

Contraindications to surgery include benign childhood epilepsy syndromes, presence of neurodegenerative disease, and lack of adequate demonstration of medical intractability. There are some controversial contraindications as well. One is intercurrent psychiatric illness. There have been reports that surgery worsens a psychiatric illness such as psychosis, depression, or mania. Yet, there also have been reports in the literature stating that surgery may improve these conditions. Others believe that mental retardation may be a contraindication for surgery, since diffuse involvement of both hemispheres is implied. However, in many children with mental retardation, surgery is used as a palliative treatment for their disabling seizures. Shewmon engaged the question as to whether bilateral interictal discharges were a contraindication to surgery (659). It was felt that this does not represent widespread disease in younger children and therefore, if a focal region can be proved, the patient may still benefit from surgery. Finally, some feel that a lack of social support may be a contraindication if the child will not be able to participate in the intensive rehabilitation that is required for some surgeries, such as a hemispherectomy.

Neuroimaging aspects of presurgical evaluation of candidates have been reviewed (660,661), and the special considerations for presurgical noninvasive and semi-invasive neurophysiologic tests are outlined by Duchowny and colleagues (662). In neuroimaging infants with cortical microdysplasias, the timing of the MRI is of crucial importance because of the progression of myelination affecting the MR signal characteristics. One of us (R.S.) has demonstrated that even microscopic cortical abnormalities may become discernible by changes in the maturation of the adjacent white matter (274). Thus, an infant with West syndrome and a normal MRI at the age of 4 to 6 months may have to be reimaged
several months later if the clinical course warrants considering surgical treatment.

Long-term video-EEG evaluation is preferred to routine scalp EEG in order to establish whether an event is actually a seizure and to interpret the semiology that determines the origin of the seizure. The interictal background as well as the ictal onset of at least two typical seizures should be captured. If mesial temporal lobe sclerosis is considered, then sphenoidal electrodes or cheek electrodes should be placed. In addition, the timing of the seizure electrographically as well as the semiology can aid in determining whether the seizure starts in a deep focus (where there are EEG changes prior to the semiology change). Ictal epileptiform abnormalities are sometimes not clearly lateralized in children; for example, in infantile spasms, the ictal EEG tends to be generalized even when a lateralized zone of cortical abnormality exists. However, the nonepileptiform abnormalities, such as focal bursts of slowing or asymmetry in the generation of beta rhythms in response to barbiturates, or asymmetry of sleep spindles can suggest the presence of an underlying lateralized zone of cortical abnormality (659). In addition, the ictal onset is often subtle and vague in children when compared with adults. For these reasons, interpretation of the video-EEG recordings should be performed by a pediatric electroencephalographer, if possible.

With recent advances in MRI, using stronger magnet strengths and novel and varied pulse sequences, more abnormalities are evident than were seen in the past. A high resolution MRI is the preferred neuroimaging for epilepsy surgery evaluation, as it provides details of neuroanatomy that are not discernible on a CT scan. However, CT is still used occasionally, especially in etiologies where there are calcifications such as in tuberous sclerosis or cystircercosis, which is difficult to pick up on MRI. The MRI should involve FLAIR sequences and T2 coronal sequences with thin cuts (1mm) through the temporal lobe in order to clearly visualize mesial temporal lobe sclerosis. Many children evaluated for epilepsy surgery may harbor cortical dysplasia where gyral thickening and/or gray white matter blurring may be discerned. Sometimes, the dysplasia is not noticeable if the MRI is performed early, such as at 4 to 6 months of age. A child may need to be reimaged several months later in order to determine whether there is a subtle dysplasia (274).

After localizing the epileptogenic zone through surgical evaluations, other invasive studies are performed in order to confirm the location or to tailor the resection further. In addition, if there is a concern that the epileptogenic focus is involved in critical function such as motor or language, functional mapping also is executed.

A patient is said to be in status epilepticus when seizures occur so frequently that over the course of 30 or more minutes, he or she has not recovered from the coma produced by one attack before the next attack supervenes. Status epilepticus is one of the few true emergencies in the practice of pediatric neurology. Consensus states that the more prolonged the status, the worse the outcome.

As has been reviewed elsewhere in this chapter, experimental studies indicate that convulsions lasting longer than 20 to 30 minutes can induce brain damage. Several systemic factors, acting singly or in concert, are believed to be responsible. During the initial stages of status (i.e., for the first 30 minutes or so), cardiac output increases. Tachycardia and systemic hypertension lead to a two- to threefold increase in cerebral blood flow, with a marked increase in cerebral oxygen consumption. During this period, plasma glucose and glucose uptake by the brain is increased. In due time, the brain requirements for oxygen outstrip its supply, and when status lasts longer than 30 to 60 minutes, decompensation sets in (716,717,718). From that time on, cerebral autoregulation breaks down, cardiac output decreases, and there is arterial hypotension and reduced cerebral perfusion. With increased oxygen demands and inadequate oxygen delivery to the brain, cellular metabolism energy fails, and with mitochondrial damage, the organ reverts to anaerobic metabolism with consequent cellular acidosis, increased CSF lactate, and cerebral edema. Ultimately, there is respiratory failure and hyperthermia (716).

Additionally, prolonged and abnormal electrical discharges, by themselves, can cause neuronal damage. The mechanism by which this occurs has not been fully substantiated but involves enhanced glutaminergic excitatory transmission that leads to excessive depolarization of neurons and increased intracellular calcium and sodium. These ion changes initiate a cascade of events that lead to cell death (719). This subject also is reviewed in Chapters 6 and 17.

Status epilepticus is not part of the natural history of the epilepsies, but rather is a complication induced by changes in medication or intercurrent infections. Its incidence has increased with the advent of the newer anticonvulsant drugs, and in a study published in 1989, 3.7% of epileptic patients experienced one or more bouts of status epilepticus (720). In children, 85% of status epilepticus develops during the first 5 years of life and 25% during the first year of life (717).

Status epilepticus also can occur as an isolated phenomenon, particularly in children experiencing viral encephalitis or brain abscess or can occur after an open head injury, especially of the frontal lobes. It can be the initial epileptic attack in patients with secondary (symptomatic) epilepsy or can appear in the course of chronic primary (idiopathic) epilepsy. Maytal and coworkers (720) analyzed the causes of status epilepticus in children. In their series, approximately one-fourth of cases occurred without an obvious precipitating cause in patients who had a prior CNS insult. One-fourth were precipitated by fever and thus represented febrile convulsions, and one-fourth were caused by an acute neurologic insult, such as meningitis, trauma, and anoxia or were due to withdrawal of anticonvulsants. In approximately one-fourth, the precipitant for status was unknown.

Four sequential aspects to the management of the patient in status should be followed. They are (a) maintenance of vital functions, (b) institution of drug therapy to control convulsions, (c) diagnosis of the cause for the condition, and (d) prevention of further convulsions (721).

The physician who treats a child in status should act promptly to maintain an adequate airway, prevent aspiration of mucus, and secure the child from injury induced by the violence of the convulsions. Hyperthermia and hypotension should be corrected.

Several modes of therapy have been used, each with its advantages and disadvantages. We currently favor the use of lorazepam as the first-line anticonvulsant. The drug is highly lipid soluble and penetrates rapidly into the brain. In our experience and that at other centers, lorazepam is superior to diazepam in that fewer seizure recurrences
follow lorazepam, and fewer repeat dosages are required (722). Whether the use of lorazepam is accompanied by a lower incidence of respiratory depression has not been clarified (723).

Lorazepam is administered as an intravenous bolus at a dosage of 0.1 mg/kg, up to a maximum of 4 mg, or 0.06 mg/kg for children older than 13 years of age (724). The drug is administered over the course of 3 or more minutes. Should seizures continue for more than 10 minutes, the initial dose is repeated (725). In most children with generalized or focal status epilepticus, seizures stop within 5 minutes of the administration of lorazepam. The principal side effect is respiratory depression, which can require assisted ventilation. Unlike with diazepam, the likelihood of this complication is not increased by concurrent or antecedent administration of barbiturates. The serum half-life of lorazepam is approximately 15.7 hours (724), and sequential doses of lorazepam are often necessary; in the Los Angeles Children’s Hospital series, sequential doses were used in 30% of cases (724). Stewart and coworkers believe that the incidence of respiratory depression is increased by the use of multiple doses of benzodiazepines (723). The effectiveness of lorazepam diminishes with successive doses.

Diazepam is another effective agent in the treatment of status epilepticus. Like lorazepam it is highly lipid soluble and as judged from concurrent EEGs the drug enters the brain in 1.0 to 9.5 minutes (726). It is highly protein bound, and its redistribution within the body limits its duration of action within the brain to less than 30 minutes. This contrasts with an 8- to 12-hour duration of CNS action of lorazepam (727). Diazepam is administered intravenously at a dosage of 0.3 to 0.5 mg/kg, up to a maximum of 20 mg, and at a rate of 1 to 2 mg per minute (725). In one-third of patients, seizures stop within 3 minutes after diazepam is injected, and in the majority within 5 minutes (728). From the range of suggested dosages and from clinical experience, it is apparent that the amount of medication required for seizure control varies considerably among individuals. The principal advantages of diazepam are its rapid effectiveness, its margin of safety, and its ability to control seizures of cortical as well as of centrencephalic origin (729). The principal side effects of diazepam are respiratory depression and hypotension. They are most likely to occur in patients receiving a combination of drugs, particularly diazepam and phenobarbital, and multiple doses, often at the low end or less than the recommended dose (723,730).

Rectal diazepam (0.5 mg/kg for children aged 2 to 5 years, 0.3 mg/kg for children aged 6 years or older) is administered by means of a rubber tube inserted 4 to 5 cm beyond the anus. It appears to be a simple and safe means of controlling prolonged major motor seizures (731,732).

The use of intravenous phenytoin in the treatment of status epilepticus has been advocated by several groups (718,733). Phenytoin enters the brain more rapidly than phenobarbital, but not as rapidly as diazepam. Once phenytoin has controlled the seizures, immediate recurrence is unlikely. The major drawbacks to phenytoin are its delivery into the circulation and the side effects caused by its high pH level as well as the propylene glycol content needed to increase its solubility (734). Phenytoin is given as an initial intravenous bolus of 10 mg/kg. To prevent cardiovascular toxicity, the drug is given at a rate of less than 1 mg/kg per minute (725,733). A second bolus of 5 mg/kg is given an hour later. This is followed by an intravenous maintenance dose of 10 mg/kg per 24 hours. Shorvon and Richard and coworkers recommend a somewhat higher loading dose (15 mg/kg), with subsequent intravenous doses adjusted according to blood levels obtained at 2 and 8 hours (718,735). The former group recommends changing to oral maintenance phenytoin at 28 hours. We prefer the use of carbamazepine or valproate for maintenance after status, particularly in infants and small children.

The main advantages of phenytoin are a half-life of 24 or more hours and lack of significant CNS depression. For this reason, the drug is usually preferred for patients with status following a head injury, in whom preservation of consciousness is desirable. The anticonvulsant should not be administered intramuscularly because it is absorbed too slowly to result in effective anticonvulsant serum and tissue levels (736).

Fosphenytoin, a disodium phosphate ester of phenytoin, has found wide acceptance for the treatment of status and in many centers has replaced phenytoin. The drug is completely water soluble and is rapidly and completely converted to phenytoin. It can be administered both intravenously at a dosage of 10 to 20 mg/kg or intramuscularly (733,737). Holmes and Riviello recommend its use in children whose status has lasted longer than 10 minutes (738). Evrard and colleagues recommend prompt combined treatment with a benzodiazepine such as lorazepam, which has GABAergic action and counteracts the excitatory effect of released glutamate, with fosphenytoin, which antagonizes the release of excitatory amino acids, and believe that these two drugs have a synergistic effect (739).

Another means of controlling status epilepticus is by the use of sodium phenobarbital. The initial dose of 15 mg/kg is administered intravenously at a rate of 2 mg/kg per minute, intramuscularly, or even subcutaneously. Because the drug requires 15 minutes to penetrate the blood–brain barrier regardless of its mode of administration, the rate at which seizures are controlled is slower than with lorazepam or diazepam, and it is now reserved for those children whose seizures have failed to respond to a benzodiazepine within 15 to 40 minutes after its administration (738).

If phenobarbital is to be used as a backup to a benzodiazepine, the child must be intubated. Aside from its slow
rate of action, the disadvantages of phenobarbital are the depression of consciousness and respiration when the drug is used in dosages required for the control of status epilepticus (740).

Refractory seizures are seizures that have continued for 60 minutes or longer despite adequate anticonvulsant therapy. They frequently develop in the presence of acute CNS injury or a neurodegenerative disease. When brain edema is suspected, as is the case when status follows a head injury, corticosteroids or other antiedematous agents should be given. In addition, one may have to resort to midazolam or pentobarbital. Midazolam, an injectable benzodiazepine, is given in a bolus dose of 0.2 mg/kg, followed by an infusion of 1 to 5 μg/kg per minute (741,742). Pentobarbital is given in a loading dose of 5 to 15 mg/kg, followed by a maintenance infusion of 1 to 5 mg/kg per hour (738), with the dosages of pentobarbital being adjusted to maintain an EEG burst suppression pattern (743). Although monitoring the electrical activity of the brain whenever a patient is in status is always desirable, this measure becomes mandatory in the patient with refractory status (725,738).

Other drugs that have been used for the treatment of status epilepticus include intravenous valproate (10 to 20 mg/kg) and lidocaine (744,745). Intravenous valproic acid is given at a dosage of 25 mg/kg for patients who have not been on previous anticonvulsants and 10 mg/kg for patients with a breakthrough seizure. The average infusion rate is 2.5 to 3.0 mg/kg per minute (746). In the United States, clinical experience with long-term use of lidocaine has been limited. According to a survey, 76% of U.S. neurologists use intravenous lorazepam for the treatment of status epilepticus. If this fails, 95% use phenytoin or fosphenytoin. Thereafter, 43% would give phenobarbital, 16% would give intravenous valproic acid, and 19% would give continuous infusion of either pentobarbital, midazolam, or propofol (747).

Whatever the means for treating status, the physician must keep in mind that the intravenous route is the preferred way to administer anticonvulsants; that the most common mistake is to give repeated, yet insufficient, doses of anticonvulsants; and that he or she should avoid using more than one anticonvulsant drug. Transient abnormalities on neuroimaging studies, particularly on MRI, have been seen by us and have been reported in the literature (748). They may raise the suspicion of a neoplastic lesion, but in most instances, they resolve completely.

Following termination of status, the patient is maintained on parenteral anticonvulsant therapy until the patient has regained consciousness. Oral medication is then resumed. With successive doses, the effectiveness of lorazepam and diazepam is progressively reduced, and these agents cannot be used for long-term seizure control.

With prompt and appropriate therapy, mortality owing to status has fallen to essentially nil. However, a significant percentage of patients can die as a consequence of the condition that precipitated status. In the 1989 series of Maytal and coworkers, no deaths occurred in 137 children with unprovoked or febrile status (720). In the series of Phillips and Shanahan, also published in 1989, only 1% of patients died (749). With current therapy, the incidence of residua is low and is essentially nil when status results from a febrile seizure. Although some 9% of children are left with new motor or cognitive deficits, these usually are the consequence of the underlying neurologic condition that caused status. In terms of outcome, the duration of status is less important than its etiology (720). The clinician who cares for children who have experienced an episode of status should keep in mind that such children have a 4% to 6% likelihood of experiencing one or more additional episodes of status, and that children who have had more than one episode of status have approximately a one in three risk of further episodes (750).

Up to 30% of children with epilepsy do not have remission of their seizures despite apparently adequate treatment. One important, recently uncovered factor in the development of refractory epilepsy is the overexpression of a variety of drug resistance proteins or their genes in brain tissue. This overexpression has been observed in brain tissue taken from areas of cortical dysplasia, hippocampal sclerosis, and tubers from patients with refractory epilepsy. Several proteins have been implicated to date, and all belong to the ATP-binding cassette superfamily that also has been associated with resistance of cancer cells to antineoplastic agents. They are the multidrug resistance gene-1 P-glycoprotein (MDR1), multidrug resistance-associated proteins (MRP1 and MRP2), and the major vault protein (MVP). By contrast, control tissue have no or low expression of these four types of proteins (751,752,753). Because these multidrug transporters restrict the entry of lipophilic molecules such as many of the antiepileptic drugs into brain, their overexpression reduces effective drug concentration at target sites. Genetic factors may well be implicated in whether there is overexpression of one or more of these proteins, and Siddiqui and coworkers have identified a polymorphism in the gene encoding MDR1 in subjects who did not respond to anticonvulsant medications (754). It is not clear, however, whether overexpression of these proteins is the cause or the result of intractable seizures. In experimental animals, recurrent seizures have been shown to induce multidrug resistance genes, and antiepileptic drugs can up-regulate MDR1 and some of the other multidrug transporters. From these data, we suspect that there also could be a genetic contribution to drug-resistant epilepsy.

In addition to these newly uncovered genetic factors, the success of medical therapy in epilepsy depends on the type of seizure and its natural course (755). Several
studies have demonstrated that the response to the first drug trial predicts the success of anticonvulsant therapy. Thus, Dlugos and coworkers found that in children with temporal lobe epilepsy, failure to respond to the first anticonvulsant drug was highly predictable for refractory epilepsy (652). MacDonald and colleagues found that the number of seizures experienced by a child during the first six months after presentation predicts the likelihood of remission. Thus, the patient who has experienced two seizures during the first 6 months of observation has a 47% chance of five years’ remission, whereas a patient who experienced 10 or more seizures during this period only has a 24% chance of five years’ remission (756).

As mentioned previously, the series of Kwan and Brodie working in Glasgow confirm these observations. In their experience, 47% of patients became seizure-free during treatment with their first antiepileptic drug, and only a further 14% became seizure-free during treatment with a second or third drug. In 3%, treatment with two drugs succeeded in controlling seizures (650). Among patients who failed to respond to the first anticonvulsant, only 11% became seizure free when treatment failure was due to lack of efficacy rather than due to intolerable side effects of an idiosyncratic reaction.

Generally, spontaneous remission or control of seizures by anticonvulsants is greatest in children with idiopathic seizures and age of onset between 2 and 12 years (757). Clinically, such children have normal intellect, no abnormalities on neurologic examination, and no focal lesions on imaging studies. In the experience of Camfield and coworkers, who studied children with generalized tonic-clonic, partial, and partial with secondarily generalized seizures, 83% were successfully treated with a single anticonvulsant. Treatment was more likely to be successful in children with generalized tonic-clonic seizures, and treatment failures were more likely in children with complex partial seizures. Of the 17% of children who did not respond to the first anticonvulsant, only 42% ultimately became free of seizures and were in remission at the end of four or more years (651). As in most other studies, the success of anticonvulsant treatment for children with neurologic deficits was significantly less than in the remainder of the study group.

Between 25% and 83% of children younger than 16 years of age who have had a single nonfebrile seizure experience a recurrence, with the median time to recurrence being 5.7 months. In the series of Shinnar and colleagues, 88% of recurrences occurred within 2 years, and only 3% recurred after 5 years (758). The lower recurrence figures are derived from prospective studies (759), whereas the higher figures are derived from retrospective studies (760). As many as 70% of the group with recurrent seizures had only a small number of further attacks before going into remission, which in approximately 90% was permanent (761). In 30% of subjects, seizures continued. These chronic cases constitute the mainstay of a pediatric neurologist’s practice. With successful treatment of children in this group, some 30% enter a long remission, whereas 70% continue to experience occasional seizures despite adequate therapy and compliance.

Generally, the longer the period between the onset of seizures and their control, the less likely the chance for significant remissions. The remission rate is 60% if seizures do not come under control within a year of their onset. This contrasts with a remission rate of 10% if seizures remain uncontrolled for more than 4 years and 5% for those who continue to experience seizures for 10 or more years after their diagnosis (762). However, even in medically intractable seizures, the frequency of attacks tends to diminish over the years, especially in those who maintain a normal IQ (763).

Some clinical features can be singled out for prognostic significance.

Remissions occurred in 13% of neurologically intact children with grand mal epilepsy who had experienced fewer than 20 such seizures (764) and in 75% to 80% of children with childhood absence (petit mal) epilepsy (765). In our experience, it is the rare child with normal intelligence whose seizures are not completely or nearly completely controlled with adequate medical supervision and good compliance. Despite good seizure control, the social outcome is not as favorable. In part, this is because of associated learning disorders and, in part, the consequence of a chronic illness (766).

Adverse prognostic factors include partial or mixed seizures, an abnormal neurologic examination, mental retardation, and inadequate seizure control during the early years of the disorder (767). Complete remission is least likely in children with minor motor attacks.

A particularly favorable prognosis is seen in the following seizures:

Absence seizures also have a favorable prognosis. In a follow-up performed 15 years after seizure onset, 65% of
patients were in remission, and 18% were taking anticonvulsants. Of the latter group, 38% were seizure free. Seventeen percent were not taking anticonvulsants and continued to experience seizures. Of the total cohort, 15% had progressed to juvenile myoclonic epilepsy. The factors that predicted treatment failure included the presence of cognitive problems, development of nonconvulsive status, the appearance of tonic-clonic seizures after the onset of anticonvulsant therapy, an abnormal EEG background, and a history of generalized seizures in first-degree relatives (770).

Of patients with complex partial seizures, 84% experienced complete seizure control at the end of the first year (651); 61% were in remission at the end of 4 or more years; 35% experienced sporadic seizures, and 4% had intractable epilepsy. Favorable predictive factors in complex partial seizures are good intelligence, a family history of seizures, and a right temporal seizure focus. According to Lindsay and colleagues, unfavorable factors include early onset of seizures, frequent grand mal seizures, an associated hyperkinetic syndrome, and attacks of rage during childhood (771).

Approximately one-eighth of patients with nocturnal tonic-clonic seizures ultimately can be expected to experience daytime attacks as well (772). Although there are no controlled studies that determine whether early and effective anticonvulsant therapy influences the chances for a spontaneous remission, Reynolds and other members of his group, including Shorvon, believe that epilepsy must be treated early and effectively to preclude the evolution of a chronic epileptic condition (761,773). We agree with their position.

Aside from dying from unrelated disorders, epileptic patients can die in status epilepticus as a consequence of their underlying brain disease (e.g., tumor or a CNS degenerative disorder), accidentally as a result of a seizure, and, most perplexing, without any apparent cause (774). The majority of children in the last group had a severe handicap, and death in children with absence epilepsy or partial and primary generalized epilepsy was between 1% and 2% (775). In the series of Donner and coworkers, 52% of children with sudden unexplained death had symptomatic epilepsy that appeared to have been relatively well controlled. In pediatric patients, 60% had therapeutic anticonvulsant blood levels at the time of death (774). Low serum anticonvulsant levels and polytherapy, which are considered to be risk factors in adults, are not significant in the pediatric population. The most likely explanation for their demise is that it is a seizure-related event, such as aspiration, or possibly neurogenic pulmonary edema (775,776,777).

The question whether recurrent seizures can cause neuronal injury and intellectual deterioration has yielded conflicting results. This is not surprising when one considers the various confounding factors: the heterogeneity of the conditions that induce epilepsy, the effect of subclinical seizures and interictal epileptiform activity, the various comorbidities, the effect of many different treatment regimens, and the genetically determined susceptibilities (778). Neuropathologic examinations of the human brain suggest that recurrent seizures are associated with neuronal injury and reactive changes in the hippocampus. This evidence must be qualified by the lack of data on the effect of chronic epilepsy on the neocortex and cerebellum.

There are longitudinal studies indicating that recurrent seizures are accompanied by a decline in cognitive function, notably an impairment of memory, and more importantly, for the pediatric age group by disruptive behavior. Thus, in a longitudinal study conducted by Austin and Dunn, children with recurrent seizures had worse behavior scores than those children whose seizures were controlled. This study was confounded by the effects of the underlying neurologic dysfunction that resulted in treatment-resistant seizures (779). Neuroimaging studies used to identify structural changes in the brain that might objectify brain damage leading to intellectual deterioration have yielded conflicting results. Most studies have examined the effect of recurrent temporal lobe epilepsy on the hippocampal volume. Studies with contrasting results have been published. In a cross-sectional study, Cendes and coworkers could not find an effect of the duration of epilepsy and seizure frequency with the volume of the hippocampus or amygdala (780). A similar result was obtained by Liu and coworkers, who were unable to correlate changes that developed in the hippocampus, the cerebellum, or the neocortex over the course of several years with the frequency or severity of overt seizures (781). Van Paesschen and colleagues obtained conflicting data. These workers found that the severity of hippocampal damage correlated with the duration of epilepsy and the number of secondarily generalized seizures (782). In a longitudinal study, Briellmann and coworkers found that the hippocampal volume loss over the course of 3 to 4 years correlated with the number of secondarily generalized seizures between scans (783). In other studies, significant atrophy of hippocampus, neocortex, or cerebellum developed in the course of 3½ years in 16% of chronic epilepsy patients as compared with 3% of controls. However, the number of overt seizures had no effect on brain volume (778). Other cross-sectional studies have shown that childhood onset temporal lobe epilepsy was associated with a reduced volume of the corpus callosum, particularly its posterior portion (784). The reliability of functional imaging techniques such as MRS, PET, and SPECT for ascertaining cerebral damage following seizures has not been confirmed.

Several cross-sectional and longitudinal studies have found a progressive decline in cognitive function, particularly in memory in both adult and pediatric patients with poorly controlled temporal lobe epilepsy (785,786). In children with rolandic or occipital lobe epilepsy, mild and transient cognitive difficulties were documented, which
Deonna and coworkers felt had a direct relationship to the presence of paroxysmal EEG activity (787).

There is considerable evidence that implicates various drug regimens in causing intellectual deterioration of epileptic patients (788). In the experience of Trimble, some 15% of patients with recurrent seizures deteriorate intellectually over time (789). Examination of patients in Trimble’s group revealed that patients with intellectual deterioration had significantly higher serum levels of phenytoin and primidone and lower folic acid levels than the group who did not experience deterioration. When seizure frequency was factored out, these findings still persisted with respect to phenytoin and folic acid (789). Other studies also have implicated polytherapy in intellectual deterioration. Addy concluded that an association between poor cognitive function and phenobarbital or phenytoin intake has been reported too often to continue considering them as first-line anticonvulsants for the pediatric population (790). Trimble and his group have demonstrated that phenobarbital and phenytoin, given singly or jointly, impair immediate and delayed memory and interfere with performance on visual and auditory scanning tasks (789). Eliminating polytherapy or changing the anticonvulsant to carbamazepine tended to improve cognitive function within 3 months. Even with therapeutic anticonvulsant levels of drugs such as phenobarbital, children experience subtle cognitive effects (791). Lamotrigine, carbamazepine, and valproic acid produce less striking cognitive impairment than phenobarbital.

Behavior disturbances are common in epileptic children. In part, they stem from the deleterious effects of a chronic illness, but frequently they, too, are induced by drug therapy and disappear spontaneously once the offending anticonvulsant is withdrawn (792). Additionally, it is likely that children with complex partial seizures are particularly prone to disordered behavior. Abnormal behavior or psychoses also can be part of the ictal phenomenon, as in some patients with absence status, or they can follow a seizure (793).

In view of the adverse effects of long-term anticonvulsant therapy, medication should be withdrawn as soon as feasible. There has been considerable controversy about when drugs should be discontinued after prolonged seizure control. Several studies have attempted to answer this question and have found that recurrence risks are similar after seizure-free intervals of 2, 3, or 4 years (757,794,795).

Holowach-Thurston and coworkers discontinued anticonvulsants after 4 years of seizure control and encountered a relapse rate of 28%, which increased by another 4% when the follow-up was extended to 15 years (796). More than one-half of recurrences were seen during the first year after anticonvulsant withdrawal, and 85% had occurred within 5 years of withdrawal. Other studies have come up with similar recurrence rates, even though in some studies, drug withdrawal was started after a 2-year seizure-free interval (764,797). Neither the sex of the patient nor the age when drugs are withdrawn affect the relapse rate. Children with neurologic dysfunction or an abnormal neurologic examination have a higher incidence of recurrence. The relapse rate is highest in children with juvenile myoclonic epilepsy and complex partial seizures and lowest in benign rolandic epilepsy and absence seizures (797). These results probably reflect the relatively high proportion of patients with idiopathic epilepsy in the last two groups. In the experience of some workers, the age of seizure onset also influences the likelihood of relapse. The prognosis is poorest for those with onset younger than 2 years of age, less so for those with seizure onset at age 12 years or older, and best for those with onset between 2 and 12 years of age (757). Children with seizure onset after 12 years of age have a higher risk for recurrence (795). By contrast, Gherpelli and colleagues noted that the age of seizure onset did not influence the likelihood for recurrence after anticonvulsant withdrawal but found that in their series, the greater the number of seizures experienced by the patient before control, the greater the likelihood for relapse (798).

Whether the EEG can predict a relapse is still unresolved. Whereas Holowach-Thurston and coworkers (796) did not consider EEG findings to be an important predictor of seizure recurrence, Peters and colleagues, Shinnar and coworkers, and Tennison and coworkers did (795,797,799). In the experience of Shinnar and coworkers, EEG slowing, but not epileptiform features, signaled an increased risk for relapse (979). Peters and colleagues and Tennison and coworkers found that the presence of paroxysmal EEG discharges was predictive for recurrence (795,799). As a rule, the EEG is more predictive for relapse in children with idiopathic epilepsy than in those with epilepsy secondary to a known CNS abnormality (794). In all studies, risk of recurrence has been highest during the first few months after withdrawal, with 50% or more of recurrences taking place within 6 months after initiation of withdrawal (794,798). Callaghan and coworkers found that the factors predictive for relapse are the same as in an adult population (800). The rate of anticonvulsant withdrawal does not appear to influence the frequency of relapse, although withdrawal over the course of less than a month is inadvisable, and abrupt withdrawal is known to initiate status epilepticus (799).

Based on these studies, we consider withdrawal of anticonvulsants after approximately 2 years of seizure control. Generally, we withdraw medication over 2 to 3 months. When the child has been on polytherapy, one drug is completely stopped before the other is tapered.

Drug withdrawal after successful epilepsy surgery has been proposed after the patient has been seizure- and aura-free for more than one year. The recurrence rate appears to be unrelated to the duration of seizure-free postoperative
anticonvulsant therapy, and in the series compiled by Schiller and coworkers was 14% at two years and 36% at five years following complete anticonvulsant withdrawal. Even after 5 to 6 years of being seizure free on anticonvulsants, patients ran a 33% risk of seizure recurrence (801).

The social problems incurred by an epileptic child in terms of his or her relationship to family and peers are so considerable that they are beyond the scope of this book. The reader is referred to books by O’Donohoe (170) and the chapter by Craig and Oxley in the volume by Laidlaw and colleagues (802) for a more extensive discussion.

In our opinion, epileptic children should participate in all normal activities, including sports. The hyperventilation incurred in the more strenuous athletics is accompanied by a buildup of carbon dioxide and, therefore, should not induce seizures. However, the more hazardous contact sports, such as high school football, are best avoided and should be reserved for nonepileptic youngsters. As well, epileptic children should be permitted to swim with a friend under competent adult supervision (802,803). Kemp and Sibert, who reviewed drowning accidents in epileptic children, found that no child who participated in supervised swimming drowned (804). Most states and countries restrict an epileptic patient from driving a car. Considering the likelihood that a youngster who has been seizure free for 2 or more years will continue to be seizure free, we have no hesitation in recommending such a patient for a driver’s license. When it comes to withdrawing medication in such a youngster, we are reluctant to do so, for fear that seizure recurrence will result in loss of the license. The dilemma between maintaining anticonvulsants for longer than necessary and inducing a serious social inconvenience requires an individualized decision.

The term febrile seizure is used to designate seizures associated with fever, usually occuring between three months and five years of age, and excluding those caused by infections of the CNS or other defined causes. Although not part of the definitions used by the National Institutes of Health or the International League Against Epilepsy, in most instances, a minimum body temperature of 37.8° to 38.5°C or 100.1° to 101.4°F is required for a seizure to be considered febrile.

Febrile seizures represent one of the most common neurologic disorders of childhood. In 1924, Patrick and Levy found an incidence of 4.2% of febrile seizures in an unselected group of children attending well baby clinics (805). Subsequent studies have shown an incidence between 2% and 4% (806,807,808). In Japanese children, the incidence is about 7%, and in Guam, it is as high as 14%. The condition is somewhat more common in male subjects.

Because seizures occur with relatively high frequency during the neonatal period and present special problems for diagnosis and treatment, they are considered separately.