30 Seizures and related disorders in children

Epilepsy is most prevalent at each end of the age spectrum—the very young and the very old. Up to 1 per cent of the childhood population will have active epilepsy at any time. Of these 60–70 per cent will be controlled on medication or enter into spontaneous remission, however the remainder will continue to have seizures despite the range of treatment available. There will be associated comorbidity of learning and behaviour difficulties in a significant proportion, and these may take precedence in management over the seizures themselves. Careful evaluation of each individual child with regard to the possible diagnosis and associated comorbidities is required in all children presenting with recurrent paroxysmal episodes in order to optimize management.

The most common cause of poor response to treatment in epilepsy is misdiagnosis. About 20–40 per cent of children arriving in a tertiary epilepsy clinic for further assessment do not have epilepsy, but rather conditions involving paroxysmal episodes that are not epileptic in origin (Uldall et al. 2006). Common causes of misdiagnosis include misconceptions of what is important in the history. Too much weight may be put on factors other than the events themselves such as family history, or abnormal neurological examination findings. Or one may be unaware of the range of conditions that may present in childhood that are not epileptic. Epileptic seizures are defined as changes in movement or behaviour that occur as the result of a primary change in the electrical activity of the brain. Numerous other types of event may involve such changes but as secondary phenomena to a non-epileptic primary event (National Institute of Clinical Excellence 2004).

Any event that reduces the supply of oxygenated blood to the brain will result in a loss of consciousness; and if prolonged, secondary epileptic phenomena may occur as a result of the hypoxia, causing a hypoxic seizure. In childhood the most common causes are breath-holding attacks, or reflex anoxic seizures. These typically occur in young children, 6 months–5 years, in response to a noxious stimulus. Children may either have a prolonged expiratory apnoea, causing cyanosis and hence a cyanotic breath holding attack, or a reflex asystole causing ‘pallid’ attacks. The description of events is key to diagnosis; most children grow out of the tendency and often there is a family history of such events. Vasovagal attacks do occur but in the older population—typical syncope is rare under 10 years, and in such children a cardiac cause must be considered. In addition atypical features should prompt a cardiac evaluation such as occurrence on exercise, or a prolonged event. This aside it is good practice for an electrocardiogram to be performed on all new presentations of collapse (Scottish Intercollegiate Guidelines Network 2005).

Hyperekplexia is an autosomal dominant disorder, the result of a mutation in the glycine receptor gene, where children present from birth with excessive startle, resulting in tonic spasm that can be very profound and result in hypoxic seizures. The history with the supportive examination finding of startle to nose tapping will suggest the diagnosis. Although attacks may become less severe with age, in infancy they can be life threatening.

The most common phenomena to mimic an epileptic episode are tics (Section 40.6.1). Simple motor tics are common in young children, usually involve the upper body or face, and are short lived in their natural history. Many may also appear to be familial. Only a minority go on to develop Gilles de la Tourette’s syndrome (Section 40.6.3) with vocal as well as motor tics. In the very young, paroxysmal abnormal eye movements can be the presenting feature of alternating hemiplegia, the intermittent hemiplegia itself only becoming apparent beyond the first year of life. Cataplexy (Section 32.3.1), with drops from loss of muscle tone triggered by emotion, may be misdiagnosed as epileptic drop attacks. Abnormal upper body movements seen associated with feeding may be related to oesophageal reflux, Sandifer’s syndrome, and may be seen particularly in the neurologically abnormal or developmentally delayed child. The movements are thought to be induced by an attempted change in intrathoracic pressure to relieve pain from oesophagitis. Other paroxysmal episodes that may be misdiagnosed include benign paroxysmal vertigo (Section 15.3.2) and benign paroxysmal torticollis (Section 40.4.1). These conditions can come on before the age of five, with obvious torticollis or with paroxysmal vertiginous episodes of which the child will be aware, and lead to an unwillingness to move or unsteadiness on their feet. The episodes usually last minutes to hours; the conditions resolve with age. Paroxysmal movement disorders and dystonias may also be misdiagnosed as epileptic, and indeed anticonvulsant treatment may be effective in some. Therefore a response to anticonvulsants should not be used as a diagnostic tool.

Various behavioural phenomena may be misdiagnosed as epileptic attacks. The most common referral to the neurology clinic will be of day dreaming being questioned as absence attacks; in the vast majority simple day dreaming or non attention will be the diagnosis. Key features confirming this are a situational occurrence and the ability to be distracted from attacks, sometimes by physical touch. In very young children self-gratification behaviour can cause concern as the children can appear distracted, unaware of their surroundings and become flushed. They may also become distressed if distracted. Such events are very common and not abnormal although the diagnosis can be distressing to parents. Toddlers may have distal movements associated with excitement that are completely benign called ‘overflow movements’ or shuddering attacks. Stereotypies and ritualistic behaviours can cause concern but are common in the learning disabled and autistic population. Episodic rage is commonly referred for assessment as being possibly epileptic in origin but is almost never directly the result of epileptic discharge, although it may be seen as a postictal phenomenon. Pseudoseizures are seen in children, rarely under 10 years, and commonly coexist with real epileptic seizures. Their diagnosis requires careful assessment and evaluation, particularly looking for a cause, although sexual abuse is only likely to account for a small number.

A wide range of sleep phenomena (Section 32.4) exist in normal children; a parent will only become aware of this when forced to sleep with a young child. Benign sleep myoclonus is commonly misdiagnosed as epilepsy in the first 6 months of life leading to children being overtreated with antiepileptic medication. The key differentiation is that jerks only occur during sleep in developmentally normal children. Older children and adults may also experience myoclonic jerks as a normal phenomenon on going off to sleep (Section 32.4.1). Night terrors may occur at the same time each night; children awake apparently terrified and unresponsive, and will appear distressed for several minutes before settling back to sleep. These may be aborted by waking the child at a similar hour each night prior to the time of the likely attack.

Arguably there is little difference in the definition of epilepsy between adults and children: it is a condition whereby the individual is prone to recurrent epileptic seizures. By definition this means at least two epileptic seizures. In certain circumstances, particularly certain syndromes, the diagnosis of epilepsy may be made after a single seizure and a diagnostic EEG abnormality. This is probably of most relevance when considering when to treat after a single seizure if the risk of further seizures is judged to be high.

The incidence of epilepsy in the first year of life is higher than at any other age during childhood (Doose et al. 1983; Hauser 1993; Braathen and Theorell 1995; Olafsson et al. 1996). Age-specific incidence rates for the first year of life documented in studies to date show great variability from 80 to 256/100 000. Differences in geographical study areas and methodology may be responsible for such variation. Retrospective study designs and studies where records of a single centre providing services to a catchment area are the only source of cases are associated with possible under-ascertainment and bias. There is evidence that the incidence of epilepsy in childhood has been declining over the past decades (Hauser et al. 1993). The availability of diagnostic criteria in later time periods may have resulted in reduced inclusion of non-epileptic patients. A recent prospective community-based study has estimated an incidence in children under 2 years of life of 62.3 (95 per cent CI: 47.4–81.9)/100 000 (Eltze et al. 2007). Neonatal seizures may be more prevalent, with incidence figures of 70 to 270/100 000, and also higher in preterm infants at 58 to 132/1000. Incidence figures fall for the remainder of childhood, but still remain higher than in adolescence.

Population data suggest the majority of children who present with epilepsy in childhood have a good prognosis, both for remission as well as control of seizures. Studies show that two-thirds of children will be seizure free long term (Sillanpaa and Schmidt 2006). Furthermore population studies suggest the majority do well. However there are select populations in whom the risk for comorbidity and long-term seizures remain high.

Mortality is related to the risk of associated systemic disease, as well as accidents. However there is also a risk of sudden unexpected death in epilepsy amongst those who continue to have convulsive seizures, and those with associated learning and physical disability are most at risk (Forsgren et al. 2005).

The history is the key to the diagnosis of epileptic seizures, and hence of epilepsy. It is of particular value if obtained from an eye witness which is usually the parent in the case of children. However if the event has happened outside the home such as in school further information may need to be sought and results of investigations may assume greater importance in support of the diagnosis. An EEG should be performed in any child with suspected epilepsy; in most this will be following the second seizure. The most important role of the EEG is not to confirm the diagnosis of epilepsy, unless of course an event is captured during the EEG recording, but to enable a syndrome diagnosis. The child’s EEG requires skilled interpretation; developmental changes seen with age may be misinterpreted if reviewed by an individual not experienced in paediatric EEG. In a

small number of children an EEG may be considered after a first seizure; for example in a child with a history of a single sleep seizure involving facial or bulbar features where benign rolandic epilepsy may be suspected, or where a teenager presents with an initial tonic clonic seizure and generalized spike wave seen on EEG would suggest an idiopathic generalized epilepsy. If the EEG is normal and yet the diagnosis is still suspected then a sleep EEG should be performed. Specialist EEG investigations such as ambulatory EEG and video EEG telemetry (Section 3.3.5) would only be performed in specific circumstances, perhaps where diagnosis remains in question or where localization of seizure onset is to be defined as part of presurgical evaluation.

Unlike in adults, neuroimaging is not routinely performed in all children presenting with epilepsy. Usually imaging will not help in diagnosis, but may point to the aetiology. MRI is the imaging of choice, although in rare circumstances CT scan may reveal the aetiology promptly in an acute presentation of seizures. MRI should be considered in all children presenting with epilepsy under the age of 2 years, in all those presenting with a focal epilepsy where a benign syndrome cannot be confidently diagnosed and in those whose seizures continue despite a trial of two anticonvulsant medications (NICE 2004). The role of other investigation remains in determining an aetiology, and will be driven by the presentation (Section 30.6.3).

It is important to consider the underlying cause of an epilepsy. There are few neurometabolic or neurodegenerative conditions that present with seizures as a first isolated symptom. These are most common within the first year of life, and are seen with decreasing frequency with age. It is important to diagnose these in view of their different prognoses and the possible relevance to antenatal diagnosis in future pregnancies. The conditions to be considered are outlined in Table 30.1 (see also Table 10.2 VI 1–16).

All children presenting within the first year require evaluation as to whether there is an underlying neurometabolic condition. In an older child it is the presence of associated symptomatology, whether a fluctuation in neurological status or neurological decline that may point to whether further evaluation is warranted. It is often difficult to decide whether a neurological deterioration is related to the epilepsy, or due to an underlying causative process of both the epilepsy and the decline. This may take time to evaluate, so as to determine whether there is any degree of fluctuation related to seizures suggestive of epileptic encephalopathy (Section 30.5.2) as opposed to the steady deterioration with loss of skills that would suggest a neurodegenerative process. The emergence of neurological signs generally, although not exclusively (Neville and Boyd 1995; Dale and Cross 1999), suggests an underlying progressive pathology. Investigations should be targeted according to the condition suspected, and may include CSF studies, biopsies, or further genetic analysis.

Certain mitochondrial diseases have seizures as part of their clinical phenotype. Some patterns are well recognized such as MELAS, myoclonic epilepsy, lactic acidosis, and stroke-like episodes or MERRF, myoclonic epilepsy with ragged red fibres (Sections 10.5 and 24.6.3). Other less-defined disorders present with evidence of multisystem involvement with epileptic seizures and developmental regression. Key investigative marker are raised plasma and CSF lactate.

Alper’s disease, or progressive neuronal degeneration of childhood, with associated hepatic involvement may result from mitochondrial gene mutations particularly of the nuclear polymerase gamma gene, POLG (Section 10.5.6) (Ferrari et al. 2005; Horvath et al. 2006). Lactic acidosis may be absent. Children with Alper’s disease may present with an aggressive form of epilepsy, with frequent status epilepticus, epilepsia partialis continua, and developmental regression. The EEG is suggestive with spikes or polyspikes, often unilaterally and prominent posteriorly, in association with slow waves and a disorganized background (Boyd et al. 1986). Visual evoked potentials are also grossly abnormal with normal electroretinogram. Signs of liver impairment may be seen only relatively late in the disease and may be determined only on biopsy. The condition is invariably fatal; some previously reported cases of valproate-induced liver failure may have been unmanifesting cases of Alper’s.

This condition may present insidiously with onset of seizures in early to mid-childhood, and cognitive regression that may only be apparent sometime after the diagnosis of epilepsy (Section 10.4.4). A movement disorder ensues, with extrapyramidal characteristics. The condition has two hallmarks, cataplexy which may be misdiagnosed as epileptic drop attacks, and failure of upward gaze. It is a progressive autosomal recessive disease, characterized by late accumulation of multiple lipid molecules in association with abnormal tubulovesicular trafficking (Patterson and Platt 2004). At the cellular level, the disorder is characterized by accumulation of unesterified cholesterol and glycolipids in the lysosomal/late endosomal system. Approximatively 95 per cent of patients have mutations in the NPC1 gene at 18q11 which encodes a large membrane glycoprotein primarily located at late endosomes. The remainder have mutations in the NPC2 gene at 14q24.3 which encodes a small soluble lysosomal protein with cholesterol-binding properties (Vanier and Millat 2003). The gene product NPC1 protein is not suitable for transduction therapies, and gene replacement or repair is to yet be practicable for Niemann Pick disease type C or related disorders. Treatment is symptomatic, and death occurs usually before the third decade.

Late infantile neuronal ceroid lipofuscinosis often presents with seizures as the first manifestation around the third year of life; developmental plateau then follows and evidence of a progressive neurodegenerative disease manifests (Section 10.3.2). Although generalized tonic-clonic seizures may be the first manifestation, myoclonus becomes prominent. Blindness eventually develops due to retinal atrophy. This is in contrast to the early juvenile form where visual failure is likely to be the first manifestation and seizures occur relatively late in the course. The EEG in late infantile neuronal ceroid lipofuscinosis can be suggestive by demonstrating prominent posterior spikes on slow photic stimulation (Fig. 30.1). Further neurophysiological studies show enlarged visual evoked potentials and a reduced or absent electroretinogram.

For the majority of families affected by one of these diseases, a biochemical and/or genetic diagnosis can be achieved (Williams et al. 2006). Classical late infantile neuronal ceroid lipofuscinosis is caused by mutations in the CLN2 gene leading to tripeptidyl- peptidase 1 deficiency. Early juvenile neuronal ceroid lipofuscinosis is caused by mutations in CLN3 gene. Other variants may result from mutations in other CLN genes.

Children may manifest with myoclonus as a significant component of the epilepsy syndrome. In many of early onset this may be part of a static encephalopathy or one of the syndromes discussed below (Section 30.5). A small number of children, particularly those presenting in later childhood, may be manifesting a progressive myoclonic epilepsy. In some presenting in teenage years, such as those with Unverricht–Lundberg disease, initially the diagnosis of an idiopathic generalized epilepsy such as juvenile myoclonic epilepsy may have been made. However there is often keenness to make such a diagnosis, however atypical the presentation, in view of the more benign nature of the disease. Certainly a consideration of a progressive myoclonic epilepsy should be raised if there is lack of response to medication, and where myoclonus is a prominent feature to which there is no diurnal variation.

Unverricht–Lundberg disease. This may be subtle in its presentation, confused at the outset with an idiopathic generalized epilepsy. First described in 1891 by Unverricht in Estonia and later Lundborg in Sweden (1903) the diagnosis is confirmed by the finding of an increased number of dodecamer repeats in the promoter region of the EPM1 gene at 21q22.3 which encodes cystatin B (Lalioti et al. 1997). This inhibits the papain family of cysteine proteases, involved in the initiation of apoptosis. Onset is also in the teenage years. Cognitive decline is not inevitable and if present may be another pointer to a progressive disease in the absence of frequent seizures such as Lafora body disease. Where there may be suspicion, a simple screen is provided by somatosensory potentials performed in conjunction with the EEG, whence an abnormally large somatosensory evoked potential would indicate that further evaluation is warranted (Fig. 30.2). The nature of further investigation would of course depend on timing of the clinical presentation (Table 30.1). Myoclonus becomes the main problem with a pseudoataxia.

Lafora body disease. This presents in the teenage years but may be earlier. It is a myoclonic epilepsy with episodes of generalized status epilepticus. Some however may present with apparent focal occipital lobe seizures. Progressive cognitive decline ensues quite rapidly, with only a short survival. Up to 80 per cent are caused by mutations in the EPM2A gene at 6q24 which encodes laforin (Serratosa et al. 1995, 1999). Diagnosis may be made by mutation analysis although screening is by skin biopsy from an area where apocrine sweat glands are present, usually the axilla, to look for Lafora bodies, consisting of excessive abnormal glycogen with excessively long linear peripheral chains.

Neonatal seizures may take many forms, but may be subtle. Such seizures make up to 50 per cent of the total, with clonic seizures making up the majority of the remainder. Rarely myoclonic or tonic seizures may be seen. Generalized tonic clonic seizures are virtually never seen (Pitt and Pressler 2005).

Subtle seizures may involve abstruse clinical signs including ocular manifestations, mouthing, sucking, autonomic symptoms, posturing, or pedalling, which may not have been interpreted as seizural in origin. In addition the differential diagnosis of abnormal movements in neonates is important; non epileptic jerky movements or jitteriness are common, particularly when there is metabolic imbalance such as hypocalcaemia. Apnoea is common in the premature infant but very rarely in isolation the result of seizure activity. Benign neonatal sleep myoclonus may be misinterpreted as seizural in origin by the inexperienced; the clue is jerks only occurring during sleep in an otherwise normal baby. Video EEG monitoring of at-risk babies in neonatal units has shown unrecognized electrical seizures to be common. Furthermore treatment with antiepileptic drugs may lead to resolution of clinical signs despite persistence of electrical seizures. The prognostic significance of this electrical-clinical dissociation is unknown, as the electrical seizure is more likely to reflect aetiology.

Interpretation of the neonatal EEG requires experience and expertise in view of the natural changes in EEG characteristics over time. EEG features of neonatal seizures include rhythmic activity, with sudden distinct onset and ending, over a minimal duration of 10 s. Focal origin with spread, electrodecremental events, and periodic discharges are also highly suspicious of seizure discharges (Mizrahi and Claney 2000). A burst suppression pattern to the EEG will be indicative of cerebral pathology, and a predisposition to seizures. Seizures will be related to hypoxic-ischaemic encephalopathy, cerebral haemorrhage, or infarction in as many as 80 per cent. Other causes include meningitis, metabolic defects, maternal drug withdrawal, and only rarely a neonatal epilepsy syndrome. In the vast majority therefore, seizures are acutely symptomatic and antiepileptic drugs are only required in the short term.

Spasms are a particular seizure type, particularly seen in the first year of life. They involve a sudden bilateral symmetrical contraction of the muscles of the neck, trunk, and extremities. Usually they occur repetitively in clusters. They are most commonly flexor, flexor-extensor, or more rarely purely extensor, reflecting the groups of muscles involved. Their intensity may vary, and in particular when manifesting subtly, for instance as head elevation or movement of only one limb, they may be overlooked clinically. Often a cry may be heard coincidentally or just after the spasm. In 6–8 per cent of patients spasms may be unilateral; and if so an underlying structural lesion of the brain should be suspected. Typically children presenting with infantile spasms will have a typical high amplitude disorganized slow waves interspersed with spikes and sharp waves, termed hypsarrhythmia (Fig. 30.3), but this does not always occur. Hypsarrhythmia is seen in about 40–70 per cent and thus an abnormal but not truly hypsarrhythmic EEG does not exclude the diagnosis of infantile spasms.

Although the semeiology of seizures in children may be distinctive as in adults, the manifestations in young children may be subtle and their epileptic basis unclear. Younger children with discharges of occipital onset may present with autonomic symptoms, such as vomiting and colour change, rather than overt visual phenomena. Those with a temporal lobe onset of seizures may simply exhibit behavioural arrest, with a less complex pattern of automatisms than are seen in older children or adults; sucking infants or swallowing automatisms are often interpreted as normal movements. Furthermore, a preictal warning may only manifest as the child seeking out an adult immediately before the seizure. Gelastic seizures present as isolated laughter occurring out of context and without mirth; a particular association is seen when such seizures arise as the result of a hypothalamic hamartoma.

True absence seizures will be associated with generalized spike wave activity on the EEG, and generally involve behavioural arrest with unresponsiveness. They are generally abrupt in onset, and brief in duration. In some instances there may be associated phenomenology such as eyelid movement, occasional mouthing, or some loss of tone. They are most often misdiagnosed as day dreaming, but more commonly daydreaming is misinterpreted as absence seizures. It should be borne in mind that behavioural arrest in children may be the sole manifestation of a focal seizure, leading to difficulty in distinguishing whether attacks reflect generalized absence or focal discharges unless defined by EEG.

Myoclonic attacks are typically seen as brief jerks of the limbs, or head nods, representing jerky movements of antagonistic muscle groups associated with spike wave discharges on the EEG, in which the spike corresponds to the myoclonic jerk. In some patients the phenomenon of negative myoclonus occurs in which the discharge itself is not associated with any muscle contraction for 50–400 ms, and then followed by a jerk as the EEG returns to normal.

Seen almost exclusively in children, atonic seizures are most likely to cause injury. Being of abrupt onset, and with a sudden generalized loss of tone, they cause the child to drop to the floor with a high risk of traumatic injury.

Tonic seizures involve sudden sustained stiffening of the whole body, associated with generalized fast activity on EEG. During sleep they may be extremely subtle, but if occurring whilst awake they may cause a drop to the floor whose considerable force leads to injury. Although the hallmark of certain distinctive epilepsy syndromes, such as Lennox–Gastaut (Section 30.5.5), such tonic seizures may be induced by medications including carbamazepine and phenytoin. Parents may have noted a characteristic noise, usually the result of forced expiration as muscles stiffen at the onset. Trembling movements may occur toward the end of tonic seizures leading to confusion with generalized tonic-clonic attacks.

Characteristically these seizures involve stiffness, followed by clonic jerking of the limbs. In children this clonic jerking may be subtle, and simply considered to be a twitching, particularly of the face. In general tonic-clonic seizures last longer than tonic seizures. Seizures of focal onset in children may also progress to become secondarily generalized, a more common occurrence in children aged under two. In such cases the child may experience a warning prior to the manifestation of generalized stiffening, or may exhibit a prior attack typical of a focal seizure.

The International League Against Epilepsy first proposed a classification of epilepsy, primarily by seizure type (1981) and subsequently revised according to epilepsy syndromes (Proposal for revised classification of epilepsies and epileptic syndromes 1989). A syndrome is defined as a cluster of symptoms and signs that define a unique epileptic condition. By definition this must include more than the seizure phenomena alone, and may include information from other investigations including as EEG and, increasingly, MRI. Many syndromes are age related, with many of the distinctive syndromes presenting in childhood, with characteristic seizure types and specific EEG features.

There are both advantages and disadvantages to using such a classification. Advantages include the ability to predict prognosis, which is helpful to outline to the parents at the outset. In addition, there may be generally accepted optimal pharmacological approaches to specific syndromes, defining particular drug responses that may be more favourable, or indeed drugs that should be avoided (National Institute for Clinical Excellence 2004). Disadvantages include the possible operational difficulties in making the diagnosis of specific syndromes, and the academic controversy that surrounds the delineation of certain syndromes.

This original classification subdivided the epilepsies into those which were localization-related and those which were generalized, with further categories for those ‘special syndromes’ which were neither. This classification also depended on the terminologies of idiopathic, where there was presumed to be no underlying brain pathology, or symptomatic, where there was underlying brain pathology, and of cryptogenic, where underlying brain pathology was presumed but unproven. Although there has been criticism in such classification, studies have shown its broad utility in classifying children at presentation in all but 12 per cent of cases, and that such diagnoses remain durable over time (Berg et al. 2000). Nonetheless some of the terminology is recognized as confusing, for instance the distinction between symptomatic and cryptogenic. Furthermore, with time the list of syndromes has lengthened dramatically yet without clarifying the definition between the well-accepted syndromes and those which are less well defined.

Recognizing the problems of the existing classification, the International League against Epilepsy proposed a further classification in 2001 (Engel et al. 2001). This proposes an axis system, that starts with the diagnosis of an epileptic seizure, moves to that of epilepsy, and then to the syndromic diagnosis (Table 30.2). It disposes of the terminologies of simple or complex as well as cryptogenic, which it proposes replacing by presumed symptomatic, and suggests the term focal instead of partial. A full list of existing syndromes is provided (www.ILAE-epilepsy.org). It also enables aetiology to be included in the classification, as well as any comorbidity. Such a classification is easier to understand, and can be used as a basis for teaching basic diagnostic principles.

The majority of the syndromes of epilepsy are age related, and therefore are diagnosed in childhood. Diagnosing a syndrome has considerable advantages; it enables a prognosis to be outlined from the start so expectations can be outlined, and may define suitable management, and particularly treatment.

An epileptic encephalopathy is defined as a condition in which the epileptiform abnormalities themselves are believed to contribute to the progressive disturbance in cerebral function (Engel et al. 2001). It occurs relatively commonly in early-onset epilepsies, one study estimated that 39 per cent of 504 consecutive children who had their first epileptic seizure were between the ages of 28 days and 3 years (Dalla Bernardina et al. 1983). By definition, it is presumed that the pathophysiological mechanism is ongoing epileptic activity at a critical point of brain development, in view of which it has to be thought of as potentially, at least in part, reversible. Accordingly optimal if not complete seizure control should therefore be the aim in all young children with epilepsy. The syndromes listed within the International League against Epilepsy classification as epileptic encephalopathies include Infantile Spasms or West Syndrome, severe myoclonic epilepsy of infancy (Section 30.5.4), Lennox–Gastaut syndrome, Continuous Spike Wave of Slow Sleep, and Landau–Kleffner syndrome (Section 30.5.5). However any epilepsy presenting with aggressive seizure onset and cognitive decline, in the absence of a progressive aetiology, could be defined as an epileptic encephalopathy, including myoclonic astatic epilepsy and some symptomatic focal epilepsies.

The majority of seizures presenting in the neonatal period are acutely symptomatic of underlying disorders (Section 30.4.1). In addition there are a small number of rare syndromes that commence in the neonatal period and which demand recognition.

Otherwise known as fifth day fits, this disorder presents between day 4 and 6 in up to 90 per cent of cases. Seizures may be frequent, involve clonic movements with or without apnoea, and resolve spontaneously within a few days. Typically children will present with seizures of unknown origin and therefore be fully evaluated for an acute symptomatic cause; EEG has been reported to show a characteristic feature ‘theta pointu alternant’ or may be normal. The long-term outcome is debated; in one study ‘abnormalities were reported in the long term in up to half suggesting not such a benign condition (Pryor et al. 1981; North et al. 1989). However in many studies follow-up is only of short duration and outcome cannot truly be evaluated (Plouin and Anderson 2005).

These have a slightly earlier onset on day 2–3. These children usually have a family history with evidence of autosomal dominant inheritance. Linkage has been demonstrated to potassium channel genes KCNQ2 and KCNQ3. The EEG is often normal. The prevalence at 11 per cent of later epilepsy in this group is higher than expected amongst the normal population in limited studies (Plouin and Anderson 2005).

These are a rare but treatable subgroup of neonatal seizures (Section 10.7.1 and Table 10.2, V2). Accordingly any child presenting with seizures within the first year of life should be considered for a trial of pyridoxine. In general this syndrome presents with generalized seizures soon after birth with associated encephalopathy involving hyperalertness, irritability, systemic features of thermal dysregulation, respiratory distress, and sometimes metabolic acidosis. Although traditionally diagnosed after a trial and ultimately trial withdrawal of pyridoxine the metabolic defect is now known to be a deficiency of brain α-aminoadipate semialdehyde dehydrogenase. Urinary α-aminoadipate semialdehyde may be diagnostic, and mutations of the aldehyde dehydrogenase, ALDH, 7A1 gene on chromosome 5q31 encoding antiquitin, or α-aminoadipic semialdehyde dehydrogenase, have recently been discovered as a major cause of pyridoxine-dependent epilepsy (Mills et al. 2006).

Febrile seizures are listed within the epilepsy classification, although by definition are single events in the majority. They may be defined as ‘an event in infancy or childhood, usually occurring between 3 months and 5 years of age, associated with fever but without evidence of intracranial infection or defined cause’ (National Institute of Health 1981). Ninety per cent occur before 3 years of age, the majority during the second year with a peak incidence between 18 and 24 months. The majority occur in isolation, lasting less than 10 min, and with no evidence of focality. Some children develop ‘complex’ febrile seizures, defined as a seizure lasting longer than 10 min, evidence of focality, or 2 or more seizures within 24 h or within the same febrile illness. There has been much discussion as to the relationship of early febrile seizures to the later development of epilepsy, particularly around the relationship of prolonged febrile seizures to the development of hippocampal sclerosis and temporal lobe epilepsy. The risk of developing epilepsy remains higher amongst those who have a complex febrile seizure (Verity et al. 1985). Adults coming to surgery for epilepsy from hippocampal sclerosis also have a higher prevalence of a history of febrile convulsions in childhood (Section 31.11.2). There is also evidence from animal studies and possibly from imaging, that this group run the risk of hippocampal damage (Liu et al. 1995; Scott et al. 2003). However, the genetic determinants remain complex, and other syndromes have been discovered where seizures are linked to fever. There is a risk of recurrence of febrile seizures in 30–40 per cent, a higher risk seen in those with early age of onset, family history, and a shorter duration of illness prior to the first seizure. There is no evidence that prophylactic medication prevents or alters outcome from febrile seizures, or that investigation outside looking for a source of infection is helpful. The latter is particularly important in children presenting under 12 months of age as intracranial infection may be masked. The main aim of management remains temperature control. A supply of emergency rescue medication may be indicated if there is a history of a prolonged seizure or a seizure recurrence during the same illness.

At present this entity is usually diagnosed on retrospective review of the history, the response to anticonvulsants, and the developmental progress (Watanabe et al. 1990, 1993). The true prevalence of this epilepsy remains unknown since most children presenting in this age group are likely to have seizures that respond poorly to anticonvulsants and have a poor prognosis with regard to neurodevelopmental outcome. It is recognized that this category includes a small number of children who presented with focal seizures, that responded promptly to anticonvulsants, have a good prognosis for long-term seizure remission, and good developmental progress. Many have a familial component.

Neonatal or early myoclonic epileptic encephalopathy and early infantile epileptic encephalopathy or Ohtahara syndrome are rare, present in the neonatal period or early infancy, and are both associated with a burst-suppression pattern on the EEG. Early myoclonic epileptic encephalopathy is characterized by fragmentary and erratic myoclonias as well as partial seizures. The main seizure types at the outset of early infantile epileptic encephalopathy are tonic spasms and partial seizures. Structural brain abnormalities appear to be more common in early infantile epileptic encephalopathy, whereas early myoclonic epileptic encephalopathy may be a manifestation of inborn errors of metabolism (Ohtahara and Yamatogi 2003). In both prognosis is poor with regard to seizure control and neurodevelopment although probably more so in early myoclonic epileptic encephalopathy with at least 50 per cent mortality before 12 months. Some cases of early infantile epileptic encephalopathy show a characteristic evolution with age to West and Lennox–Gestaut syndromes (Section 30.5.5) suggesting that early infantile epileptic encephalopathy is an age-dependent epileptic encephalopathy.

Possible metabolic causes of epilepsy are cited in Tables 10.2 V and 30.1 and Section 10.7.1. Two are of particular note:

This syndrome starts in the first year of life and has an extremely poor prognosis for control of seizures and developmental outcome. It is likely to have multiple aetiologies. The initial presentation suggests an early onset of focal epilepsy, relatively resistant to anticonvulsant medication. As the first year evolves, these seizures of initially focal onset show evidence of ‘migration’ to the contralateral hemisphere intra-ictally, as judged by clinical or EEG criteria (Fig. 30.4). Other features are recurrent status epilepticus, and proneness to seizures with intercurrent illness. The prognosis is extremely poor both for developmental outcome and mortality, with three of the original series of 14 not surviving (Coppola et al. 1995).

The triad of infantile spasms, hypsarrhythmia, and developmental plateau is termed West Syndrome, originally described by the paediatrician Charles West in his own son in the 19th century. However the presentation of typical spasms despite the absence of true hypsarrhythmia does not exclude the diagnosis and such children should receive the same treatment relatively urgently. The described EEG pattern is a supposed interictal pattern observed mainly in the awake state. Typical hypsarrhythmia is present mainly during the early stages of the disorder and preceding the clinical manifestations by weeks. However a profoundly abnormal EEG resembling hypsaarrhythmia (Fig. 30.3) may be seen with some global developmental malformations such as lissencephaly (Section 9.2.5) and is unlikely to change with treatment.

Given the highly disorganized nature of the EEG it is not surprising that the affected child’s awareness is likely to be impaired. Developmental slowdown may precede the spasms, and is an almost invariable finding at their onset, occurring in 68–85 per cent (Matsumoto et al. 1981; Riikonen 1984). Autistic withdrawal is not uncommon at onset and persists in a high proportion of children. The prognosis is strongly determined by the underlying aetiology and some degree of learning impairment is seen in up to 90 per cent, with only 18 per cent achieving a longitudinal intelligence quotient of >51 (Riikonen 1982). The evidence suggests that a better prognosis is related to earlier treatment of spasms, and in particular a shorter duration of hypsarrhythmia (Rener-Primec et al. 2006).

Diagnosing infantile spasms is important because it justifies early aggressive treatment. The treatment of choice remains between vigabatrin and steroids. Protocols differ in dosing regimes as well as in the duration of treatment courses depending on geographical location and availability of these drugs. Vigabatrin is not licensed in the USA, natural ACTH has been replaced by synthetic ACTH in some European countries and Japan, and different types of oral steroids are in use. Despite a large number of published trials investigating efficacy and tolerability of vigabatrin, glucocorticosteroids, ACTH, and other agents for the treatment of infantile spasms, only limited conclusions can be drawn due to differences in design and treatment protocols as well as poor overall quality of most randomized studies (Hancock et al. 2003; Mackay et al. 2004). Vigabatrin remains the medication of choice if infantile spasms results from tuberous sclerosis (Section 11.1) (Hancock et al. 2003). A recent multicentre randomized controlled trial compared the relative effectiveness of Tetracosactide, a synthetic ACTH, and Prednisolone versus vigabatrin for the short-term treatment, of cryptogenic and symptomatic infantile spasms excluding patients with tuberous sclerosis (Lux et al. 2004). It was observed that 78per cent of patients enrolled to Prednisolone and Tracosactide ceased spasms by day 14 of treatment versus 54 per cent on vigabatrin (mean difference 19 per cent, 95 per cent CI 1–36 per cent, P=.045). After a year there was no difference between the treatment groups with respect to proportion of patients with absence of spasms, 75 per cent, and seizure-free patients, 57 per cent (Lux et al. 2005). More robust evidence is required to make a clear recommendation on the best treatment of infantile spasms. The choice remains between steroids and vigabatrin as first-line therapy and this may be influenced by the aetiology, and preference of the physician and parents.

This has become a well-recognized clinical syndrome shown to be associated with a sodium channel gene mutation, SCNIA, on chromosome 2 (Harkin et al. 2007). A typical course will include presentation with prolonged and probably febrile lateralized seizures in the first year with normal early development. This is followed in the second decade by the development of other seizure types, including myoclonic jerks, as well as developmental plateauing (Dravet et al. 1992). The early EEG may be normal; 40 per cent show some degree of photosensitivity that may resolve. Sometimes the condition starts relatively later, or has other atypical features, and has been termed severe myoclonic epilepsy borderline (Harkin et al. 2007). Up to 80 per cent have SCNIA gene mutations. Phenotypic correlation with specific mutations has been attempted and some of the variability may reflect mosaicism (Marini et al. 2006). Although seizures may be resistant to medication, and children may continue to be troubled by recurrent status epilepticus, particular responsiveness has been demonstrated to sodium valproate, clobazam, stiripentol (Chiron et al. 2000), and topiramate (Ceulemans et al. 2004). Certain medications, namely lamotrigine, need to be avoided as they may exacerbate seizures (Guerrini et al. 1998).

This syndrome is characterized by heterogeneous epilepsies including febrile seizures and mild to severe generalized epilepsies (Scheffer and Berkovic 1997). The most common phenotypes are febrile seizures and febrile seizures plus, in which seizures persist beyond 6 years of age. The most severe phenotypes include febrile seizures plus with absences, myoclonus, or atonic seizures. The genetics are important as they link febrile seizures with epilepsy. Mutations in sodium channels genes, SCN1A and SCN1B, and GABA_A receptor subunit genes, GABRG2 and GABRD, have been described in different pedigrees (Scheffer and Berkovic 2003).

Children with structural lesions of the brain may present with devastating onset of focal epilepsy in infancy, sometimes associated with an apparent epileptic encephalopathy. All children presenting with epilepsy in association with a unilateral structural lesion should be referred to an epilepsy surgery programme for evaluation at an early stage given the likelihood they will not respond to anticonvulsant medication; surgery should be considered early on to maximize their developmental potential (Cross et al. 2006). Furthermore, children presenting apparent focal epilepsy without an aetiology evident on MRI should be evaluated in detail given that lesions may be difficult to detect on initial imaging, or even become less apparent with normal myelination of the brain (Eltze et al. 2005).

Although initially identified as an early-onset epilepsy with occipital spikes or occipital paroxysms, with or without ‘fixation-off sensitivity’, it is now recognized that it is more akin to an ‘autonomic’ epilepsy, and is commonly called Panayiotopoulos syndrome (Panayiotopoulos et al. 1993). Children may present with only a single or with recurrent seizures whose semeiology is typically a change of awareness, and vomiting with or without other autonomic manifestations such as flushing or pallor. The age of onset is usually 3–6 years. Such episodes may be prolonged and perhaps misdiagnosed as a non-epileptic encephalopathy. Occipital spikes may be seen on EEG in up to 40 per cent. The prognosis is excellent with most children having only 1–3 seizures, with remission occurring 1–2 years from onset. Thus the need for treatment needs to be carefully considered, although most will respond promptly to carbamazepine. Brain MRI is advisable as structural occipital lesions may present with similar seizure semeiology.

This syndrome is thought to be an idiopathic epilepsy (Doose et al. 1970). It is characterized by the onset of generalized tonic clonic seizures and later myoclonic, astatic, and myoclonic-astatic seizures causing drop attacks. Seizures may occur very frequently, and children may be prone to periods of nonconvulsive status epilepticus. Although not classed as one of the epileptic encephalopathies, affected children may be obtunded during periods of very frequent seizures. Although seizures may be frequent and difficult to control in the early stages, this eases after 1–3 years. Those affected by frequent drop attacks may have been classified previously as having Lennox–Gastaut Syndrome. The prognosis for long-term cognitive outcome remains relatively better than for the other early-onset seizure syndromes with 75 per cent achieving the normal range (Guerrini et al. 2002). Poor prognostic indicators include frequent episodes of nonconvulsive status epilepticus or the development of tonic seizures. The anticonvulsants most likely to be of benefit include sodium valproate, with or without lamotrigine, benzodiazepines, ethosuximide (Doose et al. 1987), and levetiracetam (Labate et al. 2006). There is also anecdotal evidence that these children may be particularly responsive to the ketogenic diet (Section 30.7.4) (Oguni et al. 2002) and this should receive early consideration.

One of the epileptic encephalopathies, this syndrome may present first in early childhood with multiple seizure types, or may evolve in a child who has previously presented with West syndrome (Section 30.5.4). It was originally described by Gastaut as an epileptic encephalopathy with diffuse spike and wave complexes and multiple types of attack, including tonic seizures (Gastaut et al. 1966). Although the term has been broadened often to describe difficult epilepsies of childhood including drop attacks, it should be defined by a triad of symptoms including multiple seizure types, slow spike wave complexes on EEG, and impairment of cognitive function. Tonic seizures occur in a high percentage of patients with slow spike and wave complexes (Gastaut et al. 1966; Chevrie and Aicardi 1972) and their documented presence is a prerequisite for a diagnosis of Lennox–Gastaut syndrome. Atypical absence and atonic or tonic falls are also seen in the majority. Generalized paroxysmal fast activity is the second EEG hallmark of Lennox-Gastaut syndrome, with or without seizures, and a period of sleep recording may be required to reveal their presence. Effective treatment options for the multiple seizures and comorbidities of Lennox–Gastaut syndrome remain limited and long-term prognosis is poor for many patients. The newer agents topiramate, lamotrigine, and felbamate have been shown to be effective in randomized controlled trials (Hancock and Cross 2003).

These two syndromes, although often listed together are not synonymous. Continuous spike wave of slow sleep is a syndrome characterized by neuropsychological and behavioural change temporally related to the development of near continuous spike wave activity on EEG during slow wave sleep. It was first described as ‘subclinical status epilepticus’ induced by sleep in children (Patry et al. 1971). Although the terms continuous spike wave of slow sleep and electrical status epilepticus of slow sleep are used synonymously, some investigators have proposed that electrical status epilepticus of slow sleep should be used to designate the EEG abnormalities and continuous spike wave of slow sleep syndrome for the combined electroclinical picture (Galanopolou et al. 2000). Initially a proportion of 85 per cent of slow wave sleep needed to be occupied by spike and wave activity as a criterion of diagnosis, but several authors now accept a lower proportion. Clinical associations include global or selective neuropsychological impairments such as acquired aphasia, or dyspraxia. Motor features include ataxia, dystonia, or unilateral deficits. The seizure types vary with focal and or apparently generalized seizures, tonic clonic seizures, absences, partial motor seizures, complex partial seizures, or epileptic falls; however tonic seizures are never seen.

Landau–Kleffner syndrome is the condition of acquired epileptic aphasia. It is characterized by acquired language regression associated with some seizures, but more importantly an epileptiform EEG pattern, often bitemporal but in some cases typical of electrical status epilepticus of slow sleep. Whilst the clinical presentation is diverse, the treatment may remain similar, with steroids and benzodiazepines being most likely to be beneficial. Lamotrigine and levetiracetam have been reported to be effective. The prognosis regarding both seizures and the EEG abnormality remains good in both conditions, although the cognitive and language is relatively unpredictable and may relate to early treatment success.

This is the most common epilepsy presenting in childhood. Typically children present aged between 5 and 10 years with nocturnal seizures, either focally involving the mouth and face, or generalized. This benign condition carries an excellent prognosis, with all children reported to enter remission by 14 years of age (Loiseau et al. 1988). Some have deemed treatment unnecessary in view of the guaranteed remission. Despite the good prognosis for remission, there is still the risk of recurrent generalized seizures, which carry the risk of morbidity, and 10–20 per cent experience seizures exclusively from the awake state (Lerman 1985). In addition some report that benign epilepsy with centrotemporal spikes may not be so benign because of associated cognitive and behavioural effects which may be related to the frequency of seizures, therefore representing an epileptiform phenomenon (Metz Lutz et al. 1999). Thus the decision to treat will depend on the degree to which the seizures are interfering with the child’s life. The treating physician should decide whether treatment is desired together with the family and child bearing in mind that there is likely to be an excellent response to antiepileptic drugs and that these will not be needed long term. The medication of choice is carbamazepine with prompt seizure control in 80 per cent of patients.

A small group of patients presenting similarly show some atypical features. A progressive course with cognitive regression is associated with progression to electrical status epilepticus of slow sleep. Caution is required when initiating carbamazepine in such patients as it may exacerbate electrical status epilepticus of slow sleep. Nonetheless, the vast majority of children presenting with the features of benign epilepsy with centrotemporal spikes are likely to have a benign course, and any unexpected deterioration on treatment requires further evaluation by sleep EEG and a change of medication should be considered. Some consider that if an initial sleep EEG recording has demonstrated very frequent discharges and there is concern about cognition, an alternative medication such as valproate is indicated.

This has a later mean age of onset than the benign epilepsy with occipital spikes of Panayiotopoulos (Section 30.5.5). The seizures almost always involve visual phenomena at the outset, which may be well described and drawn by the child. The seizures may occur in isolation, or proceed to secondary generalization with hemiclonic seizures in about 40 per cent. One-third of the original series of patients had a severe postictal headache, at times accompanied by nausea or vomiting (Gastaut et al. 1992). Therefore the condition may be confused with migraine. The interictal EEG demonstrates unilateral or bilateral paroxysmal posterior spike waves on eye closure ‘fixation off sensitivity’, as does the benign epilepsy with occipital spikes in younger children (Section 30.5.5). Full remission by late teenage occurs in the majority.

This syndrome accounts for about 8 per cent of epilepsy in school age children (Cavazzuti 1980). Children present with brief episodes of behavioural arrest, of which the EEG correlate is a 3 Hz spike and wave (Fig. 30.5). Attacks are unprovoked, and occur in any situation. They may be provoked in clinic or during EEG recordings by hyperventilation. The syndrome is almost certainly genetically determined, with 75 per cent concordance for monozygotic twins (Hauser and Anderson 1986). The prognosis is excellent with up to 80 per cent achieving long-term remission. The children with the best outlook are those with absences commencing early in childhood, whose absences are not associated with myoclonus and who do not have tonic-clonic seizures preceding or coinciding with the onset of absences. Medications that may be effective include sodium valproate, lamotrigine, and ethosuximide. The absence attacks may be exacerbated by medications such as carbamazepine and phenytoin.

Children present with a later onset of absences than childhood absence epilepsy, but with a slightly faster frequency of slow wave activity (Section 31.5.1).

Myoclonic absence epilepsy. This may be confused at presentation with childhood absence epilepsy, especially since the EEG abnormality is also likely to be a 3 Hz spike-wave discharge. However attacks are characterized by rhythmic jerks predominantly of the upper limbs that also occur at 3 per second. Small longitudinal series suggest that learning difficulty is common in this group, even prior to presentation. The prognosis for seizure control is guarded; around 50 per cent go into remission but this is less likely if generalized tonic clonic seizures occur (Tassinari et al. 1995). The response to anticonvulsant medication is variable; good response has been seen with valproate or ethosuximide (Tassinari et al. 1995) or with either combined with lamotrigine (Manonmani and Wallace 1994).

Eyelid myoclonus associated with absences. Also known as eyelid myoclonia with absences, this was first reported by Jeavons (1977), with subsequent extensive description by Giannakodimos and Panayiotopoulos (1996). The onset is in childhood and the seizures begin with rhythmic fast brief jerking of the eyelids and upward jerking of the eyeballs, followed by brief mild absence. The attacks occur mainly on closing the eyes. The EEG shows 3–6 Hz generalized polyspike and wave, and all patients are photosensitive. Most patients experience occasional tonic-clonic seizures on sleep deprivation. Treatment is as for other idiopathic generalized epilepsies using sodium valproate either alone or in combination with ethosuximide or clonazepam. Although the photosensitivity may disappear with age, the eyelid myoclonia persists and remains resistant to treatment even if absences appear controlled.

Perioral myoclonia with absence. Myoclonus of the lower face may be associated with absence attacks. This involves rhythmic contractions causing protrusion of the lips and twitching of the corners of the mouth associated with a mild absence (Panyiotopoulos et al. 1995). These seizures may build up to absence status and then to a tonic-clonic seizure. Although similar medications should be trialled as in childhood absence epilepsy, this syndrome may not respond well to treatment.

Rasmussen syndrome is a progressive condition that most commonly presents between 5 and 10 years, although may occur outside these age limits, even in adulthood. Classically children present with a focal epilepsy that becomes increasingly difficult to treat. Commonly this evolves to a hemiepilepsy, 60 per cent experiencing epilepsia partialis continua at some time in their clinical course. This often heralds an increasing hemiparesis. Hemiatrophy is seen of the contralateral cerebral hemisphere (Bien et al. 2005) (Fig. 30.6).

The term ‘encephalitis’ originates from the original surgical studies by Rasmussen who on performing hemispherectomy on some of these individuals noted that they had pathological evidence for a chronic encephalitis of no apparent aetiology. The evidence now points to a causative autoimmune process. Original studies suggested humoural immunity with a role for GluR3 antibodies, but these have been shown to be non-specific and cellular immunity involving T-cell dysfunction is now considered likely (Li et al. 1997; Bauer et al. 2002; Watson et al. 2004). The mechanism by which such immune disorders produce a uni-hemisphere abnormality remains unexplained.

Despite variable responses to steroids and intravenous immunoglobulin (Hart et al. 1994), the only curative treatment available remains hemispherectomy. Although this abolishes seizures it necessarily results in hemiparesis and hemianopia. Detailed presurgical assessment is required for such patients to enable optimal planning of any surgical procedure, particularly given the risk of cognitive decline and the influence of cerebral dominance. The decision about when to operate needs to balance the development of cognitive decline against the risk of functional deficits should surgery be undertaken. The key determinant is language function, although some degree of transfer and recovery may be seen even after relatively late onset disease and surgery (Boatman et al.1999).

Many of the idiopathic generalized epilepsies persisting into adulthood may originally present in mid- to late childhood. Juvenile absence epilepsy (Section 31.5.1) is likely to have a slightly later onset than childhood absence epilepsy (Section 30.5.6), and the spike wave activity is of slightly faster frequency at 4–5 Hz. The prognostic implication is that although individuals may respond promptly to anticonvulsants, they are unlikely to wean successfully from the medication. Furthermore some patients may present as the onset of juvenile myoclonic epilepsy (Section 31.5.1), with the myoclonus and generalized tonic-clonic seizures only manifesting at a later stage.

It is estimated that 65 per cent of epilepsy that starts in childhood persists into adulthood. There are particular syndromes where this is likely. Of the idiopathic group this includes the later onset idiopathic generalized epilepsies, as well as the benign epilepsies with occipital paroxysms and the Lennox–Gastaut syndrome (Section 30.5.5). In absence epilepsies it can be difficult to tease out those risk factors that may make an individual less likely to grow out of epilepsy. The defining features, particular those that may indicate poor prognosis are outlined in Table 30.3. These include a very early or a later age of onset, prolonged duration of attacks, and occurrence of generalized tonic-clonic convulsions.

Studies over the past 10 years have revealed a central role for genetically determined ion channel abnormalities in the pathophysiology of the idiopathic epilepsies. However many of these epilepsies follow complex inheritance and do not have single gene association. Epilepsies with monogenic inheritance are uncommon and are associated with mutations in genes that encode subunits of voltage-gated and ligand-gated channels (Table 30.4). Whereas in some inheritance appears autosomal dominant, in others inheritance is more complex and genotype–phenotype correlation is not clear. Nonetheless, the discovery of such genetic associations has improved our understanding of these epilepsies, and sometimes contributed to management. Understanding will increase in the future although at present many genetic studies still remain only available on a research basis.

It is imperative to determine family history on evaluating a child initially presenting with epilepsy since the genetics can provide an important clue to the diagnosis. However the exact nature of the reported episodes is just as important and may be described by older members of the family.

There are many genetically determined disorders of which epilepsy is part of the symptom presentation. In children with chromosomal anomalies there is increasing recognition that epilepsy is a frequent and significant part of the clinical problem. Furthermore, underlying chromosomal anomalies are found in an increasing number of children where epilepsy had been the manifesting feature of an otherwise undetermined disorder. It is likely that loss of or abnormal functioning of genetic loci on involved chromosomes involves mechanisms leading to increased seizure susceptibility, cortical excitability, or changes in neurotransmitter functions. Alternatively epilepsy may originate in structural brain abnormalities occurring in association with the chromosomal abnormality.

Although ranging in clinical severity epilepsies occurring in the setting of chromosomal anomalies are generally difficult to treat. In a few, such as Angelman syndrome, the epilepsy or the EEG pattern is characteristic or very suggestive of that chromosomal anomaly. However in the majority of disorders this is not so although our understanding of karyotype–phenotype relationships continues to develop. Increasing knowledge of the EEG patterns and epilepsies associated with particular chromosomal disorders may contribute to diagnosis. Also the presence of a chromosomal anomaly may suggest target genes responsible for seizure pathogenesis.

Angelman syndrome occurs in 1 of 2000–12 000 in the general population and may account for up to 6 per cent of cases of severe

mental retardation and epilepsy. The genetic abnormality in the majority of cases is a large deletion of the maternally inherited chromosome 15q11-13. Yet deletion of this same region of the paternally inherited chromosome also results in Prader–Willi syndrome in which seizures are uncommon (Magenis et al. 1990). Epilepsy is very common in Angelman syndrome, reportedly occurring in up to 90 per cent, with onset usually between 18 and 24 months of age, often with convulsions and fever (Viani et al. 1995). Tonic-clonic, atypical absences, complex partial, myoclonic, tonic, and atonic have all been reported, often occurring in clusters. They may become quiescent in later childhood and only emerge again in adulthood (Laan et al. 1996). Status epilepticus of generalized tonic-clonic, absence, and myoclonic forms all occur. In particular nonconvulsive status whether myoclonic or atypical absence, may go unrecognized in the context of the associated developmental delay despite treatment potentially leading to significant benefit (Viani et al. 1995) The EEG is characteristic in Angelman syndrome (Fig. 30.7):

These EEG features may be present at different times in the same patient and suggest the diagnosis of Angelman syndrome before emergence of the full clinical phenotype. The presence of all three EEG features in the setting of developmental delay provides strong evidence for Angelman syndrome.

Cortical myoclonus is particularly well described as responding to piracetam (Guerrini et al. 1996). Initially the seizures are particularly resistant to treatment, but this improves in later childhood. Sodium valproate has been noted to be useful either with or without a benzodiazepine (Viani et al. 1995). Topiramate and phenobarbitone may also be effective (Franz et al. 2000).

This syndrome includes lissencephaly (Section 9.2.5) in association with a characteristic facies of prominent forehead, bitemporal hollowing, a short nose, and protuberant upper lip. It results from deletions or rearrangements of 17p13.3, involving the LIS-1 gene. The EEG may show widespread high amplitude spike and slow wave discharges reminiscent of hypsarrhythmia. Later the EEG may show the widespread fast activity commonly seen in association with such brain malformations. As a result of the lissencephaly, the epilepsy will present early and may prove very resistant to medication.

There are increasing reports of epilepsies associated with ring chromosome abnormalities. In the majority these are not associated with the dysmorphic features suggestive of a karyotype abnormality.

The ring abnormality of chromosome 20 is that most commonly reported. In 50 per cent of such children epilepsy presents before the age of 6 years and characteristically the seizures have a focal nature with visual symptoms, nocturnal tonic seizures, or arousals with a frontal semeiology often being reported. The patients are prone to nonconvulsive status epilepticus. The EEG may manifest as continuous bifrontal rhythmic theta/delta waves with accompanying spikes or sharp waves or may show continuous diffuse abnormalities (Ville et al. 2006). Typically the response to medication is poor. Furthermore the rate of behavioural disorder is high, with poor attention and concentration, impulsivity, disinhibition, obsessive behaviours, and aggressive outbursts being common. In addition, patients may acquire cognitive difficulties over time.

Patients with ring chromosome 14 present with epilepsy of earlier onset, severe to profound learning difficulties, speech impairment, microcephaly, and dysmorphism, particularly with ocular abnormalties (Schmidt et al. 1981).

Malformations of cortical development, ischaemic lesions, and tumours remain the most common structural causes of epilepsy in childhood. MRI has greatly enhanced the sensitivity of detecting these abnormalities, leading to further insights into the causes of childhood epilepsy. Brain malformations pose a high risk of drug-resistant epilepsy (Fig. 30.8). Although the prevalence of malformations is low in a newly diagnosed population, it rises as drug-resistant populations are reviewed and ultimately is highest amongst those being assessed for surgery. Classifications of malformations have been founded on neuroimaging and histological criteria, and more recently there has been an increasing contribution from genetics with the discovery of causal mutations. A summary of the most recent proposed classification scheme is summarized in Table 30.5 (Barkovich et al. 2005). The full range of disorders of brain segmentation, of neuronal and glial proliferation, and of neuronal migration are discussed in Sections 9.2.3, 9.2.4, and 9.2.5 respectively.

At the initial diagnosis it is important to discuss all aspects of diagnosis and management. This includes basic principles of management, such as advice on safety in the event of the child having a seizure. Children should be able to participate in as many activities as possible, although with precautions. Climbing is likely to be dangerous and therefore not advised although other activities are possible with precautions—for example swimming under one-to-one supervision or cycling provided not in busy traffic. In an older child one should consider discussing lifestyle issues such as eventual driving, alcohol consumption, contraception, and conception.

In the majority of children medication is likely to be advised after the diagnosis of epilepsy has been made, that is after the second seizure. The choice of medication and its timing will depend on the particular diagnosis of the epilepsy, and requires full discussion with the family. In certain circumstances treatment may be postponed since certain idiopathic syndromes may not warrant immediate treatment. For example, benign epilepsy with occipital spikes (Section 30.5.5) may present with single episode of status epilepticus, and no further seizures thereafter. Children with benign epilepsy with centrotemporal spikes may have infrequent seizures, although the designation of ‘benign’ refers to the prompt response to treatment if prescribed and the tendency to grow out of seizures; nonetheless, the seizures themselves may well cause morbidity. A full discussion should balance the risks of the epilepsy against those of the treatment and should involve parents, any other carers, and often the child so as to decide on the advisability of starting treatment.

In other circumstances treatment should be considered early on if the risks of recurrence are considered high after a single seizure, for instance where there is an associated neurological abnormality, or where a diagnosis of idiopathic generalized epilepsy is suspected on EEG.

When a decision has been made to start treatment, the most suitable anticonvulsant drug needs to be considered. This depends particularly on the type of epilepsy, or preferably the epilepsy syndrome, which has been diagnosed. Both conventional and newer antiepileptic drugs are used in children; their licensing position will depend on country and the availability of efficacy and safety data which has been obtained in children. Drug dosage needs to be adjusted in relation to the specific pharmacokinetic characteristics in young children, such as slower gastrointestinal absorption, higher volumes of distribution, and shorter clearance periods. In particular the reduced clearance rates can lengthen with age, and further dose reduction may be required to avoid the risk of toxicity as the child gets older. In any situation, general principles of antiepileptic drug introduction remain the same, whether a drug is required, if so which drug may be best for the child and the introduction of that drug slowly, with subsequent titration according to clinical response. The initial decision requires choice of which antiepileptic drug is indicated and to introduce this in a gradual titration to the optimal dose. If this fails, an alternative drug should be introduced slowly, and if effective, the initial medication should be slowly withdrawn. Polytherapy should be avoided; although two medications may be synergistic in their action there is little evidence to suggest more than two medications will improve efficacy. However, occasionally a small minority of children require more than two drugs for optimal seizure control. Side effects may be problematic and their possibility should be discussed prior to drug introduction; the concerns of the parents should be considered. It is important to remember that seizures may be aggravated by certain medications, such as the exacerbation of electrical status epilepticus of slow wave sleep by carbamazepine (Section 30.5.5).

The aim of treatment is always to achieve seizure control with minimal if any side effects. If full seizure control cannot be achieved it may be worth considering slow withdrawal of medication so as to achieve optimal rather than complete seizure control. Of course, in some circumstances seizure control may not be the primary area of concern of parents and the coexistent comorbidities may demand greater attention (Section 30.7.7).

The ketogenic diet has been used in the treatment of epilepsy for almost one hundred years. An osteopath in the early 20th century, prior to the introduction of anticonvulsant medication, discovered that starving people with epilepsy of all but water led to remission of seizures. However, such extremes of treatment cannot be advised. In 1921 Wilder suggested the use of a diet to mimic the metabolic effects of starvation; whereas in starvation the body breaks down body fat to produce ketones, a similar effect can be achieved by giving fat within the diet as the main energy source (Wilder 1921). Cohort studies have established that this can be effective although it lost favour when anticonvulsant drugs were introduced. Yet with time it became apparent that not all respond to anticonvulsant drugs and that the diet can be extremely effective for selected individuals. But the target population remains to be clearly delineated.

The classical ketogenic diet has provided a main fat source of long chain fat and is designed on the basis of a fat to carbohydrate ratio, including protein, of 3 or 4:1. Many of the studies in the early years report use of this diet (Freeman et al. 1998). Possible side effects of the diet provoked concern, and a more palatable way of giving the diet was proposed in which the main energy source was provided by medium chain triglyceride. It remained low in carbohydrate (Huttenlocher et al. 1971). A recent randomized controlled trial revealed a ketogenic diet to be effective over and above no change in treatment and no difference in efficacy between the classical and medium chain triglyceride ketogenic diets (Neal et al. 2008). In these circumstances parental choice or the child’s wishes may confound interpretation of efficacy; this needs further examination.

Children may be considered for a ketogenic diet if they fail to respond to anticonvulsant medication. The diet requires a high degree of dietetic advice, for meal calculation, as well as considering the commitment by the parents and the child. There is no clear evidence that any age group responds over and above any other (Freeman et al. 1998). Although there is little experience in adults; the ketogenic diet poses different compliance problems for teenagers in view of the protein restriction. In addition there is no clear evidence as to which syndromes or seizure types may respond best, although anecdotal evidence suggests particular benefit in myoclonic astatic epilepsy, and reports in younger children suggest a particular benefit in infantile spasms (Kossoff et al. 2002; Nordli et al. 2002).

Status epilepticus is defined as recurrent repetitive seizures without regaining full awareness in between seizures or as continuous seizure activity for more than 30 min. Aggressive treatment of prolonged convulsive seizures is justified in view of the likelihood of increased morbidity in seizures lasting longer than 30 min. Treatment is determined by protocols, with defined times after which poor response to treatment requires intensive care (Appleton et al. 2000) (Fig. 30.9). Emergency treatment of seizures in the community has been enhanced by the availability of rescue medication in the form of rectal diazepam, or more recently buccal midazolam (Scott et al. 1999). However once in hospital intravenous protocols should be used; the most common cause of admission to the intensive care unit is overmedication with benzodiazepines (Chen et al. 2004). Initial treatment should be with a benzodiazepine, but following the protocol to load promptly with phenytoin if there is no response to the benzodiazepine; repeated dosage with benzodiazepines is undesirable.

A distinction should be made between convulsive and nonconvulsive status. Nonconvulsive status is a change in clinical state or behaviour, in association with change in EEG from the usual baseline. When occurring within syndromes where it may be more prevalent, such as Lennox–Gastaut Syndrome, there is only slim evidence that similarly aggressive treatment to that used in convulsive status is justified; indeed further change from baseline may be difficult to determine. Nevertheless it is important to be vigilant as to the possible occurrence of nonconvulsive status in view of the possible impact on cognition and learning.

Children with symptomatic focal epilepsy which is unresponsive to medication should be considered early on for curative surgical resection. Traditionally there has been a reluctance to consider children as candidates for surgery due to the perception of uncertain seizure prognosis, and the relatively invasive nature of the presurgical evaluation. However, with improvements in syndromic diagnosis, and the definition of aetiology allowed by modern imaging, along with advances in neurosurgery and neuroanaesthesia, risks have been minimized and surgery has become available to a wider range of children with epilepsy.

Like adults, children may be considered either for resective surgery, or in certain circumstances for functional procedures in which tissue is not removed but function modified (Section 31.11). In children the range of potential procedures is wide, particularly the resections, and these differ proportionally from adult practice. Similar principles apply, particularly that seizures should be proven to arise from one functionally silent area of the brain, of variable size. Hemispherectomy or multilobar resection account for about two-thirds of procedures undertaken in children (Harvey et al. 2008). Moreover children with early-onset catastrophic epilepsy of focal origin, and those with early-onset focal epilepsy associated with certain syndromes, such as Sturge–Weber syndrome (Section 11.3) or tuberous sclerosis, (Section 11.1) should be considered for surgery early in the natural history of their epilepsy (Cross et al. 2006).

Freedom from seizures is the primary aim of paediatric epilepsy surgery. However secondary aims may be of equal, if not of primary, importance to some families and require careful consideration with counselling prior to surgery. Seizure outcome from focal resection relates to the completeness of the resection, whether determined structurally (Edwards et al. 2000) or electrographically (Paolicchi et al. 2000). Wider procedures such as hemidisconnection primarily relate to the underlying pathology, with lower rates of seizure freedom seen for developmental malformations, particularly hemimegalancephaly (Section 9.2.4) (Devlin et al. 2003). Developmental and behavioural outcomes are often of concern to parents; an argument in favour of early surgery has often been that of improved developmental outcome. Yet objective support for this is scarce. Much of the data on neurodevelopmental progress appears to show a maintained intelligence quotient trajectory against peers (Devlin et al. 2003; Freitag and Tuxhorn 2005). Small prospective series suggest that surgery performed under the age of 1 year is likely to improve developmental trajectories than surgery performed later (Loddenkemper et al. 2007). This is particularly important given that almost 50 per cent of children coming to surgery have seizure onset under 12 months of age (Harvey et al. 2008). Little data is available on behavioural outcomes. Rates of behavioural difficulty are high in this population, but studies show that little prediction can be made as to possible improvement or deterioration after surgery (Devlin et al. 2003; McLellan et al. 2005).

Vagal nerve stimulation (Section 31.11.5) has a place in the management of children with drug-resistant epilepsy, although its exact role remains unclear. Data overall suggest that in children whose epilepsy is not associated with an identifiable underlying lesion, and who are not resective surgical candidates, an improvement of at least 50 per cent is likely to be seen in 40–45 per cent (Wheless and Maggio 2002; Uthman et al. 2004). Freedom from seizures cannot be a goal of this procedure and realistic expectations about outcome must be discussed with families. Some early data on small numbers of children suggested accumulated benefit over time but this has not been replicated in larger series. Benefits beyond reduced seizure frequency may also be seen such as reduced seizure duration or increased awareness.

Integral to the management of children with epilepsy is the recognition and management of coexisting learning or behaviour difficulties. The occurrence of learning difficulty in children with particularly early-onset epilepsy is high, although the pathophysiological explanation remains unclear in many. In those where it precedes the onset of the epilepsy it is likely to be related to the underlying aetiology. In others where developmental plateau or regression occurs at the onset or during the course of the epilepsy, it is presumed to be related to the underlying epileptic activity, an epileptic encephalopathy (Section 30.5.2). In many it is likely to be a combination of factors, with possible contributions also from medication side effects. It is important to recognize such difficulties in view of the likely impact on schooling and possibly behaviour.

The rate of behavioural difficulty seen in children with epilepsy is greater than in children with other non-neurological chronic diseases (Davies et al. 2003). There is also data to suggest such behaviour difficulties antedate seizure presentation (Austin et al. 2001). Whether such difficulties reflect part of the underlying aetiology to the epilepsy itself or represent sub-clinical seizure activity often remains unclear. Higher rates of behaviour disorder are seen in children with ‘complicated’ as opposed to ‘pure’ epilepsy (Davies et al. 2003), suggesting a major contribution from the underlying aetiology. Other compounding factors include seizure occurrence, other ongoing electrical activity, medication, and psychosocial influences. There is no reason why such children should not be assessed as any other child with behaviour disorder, and behaviour modification programmes tried prior to consideration of medication.