Hong Kong Journal of Paediatrics [HK J Paediatr (New Series) 1996;1:169-183]

Department of Paediatrics Neurology, Royal Hospital for Sick Children, Edinburgh, U.K.

J Keith Brown MB, DCH, FRCPE
Mary O'Regan MB, DCH, MRCP(UK)

Department of Paediatrics, Queen Elizabeth Hospital, H.K.

Shun-Ping Wu MB, MRCP(UK)

Correspondence to: Dr J Keith Brown

Received Jan 24, 1996

Introduction

Epilepsy is a symptom of recurrent cerebral electrical disturbance. Not infrequently it is accompanied by other symptoms and signs of cerebral dysfunction. However, in some children epilepsy, is the only symptom and therefore it becomes a disease entity on its own right.

The clinical manifestation of epilepsy in children is much more intriguing than in adults since the brain is a developing organ and not static so the pattern of fits changes with age. For instance, infantile spasms can evolve into Lennox-Gastaut syndrome and then complex partial epilepsy, and febrile convulsions can develop into absence epilepsy and then complex partial epilepsy. Many epileptic syndromes are restricted to children only, and it is estimated that at least 50 per cent of childhood epilepsies last no more than two years.¹ This is the reason why paediatric neurologists always attempt a withdrawal of antiepileptic drugs after a two year seizure-free period. The chance of successful withdrawal is as good as 75 per cent.

This article discusses the present classification system and proposes a pragmatic method of classifying epileptic syndromes and its implication for treatment.

Terminology

Antiepileptics are really "anti-electric" drugs and if a seizure is not secondary to a disturbance in the brain electrical system little is gained by risking toxic effects. Most clinicians make little distinction between (a)fit, (b)seizure, (c)convulsion and use these terms interchangeably. To avoid confusion they should be defined as follows:

It is the clinical manifestation of a cerebral dysrhythmia, which may be convulsive (as in generalized tonic or tonic-clonic convulsions) or non-convulsive (as in absence seizures or complex partial seizures). A temporal classification can be applied to cerebral dysrhythmias, that is acute, chronic or continuous.

Acute cerebral dysrhythmias complicate acute disease of the nervous system, such as trauma (birth trauma, accidental and non-accidental head injuries), acute asphyxial episodes (neonatal asphyxia, cardiac arrest, complications of cardiac surgery, drowning and smothering), meningitis, encephalitis, metabolic upsets (hypocalcaemia, hyponatraemia, hypoglycaemia), inborn errors of metabolism (such as glycine encephalopathy, hyperammonaemia and mitochondriopathies) and poisoning (including bacterial toxins from shigella and campylobacter).

Dysrhythmias do not usually recur once the acute encephalopathy settles. We should regard febrile convulsions in the young child as an acute encephalopathy which occurs in response to an acute viral infection in a genetically predisposed child at a physiologically vulnerable age. It is unusual for a second bouts of fits to occur from a second acute encephalopathy unless from a metabolic cause such as recurrent hypoglycaemia, recurrent febrile convulsions.

Acute cerebral dysrhythmias can be totally intractable, drug resistant or fatal as after the inadvertent injection of an overdose of intrathecal penicillin (Fig. 1).

Chronic recurrent brain dysrhythmias is what we mean by the term epilepsy. By recurrent fits we mean recurrent cerebral dysrhythmias which do not depend upon a recurrent pathology such as electrolyte or glucose disorder. However there may be a recurrent precipitant such as smell, sound, flashing lights, television, video display unit, hyperventilation or strong emotion.

Recurrent dysrhythmias may occur as a single ictus at intervals of varying frequency, episodes of "grouping" of multiple fits or an intractable epilepsy with fits everyday of the child's life.

Continuous cerebral dysrhythmia Status epilepticus as usually defined, is when there are repeated tonic-clonic fits without recovery of consciousness in between. However, any type of seizure disorder can become continuous and the concept of status epilepticus has in the past been too narrowly restricted to convulsive status. Non-convulsive status lasting hours or days complicates many of the syndromes responsible for intractable epilepsy in childhood. Cognitive status epilepticus may occur in West syndrome with severe autism in Landau-Kleffner syndrome with aphasia and the Lennox Gastaut syndrome (Table I).

A seizure describes a paroxysmal alteration in behaviour due to any transient brain pathology. It therefore includes cerebral dysrhythmias and transient ischaemic or anoxic attacks, the latter being referred as "faints". The commonest examples of faints are reflex anoxic seizures (cyanotic breath-holding attacks) and reflex asystole (pallid syncopal attack). They should not be treated as epilepsy despite their recurrent nature.

A convulsion is the motor manifestation of generalized tonic, clonic or tonic-clonic seizure. It may be due to cerebral dysrhythmia, transient ischaemic attack, raised intracranial pressure or toxic state leading to the release of decerebration. Table II shows the various causes of paroxysmal extensor hypertonus.

What Constitute A Fit ?

1. The early prodrome is usually indiscernible to a casual observer but often becomes the reliable herald to an impending fit for the patient or his parents. It is a behavioural or mood change which may last for hours, with no abnormality on electroencephalography (EEG) detectable at this stage. This may, however, be important since any lesion, be it a genetic predisposition or brain tumour, is present 24 hours a day and this prodromal period represents the change in brain biochemistry which allows a fit to occur at a particular time on a particular day.

2. When the brain starts to generate a localized electrical disturbance, an aura occurs. It is usually brief and represents a circumscribed partial fit. Its presence indicates partial onset epilepsy. However many patients cannot appreciate it due to its brevity. It is a common observation that patients with complex partial seizures do not report auras. Therefore the absence of aura does not imply generalized onset fit. This localized disturbance may interfere with cognition which is not recognized by the child but affects learning as a transient cognitive impairment.

3. The phase of ictus then follows, which represents the fit itself. Its character depends on the spread of electrical disturbance. Most frequently it is an irritative event with positive symptoms like muscular twitching or hallucinations. But it can also be a paralytic event with negative symptoms such a loss of muscle tone or a release of inhibition over the primitive reflexes. For instance, infantile spasms are startle reflex and the adversive seizures are asymmetric tonic neck reflex released from cortical inhibition. If the discharge is localized and slow at 1-3 cycles per second (see Fig. 1) then a following response occurs. When the motor system is involved, it becomes a simple partial seizures, and when the reticular formation is involved bilaterally then consciousness is lost. Fast spikes of 10 cycles per second or more (Fig. 2) "tetanize" the brain causing loss of cortical inhibition over brain stem. When the midbrain centres are released it can manifest as doggy paddling, leg cycling or decerebrate rigidity (Fig 3). Pulsed three per second discharges generalized to all parts of the brain and with a synchronous spike and wave in all areas is thought to arise from oscillation between the cerebral cortex and the thalamus as a result of genetic abnormality in the T type calcium ionophores in the thalamus. This blocks incoming perception with loss of awareness but the motor cortical areas are spared so that there is no loss of posture or jerking.

There is disagreement as to the importance of the cerebellum in the genesis of fits such as atonic fits where there is a sudden loss of muscle tone leading to violent falls. However cerebellar dysfunction in the form of ataxia is a commonly seen feature of Lennox-Gastaut syndrome. Cerebellar pacing was proposed in the past as a means of controlling severe epilepsy. Experimental unilateral cerebellar ablation increases the epileptogenicity of the corresponding cerebral hemisphere. All these observations suggest the importance of cerebellum in the intrinsic antiepileptic mechanism of the central nervous system.

4. The postictal period is often marked by confusion and drowsiness. As a rule of thumb, the patient should have no memory of the ictal event except for simple partial seizures.

How Should Epilepsy be Diagnosed ?

Epilepsy is a clinical diagnosis. No investigation can take the place of careful history taking and physical examination. Rarely will a physician witness a fit, and many children are unable to describe their experiences. Clinicians will have to rely heavily on the account of the witness, which might add to the diagnostic difficulty. It also makes paediatricians particularly vulnerable if the parents deliberately falsify statements as in Munchausen syndrome by proxy.

Many paroxysmal conditions simulate epileptic fits. (Tables III and IV) They are described by Stephenson as faints and funny turns.² Choreoathetosis, dystonias (e.g. due to gastroesophageal reflux in Sandifer syndrome), tics, breath-holding spells, behavioural stereotypies in mentally retarded or cerebral palsied children, sleep phenomena and pseudoseizures are notable examples that community paediatricians may come across. In the intensive care setting, decerebrate rigidity due to raised intracranial pressure (as in head injury, meningitis and cerebral oedema) is often a cause of confusion. Dystonia from drugs like fentanyl or an overdosage of phenytoin may cause unnecessary administration of further anticonvulsant medication. Their recognition depends on high index of suspicion. Doubt should always be raised if the phenomenon is resistant to antiepileptic drugs and not accompanied by coincident abnormalities on EEG.

The clinical manifestation of a cardiac arrhythmia, i.e. palpitation, dyspnoea or syncope, are not specific to the type of cardiac arrhythmia and neither is the type of fit specific to a particular cerebral dysrhythmia. Since no cardiologist would diagnose a cardiac arrhythmia without an electrocardiography, the neurologist or paediatrician should use the EEG in the same manner. Unequivocal diagnosis of epilepsy can only be made if an ictal record shows definite simultaneous changes. Although it is extremely unlikely for that to be available to clinicians in many units, it should be the aim for good therapy.

Abnormal discharges in the interictal record substantiates the diagnosis, but does not make the diagnosis of epilepsy without clinical fits. Three percent of normal population may have abnormal EEG and some children with febrile convulsions may show 3 cycles per second spike and wave discharges at the age of five years and yet never develop epilepsy.³ Conversely, a negative EEG does not exclude the diagnosis of epilepsy because the routine wakeful record can be normal in up to 50 to 60% of epileptic children.⁴ The detection rate of abnormal discharges can be increased by various provocation methods like hyperventilation, photic stimulation, sleep, sleep deprivation or chlorpromazine. Repeated and prolonged recording may be required in some cases. If non-epileptic disorders cannot be confidently ruled out, polygraphic monitoring and video telemetry should be employed.

Electroencephalography nonetheless is indispensable in order to classify epileptic syndromes by virtue of the detection of characteristic changes. Its use in monitoring the treatment of epilepsy is more controversial; there is no clear relationship between interictal spike frequency and seizure frequency. Patients with benign partial epilepsy with centrotemporal spikes ("benign rolandic seizures") can have multiple spikes on EEG and yet maintain a very low seizure frequency.

Classification of Epilepsy in Childhood

A good classification system should serve two purposes. Firstly it~ should be a framework to clinicians for guiding diagnosis, investigation, treatment and prognostication. Secondly it should serve as a common language to researchers worldwide for meaningful comparison in all aspects of investigations of these diseases.

In the past, we classified epilepsy according to the seizure type as if each represented a disease entity, such a grand mal, petit mal, myoclonic epilepsy and jacksonian epilepsy. The International Classification of Epilepsy and Seizure Disorders devised by the International League Against Epilepsy is the most widely used.⁵ However, to classify epilepsy by the ictal manifestation was soon found to be inadequate, especially in children where one-third were not classifiable according to this system. Children often have more than one type of fit (polymorphic epilepsy) and cannot be conveniently classified by seizure type. Equally confounding is the fact that the seizure type changes with age and brain maturation.

On the other hand, the same type of fit can occur in different conditions which have vastly different prognosis. Using absence epilepsy as an example, it can occur in (a) childhood pyknoleptic absence epilepsy ("true petit mal") with good response to treatment and an excellent outcome, (b) minor complex partial seizure with a worse outcome, associated behavioural and learning difficulty and a different therapy, (c) Lennox-Gastaut syndrome or severe myoclonic epilepsy of infancy with drug resistance and a horrendous outcome, (d) juvenile myoclonic epilepsy as the presenting feature which requires life-long anticonvulsant, (e) subacute sclerosing panencephalitis, which is a fatal disease.

A new classification of epilepsies, epileptic syndromes and other related seizure disorders emerged in 1989.⁶ The new classification is not without drawbacks. Classifying a syndrome under the localization-dependent category implies an anatomical pathology. Nonetheless, this is not always true, such as the benign partial epilepsies with centrotemporal spikes or occipital paroxysms are not associated with abnormalities on brain imaging. Similarly, generalized epilepsy like infantile spasms may have focal lesions identifiable on imaging, positron emission tomography (PET) or EEG.^4,7 As yet no perfect classification system exists, but there is general agreement that the syndromic approach is most applicable in children.

What are Epileptic Syndromes ?

Epileptic syndromes are neurological conditions that embody a non-random combination of nosology, genetics, fit type and neurophysiological abnormalities with the predominant feature being epilepsy. An epileptic syndrome usually includes the following nosological features:

These can be varied and polymorphic. Tonic-clonic, myoclonic, atonic, absences, complex or simple partial seizures are the best known ones. Other rare types include sensory, cognitive, behavioural or odd motor behaviours such as laughing (gelastic seizures), running round in circles (cursive epilepsy) or "happy clappers" (frontal lobe epilepsy). The type of fit is often the main feature by which the syndrome is classified.

2. Ictal and interictal electroencephalographic findings. Some syndromes have characteristic EEG changes, like the 3 per second spike-and-wave discharges in childhood absence epilepsy, hypsarrhythmia in infantile spasms and slow spike-and-wave complexes in Lennox-Gastaut syndrome. However, these features are not exclusive to one particular syndrome, just as hypsarrhythmic EEG can also occur in Lennox-Gastaut syndrome.

Many childhood epilepsy syndromes manifest themselves within a remarkably narrow period of time. This essentially reflects the effect of brain development on cerebral dysrhythmias.

Correctly diagnosing an epileptic syndrome will allow prediction of the chance of remission and the likely intellectual outcome of the child.

Some epileptic syndromes are genetic in origin, but the penetrance is often incomplete in autosomal dominant syndromes. Family history of epilepsy and mental retardation will also prompt the investigations for possible underlying metabolic disorders.

Many epileptic syndromes are accompanied by other clinical features that may dominate the picture, such as acquired aphasia in Landau-Kleffner syndrome.

Some form of brain imaging is an investigation most clinicians will perform for childhood epilepsy as their prime concern is missing a treatable intracranial lesion. In one study 46% of epileptic children were found to have some form of abnormalities on brain imaging, although most were not amenable to surgery.⁸ PET scan and single photon emission computed tomography (SPECT) scan will increase the sensitivity of localizing lesions.

These not only help diagnosis, but also allow reduction of seizure frequency by avoidance of them. For example photosensitive seizures can be reduced by avoiding flickering light, sitting away from television in a well lit room and avoiding computer games with a known provocative flash frequency. Some precipitants are less well-known, like certain patterns in window, radiator grills or clothing, prolonged reading, startle, bathing, water immersion, touch, tapping on head, brushing hair, concentrated thought and body movements.

Correctly diagnosing an epileptic syndrome allows a wise choice of antiepileptic drugs. Some epileptic syndromes do not respond to certain medication, just as febrile convulsions treated with carbamazepine or phenytoin, while others may in fact be made worse, such as juvenile myoclonic epilepsy being exacerbated by carbamazepine.

In daily clinical practice, it is advisable not to affix a syndromic label if the characteristics of the patients does not fit into widely accepted criteria. Each patient is unique and squeezing them into a diagnosis for convenience will lead to therapeutic and prognostic inaccuracies.

Three Types of Epileptic Syndromes

It is difficult to remember the large number of epilepsy syndromes in routine clinical practice unless they are further classified in some way. We propose a classification system for the epileptic syndromes in childhood based on aetiology. It is in no way superior to the system in use, but it provides a pragmatic perspective in daily patient management.

Childhood epileptic syndromes can be classified into lesional, genetic and malignant epilepsies. Each group has its characteristic profile of aetiology and management. The clinical manifestations, however, overlap.

(A) Lesional Epileptic Syndromes of Childhood (Table V)

Epilepsy can be caused by virtually any lesion in the cerebral cortex. Lesions in the thalamus, basal ganglia, brain stem and cerebellum do not cause epilepsy as shown by the distribution of brain tumours which presented with seizures (Table VI). In the past, it is fashionable to talk about diencephalic epilepsy and although the thalamus is involved in the generation of the genetic 3 per second spike-and-wave discharges, thalamic lesions do not cause epilepsy. Gillingham in Edinburgh made over 1 000 stereotactic lesions in the thalamus and only one patient had fit (personal communication).

The epileptic manifestation of lesional epilepsies depend on the locus of electrical disturbance. If it is circumscribed, the seizure will be a simple partial one without disturbance of consciousness. However, when the electrical activity spreads to other parts of the central nervous system, tonic-clonic, absence, complex partial or myoclonic seizures may occur. In other words, generalized seizure does not preclude focal lesion.

If focal or multifocal abnormal electrical activity is found, it might be expected to be a prima facie case for lesional epilepsy. Unfortunately, metabolic diseases such as hypocalcaemia, hypomagnesaemia and hypoglycaemia or even a simple febrile convulsion may present with focal seizure. Lesional epilepsy should therefore not be diagnosed without brain imaging, be it MRI, SPECT or PET, but a combination of methods is usually required.

The commonest lesional epilepsy in childhood is that associated with hemiplegic and quadriplegic cerebral palsy. Pure dyskinetic (e.g. post-kernicteric) and ataxic cerebral palsy are only exceptionally associated with epilepsy. ft is also uncommon in the pure diplegic cerebral palsy of prematurity.

Cerebral palsy, by definition, is always associated with brain damage and so the epilepsy is always secondary to a lesion. The co-existence of cerebral palsy reflects more extensive brain damage and thus greater neurological and cognitive disability and a poorer prognosis. About one-third of hemiplegics and 90% of quadriplegics have fits.

In a recent review of our department of 87 children with epilepsy and cerebral palsy, all had onset before 5 years old. Sixty per cent had polymorphic fits and the seizure type varied with age. Polymorphic seizure is more common in patients with mixed forms of cerebral palsy. Only 25% of patients had a single seizure type. In the small group of children with hemiplegia complex partial epilepsy was the commonest type. Polypharmacy was a common problem with one half of cases on more than two drugs. One-third of patient had daily fits despite treatment. Poor outcome was associated with occurrence of status epilepticus and high seizure frequency. Status epilepticus was common, occurring in 40% of cases, and in half of these subjects, status epilepticus was recurrent (unpublished data, Kwong and Livingston).

Neocortex develops during intrauterine life by neuronal proliferation at 8 to 16 weeks followed by migration at 12 to 24 weeks. Any interference to these processes will lead to disorganized brain growth with neurons in white matter, subpial neuronal collections, inverted neurons and vertical rather than horizontal laminar structure of the cortex, and hence the condition of cortical dysplasia. They may be macroscopic as in lissencephaly, microcephaly, pachygyria, polymicrogyria, heterotopias and hemimegalencephaly (fig. 4). Their recognition in clinical practice has risen considerably with the advent of MRI scanning. Microscopic dysplasia (microdysgenesis) is also being increasing found to be the cause of epilepsy, but they require histological examination and may escape detection even by the most sensitive neuroimaging technique.

Many neurocutaneous syndromes are associated with cortical dysplasia. Neurofibromatosis, tuberous sclerosis, Sturge-Weber syndrome and linear sebaceous naevus syndrome are the best known ones. Other examples include Zellweger syndrome, Miller-Dieker syndrome, Aicardi syndrome, Angelman syndrome and bilateral opercular dysplasia syndrome. Maternal abuse of alcohol or cocaine is also a known cause.

Cortical dysplasia causes partial fits with or without generalization. They are also causes of many malignant epileptic syndromes. Apart from causing electrical instability, they will also disrupt the function of the affected parts of the brain. Mental retardation, speech difficulty and cerebral palsy are common accompaniments. The epileptic fits are usually resistant to antiepileptic drugs.

The medial temporal lobe is the junction of emotion, memory, consciousness and autonomic function. It connects the cerebral cortex, limbic system and the reticular formation of the brain stem. A lesion in this area can produce very complex symptomatology and affected children will also show behavioural and learning difficulties in addition to their epilepsy. The area may be the site of microdysgenesis or hamartoma. There may be collection of Langhans giant cells suggesting forme fruste of tuberous sclerosis. Small tumours may remain dormant for years or even decade. However, the commonest pathology is that of mesial temporal sclerosis (Table VI).

Mesial temporal sclerosis has long been regarded as the most important cause of refractory complex partial epilepsy in adulthood. It is a disruption of the normal architecture of the hippocampal formation associated with neuronal loss and gliosis. Its strategic location in the hippocampus accounts for the limbic system involvement during seizures and their distinctive affective overtone, with hallmarks of sudden emotional changes, distorted perception of self and environment and memory disturbance like amnesia and deja vu. A vague abdominal sensation commonly heralds alteration in consciousness and automatism. Abdominal epilepsy, which represents the phase of aura in these seizures, seldom occur without subsequent loss of consciousness. This is a useful distinguishing feature from recurrent functional abdominal pain.

The medial temporal structures, like the basal ganglia and cerebellum, are very sensitive to hypoxia. They possess a high density of glutamate receptors and the possibility of excitotoxic damage is greatest in this part of the brain. Characteristic pathological findings are hypoxic-ischaemia lesions of varying ages and therefore some of the investigators suggested that mesial temporal sclerosis is a progressive lesion. Recent MRI study have confirmed the early pathological findings.⁹ Up to 80% of adults with histologically proven mesial temporal sclerosis had convulsion lasting for more than 30 minutes under the age of three.^10,11

Antenatal and perinatal hypoxic and ischaemia insults are also possible causes,¹² although a different part of the hippocampus is affected.

Cerebral oedema resulting in tentorial herniation causes ischaemia of the medial temporal structures.¹³ This is a further possible mechanism of mesial temporal sclerosis.

Discharges from the deep structures in temporal lobe cannot always be readily picked up by conventional surface electrodes. Deep electrodes like sphenoidal, foraminal, subdurally implanted electrodes and intraoperative electrocorticography are sometimes employed to localize the focus. The abnormality seen on surface EEG may not be the site of origin of the seizure and so an actual ictal EEG, provoked by stopping all anticonvulsant medication, if necessary, may be justified. Structurally, mesial temporal sclerosis can be reliably diagnosed by MRI in coronal sections. New methods like volumetric analysis and high resolution MRI can increase the sensitivity and reliability.¹⁴ SPECT scan with isotope injection given during the fit may reveal the focus.

There are two main messages for paediatricians: firstly, prolonged convulsion in small children should be avoided, and secondly, that mesial temporal sclerosis is the lesion par excellence which responds to surgery with an 85 percent cure rate.

Unlike adults, tumours are uncommon causes of epilepsy in children. Brain tumours like astrocytoma and oligodendrocytoma, however, have been reported. Hamartoma and vascular malformation are more common causes.

Lesional epilepsy responds favourably to antiepileptic drugs which act on voltage-gated sodium ionophores. These include carbamazepine, phenytoin and lamotrigine, which are the drugs of first choice. Sodium valproate and vigabatrin are also commonly used. New drugs like gabapentin and oxycarbazepine are reported to be effective.

Epilepsy surgery is most frequently performed in this group of patients. In spite of the tremendous advance in surgical technique and preoperative planning, the decision for surgery is never taken lightly. Intractability with significant impairment of quality of life is almost always a prerequisite. Nonetheless it should be considered early in some conditions that are known to be refractory to conventional treatment. Epilepsy due to hemimegalencephaly, Sturge-Weber syndrome or Rasmussen's encephalitis is best treated by timely hemispherectomy.

Careful selection of patient for surgery is the first step to ensure good results. Good candidates are those with congruent localization on pre-operative clinical, neurological, psychological, neurophysiological and imaging assessments. If localization is not conclusive, intraoperative electrocorticography can be employed. It is important to preserve the areas responsible for speech and memory. Intracarotid infusion of amylobarbital (Wada test) and complicated intraoperative psychometric assessment are often necessary to identify the dominant hemisphere. During the surgery electrical cortical stimulation and psychometric tests are done to map out the areas for resection while preserving the speech and memory areas. Since psychometric tests are difficult to interpret before the age of eight, epilepsy surgery has practical difficulty for the very young patients. Anterior temporal lobectomy is the most common technique. It removes the hippocampus and part of amygdala while the language and memory areas are spared. Amygdalohippocampectomy is a more restricted procedure that serves the same purpose. Other types of lesionectomy and tailored cortical resection are also possible. Postoperative rehabilitation is necessary in most patients.¹⁵

(B) Genetic Epileptic Syndromes of Childhood

We define these as inheritable epileptic syndromes not associated with structural lesions. For instance, epilepsy due to tuberous sclerosis with intracranial hamartoma, although is inherited as an autosomal dominant condition, should be classified as lesional epilepsy instead of genetic epilepsy.

There are two types of genetic epileptic syndromes. The first group are those with known metabolic derangements, and the second are those without. The dividing line between the two groups is arbitrary. It is only based on our current understanding of these disorders.

We would not include causes of hypoglycaemia, hypocalcaemia or hypomagnesaemia in this group. These are acute dysrhythmias which complicate acute encephalopathy and it is the encephalopathy which is recurrent and not simply the fits. Conditions such as mitochondrial encephalopathy with lactic acidosis and stroke like episodes syndrome (MELAS syndrome) and myoclonic epilepsy with ragged red fibres (MERRF) are included. These disorders have either known enzyme defect or evidence of storage abnormality morphologically but the enzyme defect is still unknown. They are almost invariably associated with other neurological abnormality, particularly intellectual deterioration. Thus, the presence of family history with a symptomatic combination of fits and mental retardation should trigger a search for inborn errors of metabolism as this has implications of treatment and genetic counselling.

Instead of deficiency of enzymes, these syndromes are probably caused by faulty neurotransmitter receptors and transmembranous ion channels. Recent advances in human genome mapping, gene cloning and molecular genetics have spurred exciting development in their understanding. Table VIII listed the various epileptic syndromes in this category.

This is the first epileptic syndrome to be linked to a fault of the neurotransmitter mechanism. The gene was mapped to chromosome 20q13.2-q13.3, which was found to be the region where the a4 subunit of the neuronal nicotinic acetylcholine receptor was encoded.^16,17 It is transmitted in an autosomal dominant manner. However there are reports of autosomal recessive inheritance in a closely consanguineous family,¹⁸ suggesting heterogeneous nature of the syndrome. Nonetheless for practical purposes the diagnosis always requires a positive family history in parents.

These fits start during the first two weeks of life and resolve by three months. They are resistant to phenobarbitone but do not seem to leave any neurological sequelae. Although 14% of patients were reported to develop epilepsy in later life,¹⁹ there is no explanation why a permanent genetic defect should cause a transient seizure disorder in the majority.

Epileptic syndromes related to chromosome 6p (Genetic Generalized Epilepsy)

By discovering a linkage between juvenile myoclonic epilepsy and HLA markers, the gene responsible was successfully mapped to human chromosome 6p.²⁰ The locus was subsequently designated EJM1. Then by virtue of the close association of juvenile myoclonic epilepsy and other generalized epilepsy disorders like childhood pyknoleptic absence epilepsy (typical absence epilepsy), grand mal seizures on early morning awakening, photoconvulsive television or video game epilepsy together with the high incidence of EEG abnormality in the close relatives, investigators concluded that the gene imparts a proclivity for abnormal electrical activity to synchronize and generalize in histologically normal cerebral cortices. The various epileptic syndromes are actually different ways of the proclivity expresses itself.^21,22 Juvenile myoclonic epilepsy can be considered the full-blown prototype which has a mixture of all three types of absence, myoclonic and generalized tonic-clonic fits, while the rest are milder variants or forme fruste. The expression of the gene is highly age-dependent, just as childhood absence epilepsy tends to resolve after adolescence. Family screening cannot be reliably done by EEG alone as it may be absent in older individuals. Linkage or DNA analysis might be necessary.

A recent report of linkage study showed no evidence for a locus on chromosome 6p in some British and Swedish families with juvenile myoclonic epilepsy and primary grand mal seizures.²³ It is therefore reasonable to conclude that there are more than one gene carrying the proclivity for generalized fits.

It is interesting to note that valproic acid is an effective antiepileptic drug for these syndromes. Valproic acid is different from other drugs in that it is a fatty acid with 8 carbon molecules. Eight-carbon fatty acids are anticonvulsants and ten-carbon ones anaesthetic agents. Octanoic acid was thought to be the fatty acid with antiepileptic properties in the ketogenic diet. The fact that it appears specific for many genetic epilepsies suggests that it may be correcting an underlying metabolic abnormality. Hopefully, gene cloning will provide insight to the mechanism of the fits as well as he antiepileptic mechanism of valproic acid.

Apart from childhood pyknoleptic absence epilepsy the rest are all less amenable to antiepileptic treatment. Adolescence onset absence epilepsy may persist into adulthood. The treatment of choice, however, remains valproic acid. Carbamazepine is contraindicated because it can exacerbate these fits.

These include the benign partial epilepsy with centrotemporal spikes ("rolandic seizures"), benign partial epilepsy with occipital paroxysms and benign psychomotor epilepsy. The underlying genetic mechanism of these syndromes is unknown. They are amongst the commonest forms of epilepsy in children between the age of 3 - 15. They are all characterized by the absence of neurological and intellectual deficit. Seizure frequency is usually low and they are very sensitive to carbamazepine treatment. Carbamazepine, however, may precipitate continuous spike discharge in slow wave sleep (electrical status in sleep) in some patients. The prognosis is good, particularly for those with centrotemporal spikes where all cases resolve. But if the abnormal EEG discharge originates from areas other than this, the prognosis is slightly less favourable.^24-26 Recently families of benign rolandic seizures with articulatory dyspraxia have been described. Speech arrest may occur in benign rolandic seizures and suggestion has been made that some cases of Landau-Kleffner syndrome who also developed electrical status in sleep may represent a not so benign variant.²⁷

This distinctive group consists of Ohtahara syndrome, infantile spasms, Lennox-Gastaut syndrome, Landau-Kleffner syndrome and polymorphic myoclonic epilepsy of Dravet (severe myoclonic epilepsy of infancy).

All these syndromes are characteristically resistant to all antiepileptic treatment with high seizure frequency. Cognitive impairment is common and there may be a progressive dementia suggesting an underlying degenerative disease. There is a high propensity to both convulsive and non-convulsive status epilepticus. While the former will not be missed by parents and clinicians, the latter is often subtle and not recognized. Non-convulsive status epilepticus may take the form of slight retardation of responsiveness or inappropriately altered affect. Motor activity is limited to sporadic myoclonic jerks or eyelid flutter, or is absent altogether. Ataxia and uncontrollable drooling are frequently observed signs. Autonomic changes like pupillary dilatation and borborgymi occasionally happen. Non-convulsive status can last for months without recognition and the external changes usually become less obvious with time. It is believed that the intellectual deterioration of these malignant epileptic syndromes is caused by prolonged spells of non-convulsive status epilepticus. The brain is "switched off' with the gross disturbance of electrical activity. Functional studies show hypometabolism of the cerebral cortex and therefore cannot acquire new experience.

All these syndromes have multifaceted aetiology: metabolic disorders, congenital and acquired lesions have been reported. There remains a group of about 30 to 40% of patients where the aetiology is not known. They are known as the cryptogenic group, where the presence of underlying cause is assumed. Histological section of resected brain often shows microdysgenesis.

Tonic spasms, quite similar to those of infantile spasms, occur in the first days or weeks of life. EEG shows a persistently discontinuous pattern irrespective of sleep or arousal (Fig. 5), with multifocal spikes interspersed haphazardly. For most patients it heralds infantile spasms and the prognosis is poor.

Infantile spasms occur between 3 - 12 months of age. There are massive forceful flexions of limbs and trunk that occur in cluster, which are followed by crying and distress of the infant. During a spasm, the EEG shows electrodecremental phenomenon while the interictal EEG shows hypsarrhythmia. Extensor and tonic spasms can also occur. Unilaterality and extensor spasms often suggest symptomatic causes.

All drugs that act on the gamma-aminobutyric acid (GABA) A receptor are effective in infantile spasms. This receptor carries receptor sites for adrenocorticotrophic hormone (ACTH) and this explains why ACTH is an effective treatment. In fact for most patients ACTH and steroids remain the treatment of choice. We would use ACTH to prevent subsequent dementia if the child has "switched off' with reduced responsiveness, autistic features and a continuously hypsarrhythmic EEG suggestive of non-convulsive status epilepticus. Tuberous sclerosis is an important symptomatic cause. Its identification is important as vigabatrin is nowadays often recommended as drug of first choice. Valproic acid, nitrazepam and lamotrigine are possible, but less effective alternatives.

Surgical treatment is possible in some patients where hypometabolic lesions are identified on PET scan.⁷

At least 80% of patients suffer from cognitive difficulty after infantile spasms. One-third will evolve into Lennox-Gastaut syndrome or complex partial epilepsy. Symptomatic cases with early onset and polymorphic seizures tend to perform poorly.

This syndrome has a triad of "stare (atypical absences with minor automatism), jerks (myoclonic jerks) and fall (atonic seizures)". Axial tonic spasms also occur in sleep. EEG shows slow irregular spike and wave complexes at one to 2.5 Hz interictally and fast spikes of 10Hz during sleep. Both electrodecremental phenomenon and fast spikes of 15 - 25 Hz can occur during fits.

Non-convulsive status is common in Lennox-Gastaut syndrome. Instead of losing skills, these patients only fail to acquire new skills. Attention deficit, autism and personality disorders are commonly seen, which often dominate the picture when seizures become less frequent with age.

Polypharmacy is a common problem in these patients. Palliative corpus callosotomy is useful for the control of atonic seizures and as the child becomes older it makes the life of carer a little easier. Steroid, immunoglobulins and other less conventional treatments are used, but none was shown to have altered the dismal intellectual prognosis.

Malignant Polymorphic Myoclonic Epilepsy of Dravet (Severe Myoclonic Epilepsy of Infancy)

The syndrome often declares itself by atypical recurrent febrile convulsions at about 4 months old. Myoclonic seizures gradually evolve together with atypical absences, simple and complex partial seizures. Spike and polyspike discharges are only detectable after the first year of life. One half of cases also show photosensitivity.

Non-convulsive status epilepticus occurs commonly with slow dementia, escalating ataxia and chorea. Autonomic derangements during these status epilepticus like cardiac arrhythmia and hypotension may account for the 15.9% mortality.³² Prognosis is unfavorable and no treatment is effective in changing the clinical evolution.

This syndrome is characterized by onset aphasia and receptive verbal disorder. EEG shows multiple spike or sharp wave discharges over both hemispheres and the extent of abnormality is not directly correlated to the severity of language difficulty. A subgroup shows continuous spikes during slow wave sleep (electrical status in sleep). The language disorder is thought to be due to an on-going non-convulsive status epilepticus that leads to a block of auditory speech decoding. Some patients however can be taught lip-reading.

Anecdotal reports incriminate cortical dysplasia in planum temporale, inflammatory, vasculitic or demyelinating process of unknown cause. The outcome is variable, and residual disability like frequent grammatical errors and word idiosyncrasy are commonly seen even after recovery. Although steroid and ACTH can be effective, it is not entirely clear that they alter the prognosis. Because of the severe disruption of all language skills, speech, reading, writing and spelling as well as sometimes the onset of autistic behaviour, we would give a trial of ACTH in all cases. Some show sustained improvement but require continuation of a small dose of ACTH for a year or more.

No anticonvulsant combination was found to be superior in controlling seizures of these syndromes. More importantly no treatment to date was shown to alter the intellectual decline. Therefore a balance of treatment benefit and side effects should always be contemplated. Unduly aggressive treatment at the expense of the consciousness of the child would only increase difficulty of care and danger of fatal chest infection.

Benzodiazepines, which are frequently used in these situations, have the disadvantage of habituation, which can occur in as soon as 8 weeks. Drug holidays can then be given. It should be noted that benzodiazepines can precipitate non-convulsive status epilepticus in Lennox-Gastaut syndrome.

Immunomodulating treatment are increasingly employed. ACTH and steroid are long proven to be useful in infantile spasms, and there are reports attesting their efficacy in Lennox-Gastaut syndrome and Landau-Kleffner syndrome.^38,39 Callostomy and multiple subpial resections are the most widely used surgical technique in their treatment. They aim at reducing the spread of epileptic discharge and hence reduce the amplitude of seizures.

Conclusion

Without accurate diagnosis and classification, rational decisions in treating epilepsy will not be possible. It is important not to lose sight of the associated neurological and intellectual abnormalities associated as they often have a more significant impact on the child's life.

Clinicians should balance between the therapeutic benefit and undesirable adverse effects of all antiepileptic treatment. Failure to abolish all seizures should not at the present time be regarded as therapeutic failure in patients with certain epileptic syndromes. Increasing polypharmacy at all costs should be avoided. In the future, more detailed study of the cognitive impairment may direct us at better electrical control of the dysrhythmia rather than to manage the motor manifestation of fits alone.

Hopefully, the study of genetic epileptic disorders will give us better understanding of the natural antiepileptic mechanisms and the biochemical basis of electrocerebral disturbance. This would give us antiepileptic drugs with more refined action and less side effects.

Acknowledgements

The authors would like to thank the George Guthrie Research Fellowship, Faculty of Medicine, University of Edinburgh for supporting Dr. Mary O'Regan in publishing this paper.

References

3. Lennox-Buchthal MA. Febrile convulsions: a reappraisal. Electroencephalography and Clinical Neurolphysiology 1973; suppl 32: S80-S89. Amsterdam. Elsevier.

5. Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epilepsia 1981;22:489-501.

6. Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 1989;30:389-99.

7. Chugani HT, Shields WD, Shewmon DA, Olson Din, Phelps ME, Peacock WJ. Infantile spasms: I PET identifies focal cortical dysgenesis in crytogenic cases for surgical treatment. Ann Neurol 1990;27:406-13.

8. Lagenstein I, Sternowski H, Rothe M, Bentele KHP, Kuhne G. Cranial computerized tomography in different epilepsies with grand mal and focal seizures in 309 children: relation to clinical and electroencephalographic data. Neuropediatrics 1980;11:323-38.

9. Harvey AS, Grattan-Smith JD, Desmond PM, Chow CW, Berkovic SF. Febrile seizures and hippocampal sclerosis: frequent and related findings in intractable temporal lobe epilepsy of childhood. Paediatr Neurol 1995;12(3):210-06.

10. Falconer MA, Serafetinides EA, Corsellis JAN. Aetiology and pathogenesis of temporal lobe epilepsy. Arch Neurol 1964; 19:353-61.

11. Ounsted C, Lindsay J, Richards P. Temporal lobe epilepsy, a biographical study 1948 - 1986. Clinics in Developmental Medicine No. 103. 1st ed. London. Mac Keith Press, 1987:109.

12. Jensen I. Temporal lobe epilepsy. Aetiological factors and surgical results. Acta Neurol Scand 1976;53: 103-118.

13. Brown JK. The pathological effects of raised intracranial pressure. In Minns RA, editor. Problems of intracranial pressure in childhood. 1st ed. London. Mac Keith Press, 1991:38-76.

14.Jackson GD. New techniques of MRI in epilepsy. Epilepsia 1994;35(suppl 6):S2-513.

15. Arroyo 8, Freeman JM. Epilepsy surgery in children: state of the art. Adv in Paediatr 1994;41:53-81.

16. Leppert M, Anderson VP, Quattelbaum T, et al. Benign familial neonatal convulsions linked to genetic markers on chromosome 20. Nature 1989;337:647-8.

17. Steinlein O, Smigrodzki R, Lindstrom J, et al. Refinement of the localization of the gene for neuronal nicotinic acetylcholine receptor a4 subunit (CHRNA4) to human chromosome 20q13.2-q13.3. Genomics 1994;22:493-5.

18. Schiffmann R, Shapira Y, Ryan 8G. An autosomal recessive form of benign familial neonatal seizures. Clin Genet 1991;40:467-70.

19. Plouin P. Benign neonatal convulsions. In: Wasterlain CG, Vert P, editors. Neonatal seizures. New York: Raven Press 1990:51-9.

20. Durner M, Sander T, Greenberg DA, Johnson K, Beck-Mannagetta G, Janz D. Localization of idiopathic generalized epilepsy on chromosome 6p in families of juvenile myoclonic epilepsy patients. Neurology 1991;41:1651-5.

21. Delgado-Escueta AV, Greenberg D, Weissbecker K et al. Gene mapping in the idiopathic generalized epilepsies: juvenile myoclonic epilepsy, childhood absence epilepsy with grand mal seizures and early childhood myoclonic epilepsy. Epilepsia 1990;31(suppl 13):S19-S29.

22. Greenberg DA, Durner M, Resor S, Rosenbaum D, Shinnar S. The genetics of idiopathic generalized epilepsies of adolescent onset: differences between juvenile myoclonic epilepsy and epilepsy with random grand mal and with awakening grand ml. Neurology 1995;45:942-6.

23. Whitehouse WP, Rees M, Curtis D, et al. Linkage analysis of idiopathic generalized epilepsy and marker loci on chromosome 6p in families of patients with juvenile myoclonic epilepsy: no evidence for an epilepsy locus in the HLA region. Am J Hum Genet l993;53:652-62.

24. Lerman P. Benign partial epilepsy with centro-temporal spikes. In: Roger J, Dravet C, Bureau M, Dreifuss FE, Wolf P, editors. Epileptic syndromes in infancy, childhood and adolescence. London: John Libbey 1985:150-8.

25. Gastaut H. Benign epilepsy of childhood with occipital paroxysms. In: Roger J, Dravet C, Bureau M, Dreifuss FE, Wolf P, editors. Epileptic syndromes in infancy, childhood and adolescence. London: John Libbey 1985:159-70.

26.Dalla Bernardina B, Sgro V. Fontana E, Coamaria V, Selva LL. Idiopathic partial epilepsies in children. In: Roger J, Bureau M, Dravet C, Dreifuss FE, Ferret A, Wolf P, editors. Epileptic syndromes in infancy, childhood and adolescence. London: John Libbey 1992:173-88.

27. Scheffer IE, Jones L, Pozzebon M, Howell RA, Saling MM, Berkovic SE. Autosomal dominant rolandic epilepsy and speech dyspraxia: a new syndrome with anticipation. Ann Neurol 1995;38:633-42.

28. Vigevano F, Fusco I, Di Capua M, Ricci S, Sebastinanelli R, Lucchini P. Benign infantile familial convulsions. Eur J Paediatr 1992;151:608-12.

29. Malafosse A, Beck C, Bellet H et al. Benign infantile familial convulsions are not an allelic form of the benign familial neonatal convulsion gene. Ann Neurol l994;35:479-82.

32. Dravet C, Bureau M, Roger J. Severe myoclonic epilepsy in infants. In: Roger J, Bureau M, Dravet C, Dreifuss FE, Perret A, Wolf P, editors. Epileptic syndromes in infancy, childhood and adolescence. 2nd ed. London, John Libbey, 1992:67-74.

33. Doose H. Myoclonic astatic epilepsy of early childhood. In: Roger J, Bureau M, Dravet C, Dreifuss FE, Perret A, Wolf P, editors. Epileptic syndromes in infancy, childhood and adolescence. 2nd ed. London, John Libbey, 1992:103-14.

34. Holmes GL. Benign focal epilepsies of childhood. Epilepsia 1993;34(suppl 3):549-561.

36. Lehesjoki AE, Koskiniemi M, Sistonen P. Localization of a gene for progressive myoclonus epilepsy to chromosome 21q22. Proc Nat Acad Sci USA 1991;88:3696-9.

37. Scheffer IE, Bhatia KP, Lopes-Cendes I, et al. Autosomal dominant frontal epilepsy misdiagnosed as sleep disorder. Lancet 1994;343:515-7.

38. Lerman F, Lerman-Satgie T, Kivity S. Effect of early corticosteroid therapy for Landau-Kleffner syndrome. Dev Med Child Neurol 1991;33:257-60.

39. Yamatogi Y, Ohtsuka Y, Ishida T, et al. Treatment of Lennox-Gastaut syndrome with ACTH: a clinical and electroencephalographic study. Brain Dev 1979;1:267-76.