31 Seizures, epilepsy, and other episodic disorders in adults

Epilepsy, or more correctly a seizure, is most easily defined in physiological terms, being ‘the name for occasional sudden, excessive, rapid, and local discharges of grey matter’ (Jackson 1873). It is more difficult to offer a comprehensive clinical definition of epileptic seizures and epilepsy because of the varied clinical manifestations produced by cerebral neuronal discharge. However, an epileptic seizure can be defined as an intermittent and stereotyped disturbance of consciousness, behaviour, emotion, motor function, or sensation that on clinical grounds is believed to result from cortical neuronal discharge. Epilepsy can then be defined as a condition in which seizures recur, usually spontaneously.

The differential diagnosis of epilepsy is large because of the enormous range of symptoms that can occur during seizures. Inevitably, the differential diagnosis for tonic-clonic seizures is very different from that for simple partial seizures with autonomic symptoms. The most common clinical problem is the differential diagnosis from other causes of transient loss of consciousness associated with collapse, the commonest cause of which is syncope.

Syncope defined as a sudden loss of consciousness associated with the inability to maintain postural tone, followed by spontaneous recovery. It is common and results from a wide variety of causes (Table 31.1). It occurs across the age range, with an incidence of 4.7/1000 in women in the third decade of life rising to 11.1/1000 in the eighth decade and has a varied prognosis. In the elderly it is associated with significant morbidity and mortality (Soteriades et al. 2002).

The commonest cause of syncope is vaso-vagal in young people who have no serious underlying pathology. It is precipitated by unpleasant sights or pain, standing for prolonged periods, or after exposure to heat, hunger, dehydration, and alcohol excess. It is posture dependent: symptoms can start while sitting, but loss of consciousness usually occurs when the individual stands. The mechanisms leading to fainting are an abnormal autonomic response consisting of vasodilation and increased vagal tone. Pre-syncopal symptoms appear before hypotension and bradycardia and persist after the blood pressure returns to normal. The subject usually has a feeling of warmth with a dry mouth and a desire for fresh air or a drink of water. Nausea can develop quickly along with deep, sighing respiration, blurring of vision with spots in front of the eyes and loss of colour vision, noises in the ears, vertigo, and depersonalization. It should be noted that the majority of these symptoms do not occur in seizures.

The onset of these symptoms is usually gradual and eye witnesses comment on pallor and sweating. The subject will collapse if they remain standing. The collapse may be rigid or flaccid and some form of clonic or other positive motor phenomena are common, ‘convulsive syncope’, (Lempert et al. 1994) raising the spectre of epilepsy for the inexperienced diagnostician. The EEG shows only generalized slow activity at this time, without any epileptiform features. It is controversial as to whether true secondary epileptic seizures can occur in as a result of prolonged cerebral hypoperfusion because of syncope.

Loss of consciousness is brief and on recovery the subject is usually nauseated and tremulous with continued pallor and sweating. Crucially the syncopal individual recalls recovery at the site of their collapse, while someone with a tonic-clonic seizure has their first recall on the way to hospital. They may, however, feel washed out and unwell for some time afterwards.

Whilst compression of the carotid sinus in young people rarely causes any symptoms, in the elderly, particularly those with heart disease, it can cause bradycardia and syncope often presenting as ‘falls’. Carotid sinus hypersensitivity may be diagnosed when 5–10 s of carotid sinus massage results in a 50mmHg or greater drop in systolic blood pressure or a 3 s or greater ventricular pause (Brignole et al. 2004). Most commonly this is due to reflex vagal inhibition of the heart. However, more rarely it may be caused by a vasopressor effect leading to a fall in blood pressure independent of heart rate. Very rarely, pressure on one carotid can lead to almost immediate loss of consciousness if there is a grossly stenotic or occluded contralateral artery so that ipsilateral carotid compression in effect leads to a standstill in much of the cerebral circulation.

Carotid sinus syncope is most commonly seen in patients with arteriosclerosis and hypertension but may sometimes occur with local neoplastic disease in the neck or aneurysmal dilatation of the sinus. The usual precipitant in all cases is a sudden turn of the head inducing dizziness and fainting.

Cough syncope is the commonest respiratory cause of syncope. It tends to be seen in middle-aged smokers, usually male, who are usually overweight and have chronic obstructive airways disease. Fainting is precipitated by a paroxysm of continuous coughing. The individual experiences light-headedness followed by unconsciousness. Recovery is usually quick. The prolonged bout of coughing elevates intrathoracic pressure and thereby impedes venous return and cardiac output. It is doubtful that people with normal respiratory function can cough for a sufficiently long period to induce cough syncope (Pederson et al. 1966). More rarely, cough syncope, as well as cough headache, have been associated with cerebellar ectopia and syringomyelia.

Primary or secondary pulmonary hypertension may be associated with syncope because of a failure of the right ventricle to increase output on demand. Syncope may therefore occur during effort in a fashion similar to that seen with aortic stenosis (Ross 1988), with Eisenmenger syndrome and with pulmonary embolus (Soloff and Rodman 1967).

Syncope may occur as a reflex response to sensory stimuli within the territories of the glossopharyngeal or vagus nerves. Guberman and Catching (1986) described 29 patients with syncope associated with swallowing. Almost all patients had some form of oesophageal disorder or cardiac disease, such as ischaemic heart disease or heart block. Rarely cancer of the head and neck can be associated with syncope occurring spontaneously or on swallowing.

Syncope in the period after eating may not be uncommon, particularly in the elderly or in patients with autonomic failure. Syncope can occur as an accompaniment of glossopharyngeal neuralgia (Riley et al. 1942) (Sections 19.2.2; 20.3.3). The pain is precipitated by stimulation, or movement of the oropharynx during chewing, swallowing, or coughing, and precedes syncope. Carbamazepine can be used, although microvascular decompression has also been successful (Tsuboi et al. 1985).

Syncope during and immediately after micturition is not uncommon. It seems to occur in two different groups of people: young healthy men and older people of either sex who have concurrent medical problems (Kapoor et al. 1985). In the first group, syncope usually occurs at the end of voiding after the patient has got out of bed in the middle of the night. It seems to be predisposed to by sleep deprivation, hunger, and intercurrent infection. Bladder distension may be one means of reflexly precipitating fainting which can also occur following the decompression of a painfully distended bladder. The role of the Valsalva manoeuvre during voiding is uncertain.

Syncope may on occasions occur with defaecation in older patients (Pathy 1978). Other pelvic examinations and interventions such as prostatic examination and insertion of intra-uterine devices are associated occasionally with syncopal episodes.

Cardiac syncope may differ from vasovagal syncope as its onset is often rapid and without warning other than palpitation (Ross 1988). It should be suspected in patients with increasing age, vascular risk factors, and known cardiac disease. On rare occasions life threatening dysrhythmias are seen in young people with no previous cardiac history. These patients may undergo neurological referral. Both the congenital long QT syndrome (Schwartz 1985) and Brugada syndrome (Antzelevitch et al. 2002) result from cardiac channelopathies and should be suspected when syncope, collapse, or atypical convulsions are precipitated by exercise, or occur during rest or sleep, or when there is a family history of sudden death. In these circumstances, a routine electrocardiogram should be examined for evidence of prolonged QTc and the characteristic ST segment elevation in leads V1-3 with apparent right bundle branch block, respectively.

Syncope in syndromes of autonomic failure develops slowly over several minutes. A large number of central and peripheral nervous system disorders are associated with symptoms of autonomic failure in which syncope is prominent (Sections 2.7.3 and 21.11.7). Syncope occurs on standing, beginning with a feeling of light- headedness which progresses slowly. Visual obscurations may occur with other central symptoms of cerebral hypo-perfusion but patients show an absence of sweating, bradycardia or pallor.

For the great majority of subjects a well-taken history and eye witness account will allow confident clinical differentiation between simple syncope and seizures. Difficulties may arise in cases of atypical or convulsive syncope where, no single clinical feature will categorically allow a differentiation. The major differences are summarized in Table 31.2. It is of paramount importance to recognize that while seizures characterized by sudden collapse, loss of consciousness, and rapid recovery, but without significant positive motor phenomena, are seen commencing in childhood, they rarely begin for the first time in adults.

Neurologists will see a number of patients where differentiation between syncope and seizures is difficult. For this reason they should be familiar with guidelines on the investigation of syncope (Brignole et al. 2004) and be able to recognize serious cardiac conditions requiring intervention. When syncope occurs with suspected or certain heart disease, cardiac evaluation will include echocardiography, stress testing, and prolonged electrocardiographic monitoring. If severe or recurrent syncope occurs in the absence of heart disease, investigation is required with electrocardiogram, tilt testing, carotid massage, and if negative, prolonged electrocardiographic monitoring. Guidelines for epilepsy draw attention to the need for standard electrocardiogram recordings in all subjects with a diagnosis of epilepsy or suspected epilepsy (NICE 2004). The electrocardiographic abnormalities which suggest an arrhythmic cause of syncope are listed in Table 31.3.

In considering the differentiation between syncope and seizures it has to be recognized that seizures can themselves be associated with significant cardiac dysrhythmia (Rugg-Gunn et al. 2004). Some complex partial seizures are described as ‘temporal lobe syncope’ (Delgado-Escueta et al. 1982). In these, sudden falls occurred without warning, but were followed by amnesia and gradual recovery. They are, however, exceptionally rare as a new seizure type in adults, and often accompanied by other seizure types in keeping with a frontal onset.

The misdiagnosis of epilepsy is common. Psychogenic non- epileptic attacks, or pseudoseizures, cause the greatest diagnostic difficulty, most commonly mimicking tonic-clonic seizures, or more rarely minor convulsive seizures and syncope (Groppel et al. 2000) (Section 4.7.12). Their incidence in the community is difficult to ascertain but could represent 4 per cent of all cases of epilepsy (Sigurdardottir and Olafsson 1998). The incidence increases dramatically as more selected and apparently refractory populations are examined. Thus up to 20 per cent of referrals to tertiary centres are misdiagnosed, and up to 50 per cent of cases of drug-refractory status epilepticus are due to non-epileptic attacks (Howell et al. 1989). Their recognition is important because of the potentially tragic consequences of both inappropriate long-term anti-epileptic drug treatment (Smith et al. 1999) and intensive care unit admission (Reuber et al. 2004).

The diagnosis of pseudoseizures may be difficult, particularly when they occur in patients who also have a history of true epileptic seizures. Eye witness accounts and videotapes of events can raise suspicions, but no single clinical feature differentiates non- epileptic attacks from epilepsy, although a number of factors may be of value (Table 31.4). The most useful clinical features of attacks are resistance to eye opening in pseudoseizures and the absence of pupillary dilatation which is invariably present in tonic-clonic seizures.

The temporal pattern of attacks should, however, alert the clinician to the possibility of non-epileptic attacks. They are refractory to anticonvulsant therapy, in contrast to epilepsy where convulsive seizures in particular are likely to be well controlled. Failure to control apparent tonic-clonic seizures in a patient who develops attacks after the first decade of life, in whom there is no identifiable cerebral disease and in whom interictal EEG recordings have never shown significant epileptiform abnormalities is almost pathonemonic of non-epileptic attacks. With modern anti-epileptic drugs very few patients continue to have frequent tonic-clonic seizures and those who do have preceding severe cerebral insults resulting in intellectual and neurological impairments. In short they have ‘bad brains’; evidence of which is strikingly absent in the majority of people most with non-epileptic attacks.

Some positive features allow identification of subjects at risk. Non-epileptic attacks most commonly occur in women, with onset most commonly in the second or third decades of life (Roy 1979; Howell et al. 1989). Individuals often have a significant history of self-poisoning and self-injury, and previous episodes of unexplained physical symptoms. They come from dysfunctional families, and there is often a history of physical and sexual abuse (Francis and Baker 1999; Reuber and Elger 2003).

The management of non-epileptic attacks is more difficult than the diagnosis. Symptomatic events will usually need to be recorded by videotelemetry and be shown to be characteristic of the usual attacks. An open discussion of the non-epileptic nature is necessary after which anti-epileptic drugs can be withdrawn, usually with advice that this alone will often result in an improvement in attacks. The psychological and psychiatric background needs to be explored by people with experience of the area and an on going programme of treatment and support instituted. The prognosis is variable, and other non-physical symptoms commonly occur in the long term (Reuber et al. 2003). There is controversy about the incidence of epilepsy in people with non-epileptic attacks. It is probably low but those with both problems are exceptionally difficult to manage.

Hyperventilation is common and the bulk of cases may go unrecognized but some may be misdiagnosed as epilepsy (Riley 1982). Common manifestations include dizziness, detachment, blurred vision, tingling, muscle spasm, tetany, palpitation, dyspnoea and chest pain, heartburn, epigastric pain, muscle cramps, and fatigue (Section 4.3.4). Some form of alteration in consciousness is common and up to 15 per cent of patients may lose consciousness during attacks (Pincus 1978). The wide variety of symptoms experienced by patients with hyperventilation will most commonly be confused with complex partial seizures. The most useful factors, which allow differentiation are that hyperventilation attacks are commonly precipitated by stressful circumstances and that they lack a stereotype nature with different types of symptoms referable to different systems occurring on different occasions. A simple diagnostic test involves asking the patient to re-breathe from a paper bag held over the mouth and nose during attacks induced by voluntary hyperventilation.

Panic attacks can be mistaken for complex partial seizures. They often include hyperventilation but also commonly encompass abdominal discomfort, a choking feeling, fear, autonomic symptoms, and sometimes even loss of consciousness. The episodes, however, are usually clearly precipitated and often more prolonged than seizures. Panic attacks are most likely to occur in association with phobic anxiety states and patients usually have considerable insight into the nature of their attacks.

It is controversial whether violent behaviour is common in people with epilepsy and complex partial seizures (van Elst et al. 2000). It is common for individuals who describe sudden outbursts of violent behaviour with minimal provocation, often with some associated patchy amnesia, to be referred for neurological evaluation with a presumptive diagnosis of a seizure disorder. It is striking that individuals are invariably young men from deprived backgrounds who have often themselves been abused. Violence occurs in response to minimal stimuli and is often directed against a specific family member. This is clearly distinct from epilepsy and the term ‘episodic dyscontrol’ has been used (Maletzky 1973). Strict guidelines must be applied before ever accepting that aggressive or violent behaviour is part of a seizure disorder (Treiman and DelgadoEscueta 1983). An epileptic basis to such attacks should only be accepted where definite seizures occur at other times and where violent behaviour is a consistent feature of that individual’s seizures. Violence can only be accepted as seizure related where it is brief and poorly directed.

States of psychogenic wandering are prolonged, usually with sudden recovery of awareness (Section 4.7.12). It is often impossible to obtain a clear account of behaviour during such attacks but this usually appears to have been quite normal. Subjects have a dense amnesia for the period of time concerned and individuals usually have an associated depression and the need to escape from some stressful life situation (Stengel 1943). Such episodes may be confused with complex partial status or other forms of non-convulsive status epilepticus (Mayeux et al. 1979), but here there is usually evidence of abnormal behaviour during the amnesia, which is briefer and more frequent (Kopelman et al. 1994), and more akin to transient global amnesia (Section 31.1.5).

On rare occasions it is difficult to differentiate between focal seizures and focal ischaemia due to either migraine or thrombo- embolic disease.

Transient ischaemic attacks will rarely be confused with seizures because they develop more slowly and last longer. They are virtually never accompanied by altered consciousness, and motor and sensory phenomena that occur are almost uniformly negative. Whilst rarely focal seizures may rarely be accompanied by predominantly negative motor or sensory problems (Lesser et al. 1987), the greatest difficulties occur in rare haemodynamic transient ischaemic attacks in which weakness may be accompanied by some shaking and tremor (Yanagihara et al. 1985).

Migraine and epilepsy occur in the same individual more commonly than would be expected by chance. Loss of consciousness is not uncommon in migraine though usually it takes the form of syncope associated with nausea and hypotension (Risser 1985). The complex relationship between migraine and epilepsy has been reviewed (Haut et al. 2006). On occasions migrainous episodes may induce frank seizures and arteriovenous malformations may cause both seizures and migraine-like phenomena. It does seem that migraine and benign Rolandic and occipital epilepsies commonly co-exist, as they may in some mitochondrial disorders such as the MELAS syndrome of Mitochondrial Encephalopathy, Lactic Acidosis, and Stroke-like episodes (Section 35.4.8).

The syndrome of transient global amnesia describes an abrupt onset of amnesia usually accompanied by repetitive questioning, in an individual who remains alert and communicative (Fisher and Adams 1958). The amnesia lasts for hours and attacks recur only rarely. The aetiology of this syndrome remains controversial but it is highly likely that it has different causes which may include thrombo-embolic disease, migraine, and epilepsy (Hodges and Warlow 1990). Up to 7 per cent of patients with transient global amnesia are likely to have an epileptic basis to their attacks. These can usually be identified by attacks that are brief and which recur over a short period of time. They are common on wakening, and individuals may have some partial recall. Sometimes such individuals will also describe some features at the start of the attacks, which would support focal seizure onset, for example, olfactory hallucination. Other types of simple or complex partial seizures occur, although transient amnesia can be the sole manifestation of seizures (Zeman et al. 1998).

A number of sleep phenomena may be confused with seizures (Section 32.4). Hypnic jerks are usually single jerks that occur in the very early stages of sleep (Section 32.4.1). They usually lead to arousal and may be accompanied by a feeling of falling, a cry, or some other kind of brief sensory disturbance. Periodic movements of sleep are rhythmic and repetitive leg movements more commonly seen in later life (Section 32.4.5). They usually consist of jerking movements, usually dorsiflexion of the foot and extension of the toes that can occur repetitively during non-rapid eye movement sleep (Coleman et al. 1980). Sleep walking is a form of automatic behaviour occurring during deep non-rapid eye movement sleep and is much more common in children than in adults (Section 32.4.2). It usually ceases by the mid-teens. In adults it must be distinguished from post-ictal automatism following sleep seizures or from complex partial seizures resulting in automatic behaviour (Pedley and Guilleminault 1977).

Abnormalities of sleep can also lead to confusion with seizures because of daytime disturbances. Narcolepsy should not lead to significant confusion (Section 32.3.1). However, cataplexy (Section 32.3.1) in which there may be sudden collapse with loss of postural tone triggered by emotion or startle or loud noise could potentially be confused with atonic drop attacks. However, the age of onset of such episodes should preclude real confusion; atonic seizures most commonly occur during the first decade of life whereas symptoms of the narcoleptic syndrome rarely begin before the second or third decades of life. Some subjects with narcolepsy can also exhibit automatic behaviour when they appear to be only half awake. The individual appears drowsy and absent minded, though may be capable of carrying on relatively complex tasks for which they are subsequently amnesic.

Sleep apnoea, which is most commonly obstructive in nature and associated with obesity and night-time snoring leads to day-time drowsiness (Section 32.3.4). Occasionally episodes of day-time sleepiness may be associated with respiratory obstruction and jerks that can lead to referral with the suggestion of a seizure disorder.

Some unusual movement disorders may on occasion cause confusion with seizures. Non-epileptic myoclonus must be differentiated from epileptic myoclonus (Section 40.7). Paroxysmal kinesogenic choreoathetosis is a rare disorder in which short-lasting tonic spasms with writhing movements, usually affecting an arm or a leg occur (Kertesz 1967) (Section 40.4.7). The onset is usually in adolescence and there is a strong male predominance, and often a family history. The attacks are precipitated by sudden movement after a period of rest and are often preceded by a peculiar sensation in the limb before the movement commences. They can occur frequently but respond readily to antiepileptic drugs, though there is no evidence to suggest that they are epileptic in nature. The syndrome is so striking that it should not easily be confused with a seizure disorder although rarely some patients with movement-induced seizures and abnormal ictal EEGs have been described. A non-kinesigenic form is described with very similar movements but usually of much longer duration (Mount and Reback 1940). This is familial with onset in infancy or childhood and there is no male predominance. Tonic spasms induced by movement can occur in multiple sclerosis as may paroxysmal episodes of dysarthria and ataxia (Section 37.5.3).

A number of metabolic disturbances may result in acute symptomatic seizures (Section 31.7). Hypoglycaemia is unusual in that it may be recurrent and give rise to diagnostic confusion with epilepsy. Hypoglycaemia is most commonly seen in diabetics receiving insulin or oral hypoglycaemic agents. Diabetics may be particularly sensitive to hypoglycaemia and can experience symptoms at higher blood glucose levels than non-diabetic subjects. Marks (1981) describes different types of neuroglycopenia, the commonest of which is acute. As blood sugar falls, pallor, sweating, and tachycardia develop associated with confusion, collapse, and occasionally coma. True seizures may occur during the course of hypoglycaemia further complicating diagnosis. Hypoglycaemia must always be considered in the differential diagnosis of epilepsy in a diabetic population but perhaps the greatest diagnostic difficulty will arise in the rare cases of insulin-secreting tumours in non-diabetic patients. Here, hypoglycaemia is most likely to occur during the course of the night. Less common is a syndrome of ‘subacute neuroglycopenia’ characteristic of insulinoma. Subjective symptoms are not marked but there is mild confusion and clumsiness. There may be disinhibition, suggesting intoxication with ataxia and slurred speech. This confused behaviour of hypoglycaemia may be difficult to differentiate from complex partial seizures.

A negative motor phenomenon that may sometimes be confused with epilepsy is the syndrome of cryptogenic or benign drop attacks of middle-aged women (Stevens and Matthews 1973) (Section 2.6.5). These result in a sudden fall, usually onto the knees, without any clouding of consciousness. The episodes tend to be infrequent and the outcome seems to be quite benign. It must be emphasized that while atonic, tonic, and myoclonic seizures can cause brief episodes of falling with rapid recovery in children and occasionally in adolescence, such attacks do not commence in adult life.

McCrory (1997) drew attention to concussive convulsions as a non-epileptic phenomenon in Australian rules football. Convulsions began within 2 sec of the head injury and usually consisted of a brief tonic phase followed by bilateral myoclonic jerks. Some versive head movements and asymmetric posturing were seen in some individuals. No convulsion lasted for more than 150 sec and none of the players had any behavioural or neuropsychological features that suggested anything other than a mild concussion. While these events have been widely assumed to represent a form of post- traumatic epileptic seizure, there is now considerable evidence to the contrary. Follow-up of subjects with such events by Jennett indicated that immediate convulsions were not a predictor of late post-traumatic epilepsy, in contradistinction to other seizures occurring in the first post-traumatic week, and that on the whole they tended to be associated with relatively mild concussive head injury (Jennett 1975). Similar events can occur with head injuries during other sports such as horse racing (Chadwick 1997). It is important that they are recognized as non-epileptic in order that inappropriate restrictions are avoided.

In man spikes and sharp waves are the electroencephalographic hallmarks of interictal recordings of many patients with epilepsy. Such activity appears to be due to a hypersynchronization of electrical activity within an abnormal pool of neurones, and they are rarely seen in the EEGs of non-epileptic patients (Section 31.9.1). Simplistically, epileptic seizures occur when excitatory influences in the cerebral hemispheres outweigh inhibitory influences. Study of the basic mechanisms of the human epilepsies is, of course, fraught with ethical and practical difficulty. Thus much knowledge has been accumulated from a number of animal models of seizures and epilepsy. While the direct relevance to the human epilepsies remains in some doubt, it seems likely that knowledge gained in this way will be highly informative.

There is considerable evidence that the fundamental building block of most seizure disorders is the paroxysmal depolarization shift and an associated high frequency burst firing of neurones (Prince 1978). This phenomenon is one that can be observed in isolated cells, within simple neuronal circuits, in animal models of epilepsy, and indeed in human post-operative material (Fig. 31.1). Paroxysmal depolarization shift may occur normally as part of the spontaneous activity of CA3 hippocampal pyramidal neurones (Wong and Prince 1981), and in pyramidal cells of layers 4 and 5 of the neocortex (Gutnick et al. 1982). While paroxysmal depolarization shift and burst firing can be intrinsic properties of neurones, the propagation and synchronization of this kind of activity to produce either interictal spikes or seizures, requires a contribution from neuronal circuits, which may themselves exhibit abnormalities predisposing neurones within them to behave in an abnormal fashion. The concept of paroxysmal depolarization shift and burst firing underlies not only to our understanding of the basic mechanisms underlying epilepsy, but also to the mechanisms of and targets for anti-epileptic drug action (White 1997).

A number of factors control neuronal excitability. These include voltage-gated ion channels, neurotransmitter ligand activated ion channels, neuromodulators, and second messenger systems. Ligand-gated ion channels are responsible for communication between cells while voltage-gated channels determine how inhibitory and excitatory influences are integrated in a way that determines the propagation of impulses to other neurones.

Neuronal membranes are usually polarized to a potential of -90 mV by the activity of Na⁺–K⁺–ATPase transporter systems. Voltage-gated ion channels are membrane-spanning proteins composed of different sub-units that when open, permit the passage of ions. Openings may be transient or persistent depending on the nature of the channel. Most channels will open on depolarization of the membrane, but some open when the membrane is hyperpolarized. Channels may exhibit a number of different states (Fig. 31.2).

Voltage-gated sodium channels are intimately involved in the propagation of action potentials, the rapid upstroke being due to an opening of fast transient channels at about – 60 mV. In addition there is also a persistent component of the current that may be of greater relevance to epilepsy. Toxins that prolong sodium channel opening cause burst firing and seizures (Mantegazza et al. 1998). Phenytoin and carbamazepine are able to block repetitive firing of neurones effect on voltage-gated sodium channels that is both use-dependent and voltage-dependent; sodium valproate has a similar action also (Fig. 31.3). For a number of reasons, this mechanism of action may be seen as having ideal properties. It will tend to block the pathological neuronal activity of repetitive firing with relatively little effect on more physiological patterns of activity, and would also be predicted to be highly effective in preventing the spread of seizure activity. The importance of voltage-gated sodium channels in human epilepsy is further emphasized by the finding of molecular abnormalities in families with generalized epilepsy and febrile seizures and patients with severe myoclonic epilepsy of infancy (Escayg et al. 2000; Claes et al. 2001).

Voltage-dependent calcium channels contribute to dendritic spikes, slow somatic depolarizations, and associated burst discharges, and by doing so, trigger neurotransmitter release. Six sub-classes of calcium channels are known to exist: L, N, T, P, Q, and R. T channels have a low threshold of activation at around –70 mV. They inactivate relatively rapidly. These channels are found in high concentrations in thalamic neurones and play an important role in the generation of generalized spike wave discharges. They appear sensitive to anti-absence drugs such as ethosuximide, methadione derivatives, and sodium valproate (Coulter et al. 1990). Over-expression of low-threshold Ca⁺ channels appears to be implicated in animal models of absence and generalized spike wave (Burgess and Noebels 1999). A mutation of CACNA1A has been found in a patient with absence epilepsy (Jouvenceau et al. 2001).

Voltage-gated potassium channels are very diverse in their nature. Delayed rectifying potassium currents, I_k, activate at potentials more positive than –40 mV and seem to contribute to spike repolarization. Fast transient currents activate at –45 mV to –60 mV and they appear to play an important role in the regulation of repetitive firing by prolonging after-spike hyperpolarization and slowing down firing rate. Inward rectifying potassium currents activate in response to hyperpolarization, and help regulate resting membrane potentials. Potassium currents can be blocked with tetraethylammonium and 4-aminopyridine (Jones and Heinemann 1987).

Mutations of KCNQ2 and KCNQ3 encoding K⁺ channel subunits can lead to benign neonatal convulsions (Biervert et al. 1998; Singh et al. 1998). The potentiation of voltage-dependent potassium currents is currently being explored as a potential target for new anti-epileptic drugs.

γ-aminobutyric acid, GABA, is the major inhibitory neurotransmitter in the forebrain, being present at approximately 30 per cent of all synapses in the central nervous system. The majority of GABAergic neurones are short axon interneurones forming local inhibitory loops. Three distinct types of GABA receptor are recognized. GABA_A and GABA_C receptors are receptors linked to chloride ion channels. Metabotropic GABA_B receptors may be pre- or post-synaptic and are coupled to calcium or potassium ion channels via GTP proteins (Bowery 1993; Macdonald and Olsen 1994). GABA_A receptors are large molecular weight proteins that contain a number of binding sites, not only for GABA but also for barbiturates, benzodiazepines, picrotoxin, and anaesthetic steroids. Binding of GABA leads to an opening of the chloride and potassium ion channels and resultant hyperpolarization, known as inhibitory post-synaptic potentials. The receptor has been cloned and five different subunit families isolated. Receptors constructed from different combinations of sub-units appear to exhibit differing pharmacological properties, though binding sites for GABA and barbiturates appear highly conserved.

Drugs such as allylgylcine, which prevents the synthesis of GABA, picrotoxin, and bicucculine which block GABA receptors, as well as penicillin, are all potent convulsant agents. The activity of benzodiazepines and barbiturates at GABA_A receptor appears responsible for their anti-epileptic activity (White 1997). Vigabatrin, a rationally developed anti-epileptic drug, is a suicidal inhibitor of GABA-transaminase, the enzyme responsible for GABA metabolism. Tiagabine potentiates GABAergic activity by blocking the re-uptake of GABA into neurones and glia. Mutations of the GABRG2 gene encoding for GABA_A subunits have been found in generalized epilepsy and febrile seizures (Wallace et al. 1998; Baulac et al. 2001). On the other hand, enhanced GABAergic inhibition via GABA_B receptors appears to worsen absence seizures in humans and experimental animals (Section 31.2.2).

The major excitatory neurotransmitters are the amino acids L-glutamate and L-aspartate. They exert their synaptic influences by interacting with a number of different types of receptors, which are identified because of specificity for binding different molecules. Binding at the DL-α-amino-3-hydroxy-5-methyl-isoxazolepropionic acid, AMPA, receptor makes its associated ion channel permeable to both sodium and potassium. It desensitizes rapidly (Tang et al. 1989) and is probably responsible for the majority of rapid excitatory neurotransmission. Four glutamate receptors genes, GluRs1-4, have now been cloned and appear to encode sub-units that can express the known electrophysiology and pharmacology of the AMPA receptor. The kainate receptor is also coupled to a channel permeable to sodium and potassium. It does not, however, appear to desensitize at the same rate as the AMPA receptor. Both these channels differ strikingly from the N-methyl-D-aspartate, NMDA, receptor. This appears to be a much more complex receptor site that has an absolute requirement for the presence of a co-agonist, glycine, in order to result in channel opening (Johnson and Ascher 1987). Magnesium, zinc, polyamines, and steroids can also modulate the site. When membranes are hyperpolarized the channel is blocked by magnesium, a blockade that is reversed when the membrane depolarizes. Opening of the channel allows the entry of both sodium and calcium. This acts as an amplification mechanism that leads to prolonged activation of already excited neurones and associated burst firing (Williamson and Wheal 1992). Calcium entry may also ultimately result in excitotoxicity and cell death. Like the AMPA and kainate receptor, the NMDA receptor exists as a number of sub-families.

Excitatory amino acids are also able to interact with metabotropic receptors that activate second messenger systems to influence biochemical pathways and ion channels. These receptors are found both pre-synaptically and post-synaptically. Activation usually results in pre-synaptic inhibition and post-synaptic excitation. These receptors may have an important role in supporting epileptic activity (Arvanov et al. 1995).

The role of nicotinic acetylcholine receptors in the central nervous system excitabailty is poorly understood, though they may serve as ligand-gated sodium channels. Mutation of a gene, CHRNA4, which encodes for the β2 sub-unit of the receptor has been associated with autosomal dominant nocturnal frontal lobe epilepsy (Steinlein et al. 1995).

While molecular changes may predispose to burst firing of neurons, synchronization of such activity, necessary for seizures also requires the involvement of neuronal circuits. Here our knowledge of the way in which circuits operate to cause or suppress seizures is fragmentary. We have little indication of why a fixed alteration in a gene-product may give rise to an intermittent and paroxysmal disturbance. While we may be able to identify the causes of epilepsy in many more patients at molecular or lesional levels, what are the changes within the brain that result in clustering of seizures on 2 or 3 days, but not on others in a month? This remains the fundamental clinical question in epilepsy.

The most studied and clinically relevant model of focal seizures and epileptogenesis is the hippocampus and hippocampal sclerosis, the most common form of focal epilepsy in man. Mesial temporal structures are interconnected by a reverberating loop involving enterorhinal cortex, dentate gyrus, CA3, CA1, enterorhinal cortex (Fig. 31.4). Normal spontaneous activity of CA3 pyramidal neurones consists of paroxysmal depolarization shifts and associated burst firing of the cell body and apical dendrites (Wong and Prince 1979; 1981). Function of the hippocampal brain slice can be studied in material from normal animals exposed to a variety of chemical manipulations as well as in slices from animals expressing a chronic model of epilepsy (Traub and Jefferys 1998). In normal brain material, bursting activity in CA3 neurones has about a 30 per cent chance of evoking bursts in connected neurones. This probability increases when the pyramidal cell is excited by many inputs simultaneously.

This system is also subject to plasticity in response to a number of stimuli or to damage. Three phenomena; cell loss, mossy fibre sprouting, and neurogenesis may all enhance the potential for seizure generation while resulting in pathological change that can lead to hippocampal sclerosis. In the kindling model of epilepsy, repeated sub-convulsive electrical stimulation, usually to the amygdala, leads to increasing after-discharge and ultimately to behavioural seizures (Goddard et al. 1969). Repeated focal applications of convulsant agents can lead to a similar phenomenon. It appears that limbic structures are particularly sensitive to the development of kindling when compared to neocortex, a situation that is reflected in man, where the temporal lobe is by far the most common site of seizure onset (Section 31.4.2).

At a pathological level, kindled animals show evidence of neuronal loss in the hippocampus accompanied by sprouting of the mossy fibre axons of the dentate granule cells (Sutula et al. 1988). It seems that sprouting probably requires the death of neurones and that the sprouting fibres take up synaptic sites that are thereby vacated. Similar changes are seen after experimental damage from status epilepticus (Tauck and Nadler 1985). If new synaptic connections are made to other excitatory cells, this would represent a process potentially contributing to the hyperexcitability of the kindled brain. More recently there has been evidence of post-natal neurogenesis occurring in the dentate gyrus through life (Eriksson et al. 1998). In the pilocarpine model of chronic epilepsy, seizures can induce neurogenesis, neurones being abnormally integrated into existing circuits (Parent and Lowenstein 1997). These changes therefore provide a potential basis for the clinical predisposition of the mesial temporal structures to produce seizures and a plausible mechanism for some of the progressive changes that may be seen as part of drug-resistant epilepsies. Full proof of the concept is however, lacking.

Here the sudden onset, the bilateral synchrony, and in the case of simple absences and myoclonus, the non-evolving pattern of electrical activity clearly suggests some very generalized disturbance of neuronal activity in both hemispheres. Historically, there have been two schools of thought that would explain such a phenomenon; that it is a primary cortical process or that there is an unspecified ‘centrencephalic’ system in which structures of the upper brainstem and thalamus are responsible for generating the spike wave discharge and driving a cortical synchrony. Gloor (1968) has pointed out that these two hypotheses are not mutually exclusive and has developed a ‘generalized cortico-reticular’ hypothesis.

The cellular substrate for these phenomena is now well- understood (Fig. 31.5). It is dependent on a thalamocortical circuit that includes the nucleus reticularis thalami. The circuits involve excitatory glutaminergic synapses and inhibitory gabergic synapses. The behaviour of thalamic neurones and the circuit is largely determined by the presence of a high density of calcium T chanels, which results in Ca²⁺/K⁺-dependent burst firing. These in turn can give rise to strong inhibitory postsynaptic potentials mediated by GABA_B receptors in thalamocortical relay neurones.

This circuitry is important in activating cortical neurones during sleep–waking cycles. Tonic activity in the relay neurones occurs during wakefulness and rapid eye movement sleep, but they fire in the burst mode during non-rapid eye movement sleep. In the awake animal, thalamic neurones are maintained at a resting potential of approximately –50 mV because of the effects of normal afferent activity of brainstem-activating systems. In this state, calcium T channels are not activated. During drowsiness and sleep, however, thalamic neurones hyperpolarize and begin to exhibit typical repetitive burst firing that contributes to sleep spindles in the EEG (Steriade et al. 1993). The classical anti-absence drug ethosuximide acts by causing a voltage-dependent blockade of T-type calcium currents a property shared by valproate. Hyperpolarization and burst firing is greatly facilitated by GABAergic activity via GABA_B mechanisms. Thus, GABA_B receptor agonists such as baclofen, exacerbate absence and GABA_B antagonists have anti-absence properties in animal models (Hosford et al. 1992; Liu et al. 1992). These phenomena probably also explain the effects of vigabatrin in exacerbating absence seizures.

Despite problems with differing definitions of epilepsy and case ascertainment methods, there is remarkable agreement about the epidemiology of epilepsy in different populations in the developed world (Sander et al. 1990; Sander 2003) Incidence rates vary in the range of 20–55/100 000 per year whereas the prevalence for active epilepsy is in the range of 4–10/1000. Age-specific incidence, prevalence, and cumulative incidence are described for a population in Rochester, Minnesota (Fig. 31.6) and replicated in other populations. It can be seen that the incidence of epilepsy is highest at the extremes of life but that there are significant differences between the cumulative incidence and prevalence of epilepsy, indicating that the majority of patients who develop epilepsy do not suffer from a chronic disorder. The cumulative incidence of epilepsy by the age of 70 may be as high as 2–3 per cent of the population. There is evidence that the incidence in children may be falling with time (Section 30.3.2), although it may be rising in the elderly. Incidence and prevalence, is higher in third world countries than the developed world, with higher rates usually found in rural as opposed to urban communities.

Most studies suggest a slightly higher incidence and prevalence in men than women, with approximately two-thirds of cases being partial epilepsies and approximately one-third generalized. Epidemiological studies identify a wide range of risk factors for the development of epilepsy (Fig. 31.7).

Community rather than hospital-based studies (Annegers et al. 1979b) and the National General Practice Survey of Epilepsy (Cockerell et al. 1997) are the best source of information in this area. Four hundred and fifty seven patients identified in Rochester, with a history of two or more non-febrile seizures were followed for at least 5 years, and in the case of 141 for 20 years. The probability of being in a remission lasting for 5 years or more was 61 per cent at 10 years, and as high as 70 per cent at 20 years (Fig. 31.8). Similarly in the general practice survey 68 per cent of patients achieved a 5-year remission by 9 years of follow-up. Further support for such high rates of remission is obtained from studies of patients followed prospectively from diagnosis and the commencement of therapy, which show that between 50 and 77 per cent of such patients are ‘controlled’ depending on how control is defined (Reynolds 1987).

The converse of this information is that 20–30 per cent of patients with epilepsy never achieve remissions; they have a refractory epilepsy that is associated with psychosocial handicap (Jacoby et al. 1996). Relative few patients switch between seizure and seizure-free states, indicating that epilepsy is bimodal in its outcome. The question then arises as to what determines prognosis?

The type of epilepsy or epilepsy syndrome is of considerable importance. Thus, juvenile myoclonic epilepsy is life long, but benign Rolandic epilepsy never recurs during adult life. However, many children and adults with epilepsy cannot be classified by epilepsy syndrome and even within syndromes there may be considerable variation in outcome. In West’s syndrome, follow-up for up to 35 years showed that 30 per cent had died, but 24 per cent survived with normal intelligence and 35 per cent were seizure-free (Riikonen 1996). Partial epilepsies have a poorer prognosis than generalized epilepsy (Cockerell et al. 1997).

The age of onset of epilepsy is perhaps one of the most important factors affecting outcome. The commencement of seizures within the first year of life, when it is usually symptomatic of cerebral pathology and indicative of one of the malignant childhood epilepsies, carries a particularly adverse prognosis (Sofijanov 1982). In childhood the risk of intractability falls strongly with each additional year of age of onset (Berg et al. 1996). The same effect of age is seen in studies which include adults, though the effect is weaker.

Whatever the age of onset, the long-term outcome of epilepsy can usually be predicted around the time of diagnosis. The outcome is worse for the more seizures that occur before diagnosis (Cockerell 1997). Annegers et al. (1979a) showed that most patients who achieve remission do so early during the course of treatment. With continuing seizures it becomes progressively less likely that an

individual patient will enter remission. Thus, there is a plateau in the number of patients in remission 15–20 years after the onset of epilepsy.

Most studies indicate that symptomatic epilepsy, with an identified cause, carries a poorer prognosis and individuals with associated neurological, cognitive, behavioural, and psychiatric impairment also fare worse (Hauser and Hesdorffer, 1990).

All available studies show an increased mortality ratio for epilepsy of between 2 and 3 times the expected (Cockerell 1997). A number of factors appear to contribute to this excess mortality. The greatest excess occurs in the early years of life and is more obvious in men than women. The risk of mortality is greatest in the early years following diagnosis, and is highest for patients with tonic-clonic seizures, seizures that recur frequently, and for those with remote symptomatic epilepsy.

Some of the excess mortality seems to be associated with the underlying aetiology of an epilepsy rather than the occurrence of seizures themselves. The association of epilepsy with cerebrovascular disease in later life probably accounts for the excess mortality associated with the diagnosis of epilepsy in the elderly. The association with mental handicap and cerebral palsy in younger age groups seems to be of considerable importance. Both neoplasms and arteriovenous malformations contribute to the excess mortality from symptomatic epilepsies.

Some controversy surrounds the role of sudden unexpected death in people with epilepsy. A population-based incidence study (Ficker et al. 1998) indicates that the incidence of sudden unexplained death was over 20 times higher in people with epilepsy than in the community as a whole. The overall incidence of sudden death in this study in people with epilepsy was 0.35 per 1000 patient years. Thus, it remains an extremely rare event. The major risk factors for sudden unexpected deaths appear to be young age, early onset of seizures, the presence of tonic-clonic seizures, the male sex, and being in bed. While status epilepticus continues to be associated with mortality its rarity means that it does not contribute significantly to the excess mortality associated with epilepsy.

Accident is a not uncommon cause of death in epilepsy, accounting for up to 10 per cent of all deaths, drowning being responsible for the great majority (Blisard and McFeeley 1988). Accidents are also a common cause of injury. In a community-based study in the United Kingdom, 24 per cent of those with active epilepsy-reported head injuries within the previous year, and 16 per cent burns (Buck et al. 1997).

The great heterogeneity of clinical phenomena that can be associated with seizure discharge necessitates some system of classification. An international classification of epileptic seizures was proposed in 1981 (Table 31.5) (Commission on classification and terminology of the International League Against Epilepsy 1981). This classification broadly divides seizures into focal or partial seizures, which begin locally, and which may spread or evolve into secondary generalized tonic-clonic seizures, and generalized seizures in which the onset is sudden and in which both cerebral hemispheres are involved in the discharge from a very early point in the seizure. The classification makes use of both clinical and electroencephalographic information. However, similar seizures may occur at different ages and have very different implications. Conversely, a patient may experience differing seizures during the course of their life so that a classification of different epileptic syndromes based on seizure types occurring within the syndrome, age of onset, and aetiology will also be of vital importance in the management of patients with epilepsy. A classification of epilepsy syndromes is presented in Table 31.6 (Commission on

Classification and Terminology of the International League Against Epilepsy 1989b). There has been further discussion about the adequacy of these classifications (Engel 2001), but they have not been replaced. It is must be recognized that aetiological classification may be as important as classification based on clinical phenomenology.

Classical neurological teaching has suggested that motor and sensory phenomena may be used to infer precise localizing value. This has increasingly been questioned as more detailed intracranial recording has become available to correlate with observed clinical phenomena. Thus, the clinical phenomena of seizures reflect not only the site of origin, but also the structures through which seizure discharge spreads during the seizure.

Simple partial seizures with motor signs may give rise to clonic or tonic movements involving any part of the body. Seizures most often involve the face or hand area, because of the disproportionate amount of the motor cortex occupied by the somatotopic representation of these parts of the body. Whilst clonic seizures involving an arm or a leg may be taken as a reasonably satisfactory indication of seizures involving the motor strip, movements of the eye or facial muscles around the eye can be produced by occipital discharge and clonic movements of the mouth, tongue, or pharynx by temporal discharge (Lesser et al. 1987). True Jacksonian ‘march’ with a slow spread of clonic activity from one muscle group to another is uncommon, but when seen does seem to imply relatively specific localization to the pre-central gyrus.

It is evident that tonic motor seizures resulting in version or dystonic posturing of limbs have considerably less localizing value than was previously understood. They may arise from involvement of wider areas of cortex that include the pre-motor region and supplementary motor area. Versive movement, however, can occur in seizures arising in temporal, parietal, and occipital lobes as well as in generalized seizures. Ipsilateral head turning can be as common as contralateral head turning (Ochs et al. 1984). However, version occurs earlier in seizures with frontal onset than in those with temporal onset.

Negative motor phenomena may occur during simple partial seizures. Speech arrest may be the most common manifestation. Whilst speech arrest with a preserved ability to understand speech suggests a seizure in the dominant inferior frontal gyrus, less specific forms of speech arrest can occur with seizure onset in the supplementary motor areas of the dominant or non-dominant hemisphere (Geier et al. 1977). Very rarely, inhibition of movement has been described as part of simple partial seizures. Post-ictal paralysis, or Todd’s paralysis, seems specific for seizures involving the contralateral motor strip at some point. Versive seizures tend not to be followed by such paralysis.

These are rare and occur in no more than 2 per cent of patients with epilepsy. Somatosensory seizures usually arise in the post-central area, but almost always spread to the pre-motor strip at an early stage so that motor phenomena dominate the seizure semiology. They most commonly affect the face or hand and may occasionally spread as do Jacksonian seizures. Sensations are usually those of paraesthesiae or numbness. Rarely, disturbing or painful sensations can occur. Post-ictal numbness similar to Todd’s phenomena may occur.

Primitive visual symptoms including spots, flashes of light, or patterns in one visual field are most commonly associated with occipital seizures. On occasion occipital seizures may produce visual symptoms that involve both half visual fields. More complex visual hallucinations are less likely to have an occipital onset.

Whilst non-specific ‘dizziness’ is often described by patients as part of a simple partial seizure, it usually seems that this term is used because of difficulties in describing complex sensory disturbances. True vertigo must be exceptionally uncommon. Buzzing, hissing, whistling, and ringing noises can be experienced most commonly with involvement of the lateral parts of the temporal lobe.

Olfactory and gustatory symptoms are commonly associated with medial temporal involvement or involvement of the frontal orbital regions. Smells and tastes are usually unpleasant but may be difficult to characterize further than this.

Visceral symptoms are a common component of simple partial seizures. They are most often associated with involvement of limbic structures of the temporal and frontal lobes. Most common is an epigastric sensation, sometimes described as butterflies or nausea which tends to rise characteristically into the throat or mouth. More rarely stomach pain, belching, or even vomiting can occur. Autonomic symptoms can include pallor, flushing and sweating, pupillary dilatation, and increases in heart rate. Involuntary micturition or defaecation whilst not uncommon in seizures associated with loss of consciousness is extremely rare in simple partial seizures.

Complicated psychic symptoms are not uncommon during simple partial seizures. Again they usually indicate involvement of mesial temporal or frontal limbic structures. Psychic symptoms are often associated with other olfactory, gustatory, or autonomic disturbances. Dysmnesic symptoms are perhaps the most common with sensations of familiarity, known as déjà vu, or strangeness, jamais vu. Memory flashbacks or playbacks may occur and may merge into rather non-specific symptoms of dream-like states, unreality and depersonalization on the one hand, and more formed illusions and hallucinations combining visual and auditory aspects on the other. A variety of perceptual changes may occur during seizures with objects appearing larger or smaller or changing in shape or being perseverated (Gloor et al. 1982). Emotional experiences are often described, the most frequent being intense fear. Pleasurable sensations are much rarer but laughter can occur. Anger or rage is extremely rare as a true ictal disturbance.

Complex partial seizures in adults are of considerable importance. They are the predominant seizure type in approximately 40 per cent of patients with epilepsy (Juul-Jensen and Foldspang 1983). They tend to be more resistant to anti-epileptic drug treatment with at best only 50 per cent of patients attaining long-term remissions. Patients with this seizure type commonly present additional psychological or psychiatric handicap, which greatly adds to the complexity of the management problems that they present. Additionally, it is in this group of patients that a surgical approach to the treatment of epilepsy is likely to be most successful.

The international classification of seizures emphasizes that complex partial seizures must include some impairment of consciousness and amnesia. They may or may not be preceded by symptoms of a simple partial seizure and they may or may not be associated with automatism. The site of origin for complex partial seizures is most commonly in the medial temporal lobes in 70 per cent of cases, and the frontal lobes in 20 per cent. The onset of the complex partial seizure occurs when the seizure discharge spreads to involve the limbic system bilaterally. It is important to emphasize that complex partial seizures are not identical or synonymous with temporal lobe seizures and that there is no absolute correlation between the phenomenology of complex partial seizures and their site of origin.

The most typical form of temporal lobe complex partial seizure will start with a visceral aura of a simple partial seizure. There is then an arrest of activity and a motionless stare followed by a phase of stereotyped automatism. Such automatisms occur in a similar fashion in most of an individual patient’s seizures and most commonly consist of lip smacking, chewing, or swallowing, or picking at clothes or fidgeting with objects. Walking or running or verbal automatisms are less frequent. Dystonic asymmetrical posturing is common. Stereotyped automatisms are usually followed by a phase of confusion associated with reactive automatisms. In these, the patient may continue with a previous activity or begin some form of activity that may be considerably influenced by the patient’s immediate environment. Restraint during this phase of a seizure may sometimes give rise to reactive violent behaviour. It is probable that the majority of such typical complex partial seizures arise from the hippocampus, but by no means all do so. Patients with this seizure type often have well localized anterior temporal spikes and sharp waves that may be unilateral or bilateral.

Complex partial seizures arising from the frontal regions are more likely to exhibit some of the following phenomena: they are usually frequent, brief, and less commonly followed by post-ictal confusion, the onset of seizures is without warning and automatisms begin immediately without a preceding motionless stare, automatisms tend to be bilateral and include thrashing, rolling, kicking, or bicycling movements, sexual automatism with pelvic thrusting may be more common with seizures of frontal onset. There is a marked predominance of sleep seizures and complex partial status seems to be more common with frontal seizures (Williamson et al. 1985; Quesney 1986). Inter-ictally the EEG may be unremarkable or show apparently generalized epileptiform abnormalities.

Secondarily generalized seizures are relatively uncommon and may only contribute approximately 9 per cent of seizures in adults compared to approximately 16 per cent in children (Gastaut et al. 1975). Furthermore, they are particularly amenable to anti-epileptic drug treatment and the majority of patients developing partial or focal epilepsy in adult life have very few secondarily generalized seizures even though their partial seizures remain a management problem. In adults, secondary generalized seizures not infrequently occur during sleep. Late-onset tonic-clonic seizures occurring during sleep must be regarded as having a focal onset until proved otherwise.

Whilst typical absence, myoclonus and generalized tonic-clonic seizures may begin in adolescence and early adult life, other forms of generalized seizures such as atypical absence, tonic and atonic seizures are features of age-related childhood epilepsies. While they may persist into adult life, they rarely if ever commence in adult life. However, many children with severe childhood epilepsies tend to develop more typical partial seizures (particularly complex partial seizures) or may expect remission when they enter adult life (Huttenlocher and Hapke 1990).

Almost all seizures are self-limiting events, though the mechanisms leading to termination of seizures are poorly understood. Status epilepticus can be defined as two or more seizures occurring without full recovery of function, or as more continuous seizure activity for 30 min or more. While it is true that there are as many types of status epilepticus as there are types of seizures, it is increasingly apparent that status has a dynamic of its own that goes beyond the semiology of seizures themselves. At a simplified level, status may be classified into generalized convulsive status epilepticus, non-convulsive status epilepticus including complex partial and absence status, and simple partial status.

Generalized convulsive status epilepticus is most common. Best estimates of its incidence lie in the range of 200–300/1 000 000 (Shorvon 1994). It probably makes up about 70 per cent of all cases of status and is associated with the highest morbidity and mortality. Mortality rates for generalized convulsive status epilepticus remain high with most series demonstrating rates of 10–20 per cent. It is more common in children than in adults.

Generalized convulsive status epilepticus may have many different causes. The number of cases with a previous history of chronic epilepsy may be falling and less than 5 per cent of people with epilepsy ever experience generalized convulsive status epilepticus. In this group, discontinuation of anti-epileptic drugs is by far the most common provocative factor. Other causes of status in those without epilepsy include drug abuse particularly of alcohol, metabolic disorders, cerebral hypoxia, stroke, trauma, tumour, and neurological infection. The outcome is largely determined by the aetiology and by the speed at which effective treatment is instituted.

At onset of generalized convulsive status epilepticus, typical and relatively discrete tonic-clonic seizures may be observed. However, with progression, convulsive activity commonly becomes less marked and may eventually be confined to myoclonic jerks of the head, eyes, and face. At this stage, the usual EEG characteristics may be lost and more periodic activity is seen (Treiman et al. 1990). This has been called subtle convulsive status and may represent a terminal state. It is most frequently described in series that include large numbers of cases of status following cardiorespiratory arrest during anoxic coma, and it remains somewhat controversial whether it really represents an anoxic rather than a seizure-related state.

Early in generalized convulsive status epilepticus there is immediate catecholamine release resulting in tachycardia, hypertension, and hyperglycaemia. Pulmonary oedema may occur as may a combined respiratory and metabolic acidosis. Hyperpyrexia commonly occurs and appears important in determining poor outcomes. This, combined with blood and CSF leucocytosis, can suggest the presence of infection when they are, in fact, the result of the status. Late in status epilepticus the blood pressure and heart rate begin to fall, hypoglycaemia may occur, and renal failure may be caused by rhabdomyolysis.

Few conditions should be confused with generalized convulsive status epilepticus though it is not unusual for recurrent pseudoseizures to be managed by the inexperienced as true status epilepticus (Howell et al. 1989). Acute encephalopathies due to hypoxia, uraemia or other metabolic causes, which are associated with multifocal myoclonus may also on occasions be confused with status epilepticus.

Complex partial status epilepticus is of uncertain incidence, largely because it is under-diagnosed (Shorvon 1994). Probably one-third of cases occur without a previous history of epilepsy and often start with one or two discrete tonic-clonic seizures. It may well be that the condition is more common in subjects with frontal than temporal lobe epilepsy. The typical picture is of a twilight state with varying degrees of confusion. Typical automatisms of complex partial seizures may or may not be seen. The EEG is crucial for a diagnosis but a variety of changes occur from localized seizure discharges through periodic lateralized epileptiform discharges to generalized patterns of disturbance. Episodes vary in length, some examples lasting months are documented (Shorvon 1994).

The differential diagnosis of complex partial status epilepticus is exceptionally wide and includes a variety of psychiatric disorders as well as any unexplained confusional state or alteration of consciousness. There may be surprisingly little morbidity to even prolonged episodes of complex partial status epilepticus and there are very few case reports of significant memory impairment following such episodes.

Simple partial status epilepticus is probably even rarer and commonly takes a somatomotor form. In adults, stroke, tumour, and non-ketotic hypoglycaemia are probably the most common causes.

A systematic classification of epilepsy syndromes has been proposed as have modifications of these (Commission on classification and terminology of the International League Against Epilepsy 1989a; Engel 1998; Soteriades 2002). Most epilepsy syndromes are age-related and occur in childhood, whilst the majority of the epilepsies of adult life are symptomatic partial or focal epilepsies. For this reason the classification has less relevance to adults, in whom accurate localization of the onset of partial seizures and determination of their aetiology is of greater importance. The classification has been subject to considerable criticism as many epidemiological studies fail to classify syndromes in the majority of patients with epilepsy. Some reference does, however, need to be made to those specific epilepsy syndromes where seizures have a predilection to continue into adult life.

These are a relatively well-defined group of syndromes with age-specific onsets and generalized spike wave abnormality in the EEG. They are primarily genetic though the mode of inheritance is likely to be complex and heterogeneous. They account for at least 15–20 per cent of human epilepsy (Jallon and Latour 2005). Often they are of childhood onset but may persist into adulthood (Sections 30.4 and 30.5).

This is discussed in Section 30.4.4. In those with absence persisting into adult life there may be features suggesting alternative diagnoses such as eye-lid or peri-oral myoclonus with absence (Panayiotopoulos et al. 1992; 1994; 1995), or that the absences were part of a juvenile myoclonic or absence epilepsy. When tonic-clonic seizures occur in subjects with this syndrome in adult life, they occur without an aura and have a tendency to occur within 1–2 h of wakening. Sodium valproate is undoubtedly the drug of choice for the treatment of persisting seizures in adults with this syndrome.

This syndrome is less common than childhood absence epilepsy and may account for 20 per cent of patients with typical absence. Absence seizures begin between 7 and 16 but tend to be much less frequent than those of childhood absence epilepsy, while a greater proportion, 80 per cent, have tonic-clonic seizures. Seventy- five per cent of these latter seizures occur on awakening. Myoclonic seizures can also occur in this syndrome, perhaps in as many as 16 per cent of patients (Wolf and Inoue 1984). The EEG shows a high incidence of poly-spike and wave and photosensitivity. Valproate probably remains the drug of choice with remission in 85 per cent of patients In distinction from childhood absence epilepsy, the syndrome is probably lifelong (Bouma et al. 1996).

A number of other described absence epilepsy syndromes exist that present in childhood, but are likely to persist into adult life. These are discussed elsewhere (Section 30.4.4).

This is a common epilepsy syndrome, which was first comprehensively described by Janz and Christian (1957). It is a genetic disorder and up to 25 per cent of patients have a family history. In some pedigrees the gene encoding for this disorder has been shown by linkage analysis to lie on the short arm of the 6th chromosome (Bai et al. 2002), while in others linkage to chromosome 15 has been found (Taske et al. 2002) suggesting that genetic heterogeneity exists.

It accounts for approximately 5–10 per cent of the epilepsies. Seizures can begin between 8 and 26, but 80 per cent commence between the ages of 12 and 18 (Section 30.4.5). Myoclonic jerks are usually symmetric and mostly affect the upper limbs. On occasion lateralized myoclonus can occur mimicking focal motor seizures. Myoclonus can on occasions be associated with absence or typical absence seizures can occur independently. Seizures most commonly occur after wakening but there may also be a second peak in seizure susceptibility during the evening. Sleep deprivation appears to be a potent provocative factor.

Ninety per cent of patients have generalized tonic-clonic seizures and 10 per cent of patients have absence seizures. It is usually the tonic-clonic seizures that precipitate medical referral and patients may not recognize the important association with jerks which may have preceded the first tonic-clonic seizure by some time. In identifying this syndrome it is important to ask specifically for a history of myoclonus.

The EEG typically shows polyspike and spike wave activity at a more rapid rate than the classic 3 cycle per second spike wave. Photosensitivity is extremely common and is usually identified in about 30 per cent of patients, though this phenomenon may frequently be blocked by treatment with valproate (Goosses 1984). Valproate is the drug of choice in this syndrome. Remission may be uncommon if patients are treated with other anti-epileptic drugs (Delgado-Escueta and Enrile-Bacsal 1984) and carbamazepine and vigabatrin may exacerbate myoclonus. Remissions of this syndrome are drug-dependant and 90 per cent of patients with it will relapse if drugs are withdrawn.

Tonic-clonic, or clonic-tonic-clonic, seizures predominate in this syndrome although the presence of occasional myoclonus or absence does not preclude the diagnosis. Janz (1962) examined the timing of tonic-clonic seizures in 2825 patients. Thirty-three per cent had tonic-clonic seizures on awakening compared to 44 per cent occurring during sleep and 23 per cent occurring at random. There was a strong association between tonic-clonic seizures occurring on wakening and generalized spike-wave activity in the EEG. The syndrome has the widest range of age at onset of the idiopathic generalized epilepsies, from 5 to 30 years, with peak onset at 17–20 years.

Precipitating factors include sleep deprivation, sudden arousal, and alcohol intake but catamenial exacerbations are also prominent. Avoidance of precipitating factors is important in management and valproate is the drug of choice for this syndrome.

These childhood epilepsies frequently carry a malign prognosis and are associated with a significant mortality (Section 30.5). However, many children with such epilepsies will survive into the adult age range. They frequently exhibit multiple handicaps. In adult life there is often a change in the nature of seizures with myoclonic, tonic, atonic, and atypical absence seizures becoming less frequent. More typical simple partial and complex partial seizures become evident. The characteristic of these seizures is often that they appear multi-focal from a clinical and electroencephalographic point of view. Whilst such patients may continue to have a severe epilepsy during adult life, seizure frequency tends to be less than in childhood.

Some idiopathic age-related syndromes, such as benign Rolandic epilepsy, are not seen in adult life as they have an early age of onset, a uniformly excellent prognosis, and do not continue into adult life. However, the range of genetically determined partial epilepsies seen in adults is increasing.

Autosomal dominant nocturnal frontal lobe epilepsy has been described in a number of families from Australia, United Kingdom, and Canada (Scheffer et al. 1995). Many patients previously diagnosed as having paroxysmal nocturnal dystonia probably have this seizure disorder. Seizures begin in childhood and persist into adult life. They occur during sleep, often in clusters and are typical frontal lobe seizures that are brief, associated with vocalization, and include varieties of motor activity including thrashing, tonic stiffening, and clonic jerks. Occasional secondarily generalized tonic-clonic seizures occur. The interictal EEG is usually normal as can be ictal EEG recording. Most patients are responsive to carbamazepine. Some families show linkage to markers on chromosome 20q. A mis-sense mutation has been demonstrated in the gene encoding the α-4 sub-unit of the nicotinic acetylcholine receptor (Steinlein et al. 1995). Other families do not show this linkage and there may be genetic heterogeneity.

Other genetically determined partial epilepsies probably occur in adult life. Berkovic et al. (1994) described 19 monozygotic twins concordant for a temporal lobe epilepsy with a very benign course. Similar cases have been reported in non-twin families. The age of onset was anywhere between 10 and 60 years. Seizures tended to be brief simple or complex partial temporal lobe seizures, which were very easily controlled with carbamazepine or phenytoin. Ottman et al. reported a family with characteristic temporal lobe seizures associated with an auditory aura. There was evidence of linkage to chromosome 10q with mutations in LGI1, the function of which is unclear (Ottman et al. 2004).

The syndrome of mesial temporal lobe epilepsy remains best described as a cryptogenic epilepsy in spite of increasing understanding of its aetiological and pathological mechanisms. There is no good epidemiological data about its incidence and prevalence, but information from surgical series would suggest that it probably is a major contributor to the total number of cases of temporal lobe epilepsy. Up to two-thirds of cases have a history of prolonged febrile convulsions in childhood (Williamson et al. 1993) and in others there may be a history of trauma or infection. A case-controlled study has suggested that complicated febrile seizures during childhood may be the aetiological factor in up to 20 per cent of all patients of complex partial seizures (Rocca et al. 1987). The pathology shows neuronal loss and gliosis in the hippocampus, known as hippocampal sclerosis, associated with synaptic reorganization resulting in functional hypersynchronization and hyperexcitability. This pathological change can be seen following seizures in experimental animals (Sutula et al. 1988) and on occasions is also seen in man associated with hamartomas and cortical dysplasias.

The most striking clinical feature of mesial temporal lobe epilepsy is its progressive nature. Typical temporal lobe seizures may be noted in as early as the first decade of life. However, when they begin they are often simple partial seizures that are brief, and may continue for years before the diagnosis is made. When recognized in childhood, seizures are often suppressed only to return in adolescence or early adult life (French et al. 1993). By this time, the seizures are more severe and commonly are typical temporal lobe complex partial seizures and some episodes of secondary generalization start to appear. Ninety per cent of subjects describe an aura, most commonly visceral sensations, olfactory hallucination, and memory disturbance. Subsequently there is often a progression to complex partial seizures beginning with motor arrest and staring, oro-alimentary automatisms, and fumbling and picking automatisms of the hands. Versive symptoms may occur, either early or late in the seizures. There is also evidence of a progressive disturbance of cognitive function, particularly memory (Elger et al. 2004). which can be associated with increased severity of MRI changes with the duration of disease (Salmenpera et al. 1998).

Investigations may show a normal inter-ictal EEG or alternatively characteristic anterior temporal sharp waves which are bilaterally independent in up to a third of patients. High resolution T1-weighted MR imaging shows hippocampal atrophy that is most commonly unilateral but can be bilateral (Fig. 31.9). T2-weighted MRI and FLAIR sequences show high signal in the affected hippocampus.

Whilst anti-epileptic drug treatment reduces or abolishes secondarily generalized tonic-clonic seizures, the complex partial seizures tend to be refractory and many patients exhibit considerable psychosocial disadvantage. Surgical treatment will usually abolish disabling seizures in about 60–70 per cent of subjects (Section 31.11.2). For this reason, it is important that the syndrome is recognized early so that the impact of the disorder can be minimized.

This syndrome is fortunately rare and although usually seen in children (Section 30.5.4), it can present in adults. It comprises epilepsia partialis continua, slowly progressive hemiplegia, and intellectual impairment with progressive atrophy of one cerebral hemisphere. Histological examination reveals a low grade inflammatory response (Rasmussen et al. 1958). While a variety of viruses have been suggested as an aetiological agent, there are no consistent findings in this area. Recently some cases have demonstrated antibodies to GluR3, glutamate receptor, and immunosuppression can be helpful (Leach et al. 1999). Many cases require hemispherectomy for seizure control (Section 31.11).

A number of factors may appear to precipitate seizures in susceptible individuals. These may be classified as either specific sensory stimuli or actions, causing reflexly induced seizures, or non-specific precipitants.

Whilst the term ‘reflex epilepsy’ has been widely applied, two-thirds of patients who have reflexly induced seizures also have apparently spontaneous seizures occurring at other times.

Photosensitive epilepsy is the most common reflex epilepsy, accounting for 0.5–8 per cent of patients with epilepsy, but a much greater proportion of those with idiopathic generalized epilepsy (Jallon and Latour 2005). The crudest visual stimulus to evoke seizures is flicker or flash, a factor that is made use of in the routine recording of most EEGs. This form of sensitivity is most common in childhood and juvenile absence epilepsy and juvenile myoclonic epilepsy and some forms of progressive myoclonic epilepsy. It is a much rarer accompaniment of symptomatic generalized epilepsies and is seen rarely in partial epilepsy. Flicker stimuli may be produced in the environment by television and video games, stroboscopic illumination or sunlight passing through trees or railings or other regularly spaced objects. Most individuals are maximally sensitive between 15 and 20 Hz. Maximum sensitivity often occurs just after the eyes are closed or when the eyes are open. Stimulation of one eye only reduces sensitivity. The greater the proportion of the visual field taken up by stimuli and the greater the luminance then the greater the potential for photosensitivity. In many individuals patterned flash is a potent stimulus.

The electrophysiological correlate of photosensitivity is the photoconvulsive response. This is most commonly seen in females during adolescence and may disappear in adult life. A photoconvulsive response consists of bilaterally synchronous spike-wave activity, which persists for a second or more after the cessation of the flash stimulus. It must be clearly differentiated from photic following and photomyoclonic responses, which have no significant association with photically induced seizures.

About a third of flash-sensitive patients exhibit sensitivity to patterns, the most potent of which are strong stripes, and 70 per cent do so if the pattern oscillates (Wilkins and Lindsay 1985). The most important practical implication of pattern sensitivity is television epilepsy. A television picture is created by variations in the brightness of a spot that scans the screen repeatedly from left to right. The pattern that is generated in this way is similar to a vibrating pattern, which is a very potent stimulus to pattern-sensitive individuals.

Seizures can be prevented in susceptible individuals by maintaining a satisfactory distance from the television set and using the remote control to adjust the picture. More complex preventative methods involve viewing the screen through polarized spectacles so as to produce only monocular stimulation (Wilkins and Lindsay 1985). Most anti-epileptic drugs block photic sensitivity. Valproate is the treatment of choice.

Particular interest has been aroused by the occurrence of seizures with computer games (Fish et al. 1994). The incidence has been estimated as approximately 1.5 per 100 000 of the population between 7 and 19 years. Most subjects show sensitivity to flicker or pattern, but in some cases it may be absent, suggesting a contribution from more complex visual or cognitive stimuli.

Primary reading epilepsy can be viewed as a form of visually induced epilepsy. The characteristic seizures in this disorder are myoclonic jerks of the jaw, which may proceed to tonic-clonic seizures. Both focal and paroxysmal EEG abnormalities have been described in this condition (Wilkins and Lindsay 1985). A number of different mechanisms may be involved in producing seizures with reading. In some patients, the lines of print may act as patterns, in others eye movements may provoke the seizures. In some patients neither pattern sensitivity nor eye movements appear important and in these comprehension of the written material may be the important provocative factor.

Sudden noise or other startle may give rise to seizures, particularly in mentally handicapped patients (Anderman and Andermann 1986). More complex auditory stimuli can also provoke seizures in  musicogenic epilepsy (Poskanzer et al. 1962), where seizures are usually complex partial in type. A variety of other complex reflex epilepsies have been described including eating epilepsy which gives rise to complex partial seizures and writing epilepsy producing jerking of the writing hand. Other cognitive functions such as arithmetic and listening to spoken language can evoke seizures in rare cases.

Touch or muscle stretch may occasionally provoke seizures although more typically evoking myoclonic jerking in patients with progressive myoclonic or post-hypoxic myoclonus. Immersion in hot or cold water can act as seizure precipitant, particularly in the Indian sub-continent (Gururaj and Satishchandra 1992). Rarely seizures may be provoked by movement though this is much less common than paroxysmal choreathetosis.

Sleep epilepsy. The sleep–waking cycle can have a profound influence on the occurrence of seizures in susceptible individuals. Many patients have tonic-clonic seizures only during sleep (Gibberd and Bateson 1974; D’Alessandro et al. 2004). A number of epilepsy syndromes show a predilection for seizures during non-rapid eye movement sleep including the idiopathic partial epilepsies, electrical status epilepticus during slow wave sleep, and temporal and frontal lobe epilepsies, with greater likelihood of secondary generalization (Bazil and Walczak 1997). Sleep may enhance focal epileptogenic discharges and tonic-clonic seizures limited to sleep in the adult should usually be regarded as having a partial onset until proved otherwise. When a pattern of only sleep seizures has been established, the risk of waking seizures is low at 13 per cent over 6 years (D’Alessandro et al. 2004).

Awakening seizures. Seizures occurring shortly after wakening are common in the idiopathic generalized epilepsies: juvenile myoclonic epilepsy, tonic-clonic seizures on awakening, and childhood and juvenile absence epilepsy. These epilepsies are particularly sensitive to sleep deprivation or to sudden rousing from deep sleep. The symptomatic generalized epilepsies are characterized by occurrence of seizures that are independent of the sleep–waking cycle.

Catamenial epilepsy. Many women with epilepsy are subject to catamenial exacerbation of seizures although it is uncommon to see women who only have seizures corresponding to such a pattern. The time of greatest susceptibility seems to be in the few days preceding the onset of menstruation. There is evidence of oestrogens being potentially epileptogenic and progestogens potentially anticonvulsant (Scharfman and MacLusky 2006). In spite of this, regulation of periods using oral contraceptive preparations has not shown any benefit in suppressing seizures. Where periods are regular the prescription of a benzodiazepine such as clobazam for a number of days around the period of maximum risk can be beneficial (Feeley et al. 1982). The effects of pregnancy on seizure control in women with epilepsy are unpredictable. In some individuals seizures may increase in frequency, in some they may decrease but in the majority there is no significant change in seizures frequency.

The great majority of seizures and epilepsies developing in adult life will be regarded as symptomatic, a view reinforced by the advent of modern MRI. It may be useful to divide causes of seizures and epilepsy in adult life into:

Some aetiologies such as head injury, stroke, and intracranial infections may cause both acute symptomatic seizures and remote symptomatic epilepsy. The presence of the one is not necessarily associated with the other. Sander et al. (1990) found that the commonest remote symptomatic causes of epilepsy were vascular disease in 15 per cent and tumour in 6 per cent. Remote symptomatic epilepsy was commonest in the elderly where vascular disease accounted for 49 per cent of cases. Tumour causes only 1 per cent of epilepsy below 30 years of age, but accounts for 19 per cent of cases between 50 and 59 years. Trauma caused 3 per cent of cases, infection 2 per cent. Acute symptomatic seizures occurred in 15 per cent, and alcohol was the commonest single cause at 6 per cent, its highest incidence of 27 per cent occurring between 30 and 39 years of age.

Acute symptomatic seizures occur commonly: of 1758 patients admitted to an intensive care unit, 217 exhibited neurological complications, 61 of whom had seizures (Bleck et al. 1993). When they occur as a result of systemic disorders they are associated with an acute encephalopathy and most commonly seizures are tonic-clonic. Patients may also exhibit tremor, asterixis, and multifocal myoclonus. The risk of tonic-clonic seizures is more determined by the rate of a metabolic change than the absolute levels reached. They are particularly common in the elderly, where they may account for as much as 77 per cent of the incidence of seizures (Loiseau et al. 1990). Focal seizures can be seen in association with acute cerebral insults when they are usually recognized as focal motor seizures, sometimes as epilepsia partialis continua. It is suggested that complex partial seizures rarely occur as acute symptomatic seizures but such seizures could be difficult to recognize when accompanied by an acute confusional state or coma.

Whilst acute symptomatic seizures may be suppressed by short-acting anti-epileptic drugs such as benzodiazepines they do not usually require longer term anti-epileptic drug treatment. Acute symptomatic seizures are often resistant to benzodiazepines and other anti-epileptic drugs requiring correction of the underlying metabolic abnormality to suppress seizures.

Hypernatraemia may occur in gastroenteritis, fever, sweating, burns, and diabetes, and due to gross fluid restriction or excessive salt intake. It is usually defined as a serum sodium concentration above 145 mEq/l. Much higher concentrations can be tolerated in chronic hypernatraemia, with seizures occurring more commonly during the phase of correction. Altered consciousness is common and focal or generalized tonic-clonic seizures occur most commonly in patients who are also uraemic or acidotic or in whom there is acute elevation towards 160 mEq/l (Riggs 2002).

Hyponatraemia is more common than hypernatraemia, occurring with congestive cardiac failure, liver disease, nephrotic syndrome, water-overload, diuretic misuse, as well as with renal disease and syndromes of inappropriate antidiuretic hormone secretion which can result from neurological disorders. Tonic-clonic seizures occur and occasionally status epilepticus. Again the rate of change in serum sodium concentration is of prime importance, with encephalopathy and seizures seen with rapid falls to 115 mEq/l or less (Adrogue and Madias 2000). Hyponatraemia is associated with a high mortality and too rapid a correction by the over-enthusiastic use of hypertonic saline has been associated with the occurrence of central pontine myelinolysis (Section 37.6).

Hypocalcaemia may be seen in hypoparathyroidism, vitamin D deficiency, acute pancreatitis, and pseudohypoparathyroidism. Up to 70 per cent of patients with hypoparathyroidism may have seizures (Messing and Simon 1986) associated with tetany, altered consciousness, and abnormal behaviour and dyskinesia. Both tonic-clonic and focal motor seizures are described.

Hypercalcaemia most commonly occurs in disseminated malignant disease and hyperparathyroidism. It results in weakness, drowsiness, and confusion, but seizures are uncommon. When they occur they may be generalized or focal motor seizures (Riggs 2002).

Hypomagnesaemia may be seen in inflammatory bowel disease, bowel resection, and other malabsorption syndromes, and is often associated with other electrolyte disturbance. Neurological syndromes may be seen with levels below 1.0 mmol/l. The clinical state, with startle, tremor, and myoclonus and Chvostek’s sign, may be indistinguishable from hypocalcaemia, although tetany may be less common. Hypomagnesaemia should be considered if seizures continue in a treated hypocalcaemic patient.

Seizures seem particularly common in association with hyperosmolar non-ketotic hyperglycamia (Venna and Sabin 1981). They may occur in up to a quarter of such patients and simple partial motor seizures account for approximately 80 per cent of such seizures. Epilepsy partialis continua can occur. Such seizures may be very resistant to anti-epileptic drug treatment but seem to respond rapidly to the correction of the hyperglycaemia. By contrast, seizures seem to be very rare in ketoacidotic diabetic coma (Messing and Simon 1986).

This is usually seen in diabetic patients using insulin or hypoglycaemic drugs. It occurs more rarely with insulinoma, other neoplasms, or severe liver disease. Seizures, usually tonic-clonic seizures without an aura, may occur in 7 per cent of patients (Malouf and Brust 1985).

Seizures appear extremely uncommon in hyperthyroidism. They do occur occasionally as a feature of Hashimoto’s encephalopathy (Section 38.3.1) that can precede other manifestations of thyrotoxicosis. Myoclonic seizures may occur along with tonic-clonic seizures (Ghika-Schmid et al. 1996). Seizures seem more common in patients with hypothyroidism and may occur in as many as a quarter of patients with myxoedema coma (Jellinek 1962). Patients do not seem to be at risk of continued seizures after the underlying thyroid abnormality has been corrected.

Seizures may occur in approximately 15 per cent of patients during episodes of acute intermittent porphyria and may be a presenting feature (Section 21.8.6). Drug control of seizures can be a significant management problem as phenytoin, barbiturates, and carbamazepine can all induce attacks of porphyria. Benzodiazepines may be used with caution but it has also been suggested that magnesium sulphate may also be effective in controlling seizures. Of the newer anti-epileptic drugs, pregabalin and levetiracetam may be safe.

Seizures are a feature of acute hepatic failure but not of chronic hepatic dysfunction. Seizures may be focal but are more commonly tonic-clonic seizures often preceded by multifocal myoclonus. Neurological features of encephalopathy and seizures are common following liver transplantation (Saner et al. 2006).

Acute uraemic encephalopathy commonly presents with motor excitability including tremors, asterixis, multifocal myoclonus, chorea, and dystonia. Convulsions occur in as many as a third of patients; most are tonic-clonic seizures but focal motor seizures and epilepsia partialis continua can occur (De Deyn et al. 1992). Seizures have also been reported during dialysis, in the dialysis disequilibrium syndrome, and as part of dialysis encephalopathy, a sub-acute progressive disorder in which speech disorders, dementia, and myoclonus are prominent.

Drugs, and particularly alcohol, are a common cause of seizures (Messing et al. 1984) and many different drugs have been associated with seizures (Table 31.7). The Boston collaborative drugs surveillance programme (1972) reported 26 cases of drug-induced convulsions in approximately 33 000 in-patients, an incidence of 0.08 per cent. Most commonly implicated were penicillin, hypoglycaemic drugs, lignocaine, and psychotropic agents. Messing et al. (1984) reviewed case records of over 3000 patients presenting with seizures and found that they were drug-related in 1.7 per cent, the most common drugs involved being isoniazid, psychotropic drugs, bronchodilators, hypoglycaemic agents, lignocaine, and penicillin. The majority of seizures are tonic-clonic but whilst most begin without an aura, there was a simple partial motor onset in nine patients.

Seizures may be provoked in two ways: there may be specific neural excitatory effects for some drugs or alternatively there may be non-specific effects resulting from high doses of drugs often administered during self-poisoning. Most drug-induced seizures are dose-related and particular care must be exercised when drugs are administered parenterally or intrathecally. Patients with renal or hepatic failure may be at risk because of inability to metabolize potentially convulsant drugs and individuals with a previous history of epilepsy or pre-existing brain disease may be particularly at risk. The subject has been recently reviewed (Ruffmann et al. 2006).

Penicillin has potent epileptogenicity in animals and has been widely used as a model for both focal and generalized epilepsies. It acts as a GABA antagonist and may also bind to benzodiazepine receptors (Curtis et al. 1972; Antoniadis et al. 1980). Benzylpenicillin is probably the most potent antibiotic in causing seizures but ampicillin and cephalosporins carry some risk. Newer quinalone antibiotics may similarly interfere with GABAergic mechanisms. Isoniazid may cause seizures because of its action in antagonizing pyridoxine, a co-enzyme required for the synthesis of GABA (Blakemore 1980).

Tricyclic antidepressants are particularly likely to cause myoclonus and convulsions when taken in overdose but seizures may occur in up to 1 per cent of patients taking therapeutic dosages (Lowry and Dunner 1980). Amitriptyline is probably the antidepressant with the highest risk, but monoamine oxidase inhibitors and selective serotonin uptake inhibitors appear relatively safe (Dailey and Naritoku 1996). Antipsychotic usage may be complicated by seizures. Phenothiazines, particularly chlorpromazine, are associated with a 1–2 per cent incidence of seizures (Logothetis 1967). Of the atypical antipsychotics, clozapine may have a high risk of seizures (Pacia and Devinsky 1994) but pimozide and sulpiride seem relatively safe. Lithium toxicity may be associated with seizures (Section 5.5.3).

Most central nervous system stimulant drugs are capable of causing seizures and problems most commonly arise with theophylline and its derivatives which significantly lower seizure threshold possibly by elevating cyclic GMP levels in brain (Walker 1981).

Cocaine, amphetamines, narcotics, and phencyclidine have been associated with seizures (Holland et al. 1993) (Section 5.3). They have stimulant actions on the central nervous system and lower seizure threshold. Cocaine presents the most frequent problem with tonic-clonic seizures occurring in 5 per cent of emergency department attendances with acute toxicity (Pascual-Leone et al. 1990). Organic solvents have been reported to cause epilepsy (Jacobsen et al. 1994). It should be noted that seizures are rarely seen with narcotic abuse.

Seizures can occur in up to 20 per cent of children undergoing renal transplantation (McEnery et al. 1989), but only 4 per cent of liver transplants (Wijdicks et al. 1996). Cyclosporin and some other immunosuppressant drugs give rise to seizures (Section 5.6.3). Inevitably, drug effects may interact with other factors such as metabolic disturbance and infection. Cyclosporin-induced seizures may occur in 1–2 per cent of renal transplants and 5 per cent of bone marrow transplants (Wijdicks et al. 1995).

The withdrawal of chronically administered sedative drugs which show tolerance is a well-recognized cause of seizures and may occur with alcohol, barbiturates, and benzodiazepines. The best studied of withdrawal seizures are those that occur with alcohol and which may be a part of the delirium tremens syndrome (Section 5.2.1). The abuse of alcohol is an important cause of seizures in the community and it must be considered in adults developing tonic-clonic seizures for the first time (Hillbom 1980). The risk is clearly related to the dose of alcohol consumed (Lechtenberg and Worner 1992). It seems that abrupt, absolute, or relative withdrawal of alcohol is most commonly responsible for causing seizures. Clustering of seizures occurs between 7 and 48 h after the withdrawal of alcohol. Sixty per cent of patients have more than 1 seizure, but status epilepticus occurs in less than 5 per cent of patients. During the withdrawal period photo-myoclonic and photo-convulsive responses may be seen in the EEG. On occasions seizures can occur in patients whilst intoxicated with alcohol.

Non-compliance with anti-epileptic drug medication is well recognized to be a common cause of seizures in people with epilepsy and is an important cause of status epilepticus (Aminoff and Simon 1980). Abuse of barbiturates and subsequent withdrawal in non-epileptics has been an important cause of seizures. This effect is dose-related and seizures occur with other withdrawal symptoms such as insomnia, tremor, anorexia, and autonomic over-activity. The EEG of patients undergoing barbiturate withdrawal shows features of both photo-myoclonic and photo-convulsant responses. Benzodiazepines can also be associated with withdrawal seizures.

It is well recognized that a number of cerebral insults and pathologies predispose to the development of epilepsy.

A static encephalopathy, manifesting as learning disability and motor handicap present from birth, is commonly associated with seizure disorders. As many as 50 per cent of individuals with both mental handicap and cerebral palsy have seizures (Hauser et al. 1987). Some but by no means all may have been caused by cerebral hypoxia. The more severe the mental and physical handicap the higher the risk of epilepsy (Edebol-Tysk 1989). The great majority will develop seizures early in life but up to 15 per cent of patients may have a seizure disorder that starts after the age of 15 (Forsgren et al. 1990). Whereas in childhood this symptomatic epilepsy commonly involves generalized seizures including myoclonus, tonic and atonic seizures, and infantile spasms, there is an increasing predominance of partial seizures and secondary generalized tonic-clonic seizures as individuals mature. In this population the outcome for epilepsy is poor, only a third achieving seizure remissions of a year or more and a third having at least one seizure per month. Early brain damage is one of the strongest factors predicting a poor outcome of epilepsy.

Generalized cerebral hypoxia during adult life seems much less likely to result in seizures. Acute hypoxia, when severe, is commonly associated with convulsions and multifocal myoclonus (Wardrope et al. 1991). In patients with post-hypoxic coma following an anoxic insult, seizures seem to be much rarer and when they occur they may be associated with an adverse prognosis (Bates et al. 1977).

The relationship between head injury, acute symptomatic seizures occurring within the first week of injury, and late post-traumatic epilepsy represents perhaps the best studied of all causes of epilepsy. Head injury is a well-known cause of epilepsy and most patients developing epilepsy will recall the occurrence of a minor head injury sometime before the development of seizures. However, it is clear that only specific types of head injury carry a significant risk of post-traumatic epilepsy.

Although perhaps 2 per cent of all concussive head injuries result in epilepsy (Annegers et al. 1980), head trauma was the cause of seizures in 3 per cent of patients registered in the National General Practitioners Survey of Epilepsy (Sander et al. 1990). There can be no doubt that the more severe the head injury, the higher the risk of post-traumatic epilepsy (Annegers et al. 1975b, 1998).

Brain injuries caused by missiles provide a well-defined and relatively homogenous group of injuries that are fortunately rare in civilian life. They have been fully studied in cohorts of patients from the First World War through to the Vietnam War. Overall, it would seem that 50 per cent of patients surviving such injuries will develop post-traumatic epilepsy and that the relative risk of developing epilepsy will initially be 580 times higher than a general age-matched population during the first year, falling to 25 times higher after 10 years (Salazar et al. 1985). The risk of epilepsy correlates with the severity of the injury, the amount of tissue loss, and the severity of neurological impairment. Infection and abscess formation probably exert an important influence on the incidence of epilepsy. Seizures during the first year seems to predict the duration and frequency of subsequent epilepsy.

The most satisfactory unselected population of patients developing post-traumatic epilepsy has been studied by Annegers et al. (1998). The relationship of the severity of head injury to subsequent risk of epilepsy over 15–20 years was:

Subdural haematoma and brain contusion were the factors that greatly increased this risk. Within the neurosurgical population studied by Jennett (1975), an increased risk of epilepsy was determined by the presence of a compound depressed fracture, intracranial haemorrhage, or the occurrence of early acute symptomatic seizures within the first post-traumatic week. In those few subjects with all three risk factors, the likelihood of post-traumatic epilepsy could be as high as 50–80 per cent.

Annegers et al. have further clarified the importance of early, as opposed to immediate, seizures in predicting late post-traumatic epilepsy by showing that early seizures have no independent effect on the risk of late post-traumatic epilepsy, but merely act as a marker of injuries of sufficient severity to cause late epilepsy.

Jennett recognized that seizures occurring immediately on impact do not seem to carry any subsequent excess risk of epilepsy and indeed characterize relatively mild concussive injuries. McCrory et al. (1997) studied such immediate ‘concussive convulsions’ in Australian sportsmen and came to the same conclusions. They questioned whether these events are seizures rather than a form of acute temporary decerebration.

A systematic review of 10 randomized controlled trials showed there was consistent evidence that the early treatment with anti-epileptic drugs, most commonly phenytoin but also carbamazepine and phenobarbitone, resulted in a relative risk for early seizures of 0.34 (95 per cent CI 0.21-–0.54). Treatment of 100 patients would result in 10 patients avoiding seizures during the first week, who would otherwise have had them. However, the studies did not provide evidence that this benefit was accompanied by any reduction in mortality and there is no evidence that a reduction in early seizures is accompanied by a reduction in late post-traumatic epilepsy (Schierhout and Roberts 1998). Current evidence from intervention studies and community-based epidemiology would indicate that early seizures act simply as a marker for more severe head injuries rather than an important part of a pathogenic process that leads from severe head injury to either late post-traumatic epilepsy or death. More recently, studies of the potential effects of valproate prophylaxis have been undertaken (Temkin et al. 1999). Slightly under 15 per cent of patients had late seizures when treated with phenytoin compared to approximately 20 per cent of patients receiving valproate treatment, a relative risk of 1.4 (95 per cent CI 0.8–2.4) and the study was stopped because of excess deaths in the valproate-treated group. Thus, there is reasonable evidence that immediate administration of anti-epileptic drugs reduces the incidence of early seizures, and that phenytoin has the best evidence to support its use in this role. This effect will, however, carry with it a small risk of acute idiosyncratic reactions, as long as phenytoin is only used during the first post-traumatic week. There seems no justification for the routine use of anti-epileptic drugs in patients with severe head injuries.

Craniotomy can be viewed as a type of head injury, although the pathology underlying its necessity contributes significantly to risk. The overall incidence of seizures occurring after supratentorial craniotomy has been estimated at 17 per cent during a follow-up period of at least 5 years (Foy et al. 1981a). The incidence varied from 3 to 92 per cent depending on the condition for which the craniotomy was undertaken.

Approximately one-fifth of patients undergoing aneurysm surgery develop post-operative seizures (North et al. 1983). The incidence varies according to the site of the aneurysm; 7.5 per cent from internal carotid aneurysm, 21 per cent from anterior communicating aneurysm, and 39 per cent for a middle cerebral artery aneurysm (Foy et al. 1981a). Additional factors influencing the incidence may be the presence of an intracerebral haematoma, cortical damage, splitting the silvian fissure, cerebral swelling and peri-operative aneurysmal rupture, and the length of surgery (Foy et al. 1981b). That surgery is responsible for at least part of the risk of epilepsy associated with aneurysm is suggested by the halving of risk in conservatively managed survivors following aneurysmal subarachnoid haemorrhage and in those treated by endovascular coiling (Molyneux et al. 2005). Arteriovenous malformations and spontaneous intracerebral haematoma from other causes carry risks of epilepsy of 50 and 20 per cent respectively and surgical treatment does seem to be an additional risk factor for these conditions (Crawford et al. 1986). Cavernous angiomas probably represent 10–20 per cent of all central nervous system vascular malformations, and seizures are the only symptom in up to 70 per cent (Farmer et al. 1988). The MR changes, with high T2 signal core and halo of low signal due to haemosiderin, suggests that many may bleed asymptomatically. Here surgery seems highly effective in seizure control.

The incidence of new seizures following meningioma surgery is of the order of 20 per cent (North et al. 1983). The incidence is higher for parasagittal lesions than for convexity or basal tumours. Some 44 per cent of patients who have pre-operative seizures do not have any further seizures post-operatively. Surgery for supratentorial abscess carries a very high risk. With sufficiently long follow-up virtually all patients develop epilepsy. Ventricular shunting procedures can be associated with a 24 per cent risk of seizures and multiple shunt revisions and shunt infections significantly increase the risks (Copeland et al. 1982).

Thirty-seven per cent of all patients who experience post-operative seizures do so within the first week and 40 per cent of this group continue to have later seizures. By the time 1 year has elapsed 77 per cent of those who will develop seizure disorders will have done so and by 2 years, 92 per cent will have had their first seizure (Foy et al. 1981b).

The possible effects of prophylaxis in high risk patients has been studied by a number of authors (Temkin et al. 1998). There is no evidence that phenytoin or carbamazepine significantly reduce the incidence of post-craniotomy seizures in high-risk groups of patients, nor do they seem to effect the likelihood of persistence of the seizure disorder over a period of time. This use of prophylactic anti-epileptic drugs can be associated with a high incidence for adverse effects, particularly drug-induced rash (Chadwick et al. 1984). For this reason prophylactic treatment does not seem to be justified.

The relationship between intracranial tumours and epilepsy is well established. Consequently there is considerable pressure to image patients presenting with epilepsy. In fact, brain tumours are responsible for late-onset epilepsy in only about 10 per cent of cases. They were found in 6 per cent of patients registered with the National General Practitioner Survey of Epilepsy (Sander et al. 1990). The incidence of tumours rises steeply where seizures are clearly focal in nature (Sumi and Teasdall 1963). Tumours of the frontal, parietal, and occipital lobes seem to carry the highest risk of epilepsy (Mauguiere and Courjon 1978).

In general, about 40 per cent of those with seizures due to tumour have seizures as the first manifestation. The interval between the first seizure, the diagnosis of the tumour, and the development of further neurological problems is commonly prolonged (Smith et al. 1991). This reflects the fact that the majority of tumours that present only with epilepsy tend to be benign. Indeed, presentation with an initial symptom of epilepsy is one of the most powerful prognostic factors indicating a good prognosis (Smith et al. 1991). Such intracerebral tumours are most commonly shown to be low density, non-enhancing lesions on initial CT scans and to be relatively low grade astrocytomas or oligodendrogliomas. The incidence of seizures with cerebral metastases is lower, probably about 20 per cent at the time of presentation (Cohen et al. 1988). The prognosis for tumour-associated epilepsies is poor. Only 11 of 164 patients achieved a 1-year remission of epilepsy with anti-epileptic drug treatment and 50 per cent of patients with tumour epilepsies in adult life die within 4 years. However, 20–30 per cent show prolonged survival (Smith et al. 1991).

There is a considerable dilemma about the management of intracerebral tumours. Meningiomas and other well-defined intracerebral lesions should be treated surgically where this is practical and where the patient is not old or infirm. Seizures will be suppressed in about 40 per cent of patients with meningiomas, and up to 80 per cent of other patients undergoing lesionectomy of other discreet tumours (Cascino et al. 1990). In contrast many neurologists find it difficult to recommend aggressive treatment with biopsy or tumour debulking, and radiotherapy in a patient with a low-grade infiltrative tumour whose only symptom is epilepsy and in whom there is a good prospect of good quality survival for many years. In such cases the tumour is rarely fully resectable and the risk-benefit for early surgical treatment remains uncertain (Karim et al. 1996; Shaw et al. 2002) (Section 27.8.1).

Cerebral tumours are another area in which prophylactic treatment with anti-epileptic drugs has been advocated for those even before the first seizure. Trials have to date failed to show any benefit in preventing the first or later seizures (Weaver et al. 1995; Sirven et al. 2004). The studies available are small and do not exclude clinically important benefits. The striking factor was the low risk of de novo seizures in the studies in treated and untreated patients. Given that anti-epileptic drugs have a considerable propensity to interact with chemotherapeutic agents and steroids, and that tumour-related epilepsies tend to be drug resistant, there seems no justification for this approach.

Cerebrovascular disease and stroke become an increasingly common cause of epilepsy in the later years of life (Loiseau et al. 1990). It has been estimated that cerebrovascular disease may account for 15 per cent of new cases of epilepsy (Sander et al. 1990) and more than 50 per cent of new cases in the elderly (Camilo and Goldstein 2004). A community-based study of stroke showed a 5-year actuarial risk of seizures of 11.5 per cent, over 30 times that expected. The risk was highest after subarachnoid haemorrhage, 30 per cent, and primary intracerebral haematoma, in 25 per cent, while the risk for ischaemic stroke of 9 per cent was restricted to survivors of large anterior circulation stokes. Seizures within 24 h of stroke onset were a risk factor for late seizures, which occurred in 35 per cent of such patients (Burn et al. 1997). Risk factors for late epilepsy include cortical involvement, stroke severity, and haemorrhagic stroke (Camilo and Goldstein 2004). However, asymptomatic carotid occlusion (Cocito et al. 1982) and asymptomatic cerebral infarction (Shorvon et al. 1984) may be found in patients presenting with epilepsy in later life and seizures may also precede a stroke (Burn et al. 1997; Cleary et al. 2004).

Cerebral venous sinus thrombosis is increasingly recognized as an important type of stroke, in younger people (Section 35.15.1). Seizures commonly complicate the acute illness, but the long-term prognosis appears good, with few survivors developing epilepsy. In a series of 77 patients 28 had acute symptomatic seizures, but only four had late seizures and epilepsy, all of whom had acute seizures (Preter et al. 1996).

Arteritic disorders can be accompanied by seizures as part of stroke-like syndromes or acute encephalopathies (Sections 36.2 and 36.3). Anywhere between 17 and 50 per cent of patients with systemic lupus erythematosis and central nervous system involvement have seizures, Seizures may also complicate, to a lesser degree, involvement in polyarteritis nodosa, Behcet’s disease, and mixed connective tissue disease (Shannon and Goetz 1989). Seizures can occur in hypertensive encephalopathy and in subacute bacterial endocarditis.

A rare cerebrovascular cause of seizures is the hyper-reperfusion syndrome occurring after carotid endarterectomy where seizures can occur on the first post-operative day. It is doubtful that this has a significant risk for late epilepsy (Nielsen et al. 1995).

A wide range of viral, bacterial, opportunistic, and parasitic infestations can be associated with seizures. Infections accounted for 3 per cent of seizure disorders in the epidemiological study in Rochester, Minnosota (Hauser and Kurland 1975). The 20-year risk of developing unprovoked seizures following common central nervous system infections was 6.8 per cent, almost 7 times the expected rate. Increased incidence of seizures was highest during the 5 years after a neurological infection but continued to be elevated for as long as 15 years (Annegers et al. 1988). The risk was highest, at 22 per cent, for patients with a viral encephalitis associated with acute symptomatic seizures, and 10 per cent for patients with viral encephalitis without early seizures. For bacterial meningitis associated with early seizures the risk was 13 per cent, but only 2.4 per cent for those without early seizures. Aseptic meningitis does not pose an increased risk of seizures.

Seizures commonly occur during acute viral encephalitis, particularly with herpes simplex encephalitis when the seizures are frequently focal in nature. Pre-natal infection with cytomegalovirus, rubella, and herpes can produce retardation associated with late epilepsy (Forsgren et al. 1990). Seizures can also occur with subacute measles encephalitis and subacute rubella encephalitis as well as with ‘slow virus infections’ including subacute sclerosing panencephalitis and Creutzfeld–Jacob disease, but in both these conditions myoclonus tends to dominante the picture. Infection with HIV can be associated with seizures not only because of an increased risk of opportunistic infections, but also because the direct neurotropic effects of the virus (Wong et al. 1990). Of 100 patients who were HIV positive and had seizures, 45 had evidence of opportunistic infection or central nervous system lymphoma, 24 evidence of the AIDS–dementia complex, but 23 had no other identifiable cause for their seizures (Holtzman et al. 1989).

Bacterial infections causing meningitis are occasionally associated with seizures, particularly if complicated by cortical vein or sinus thrombosis or cerebral abscess. Tuberculous meningitis may present with seizures. In the Indian subcontinent tuberculomas may be a common cause of epilepsy associated with disappearing ring-enhancing CT lesions (Goulatia et al. 1987). Other causes of chronic meningitis are not infrequently associated with seizures, for instance cryptococcus, and candida. Perhaps the commonest infestation associated with seizures is cysticercosis or which may account for 50 per cent of incident cases in adults in developing countries (Medina et al. 1990) (Section 43.2.9). Others include schistosomiasis, hydatid, malaria, and toxoplasmia (Bittencourt et al. 1988).

The introduction of modern high resolution MRI has shown that malformations of cortical development are important and common causes of epilepsy (Section 9.2). A series from Norway found that 13 of 303 patients had such developmental anomalies (Brodtkorb et al. 1992), while 16 of 222 patients with temporal lobe epilepsy had developmental abnormalities in another series (Lehericy et al. 1995). Malformations of cortical development may be generalized, when they are usually associated with developmental delay and a static encephalopathy from birth as well as a severe epilepsy, or they may be focal in nature with seizures as the only symptom. Such focal abnormalities may account for 15–20 per cent of the adult population with intractable epilepsy (Kuzniecky and Jackson 1997). Many have a genetic basis and some form a part of other well-recognized symptom complexes such as the neurocutaneous syndromes, for example tuberose sclerosis (Section 11.1). They may also be environmental and related to infection, toxins, and radiation.

Focal cortical dysplasia is usually frontal or temporal with MRI findings of abnormal gyral thickening often associated with abnormalities of the underlying white matter with blurring of the grey–white matter interface (Fig. 31.10). On occasions, the abnormality may be ‘transmantle’ extending from the ventricle to the cortical surface. The changes may be difficult to differentiate from those of tuberose sclerosis. Seizures usually present in the first decade and are often refractory to drug treatment. Surgical resection offers benefits in many cases.

The heterotopias are, by definition, normal cells present in an abnormal location (Section 9.2.5). Subcortical band heterotopia is seen in women and consists of MR changes with a circumferential band of subcortical grey matter (Fig. 31.11). Lisencephaly results in an abnormally smooth cortical surface and may be incompatible with survival. Both conditions can be seen within the same family and segregate with familial epilepsy. Mutations in the DCX gene on the X chromosome and LIS1 gene on chromosome 17 have been implicated in the disorders (Sisodiya 2004).

Periventricular or subependymal nodular heterotopia is probably the commonest developmental disorders seen in patients with epilepsy. MR appearances are of multiple smooth nodules of grey matter lining the lateral ventricles, which may be associated with other focal subcortical heterotopias (Fig. 31.12). On rare occasions, focal subcortical heterotopia can be seen in the absence of subependymal heterotopia. Most patients have relatively normal development but perhaps 80 per cent of patients with the disorder have epilepsy. When familial there is evidence of more women being

affected, with fewer surviving males. The condition can be caused by mutations of FLNA on Xq28, encoding for filamin A, a protein required for neuronal migration.

Polymicrogyria refers to the presence of an area with many abnormally small gyri with an irregular cortical surface. When diffuse, it can be associated with severe developmental delay, with localized epilepsy. Schizencephaly describes the presence of grey matter lined clefts that extend from the cortical surface to the ependymal lining and may represent a form of polymicrogyria. The condition can be unilateral or bilateral. Developmental delay and contralateral hemiparesis are common.

The importance of the recognition and accurate diagnosis of cortical dysplasia lies in the potential for surgical treatment in some patients. Cortical dysplasias of the temporal lobe seem to have by far the best surgical outcome with up to 50 per cent being seizure-free (Kuzniecky and Jackson 1997). Unfortunately, the majority of cortical dysplasias are extra-temporal and here lower seizure-free rates are seen when localized resection of focal dysplasias is attempted.

Dysembrioplastic neuro-epithelial tumours are present in many patients diagnosed as having low-grade temporal gliomas or hamartomas in early series (Daumas-Duport et al. 1988). They can be reasonably described as malformations of cortical development due to abnormal proliferation or apoptosis. The lesions are usually seen within areas of dysplastic cortex and seem to have a predilection for the temporal lobes though they may also occur in the frontal lobes. They may make up between 5 and 10 per cent of pathology found in temporal lobectomy series and virtually never show any form of malignant transformation.

Neurodegenerative disorders can be associated with epilepsy. In Alzheimer’s disease seizures occur usually late in the illness in up to 15 per cent of patients (Romanelli et al. 1990) (Section 34.6.2). The relative risk of epilepsy for autopsy-proven causes of Alzheimer’s disease is around 10 (Hauser et al. 1986). Myoclonus is also evident, particularly in patients with familial Alzheimer’s disease and with Alzheimer’s change complicating Down’s syndrome. In contrast, seizures appear rare in frontotemporal degenerations (Section 34.6.4).

An increased incidence of seizures has been noted in association with multiple sclerosis, at 3–6 times the expected rate (Poser and Brinar 2003) (Section 37.5.3). It may be that seizures are particularly likely to occur as an acute symptomatic phenomenon related to clinical relapse and plaque formation. More chronic epilepsy can be seen in severely disabled patients with frontal lobe syndromes and numerous frontal subcortical plaques (Moreau et al. 1998).

The diagnosis of epilepsy in the adult is essentially clinical, and based on a detailed description of events experienced by the patient before, during, and after a seizure. An eye witness account is particularly important. In view of the social and economic implications, diagnostic errors must be avoided at all costs. Thus, the first rule about diagnosing epilepsy is never to make the diagnosis without incontrovertible clinical evidence. If there is any doubt, the clinician should resist the temptation to attach a label and should rely on the passage of time and the further description of symptomatic events to reach a firm conclusion. Few people with epilepsy will come to harm from a delay in diagnosis whereas a false- positive diagnosis is always gravely damaging. In patients with possible syncope there is an imperative to reach a clear diagnosis as this group have excess morbidity and mortality (Section 31.1.2).

However, it is not enough simply to decide that a patient’s attacks are epileptic in nature. Other considerations must be addressed in the diagnostic process.

While the differentiation of seizures from other episodic events is made on clinical grounds, investigations have particular importance in answering the subsequent diagnostic questions.

The EEG may rarely provide information that adds weight to the clinical diagnosis, but more importantly aids the classification of epilepsy. Routine inter-ictal EEG recording is one of the most abused investigations in clinical medicine and wrong interpretation of it use results in great human suffering. The diagnostic value of an inter-ictal EEG is widely misunderstood (Binnie 1997). Often the EEG is requested either to exclude or to prove a diagnosis of epilepsy— something that can seldom, if ever, be done. Erroneous interpretation of the EEG is probably the commonest reason for non-epileptic events being diagnosed as seizures. Particular problems are caused by misinterpretation of non-epileptiform sharp transients, such as 6 and 14 per second positive spikes and benign epileptiform transients of sleep, and responses to hyperventilation and photic stimulation.

In a population of patients with definite epilepsy from a tertiary referral centre, 30 per cent exhibited epileptiform activity in every routine inter-ictal record, while 11 per cent never did (Ajmone-Marsan and Zivin 1970). Sleep recording may increase sensitivity; epileptiform abnormalities occurred in 63 per cent of subjects without discharges in an intial awake record. When it is recognized that simple partial and brief complex partial seizures can occur without detectable changes at scalp electrodes in ictal records, the sensitivity of standard inter-ictal EEG recording will always be poor. It must be remembered that these figures apply to populations with clinically definite epilepsy, and considerably overestimate the value of the EEG in populations for whom there is diagnostic uncertainty. Interictal epileptiform features were found in 39 per cent of adults with a first presentation of seizures or epilepsy, many of whom had recordings within 24 h of a first seizure (King et al. 1998). This rate was again increased by subsequent sleep-deprived recording. The specificity of the inter-ictal EEG is best demonstrated by series that screened military personnel (Gregory et al. 1993). In a total of 21 000 individuals epileptiform abnormalities were found in only 2.4 per 1000.

The inter-ictal EEG is potentially important in two clinical settings. First, it may be difficult to differentiate absence seizures from complex partial seizures. In patients with seizures occurring without an aura that are characterized by a brief period of absence with or without automatism. The finding of generalized spike wave or focal spike activity, respectively, will clarify the diagnosis. This differentiation has important implications for treatment and prognosis. Second, in patients with tonic-clonic seizures without an aura, especially when these occur during sleep, the EEG can again differentiate between generalized epilepsies characterized by generalized spike wave and seizures with a focal onset in which there may be localized abnormalities.

Ictal EEG, using ambulatory or more satisfactorily videotelemetry techniques, is important in distinguishing epileptic seizures from non-epileptic attacks. However, movement and other artefacts may complicate interpretation. Identifying tonic-clonic seizures should, however, present no problems because of the post-ictal changes. Differentiating between non-convulsive psychogenic episodes and complex partial seizures, particularly those of frontal origin can still be difficult.

Over the past decade the role of neurological imaging in the diagnosis and management of epilepsy has changed considerably. On the one hand CT scanning is universally available, but MR scanning has become enormously sophisticated and capable of demonstrating many abnormalities not previously recognized by CT.

MRI is more sensitive for most of the cerebral pathologies associated with chronic epilepsy, with the exception of calcification which it does not demonstrate well. Mesial temporal sclerosis, low-grade neoplasia, vascular lesions, particularly caveromas, and developmental abnormalities are all likely to be missed on CT, but readily demonstrated by MRI. In the elderly, however, MRI often shows a high incidence of lesions, whose clinical relevance to the seizures may be uncertain. In a series of 300 children and adults presenting with first seizures, MRI identified epileptogenic lesions in 17 per cent of 154 patients with definite partial epilepsy and 18 per cent with unclassified epilepsy, but was normal in all patients with clinical and EEG evidence of a generalized epilepsy syndrome (King et al. 1998). MRI was more sensitive than CT in the early detection of causative lesions. Indeed, CT detected only half the 17 tumours found, although it was uncertain how many of these would have been optimally treated by surgery if diagnosed.

At present a reasonable approach is to undertake imaging at the point of diagnosis in all adults except those who can be identified clinically and neurophysiologically as having one of the syndromes of idiopathic generalized epilepsy. While MRI is preferable, CT scanning is effective in identifying the more aggressive tumours causing epilepsy, but could miss more indolent gliomas. MRI is indicated in all patients with epilepsy who appear refractory to pharmacological treatment irrespective of previous imaging studies. Here MRI should be tailored to the individual, but will usually include high definition T1-weighted thin-slice scans, often with hippocampal volumetry, T2-weighted and FLAIR images.

Other technologies that produce ‘functional’ images such as positron emission tomography, PET, single photon emission computerized tomography, SPECT, and functional MRI and spectroscopy are largely experimental and to date have usually been used in the assessment of patients for surgical treatment of their epilepsy (Sections 3.1.5 and 3.1.6).

At a time when there is a sudden and dramatic increase in the number and choice of drugs to treat epilepsy, it is perhaps important to begin by considering some broad principles that need to be applied to the treatment of an individual patient:

Most comparative studies of anti-epileptic drugs identify the spectrum of adverse effects of anti-epileptic drugs as the factor contributing most to determining their relative effectiveness. It has proved difficult to detect significant differences in efficacy outcomes in comparative monotherapy studies, because of inadequate design and powering of studies (Glauser et al. 2006). Anti-epileptic drugs possess dose-related adverse effects, largely affecting the central nervous system, as well as idiosyncratic side effects. In addition, because of the long periods of time for which they may be taken, these drugs have also been associated with chronic toxicity, as well as teratogenicity as they may be taken through the child-bearing years. All these issues need to be taken into account in choosing drug treatment.

In the past anti-epileptic treatment has been advocated before seizures occur. Such prophylactic treatment has been recommended for patients with a high prospective risk of epilepsy after head injury and craniotomy for various neurosurgical conditions. Because no clear evidence exists that anti-epileptic treatment is effective in preventing late epilepsy (Section 31.8.2) (Temkin et al. 1998), it seems better to delay treatment until seizures have occurred rather than to adopt a policy of treatment of all those at risk— particularly as there may be a high incidence of side effects with prophylactic treatment and of poor compliance.

It is also common for clinicians to see patients with a first tonic-clonic seizure, infrequent such seizures, or seizures with minor symptomatology for whom there would be uncertainty about commencing treatment. Such patients have been studied in two randomized controlled trials (Marson et al. 2005; Leone et al. 2006). These provide evidence that early treatment fails to modify the longer term prognosis which is universally good, but that it can reduce the incidence of seizures in the shorter term. The chief factors affecting the risk of seizure recurrence in the short term are increasing numbers of seizures, an abnormal EEG, and the presence of a history of, or current evidence of underlying brain disease of which seizures are symptomatic. The presence of these factors make it possible to identify a small number of higher risk patients for whom treatment may be worthwhile (Kim et al. 2006).

When two or more unprovoked seizures have occurred within a short interval, anti-epileptic therapy is usually indicated. Problems do, however, arise in defining a short interval. Most would include periods of 6 months to 1 year within the definition. Even where seizures occur close together, the identification of specific precipitating factors may make it more important to counsel patients than to commence drug therapy. The most common examples are photically induced and alcohol-withdrawal seizures in adults.

There is agreement that patients with newly diagnosed epilepsy should be treated with a single drug. The key issue in the choice of this first drug at diagnosis is an accurate and adequate diagnosis of seizure type and, if possible, epilepsy syndrome. By no means all drugs are effective against all seizure types. The spectrum of efficacy of drugs is represented graphically in Fig. 31.13. It is particularly important to avoid the use of drugs that may themselves exacerbate seizures. Hence, there is evidence that both carbamazepine and vigabatrin (Perucca et al. 1998) may exacerbate absence and myoclonic seizures in the generalized epilepsy syndromes. It is here that syndromic classification of epilepsy becomes most important as a drug should be chosen that would be effective against all seizure types known to occur in that syndrome, rather than only those that have occurred in an individual patient to date.

It is now becoming possible to associate particular anti-epileptic drug mechanisms with effects against different seizure types and different adverse effects. Inevitably, many anti-epileptic drugs, may have multiple mechanisms of action, some of which are not fully understood. Table 31.8 summarizes different mechanisms of action for anti-epileptic drugs (Perucca 2005). Several drugs modify ionic sodium conductance across membranes, binding to ion channels in order to maintain them in an inactivated state thereby blocking repetitive neuronal firing. Drugs that possess this property include phenytoin, carbamazepine, lamotrigine, oxcarbazepine, topiramate, valproate, and zonisamide. All are effective in preventing partial seizures and both generalized and secondarily generalized tonic-clonic seizures. Most of these drugs, with the possible exception of valproate, can be associated with dose-related neurotoxicity syndromes that include ataxia, nystagmus, and diplopia. The second direct membrane effect is displayed by drugs such as ethosuximide and possibly valproate which modify slow or calcium-T currents in the thalamus. This mechanism seems particularly effective against spike wave mechanisms and absence seizures. These same calcium conductances can be enhanced by gabergic inputs to the thalamus via GABA_B receptors (Crunelli and Leresche 1991), which may explain the exacerbation of absences by vigabatrin.

A number of anti-epileptic drugs exert their properties through modulation of the GABA_A receptor/chloride ionophore. Thus, both benzodiazepines and barbiturates bind close to this site to increase chloride-ion conductance and maintain membrane hyperpolarization. Newer drugs such as vigabatrin and tiagabine may have more direct effects in prolonging the synaptic action of GABA. To date evidence suggests that these drugs are effective against partial and secondary generalized seizures but may exacerbate spike wave epilepsies. They are less likely to cause sedation and ataxia but may have a higher risk of psychiatric disorder including depression. Drugs, which interfere with excitatory neurotransmission via glutamate and aspartate receptors, may yet prove valuable anti-epileptic drugs. Some of the anti-epileptic properties of felbamate may be due to its ability to interfere with the action of glycine in facilitating glutaminergic activity, and a number of drugs with potential glutaminergic activity have entered clinical trial programmes with varying success.

Actions, common side-effects and indications for use are briefly summarized in Table 31.9. Further details of the pharmacokinetics of these drugs are also presented later in the chapter.

Currently, a large number of drugs can be considered for treating patients with cryptogenic and symptomatic partial epilepsies. These include both older and newer drugs, including those with a spectrum of efficacy limited to the partial epilepsies and those with a broader spectrum of effects (see Fig.13.13). Evidence-based guidelines have indicated that carbamazepine is for most patients the optimal first choice drug (NICE 2004). This guideline will require updating following the outcome of a large study comparing carbamazepine with newer anti-epileptic drugs. This showed that lamotrigine was superior to carbamazepine for time to treatment failure, being better tolerated and of equivalent efficacy to carbamazepine (Marson et al. 2007b) and was cost effective. Gabapentin and topirimate were inferior to lamotrigine, but there was uncertainty about the place of oxcarbazepine. The study did not include pregabalin, zonisamide, or levetiracetam for which there is some limited evidence that its effectiveness may be similar to carbamazepine (Brodie et al. 2007). The newer drugs are likely to remain second choice montherapies or will be used as adjunctive therapy with one of the first-choice sodium channel drugs, lamotrigine or carbamazepine or oxcarbazepine.

The management and treatment of the cryptogenic and symptomatic generalized epilepsies is more relevant to childhood epilepsy and will not be discussed further here (Section 30.7.3). The choice of drug therapy in the idiopathic generalized epilepsies is, however, a major issue as collectively these may represent between 20 and 30 per cent of all human epilepsies. Evidence-based guidelines have indicated that valproate is for most patients the optimal first choice drug (NICE 2004). This guideline will require updating following the outcome of a large study comparing valproate with lamotrigine and topiramate (Marson et al. 2007a). This shows that valproate remains the most clinically effective drug for this group of epilepsies, topiramate being poorly tolerated and lamotrigine lacking efficacy compared to valproate. These results create particular difficulties in decision making for women in their child-bearing years, for whom valproate may present problems during pregnancy (Section 31.10.5).

Decisions about starting anti-epileptic drug treatment often have to be made in the face of some uncertainty concerning a syndromic classification. While the clinician may be certain that seizures have occurred, there may be insufficient information available from a few poorly witnessed events to provide a definite syndrome diagnosis. Common situations in which this occurs include the patient with witnessed tonic-clonic seizures during sleep, and infrequent day-time trance-like episodes reflecting absence or complex partial seizures. Where this uncertainty exists it is relatively unusual for the EEG or other investigations to provide definitive information. In these circumstances, a broad spectrum anti-epileptic drug, such as valproate is to be preferred (Marson et al. 2007a).

When prescribing anti-epileptic drugs it is important that patients are counselled about the possible adverse effects and their significance.

Most anti-epileptic drugs including phenytoin, carbamazepine, lamotrigine, barbiturates, and benzodiazepines, give rise to a non-specific encephalopathy associated with high blood concentrations. Patients exhibit sedation and nystagmus and, with increasing blood levels, ataxia, dysarthria, and ultimately confusion and drowsiness. In some instances seizure frequency may increase with high blood levels and occasionally involuntary movements are seen, particularly with phenytoin. Phenytoin is especially likely to result in dose-related toxicity because of its unusual pharmacokinetics (Section 31.10.8). Carbamazepine may cause similar symptoms if the dose is not built up slowly, because of its ability to auto-induce liver microsomal enzymes. Valproate does not appear to be associated with this typical syndrome of neurotoxicity, but some patients with high blood levels may exhibit restlessness and irritability, sometimes with a frank confusion state, sometimes with elevated blood ammonia concentrations. Postural tremor is a more common dose-related adverse effect.

All anti-epileptic drugs can have adverse effects on cognitive function and behaviour with increasing dose and blood concentrations, although there is little evidence that they are common in patients using therapeutic doses as monotherapy (Loring et al. 1994). While carbamazepine and valproate may have smaller risks of this type than barbiturates and phenytoin, the newer generation of drugs may be better tolerated (McCorry et al. 2004).

Drug interactions may increase the risk of dose-related toxicity. Thus valproate may greatly prolong the half-life of lamotrigine, making dosage reduction necessary during co-medication. Similarly, withdrawal of enzyme-inducing drugs can give rise to similar problems.

Most anti-epileptic drugs, particularly phenytoin, carbamazepine, and lamotrigine, may cause a delayed hypersensitivity reaction consisting of maculopapular erythematous eruption which, in more severe cases, may be associated with fever, lymphadenopathy, and hepatitis. The incidence of allergic skin reaction with phenytoin may be as high as 10 per cent and with carbamazepine up to 15 per cent with rapid introduction of the drug (Chadwick et al. 1984). Lamotrigine is also associated with similar problems, requiring slow titration particular with co-medication with valproate. The incidence of serious reactions such as Stevens Johnson syndrome or acute epidermal necrolysis is much lower.

Marrow aplasia is a rare complication of carbamazepine, but more common with felbamate (Kaufman et al. 1997). Reports of fatal cases of liver failure in association with valproate therapy largely concern children under the age of 2 years who are often multiply handicapped and receiving many different anti-epileptic drugs. It may be that they have an underlying error of metabolism that predisposes them to liver failure (Dreifuss et al. 1987). The potential for rare idiosyncratic side-effects of levetiracetam, topiramate, and gabapentin is currently uncertain, but probably small.

Anti-epileptic drugs are unusual in that they may be administered to patients over a long period as treatment for chronic epilepsy. This may lead to the development of a wide variety of syndromes of chronic toxicity (Table 31.9). A number of factors seem to predispose to the development of these disorders, including polypharmacy, dosage, and duration of therapy. Whilst valproate and carbamazepine may have fewer chronic toxic effects than barbiturates and phenytoin, the delay in recognizing quite common chronic toxic effects with the older agents should warn us that continued vigilance is needed in the use of the newer anti-epileptic drugs. Some patients exposed to long-term treatment with vigabatrin have developed severe irreversible concentric visual field loss (Eke et al. 1997). Quantitative visual field assessment can reveal asymptomatic, usually nasal, visual field constriction, associated with electroretinographic changes in keeping with retinal cone system dysfunction (Krauss et al. 1998), in larger numbers of patients treated with vigabatrin. Further follow-up of 32 patients continuing monotherapy with vigabatrin and 19 patients continuing carbamazepine from the randomized controlled trial reported by Kalviainen et al. (1998) showed that 41 per cent of vigabatrin-treated patients had visual field constriction, compared to no carbamazepine treated patients, indicating a causal relationship between long-term vigabatrin exposure and visual field loss.

There is some concern about the incidence of weight gain on anti-epileptic drugs. Valproate, gabapentin, and pregabalin are perhaps most troublesome in this respect, although topiramate and zonisamide may be associated with weight loss.

The polycystic ovary syndrome of amenorrhea, oligomenorrhea, abnormal cycle intervals, or menometrorrhagia, and signs of hyperandrogenism, including hirsuitism, acne, and alopecia is more common in women with epilepsy. Obesity is a common association. The symptomatic syndrome is much less common than the finding of polycystic ovaries on ultrasound. A number of studies have shown that polycystic ovary syndrome is over-represented among women with epilepsy, but there is controversy as to whether this is due to drug effects (Bilo et al. 2001; Isojarvi et al. 2001). Epilepsy itself may be related to polycystic ovary syndrome via epileptiform discharges disrupting normal hypothalamic function.

There is evidence that populations of people with epilepsy have a higher prevalence of osteopenia and or osteoporosis than would be expected. Enzyme-inducing anti-epileptic drugs may contribute to this as may reduce mobility in people with epilepsy.