OUP user menu

Benign childhood focal epilepsies: assessment of established and newly recognized syndromes

Chrysostomos P. Panayiotopoulos, Michael Michael, Sue Sanders, Thalia Valeta, Michael Koutroumanidis
DOI: http://dx.doi.org/10.1093/brain/awn162 2264-2286 First published online: 21 August 2008

Summary

A big advance in epileptology has been the recognition of syndromes with distinct aetiology, clinical and EEG features, treatment and prognosis. A prime and common example of this is rolandic epilepsy that is well known by the general paediatricians for over 50 years, thus allowing a precise diagnosis that predicts an excellent prognosis. However, rolandic is not the only benign childhood epileptic syndrome. Converging evidence from multiple and independent clinical, EEG and magnetoencephalographic studies has documented Panayiotopoulos syndrome (PS) as a model of childhood autonomic epilepsy, which is also common and benign. Despite high prevalence, lengthy and dramatic features, PS as well as autonomic status epilepticus had eluded recognition because emetic and other ictal autonomic manifestations were dismissed as non-epileptic events of other diseases. Furthermore, PS because of frequent EEG occipital spikes has been erroneously considered as occipital epilepsy and thus confused with the idiopathic childhood occipital epilepsy of Gastaut (ICOE-G), which is another age-related but rarer and of unpredictable prognosis syndrome. Encephalitis is a common misdiagnosis for PS and migraine with visual aura for ICOE-G. Pathophysiologically, the symptomatogenic zone appears to correspond to the epileptogenic zone in rolandic epilepsy (sensory-motor symptomatology of the rolandic cortex) and the ICOE-G (occipital lobe symptomatology), while the autonomic clinical manifestations of PS are likely to be generated by variable and widely spread epileptogenic foci acting upon a temporarily hyperexcitable central autonomic network. Rolandic epilepsy, PS, ICOE-G and other possible clinical phenotypes of benign childhood focal seizures are likely to be linked together by a genetically determined, functional derangement of the systemic brain maturation that is age related (benign childhood seizure susceptibility syndrome). This is usually mild but exceptionally it may diverge to serious epileptic disorders such as epileptic encephalopathy with continuous spike and wave during sleep. Links with other benign and age-related seizures in early life such as febrile seizures, benign focal neonatal and infantile seizures is possible. Overlap with idiopathic generalized epilepsies is limited and of uncertain genetic significance. Taking all these into account, benign childhood focal seizures and related epileptic syndromes would need proper multi-disciplinary re-assessment in an evidence-based manner.

  • rolandic epilepsy
  • Panayiotopoulos syndrome
  • occipital epilepsy of Gastaut
  • autonomic seizures
  • autonomic status epilepticus
  • ictus emeticus
  • visual hallucinations
  • childhood seizure susceptibility
  • migraine, cyclic vomiting syndrome

Benign childhood focal seizures and related idiopathic epileptic syndromes affect approximately 22% of children with non-febrile seizures and constitute a significant part of the everyday practice of paediatricians, neurologists and electroencephalographers. They comprise three identifiable electroclinical syndromes recognized by the International League against Epilepsy (ILAE) (Engel, 2006): rolandic epilepsy which is well known, Panayiotopoulos syndrome (PS), a common autonomic epilepsy, which is currently more readily diagnosed and the idiopathic childhood occipital epilepsy of Gastaut (ICOE-G) including the idiopathic photosensitive occipital lobe epilepsy, a less common form with uncertain prognosis. There are also reports of children with benign focal seizures of predominantly affective symptoms, and claims have been made for other clinical phenotypes associated with specific inter-ictal EEG foci, such as frontal, midline or parietal, with or without giant somatosensory evoked spikes (GSES). Neurological and mental states and brain imaging are normal, though because of their high prevalence any type of benign childhood focal seizures may incidentally occur in children with neurocognitive deficits or abnormal brain scans. The most useful diagnostic test is the EEG. In clinical practice, the combination of a normal child with infrequent seizures and an EEG showing disproportionately severe spike activity is highly suggestive of these benign childhood syndromes (Panayiotopoulos, 1999a).

All these conditions may be linked together in a broad, age-related and age-limited, benign childhood seizure susceptibility syndrome (BCSSS), which may be genetically determined (Panayiotopoulos, 1993).

This appraisal is based on over 30 years clinical research, extensive review of English and other journals including relevant book chapters and personal communications with experts. Details of original studies, numerous case histories and published reports not cited here can be found in our previous reviews (Panayiotopoulos, 1999a, 2002, 2007; Koutroumanidis, 2007).

Rolandic epilepsy (benign childhood epilepsy with centro-temporal spikes)

Rolandic epilepsy is the best known and most common benign childhood focal epilepsy ((Beaussart and Faou, 1978; Loiseau et al., 1988; Bouma et al., 1997; Beaussart et al., 1999; Panayiotopoulos, 1999a; Dalla Bernardina et al., 2005; Wirrell et al., 2006; Fejerman, 2008). The age of onset ranges from 1 to 14 years with 75% starting between 7 and 10 years. There is a 1.5 male predominance, prevalence is around 15% in children aged 1–15 years with non-febrile seizures and incidence is 10–20/100 000 children aged 0–15 years (Heijbel et al., 1975; Cavazzuti, 1980; Sidenvall et al., 1996; Astradsson et al., 1998; Berg et al., 1999; Larsson and Eeg-Olofsson, 2006).

Clinical manifestations

The cardinal features of rolandic epilepsy are focal seizures consisting of unilateral facial sensory-motor symptoms (30% of patients), oro-pharyngo-laryngeal symptoms (OPLS) (53%), speech arrest (40%) and hypersalivation (30%) (Beaussart, 1972; Loiseau and Beaussart, 1973; Lerman and Kivity, 1975; Bouma et al., 1997; Panayiotopoulos, 1999a; Dalla Bernardina et al., 2005; Wirrell et al., 2006; Fejerman, 2008). Ictal manifestations indicative of temporal lobe involvement do not occur in rolandic epilepsy, and the term ‘centro-temporal’ refers only to the spike topography where it is partly a misnomer (see EEG section below).

Hemifacial sensory-motor seizures are mainly localized in the lower lip and may spread to the ipsilateral hand. Motor manifestations are clonic contractions sometimes concurrent with ipsilateral tonic deviation of the mouth, and sensory symptoms consist of numbness in the corner of the mouth. OPLSs are unilateral sensory-motor symptoms of numbness or paraesthesias (tingling, prickling or freezing) inside the mouth, associated with strange sounds, such as death rattle, gargling, grunting and guttural sounds.

Hypersalivation, a prominent autonomic manifestation, is often associated with hemifacial seizures, OPLSs and speech arrest. In speech arrest, the child is actually anarthric, unable to utter a single intelligible word and attempts to communicate with gestures.

Consciousness and recollection are fully retained in more than half (58%) of rolandic seizures. In the remainder, consciousness becomes impaired during the ictal progress and in one-third there is no recollection of ictal events. Progression to hemiconvulsions or generalized tonic–clonic seizures (GTCS) occurs in around half of children and hemiconvulsions may be followed by postictal Todd's hemiparesis (Panayiotopoulos, 1999a).

Three-quarters of rolandic seizures occur during non-REM (rapid eye movement) sleep, mainly at sleep onset or just before awakening.

Rolandic seizures are usually brief lasting for 1–3 min. Focal motor, hemiconvulsive and generalized convulsive status epilepticus are rare at around 5% (Deonna et al., 1986; Wirrell et al., 1995; Panayiotopoulos, 1999a). Opercular status epilepticus usually occurs in children with atypical evolution (Colamaria et al., 1991; Deonna et al., 1993; Fejerman et al., 2000) or may be induced by carbamazepine or lamotrigine (Caraballo et al., 1989; Parmeggiani et al., 2004). This state lasts for hours to months and consists of ongoing unilateral or bilateral contractions of the mouth, tongue or eyelids, positive or negative subtle perioral or other myoclonus, dysarthria, speech arrest, difficulties in swallowing, buccofacial apraxia and hypersalivation.

Other seizure types

Despite prominent hypersalivation, focal seizures with primarily autonomic manifestations (autonomic seizures) are not considered part of the core clinical syndrome of rolandic epilepsy. However, some children may present with independent autonomic seizures or seizures with mixed rolandic-autonomic manifestations including emesis (see below in the relations between rolandic epilepsy and PS).

Primarily GTCS are considered part of rolandic epilepsy by the ILAE (Engel, 2006) and their occurrence can not be excluded. However, from the published ictal recordings (Watanabe, 1996; Panayiotopoulos, 1999a; Wirrell et al., 2006) and the electroclinically unequivocal focal nature of rolandic epilepsy, it can be inferred that at least the majority of the GTCS follow rolandic activation, and are therefore secondarily GTCS. Short-lived initial focal symptoms may pass unnoticed in daytime GTCS and are bound to be missed in nocturnal GTCS.

Typical absence seizures are considered rare (Beydoun et al., 1992; Gelisse et al., 1999; Panayiotopoulos, 1999a) though a high incidence of them has also been reported (Beaumanoir et al., 1974).

Electroencephalography

By definition, centro-temporal spikes (CTS) are the hallmark of benign childhood epilepsy with CTS (Fig. 1). However, although called centro-temporal, these spikes are mainly localized in the C3 and C4 (high central) or C5 and C6 (low central) supra-sylvian and not temporal electrodes (Legarda et al., 1994; Panayiotopoulos, 1999a). CTS are often bilateral and typically activated by drowsiness and slow (non-REM) sleep, but not by overbreathing (Smith and Kellaway, 1964; Blom and Brorson, 1966; Clemens and Majoros, 1987). In serial EEGs of the same child, CTS may occur right or left, infrequently or frequently, and appear small or giant, alone or with spikes in other locations. Rarely, children with rolandic epilepsy may have normal EEG or CTS may appear only during non-REM sleep (3–35%) (Panayiotopoulos, 1999a). The incidence of extra-rolandic spikes in rolandic epilepsy is not precisely known but may be significant when these are sought (Drury and Beydoun, 1991).

Fig. 1

Interictal EEG in rolandic epilepsy (top), PS (middle) and idiopathic childhood occipital epilepsy of Gastaut (bottom). (A) Interictal EEG with CTS of two children with rolandic epilepsy separated by the vertical line. Left: spontaneous and GSES elicited by the patient simultaneously tapping the tip of the fingers of both hands (arrows). Right: the same EEG sample at two different montages (extreme right is laplacian montage) to show that what appears as CTS are not temporal which on this occasion are localized in the lower central electrode (C5) of the laplacian montage. (B) Interictal EEG in five patients with PS separated by vertical lines. Despite similar clinical features, spikes are localized in the occipital, centro-temporal and frontal regions or they are frequently multifocal and may appear as clone-like, repetitive, spike–wave complexes. Brief generalized discharges of slow wave with small spikes (extreme right) are sometimes an interictal EEG feature. (C) Interictal EEG in two patients with ICOE-G separated by a vertical line. Left: Classical occipital paroxysms demonstrating FOS. Right: spontaneous scattered occipital spikes and occipital photosensitivity.

Dipole EEG (Gregory and Wong, 1992; Tsai and Hung, 1998; Jung et al., 2003), magnetoencephalography (MEG) (Minami et al., 1996; Huiskamp et al., 2004) and functional MRI (Boor et al., 2007) studies have demonstrated that the main negative spike component of CTS is usually modelled by a single and stable tangential dipole source with the negative pole maximum in the central region and the positive pole maximum in the frontal region.

Brief 1–3 s generalized bursts of 3–5 Hz slow waves with intermixed small spikes without associated overt clinical symptoms may occur in about 4% of patients with rolandic epilepsy (Gelisse et al., 1999; Panayiotopoulos, 1999a). Typical 3 Hz spike–wave discharges are probably rare (see absence seizures above).

CTS are diagnostic markers of benign rolandic epilepsy only in a suggestive clinical presentation. Their frequency, location and persistence do not determine the clinical manifestations, severity and frequency of seizures or the prognosis. It is well established that CTS are not specific to rolandic epilepsy (Kellaway, 1980; Panayiotopoulos, 1999a) as they:

  • occur in 2–3% of normal school-aged children, of whom <10% develop rolandic seizures (Gibbs and Gibbs, 1967; Petersen and Eeg-Olofsson, 1971; Cavazzuti et al., 1980; Okubo et al., 1994).

  • are common among relatives of children with rolandic epilepsy (Bray and Wiser, 1965; Bali et al., 2007)

  • may occur in a variety of organic brain diseases with or without seizures, such as cerebral tumours, Rett syndrome, fragile X syndrome and focal cortical dysplasia (Kellaway, 1980; Panayiotopoulos, 1999a)

  • may incidentally be found in non-epileptic children with various symptoms, such as headache, speech and behavioural and learning difficulties (Gibbs and Gibbs, 1967).

Somatosensory stimulation is common form of activation of CTS (10–20%) (De Marco and Tassinari, 1981; Panayiotopoulos, 1999a; Fonseca and Tedrus, 2000; Kubota et al., 2000; Langill and Wong, 2003) and evokes GSES (Fig. 1), which correspond to mid- or long-latency somatosensory evoked potentials (Manganotti et al., 1998). GSES, like spontaneous CTS, occur in children with or without seizures and disappear with age.

There have been around 20 reported ictal EEGs of rolandic seizures showing an initial paucity of spontaneous CTS before the onset of the ictal discharge, which appears in the contralateral to the clinical manifestations rolandic regions and consists of slow waves intermixed with spikes (Panayiotopoulos, 1999a; Wirrell et al., 2006; Fejerman et al., 2007) (Fig. 2). GTCS, when occurred, were preceded by focal clinical and EEG features. (Watanabe, 1996; Panayiotopoulos, 1999a; Wirrell et al., 2006).

Fig. 2

Ictal EEG of rolandic (A), autonomic (B) and occipital lobe (C) seizures. (A) Ictal EEG of a rolandic seizure. The onset of ictal discharge becomes apparent at the right central regions (black arrow) and the first clinical events (open arrow) occur 12 s later and consist of hemifacial motor symptoms indicated by the muscle artefacts. The seizure ended with a GTCS. (B) Ictal EEG of an autonomic seizure. The onset of ictal discharge becomes apparent at the right frontal regions (black arrow) and the first clinical event (open arrow) occurs 12 min later and consists of coughing. The seizure was prolonged for over 40 min terminated with intravenous benzodiazepine. Modified with permission from Koutroumanidis et al. (2005) where further details can be found. (C) Ictal EEG of an occipital seizure. The onset of ictal discharge becomes apparent at both occipital regions (black arrow) nearly simultaneously with the first clinical events (open arrow) which consist of blurring of vision. The seizure also consisted of forced eyelid closures and lasted for less than a minute.

Aetiology

Rolandic epilepsy is genetically determined although conventional genetic influences may be less important than other mechanisms (Vadlamudi et al., 2004, 2006). There is evidence of linkage with chromosome 15q14 (Neubauer et al., 1998). Autosomal dominant inheritance with age-dependent penetrance refers to the EEG CTS and not to the clinical syndrome of rolandic epilepsy (Bray and Wiser, 1965; Bali et al., 2007).

Siblings or parents of patients with rolandic epilepsy may rarely have the same type of seizures or other phenotypes of BCSSS, such as PS. Reported occurrence of febrile seizures ranges from 10% to 20% (probably around 18%) of patients (Kajitani et al., 1992).

Pathophysiology

As indicated by the distribution of CTS, the epileptogenic zone in rolandic epilepsy involves neuronal networks within the rolandic cortex surrounding the central fissure bilaterally. This is congruent with the seizure symptomatogy (symptomatogenic zone) and in agreement with those described by Penfield and Rasmussen (1957) during electrical stimulation of the lower part of the pre-central and post-central gyrus in man.

The speech arrest is due to anarthia attributed to loss of the power and coordination of the musculature responsible for the articulation of words. There is no impairment of the cortical language networks. Hypersalivation most probably relates to the involvement of the superior bank of the sylvian fissure (Luders et al., 1987), but defining ictal symptomatogenesis by plotting the simple topographic coordinates of an ictal discharge can hardly explain the high prevalence of hypersalivation in benign rolandic epilepsy compared to its exceptional only occurrence in adults with symptomatic foci of similar topography. Nor can it explain the opercular status epilepticus, with the lengthy for hours speech arrest, drooling and bilateral regional twitching that is associated with diffuse or bilateral rolandic spike wave activity, but does not propagate in a conventional way and does not involve other systems like for instance the motor strip or the language function. Therefore, at variance with the symptomatic adult focal epilepsies of comparable but more discretely localized topography, rolandic epilepsy reflects an age-related maturational instability of the lower rolandic (somatosensory) cortex that represents the face and the oropharynx bilaterally (Koutroumanidis, 2007).

Evolution and prognosis

The prognosis for rolandic seizures is almost invariably excellent, with probably <2% risk of developing absence seizures and less often GTCS in adult life (Beaumanoir et al., 1974; Lerman and Kivity, 1975; Blom and Heijbel, 1982; Loiseau et al., 1983; Panayiotopoulos, 1999a; Datta and Sinclair, 2007; Koutroumanidis et al., 2008; Caraballo et al., 2008b). Remission occurs within 2–4 years from onset and before the age of 16 years. The total number of seizures is low, the majority of patients having fewer than 10 seizures; 10–20% have just a single seizure. About 10–20% may have frequent seizures, but these also remit with age.

Children with rolandic seizures may develop usually mild and reversible linguistic, cognitive and behavioural abnormalities during the active phase of the disease (Giordani et al., 2006; Nicolai et al., 2006; Riva et al., 2007; Kossoff et al., 2007; Kavros et al., 2008; Perkins et al., 2008). These may be worse in children with onset of seizures before 8 years of age, high rate of occurrence and multifocal EEG spikes (Bulgheroni et al., 2008; Piccinelli et al., 2008; Boatman et al., 2008). The effect of anti-epileptic drugs (AED), the impact of stigmatizing because of epilepsy, bias in selection of the most serious cases and other factors have not been excluded in most of these studies. The development, social adaptation and occupations of adults with a previous history of rolandic seizures was found normal (Blom and Heijbel, 1982; Loiseau et al., 1983).

Rarely (<1%), rolandic epilepsy may evolve to more severe syndromes with linguistic, behavioural and neuropsychological deficits, such as Landau–Kleffner syndrome, atypical focal epilepsy of childhood or epilepsy with continuous spike and wave during sleep (CSWS) (Fejerman et al., 2000) as explained in the relevant section of this assessment.

Panayiotopoulos syndrome

PS is a common, childhood-related, susceptibility to autonomic seizures confirmed in long term studies of over 800 children worldwide (Panayiotopoulos, 1981, 1988, 2002; Ferrie et al., 1997; Oguni et al., 1999; Kivity et al., 2000; Lada et al., 2003; Ohtsu et al., 2003, 2008; Covanis et al., 2003; Caraballo et al., 2007; Dura-Trave et al., 2008).

The appraisal of PS here is based on in-depth analysis of our own patients and review of the related literature up to 2002 (Panayiotopoulos, 2002). Subsequent reports of around 60 publications on this and related aspects have also been thoroughly studied.

PS is defined as ‘benign age-related focal seizure disorder occurring in early and mid-childhood. It is characterized by seizures, often prolonged, with predominantly autonomic symptoms, and by an EEG that shows shifting and/or multiple foci, often with occipital predominance’ (Ferrie et al., 2006). ‘Early onset benign childhood occipital epilepsy’ often used as synonymous with PS (Taylor et al., 2003, 2008; Engel, 2006) does not represent the wide clinical, EEG and pathophysiological spectrum of PS, which is far beyond the occipital neocortex (Martinovic, 2007).

Onset is from age 1 to 14 years with 76% starting between 3 and 6 years. Both sexes are probably equally affected, though a female preponderance was found in some studies (Lada et al., 2003; Dura-Trave et al., 2008). Prevalence of PS may be high though this is practically absent in designed controlled epidemiological studies (Berg et al., 1999; Jallon et al., 2001; Shinnar and Pellock, 2002; Cowan, 2002; Forsgren et al., 2005), which is understandable as this syndrome was only recently formally recognized, its features imitate many other conditions and often manifests with a single seizure only. In the original cohort of Panayiotopoulos (1988), prevalence was around 13% in children aged 3–6 years with one or more non-febrile seizures, and 6% in the age group 1–15 years. These figures may be higher if children who are currently considered to have atypical clinical presentation are included in the syndrome (Panayiotopoulos, 2002; Covanis, 2006). PS is the most common specific cause of non-febrile, non-convulsive status epilepticus in childhood (Okanishi et al., 2008).

Clinical manifestations

The hallmark of PS is ictal autonomic aberrations that may involve any function of the autonomic system and mainly emesis (70–80% of seizures). The following description of clinical manifestations of PS are based on a synthetic analysis of available clinical historical data as perceived by patients and witnessed by observers from our records and those provided in the literature (Panayiotopoulos, 2002). Therefore, they may not accurately represent their true prevalence and sequence in PS.

Ictal autonomic symptoms

Seizures commonly commence with autonomic manifestations (80–90%), while consciousness and speech, as a rule, are preserved. Ictus emeticus (nausea and retching) culminates in vomiting in 74–82% of seizures; in others, only nausea or retching occurs and, in a quarter, emesis may not be apparent. Emesis is usually the first apparent ictal symptom, but it may also occur long after the onset of other manifestations. Other autonomic manifestations include pallor (28%), incontinence of urine (19%) and faeces (3%), hypersalivation (10%), cyanosis (12%), mydriasis (7%) and less often miosis (2%), coughing and abnormalities of intestinal motility (3%). Breathing (7%) and cardiac irregularities may be more common than reported. Tachycardia is usually found, sometimes at the onset, in ictal EEG (Beaumanoir, 1993b; Oguni et al., 1999; Koutroumanidis et al., 2005; Parisi et al., 2005). Cardiorespiratory arrest is rare probably occurring in 1 per 200 individuals; four cases out of around 1000 patients with PS have been reported but they all had complete recovery (Panayiotopoulos, 2002; Verrotti et al., 2005; Ferrie et al., 2006). Raised temperature has been documented in a few cases (2%) after seizure onset. Cephalic auras of discomfort and odd sensations or headache commonly are described with other autonomic symptoms at seizure onset.

Syncopal-like manifestations occur in at least one-fifth of seizures (Panayiotopoulos, 2002; Ferrie et al., 2006, 2007; Covanis, 2006; Caraballo et al., 2007). The child becomes ‘completely unresponsive and flaccid like a rag doll’, which may precede, be concurrent with other seizure symptoms or be the sole manifestation of a seizure (Oguni et al., 1999; Panayiotopoulos, 2002). They may occur while the patient is standing, sitting, lying down or asleep and last from 1–2 min to half an hour.

Ictal behavioural changes

Restlessness, agitation, terror or quietness, may occur at the onset of seizures, often in combination with other autonomic manifestations.

Ictal non-autonomic symptoms

Pure autonomic seizures and pure autonomic status epilepticus appear to occur in 10% of patients. They commence and terminate solely with autonomic symptoms. However, in the majority of seizures, autonomic manifestations are followed by conventional seizure symptoms. Nearly always, the child gradually or suddenly becomes confused or unresponsive. Other non-autonomic manifestations include in order of prevalence unilateral deviation of the eyes or eyes opening (60–83%), speech arrest (8–13%), hemifacial convulsions (6–13%), visual hallucinations (6–10%), OPLS (3%), unilateral drooping of the mouth (3%) and rarely (1%) eyelid or limb jerks, nystagmus or automatisms. The seizures may end with hemiconvulsions often with Jacksonian marching (19–30%), or generalized convulsions (21–36%).

Autonomic manifestations may not be apparent at seizure onset even in witnessed diurnal seizures (<10%). They may be absent, mild or missed in clinical observation. In these cases, eye deviation is the more common symptom. Visual symptoms are rare (1%) and not present in recurrent seizures (Ferrie et al., 1997; Yalcin et al., 1997; Panayiotopoulos, 2002; Caraballo et al., 2007).

Duration of seizures and precipitating factors

The seizures are usually lengthy of over 6 min and almost half of them last for >30 min to many hours, thus constituting autonomic status epilepticus (Panayiotopoulos, 2002; Ferrie et al., 2007). Lengthy seizures are equally common in sleep and wakefulness. Even after the most severe seizures and status, the patient is normal after a few hours’ sleep. There is no record of residual neurological abnormalities. Hemiconvulsive or convulsive status epilepticus is rare (4%).

Two-thirds of seizures start in sleep. Many seizures have been witnessed while travelling in a car, boat or aeroplane. The reason for this may be because in these circumstances children easily fall asleep, seizures are more likely to be witnessed and because travelling also precipitates motion sickness, to which children are particularly susceptible.

Intra-individual seizure variability

The same child may have brief and lengthy seizures, diurnal and nocturnal, with marked, inconspicuous, or even without any autonomic changes (Ferrie et al., 1997; Oguni et al., 1999; Kivity et al., 2000; Panayiotopoulos, 2002; Lada et al., 2003; Ohtsu et al., 2003, 2008; Covanis et al., 2003; Caraballo et al., 2007). Even cardinal symptoms (such as vomiting or eye deviation) may be present in one but absent in another seizure. Seizures without autonomic manifestations are rare (7%) and occur in patients who also have additional autonomic seizures (Panayiotopoulos, 2002). Ictal video EEG recordings have documented that autonomic symptoms and signs may vary between seizures of the same child (Koutroumanidis et al., 2005). There is no correlation between ictal semiology and topography of interictal spikes.

Aetiology

PS, like rolandic epilepsy, is probably genetically determined though conventional genetic influences may be less important than other mechanisms (Taylor et al., 2008). Usually, there is no family history of similar seizures, although siblings with PS or PS and rolandic epilepsy have been reported (Ferrie et al., 1997; Lada et al., 2003; Covanis et al., 2003; Caraballo et al., 2007; Livingston et al., 2008, Taylor et al., 2008). There is a high prevalence of febrile seizures (about 17%) (Panayiotopoulos, 2002).

SCN1A mutations have been recently reported in a child (Grosso et al., 2007) and two siblings (Livingston et al., 2008) with relatively early onset of seizures, prolonged time over which many seizures have occurred and strong association with febrile precipitants even after the age of 5 years. This is an area that needs further attention but may indicate that SCN1A mutations contribute to a more severe phenotype of PS.

Pathophysiology

Autonomic symptoms of any type are often encountered in seizures, whether focal or generalized, in adults or children (Freeman, 2006; Ferrie et al., 2007; Goodman et al., 2008). They are generated by activation or inhibition of parts of the central autonomic network that involves the insular cortex, medial prefrontal cortex, amygdala, hypothalamus and ventrolateral medulla (Goodman et al., 2008). The resultant autonomic disturbances depend on the brain areas involved in seizure onset or propagation, and appear as single or multiple symptoms some of which may be of localizing value (Elger, 2000).

In PS, the neuroanatomical and neurophysiological underpinnings of autonomic manifestations are unknown. Any hypothesis of the pathophysiology of PS should explain significant pieces of evidence that converge from clinical, EEG and magnetoencephalographic studies.

First, autonomic seizures and autonomic status epilepticus with the symptomatology and sequence as in PS appear to be specific for childhood (Panayiotopoulos, 2004; Ferrie et al., 2007). For example, in adults, ictal vomiting occurs rarely, and as a rule when consciousness is impaired following other focal mainly temporal lobe symptoms, and is attributed to non-dominant mesial temporal lobe involvement (Kramer et al., 1988; Schauble et al., 2002; Koutroumanidis, 2003) In contrast, ictal vomiting in children is common, usually occurs when consciousness is intact without preceding focal cortical symptoms, and probably has no localizing or lateralizing value (see ictal EEG). A possible explanation for this discrepancy may relate to the fact that children are constitutionally more vulnerable to emetic disturbances as exemplified by the ‘cyclic vomiting syndrome’, a non-seizure disorder of unknown aetiology that is also specific to childhood (Li et al., 1999) and associated with autonomic dysfunction (Chelimsky and Chelimsky, 2007). Thus, the preferential involvement of emetic and other autonomic manifestations in PS may be attributed to a maturation-related susceptibility of the central autonomic network (Panayiotopoulos, 2002, 2004).

Second, the epileptogenic zone in PS is wide and bilateral with multifocal pockets in cortical areas surrounding major fissures such as central, sylvian and mainly calcarine (Kanazawa et al., 2005; Yoshinaga et al., 2006; Saitoh et al., 2007, 2008).

Third, ictal autonomic symptomatology appears to pertain to any epileptogenic cortical onset zone, be this occipital, frontotemporal or frontal (Beaumanoir, 1993b; Oguni et al., 1999; Demirbilek and Dervent, 2004; Koutroumanidis et al., 2005; Parisi et al., 2005) (Fig. 2) and usually precede other focal cortical semiology. It is likely, that central autonomic networks have a lower threshold to epileptogenic activation than those producing focal cortical semiology (occipital, frontal, central, parietal and less often temporal). Irrespective of the localization of their onset, ictal discharges may activate the lower threshold autonomic centres (and therefore produce autonomic manifestations) commonly before other cortical regions of relatively higher threshold that generate focal cortical symptoms (sensory, motor, visual or other). Seizures remain purely autonomic if ictal neuronal activation of non-autonomic cortical areas fails to reach symptomatogenic threshold; otherwise they consist of autonomic and localization-related cortical symptoms and signs that may only rarely occur from onset. This hypothesis may explain why similar autonomic manifestations may appear from anterior or posterior, right or left brain onsets. As seizures primarily involve a particular system (the autonomic), PS may be considered as an electroclinical example of ‘system epilepsy’ (Koutroumanidis, 2007).

To explain the paradoxical discrepancy between the prolonged and ample-looking ictal discharges that have invariably featured in all published ictal recordings (Beaumanoir, 1993b; Oguni et al., 1999; Vigevano et al., 2000; Demirbilek and Dervent, 2004; Koutroumanidis et al., 2005; Parisi et al., 2005) and the also consistent lack of conspicuous cortical (motor or sensory) manifestations for several minutes into the seizure, one may also hypothesize a suboptimal ‘strength’ of the ictal electrical activity. Despite their scalp EEG phenomenology discharges presumably fail to transform into dynamic cortico–cortical propagation and generate conventional cortical symptoms according to their distribution over the cerebrum (being though still capable of activating a more hyperexcitable autonomic network). After all, the magnitude (amplitude) of ictal discharges as they appear in scalp recordings is hardly a reliable indicator of the ‘vigour’ of the clinical manifestations: for example, dramatically appearing hypermotor frontal lobe seizures may have no scalp EEG correlates and high voltage generalized or diffuse spike wave activity (such as in atypical absence seizures of epileptic encephalopathies) may have only few and mild clinical correlates (Koutroumanidis, 2007).

Syncopal-like attacks are difficult to explain. They may be a distinct seizure-type symptom similar to atonic seizures, but on some occasions they may be due to cardiac asystole (ictal syncope) generated by the seizure discharge.

Electroencephalography

Inter-ictal EEG findings show great variability (Fig. 1) (Panayiotopoulos, 1988, 2002; Oguni et al., 1999; Lada et al., 2003; Ohtsu et al., 2003, 2008; Covanis et al., 2003; Sanders et al., 2004; Ferrie et al., 2006; Caraballo et al., 2007). In about 90% of cases, the EEG reveals mainly multifocal, high amplitude, sharp slow wave complexes that may appear in any area, often shifting from one region to another in the same or the contralateral hemisphere in sequential EEGs of the same child. Occipital spikes predominate but they do not occur in a third of patients. Occipital paroxysms in their classical form with fixation off sensitivity (FOS) are even rarer. Clone-like multifocal spikes are intriguing features when they occur (19%). In routine EEG, these spikes appear to occur simultaneously at various locations in one or both hemispheres but usually they are driven (secondarily activated) by a primary spike generator which is predominantly posterior, leads the other spikes by a few ms and may also be the smallest of all or inconspicuous (Panayiotopoulos 2002; Leal et al., 2007, 2008). Brief generalized discharges of slow waves, intermixed with small spikes, may occur either alone (4%) or more often with focal spikes (15%). A single routine EEG may be normal in 10% of patients, and a few children have consistently normal wake EEGs before a diagnostic sleep recording. Sleep typically accentuates the spike abnormalities, and photosensitivity is practically absent.

As in benign rolandic epilepsy, the frequency, location and persistence of spikes do not determine the clinical manifestations, the duration, the severity and frequency of seizures or their prognosis. For instance, spikes may persist for many years after clinical remission or appear only once despite multiple EEGs.

The multifocal potential for epileptogenesis in PS has also been documented by EEG dipole analysis (Yoshinaga et al., 2006) and magnetoencephalography, which have implicated areas along the parieto-occipital, calcarine and central sulci or in the frontal lobes (Kanazawa et al., 2005; Saitoh et al., 2007; Saito et al., 2008).

In the few reported ictal EEGs, the discharges consist mainly of unilateral rhythmic slow activity, usually intermixed with fast rhythms and small spikes. They start in wider more often in posterior than anterior regions, quickly become diffuse and last for many minutes (Beaumanoir, 1993b; Oguni et al., 1999; Demirbilek and Dervent, 2004; Koutroumanidis et al., 2005; Parisi et al., 2005) (Fig. 2). The first ictal clinical symptoms become apparent usually long after the onset of the electrical discharge and present as tachycardia, breathing irregularities, coughing or emesis, which would be unlikely to consider as seizure events without an EEG.

Differential diagnosis

PS is easy to diagnose because of the characteristic clustering of clinical seizure semiology, which is often supported by inter-ictal EEG findings. The main problem is to recognize emetic and other autonomic manifestations as seizure events, and not to dismiss them or erroneously consider them as unrelated to the ictus and as a feature of encephalitis, migraine, syncope or gastroenteritis, which is the reason of the belated recognition of this common syndrome (Panayiotopoulos, 1988, 2002; Kivity and Lerman, 1992; Covanis, 2006).

A most difficult situation that demands experienced evaluation is when a child is seen at the acute stage of a seizure when symptoms may dramatically accumulate in succession and the diagnosis of true encephalitis is possible. A history of a previous similar seizure or full recovery after a few hours of sleep is reassuring and may help to avoid unnecessary investigations and promote withdrawal of any medication that may have been initiated (Kivity and Lerman, 1992; Sanders et al., 2004).

Approximately 10–20% of autonomic seizures and autonomic status epilepticus in children are due to heterogeneous cerebral pathology (Panayiotopoulos, 1988, 2002). These symptomatic cases are betrayed by abnormal neurological or mental state, abnormal brain imaging and background EEG abnormalities.

PS is significantly different from the rolandic epilepsy and the ICOE-G despite some overlapping clinical and/or EEG features. These are detailed in the relevant section of this article.

Prognosis

PS is remarkably benign in terms of its evolution (Panayiotopoulos, 1988, 2002; Ferrie et al., 1997; Oguni et al., 1999; Kivity et al., 2000; Lada et al., 2003; Ohtsu et al., 2003; Covanis et al., 2003; Caraballo et al., 2007) but autonomic seizures are of concern in the rare context of cardiorespiratory arrest though all four reported cases recovered completely (Panayiotopoulos, 2002; Verrotti et al., 2005; Ferrie et al., 2006, 2007). The majority of patients have a single or less than five seizures until remission. Only one quarter have multiple and sometimes very frequent and prolonged seizures that may be resistant to treatment. Remission often occurs within 1–2 years of onset but probably 10% may have more protracted active seizure periods. A fifth of patients develop rolandic and less often occipital or other seizures but these are also age-related and remit (Panayiotopoulos, 2002). Atypical evolution of PS similar to those described in rolandic epilepsy is rare, probably <3% (Caraballo et al., 2001, 2007; Ferrie et al., 2002; Kikumoto et al., 2006).

The risk of epilepsy in adult life appears to be no higher than in the general population (Panayiotopoulos, 2002; Ferrie et al., 2006; Caraballo et al., 2007).

Subtle neuropsychological deficits in some children during the active phase (Germano et al., 2005) may be syndrome-related but may also reflect effects of AED (most of the children were on AEDs including phenobarbital and vigabatrin) and/or other contributing factors. Prognosis of cognitive function is good even for patients with atypical evolutions (Caraballo et al., 2007).

Idiopathic childhood occipital epilepsy of Gastaut

The ICOE-G is a relatively rare form of pure occipital epilepsy accounting for about 2–7% of benign childhood focal seizures (Gastaut, 1981, 1982a, b; Panayiotopoulos, 1981, 1999a, b; Beaumanoir, 1983; Gastaut and Zifkin, 1987; Gastaut et al., 1992; Ferrie et al., 1997; Kivity et al., 2000; Covanis et al., 2005; Gobbi et al., 2008; Caraballo et al., 2008a). Age at onset ranges from 3 to 15 years, but most start between 8 and 11 years. Both sexes are equally affected.

Clinical manifestations

Seizures are occipital and primarily manifest with elementary visual hallucinations, blindness or both (Gastaut, 1982a; Gastaut and Zifkin, 1987; Gastaut et al., 1992; Panayiotopoulos, 1999a, b; Gobbi et al., 2008; Caraballo et al., 2008a). They are usually frequent, brief and diurnal.

Visual ictal symptoms

Elementary visual hallucinations are the most common and characteristic ictal symptom of ICOE-G. They are frequently the first and often the only seizure symptom. They develop rapidly within seconds and consist mainly of small multicoloured circular patterns that often appear in the periphery of a visual field, becoming larger and multiplying during the course of the seizure, frequently moving towards the other side.

Ictal blindness is probably the second most common symptom after visual hallucinations. It is sudden, usually total and it is frequently the first and often the only seizure symptom in patients who may also have other visual seizures without blindness. Impairment of visual awareness is consistently reported by some patients before the appearance of visual hallucinations.

Complex visual hallucinations such as faces and figures, and visual illusions such as micropsia, palinopsia and metamorphopsia occur in <10% of patients and mainly after the appearance of elementary visual hallucinations (Gastaut and Zifkin, 1987).

Non-visual ictal occipital lobe symptoms and signs

Non-visual occipital symptoms usually appear after the elementary visual hallucinations and these in order of prevalence are deviation of the eyes, eyelid fluttering or repetitive eye closures, pain and sensory hallucinations of ocular movements (Gastaut, 1982a; Gastaut and Zifkin, 1987; Panayiotopoulos, 1999a, b; Gobbi et al., 2008; Caraballo et al., 2008a).

Deviation of the eyes, often associated with ipsilateral turning of the head, is the most common (in about 70% of cases) non-visual ictal symptom. It usually starts after the commencement of visual hallucinations and may be mild, but more often it is forceful tonic and may progress to hemiconvulsions and GTCS. Some children may have seizures of eye deviation from the start without visual hallucinations and it is likely that these cases have a better prognosis (Beaumanoir, 1983; Ferrie et al., 1997). Other ocular manifestations may include unidirectional ocular clonic seizures (oculoclonic seizures) that are rare, and eyelid fluttering or repetitive eye closures that occur in about 10% of patients, usually at a later stage when consciousness is impaired. They signal an impending secondary GTCS.

Ictal headache, or mainly orbital pain, is a common ictal symptom, and in a small number of patients it may start before the first visual or other ictal occipital symptoms.

Consciousness

Consciousness is intact during the visual symptoms (simple focal seizures), but may be disturbed or lost in the course of the seizure, usually before or at the time of eye deviation or convulsions. Syncopal-like attacks are rare (Panayiotopoulos, 2002).

Extra-occipital seizure progression

Elementary visual hallucinations or other ictal symptoms may progress to complex focal seizures (14%), hemiconvulsions (43%) or GTCS (13%) (Gastaut and Zifkin, 1987). Complex focal seizures of temporal lobe symptomatology are extremely rare and may indicate a symptomatic cause (Panayiotopoulos, 1999b). Ictal vomiting may occur with progression to the non-dominant temporal lobe (Guerrini et al., 1995).

Post-ictal headache

Post-ictal headache, mainly diffuse, but also severe, unilateral, pulsating and indistinguishable from migraine headache, occurs in half the patients, in 10% of whom it may be associated with nausea and vomiting (Gastaut and Zifkin, 1987; Panayiotopoulos, 1999a, b; Caraballo et al., 2008a). This occurs immediately, or 5–10 min after the end of the visual hallucinations. The duration and severity of the headache appears to be proportional to the duration and severity of the preceding seizure although it may also occur after brief simple visual seizures.

Seizure stereotype

For any one patient, in every seizure, the elementary visual hallucinations have a fingerprint with a stereotypic appearance regarding morphology, colours, location, movement and other characteristics. Most of patients also know at what stage of their ictal manifestations a secondary GTCS may occur.

Duration and circadian distribution

Visual seizures are usually brief, lasting from a few seconds to 1–3 min if they occur alone without other occipital or extra-occipital spreading (Gastaut, 1982a; Gastaut and Zifkin, 1987; Panayiotopoulos, 1999a, b; Covanis et al., 2005; Gobbi et al., 2008; Caraballo et al., 2008a). However, a few patients with brief visual seizures may later develop lengthy visual seizures lasting for 10–20 min.

Visual seizures are predominantly diurnal and occur at any time of the day but some patients may also have infrequent seizures in sleep or on awakening.

Frequency of seizures

If untreated, the majority of patients experience frequent brief visual seizures ranging from several every day to one per week or month. However, propagation to other seizure manifestations, such as focal or generalized convulsions, is much less frequent occurring once per month, year or even rarer.

Precipitating factors and idiopathic photosensitive occipital epilepsy

This is a matter of inclusion criteria. Gastaut considered photosensitivity as part of ICOE-G (Gastaut, 1982a; Gastaut and Zifkin, 1987), while the ILAE Task Force recognizes ‘idiopathic photosensitive occipital lobe epilepsy’ as a syndrome of reflex epilepsy with age-related onset (Engel, 2001, 2006). Reflex occipital seizures induced by television, video games and intermittent photic stimulation (IPS) manifest with similar semiology as the spontaneous visual seizures (Aso et al., 1987; Michelucci and Tassinari, 1993; Guerrini et al., 1995, 1998; Yalcin et al., 2000; Panayiotopoulos, 2007). Deviation of the eyes, epigastric discomfort and vomiting, headache and generalized convulsions may follow. Prognosis is uncertain. Some children may have only 1 or 2 seizures, but others may not remit. Interictal EEG shows spontaneous and photically induced occipital spikes (Fig. 1). CTS may coexist. Ictal EEG documented the occipital origin and the spreading of the discharges to the temporal regions (Guerrini et al., 1995, 1998). There remain no other significant precipitating factors in ICOE-G if photosensitive patients are excluded. Despite FOS in EEG, only a few patients report seizure precipitation by going from bright light to darkness or by darkness itself (Beaumanoir et al., 1989).

Aetiology

There is an increased family history of epilepsies (21–37%) or migraine (9–16%) (Gastaut and Zifkin, 1987; Caraballo et al., 2008a) but familial ICOE-G appears to be rare (Nagendran et al., 1990; Grosso et al., 2008; Taylor et al., 2008).

Pathophysiology

The seizures are purely of occipital lobe origin. The epileptogenic zone involves networks within the occipital lobes and this localization is congruent with the symptomatogenic zone. Elementary visual hallucinations originate from the primary visual cortex, complex visual hallucinations from the junction of the occipital with the parietal and temporal lobes, formed visual illusions from the lateral occipital–posterior temporal junction and tonic deviation of the eyes from the medial occipital cortex, above or below the calcarine sulcus. Ictal blindness may reflect bi-occipital seizure spreading but this may not explain its sudden onset, without any other preceding manifestations. From the EEG standpoint, the occipital paroxysms are usually bilateral and synchronous because they are activated in both occipital regions by the elimination of fixation and central vision (FOS) (Panayiotopoulos, 1981) and not by thalamocortical activation proposed by Gastaut and Zifkin (1987).

The mechanisms for post-ictal headache are unknown. It is likely that the occipital seizure discharge triggers a genuine migraine headache through trigeminovascular or brain-stem mechanisms (Panayiotopoulos, 1999b, c).

Diagnostic procedures

By definition, all tests other than the EEG are normal. However, high-resolution MRI is mandatory, because symptomatic occipital epilepsy present with the same clinical EEG manifestations.

Electroencephalography

The inter-ictal EEG shows occipital paroxysms (Gastaut, 1982a; Gastaut and Zifkin, 1987), often demonstrating FOS (Panayiotopoulos, 1981) (Fig. 1). Because terminology is often unclear and FOS is not always tested, the prevalence of classical occipital paroxysms with FOS is uncertain and ranges from 100% (Gastaut and Zifkin, 1987), 88% (Caraballo et al., 2008a) to 19% (Panayiotopoulos, 1999a). Some patients may have only random occipital spikes, whereas others may have occipital spikes only in sleep EEG and some may have a consistently normal EEG (Panayiotopoulos, 1999b). Centro-temporal, frontal and GSES occur together with occipital spikes in around 20% of patients (Herranz Tanarro et al., 1984; Gastaut and Zifkin, 1987). IPS consistently elicits occipital spikes and/or generalized discharges in photosensitive patients.

As it happens with the rolandic spikes, occipital spikes are not pathognomonic of any particular syndrome, because they also occur in a variety of organic brain diseases with or without seizures, in children with congenital or early onset visual and ocular deficits, and even in 0.5–1.2% of normal pre-school age children (Gibbs and Gibbs, 1952, 1967; Kellaway, 1980). They are common in young children with a peak age at first discovery of 4–5 years, and ‘tend to disappear in adult life, and the subsidence of the EEG abnormality is usually accompanied by a cessation of seizures’ (Gibbs and Gibbs, 1952, 1967).

There are many reported ictal EEGs (Gastaut, 1982a; Aso et al., 1987; Gastaut and Zifkin, 1987; De Romanis et al., 1988, 1991; Beaumanoir, 1993a, b; Thomas et al., 2003). Seizure onset is preceded by regression of occipital paroxysms, and is characterized by the sudden appearance of an occipital discharge that consists of fast rhythms, fast spikes or both and is of much lower amplitude than the occipital paroxysms. Elementary visual hallucinations relate to the initially fast spike activity and complex visual hallucinations may occur when the ictal discharge is slower. In oculoclonic seizures, spikes and spike–wave complexes are slower, and a localized ictal fast spike rhythm may occur before deviation of the eyes. Ictal EEG during blindness is characterized by pseudo-periodic slow waves and spikes, which differ from those seen in ictal visual hallucinations. There are usually no post-ictal abnormalities.

Differential diagnosis

The differential diagnosis of ICOE-G is mainly from symptomatic occipital epilepsy, migraine with aura, acephalgic and basilar migraine where misdiagnosis is very high (Panayiotopoulos, 1999a, b).

Patients with symptomatic occipital epilepsy may often have identical symptoms as ICOE-G with normal neuro-ophthalmological examination and routine brain imaging. Thus, high-resolution MRI is required to detect subtle lesions (Kuzniecky et al., 1997). Occipital seizures of mitochondrial disorders, Lafora disease and coeliac disease should be considered (Panayiotopoulos, 1999a; Taylor et al., 2003).

The differential diagnosis of ICOE-G from migraine is usually easy if all clinical elements are properly assessed and synthesized (Table 1). Contrary to visual seizures, visual aura of migraine develops slowly within minutes, lasts long for 10–20 min and consists of mainly achromatic and linear patterns (Panayiotopoulos, 1994; Russell and Olesen, 1996; Schott, 2007). Illustration of the visual symptoms of the attacks by the patient is a powerful tool in differential diagnosis and objective analysis. Orbital pain in the ictal phase of visual hallucinations is typical of occipital seizures and does not occur in migraine. However, post-attack headache is common and similar for both occipital epilepsy and migraine. Basilar migraine attacks also develop slowly within minutes, last for 30–60 min and consist of mainly bilateral impairment of vision associated with, or followed by, neurological symptoms such as vertigo, tinnitus, ataxia, bilateral weakness and dysaesthesiae, which do not occur in occipital lobe epilepsy (Panayiotopoulos, 1999c).

View this table:
Table 1

Key differences between occipital seizures and migraine

Elements of visual hallucinationsOccipital seizuresMigraine
Speed of development from onset to full imageFast in secondsSlow in minutes
Speed and direction of movementFast in seconds and usually towards the centre of the visual field and contalateral to the side of onsetSlow in minutes and usually towards the periphery of the visual field and ipsilateral to the side of onset
QualityUsually with bright colours and circular shapesUsually achromatic or black and white linear zigzag patterns
DurationUsually 1–3 minUsually over 15 min
Progression to transient neurological symptomsEye deviation, eyelid closures and sometimes convulsionsScotoma, hemianopia, hemi-anaesthesia or hemi-paresis and for basilar migraine, vertigo, ataxia, bilateral weakness and dysaesthesiae

Mistaking visual seizures as migraine attacks may be common in publications referring to controversial diagnostic terms such as ‘migralepsy’ and ‘basilar migraine with occipital paroxysms’. A critical review of such reported cases indicates that these are likely to be genuine occipital seizures imitating migraine (Panayiotopoulos, 1999c).

Despite some overlapping features, ICOE-G is distinctive from PS (Table 2) and the differences have been statistically validated (Panayiotopoulos, 1999a). As a rule of thumb seizure onset is primarily with visual symptoms in ICOE-G and with autonomic manifestations in PS. On the rare occasion of ICOE-G with autonomic manifestations, these always occur after occipital lobe symptomatology.

View this table:
Table 2

Main features of rolandic epilepsy, Panayiotopoulos syndromes and idiopathic childhood epilepsy of Gastaut

Rolandic epilepsyPanayiotopoulos syndromeIdiopathic childhood occipital epilepsy of Gastaut
Prevalence amongst children aged 1–15 years with non-febrile seizures (%)1560.5–1
Peak age at onset (years)7–103–68–11
Male to female ratio1.511
Seizure characteristics
Typical onset withHemifacial sensory-motor or oropharyngolaryngeal symptomsAutonomic symptoms mainly with emesisVisual symptoms mainly with elementary visual hallucinations
Hemifacial sensory-motor symptomsCommon and often from onsetRare and not from onsetRare and not from onset
Oropharyngolaryngeal symptomsCommon and often from onsetRare and not from onsetHave not been reported
Speech arrestCommon and often at onsetRare and not from onsetHas not been reported
HypersalivationCommon and often at onsetRare and not from onsetHas not been reported
Ictus emeticusRare and not from onsetCommon and often at onsetRare and not from onset
Autonomic disturbances other than vomiting and hypersalivationExceptional and not from onsetCommon and often at onsetExceptional and not from onset
Visual symptomsHave not been reported7% but exceptional at onsetCommon and often at onset
Deviation of the eyesFrequent during sensory-motor symptomsCommon but rarely at onsetCommon but rarely at onset
Ictal behavioural changesExceptional and not from onsetCommon and often at onsetHave not been reported
Duration for 1–3 minAs a ruleRareAs a rule
Duration of more than 5 minRareCommonRare
Partial status epilepticus (>30 min)Exceptional40%Exceptional
Total number of seizures 1–15As a ruleAs a ruleRare
Single seizures only (%)10–2030Exceptional
Frequent seizures (%)101090
Nocturnal (sleep only) (%)7064Exceptional
Febrile convulsions181710
PrognosisExcellentExcellentUncertain
Remission within 1-2 years from first seizureCommonCommonExceptional or rare
Seizures after the age of 13 yearsRareExceptionalCommon
Interictal EEG
Centrotemporal spikes aloneAs a rule and characteristicRareHave not been reported
Occipital spikesHave not been reported65%Probably 90%
Spikes in other locationsProbably uncommonFrequentExceptional
Brief generalised discharged of 3–5 Hz slow waves with small spikes (%)510Exceptional
Somatosensory evoked spikesCommonRareHave not been reported
Fixation-off sensitivityHas not been reportedRareMay be less common than reported
PhotosensitivityHas not been reportedExceptionalProbably 20–30%
Normal EEG or focal slow after first seizure (%)∼10∼10∼10
Ictal EEGSlow activity with spikesSlow activity with spikesFast spikes and fast rhythms
Ictal onsetRolandic regionsAnterior or posterior regionsOccipital regions

Prognosis

The prognosis of ICOE-G is unclear, although available data indicate that remission occurs in 50–60% of patients within 2–4 years of onset (Gastaut and Zifkin, 1987; Panayiotopoulos, 1999b; Caraballo et al., 2008a). Seizures show a dramatically good response to carbamazepine in >90% of patients. However, 40–50% of patients may continue having visual seizures and infrequent secondarily GTCS. Rarely, atypical evolutions to epilepsy with CSWS and cognitive deterioration have been reported (Tenembaum et al., 1997). Also, rarely children with ICOE-G may manifest with typical absence seizures, which usually appear after the onset of occipital seizures (Caraballo et al., 2005).

The performance scores for attention, memory and intellectual functioning were lower in patients with ICOE-G than control subjects though basic neurophysiological functions did not differ significantly (Gulgonen et al., 2000).

Other phenotypes of BCSSS

There are reports of children suffering from benign childhood focal seizures with clinical EEG manifestations that cannot be classified as rolandic epilepsy, PS or ICOE-G. They may represent rare, atypical or overlapping presentations of BCSSS.

Benign childhood seizures with affective symptoms

Benign childhood epilepsy with affective symptoms, reported in less than 40 patients, is a clinical phenotype of BCSSS with features common in both PS (behavioural and autonomic symptoms) and rolandic epilepsy (speech arrest and hypersalivation) (Dalla Bernardina et al., 1992, 2007). Onset is between 2 and 9 years of age and both sexes are equally affected.

Seizures manifest with terror and screaming, autonomic disturbances (pallor, sweating, abdominal pain and hypersalivation), chewing and other automatisms, speech arrest and mild impairment of consciousness. These are usually brief, lasting 1–2 min, and frequent, occurring several times a day in wakefulness or sleep. A fifth of patients have febrile seizures and some may also have infrequent rolandic seizures. Generalized seizures have not been reported.

The inter-ictal EEG shows high-amplitude frontotemporal and parietotemporal spikes that are exaggerated by sleep. Ictal EEG discharges are mainly localized in the frontotemporal, centro-temporal or parietal regions and are stereotypical for each patient.

The response to treatment is excellent and remission occurs within 1–2 years from onset. Behavioural problems may be prominent during the active stage of the disease, but subside later with seizure remittance.

Benign childhood epilepsy with parietal spikes and frequent extreme somatosensory-evoked spikes

Benign childhood epilepsy with parietal spikes and frequent GSES (De Marco and Tassinari, 1981; Tassinari and De Marco, 1992; Fonseca and Tedrus, 2000) has been proposed as another phenotype of BCSSS. The defining features are EEG spikes in the parietal regions, which are often elicited by tactile stimulation (Fig. 1). However, GSES are not specific for any syndrome because they also occur in 10–20% of children with rolandic seizures (Fonseca and Tedrus, 2000), in a few patients with PS (Panayiotopoulos, 1999a, 2002) and in children with no seizures (Negrin and De Marco, 1977).

Versive seizures of the head and body, often without impairment of consciousness, are mainly diurnal and infrequent. Frequent seizures and focal status epilepticus are exceptional.

Remission usually occurs within 1 year from seizure onset, but EEG abnormalities may persist for longer.

Benign childhood focal seizures associated with frontal or midline spikes

Benign childhood focal seizures associated with frontal (Beaumanoir and Nahory, 1983; Martin-Santidrian et al., 1998; Panayiotopoulos, 1999a) or midline spikes (Bagdorf and Lee, 1993; Panayiotopoulos, 1999a) have been described and long follow-up reports have confirmed a benign course, although no systematic studies have been published. However, EEG spike foci specificity is questionable, as spike foci of various locations (including frontal and midline) are also seen in rolandic epilepsy and more frequently in PS, and midline spikes are more common in children than in adults (Kutluay et al., 2001; Sanders et al., 2002).

Recently, ‘benign infantile focal epilepsy with midline spikes during sleep’ has been described as a new syndrome of BCSSS (Bureau et al., 2002; Capovilla et al., 2006). Age at onset is in the first 3 years of life and both sexes are equally affected. Seizures consist mainly of staring, motion arrest, cyanosis, loss of consciousness and stiffening of the arms. Clonic convulsions and automatisms are rare. Seizures are brief from 1 to 5 min, mainly diurnal and are generally infrequent from one to three per year. There is a strong family history of undefined types of epileptic seizures with benign epilepsies prevailing.

Inter-ictal EEG abnormalities are seen only in non-REM sleep and consist of small, mostly singular, midline spikes. The prognosis is excellent, with remission of seizures, normal development and normalization of the EEG before the age of 4 years.

Differential diagnosis between seizures and syndromes of BCSSS

The differential diagnosis between the main phenotypes of BCSSS is easy in their typical presentations (Table 2). Problems may arise in children with clinical symptoms that fall into two (or more) phenotypes or from overemphasizing on EEG localization. As in any other medical condition, a single symptom is of limited syndromic significance. The differential diagnosis requires that symptoms are meaningfully synthesized in regard to quality and quantity, chronological sequence, consistency, relation to other seizure manifestations, the circumstances of their appearance and the overall clustering of clinical EEG manifestations.

Rolandic epilepsy versus PS

Their differential diagnosis is usually easy (Table 2). However, there are some cases with overlapping features:

  1. One-tenth of children with PS often have typical and lengthy autonomic seizures with concurrent rolandic features such as speech arrest, hemifacial convulsions, hypersalivation and OPS but these appear after the onset of autonomic symptoms and emesis (Ferrie et al., 1997; Oguni et al., 1999; Kivity et al., 2000; Panayiotopoulos, 2002; Lada et al., 2003; Caraballo et al., 2007). Conversely, these ictal symptoms occur at onset and usually without autonomic symptoms in rolandic epilepsy.

  2. One-tenth of children with PS develop pure rolandic seizures, either in parallel with autonomic seizures, or at a later age prior to final remission (Oguni et al., 1999; Panayiotopoulos, 2002; Lada et al., 2003; Caraballo et al., 2007).

  3. The topography of interictal spikes may overlap. Covanis et al. (2003) studied 24 otherwise normal children with focal non-febrile seizures who had emetic manifestations in at least one seizure and CTS in at least one EEG; 21 (83%) had ictal semiology typical of PS but five also had concurrent rolandic symptoms and four later developed pure rolandic seizures. The other four children (17%) had typical rolandic seizures with concurrent ictus emeticus. Ohtsu et al. (2007) found that in early-onset rolandic epilepsy vomiting usually happened in the middle of the ictus, seizures, neurocognitive and behavioural abnormalities were more frequent, while focal status epilepticus and prolonged seizures were less common than in PS.

  4. Of siblings one may have rolandic seizures and another PS, and there is a high prevalence of febrile seizures in both (Ferrie et al., 1997; Oguni et al., 1999; Panayiotopoulos, 2002; Caraballo et al., 2007).

Idiopathic childhood occipital epilepsy of Gastaut versus PS

The differentiation here is straightforward (Table 2) and statistically validated (Panayiotopoulos, 1999a). The seizures of ICOE-G are purely occipital and as such start and often end only with occipital lobe symptomatology. Further, seizures are mainly brief, frequent and diurnal. Rarely, seizures may be longer and also occur in sleep but these are also fundamentally different to the rolandic epilepsy or the autonomic seizures and autonomic status epilepticus of PS.

Exceptionally, ictal vomiting may occur in ICOE-G but this follows the appearance of visual symptomatology as it happens with reflex photosensitive occipital seizures (Guerrini et al., 1995, 1998) and the same patient usually has frequent brief occipital seizures. Conversely, visual symptoms in PS, when present, are not the sole manifestation of a seizure or stereotypical; only exceptionally (1%) are they reported at seizure onset (Ferrie et al., 1997; Yalcin et al., 1997; Panayiotopoulos, 2002; Caraballo et al., 2007). From the EEG standpoint, occipital spikes which characterize ICOE-G are also common in PS but these often occur with extra-occipital spikes and with shifting locations in sequential EEG (Fig. 1). Further, ictal EEG is markedly different between these syndromes (Fig. 2).

Reported difficulties in differentiating ICOE-G from PS (Genizi et al., 2007; Taylor et al., 2008) may arise when emphasis is unduly placed on individual symptoms that may overlap rather than a comprehensive synthetic analysis of their quality, chronological sequence and other clustering features in the respective electro-clinical phenotypes, which is the basis for precise differential diagnosis in clinical practice. If any other diagnostic approach is followed, then even non-epileptic disorders such as migraine with aura could be deemed as overlapping with ICOE-G (visual hallucinations and headache), PS (lengthy duration and vomiting) or both (age, family history of epilepsies) (Table 1). It may be because of these limitations and the retrospective character of their study that Taylor et al. (2008) found that only one of their 16 patients was typical in all respects of PS and that ICOE-G was as frequent as PS, which contrasts all previous prospective studies detailed in this assessment. Such a discrepancy may indicate that PS is still unrecognized and that the study does not represent the vast majority of typical PS.

Further, the commonly quoted argument that PS is not essentially different from ICOE-G considering that ‘the younger the children are, the less likely they are to describe visual symptoms’ (Andermann and Zifkin, 1998) is not tenable: (i) more than two-thirds of children with PS are older than 4 years and therefore able to describe their visual experiences and (ii) there is no difference in seizure presentation between younger and older children with PS.

A few patients with either PS or rolandic epilepsy may later develop purely occipital seizures as of ICOE-G (Guerrini et al., 1997; Parmeggiani and Guerrini, 1999; Panayiotopoulos, 2002). These cases are easy to diagnose and indicate the intimate links of these disorders within the framework of BCSSS.

Benign childhood seizure susceptibility syndrome: a unified concept of benign childhood focal seizures

Rolandic epilepsy, PS, ICOE-G and other possible clinical phenotypes of benign childhood focal seizures are likely to be linked together by a genetically determined, functional derangement of the systemic brain maturation that is mild and age related (benign childhood seizure susceptibility syndrome) (Panayiotopoulos, 1993, 2002). They have distinctive characteristics but they also share common clinical and EEG features: seizures are infrequent, usually nocturnal and remit within a few years from onset. Brief or prolonged seizures, even focal status epilepticus, may occur only once in the patient's lifetime. Despite the distinctiveness of their core clinical and EEG features, the natural histories of these syndromes may show significant reciprocity: some children with rolandic epilepsy may present autonomic seizures referable to PS (and vice versa) before remittance, while other may have alternate autonomic and rolandic seizures. Some seizures may be of mixed character, and certainly ictal autonomic manifestations, such as hypersalivation, emesis, headache and syncopal-like attacks that are unusual in other epileptic syndromes in children or adults, are frequent in BCSSS, and may predominate. Affected siblings may have the same or another type of benign childhood focal seizures, and febrile seizures are common. EEG spikes are regional (bilateral and multifocal) than focal—and as a rule disproportionately abundant to the frequency of seizures—and there is a significant overlap of interictal topographies.

There is no reason to suggest that these syndromes differ merely because an ‘epileptogenic’ focus is slightly anterior or posterior, lateral or medial to the central regions. The relevant ictal semeiologies and EEG findings suggest that each one of these forms reflects constitutional hyperexcitability of a particular functional brain area or system: the lower rolandic (somatosensory) cortex that represents the face and the oropharynx bilaterally in benign rolandic epilepsy, the occipital areas (cortical visual system) in ICOE-G and the central autonomic network bilaterally and diffusely in PS (Koutroumanidis, 2007). All these conditions appear to be linked together by a genetically determined, functional derangement of the systemic brain maturation that is mild and age related (Panayiotopoulos, 1993, 1999a). This derangement is often clinically silent and presents in >90% of the susceptible children only with—also age related—EEG sharp and slow waves; the remaining one-tenth of these children have infrequent focal seizures. A small number of susceptible children, with or without seizures, may also have minor and fully reversible neuropsychological symptoms that are rarely clinically overt and can be detected only by formal neuropsychological testing. Finally, in a very small number of patients (probably <1%) this disturbance of brain maturation may further evolve into a more aggressive clinical state with seizures, neuropsychological manifestations and EEG abnormalities of various combinations and severity, such as atypical benign focal epilepsy of childhood, Landau–Kleffner syndrome and epilepsy with CSWS.

This concept of BCSSS is in agreement with previously expressed views of ‘functional epilepsies of maturation’ (Sorel and Rucquoy-Ponsar, 1969), ‘multifactorial pathogenesis of epilepsies with benign focal epileptiform sharp waves’ (Doose et al., 1997, 2002), ‘selective rates of maturation of the different cortical areas (Luders et al., 1987)’ and more recently of possible ‘neurobiological spectrum’ between BCSSS and IGE (Taylor et al., 2003, 2008).

BCSSS, febrile and other idiopathic focal seizures in neonates and infants

One of the most interesting aspects of benign childhood seizures is their striking age-related sequence that appears to reflect enhanced epileptogenicity of the developing brain, as a whole and also of its functional systems, in different stages of maturation. Benign neonatal and infantile seizures, rolandic epilepsy, PS, ICOE-G and other clinical phenotypes of BCSSS are specific to early life and do not occur in adults. This is also the case with most febrile seizures whose different genetic influences may explain their high prevalence amongst patients with BCSSS and other more severe types of epilepsy, including the febrile plus phenotypes and genotypes (Scheffer et al., 2007; Harkin et al., 2007). It appears that there are three main periods of age-related childhood susceptibility to benign seizures: febrile, mainly generalized, convulsions first appear in early childhood at a peak age of around 18–22 months. Rolandic epilepsy and ICOE-G manifest with purely focal seizures and occur in late childhood age. PS presents with mainly autonomic seizures and covers the intermediate early childhood period with peak at 4 or 5 years. The neonatal and early infantile periods are not immune to focal seizure susceptibility either, as indicated by the benign neonatal seizures of the first few days of life, (Plouin and Anderson, 2005) and the benign infantile focal seizures of Watanabe and Vigevano (Vigevano et al., 2008). This point is exemplified by reports of children with neonatal seizures who later developed rolandic epilepsy (Maihara et al., 1999) or PS (Lada et al., 2003; Panayiotopoulos, 2007).

BCSSS, Landau–Kleffner syndrome, epilepsy with CSWS and atypical benign partial epilepsy of childhood

Landau–Kleffner syndrome and epilepsy with CSWS (Beaumanoir et al., 1995) are now considered by the ILAE as an entity named ‘epileptic encephalopathy of CSWS including Landau–Kleffner syndrome’ with a common pathophysiological mechanism (Engel, 2006). Epileptic encephalopathy of CSWS including Landau–Kleffner syndrome are functional disorders occurring at an age where cortical synaptogenesis with abundant axonal sprouting and elemental functional network is being established in the brain. Aggressive epileptogenic activity at this active period of brain organization is detrimental for the establishment of appropriate neuronal connections, normal brain development and functioning (Smith and Hoeppner, 2003). These disorders may constitute a rare and extreme derailment of BCSSS. EEG manifests with abundant and often continuous high amplitude sharp waves morphologically similar to the CTS and occipital paroxysms. Seizures are predominantly nocturnal and often resemble rolandic seizures. Otherwise typical rolandic epilepsy (Fejerman et al., 2000), PS (Caraballo et al., 2001, 2005; Ferrie et al., 2002) and ICOE-G (Tenembaum et al., 1997) evolve to clinical and EEG features of epileptic encephalopathy of CSWS including Landau–Kleffner syndrome and atypical partial epilepsy of childhood (Saltik et al., 2005). Atypical benign partial epilepsy of childhood (Aicardi and Chevrie, 1982) is considered either a mild form of epileptic encephalopathy of CSWS (Arzimanoglou et al., 2004) or a more severe form of rolandic epilepsy (Doose et al., 2001; Hahn et al., 2001). The reason for this derailment of BCSSS is unknown, but may be related to location, epileptogenic threshold and other intrinsic genetic and external superimposed factors. Intense epileptic activity in the dominant temporal region would affect linguistic capabilities as in Landau–Kleffner syndrome (Morrell et al., 1995). Conversely, the mainly frontal localization of CSWS primarily affects higher cognitive and executive functioning (Galanopoulou et al., 2000; Smith and Hoeppner, 2003; Tassinari et al., 2005).

BCSSS and idiopathic generalized epilepsies

The majority of BCSSS if properly diagnosed do not have any clinical or EEG resemblance to idiopathic generalized epilepsies, though others may disagree (Taylor et al., 2003, 2008). Overlap of BCSSS with IGE is limited (Kivity et al., 2000). However, a possible link, the type and extent of which should be explored further with clinical and genetic studies may be suggested by

  1. the occurrence of EEG generalized discharges in BCSSS (though these are often markedly different from the classical generalized spike or polyspike discharges of IGE);

  2. an undetermined but probably small proportion of patients with any type of BCSSS that may also suffer typical generalized convulsive or absence seizures either during the active phase of BCSSS or more often at a later stage and

  3. an undetermined but probably small proportion of patients with syndromes of IGE including childhood absence epilepsy that may also have EEG focal spikes only or together with any type of seizures of BCSSS (Beaumanoir et al., 1974; Taylor et al., 2003, 2008; Caraballo et al., 2008b).

Management of benign childhood focal seizures

Short- and long-term treatment strategies of benign childhood focal seizures are largely empirical.

In the acute stage, control of the seizure is of paramount importance. Benzodiazepines are used to terminate long-lasting seizures (>10 min) and status epilepticus. Autonomic status epilepticus needs thorough evaluation for proper diagnosis and assessment of the neurological/autonomic state of the child. Aggressive treatment should be avoided because of the risk of iatrogenic complications (Ferrie et al., 2006).

Continuous anti-epileptic medication is not usually recommended. Although there may be effective AED that may prevent the occurrence of additional seizures, potential adverse effects may not commensurate with the benefit. Decisions on management must take into account the following (Bourgeois, 2000; Wirrell et al., 2006; Panayiotopoulos, 2007; Fejerman, 2008):

  1. Most children have excellent prognosis: 10–30% may have only a single seizure and 60–70% may have less than 10 in total. However, 10–20% of children may have frequent seizures, which are sometimes resistant to treatment.

  2. Remission of benign childhood focal seizures is expected in all patients by the age of 15–16 years at the latest.

  3. There is no evidence that the long-term prognosis is worse in untreated children (Ambrosetto and Tassinari, 1990), although they may not be protected against seizure recurrences and particularly GTCS (Peters et al., 2001).

  4. Some children become frightened, even by simple focal seizures, and some parents are unable to cope with the possibility of another fit despite firm reassurances.

  5. Persistence and frequency of EEG functional spikes do not predict clinical severity, frequency or degree of liability to seizures

  6. In contrast to the other phenotypes of the BCSSS, patients with ICOE-G often suffer from frequent seizures and therefore prophylactic AED treatment may be mandatory. Secondarily, GTCS are probably unavoidable without medication.

Continuous prophylaxis consists of daily monotherapy using any AED that has proven efficacy in focal seizures and minimal adverse effects in children. The 2006 ILAE treatment guidelines found that ‘no AED had level A or level B efficacy and effectiveness evidence as initial monotherapy’ in rolandic epilepsy (Glauser et al., 2006). Of older AED most authorities prefer carbamazepine in USA (Wheless et al., 2005) and valproic acid in Europe (Wheless et al., 2007), though these may have equivalent efficacy with all others (phenobarbital, phenytoin and clonazepam) (Ferrie et al., 1997; Bouma et al., 2002); carbamazepine may exaggerate seizures in a minority of children with BCSSS including PS (Kikumoto et al., 2006) and valproate is associated with significant adverse effects. Recently, sulthiame (available only in a few countries) has been revived as an excellent drug for the treatment of rolandic epilepsy with EEG normalization (Bast et al., 2003; Ben Zeev et al., 2004) but this may be associated with cognitive abnormalities (Wirrell et al., 2008). Recommended newer AED include levetiracetam (Verrotti et al., 2007; Kossoff et al., 2007), gabapentin (Bourgeois, 2000) and lamotrigine (Wheless et al., 2005). Lamotrigine on rare occasions may cause seizure exacerbation and cognitive deterioration (Cerminara et al., 2004).

Opinions of when to withdraw medication differ among experts, although all agree that there is no need to continue with AED after the age 14 when most benign childhood focal seizures remit or 16 when they are practically non-existent; a common recommendation is to start slow tapering the AED after a 1–3 years seizure free period (Bourgeois, 2000).

Parental needs and support

Despite their excellent prognosis, benign childhood focal seizures usually have a dramatic impact to parents. The most dominant points of parental anxiety and concerns refer to their uncertainty of the nature, the cause and the impact of the events on their child's development as well as the lack of appropriate information (Valeta, 2005). There is a need for supportive family management, education and specific instructions about emergency procedures for possible subsequent seizures. Education about BCSSS is the cornerstone of the optimal management. Parents should be given general information about benign childhood focal seizures and, in particular PS, in which seizures may have dramatic features and last for many hours; the situation is frequently compounded by physicians’ uncertainty over diagnosis, management and prognosis. Parents who have watched their child during a seizure may need specific psychological support to overcome anxiety and panic that may result in overprotection and interfere in turn with parent–child separation and independence (Valeta, 2007).

Conclusions and recommendations

It is apparent from this assessment that BCSSS is a formidable challenge in childhood epileptology. Despite significant progress made, we may have presented only the tip of the iceberg and it is likely that significant pieces are still missing or we may have misplaced them in the jigsaw puzzle. There are many practical, diagnostic, genetic, aetiological, electrophysiological, pathophysiological, theoretical, epidemiological, social and management issues still to be resolved. This will require a prospective multidisciplinary and evidence-based approach. Significant evidence of the processes underpinning these diseases may emerge and our old notions discarded. Furthermore, revisions should be proposed in the best spirit of unbiased work with an open-minded and dispassionate scientific decorum. Meanwhile, from the clinical point of view their phenotypic recognition and their differentiation from other epileptic and non-epileptic disorders have significant prognostic and management implications.

Footnotes

  • Abbreviations:
    Abbreviations:
    AED
    anti-epileptic drugs
    BCSSS
    benign childhood seizure susceptibility syndrome
    CTS
    centro-temporal spikes
    CSWS
    continuous spike and wave during sleep
    FOS
    fixation off sensitivity
    GSES
    giant somatosensory evoked spikes
    GTCS
    generalized tonic–clonic seizures
    ICOE-G
    idiopathic childhood occipital epilepsy of Gastaut
    ILAE
    International League against Epilepsy
    IPS
    intermittent photic stimulation
    OPLS
    Oro-pharyngo-laryngeal symptoms
    PS
    Panayiotopoulos syndrome

References

View Abstract