JLE

Epileptic Disorders

MENU

Controversies regarding neonatal seizure recognition Volume 4, numéro 2, June 2002

Diagnostic dilemmas: reliance on clinical versus EEG criteria

Since neonatal seizures are generally brief and subtle in clinical appearance, unusual behaviors may be difficult to recognize and classify. While some claim that seizures are relatively common events in symptomatic neonates suffering from medical conditions, medical personnel vary significantly in their ability to recognize suspicious behaviors, contributing to either overdiagnosis or underdiagnosis. The most common practice in many neonatal ICUs is to rely on clinical behaviors to identify seizures, without EEG confirmation. Motor or autonomic behaviors however, may represent either normal, age and state-specific behaviors in healthy infants. Alternatively, nonepileptic paroxysmal conditions are noted with encephalopathic infants. Therefore, confirmation of suspicious clinical events as seizures, using EEG recordings is now strongly recommended. While routine EEG studies [7] may miss isolated seizures in certain patients, continuous synchronized video/EEG/polygraphic recordings establish a more reliable start and endpoint of electrically-confirmed seizures [8]. Rigorous physiological monitoring also assists the clinician to integrate seizure diagnosis with etiology and the neurobiology of the immature brain.

Clinical seizure criteria

Neonatal seizures are presently listed separately from the traditional classification of seizures and epilepsy during childhood. The International League Against Epilepsy's classification adopted by the World Health Organization still considers neonatal seizures within an unclassified category [9]. A recent classification scheme now suggests a more strict distinction of clinical seizure (nonepileptic) events from electrographically-confirmed (epileptic) seizures, with respect to possible treatment [10]. Continued refinement of novel classifications are needed to reconcile disagreements between clinical and EEG criteria which impede seizure diagnosis [11, 12], in the context of nonepileptic movement disorders caused by acquired diseases, malformations and/or medications.

Several caveats (table I), may be applied to the identification of suspected neonatal seizures.

The clinical criteria for neonatal seizure diagnosis was historically subdivided into five clinical categories: focal clonic, multifocal or migratory clonic, tonic, myoclonic and subtle seizures [13]. A more recent classification expands the clinical subtypes, adopting a strict temporal occurrence of specific clinical events with coincident electrographic seizures, to distinguish neonatal clinical "nonepileptic" seizures from "epileptic" seizures (table II) [10].

Subtle seizure activity

This is the most frequently observed category of neonatal seizure. Repetitive, buccolingual movements, orbital-ocular movements, unusual bicycling or pedaling and autonomic findings are examples of this seizure category (figure 1A). Any subtle paroxysmal event which interrupts the expected behavioral repertoire of the newborn infant, and appears stereotypical or repetitive should alert the clinician to the possibility of seizures. However, alterations in cardiorespiratory regularity, body movements and other behaviors during normal active (REM), quiet (NREM) sleep or waking states must first be considered before initiating a seizure evaluation [14, 15].

Within the subtle category of neonatal seizures are stereotypical alterations in heart rate, blood pressure, oxygenation or other autonomic signs which are particularly helpful during pharmacological paralysis for ventilatory care. Unusual autonomic events include penile erections, skin changes, salivation, and lacrimation. Autonomic phenomena are often intermixed with motor findings. Isolated autonomic signs such as apnea, unless accompanied by other clinical findings are rarely associated with coincident electrographic seizures (figure 1B) [16, 17]. Since subtle seizures are both clinically difficult to detect and are only variably coincident with EEG seizures, synchronized video/EEG/polygraphic recordings are recommended to document temporal relationships between clinical behaviors and coincident electrographic events [8, 10, 18, 19]. Despite the "subtle" expression of this seizure category, these children may have suffered significant brain injury.

Clonic seizures

Rhythmic movements of muscle groups in a focal distribution, which consist of a rapid flexion phase followed by a slower extensor movement may be clonic seizures, to be distinguished from the symmetric "to and fro" movements of nonepileptic tremulousness or jitteriness [2]. Gentle flexion of the affected body part easily suppresses the tremor, while clonic seizures persist. Clonic movements can involve any body part such as the face, arm, leg, and even diaphragmatic or pharyngeal muscles. Generalized clonic activities can occur in the newborn but rarely consist of classical tonic followed by clonic phases, characteristic of the generalized motor seizure noted in older children and adults. Focal clonic and hemiclonic seizures have been described with localized brain injury, usually from cerebrovascular lesions [8, 20-22] (figure 2A) but they can also be seen with generalized or multifocal brain abnormalities. As with older patients, focal seizures in the neonate can be followed by transient motor weakness, historically referred to as a transient Todd's paresis or paralysis [23], to be distinguished from a more persistent hemiparesis over multiple days to weeks. Clonic movements without EEG-confirmed seizures have been described in neonates with normal EEG backgrounds and their neurodevelopment outcome can be normal [18]. The less experienced clinician may misclassify myoclonic as clonic movements. Recent computational studies suggest strategies to extract quantitative information from video recordings of neonatal seizures as a method by which clinicians can differentiate myoclonic from focal clonic seizures, as well as from normal infant behaviors [24].

Multi-focal (fragmentary) clonic seizures

The word "fragmentary" was historically applied to distinguish this event from the more classical generalized tonic-clonic seizure seen in the older child. Multifocal or migratory clonic activities spread over body parts, either in a random or anatomically appropriate fashion. Seizure movements may alternate from side to side, and appear asynchronously between the two halves of the child's body. Multifocal clonic seizures may be misclassified as myoclonic seizures, and video-EEG documentation helps with proper identification. Neonates with this seizure description often suffer death or significant neurological morbidity [25].

Tonic seizures

Tonic seizures refer to a sustained flexion or extension of axial or appendicular muscle groups. Tonic movements of a limb or sustained head or eye turning may also be noted. Documentation of tonic activity during coincident EEG recordings are needed, since 30% of such movements lack a temporal correlation with electrographic seizures [26] (figures 3A, B, C). Such nonepileptic activity is referred to as "brainstem release" resulting from functional decortication after severe neocortical dysfunction or damage. Extensive neocortical damage or dysfunction permits the emergence of uninhibited subcortical expressions of extensor movements [27]. Tonic seizures may also be misidentified when nonepileptic movement disorders consisting of dystonia are more appropriate behavioral descriptions (figure 6). Both tonic movements and dystonic posturing may also occur simultaneously.

Myoclonic seizures

Myoclonic movements are rapid, isolated jerks which can be generalized, multifocal or focal in an axial or appendicular distribution. Myoclonus lacks the slow return phase of the clonic movement complex. Healthy preterm infants commonly exhibit myoclonic movements without seizures or a brain disorder. EEG, therefore, is recommended to confirm the coincident appearance of electrographic discharges with these movements (figure 4A). Pathological myoclonus in the absence of EEG seizures can also occur in severely ill preterm or fullterm infants after suffering severe brain dysfunction or damage [28]. As with older children and adults, myoclonus may reflect injury at multiple levels of the neuraxis from the spinal cord, brainstem to cortical regions. Stimulus-evoked myoclonus with either coincident single coincident spike discharges or sustained electrographic seizures have been reported [29] (figures 4A and B). An extensive evaluation must be initiated to exclude metabolic, structural and genetic causes. Rarely, healthy sleeping neonates exhibit abundant myoclonus which subside with arousal to the waking state [30, 31]. These are termed benign sleep myoclonus of the newborn.

Nonepileptic behaviors of neonates

Nonepileptic neonatal movement repertoires continue to challenge the physician's attempt to accurately diagnose seizures. Coincident synchronized video/EEG/polygraphic recordings are now suggested to confirm the temporal relationships between the suspicious clinical phenomena and electrographic expression of seizures [32]. The following three examples of nonepileptic movement disorders incorporate a new classification scheme [10], based on the absence of coincident EEG seizures.

Tremulousness or jitteriness without coincident EEG seizures

Tremors are frequently misidentified as clonic activity by inexperienced medical personnel. The flexion and extension phases of tremor are equal in amplitude, unlike the unequal phases of clonic movements. Children are generally alert or hyperalert but may also appear somnolent or lethargic as part of an encephalopathy. Passive flexion and repositioning of the affected tremulous body part diminishes or eliminates the movement. Such movements are usually spontaneous but can be provoked by tactile stimulation. This movement may also appear asymmetric with less expression in a weak limb following a brain injury or peripheral neuropathy. Metabolic or toxin-induced encephalopathies including mild asphyxia, drug withdrawal, hypoglycemia-hypocalcemia, intracranial hemorrhage, hypothermia and growth restriction are common clinical scenarios when tremulous movements occur. Neonatal tremors generally diminish with increasing post-conceptional age. For example, thirty-eight full-term infants with excessive tremulousness resolved spontaneously over the next six-week period, with 92% neurologically normal at three years of age [33]. Medication is rarely considered to treat this particular movement disorder [34].

Neonatal myoclonus without coincident EEG seizures

Myoclonic movements can be bilateral and synchronous, or asymmetric and asynchronous in appearance. Clusters of myoclonic activity occur predominantly during active (REM) sleep, and are predominant in the preterm infant [15, 35] (figure 4B), as well as rarely in healthy full-term infants. Benign myoclonic movements are not stimulus-sensitive, have no coincident electrographic seizure correlates and are not associated with EEG background abnormalities. When these movements occur in the healthy full-term neonate, the activity is suppressed during wakefulness. The clinical description of benign neonatal sleep myoclonus must be a diagnosis of exclusion, after a careful consideration of pathological diagnoses [30].

Infants with severe central nervous system dysfunction may also present with nonepileptic, spontaneous or stimulus-evoked pathological myoclonus. Neonates with different forms of metabolic encephalopathies such as glycine encephalopathy, cerebrovascular lesions, brain infections or congenital malformations can present with nonepileptic pathological myoclonus (figure 4C) [28]. Encephalopathic neonates may respond to tactile or painful stimulation by either isolated focal, segmental or generalized myoclonic movements. Rarely, cortically-generated spike or sharp wave discharges, as well as seizures, may also be noted on the EEG recordings which are coincident with these myoclonic movements (figures 5A and B) [28]. Medication-induced myoclonus, as well as other stereotypical movements, have also been described [36], which resolve when the drug is withdrawn.

A rare familial disorder termed hyperekplexia has been described in the neonatal and early infancy periods. These movements usually are misinterpreted as a hyperactive startle reflex, or whole body myoclonus. Infants are also stiff, with severe hypertonia which may lead to apnea and bradycardia. Forced flexion of the neck or hips sometimes alleviates these events. EEG background rhythms are generally age-appropriate. The postulated defect for these individuals involves regulation of brainstem centers which facilitate myoclonic movements [37], but molecular defects of ion channels are also considered. Occasionally, benzodiazepines or valproic acid lessen the startling, stiffening or falling events [38]. Neurological prognosis is reported to be variable.

Neonatal dystonia
or choreoathetosis without coincident EEG seizures

Dystonia and choreoathetosis are commonly occurring movement disorders that are often misrepresented as seizures. These nonepileptic movement disorders are associated with either acute or chronic disease states involving basal ganglia structures, or the extrapyramidal pathways which innervate these regions. Antepartum or intrapartum adverse events such as from severe asphyxia can damage the basal ganglia (i.e. termed status mamoratus) [13], and rarely, specific inherited metabolic diseases [39, 40] (e.g. glutaric aciduria) can also injure these structures. Sustained posturing or dystonia may also reflect subcortical motor pathways that become functionally unopposed because of diseased or malformed neocortex [27] (figure 6). Documentation of electrographic seizures by coincident video-EEG-polygraphic recordings will help avoid misdiagnosis and inappropriate treatment.

Electrographic seizure criteria

Over the last several decades electrographic/polysomnographic studies have become invaluable tools for the assessment of suspected seizures [2, 8, 10, 26, 32, 41, 42]. Technical and interpretative skills of normal and abnormal neonatal EEG-sleep patterns must be mastered before one can develop a confident visual analysis style for seizure recognition [14, 15, 43-45].

Corroboration with the EEG technologist is always an essential part of the diagnostic process, since physiological and nonphysiological artifacts can masquerade as EEG seizures. The physician must also anticipate expected behaviors for the child for gestational maturity, medication use, and state of arousal, in the context of potential artifacts. Synchronized video-EEG documentation permits careful off-line analysis for more accurate classification.

As with the epileptic older child and adult, it is generally accepted that the epileptic seizure is a clinical paroxysm of altered brain function with the simultaneous presence of an electrographic event on an EEG recording. Therefore, when assessing the suspected clinical event in the neonate, synchronized video/EEG/polygraphic monitoring is a useful tool to distinguish an epileptic from a nonepileptic event. Some advocate the use of single channel computerized devices for continuous prolonged monitoring [46] given the multiple logistical problems inherent with the use of conventional multichannel recording devices. This specific device nonetheless may fail to detect focal or regional seizures if the single channel recording is not near the brain region involved with seizure expression. For example, a recent study reported that fewer than three out of ten neonates with suspected seizures on a single channel device could be verified by conventional EEG [47]. Others have suggested that a four-channel EEG monitoring device can more efficiently detect seizures, which can then be verified by continuous video-EEG monitoring with more extensive electrode placement [48].

Epilepsy monitoring services for older children and adults readily utilize intracerebral or surface electrocorticography to detect seizures. Such recording strategies, however, are not ethically appropriate or practical for use with the neonatal patient. Subcortical foci are consequentially difficult to definitively eliminate from consideration.

Ictal EEG patterns - a more reliable marker for seizure onset, duration and severity

Neonatal EEG seizure patterns commonly consist of a repetitive sequence of waveforms which evolve in frequency, amplitude, electrical field and/or morphology. Four types of ictal patterns have been described: focal ictal patterns with normal background, focal patterns with abnormal background, multifocal ictal patterns [44] and focal monorhythmic periodic patterns of various frequencies. It is generally suggested that a minimum duration of ten seconds for the evolution of discharges is required to distinguish electrographic seizures from repetitive but nonictal epileptiform discharges [8, 36, 49]) (figures 1A, 2A, 2B). Clinical neurophysiologists should separately classify brief or prolonged repetitive discharges which lack an electrographic evolution as nonictal abnormal epileptiform patterns and do not confirm seizures [50]. The unique features of neonatal electrographic seizure duration and topography are discussed below.

Seizure duration and topography

Few studies have quantified minimal or maximal seizure durations in neonates [8, 49, 51]. The definition of the most severe expression of seizures, status epilepticus, which potentially causes brain injury is problematic for the newborn. In the older patient, status epilepticus can be defined as at least 30 min of continuous seizures or two consecutive seizures with an interictal period during which the patient fails to return to full consciousness. This definition, however, is not easily applied to the neonate for whom the level of arousal can be difficult to assess, particularly if sedative medication is given. One study arbitrarily defined neonatal status epilepticus as continuous seizure activity for at least 30 min, or 50% of the recording time [51]; 33% or 11 of 34 full-term infants had SE with a mean duration of 29.6 min prior to antiepileptic drug use, with another 9% or 3 out of the 34 preterm infants who also had SE with an average duration of 5.2 min per seizure (i.e. 50% of the recording time). The mean seizure duration was longer in the full-term infant (i.e. 5 min) compared to the preterm infant (i.e. 2.7 min). Given that more than 20% of this study group fit the criteria for SE by EEG documentation, clinicians who rely only on clinical criteria will underdiagnose prolonged seizures that potentially have the greatest likelihood to contribute to brain injury. This subject is explored in the section regarding the consequences of seizures on brain development.

Uncoupling of the clinical and electrographic expressions of neonatal seizures after antiepileptic medication administration also contributes to the underestimation of the true seizure duration, including SE (figure 7). One study estimated that 25% of neonates expressed persistent electrographic seizures despite resolution of their clinical seizure behaviors after receiving one or more antiepileptic medications [52], termed electroclinical uncoupling. Other pathophysiological mechanisms besides medication effect also might explain uncoupling [11].

Neonatal electrographic seizures can arise focally from one brain region, or be expressed as generalized synchronous and symmetrical repetitive discharges. In one study, 56% of seizures were seen in a single location at onset; specific sites included temporal-occipital (15%), temporal-central (15%) central (10%), frontotemporal central (6%), frontotemporal (5%) and vertex (5%). Multiple locations at the onset of the electrographic seizures were noted in 44% [8]. Electrographic discharges may be expressed as specific EEG frequency ranges from fast to slow, including beta, alpha, theta or delta activities. Multiple electrographic seizures can also be expressed independently in anatomically unrelated brain regions.

Periodic discharges - prolonged repetitive discharges; ictal or interictal?

Clinical neurophysiologists distinguish periodic non-seizure EEG patterns from electrographic seizures for patient populations of varying ages. As with older patients, the neonate may express or sustain repetitive or periodic discharges which do not satisfy electrographic criteria for seizures (figure 8A). Sustained periodic lateralized epileptiform discharges of
10 min or 20% of the recording time (i.e. defined as PLEDS on recordings of older children and adults) are rarely noted for the newborn [53]. This particular repetitive pattern of electrographic discharges in older patients are commonly associated with acute brain injuries from stroke, hemorrhage or trauma, and may also follow or precede electrographic seizures. Periodic discharges are less commonly noted in neonates (i.e. 5% of the 1,114 neonatal recordings), with PLEDS noted in only 4 of 34 infants. Most newborns with periodic discharges were expressed as shorter durations than classically defined PLEDS. However, nearly half of neonates with periodic discharges also expressed electrographic seizures at other times during the same EEG recording. Cerebrovascular lesions were the most common brain lesion in 53% (18 of 34) of this group of newborns. Periodic EEG discharges for 26 preterm neonates lasted less than 60 s, and were located in the parasagittal regions; whereas discharges in eight-term neonates were longer than a minute and located in the temporal regions. Forty-four percent of neonates died (15 of 34); 58% (11 of 19 infants) survived with neurological sequela. Therefore, neonates with periodic discharges identified on EEG recordings, require aggressive investigation for underlying brain injuries and prognostic considerations for neurological sequelae, independent to the decision to treat with AEDs.

Brief rhythmic discharges - ictal or interictal?

At the opposite end of the spectrum from periodic discharges, brief rhythmic discharges that are less than 10 s in duration have also been addressed with respect to an association with seizures and outcome (figure 8B). Some neonates with electrographic seizures may also exhibit these brief discharges, while other neonates only express isolated discharges without seizures. Neonates with brief discharges can suffer from specific etiologies or injuries such as hypoglycemia or periventricular leukomalacia, which carry a higher risk for neurodevelopmental delay [54]. These children will not benefit from the use of antiepileptic medication, whether suppression of brief discharges occurs or not.

Subcortial seizures versus nonictal functional decortication

Experimental animal models suggest conflicting neuronal mechanisms to explain clinical events which do not have coincident EEG confirmation. Most clinical neurophysiologists require documentation of an ictal pattern by surface EEG electrodes. However, subcortical seizures with only intermittent propagation to the surface may occur. At the other end of the spectrum, non-ictal "brain-stem release" phenomena must be considered, particularly if EEG seizures are never expressed [26]. A more integrated electroclinical approach has been suggested to classify clinical events as seizures versus nonepileptic movement disorders, based on documentation by synchronized video-EEG monitoring [10].

Brainstem release phenomena

Synchronized video EEG polygraphic monitoring provides the physician with documentation of a suspicious event with a concurrent electrographic pattern on surface recordings [19]. The temporal relationship between clinical and electrographic phenomena has been described, based on the synchronized video EEG polygraphic monitoring. Based on 415 clinical seizures in 71 babies, clonic seizure activity had the best correlation with coincident electrographic seizures. "Subtle" clinical events, on the other hand, had a more inconsistent relationship with coincident EEG seizure activity suggesting a nonepileptic brainstem release phenomena for at least a proportion of such events. Functional decortication resulting from neocortical damage without coincident EEG seizures [26] has therefore been suggested, such as with tonic posturing, as illustrated in figure 3A. Newborns with nonseizure brainstem release activity may express a different functional pattern of metabolic dysfunction, detected as altered glucose uptake on single photon emission tomography studies than neonates with seizures [55]. A recent suggestion to document increased prolactin levels in neonates with clinical seizures has also been reported [56], but such levels have not yet been correlated with electrographic seizures.

Electroclinical dissociation suggesting subcortical seizures

Experimental studies of immature animals also support the possibility that subcortical structures may initiate seizures, which only intermittently propagate to the cortical surface [57-59]. While EEG depth recordings in adults and adolescents help document subcortical seizures, both with and without clinical expression, this technology is not applicable or appropriate to the neonate. One anecdotal report of a human infant documented seizures, possibly emanating from deep gray matter structures [60].

Electroclinical dissociation (ECD) is one proposed mechanism by which subcortical seizures may also intermittently appear on surface-recorded EEG studies [12]. ECD has been defined as a reproducible clinical event that occurs both with and without coincidental electrographic seizures. In one group of 51 infants with electroclinical seizures, 33 infants simultaneously expressed both electrical and clinical seizure phenomena. Extremity movements were more significantly associated with synchronized electroclinical seizures. However, a subset of 18 of 51 neonates, 34%, also expressed ECD on EEG recordings. For neonates who expressed ECD, the clinical seizure component always preceded the electrographic seizure expression, suggesting that a subcortical focus may have initiated the seizure state. Some of these children also expressed synchronized electroclinical seizures, even on the same EEG record.

Controversy remains whether subcortical seizures or non-ictal functional decortication best categorize suspicious clinical behaviors in neonates without coincident EEG seizure documentation. This dilemma should encourage the clinician to use the EEG as a neurophysiological yardstick by which more exact seizure start and endpoints can be assigned, before offering pharmacological treatment with AEDs [32]. Neonates certainly exhibit electrographic seizures that go undetected unless EEG is utilized [61-65]. Two examples are neonates who are pharmacologically paralyzed for ventilatory assistance (figure 9A), or clinical seizures which are suppressed by the use of antiepileptic drugs (figure 9B) [6, 8, 52, 63-65]. In one cohort of 92 infants, 60% who were pretreated with antiepileptic medications, 50% of neonates had electrographic seizures with no clinical accompaniment [6]. Both clinical and electrographic seizure criteria were noted for 45% of 62 preterms and 53% of 33 full-term infants. Seventeen infants were pharmacologically paralyzed when the EEG seizure was first documented. A subsequently reported cohort of 60 infants, none of whom were pretreated with antiepileptic medication, included 7% of infants with only electrographic seizures prior to AED administration [52], and 25% who expressed electroclinical uncoupling after AED use.

The underestimation of seizures in the newborn period may also result from inadequate monitoring for specific neurological signs. Autonomic changes in respiration, blood pressure, oxygenation, heart rate, pupillary size, skin color and salivation are examples of subtle ictal signs (figures 10A and B). In one study, autonomic seizures accompanied electrographic seizures in 37% of 19 preterm neonates [6]. Newer classifications of neonatal seizures emphasize documentation of autonomic findings on EEG recordings [10].

Variation in the incidence of neonatal seizures based on clinical versus EEG criteria

Overestimation and underestimation of neonatal seizures are reported depending on whether clinical or electrical criteria are used. Based on clinical criteria, seizure incidences ranged from 0.5% in term infants to 22.2% in preterm neonates [66-69]. Discrepancies in incidence reflect not only varying post-conceptional ages of the study populations chosen, but also poor interobserver reliability [70] and the hospital setting in which the diagnosis was made. Hospital-based studies [6], which include high-risk deliveries, generally report a higher seizure incidence. Population studies which [71] include less ill infants from general nurseries report lower percentages. Incidence figures based only on clinical criteria without EEG confirmation include "false positives", consisting of the neonates with either normal or nonepileptic pathological neonatal behaviors. Conversely, the absence of scalp generated EEG seizures may include a subset of "false negatives" who express seizures only from subcortical brain regions without expression on the cortical surface. Consensus between clinical and EEG criteria is still needed for a more accurate incidence estimate.

Interictal EEG pattern abnormalities

Interictal EEG abnormalities (including nonictal repetitive epileptiform discharges) remain the most important contribution by the electroencephalographer for prognosis for either preterm or full-term infants [72, 73]. Severely abnormal bihemispheric patterns include suppression burst (i.e. paroxysmal) pattern (figure 11), electrocerebral inactivity, low voltage invariant pattern (figure 4C), persistently low background pattern (figure 6), multifocal sharp waves, and marked asynchrony [8]. For newborns who are being assessed for the presence and severity of neonatal encephalopathies, recognition of EEG pattern abnormalities are essential. There may be subclassifications for specific EEG patterns such as burst suppression with or without reactivity; the former suggesting a more favorable outcome than the latter [74]. Other descriptions of the EEG background abnormalities are not as readily recognized, such as dysmaturity of the EEG (i.e. background inappropriate for the child's postconceptional age). Dysmaturity may include discordance between cerebral and noncerebral components of sleep state organization as well as immaturity of the EEG patterns, both of which may predict neurological sequela [4, 75-77]. Focal or regional pattern abnormalities also have prognostic significance such as epileptiform patterns separate from seizures in preterm infants (i.e. positive sharp waves at the midline or central regions in children who have had either of two forms of cerebrovascular disease of the newborn, intraventricular hemorrhage or periventricular leukomalacia (figure 8A) [14].

Screening encephalopathic infants who are at risk for neonatal seizures will benefit from even a short routine EEG soon after birth. Identification of severe interictal EEG abnormalities may predict seizure occurrence on subsequent EEG records [78].

Interictal EEG findings are not pathognomonic for particular etiologies, mechanisms or timing [4]. One must integrate historical, physical examination and laboratory findings with the electrographic interpretation of both seizure and nonseizure pattern findings for the particular child. Serial EEG studies allow the clinician a more accurate diagnostic and prognostic interpretation [72, 73]. With the persistence of electrographic abnormalities on these recordings after the first week of life, chances of neurological sequelae are greater, even with resolution of clinical abnormalities. Conversely children who recover rapidly from a significant brain encephalopathy with the reemergence of normal EEG features during the first week after birth may experience comparatively less severe neurological sequelae. Both neonatal seizure patterns as well as interictal EEG pattern abnormalities may also reflect fetal brain disorders that preceded labor and delivery. The depth and the severity of neonatal brain disorders, as defined by clinical and electrographic criteria, should therefore be considered to estimate the timing of brain injury in the context of maternal, fetal, placental and neonatal histories [4].

Effects of neonatal seizures on brain development - consequences of under-diagnosis

Embedded within the controversy regarding how to diagnose neonatal seizures is the association of repetitive or prolonged seizures with brain damage and altered brain development. Linked to the clinician's concern for seizure duration is an appreciation of the diverse neuropathological processes and etiologies which cause neonatal seizures and neurological sequelae [79]. Central nervous system infections and severe asphyxia are two etiologies which exemplify underlying pathophysiological mechanisms responsible for brain damage in neonates independent of seizure expression.

Adverse effects of the seizure state on developing brain [80] have been recently reviewed. Seizures can disrupt a cascade of biochemical/molecular pathways which are normally responsible for the plasticity or activity-dependent development of the maturing nervous system. Depending on the degree of brain immaturity, seizures may disrupt the processes of cell division, migration, myelination, sequential expression of receptor formation and stabilization of synapses, each of which contributes to varying degrees of neurological sequelae [81].

Experimental models of seizures in immature animals suggest comparatively less vulnerability to seizure-induced brain injury than mature animals [82]. In adult animals, seizures alter growth of hippocampal granule cells, axonal and mossy fiber growth resulting in long-term deficits in learning, memory and behavior. A single prolonged seizure in an immature animal, on the other hand, results in less cell loss or fiber sprouting, and consequentially fewer deficits in learning memory and behavior. Resistance to brain damage, from prolonged seizure activity, however, is age-specific, as demonstrated by increased cell damage after only two weeks of age [83]. A recent study examined developmental changes in epileptiform activity in neocortical preparations using four different ages and four different pharmacological models. These authors confirmed that there are definite age-dependent differences in the susceptibility to epileptiform activity in the neocortex. These developmental changes seem to relate to intrinsic network properties of the neocortex that are independent of ontogenetic differences in any specific neurotransmitter system [84].

Repetitive or prolonged neonatal seizures alternatively can increase the susceptibility of the developing brain to suffer subsequent seizure induced brain injury during adolescence or early adulthood, by altering neuronal connectivity rather than increasing cell death [80, 81, 85, 86]. Neonatal animals subjected to status epilepticus have reduced seizure thresholds at later ages and impairments of learning, memory and activity levels after suffering seizures as adults. Proposed mechanisms of injury also include reduced neurogenesis in the hippocampus, for example, possibly because of ischemic-induced apoptosis as well as necrotic pathways [87]. Other suggested mechanisms of injury include effects of nitric oxide synthetase inhibition on cerebral circulation, which then contributes to ischemic injury [88]. Neonatal seizures therefore may initiate a cascade of diverse changes in brain development that become maladaptive at older ages, and increase the risk of subsequent damage after subsequent insults. Destructive mechanisms such as mossy fiber sprouting in the hippocampus or increased neuronal apoptosis may explain mutually exclusive pathways by which the immature brain suffers altered connectivity and reduced cell number, which is then "primed" for later seizure-induced cell loss at older ages.

The critical duration of seizures, whether cumulative or continuous, remains elusive with respect to resultant brain injury. Given that as many as one-third of full-term infants may satisfy a definition of status epilepticus [51] using EEG, EEG documentation appears crucial. A recent study of ten-day-old rat pups indicated that prolonged seizures for 30 min after asphyxia resulted in exacerbation of brain injury specific to the hippocampus, while sparing the neocortex. Prolonged neonatal seizures do worsen damage incurred by an already compromised brain [89] in a region-specific manner. In a neonatal rodent model of brief recurrent seizures [90], Landrot demonstrated increased mossy fiber sprouting in the granule cells of the hippocampus, which correlated with impaired cognition and reduced EEG power spectra during adolescence. A companion study from the same laboratory demonstrated alterations in cognition and seizure susceptibility within two weeks of the last seizure before the adult pattern of mossy fiber distribution is achieved. Therefore, therapeutic strategies to alter the adverse outcome of neonatal seizures must be initiated during or shortly after the seizures.

The overlapping effects of brain dysgenesis or injury from specific etiologies versus seizure-induced brain damage make it difficult to differentiate preexisting brain lesions from the direct, injurious effects of seizures themselves. The use of microdialysis probes in white and gray matter of piglet brains subjected to hypoxia indicate elevated lactate/pyruvate ratios after hypoxia but no direct association with seizure activity [91]. These findings support the conclusion that seizures themselves may not always be injurious to brain.

The lack of well-designed clinical investigations of outcome after neonatal seizures unfortunately fail to support these experimental findings [79]. Better definitions of neonatal seizure severity, including electrographic expression and seizure duration are still required to resolve this controversy. A recently published investigation of human newborns with perinatal asphyxia suggests that seizure severity is independently associated with brain injury, as measured by magnetic resonance spectroscopy however, strict EEG definition were not utilized [92].

Aggressive use of antiepileptic medication without EEG confirmation contributes to the inaccurate estimate of seizure severity in neonates and possible medication-induced brain injury. Intractable seizures generally require the use of multiple antiepileptic medications, which may still not effectively control seizures [93]. Drugs may also impede the recognition of persistent seizures, because of the uncoupling phenomenon by which the clinical expression is suppressed while the electrical expression of seizures continues. Clinical definitions of seizure occurrence and duration consequently underestimate seizure severity, which may be associated with increased risk of damage [92]. AED use also has secondary harmful effects on cardiac and respiratory functions, with resultant circulatory disturbances that may can contribute to brain injury because of hypoperfusion [79]. Finally, AED use may have teratogenetic effects on brain development as a result of exposure over long periods of time.

Received March 19, 2002 / Accepted April 21, 2002