Nov 22, 2007

Abnormal Neonatal EEG

The EEG prognostic value at the time of continuous development is often greater than at a later stage. EEG testing can provide reassurance to the physician and parents at a time of potential catastrophic damage.

The continuous changes that occur during early brain development are often associated with striking changes in EEG patterns over short periods. This makes it difficult to interpret EEG results, which can discourage the use of EEG testing.

Given the close relationships between certain morphological aspects of the developing brain and EEG results, gestational age (GA) can be reliably estimated (to ±1 wk) by EEG criteria. In fact, CNS development of the immature brain proceeds at about the same rate during fetal development as in the postnatal environment.

The physiological substrate for these early EEG patterns is unknown, but is probably derived from cortical generators that are strongly influenced by subcortical (primarily thalamic) afferent input. Rapid maturation of these structures (and not the corpus callosum) is most likely responsible for the interhemispheric synchrony that occurs close to full-term GA; in particular, rapid dendritic spine development and synaptogenesis are typical of the last month of fetal development. The complex development of cerebral sulci during this same period is probably responsible for the neonatal EEG results showing complex, more definitive patterns at term. At this age, easily recognizable and organized activity patterns appear. These continue with little change during the first month of life and are strictly characteristic of neonatal EEG.

There are several technical considerations when recording from a small (neonate) scalp. High skin resistance impedes low-resistance scalp-to-electrode contact. The state of activity (awake or quiet vs active sleep) can be selectively bound to certain aspects of pathology. It is important to annotate the tracing with particular attention to the presence and type of eye movements, facial movements, respiration (regular or irregular), sucking, crying, grimacing, and so on. Extracerebral monitors are needed in routine recordings, including at least electrooculogram (EOG), respiration rate measurement, and electrocardiogram (ECG). Only a reduced number of scalp electrodes, generally never more than the set in a 16-channel recording, are applicable. A low time constant (0.25-0.60 seconds) is preferable to record the low-frequency background activity. Slow paper speed maximizes the slow background and the degree of interhemispheric synchrony.

Active sleep (AS), the antecedent of rapid eye movement (REM) sleep, is usually indicated by irregular respiratory patterns with interspersed, brief apneic episodes that often precede clusters of eye movements. Contrary to adult physiology, prominent, subtle motor activity, especially of the face (eg, grimacing, smiling), accompanies this state. These results are often interpreted as seizure activity by the inexperienced reader. On EEG, 2 patterns are observable, as follows:

Sleep state cyclicity is certainly achieved after 30 weeks' postconceptional age (PCA), with stability over multiple cycles only at 36 weeks' PCA.

Lately, the increased survival of extremely young premature babies has allowed to assess very early expression of sleep cyclicity by combining measures of REM and EEG discontinuity (Sher, 2005) between 25 and 30 weeks' PCA. Early forms of "transitional" sleep akin to "seismic sleep" in the rat represents an immature form of paradoxical sleep with mixed features of active and quiet sleep. It probably corresponds to a primitive form of brain activity characterized by a low level of inhibition progressively declining toward term (Biagioni, 2005).

Near the end of the first month, a more diffuse pattern usually appears, consisting of continuous, high-to-moderate amplitude, slow activity that is not seen in the preterm infant. A high degree of synchrony between burst and interburst activity is desirable at term. This usually confirms normal maturational patterns.

Several morphological figures may occur with variable frequency. Random sharp-waves, most commonly temporal or rolandic, are sporadically seen in QS. Nonrolandic, repetitive, highly focal spikes confined to a single location that occur during wakefulness usually indicate abnormalities. A burst of frontal delta and synchronous, frontal sharp waves are still abundant in the FTBI during AS. Spindle delta bursts (brushes) are seen with decreasing frequency in the FTBI and are usually confined to the central and temporal leads during QS. This state is the most vulnerable, being susceptible to various minor CNS insults that are only transiently apparent, depending on their expression. It is important to perform prolonged recordings, especially in stressed infants as they are likely to express less QS.

Several important milestones characterize EEG maturation patterns during the first months of life. The newborn progressively develops a circadian rhythm, resulting from the interaction of endogenous factors with external synchronizers such as light, eating, and sensory stimulation over the course of a day. At approximately the third month, sleep efficiently occurs in nocturnal intervals of at least 8 hours, reflecting mother-child interactions and the established activity of endogenous pacemakers.

With regard to EEG results, several important changes accompany this phase. From the second week of life, slow and continuous background activity (consisting of increasing amplitude delta waves whose frequency also decreases with the approaching first month of life) progressively replaces the discontinuous pattern (tracé alternante) that is typical of QS. Typical EEG characteristics disappear within the second month of life, including slow frontal biphasic spikes (encoches frontales) and negative rolandic spikes. The newborn still falls asleep in AS until the end of the third month. AS decreases from 50% to 40% by the end of the fourth month; likewise, QS progressively increases and becomes more defined due to the appearance of EEG hypnic features that are typical of adults. Vertex waves can be noted in the rolandic regions after the third month; sleep spindles appear earlier, at about the sixth week, over the central regions.

The first sleep spindling samples are slower in frequency and more anteriorly distributed in newborns compared with older infants. These infrequently appear at the beginning of QS as rudimentary, low-voltage (<25>

Around the sixth postnatal week, 75 mV occipital sharp waves characterize AS and increase in frequency from 2 to 4 Hz toward the end of the third month. At 3 months, a clearly defined 3-4 Hz, 50-75 mV occipital rhythm appears during wakefulness; this is interrupted by eye opening. It progressively evolves at about 5 months to a faster frequency of 6-7 Hz.

ABNORMAL NEONATAL EEG
Even more than in other epochs of life, in neonates the abnormal neonatal EEG has a prognostic value as opposed to a diagnostic value. Rarely, specific EEG patterns correspond to typical syndromes. Prognostic value can be increased with the following methods:
Because EEG abnormalities in neonates cover a broad spectrum, any classification is difficult and, in some cases, arbitrary. One possibility for classification would be to distinguish between diffuse and focal abnormalities and to categorize separately ictal and paroxysmal patterns in the presence of neonatal convulsions.

DIFFUSE EEG ABNORMALITIES
With regard to severity and prognosis, severe and irreversible abnormalities should be distinguished from moderate, reversible abnormalities. Severe abnormalities correspond to 2 main EEG patterns, inactive and paroxysmal, both of which are accompanied by a lack of sleep cycles and a lack of reactivity to internal or environmental stimuli.

The inactive or isoelectric pattern consists of cerebral activity below 10 mV that is continuously present throughout the record. Brief intervals of low-voltage activity, which are located over the posterior head regions, may occasionally be present. This pattern may occur in varying clinical conditions and often occurs with the following states:

In the absence of a drug-induced state, hypothermia, or postictal recording, the prognosis is poor but not necessarily fatal.

The paroxysmal or burst suppression EEG pattern is characterized by intervals of inactive background activity (<10-15 href="http://www.emedicine.com/neuro/topic545.htm#target3">Picture 3). This pattern, which carries a highly unfavorable prognosis, must be clearly distinguished from a full-term newborn's tracé alternante and a preterm infant's tracé discontinue (TD), both of which are normal patterns. Serial recordings are essential to reach a reliable prognosis. Certain conditions (eg, Aicardi syndrome or uncommon dysgenetic conditions that involve the corpus callosum) rarely present as hemihypsarrhythmia.

Severe but reversible diffuse abnormalities can occur and are exemplified by the so-called low-voltage pattern throughout the EEG record. QS and AS are only distinguishable by the slightly higher voltage in QS, where mixed frequencies under 10-50 mV are almost continuously recorded. This finding and a diffuse delta pattern with minimal theta rhythms throughout the entire EEG record hold an intermediate prognosis. When the abnormalities are compatible with these changes seen in sleep, they are generally considered moderate and reversible.

Diffuse EEG abnormalities can also be seen as irregularities in maturational indices and organizational states. In addition to the patterns of profound disruption to the ability to organize cyclic states (which are typical of the most severe abnormalities), several patterns of EEG dysmaturity can be recognized and identified. In newborns who are small for their gestational age, transient or persistent dysmaturity patterns can be distinguished by their duration. Quantification may include assessment of interhemispheric synchrony in tracé alternante, typical of QS, or the counting of premature features such as delta brushes.

Abnormalities of EEG patterns, noted in relation to sleep states and the instability of sleep-wake states during the newborn period, have some prognostic value. When different etiologies of the EEG pattern are considered, a few fundamental groups can be distinguished.

Transient metabolic disorders

Neonatal hypoglycemia can range from an asymptomatic state with a minimal EEG correlation to late-onset, idiopathic hypoglycemia accompanied by neurological symptoms and seizures. Toxemia and maternal diabetes are often encountered in high-risk pregnancies. These newborns usually present with decreased QS with a relative increase in AS. Transient hypocalcemia is often associated with barely abnormal interictal EEG and variable focal seizures (in 20% of patients).

Inborn errors of metabolism

Periodic EEG patterns in newborns with uneventful deliveries strongly suggest the possibility of an inborn error of metabolism. The most frequent neurological symptoms are early movement disorders, convulsions, and cognitive dysfunction. In 1977, Mises accurately described periodic EEG patterns in methylmalonic aminoacidopathy.

High interindividual variability characterizes a pattern of periodic frontal or occipital sharp waves that are interspersed with rapid rhythms. In maple syrup urine disease, EEG complexes are low-voltage and less periodic; background activity is less depressed. Comb-like rhythms during the second and third postnatal weeks are pathognomonic of this disorder.

The highly peculiar EEG pattern of non-ketotic hyperglycemia distinguishes it from other forms. During the first 10 postnatal days, these infants, who present with hypotonia, respiratory distress, and myoclonic seizures, have EEGs characterized by periodic, highly stereotyped 1-3 Hz complexes with 4- to 18-second interburst intervals. Frontal, high-voltage slow waves are associated with characteristic rolandic and occipital early alpha rhythms.

Pyridoxine dependence (not to be confused with pyridoxine deficiency) is inherited as an autosomal recessive trait and is accompanied by severely abnormal EEGs and refractory seizures that only respond to pyridoxal supplementation.

CNS INFECTIONS
An important distinction must be made between prenatal and postnatal infections. No specific or typical EEG patterns exist for the first group. Severity and extent of CNS involvement is more significant compared to noninfectious etiologies. Rubella and toxoplasmosis are the most common causative agents.

Infants with congenital rubella and cataracts present with the most consistent EEG abnormalities in this group (ie, prolonged subclinical seizures expressed as paroxysmal bursts without interburst intervals or alpha-like activity). In the occipital areas, there is marked asynchrony between burst and interburst intervals. Slow anterior dysrhythmia with excessive frontal sharp waves is present. Sleep states are not well defined, given the absence of recorded REM and the paucity of QS.

Fetal toxoplasmosis is less disruptive, at least in terms of ultradian sleep cycle organization. It is associated with more variable EEG patterns. In cytomegalovirus, the EEG is frequently inactive.

For postnatal infections (usually meningitis), the prognostic value of EEG is linked to the severity and extent of the abnormalities rather than their specificity. In most cases, they are associated with early status epilepticus (SE) or single seizures with a definite interval following birth that is unlike postanoxic SE. Consistently abnormal recordings (rather than merely an initial abnormal recording) are linked to an unfavorable prognosis. Three distinctive patterns are associated with type 1 and type 2 herpes simplex virus (HSV) encephalitis. For pregnant women in many countries, HSV is still the most common (and preventable in neonates by means of cesarean section) genital infection. HSV is easily transmitted to the newborn during vaginal delivery. One type of EEG abnormality consists of continuous, sharply contoured unifocal or multifocal delta activity with a typical asynchronous distribution across several foci (each with a specific rate). Foci are predominantly temporal, frontal, or central in distribution. In older infants, hemispheric, monomorphic slow waves appear interspersed on a low-voltage or suppressed background. They may recur as periodic lateralized epileptiform discharges (PLEDs) within several seconds. Typical electroencephalographic seizures are associated with positive, multifocal sharp waves on a background that is characterized by significant interhemispheric voltage asymmetries and asynchrony.

FOCAL ACUTE NEUROLOGICAL ABNORMALITIES
The following conditions may cause focal neurologic abnormalities: trauma, primary subarachnoid hemorrhage, intraventricular hemorrhage (IVH), intraparenchymal hemorrhage, and cerebral infarction. EEG abnormalities include an interhemispheric amplitude asymmetry pattern that is mainly seen with intraparenchymal hemorrhage or with a prenatal or postnatal ischemic insult. A wider criterion (>50%) is usually applied to preterm (PT) infants in whom significant hemispheric voltage alteration has been found to strongly correlate with contralateral hemiparesis. In 1984, Challamel reported that transient hemispheric asymmetries were a normal variant in infants with no CNS-related ailments who later resumed fully developed normal EEG patterns.

The focal attenuation pattern refers to a single scalp region with a persistent voltage attenuation. This pattern has an inconsistent association with neuropathological lesions; both false-positive and false-negative correlations have been observed. Focal slowing is highly unusual and may be a sign of ongoing seizures.

Nonictal paroxysmal patterns include the following:

EEG IN NEONATAL SEIZURES

Seizures frequently occur in newborns (14 per 1000), often causing death or permanent neurological sequelae. The prognosis largely depends on etiologic factors and the duration of convulsive activity. It should be noted that generalized tonic-clonic seizures are not seen in the immature brain.

Seizure patterns can be distinguished into subtle, classic, tonic, and myoclonic types. Among these, some have a fairly consistent EEG correlation. For example, tonic spasms, unlike those that occur in older children, are often associated with rhythmic delta wave activity. Clonic seizures frequently correspond to repetitive spike discharges. Myoclonic seizures, which are often erratic or fragmented and which should be distinguished from fragmentary neonatal sleep myoclonus, are often associated with other seizure types (eg, tonic or clonic) and with a burst suppression pattern with or without ictal correlation.

Ictal intervals, apnea, or respiratory disturbances often correlate with alphalike EEG patterns. These may be ictal discharges without clinical seizures that are not limited to iatrogenically paralyzed infants.These occult or electrographic seizures without clinically detectable signs may result from iatrogenic loading, causing serious neurological injury that disables the effector structures or silent cortical areas, which might be more generalized in newborns than in older patients.

Conversely, clinical seizures in the absence of EEG discharges suggest nonepileptic events that should be closely monitored to avoid misdiagnosis. Minimal seizure behavior, uncoupled to ictal EEG patterns, can be seen in healthy neonates and especially in encephalopathic neonates whose brains are seriously compromised by hypoxic-ischemic insults. Severe neurological injury seen in these cases causes severe background EEG abnormalities.

Several ictal discharge patterns have been identified and reported, including the following patterns:

  • Focal spikes or sharp wave discharges of progressively increasing amplitude over the course of the seizure - These discharges correspond to contralateral jerking and occur predominantly in the rolandic and temporal regions.
  • Multifocal spike and sharp wave discharges, which are often erratic with independent frequencies in multiple foci, are associated with variable seizure types. The underlying cause may range from benign conditions to CNS infection to various hypoxic-ischemic injuries. The prognosis is dependent on the background EEG abnormalities and the specific underlying etiology.
  • Prehypsarrhythmic or hypsarrhythmic patterns can be seen early in compromised newborns, representing the most severe examples of the previous pattern, and are usually associated with anarchic and refractory seizures. A separate group may be the brief ictal discharge pattern and questionable EEG ictal discharges. Decremental discharges, which sometimes accompany neonatal tonic seizures, must be distinguished from the normal arousal response that follows postural change or stimulation.

The expression of ictal activity in relation to sleep stages (REM/NREM) may have age-dependent mechanisms in the developing brain. Schmutzler et al 2005 tried to assess the relationship between ictal activity and sleep stages in the newborn EEG, finding the highest association with unrecognizable sleep states where sleep organization is already disrupted. Ictal activity predominates otherwise in REM sleep (p=0.01) with longer duration of discharges, contradicting findings in adults with epilepsy. However, the mechanisms responsible for increased seizures during NREM in adults (synchronous EEG oscillations promoting electrographic seizure propagation and asynchronous discharge patterns reducing seizures during REM sleep) cannot be extrapolated to the immature developing brain.

There is in fact an age-related differential regional distribution of GABA with excitatory and not inhibitory roles (Mosche, 2000) in subcortical areas like the substantia nigra that could facilitate the release of focal discharges during REM in newborns. Furthermore, an immaturity of REM-related inhibitory systems at a peripheral level have been shown in infants, which might affect the cortex influencing the frequency of ictal discharges during REM sleep in newborns.

Tekgul et al (2005) compared the yielding power of a reduced montage (9 electrodes) with the full 10/20 electrode montage to detect and characterize neonatal seizures and background EEG features. The sensitivity and specificity of the reduced montage for electrographic seizure detection was 96.8% and 100%, respectively, and only in 1 patient's record (over 31 pts) the single seizure was missed altogether. For assessing background abnormalities, the sensitivity and specificity of the reduced montage was 87% and 80%, respectively. The authors conclude that a reduced montage proves to be a sensitive tool for identification of neonatal seizures and grading of background EEG features in newborns.

Specific syndromes of neonatal seizures include the following: