EEG Abnormalities of Premature and Full-Term Neonates
THE TIMING OF EEG STUDIES
Ideally, the initial electroencephalogram (EEG) examination should be done within the first 24 hours after birth or after a suspected brain insult. The best estimate of the degree of damage or dysfunction can be made when the EEG documents the evolution of the abnormality over time. Typically, as time passes, the degree of abnormality lessens. The slower this change, the more severe the underlying brain abnormality. If serial studies document the rate and character of the changes, the prognostic information will be more reliable than that obtained from a single study (Briatore et al., 2013; Chequer et al., 1992; Douglass et al., 2002; Graziani et al., 1994; Hellström-Westas and Rosén, 2005; Holmes and Lombroso, 1993; Holmes et al., 1982; Klinger et al., 2001; Kumar et al., 1999; Le Bihannic et al., 2012; Menache et al., 2002; Pressler et al., 2001; Selton and Andre, 1997; Selton et al., 2010; Takeuchi and Watanabe, 1989; Tharp 1990; Tharp et al., 1989; Watanabe et al., 1999; Zeinstra et al., 2001). Serial studies also afford greater opportunity to detect electrical seizures without clinical signs (Glauser and Clancy, 1992), which themselves may be of prognostic significance (see Chapter 7).
Failure to recognize that EEG findings evolve over time may lead to a less than accurate determination of prognosis. For example, an EEG of an infant might show a suppression-burst pattern on the first day of life, a finding typically indicating the presence of severe brain dysfunction. However, hours later, after the infant’s physiological condition has stabilized, the EEG activity may become continuous with relatively normal background activity. Such changes drastically alter statements concerning the prognosis. However, a suppression-burst pattern that is sustained over several days or that changes to a depressed and undifferentiated pattern implies a poor prognosis for recovery of brain function. Failure to recognize the importance of the time course has led to contradictory statements in the literature concerning the prognostic significance of suppression-burst activity.
With this caveat, however, some statements can be made concerning the significance of the first EEG recorded early in the course of neurological illness. In infants requiring intensive care, the EEG findings obtained within the first 24 hours after birth can provide reliable prognostic information (Pezzani et al., 1986; Pressler et al., 2001); an EEG with normal findings on the first day, or with only minimally abnormal findings, reliably indicates a good prognosis unless further brain injury occurs later.
Dyschronism: Disordered Maturational Development
An experienced neurophysiologist can usually determine conceptional age (CA) to within 2 weeks between 26 and 33 weeks CA and within 1 week between 34 and 37 weeks CA, based upon expected developmental features (see Chapter 4) (Tharp, 1990). Disordered maturational development is referred to as “dyschronism” and, if present, is an important abnormality of the neonatal EEG (Hrachovy et al., 1990).
This term refers to an EEG in which the developmental characteristics in all states are immature for the reported gestational or CA. If the developmental features of an EEG are immature for the stated gestational or CA and the features of the background activity in all states are normal, the following questions must be addressed: Is the age as determined by clinical evaluation overestimated or are the immature EEG features evidence of delayed maturation? The last explanation suggests that a cerebral insult may have occurred during intrauterine life. A discrepancy of 2 weeks or less between EEG age and estimated CA most likely indicates the presence of a transient central nervous system (CNS) dysfunction. However, discrepancies of more than 3 weeks usually indicate the possibility of persistent CNS dysfunction and are frequently accompanied by other EEG abnormalities, including marked suppression of background activity or multifocal sharp waves.
This term refers to different maturational characteristics between the EEG during wakefulness and in the deepest stages of nonrapid eye movement (NREM) sleep (Figures 6.1 and 6.2). For example, characteristics of the waking EEG might be consistent with a CA of 38 weeks, whereas the background activity during NREM sleep might be consistent with a CA of 34 weeks. If dyschronism of 3 or more weeks occurs between these states, other EEG abnormalities are often present. These abnormalities are usually most apparent in NREM sleep, the state that always shows the least mature characteristics (Figure 6.3). Such findings suggest significant brain dysfunction. Therefore, an EEG of any infant should include a period of the deepest stages of NREM sleep.
Transient Maturational Abnormalities
Maturational abnormalities may occur transiently in a newborn experiencing acute and ongoing encephalopathy. Therefore, the prognostic significance of an EEG for which developmental features are immature for the stated CA can be determined only by making serial tracings.
Abnormalities of Background Activity in Diffuse Brain Disturbance
There is no consensus on the classification of background abnormalities of the EEG in the neonate, particularly in considering hypoxic ischemic encephalopathy. Walsh et al. (2011) note that most of the classification systems do agree at the extremes, but that the moderately abnormal findings cause the most controversy in terms of their clinical significance.
Prolongation of Interburst Intervals in Premature Infants
The range of duration of normal interburst intervals is CA dependent (Figure 6.4) and, in older premature infants, state dependent (Figure 6.5). The longest acceptable single interburst interval duration according to CA has been reported to be 24 weeks CA, 60 seconds (Vecchierini et al., 2003); 26 weeks CA, 46 seconds; 27 weeks CA, 36 seconds (Selton et al., 2000); less than 30 weeks CA, 30 to 35 seconds; 31 to 33 weeks CA, 20 seconds; 34 to 36 weeks CA, 10 seconds; 37 to 40 weeks CA, 6 seconds (Hahn et al., 1989). In general, in considering interburst intervals within an individual record, the longest interval is measured, rather than assessing an average of intervals. Although a prolongation of the interburst interval may be secondary to a CNS insult, it may also be due to the use of sedative medications such as morphine (Young and da Silva, 2000) and sufentanil (Nguyen et al., 2003).
Episodes of Generalized and Regional Voltage Attenuation Associated With Multifocal Sharp Waves in Term Infants
Another finding of diffuse dysfunction in the term infant is the presence of generalized or regional episodes of voltage attenuation (Figure 6.6). Although abnormal, this suggests relatively mild diffuse dysfunction compared to other findings listed later. This finding often occurs in the presence of multifocal sharp waves.
Depression and Lack of Differentiation
Depressed beta activity, either focal or diffuse, is often the first manifestation of abnormal cortical function. After hypoxia-ischemia, faster frequencies tend to be depressed or obliterated. Lack of differentiation (i.e., the “undifferentiated EEG”) refers to virtual or complete disappearance of the polyfrequency activity normally present (Figure 6.7) (Selton and Andre, 1997). A depressed and undifferentiated EEG background often accompanies other abnormalities (Figure 6.8). However, in some instances, developmental milestones may persist (Figure 6.9). Infants with hypoxic-ischemic encephalopathy may be acutely treated with hypothermia. The EEG may initially be depressed and undifferentiated and may either persist or evolve during the cooling period. This may include emergence of polyfrequencies, multifocal sharp waves, continuity and, at times, electrographic seizures (Figure 6.10) (Boylan et al., 2015; Glass et al., 2014; Low et al., 2012; Shah et al., 2014; Wusthoff et al., 2011).
A depressed and undifferentiated EEG in the newborn indicates that a severe brain insult has occurred. Disorders causing such an EEG include profound hypoxia, severe metabolic disorders, infectious processes such as meningitis or encephalitis, and intraventricular hemorrhage (IVH). A depressed and undifferentiated EEG within the first 24 hours after birth that persists signifies a poor prognosis.
The suppression-burst pattern represents an intermediate degree of diffuse brain disturbance between the depressed and undifferentiated EEG and electrocerebral silence. Activity during the bursts consists primarily of delta and theta frequencies, which, at times, are intermixed with sharp waves. The bursts are separated by periods of marked generalized voltage attenuation or electrocerebral silence (Figures 6.11–6.19). Suppression-burst patterns persist unremittingly; no change in the EEG activity is seen during the entire recording and the pattern does not react to painful stimuli. Some infants with the suppression-burst pattern may experience periodic slow myoclonic jerks (Figure 6.17). Although typically synchronuous on the two sides, the bursts may appear asynchronously, possibly in association with congenital anomalies of the brain (Figure 6.18).
Electrocerebral silence represents the ultimate degree of depression and lack of differentiation in the neonate. The transition from a severely depressed and undifferentiated background to one that is isolectric may be difficult to determine (Figure 6.20) and serial EEGs may be required to demonstrate a persistent degree of cortical inactivity (Figure 6.21). An isoelectric EEG (i.e., “electrocerebral silence”) is evidence of severe cortex dysfunction, not that of the brainstem, which, in the infant, may sustain vital functions for prolonged periods in the absence of cortical function. Indeed, prolonged survival may occur in infants whose EEGs continue to show electrocerebral silence (Mizrahi et al., 1985). The finding of electrocerebral silence on EEG in the newborn is not used to support a determination of brain death.
Specialized Generalized Patterns
The classical hypsarrhythmic pattern as described by Gibbs and Gibbs (1952) rarely appears before 44 weeks CA. However, one modification of this pattern in the neonatal period is the suppression-burst variant (Hrachovy et al., 1984) (Figure 6.19). The bursts consist of asynchronous, high-voltage slow activity mixed with multifocal spikes and sharp waves. The primary features that distinguish this variant of hypsarrhythmia are the periodicity of the bursts and the high voltage of the activity within the bursts.
Patients with infantile spasms and this variant have a poor prognosis for long-term outcome, regardless of whether the pattern develops in the neonatal period or in later months of life (Maheshwari and Leavons, 1975). In addition, when this pattern does appear in the neonatal period, it is closely associated with the presence of inborn errors of metabolism, most notably nonketotic hyperglycemia.
The term holoprosencephaly is applied to a spectrum of related cerebral malformations resulting from faulty diverticulation of the prosencephalon. The malformation varies in severity, from arhinencephaly (in which the olfactory bulbs and tracts are absent but the brain is otherwise normal) to alobar holoprosencephaly (in which lobes are not demarcated and the cerebrum is monoventricular, with or without a dorsal cyst). A variety of median facial defects (including cyclopia, orbital hypotelorism, cleft lip, cleft palate, and hypoplasia of the premaxilla) are associated with the cerebral malformations (DeMyer and Zeman, 1963; Yakovlev, 1959).
The EEG findings associated with holoprosencephaly were described by DeMyer and White (1964). They include: (a) multifocal spike and polyspike activity mixed with slow waves; (b) periods of monorhythmic beta-, alpha-, theta-, or delta-frequency activity, occurring singly or in various combinations; (c) asynchrony between hemispheres; (d) isoelectric or relatively low-voltage activity; (e) periodic patterns; and (f) lack of any normal organization (Figures 6.22–6.24). These findings occur in various combinations in a single patient. Equally dramatic are the continuous and abrupt changes from one pattern to another, such that within few minutes, most or all of the aforementioned features can be visualized. This constellation of findings is not seen in other disorders of infancy and is therefore diagnostic for holoprosencephaly. Infants with holoprosencephaly often have unusual or stereotyped movements suggestive of seizures. However, there is no correlation between these movements and EEG changes. The prognosis is poor; about 50% of infants die within the first month of life; about 80% will not survive past the first year.
PATTERNS THAT MAY BE FOCAL OR GENERALIZED
Sustained Rhythmic Alpha and Theta Activity
Sustained, rhythmic 4 to 7 Hz activity (theta) and/or 8 to 10 Hz (alpha), 40 to 100 µV activity that is generalized or focal is an abnormal finding. However, precise correlations to specific etiologic factors have not, for the most part, been determined. This activity may occur almost continuously or paroxysmally in brief runs and may occur as theta, alpha, or mixed activity (Figures 6.25–6.30).
When alpha activity occurs focally, it is most prevalent in the central or temporal regions, where it may occur independently on the two sides. Although this activity may be most prominent in the awake state, it is usually present in all states and is usually accompanied by other abnormalities such as abnormal sharp waves. Generalized rhythmic alpha activity has been associated with various underlying abnormalities, most commonly congenital heart disease; However, it also may be seen in infants who have received CNS-active drugs such as diazepam and phenobarbital (Hrachovy and O’Donnell, 1999). It is important to distinguish this pattern from the alpha seizure pattern (see Chapter 7).
Sustained Rhythmic Delta Activity
In some instances bifrontal slow (delta) activity is considered an abnormal finding. Abnormal bifrontal slow activity can be differentiated from normal bifrontal slow activity by its presence in all stages of sleep and wakefulness and its unrelenting character (Figure 6.31). Abnormal rhythmic slow (delta) activity also may be present in the occipital regions bilaterally (Figure 6.32).
Periodic and quasiperiodic discharges in a newborn’s EEG have been reported to occur in association with various CNS injuries. They are believed to be suggestive of the presence of neonatal herpes simplex virus encephalitis (HSVE) (Mikati et al., 1990; Mizrahi and Tharp, 1982) (Figure 6.33). While such periodic discharges may be associated with HSVE, some may also be interpreted as electrical seizures of the depressed brain type (see Chapter 7). In addition, periodic discharges are not, however, peculiar to HSVE and may occur with various other CNS insults such as infarction (Randò et al., 2000; Scher and Beggarly, 1989).
Unilateral Depression of Background Activity
Mild shifting asymmetry of the background EEG activity between hemispheres is a common finding in the newborn, particularly during NREM sleep. However, an abnormal finding is a marked voltage asymmetry of background rhythms between hemispheres that persists in all states (Figures 6.16 and 6.34–6.36). Unilateral depression may occur in association with a wide range of structural cerebral lesions such as infarction, hemorrhage, focal cystic lesions, and rarely, congenital malformations. In addition, other intracranial abnormalities such as subdural fluid collections may be associated with this EEG finding. However, focal depression of the background activity may also persist for variable periods after electrical seizures. A marked asymmetry of the background activity may also result from nonintracranial causes such as subgaleal swelling, scalp edema, or technical error.
Focal Slow Activity
Just as in older infants, the finding of focal slow activity that persists at a specific site may indicate the presence of a focal destructive lesion such as infarct, hemorrhage, or, more specific to neonates, congenital anomalies of the brain (Figure 6.37).
Central Positive Sharp Waves
These waves are 50 to 250 µV surface-positive transients lasting 100 to 250 ms and occurring either unilaterally or bilaterally in the central regions (Figures 6.38–6.40). A lower voltage after-going surface-negative component may be present. The waves usually occur singly or in brief runs. Central positive sharp waves are not epileptiform discharges, and the physiological processes causing them remain unknown. They have been described most notably in infants with IVH (Blume and Dreyfus-Brisac, 1982; Cukier et al., 1972; Lomboso, 1982; Tharp et al., 1981), and white matter necrosis (periventricular leukomalacia) (Lombroso, 1982; Marret et al., 1986; Novotny et al., 1987). They have also been associated with “major ultrasound lesions” such as grade III IVH, periventricular hemorrhagic infarction, cystic periventricular leukomalacia, persistent periventricular echodensities, and persistent ventricular dilatation (Castro Conde et al., 2004). The timing of the recording in relation to the occurrence of the hemorrhage may influence their appearance (Clancy and Tharp, 1984; Kidokoro et al., 2009). Current thought is that the finding of central positive sharp waves is not specific for IVH but rather for white matter necrosis, which may result from a variety of insults, including IVH (Novotny et al., 1987). In addition, the abundance and rate of recurrence of central positive sharp waves within a single record of an infant appear to correlate with long-term neurological outcome; a rate of greater than 2 per second has been associated with a poor outcome (Blume and Dreyfus-Brisac, 1982; Kidokoro et al., 2009; Nosralla et al., 2009). Other conditions such as meningitis, hydorcephalus, amino-aciduria (Tharp, 1980), and asphyxia (da Costa and Lombroso, 1980) have also been associated with central positive sharp waves.
Temporal Sharp Waves
The problems of determining whether sharp waves that occur in the temporal regions are normal or abnormal have been previously discussed (see Chapter 5). Whereas some sharp waves occurring in the temporal regions are considered normal, others may not meet the criteria to be called abnormal and are thus of questionable significance. However, there are temporal sharp waves that are clearly abnormal. Criteria for abnormality include morphology, polarity, rate of recurrence, and persistence at one site (Figures 6.41–6.44). They may be, although not always, associated with focal injury in the region of occurrence. They are not considered to be interictal epileptiform findings (see Chapter 7).
Extratemporal Focal Sharp Waves
Abnormal sharp waves that appear as slow sharp transients or rapid spikes may occur in the frontal (Figures 6.45–6.51), central, (Figures 6.52–6.54) and occipital (Figures 6.55–6.57) regions. When persistently focal, they may suggest focal brain injury, although often no well-defined structural lesion can be documented by neuroimaging.
Multifocal Sharp Waves
As already noted, sharp waves in the newborn EEG are common, with certain sharp-wave activity being considered “normal.” Multiple foci of high-voltage, long-duration, sharp-wave activity are commonly seen in infants who have experienced a diffuse CNS insult (Figures 6.58–6.61). Such abnormal sharp waves usually predominate in the temporal regions and may persist over one hemisphere. As noted, multifocal sharp waves usually accompany various other EEG abnormalities, including depressed and undifferentiated background activity, episodes of attenuation, and rhythmic bifrontal delta activity. Multifocal sharp waves may be the only abnormality after a CNS insult and may also be the last remaining evidence of CNS dysfunction in serial tracings. This abnormality is usually maximal in NREM sleep, and in some infants it may occur only in this state (Figure 6.3). Multifocal sharp waves cannot be used as evidence that a seizure has occurred or will occur, because the sharp waves do show a significant association with neonates with seizures.