Fig. 1
Burst suppression pattern seen 12 h after cardiac arrest. The patient is on TH protocol, with a core temperature of 33 °C
Anoxia and Neuronal Injury
Cerebral anoxia results in disruption of adenosine triphosphate (ATP) production, which then leads to glutamate release and activation of N-methyl-d-aspartate (NMDA) receptors. This increase in excitotoxicity results in cell death. Neurons in the cerebral cortex, cerebellar Purkinje cells, and the CA-1 region in the hippocampus are most vulnerable to neuronal death from anoxic injury due to their increased metabolic demands. The severity of injury depends upon the duration of anoxia and ability to restore cerebral blood flow [4]. Attempts have been made to identify neuroprotective agents that may ameliorate neuronal injury after anoxic injury. Barbiturates, glucocorticoids [5], sodium channel blockers [6], magnesium [7], and benzodiazepines [8] have all been proposed, but compelling clinical data are lacking.
Therapeutic Hypothermia
Approaches to neuroprotection after anoxic injury from cardiac arrest changed in 2002, with the publication of two pivotal trials. Both trials demonstrated a positive outcome when using mild TH for neuroprotection after cardiac arrest. In both trials, patients were cooled to between 32 and 34 °C for 12–24 h after cardiac arrest. Those patients that were cooled showed improved neurologic outcomes [9, 10]. As a result of these two trials, TH after cardiac arrest became standard of care. The protocols for TH often include sedation and paralytics, which confound the physical exam in this critical period.
Despite these initial large trials demonstrating improved neurologic outcomes after cardiac arrest, additional trials have shown that there may not be a significant benefit to TH. A 2013 randomized trial of 939 patients compared TH of 33 °C versus targeted temperature control at 36 °C and found no significant difference in regard to mortality or neurologic outcomes at 180 days [11]. A second randomized trial, published in 2014, compared prehospital cooling versus no prehospital cooling in 1539 patients, and found no difference in mortality or neurologic outcome at hospital discharge [12]. These data suggest that it may actually be the controlled avoidance of fever, rather than mild TH, which confers benefit in regard to neurologic outcomes after cardiac arrest.
The use of TH must be taken into account when considering the significance of EEG patterns, and data in the literature. As practice parameters change in regard to TH, published data regarding EEG findings may not be applicable across all clinical situations.
Early Descriptions of EEG After Circulatory Arrest and During Hypothermia
Experimental animal studies looking at the relationship between anoxia and EEG changes emerged in the 1930s, though human data were not published until the 1950s. Pampiglione made observations based upon intraoperative cardioplegia and restoration of cerebral blood flow with carotid massage. He described progressive diffuse slowing, followed by attenuation of cerebral activity, and return of cerebral activity after carotid massage [13]. Pampiglione expanded on his observations outside of the operating room (OR), correlating EEG findings after anoxic injury with outcomes. His descriptions of early EEG findings after cardiopulmonary arrest and clinical outcomes foreshadowed what would be repeatedly confirmed later [14].
Many of the early reports regarding EEG findings with anoxia did occur in the OR, where hypothermia was used in conjunction with circulatory arrest. These observations allowed for the determination that hypothermia had an effect on cerebral activity and EEG. In 1966, Harden published a series of OR cases, concluding that mild hypothermia alone had little effect on EEG, but that temperature affected EEG activity with circulatory arrest. EEG persisted longer after circulatory arrest with a temperature between 18.5 and 24.5 °C, versus between 28 and 32 °C. The postulation was cerebral function in moderate hypothermia had lower metabolic demands and was neuroprotective [15]. These early series provided a basis for expansion of data regarding EEG findings and prognosis after anoxic injury.
EEG After Acute Anoxic Injury
In patients who suffer an acute anoxic injury, the time period after successful resuscitation is fraught with uncertainty and waiting. EEG has been shown to provide important prognostic information in this critical situation.
Variability and Reactivity
The EEG background, in regard to variability and reactivity, can be a strong correlate of outcome. A well-modulated background contains a mix of frequencies and amplitudes, and varies in response in internal and external stimulation. In general, spontaneous variability is a positive sign, while an invariant background portends a poor prognosis. Nonreactivity as an indicator of poor prognosis has validity after TH as well (Fig. 2). After induced mild TH (33 °C), the false-positive rate of a nonreactive background and a poor outcome is 0.07 [16]. Caution needs to be taken to ensure that appropriately noxious stimulation, including deep painful stimulation, is deployed.
Fig. 2
No reactivity is seen in response to deep painful stimulation
Alpha and Theta Coma
Alpha and theta coma are specific nonreactive patterns that traditionally were correlated with poor outcome after anoxic injury. The alpha coma pattern was first described in 1975 [17]. The pattern was described as being similar to an awake pattern, with 8–13 Hz alpha activity, which was broadly distributed with slight spontaneous variability and was nonreactive (Fig. 3). All of the patients with anoxic injury in the initial study died [17]. Theta coma has been considered a variant of alpha coma. This pattern is characterized by broadly distributed 5–6 Hz theta activity with minimal spontaneous variability and reactivity. The initial patient in whom this was described after anoxic injury did not survive [18].
Fig. 3
Alpha coma. This pattern, consisting invariant diffuse alpha activity in a comatose patient, may be associated with poor outcomes if no reactivity is seen. Serial or cEEG is required to determine if the pattern resolves over time
Though alpha and theta coma were initially described as being associated with poor prognosis, further studies have shown that this is not the case, and outcomes can be more variable [19, 20]. In some instances, better outcomes may be predictable based upon features of the EEG. There are gradations of alpha or theta coma, some of which may not be as highly correlated with poor outcomes after anoxic injury. Features consistent with “incomplete” alpha or theta coma, and less likely to be associated with poor outcomes, include a pattern which is not entirely monotonous and hyporeactive (compared to nonreactive) and has a posterior, rather than anterior, distribution [21]. It is inadvisable to make determinations regarding prognosis when alpha or theta coma is seen at a single time point after acute anoxic injury. These coma patterns may evolve over time, and serial EEGs or continuous EEG (cEEG) monitoring are required to determine the true significance in an individual.
Burst Suppression
Burst suppression is defined as alternating bursts of cerebral activity with periods of background attenuation that comprises at least 50 % of the record [22]. Burst suppression can occur as a physiologic pattern, in response to certain medications, such as anesthetic agents, or moderate to severe hypothermia. It can also be pathologic, occurring after acute anoxic injury and is associated with the degree of injury. Bursts may occur in response to stimulation, variably include epileptiform activity, and may be time locked with myoclonus. Grading scales regarding EEG patterns and clinical outcomes have consistently regarded burst suppression as a malignant pattern [23–25], and electrographic seizures may arise out of this background [26]. When confounding variables, including medications and hypothermia, are absent, bursts suppression after cardiac arrest is correlated with a poor prognosis. A reactive burst suppression pattern does not portend a better outcome [27].
The specifics of the composition of the bursts may provide more refined prognostic data. The concept of “identical bursts” has been advocated as being more specific for poor outcomes [28]. On visual analysis, bursts are considered identical if the initial 500 ms of each burst are consistent in morphology. Quantitative analysis provides objective evidence of identical bursts. This specific pattern was not seen in burst suppression due to etiologies other than anoxia, including with anesthesia.
Episodic Low-Amplitude Events
Episodic low-amplitude events are characterized by brief periods of diffuse background attenuation. Compared to burst suppression, this is a relative minor feature of the record, occurring intermittently (Fig. 4). Initial descriptions of this pattern after anoxic injury associated the finding with poor outcomes [29, 30]. However, later studies have found that this pattern is common after anoxic injury and has no association with poor outcomes. In the setting of TH, this pattern may be an alerting pattern or related with hypothermia or medications commonly used as part of clinical protocols [31].
Fig. 4
Episodic low-amplitude events. These are periods of background attenuation, which occur intermittently and compromise a minority of the record. They may occur with or without TH and in some instances may be related to medication, hypothermia, or alerting
Generalized Periodic Discharges
Generalized periodic discharges (GPD), previously referred to as generalized periodic epileptiform discharges (GPED), may occur as part of a burst suppression pattern or over a continuous background (Fig. 5). Most commonly in acute anoxic injury, GPDs are superimposed on an isoelectric background and are a feature of a burst suppression pattern. In this instance, GPDs are associated with poor outcomes. GPDs can also occur with myoclonic seizures, which is consistent with a poor outcome. It is not well understood if GPDs superimposed on a continuous background with spontaneous variability correlates with a better prognosis.
Fig. 5
GPEDs superimposed on an otherwise isoelectric background
Seizures and Status Epilepticus
Postanoxic seizures and status epilepticus (SE) are common after cardiac arrest and have significance in regard to outcomes. The majority of seizures in this population are subclinical, requiring cEEG monitoring to detect them [26, 32, 33]. NCS occur in 9–30 % of patients treated with TH. NCSE occurs in 10–12 % of these patients [31, 34–36]. After cardiac arrest, the mean time for onset of NCS and NCSE is 15 h, occurring during TH [26]. Those patients at risk for NCS can often be identified by EEG, as NCS are likely to be preceded by interictal epileptiform discharges [35] (Fig. 6).
Fig. 6
Evolving electrographic seizure arising from an otherwise isoelectric background
NCS and NCSE are nearly universally associated with poor outcomes, even with attempts at treatment. In multiple studies using cEEG monitoring, the presence of NCS or NCSE were associated with mortality rates from 80 to 100 %. When poor neurologic outcomes and mortality are combined, the mortality rate approaches 100 % [16, 26, 31, 34, 36, 37]. There have been reports of exceptional recovery after anoxic injury and SE. In these cases, however, other encouraging findings were present, including a reactive EEG, preserved brainstem reflexes, and preserved cortical responses to somatosensory-evoked potentials [38]. There has been some suggestion that focal seizures after anoxic injury may be more readily controlled with antiepileptic drugs (AED) and portend a slightly better outcome, but this has not been borne out well in subsequent literature [39].
Myoclonic status epilepticus (MSE) is often grouped with NCS and NCSE, and many studies do not make definite separations between the three. MSE occurs in the first 24 h after acute anoxic injury while the patient remains comatose. Myoclonus can often be triggered by stimulation, and correlates with GPDs or the bursts in a burst suppression pattern. Occasionally, myoclonus can be seen without EEG correlate in a comatose patient. In these instances, it is thought that the myoclonus arises from the brainstem. In either case, MSE is strongly associated with a poor outcome after anoxic injury, even with attempts at treatment [40].
Low-Voltage Output Record
A low-voltage output record after acute anoxic injury may also be referred to a low-voltage pattern. This pattern consists of background activity less than 10 μV and is not due to sedation or hypothermia. There is no reactivity to stimulation (Fig. 7). In the absence of confounding factors, this pattern correlates with poor outcomes with reported specificities as high as 100 % [41].
