Challenge of Neonatal Anesthesia: Which Optimal EEG Target?

In this month’s issue of Anesthesiology, Corlette et al. present a review aiming to analyze published electroencephalography (EEG) profiles in anesthetized neonates. This article comes from a team renowned not only for their expertise in pediatric and neonatal anesthesia and their focused work on EEG monitoring in young children but also for their considerations on consciousness and its abolition in anesthetized neonates.

The authors found only nine observational studies describing the evolution of EEG in anesthetized neonates, reflecting the complexity and rarity of research in this field. Despite the weaknesses of most publications analyzed, a characteristic profile marked by the alternation of high- and low-amplitude EEG activity segments, termed discontinuity, was identified. This discontinuity pattern is characteristic of the EEG of the immature brain, and the authors discuss its presence in anesthetized full-term neonates and modulatory factors such as hypothermia or sevoflurane concentration. Several pathophysiological hypotheses are raised, and the diagnostic ambiguity with burst suppression tracings is analyzed.

EEG analysis enables visualization of the pharmacodynamic effects of γ-aminobutyric acid–mediated (GABAergic) anesthetic agents on the cerebral cortex. The complexity of interpreting EEG changes in anesthetized neonates is due to the need to consider both the particularities of the EEG trace in the neonate, whose cerebral cortex is still maturing, and the possible pharmacodynamic specificities of anesthetic agents on this immature brain.

Conventional EEG, obtained by electrodes on the scalp, captures a low-amplitude electrical signal representing the summation of the electrical potentials of cortical neurons, organized in columns. In awake adults, there is rich, desynchronized spontaneous activity, typically resulting in a low-amplitude EEG with rapid oscillations, dominated by alpha and beta types. GABAergic anesthetic agents induce dose-dependent neuronal inhibition, electrically translated by synchronization of neuronal discharges leading to characteristic profiles on the cortical EEG. These profiles show a dose-dependent slowing and increase in the amplitude of the slow oscillations, up to the occurrence of burst suppression and possible isoelectric tracing. Functionally, the adult brain is characterized by rich and complex neuronal activity, organized into networks with thalamo-cortical and cortico-cortical connections, involved in consciousness and reactivity to ascending stimuli. EEG tracings recorded in anesthetized children greater than 1 yr old are close to those observed in adults, differing mainly in oscillation amplitude changes. 

Conversely, in neonates, cerebral immaturity is associated with a distinctive and evolving EEG pattern in the first months of life. The dynamic processes of cerebral maturation result in changes in cortical neuron activity and hence in the EEG patterns.  Cerebral maturation begins in utero and continues rapidly through the first 3 months of life and then more slowly into adolescence. This postnatal maturation involves the continuation of several processes initiated in utero, such as intrinsic subcortical brain activity, white matter myelination, increase in cortical grey matter and dendritic growth of cortical pyramidal neurons, and synaptogenesis. These processes lead to the organization and progressive development of thalamocortical and corticocortical neuronal connections. Pharmacologically, γ-aminobutyric acid initially has an excitatory function, which changes to an inhibitory function during the third trimester of pregnancy.

Electroencephalographically, the premature brain is characterized by low and discontinuous activity, with the emergence of slow and large oscillations (delta) that increase with brain maturation, whereas the periods of discontinuity decrease.  In the full-term neonate, the EEG is mainly characterized by slow delta and theta oscillations. The discontinuity periods have almost disappeared, and when they persist transiently, they alternate with periods of slow activity (delta and theta). EEG evolves in the first months of life with an overall acceleration of oscillations, resulting in an increased dominant frequency and decreased amplitude, reflecting the complexification of neuronal electrical activity with cortical connection development.  Simply put, it could be proposed that there is a parallel evolution of neuronal organization, connection development, and EEG tracing complexity.

In this review, the authors discuss the presence and origin of discontinuity episodes (less than 25 µV) observed in EEG of full-term neonates anesthetized with sevoflurane. As previously seen, the term of discontinuity comes from the EEG in premature infants, reflecting limited neuronal activity due to immaturity. In full-term neonates anesthetized with sevoflurane, the presence of these discontinuous episodes could reflect dose-dependent GABAergic neural inhibition of physiologically already slow and easily depressible EEG. Similarly, hypothermia, associated with reduced cellular metabolism, could enhance neuronal inhibition, leading to a slowing of the EEG up to burst suppression. Burst suppression patterns are described in children and in adults as low-amplitude periods (less than 5 µV) possibly resulting from a decrease of the cerebral metabolic rate coupled with stabilization by adenosine triphosphate-gated potassium channels. 

In neonates, there is some ambiguity surrounding the description of these EEG segments of low activity, which can be considered either as true suppressions  or as the premise of suppressive segments or, as in this review, an EEG pattern close to suppression but specific to anesthetized neonates. The most interesting aspect of these EEG descriptions in anesthetized neonates is the almost systematic presence of these episodes of cortical inhibition at low doses of anesthetic, which in older children are associated with more classic EEG changes close to those observed in adults. Given the low complexity of interneuronal connections in the neonatal cortex, it may be hypothesized that inhibitory GABAergic action leads rapidly, at low doses of anesthetic, to a clear drop in cortical cerebral electrical activity.

Whether this cortical inhibition reflects the depth of the hypnotic component of anesthesia depends on the definition of hypnosis and the brain structures involved—isolated cerebral cortex, which is unlikely, or cortex and subcortex, which is more likely. Subcortical activity in anesthetized neonates is probably less easily inhibited than cortical activity. Indeed, the minimum alveolar concentration of sevoflurane (motor response to nociceptive simulation) is higher in young infants than in older children, whereas inhibition of EEG activity occurs at lower doses of sevoflurane.  Clinically, it is not uncommon to observe a hemodynamic or motor response to nociceptive stimuli in neonates, even with an isoelectric EEG. This suggests greater cortical sensitivity to GABAergic anesthetic effects compared to subcortical regions.

In the context of concerns about potential neurotoxicity of anesthetic agents on the developing brain, the question of target EEG effects arises. If the goal in neonate is, as in adults, to limit the hypnotic dose to avoid or minimize total cortical neuronal inhibition expressed as discontinuities or burst suppression episodes, then one risks subcortical structure reactivity during nociceptive stimulation, possibly causing adrenergic stimulation with stress hormone release or implicit memory trace formation facilitated by amygdala activation. Implicit memory, responsible for fear conditioning in animals, is present and stable from the first months of life, unlike explicit memory, which seems to emerge alongside language acquisition. Thus, the target of inhibiting subcortical reactivity in anesthetized neonates without overdosing the cerebral cortex seems to be crucial, possibly achievable through central analgesics such as opioids.

Beyond the challenges of interpreting EEG data from analyzed studies, owing to the variety of anesthetic agents used, from thiopental to sevoflurane, halothane, and propofol boluses, the main limitation of this review might be the lack of data and discussion on the hemodynamic state of anesthetized neonates. Neuronal metabolic activity and thus EEG patterns are highly dependent on cerebral blood flow, which is itself conditioned by blood pressure, especially below the autoregulation threshold. The studies of Rhondali et al.  have highlighted the vulnerability of young infants anesthetized with sevoflurane, particularly neonates, in terms of cerebral blood flow reduction, potentially occurring with a 20% decrease in mean arterial pressure. This hemodynamic approach to the effects of anesthesia on the EEG could be integrated into the overall reasoning of target effects of anesthesia in neonates, drawing inspiration from the triple (or double) low concept described in adults.