Seizures in Neonates

Pathophysiology of Newborn Seizures—Unique Maturational Features Increasing the Likelihood of Seizures in the Newborn

A seizure results from an excessive synchronous electrical discharge (i.e., depolarization) of neurons within the central nervous system. Neuronal depolarization is produced by the influx of sodium (Na ⁺ ), and repolarization is produced by the efflux of potassium (K ⁺ ). Maintenance of the potential across the membrane requires an energy (adenosine triphosphate, ATP) dependent pump, which extrudes sodium and takes in potassium. In the newborn brain, it appears that excessive depolarization may occur because of the imbalance of neural excitation over inhibition. The factors contributing to this include developmental factors unique to the immature brain with an excess of excitatory neurotransmitters, particularly in relation to the principal excitatory neurotransmitter, glutamate, and a relative deficiency of inhibitory neurotransmitters. Developmentally, this enhanced excitation is important for activity-dependent synaptogenesis but predisposes to excitation and seizures. Secondly, the common pathologic processes in the newborn brain of hypoxemia-ischemia and hypoglycemia can result in failure of the ATP-dependent sodium-potassium pump disabling the cell from maintaining a stable membrane potential. Finally, other molecules can influence the membrane's sensitivity to depolarizations, such as calcium and magnesium that interact with the neuronal membrane to inhibit Na ⁺ movement. Thus, hypocalcemia or hypomagnesemia increase Na ⁺ influx resulting in depolarization.

In understanding why and how the seizure phenomena in newborns differ from those observed in older humans, it is important to understand that in the vast majority of neonatal seizures, electrical onset is focal or multifocal with the spread of the seizure occurring within one hemisphere and secondary generalization to the contralateral hemisphere occurring only rarely. Thus, newborns rarely have well-organized, generalized tonic-clonic seizures, and premature infants have even less well-organized seizures than do term infants. The precise reasons for these differences relate to the status of neuroanatomical and neurophysiologic development in the perinatal period. Neuroanatomically, a propagation of seizures appears related to the completed cortical lamination and organization with myelination of cortical efferent systems and interhemispheric commissures. The relatively advanced cortical development apparent in limbic structures in the human newborn infant and the connections of these structures to the diencephalon and brainstem may underlie the frequency and dominance of oral-buccal-lingual movements (e.g., sucking, chewing, or drooling), oculomotor movements, and apnea as clinical manifestations of neonatal seizures.

From a neurophysiologic viewpoint, the relation of excitatory to inhibitory synapses is important in determining the capacity of a focal discharge to both form and then to spread to contiguous and distant brain regions. Strong evidence indicates that the rates of development of the excitatory and inhibitory synaptic activities differ in the newborn cerebral cortex ( Fig. 55.1 ). Excitatory activity is mediated by glutamate through two key receptor types, N-methyl- d -aspartate (NMDA) and alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA). These two excitatory receptors are the predominant neurotransmitter receptors found in the immature brain, with a relative paucity of the principal inhibitory receptors for gamma-aminobutyric acid (GABA). Indeed, the neonatal period is characterized by levels of excitatory neurotransmitter expression and function that exceed those observed in adult cortical neurons, while inhibition is not yet at adult levels (see Fig. 55.1 ).

Moreover, properties of these two glutamate receptors enhance their excitatory function. NMDA receptors in the neonatal period exhibit prolonged duration of the NMDA-mediated excitatory postsynaptic potential, reduced ability of magnesium to block NMDA receptor activity, diminished inhibitory polyamine binding sites, and a greater sensitivity to glycine enhancement. Similarly, AMPA receptors in the neonatal period are deficient in the GluR2 subunit responsible for rendering the AMPA channel impermeable to calcium. Thus these immature AMPA receptors are permeable to calcium and, as a consequence, enhanced excitation. In addition, early in development, the principal inhibitory neurotransmitter, GABA, acts at the major postsynaptic GABA _A receptor (GABA _A ) to produce excitation rather than inhibition, as occurs later in development. Consistent with these developmental phenomena, it is easier to produce epileptic activity in the immature animal than in the adult.

Insights into the critical developmental relationship between neuronal chloride (Cl ⁻ ) levels and Cl ⁻ transport in the perinatal period have major implications for understanding the basis of GABA excitation and thus key clinical and therapeutic aspects of neonatal seizures. GABA activation of the major postsynaptic GABA _A receptor causes Cl ⁻ flux. In the mature neuron, there is Cl ⁻ influx down an electrochemical gradient. However, in developing brains, at maturational stages comparable to the human perinatal period, GABA activation causes Cl ⁻ efflux, and GABA activation is, therefore, excitatory. The basis for this paradoxical effect relates to a developmental mismatch between the two Cl ⁻ transporters that determine neuronal Cl ⁻ levels. Thus in the perinatal period, in human cerebral cortex, the expression of the Na ⁺ -K ⁺ -Cl ⁻ cotransporter (NKCC1) responsible for Cl ⁻ influx reaches a peak, whereas the expression of the K ⁺ -Cl ⁻ cotransporter (KCC2) responsible for Cl ⁻ efflux is relatively low. The result is a high internal neuronal level of Cl ⁻ , so when the GABA _A receptor is activated there is efflux (rather than influx) of Cl ⁻ resulting in depolarization and excitation. These findings may explain the therapeutic inconsistency of GABA agonists, such as phenobarbital and benzodiazepines, to be effective anticonvulsants in the newborn infant with seizures. The NKCC1 inhibitor, bumetanide, has potent antiseizure properties by enhancing GABA-mediated inhibition through blockage of Cl ⁻ uptake and lowering of neuronal Cl ⁻ levels (see later). Moreover, because the maturation of the two cotransporters and neuronal Cl ⁻ levels occurs in a caudal-rostral direction, spinal cord and brainstem motor neurons would be expected to exhibit GABA-mediated inhibition before the cerebral cortical regions. This maturational process could explain the frequent occurrence of electroclinical uncoupling/dissociation in which antiseizure medications with GABA agonist mechanisms (i.e., phenobarbital and benzodiazepines) suppress motor manifestations of seizures (by spinal cord and brainstem inhibition) but not cortical EEG manifestations (due to lack of cortex inhibition). The presence of a relative overexpression of NKCC1 versus KCC2 has been documented in postmortem human neonatal brain, and combined with the efficacy in rodent neonatal seizure models, studies have been initiated to examine the safety and pharmacokinetics of the use of bumetanide as a combination anticonvulsant in human infants suffering from acute neonatal seizures. Results to date are conflicting on safety and efficacy (see later).

Biochemical and Physiologic Consequences of Seizures in the Immature Brain—Implications for the Clinician

The most prominent acute biochemical effects of seizures involve energy metabolism. Seizures are associated with a greatly increased rate of energy-dependent ion pumping, which is accompanied by a fall in the concentration of ATP and phosphocreatine, the storage form of high-energy phosphate in brain. The resulting rise in adenosine diphosphate (ADP) leads to stimulation of glycolysis and to a shift of the redox state in the cytoplasm toward reduction (i.e., NADH). The excess of lactate has the beneficial effect of causing local vasodilation and a consequent increase in local blood supply and substrate influx. In addition, seizures are associated with elevated blood pressure, which contributes to increased cerebral blood flow (CBF) and substrate influx. This pressor effect is presumed to be a central autonomic component of the seizure, because it can be interrupted by section of the spinal cord or by administration of sympathetic ganglion-blocking agents.

Despite these important compensatory factors, in experimental animals, neonatal seizures are accompanied by reductions in brain glucose concentrations. In the neonatal rat, rabbit, dog, and monkey, despite normal or slightly elevated blood glucose concentrations, brain glucose concentrations fall dramatically within 5 minutes of onset of seizure to nearly undetectable levels after 30 minutes. Concomitant with the fall in brain glucose is a rise in brain lactate, which is used readily as a metabolic fuel in the neonatal brain. This fall in brain glucose concentration and rise in brain lactate are directly reminiscent of a hypoxic-ischemic brain insult and presumably relate to the accelerated rate of glucose utilization in an attempt to preserve supplies of phosphocreatine and ATP. Glucose conversion to lactate, which is accelerated with neonatal seizures, results in only two molecules of ATP for each molecule of glucose, as opposed to the 38 molecules of ATP generated when pyruvate enters the mitochondrion and is oxidized to carbon dioxide. MRS studies by Younkin and colleagues in the human newborns demonstrate the relevance of these experimental data to the clinical situation ( Fig. 55.2 ). Four newborns had seizures during MRS imaging. The seizures resulted in substantial (~50%) decrease in the phosphocreatine to inorganic phosphate (PCr/Pi) ratio. One newborn's seizures were successfully treated with intravenously administered phenobarbital, which caused an immediate increase in the PCr/Pi ratio. Further, newborns had PCr/Pi ratios of less than 0.8 during seizures and developed long-term neurologic sequelae, indicating that neonatal seizures may increase cerebral metabolic demands above energy supply, thereby causing or exacerbating injury. These observations indicate that seizures may lead to secondary brain injury in an already injured neonatal brain and, therefore, have important implications for prognosis and therapy.

The deleterious effects of seizures may be divided into those related to prolonged seizures (in which the most prominent feature is cell loss) and those related to briefer recurrent seizures (in which the most prominent feature is altered development). While minimal data are available in human newborns, experimental studies are abundant, primarily in developing rodent models. Importantly, although the threshold for seizure generation is lower in the developing brain than in the mature brain, developing neurons are less vulnerable to injury from single prolonged seizures than are mature neurons. This may be due to a lower density of active synapses, lower energy consumption, and immaturity of relevant biochemical cascades to cell death.

For prolonged seizures, the best-documented mechanisms leading to brain injury include hypoventilation and apnea, which may result in hypoxemia and hypercapnia; cardiac dysfunction and diminished cardiac output as late complications of seizures, resulting in hypotension, diminished cerebral blood flow (CBF) and impaired energy metabolism. Importantly, increases in CBF have also been documented. In a study of 12 newborns with seizures, ictal measurements of regional CBF by single photon emission computed tomography showed a 50%-150% increase, and this increase occurred in newborns with subtle seizures and EEG-only seizures. Although the increase in CBF with seizures may be initially an adaptive response to increase substrate supply to the brain at a time of excessive metabolic demand, this response could become maladaptive in some newborns. For example, depending on such factors as the gestational age of the newborn or the neuropathologic substrate for the seizures, some newborns may have highly vulnerable capillary beds, such as the germinal matrix in the premature infants or the margins of ischemic lesions in premature infants or asphyxiated term newborns.

Repeated prolonged seizures may be deleterious for the brain, even in the absence of prominent disturbances of ventilation or perfusion. Such prolonged and repeated seizures eventually lead to decreases in brain ATP and phosphocreatine concentrations such that progressive and irreparable brain injury may result. Nevertheless, most studies indicate that the neonatal brain is more resistant to seizure-induced neuronal necrosis than is the adult brain.

An additional mechanism for the genesis of brain injury with severe seizures relates to excitatory amino acids. Injury to neuronal dendrites and cell bodies, the most prominent acute manifestations of injury from seizures, occurs particularly in limbic structures (e.g., hippocampus) and in distant sites intimately connected with limbic structures (e.g., selected areas of thalamus and cerebellum). When diminution of energy supplies is added, the energy-dependent reuptake systems for excitatory amino acids in presynaptic nerve endings and astrocytes are impaired, and the local accumulation of the neurotransmitters is accentuated. A particular vulnerability of the developing brain of the newborn may relate to the rich expression in the developing brain of glutamate receptors, which appear to play an important role in neuronal differentiation and plasticity.

Although most evidence does not suggest serious structural or functional defects from a single neonatal seizure, recurrent seizures, even if not prolonged, appear to be associated with long-term functional, morphologic, and physiologic deficits, particularly functional deficits in cognition. Visual–spatial memory and learning have been particularly involved, and these deficits are consistent with the locus of the principal structural deficits in the hippocampus. The morphologic correlates of the functional disturbances involve neuronal developmental abnormalities rather than neuronal cell loss. The most severe disturbances occur in the hippocampus and include dendritic spine loss in CA3 pyramidal cells and a distinctive pattern of synaptic reorganization of axons and terminals of the dentate granule cells (i.e., mossy fibers). The degree of this “sprouting” of mossy fibers correlates with the severity of the cognitive deficits. Additionally, dentate granule cell neurogenesis, which, unlike in other cortical areas, persists in the neonatal period, is impaired after recurrent seizures. Recurrent seizures also lead to physiologic and molecular alterations that favor subsequent neuronal excitability and, therefore, epileptogenesis ( Fig. 55.3 ), as well as the occurrence of neuronal injury with subsequent insults. Alterations include increases in excitatory amino acid receptors (NMDA and AMPA/kainate), decreases in GABA receptors, post-translational alterations in AMPA receptors (resulting in a decrease in the GluR2 subunit) that render them permeable to calcium, imbalanced excitatory and inhibitory systems, and altered intrinsic neuronal membrane properties, all of which favor excitation. A critical question, of course, is the extent to which these changes occur in the human newborn who experiences recurrent seizures and what seizure burden may produce such adverse neurobiologic consequences. Although this question remains unanswered, there is mounting evidence that seizures are associated with less favorable neurobehavioral outcomes.

Classification of Seizures in the Newborn

Seizure Etiology

The majority of neonatal seizures occur in the context of acute neurologic disorders. Thus most neonatal seizures may be considered acute symptomatic seizures, which have been defined as seizures occurring at the time of a systemic insult or in close temporal association (often 1 week) with a documented brain insult. The current International League Against Epilepsy classifies seizure causes as genetic, structural/metabolic, and unknown. Within that classification scheme, the majority of neonatal seizures are structural/metabolic in etiology.

Determination of the seizure etiology is critical, because it affords the opportunity to provide specific treatment and important prognostic information. While there are many causes for neonatal seizures, a relatively limited group of etiologies accounts for the majority of affected newborns. The most common causes and their usual time of onset in premature or full-term infants are shown in Table 55.3 . The most common underlying etiologies are hypoxic-ischemic encephalopathy, stroke, intracranial hemorrhage, intracranial infections, and cerebral dysgenesis. Less common but important etiologies include inborn errors of metabolism and neonatal epileptic syndromes, such as benign familial neonatal epilepsy, benign nonfamilial neonatal seizures, early myoclonic epilepsy, early infantile epileptic encephalopathy, and malignant migrating partial seizures of infancy.

The Neonatal Seizure Registry consortium of seven tertiary care pediatric centers in the United States prospectively collected data related to etiology in a cohort of 426 newborns with seizures who underwent cEEG. The most common seizure etiologies were hypoxic-ischemic encephalopathy in 38%, ischemic stroke in 18%, neonatal onset epilepsy in 13%, intracranial hemorrhage in 11%, neonatal genetic epilepsy syndrome in 6%, congenital cerebral malformation in 4%, and benign familial neonatal epilepsy in 3%. Additionally, for all these etiologies, the seizure burden was high, with 59% of subjects having >7 electrographic seizures and 16% having status epilepticus. There was no significant difference in seizure burden between preterm and term newborns or among the three most common causes of seizure (hypoxic-ischemic encephalopathy, ischemic stroke, and intracerebral hemorrhage). These etiologies were similar to those reported in a study by Weeke and colleagues of 378 newborns obtained over a 14-year period with seizures confirmed by EEG or aEEG from a level 3 neonatal intensive care unit. The most common etiologies identified were hypoxic-ischemic encephalopathy (46%), intracranial hemorrhage (12.2%), and perinatal arterial ischemic stroke (10.6%). These etiologies are quite similar to those found in a study by Tekgul and colleagues in which 89 newborns underwent careful etiologic evaluation. The most common etiologies were global hypoxic-ischemic encephalopathy in 40%, focal ischemic injury in 38%, intracranial hemorrhage in 17%, cerebral dysgenesis in 5%, transient metabolic disturbance in 3%, infection in 3%, and an inborn error of metabolism in 1%. The etiology remained unknown in 12%. Thus, in summary, three key conditions account for nearly 75% of neonatal seizures—hypoxic-ischemic brain injury (40%-50%), arterial stroke (10%-15%), and intracranial hemorrhage (10%-20%). The next two most common etiologies are intracranial infection (5%) and cerebral dysgenesis (5%). The remaining less common conditions, accounting for 5%-10% of all seizures, remain important due to potential therapeutic interventions in transient metabolic disorders and inborn errors of metabolism.