Outcome After Neonatal Seizures

Renée A Shellhaas and Courtney J Wusthoff

Introduction

A child is at his highest lifetime risk of seizure during the first month of life. Neonatal seizures occur in 1–5 per 1000 term newborn infants, with estimates as high as 1 per 20 among preterm or very-low-birth-weight newborn infants (Vasudevan and Levene 2013). Although neonatal seizures can occur at any point up to 4 weeks after term equivalent age, the majority of neonatal seizures occur during the first few days after birth. Seizures in the neonatal period are most often symptomatic of acute brain injury and less commonly reflect an early-onset epilepsy syndrome.

Neonatal seizures are defined electrographically as sudden, evolving, rhythmic discharges lasting longer than 10 seconds. As with seizures in older age groups, neonatal seizures may manifest clinically as a variety of motor signs, automatisms, or autonomic fluctuations. The majority of seizures are subclinical, which is unique to neonatal seizures. These require EEG for diagnosis; as such, the true incidence of neonatal seizures may be higher than previously recognized. Because the majority of neonatal seizures occur in patients with coexisting neurological disorders, it is difficult to disentangle the contribution of seizures to later outcomes from the role of the underlying brain problem that caused the seizures. Furthermore, there are concerns that chronic antiepileptic drug administration to infants may in itself negatively impact brain development. Despite these complexities, a growing body of research has begun to elucidate the relationship between neonatal seizures and neurodevelopmental outcomes.

Diagnosis of neonatal seizures

Neonatal seizures are particularly challenging to diagnose. In contrast to seizures in older children and adults, the large majority of neonatal seizures are subclinical—they have no clear outward signs. Alternately, neonatal seizures are sometimes described as “subtle” or “nonconvulsive.” Because there are not reliable outward clinical signs, EEG is required for accurate diagnosis.

Traditionally, neonatal seizures had been diagnosed primarily by clinical observation. They were widely characterized and classified according to the system proposed by Volpe (2008). This system classifies seizures by their motor signs as focal clonic, multifocal clonic, generalized tonic, myoclonic, and subtle. In this classification scheme, subtle seizures include those with smaller, distinctive signs such as eye movements, lip smacking, swimming or pedaling movements, tonic posturing of a single limb, or apnea. Early research in neonatal seizures relied on clinical features for the diagnosis of seizure. However, EEG data have clarified that not all clinically suspicious events are epileptic seizures. For example, in their study of 349 neonates, Mizrahi and Kellaway (1987) reviewed EEG monitoring to correlate the observed clinical signs on video with epileptic seizures recorded on EEG. Of 415 events characterized as seizure on the basis of signs seen on video, 296 (71%) had an inconsistent or no relationship to ictal discharges on EEG. In particular, events of motor automatisms (mouth movements, eye movements) or myoclonus were often found to lack clear EEG correlate. These findings call into question the reliability of clinical observation for the diagnosis of epileptic neonatal seizures. Although it is possible that the events without EEG correlates truly were epileptic seizures, but with foci too deep to record on scalp EEG, the authors suggest that it is more likely that many of the abnormal events diagnosed clinically as seizures were, in fact, not epileptic seizures. These findings have been replicated in other studies using video EEG monitoring to demonstrate poor accuracy of bedside observation for diagnosis of neonatal seizures.

Although clinical observation alone may lead to incorrect diagnosis of abnormal movements as seizures, conversely, many subtle and subclinical seizures are missed when diagnosis is made by clinical observation alone. One study reviewed EEG recordings among 41 patients with neonatal seizures (Clancy and Legido 1988). During the EEG recordings, experienced EEG technologists noted any observed signs suspicious for seizure as they occurred. The authors found that, overall, 79% of seizures had no distinctive clinical signs. More recently, 51 term neonates had continuous video EEG recordings reviewed to define the gap between clinical diagnosis and EEG-based diagnosis of seizure burden (Murray et al. 2008). Two-thirds of captured seizures had no clinical signs on video, whereas only 9% of all seizures were recognized and documented by the bedside nurse or physician. At the same time, 73% of the events documented by the bedside nurse or physician were not associated with electrographic seizures. Again, these findings confirm that clinical diagnosis is unreliable for neonatal seizures. That seizures in neonates are most often subclinical should be intuitive—unless the seizure originates or propagates to the motor cortex, there will be no abnormal movements, and a non-verbal newborn infant will never complain of a stereotyped sensory aura. Because of the preponderance of subclinical seizures, EEG monitoring provides the criterion standard for diagnosis. For this reason, it has been recommended that all neonates at high risk for seizures are assessed using full-array video EEG monitoring (Shellhaas et al. 2011).

More recently, amplitude-integrated EEG (aEEG) has gained popularity as a tool for seizure detection in the intensive care nursery. aEEG is a simplified trend, based on two- or three-channel EEG recording. The recordings are time compressed (6cm per hour) and processed to display a bedside trend for the nurse or neonatologist to interpret in real time. There is a wide variability in the sensitivity of aEEG for seizure detection, with reported sensitivities most often about 25–35%, and rarely as high as 85%, for individual seizure detection (Glass et al. 2013). Given this suboptimal rate of seizure detection, caution must be used in interpreting data regarding seizures diagnosed by aEEG.

Overall, significant evidence supports continuous EEG as the criterion standard for neonatal seizure diagnosis. Clinical observation alone suffers from high rates of both overdiagnosis of clinically suspected seizures and missing common subclinical seizures in these neonates. Similarly, aEEG fails to detect a significant percentage of neonatal seizures. This has implications for research regarding the outcomes following neonatal seizures. Studies that relate only clinical seizures to outcomes may be incomplete, by missing subclinical seizures. Furthermore, they run the risk of including infants with inaccurate diagnoses of neonatal seizure, based on clinical signs alone and/or including infants with only subclinical seizures in the “seizure-free” control group. The best evidence comes from those studies that apply the criterion standard for neonatal seizure diagnosis, continuous video EEG monitoring, but these typically lack long-term follow-up.

Aetiologies and comorbidities of neonatal seizures

When considering neurodevelopmental outcomes following neonatal seizures, there is a complex interplay between the role of the seizures themselves and the contribution of the underlying brain disorder that is causing the seizures. The large majority of neonatal seizures are symptomatic of an underlying acute process (Table 8.1). In North America and Europe, hypoxic-ischemic encephalopathy (HIE) is the most common cause of neonatal seizures, accounting for roughly half of term newborn infants with seizures. The incidence of associated seizures may have declined with a widespread use of therapeutic hypothermia as a neuroprotection strategy for neonates with HIE. Yet, studies using video EEG have demonstrated that about half of neonates receiving hypothermia still have seizures during the first 4 days of life (Glass et al. 2014). In these cases, the hypoxic-ischemic injury confers a risk of impaired development, independent from, and in addition to the risk, from the seizures. Stroke is the next most common aetiology in term newborn infants, causing approximately 10% to 15% of neonatal seizures (Vasudevan and Levene 2013). In preterm neonates, intraventricular hemorrhage is the second most common cause of neonatal seizures. Finally, meningitis, glucose and electrolyte derangements, and maternal drug withdrawal are all potential causes of symptomatic seizures.

Less common are neonatal-onset epilepsy syndromes. Some of these are now known to be the result of genetic mutations, inborn errors of metabolism, or cerebral malformations. Genetic disorders are increasingly identified in patients with neonatal-onset epilepsies. For example, mutations in the potassium channel gene KCNQ2 are now known to cause the majority of cases of benign familial neonatal seizures. Glucose transporter deficiency syndrome is an example of a metabolic disorder that similarly causes, difficult to control neonatal seizures. Pyridoxine-dependent epilepsy may also present in the neonatal period. Cerebral malformations have been estimated to cause up to 10% of all neonatal seizures.

Finally, because many of these children are critically ill, they often experience comorbidities, which may further affect neurodevelopment. For example, a child with seizures due to HIE is at neurodevelopmental risk due to the primary brain injury, in addition to the seizures, but may also have sepsis in the neonatal period, or feeding difficulties with poor nutritional intake, further elevating the risk of later neurodevelopmental impairment.

TABLE 8.1
Aetiologies of neonatal seizures

Categories of seizure aetiologies	Examples¹
Reversible causes	Hypoglycemia Hypocalcemia Hypernatremia Neonatal abstinence syndrome
Acute acquired brain injury	Hypoxic ischemic encephalopathy Arterial ischemic stroke Cerebral sinovenous thrombosis Intraventricular hemorrhage
Infection	Meningitis Encephalitis Toxoplasmosis, Rubella, Cytomegalovirus and Herpes simplex virus infections²
Congenital brain malformations	Malformations of cortical development (e.g. lissencephaly or focal cortical dysplasia) Disorders of prosencephalic development (e.g. holoprosencephaly)
Genetic epilepsy syndromes or metabolic disorders³	Benign familial neonatal seizures Glucose transporter deficiency syndrome Pyridoxine-dependent epilepsy Early myoclonic epilepsy of infancy Ohtahara syndrome Inborn errors of metabolism Urea cycle disorders