The incidence of neonatal seizures is 3.5/1000 live births¹, with hypoxic-ischemic brain injury the most common cause of seizures in the neonatal period. Clinical manifestations of neonatal convulsions are very diverse and not always easy to recognize, especially when they are subtle. Continuous EEG monitoring often reveals electrographic seizures (discharges without clinical manifestations), especially following initiation of therapy for clinical seizures^2,3. Only with the use of continuous monitoring can subclinical seizures be detected. So far there is no agreement as to whether these subclinical seizure discharges should be treated, although some studies suggest an adverse effect on neurodevelopmental outcome^4,5. Recent findings in animal studies show that neonatal seizures adversely affect the developing brain, as do anti-epileptic drugs^6-10.

Three recent publications^2,3,11 show that anticonvulsants, effective in adults, are not as effective in neonates. On the other hand, the neonatal brain appears uniquely susceptible to seizures because neonatal gamma-aminobutyric acid (GABA) receptors are excitatory (instead of inhibitory as in adults), and are functionally more active at this stage of life than are N-methyl-D-aspartate receptors. In this context, the hypothesis that a diuretic drug such as bumetanide could be used as an anti-epileptic drug in neonates is of particular interest^12,13. The severity of the seizures appears to be a stronger predictor of the success of treatment than the assigned agent¹¹, a hypothesis confirmed by Boylan et al², who report that electrographic seizures were common in infants who had severe hypoxic ischemic encephalopathy, particularly after treatment with phenobarbital.

Considering the potential harmful effect of anti-epileptic drugs as described by Bittigau et al¹⁰, should the clinician give a maintenance dose of anti-epileptic drugs after the initial treatment of neonatal seizures to prevent post-neonatal seizures? The question is further complicated by the report by Toet et al¹⁴, showing that post-neonatal epilepsy occurred in only 9.4 % of the patients following treatment of clinical and subclinical neonatal seizures. Following perinatal arterial stroke, the incidence of post-neonatal epilepsy was higher (18%), suggesting that that treatment of subclinical seizures may be of additional value. With the “real EEG” readings now available on the new digital aEEG machines, seizures have become easier to differentiate from artifacts, allowing for potential interventions.

The large variety of characteristics of the neonatal seizure make automatic seizure detection methods very difficult to implement, as was recently shown by Faul et al¹⁵. The authors report that neonatal seizures tend to migrate throughout the brain, contain rhythmic activity that can vary in frequency, and that multifocal seizures are common in neonates and often display simultaneous independent focal electrographic signatures. The morphology of the electrographic neonatal seizures also varies tremendously between individuals, and can vary within the seizure. Further, neonatal seizures often evolve in amplitude, but may suddenly end when reaching a maximum, or may continue and gradually wane.

There remain many questions to be answered about how to treat neonatal seizures most effectively. While continuous (a)EEG monitoring is an essential tool in providing answers, thoroughly effective and reliable automated neonatal seizure detection remains a long way in the future.

References:

The investigators found that NKCC1 was highly expressed in both neonatal rat and human cortex and that bumetanide, a diuretic drug, inhibited cortical seizure activity in neonatal rats both in vitro and in vivo at doses that have already been extensively tested/used in human neonates.

The investigators found that seizures were controlled in 13 of the 30 neonates assigned to receive phenobarbital (43%) and 13 of the 29 neonates assigned to receive phenytoin (45%). When combined treatment was considered, seizure control was achieved in 17 (57 %) of the neonates assigned to receive phenobarbital first and in 18 (62%) of those assigned to receive phenytoin first (p=0.67). The authors further note that severity of the seizures was a stronger predictor of the success of treatment than the assigned agent.

In summation, the authors found that phenobarbital and phenytoin are both equally ineffective in complete seizure control in neonates.

Toet et al addressed the incidence of post-neonatal epilepsy in full-term infants treated with anti-epileptic drugs for neonatal seizure discharges, as detected with amplitude integrated EEG (aEEG). 206 full-term infants were monitored using aEEG, and received anti-epileptic drugs for clinical as well as subclinical neonatal seizures. Follow-up data were analyzed for the development of post neonatal epilepsy, and their neurodevelopmental outcomes assessed at 3, 9, 18 months, 3 and 5 years of age. 169 (82%) neonates received 2 or more anti-epileptic drugs. Overall mortality was 39% (n=80). 41 of the 126 survivors (33%) were abnormal at follow-up, and 12 developed post-neonatal epilepsy (9.4 %).

84 children survived following hypoxic ischemic encephalopathy Sarnat grade II (n=92) and 6 (7%) developed post-neonatal epilepsy. In this subgroup no post-neonatal epilepsy was seen if seizures were controlled within 48 hours after birth and when not more than two anti-epileptic drugs were needed. Twenty-four children survived following intracranial hemorrhage (n=28) and only one (4%) developed post-neonatal epilepsy. Eleven children survived following perinatal arterial stroke (n=13) and 2 (18%) developed post-neonatal epilepsy.

In summation, the authors report that the incidence of post-neonatal epilepsy following treatment of clinical and subclinical neonatal seizures, detected with continuous aEEG, was 9.4%, and was only found in infants with an abnormal neurodevelopmental outcome. This figure is lower than previously reported in children who only received treatment for clinical seizures.

Sher et al describe a prospective study of the efficacy of seizure cessation by phenobarbital versus phenytoin in 59 term and preterm neonates. They introduced the phenomenon of uncoupling: the persistence of electrographic seizures despite the suppression of >50% of the clinical seizures following administration of either one or both anti-epileptic drugs.

The authors found that 58% of the neonates with persistent seizures after treatment had uncoupling of electrical and clinical expression of seizures, with phenobarbital and phenytoin resulting in equal rates of uncoupling. Twenty-four infants responded to the first choice of an anti-epileptic drug with no further seizures.

Boylan et al describe 3 seizure types in 24 neonates (electroclinical, electrographic, and clinical only) using video-EEG for at least 100 minutes within 12 hours of the first seizure. Griffiths neurodevelopmental outcome was assessed in all groups. One of their findings was that electrographic seizures were common in infants who had severe hypoxic ischemic encephalopathy, particularly after phenobarbital treatment.

Both authors found that electrographic neonatal seizures without clinical seizures are very common following anti-epileptic drugs.

Faul et al evaluated 3 published automated algorithms (Gotman, Liu and Celka) for detecting seizures in neonatal EEG. They tested 43 data files (one-minute, artifact-free EEG segments) containing seizure activity and 34 data files free from seizure activity. In an attempt to obtain improved detection rates, threshold values in each algorithm were manipulated and the actual algorithms were altered. The sensitivity of the 3 algorithms was respectively 62.5, 42.9 and 66.1% with a specificity of 64.0, 90.2 and 56.0%.

The authors conclude that the levels of performance achieved by automated seizure detection algorithms are not high enough for clinical use, reflecting the difficulties involved in detecting seizures in neonates and the lack of a reliable detection scheme for clinical use. The overlap of the frequency characteristics of seizure and non-seizure EEG, artifacts, and natural variances in the neonatal EEG cause great difficulties with the seizure detection algorithms. The algorithm performances figures for this neonatal data set are considerably worse than quoted in the original algorithm source papers.

Wistar rats 0-30 days old received i.p. administration of anti-epileptic drugs, β-estradiol, or normal saline, and were allowed to survive for up to 48 hours after injection. Their brains were analyzed 24 hours later. The investigators looked at histology of the whole brain to visualize degenerating cells and quantify the damage. In addition, total cellular RNA was isolated and total cellular proteins were separated.

In this animal model, the investigators show that at plasma concentrations relevant for seizure control in humans, phenytoin, phenobarbital, diazepam, clonazepam, vigabatrin, and valproate all cause apoptotic neurodegeneration. The combination of anti-epileptic drugs with different modes of action appears to result in a substantially higher apoptotic response compared to monotherapy.

LAST MONTH�S Q & A    April 2006 - Volume 3 - Issue 8

Last issue we reviewed the increasing use of continuous (a)EEG monitoring in the NICU, with a particular focus on subclinical seizures, and discussed the effectiveness and risk/reward of anti-epileptic drugs in neonates.