Neonatal seizures: Have we got the treatment right?

Abstract

Clinical seizures are a relatively common occurrence in the neonatal intensive care unit (NICU) setting with an incidence of up to 2.6 per 1000 live births in term infants and as high as 50–130 per 1000 live preterm births.1,2 With the increasing use of bedside electrophysiological monitoring of the brain within the NICU, there is a growing recognition that the true incidence of seizures in the sick newborn may be even higher. The outcome of neonatal seizures is poor with observational studies showing that seizures in the newborn period are associated with epilepsy in childhood with an incidence ranging from 17%3 to 56%.4 Electrographic seizures in neonates have also been correlated with the subsequent development of microcephaly and severe cerebral palsy.5 In both clinical and purely electroencephalographic seizures it remains unclear what the optimal management for neonatal seizures is and whether more aggressive anticonvulsant therapy would improve this high risk of adverse neurological outcome.6,7 The newborn brain has a decreased seizure threshold compared to that of the adult. This predisposition of the newborn brain to produce seizure discharges is partly related to increased excitatory amino acid activity mediated by glutamate at N-methyl D-aspartate (NMDA) receptors, together with a decreased binding at inhibitory gamma-amino butyric acid (GABA) receptors and an immature isoform of the GABAA receptor being pro-convulsant.8 Whether or not this increased propensity for seizures is coupled with an increased risk of cerebral injury from the seizures themselves is unclear. The prognosis from neonatal seizures is largely predicted by the aetiology of the seizures, rather than the frequency and length of the seizures themselves.9 Following such an aetiology–prognosis link, there are scenarios in which seizure therapy may have questionable impact on outcome. For example, for those infants with seizures resulting from structural developmental brain abnormalities there is a greater than 90% risk of neurodevelopmental impairment among survivors. In this setting, the early neonatal presentation of seizure activity represents very disturbed neuroanatomy, with this dysgenesis being the major determinant of outcome. Thus, treatment of seizures in this setting may achieve little. This view would also be more broadly supported by a Cochrane review of anticonvulsant therapy for neonatal seizures of more mixed aetiologies. Booth and Evans found that there were three randomised controlled trials eligible for inclusion, allowing data to be meta-analysed from a total of 109 preterm and term babies.10 The data from this small number of subjects showed no evidence of a reduction in mortality or neurodevelopmental morbidity amongst seizing infants successfully treated with anticonvulsants compared to those who were not. However, for those aetiologies that leave the brain in a metabolically compromised state (e.g. hypoxia-ischaemia or hypoglycaemia) there is some evidence that the additional burden of seizure activity may itself be detrimental. Despite an increased awareness of nonasphyxial causes for newborn encephalopathy,11 in the term newborn infant perinatal asphyxia remains a significant contributor to neonatal seizures.12 The incidence of perinatal asphyxia is reported at 1–2 per 1000 live births with subsequent moderate or severe encephalopathy occurring in 0.5 per 1000 live births.13 Infants with Sarnat stage 2 encephalopathy,14 a key feature of which is the presence of clinical seizures, have a risk of subsequent neurodevelopmental impairment (or conversely normal outcome) of approximately 50%.15 So with considerable scope to improve outcome for these infants the question of whether or not to treat seizures is of importance. A Cochrane review by Evans and Levene16 looked specifically at the use of anticonvulsants in the context of term-born infants with perinatal asphyxia. In this review, five trials recruited a total of 155 subjects to examine the effect of anticonvulsants in infants with asphyxia, regardless of whether or not they had clinical seizures at trial entry. There was no significant difference in either mortality or neurodevelopmental outcome detected for the infants who received anticonvulsants. The authors concluded that further randomised controlled trial data is required in order to answer the question of whether there is therapeutic benefit from anticonvulsants in the context of either neonatal seizures or perinatal asphyxia. Having failed to find definitive evidence as to the impact of anticonvulsant therapy on outcome from randomised controlled trials it is reasonable to explore other levels of evidence. Some of the most compelling evidence in this area comes from the experimental animal model work. Wirrell et al.17 provide evidence in the newborn rat pup that seizures, in the context of hypoxia-ischaemia, exacerbate neuronal injury. By exposing different groups of rats to a controlled hypoxic ischaemic insult, they demonstrated the typical microscopic changes of neuronal cell death. To one group of rat pups they also administered kainic acid – a potent seizure inducer – after an identical hypoxic-ischaemic insult. The histopathological evidence of neuronal injury was significantly greater amongst this group. Seizures in the rat pup have also been linked to subsequent impairment of memory and learning capacity.18 The extrapolation of such data to the human setting remains questionable. Seizures in the rat pup are thought to originate from the hippocampus, whereas those in the human infant are usually neo-cortically driven, providing us with some room for scepticism. However, if based on the strength of such evidence we were to assume that seizures in the context of perinatal asphyxia would be best treated, then unfortunately there remain significant challenges in achieving this goal. One randomised controlled trial examining the efficacy of anticonvulsant therapy in 59 neonates with electroencephalogram (EEG)-confirmed seizures demonstrated that adequate seizure control was achieved for less than 50% of infants in either group with monotherapy phenobarbitone or phenytoin. When the other agent was added, control of seizures was attained in approximately 60% of cases.19 Thus, with the standard anticonvulsant agents currently utilised in clinical practice in the NICU, adequate seizure control may only be achieved in 60% of infants. In this month's journal, Carmo and Barr present the results of a survey across Australia and New Zealand in which management of neonatal seizures in the context of perinatal asphyxia is examined.20 They report that we are predominantly using phenobarbitone and phenytoin as first- and second-line therapy when treating neonatal seizures in the context of perinatal asphyxia. Thus, along with the previous study we can reflect that we are currently not effectively treating a significant proportion of our newborn infants with seizures. There have been isolated case reports of the use of newer anticonvulsants21,22 in the neonatal population, but such agents remain experimental in use. In Europe there has been increased use of lignocaine rather than phenytoin as a second-line agent. Boylan et al.23 found that lignocaine was more effective than a benzodiazepine as a second-line agent in the management of EEG-confirmed seizures. However, adverse effects such as cardiac arrhythmias24 have also been reported with its use in this context. Others report that it can induce seizures when used as analgesia for circumcision.25 The utility of lignocaine as an anticonvulsant in neonates is currently being evaluated in a randomised controlled trial (L. deVries, pers. comm.). In conclusion, the question still requiring an answer is ‘do anticonvulsants given to newborns with seizures improve mortality or neurodevelopmental outcome?’ The answer to this critical question is still unknown and very worthy of study. If we continue to treat electroencephalographic seizures aggressively, conversely one could ask ‘could there be any potential harm from the use of anticonvulsant medications in the newborn during such a critical phase of cerebral development?’ There has been a suggestion that phenobarbitone may inhibit normal brain development in the rat pup26 but no data in human series to suggest harm. Whilst we await more definitive evidence on this topic a pragmatic approach must be taken, and this is reflected in Carmo and Barr's survey.20 Uncertain as to the potential complications of seizures, the majority of us elect to treat asphyxial seizures, and do so in a fairly uniform manner, despite its advertised inefficiency. This should not distract us, however, from the task of obtaining the evidence we currently lack, to ensure that our practice is both effective and based on sound evidence.