You are called to the intensive care unit to see a 6-year-old girl admitted with encephalitis. The child presented with 2 days of fever, confusion and lethargy and was admitted following a generalized tonic-clonic seizure lasting 8 min, which was treated with intravenous lorazepam followed by levetiracetam. When you see the child she is stuporous and does not respond to verbal commands but shows no clinical signs of ongoing seizures. The staff in the intensive care unit wish to know whether they should call in a technician to obtain an EEG. In a child who has already had a convulsive seizure, is on antiepileptic therapy and is showing no signs of ongoing seizure activity, will EEG monitoring add anything to her management? In the current issue of Brain, Payne et al. (2014) provide evidence to suggest that it might.
There is a long-standing controversy as to whether seizures, either clinical or subclinical, are harmful in critically ill children. Many clinicians are of the mindset that the aetiology of seizures is the determinant of outcome and that seizures themselves are simply a reflection of the severity of the underlying condition. In a condition such as encephalitis, it is argued, children with more severe disease are more likely to have seizures than children with milder forms. In other words, seizures are a marker of a severe encephalopathy but do not cause additional brain damage. As the presence or absence of seizures has no bearing on outcome, why perform expensive testing that does not alter the clinical course? According to this viewpoint, the emphasis should be on treating the cause of the seizures, not the seizures themselves. The counter-argument is that seizures, even if electrographic and without clinical manifestations, are harmful and need to be identified quickly and treated aggressively. Knowing whether seizures are harmful in the context of a severe illness is therefore of critical importance.
Payne and colleagues (2014) have now provided an answer to this contentious question. The investigators prospectively evaluated 259 children who underwent continuous video-EEG monitoring upon admission to the paediatric and cardiac intensive care units at the Hospital for Sick Children in Toronto. Strikingly, seizures occurred in 36% of the children with 9% experiencing status epilepticus. To determine whether seizures had an impact on neurological function, the investigators used a Paediatric Cerebral Performance Category (PCPC) score to estimate neurological status before and after hospitalization. Neurological decline was seen in 67% of this group of critically ill children. Moreover, neurological decline was clearly related to seizures; in the children with PCPC worsening the mean seizure burden (calculated as the maximum percentage of any given hour occupied by electrographic seizures) was 15.7% compared to 1.8% for those without PCPC worsening. Using multivariable analysis, which adjusted for diagnosis and illness severity, Payne et al. (2014) show that seizures, independent of illness severity, result in increased morbidity, but not mortality.
The finding that seizures, whether clinically evident or not, are associated with poor outcomes would not be surprising to basic science investigators who demonstrated decades ago in animals that electrographic seizures can result in substantial brain damage, even in the absence of seizure-induced systemic changes in oxygenation, temperature or blood pressure, or seizure-induced electrolyte abnormalities or hypoglycaemia. Meldrum and colleagues (1973) and Brierley et al. (1972) compared the brain damage caused by convulsive and non-convulsive seizures in adolescent baboons. Prolonged convulsive seizures lasting a number of hours were induced by the intravenous injection of bicuculline (a GABA antagonist) in Papio papio. The seizures resulted in severe damage throughout the brain including the neocortex, thalamus, cerebellum and hippocampus. When the same dosage of bicuculline was given to paralyzed and artificially ventilated animals, there were no behavioural signs of seizures but the EEG showed electrographic status epilepticus. Although homeostasis was maintained in these baboons, the animals experienced substantial brain damage, albeit not to the same degree as animals with convulsive status epilepticus. Thus, electographic seizures, in the absence of clinical signs or systemic effects, can result in neurological deterioration.
Why would seizures add to brain damage in critically ill children? It is well established that seizures can lead to a cascade of molecular events that can culminate in cell injury or death and synaptic reorganization (Holmes, 2002; Ben-Ari and Holmes, 2006). Seizures can cause massive depolarization of neurons triggering excessive glutamate release (excitotoxicity) (Olney et al. 1986; Fujikawa, 2005). This leads to an increase in intracellular calcium, causing a cascade of changes that ultimately result in cell death. Seizures can initiate processes in glia that lead to brain inflammation and can promote blood–brain barrier damage, again leading to cell death (Librizzi et al., 2012; Noe et al., 2013). The ‘sick’ brain appears to be particularly prone to the adverse effects of seizures. In the intact hippocampus model in rats, seizures aggravate the hypoxic state by accelerating rapid anoxic depolarization and associated neuronal death (Dzhala et al., 2000).
The work of Payne et al. (2014) is important in that it provides evidence from a well-studied cohort that seizures add to disease burden in a highly vulnerable group of children. Should the 6-year-old with encephalitis undergo EEG monitoring? Based on the work of Payne et al. (2014) the likelihood that the child is having subclinical electrographic seizures is substantial and monitoring should be performed. If seizures are detected should they be treated? As critically ill children with seizures fared worse than those without seizures, treatment appears warranted. Will such treatment improve outcome? Although this seems to be a reasonable assumption, Payne et al. (2014) could not answer this question. Over 80% of the children received antiepileptic drugs before or during continuous video-EEG and the association between seizure burden and neurological impairment occurred despite these interventions. Whether more aggressive or earlier therapy would have improved outcome is not known and, as the authors note, studies assessing the relationship between seizure burden and the timing, efficacy and safety of interventions are necessary. However, overall, the study by Payne et al. (2014) would indicate that critically ill children with neurological involvement should undergo EEG and video monitoring and, if seizures are detected, they should be treated aggressively.