Abstract
Torolira D, Suchomelova L, Wasterlain CG, Niquet J. Ann Neurol 2017;82:115–120.
Status epilepticus is common in neonates and infants, and is associated with neuronal injury and adverse developmental outcomes. γ-Aminobutyric acidergic (GABAergic) drugs, the standard treatment for neonatal seizures, can have excitatory effects in the neonatal brain, which may worsen the seizures and their effects. Using a recently developed model of status epilepticus in postnatal day 7 rat pups that results in widespread neuronal injury, we found that the GABAA agonists phenobarbital and midazolam significantly increased status epilepticus-associated neuronal injury in various brain regions. Our results suggest that more research is needed into the possible deleterious effects of GABAergic drugs on neonatal seizures and on excitotoxic neuronal injury in the immature brain.
Commentary
Treatment of seizures in newborns is an ongoing source of consternation for child neurologists. Armed with the knowledge that neonatal seizures are probably bad for the developing brain (1–3) and are recognized at high frequency via continuous video EEG monitoring (4), we are driven to try to stop seizures as expeditiously as possible (5, 6). In newborns with seizures due to hypoxic ischemic encephalopathy (HIE), therapeutic hypothermia has been shown to reduce both clinical and electrographic seizures and to improve neurodevelopmental outcomes. Unfortunately, when faced with neonates with HIE and seizures despite hypothermia, or neonates with seizures from other causes, we are stuck with phenobarbital as a first-line antiseizure medication, shown to be effective in fewer than half of the neonates who receive it (7, 8). We are also plagued by the knowledge that many antiseizure medications available for use in neonates, including phenobarbital, benzodiazepines, and phenytoin, cause widespread apoptotic neurodegeneration in the normal term-equivalent rodent brain when given at moderate to high doses (9). So we are left with the ongoing debate surrounding which is worse, the disease (neonatal seizures) or the remedy (antiseizure medications).
This debate has been difficult to resolve using experimental models of neonatal seizures, much less using data from human neonates. Some of the difficulty has arisen from findings that evidence of neuronal injury is often absent after seizures induced in rodents younger than 12 days old (10), even though long-term deficits in behavior and learning are produced (11, 12). In the current article of Torolira et al., the authors attempt to provide answers to the conundrum of the competing harms of remedy versus disease in neonatal seizures utilizing a rat model of neonatal status epilepticus induced by a modified lithium pilocarpine protocol. In an earlier publication (13), the investigators characterized this model, which consists of intraperitoneal (IP) injection of lithium chloride (5 mEq/kg) in rat pups on P6, followed by high-dose pilocarpine (320 mg/kg) together with scopolamine (1 mg/kg) on P7. A neonatal mouse at P7 is roughly equivalent to a late-gestation or term human infant, and is an age at which intracellular chloride levels are high in cortical neurons, leading to depolarizing responses to GABA-A receptor activation (14, 15). This protocol produced status epilepticus in 75% of tested animals, with a 24-hour survival rate of 70%. The EEG recordings showed continuous polyspike activity for an average of 54 minutes, which was correlated with behavioral seizure activity consisting of forelimb clonus or running seizures with vocalization. Importantly, unlike what has been found in many other models of neonatal seizures, extensive neuronal damage was identified using Fluoro-Jade B staining and caspase 3-a immunoreactivity in the hippocampus (CA1/subiculum and CA3), cortex (parietal, piriform, and lateral entorhinal), thalamus, basal ganglia, amygdala, and hypothalamus (13). These findings were attributed to the severity of status epilepticus in this model compared with others, combined with a high survival rate.
In the current article, Torolira et al. utilize this neonatal status epilepticus model to study the effects of two commonly used GABAergic antiseizure medications, phenobarbital and midazolam. The model was slightly altered by administration of pilocarpine 320 mg/kg subcutaneously rather than IP to P7 rats, which resulted in induction of behavioral status epilepticus in 100% of the pups. Ten minutes after onset of running seizures with vocalization, cohorts were treated with either IP saline, phenobarbital 10 mg/kg, or midazolam 3 mg/kg. In preliminary experiments, these doses were found to not cause neuronal degeneration in the absence of status epilepticus (although higher doses did). After administration of either phenobarbital or midazolam, behavioral seizure activity was reduced, and the pups appeared sedated. Survival rates were not significantly different between groups, ranging between 64 and 77 percent.
Neuronal degeneration as measured by Fluoro-Jade B staining was determined in multiple brain regions 24 hours after status epilepticus. Previous findings were confirmed (13), as animals with untreated status epilepticus again demonstrated increased numbers of degenerating neurons in the hippocampus, cortex, and subcortical structures. Importantly, administration of either phenobarbital or midazolam, at concentrations insufficient to cause neuronal degeneration on their own, resulted in significantly increased numbers of degenerating neurons throughout the brain after neonatal status epilepticus. This effect was inconsistent in the hippocampal and cortical regions but was robust for both agents in the thalamus and basal ganglia. In the thalamus, neuronal degeneration measured by Fluoro-Jade B staining was confirmed by caspase-3a immunoreactivity.
As a brief communication, the article by Torolira et al. provides some tantalizing data but also leaves many questions unanswered. Unlike their previous study (13), EEG recording during status epilepticus was not reported, so it is not known how phenobarbital or midazolam affected electrographic seizure activity. In human neonates, electrographic seizures can continue after phenobarbital treatment, while electoclinical seizures are controlled (8). Oxygen saturations were also not reported, so the incidence of hypoxia in the treated and untreated animals is not known. Given the reported sedation induced by phenobarbital and midazolam in these animals, concomitant hypoxia could potentially have contributed to the enhanced incidence of neuronal death. Although large groups of animals were studied (n = 26–36 per group), sex-specific results were not presented. This is relevant, as maturation of inhibitory GABAergic signaling and seizure susceptibility in neonatal rodents is sex dependent (14). The authors suggest that the increased neuronal death they measured after phenobarbital and midazolam may be due to GABAergic depolarization of “less mature” neurons. However, the finding of high levels of neuronal degeneration in the thalamus, an area that exhibits very early hyperpolarizing GABAergic responses in rodents (15), would argue against this hypothesis. Long-term functional outcomes were not measured. Finally, these results from lithium pilocarpine-induced status epilepticus cannot be extrapolated to neonatal seizures of other causes; for example, in a rat model of neonatal hypoxia-ischemia, phenobarbital was recently found to decrease brain damage and improve motor function (16).
Despite their limitations, the findings of Torolira et al. provide further information to be considered in the debate about whether or not currently used remedies are worse than the disease of neonatal seizures. Hopefully these findings will spur further research into the identification of safe, effective antiseizure medications for use in newborns.