[1] 8w Neonatal Seizures: To Treat Or Not To Treat?

[1] 179w The immature brain is intrinsically hyperexcitable, a feature that, despite being crucial for learning, synaptogenesis and neuronal plasticity, predisposes the neonate to seizures. Seizures represent the most common neurologic manifestation of impaired brain function in this age group. Importantly, although seizure-induced neuronal injury is minimal in the "healthy" neonatal brain, the "metabolically-compromised" brain appears more vulnerable. Even in the "healthy" brain, however, seizures result in impaired learning, enhanced susceptibility to further seizures, and increased risk of brain injury with seizures later in life, as a result of altered hippocampal circuitry. Given these findings, an aggressive approach to neonatal seizures appears warranted. However, our current conventional therapies (including phenobarbital, phenytoin, and benzodiazepines), even when used in combination, are often ineffective in controlling seizures. Lidocaine may yield better efficacy but requires more study. Recent animal data suggest that alpha-amino-3-hydroxy-5-methyl-4isoxazole proprionic acid (AMPA) antagonists such as topiramate may have a neuroprotective role. However, further work is needed to confirm the safety of excitatory amino acid antagonists in neonates because there remains a prevailing concern that such agents may impair normal neurodevelopmental processes.

[1] 141w There are many reasons for the increased susceptibility of the immature brain to seizure activity (Table 1). First, the birth process itself entails several potential risks. Traumatic delivery could lead to intracranial hemorrhage or contusion. Various antepartum, intrapartum, and postpartum factors may impair oxygen and blood delivery to the infant's brain resulting in hypoxic-ischemic brain injury. This injury may be global or focal, and indeed the neonatal period is one of the highest risk periods for ischemic stroke in children. 4,5 Neonates are relatively immunocompromised and may develop central nervous system infection from exposure to an infected birth canal, other infants, or through iatrogenic means. Importantly, the neonatal brain has a greater propensity for seizures compared with the more mature brain because of a number of developmental factors (see the article in this issue by Wong for a more detailed discussion).

[2] 214w Although the prognosis of neonatal seizures has improved over the past several decades, approximately one third of survivors are still left with neurologic sequelae including motor deficits, mental handicap, and epilepsy. 6 Holden et al 7 found that neonatal seizures are associated with a 55-to 70-fold increased risk of cerebral palsy, a 5.3-fold increased risk of mental retardation, and an 18-fold increased risk of epilepsy. Preterm infants fare most poorly, with normal outcome occurring in only 35% of those less than 2,500 g and in 19% of those less than 1,500 g. 6 Much of the disability is because of the actual cerebral insult causing the seizures because underlying etiology is a strong predictor of outcome. 6 Favorable outcome is generally associated with familial neonatal seizures, hypocalcemic seizures, or primary subarachnoid hemorrhage, whereas poor outcome is the rule with severe hypoxic-ischemic encephalopathy, central nervous system malformations, and massive intraventricular/periventricular hemorrhage. 6 There has been debate about whether seizures per se cause brain injury in neonates. [8][9][10][11] Many clinical, epidemiological studies have suggested that seizures have a minimal effect on the developing brain. 8,12 However, although the immature brain appears to be relatively protected from seizure-induced injury, there is now evidence from both animal and clinical studies that neonatal seizures are not benign (Table 2).

[3] 92w In clinical practice, neonatal seizures are usually symptomatic of an underlying brain insult; however, most animal work on the effect of neonatal seizures has been done in otherwise "healthy" brains. Hypoxic-ischemic encephalopathy is the most common underlying etiology in term neonates with seizures, accounting for approximately 60% to 65% of all cases. 6 Seizures begin within the first 24 hours of life and are often prolonged and recurrent. In a brain that is already metabolically compromised by prior asphyxia or hypoglycemia, the potential for seizure-induced brain damage may be very much increased.

[4] 248w Several studies have examined this question in animal models with varying results. Cataltepe et al 40 induced seizures with bicuculline in P7 rats 2 to 12 hours after a 2-hour hypoxic-ischemic insult and found no enhanced damage. Towfighi et al 41 induced seizures with either kainic acid or flurothyl in P7 and P13 rats 2 to 24 hours after a 60-to 90-minute hypoxic-ischemic insult and also found no increased damage. However, using a less severe hypoxic-ischemic insult and a shorter recovery period between hypoxiaischemia and seizure induction, Wirrell et al 42 found that kainic acid-induced seizures did substantially increase hippocampal injury when superimposed on a moderate hypoxic-ischemic insult in P10 rat pups. Animals were exposed to either 15 or 30 minutes of hypoxia ischemia using the modified Levine preparation 43 and were then recovered for 30 minutes before receiving either kainic acid or saline. Pups receiving kainic acid showed continuous electrographic seizure activity for a mean duration of 4 hours 42 minutes. Surviving animals were examined neuropathologically at 3 and 20 days after seizure. No brain injury was seen in animals exposed to only 15 minutes of hypoxia ischemia, regardless of whether seizures occurred, but damage was present in those receiving 30 minutes of hypoxia ischemia. However, the group receiving 30 minutes of hypoxia-ischemia and superimposed seizures showed a significantly greater degree of injury in the hippocampus but not the cortex, indicating that seizures superimposed on moderate hypoxia-ischemia exacerbate brain injury in a topographically specific manner (Fig. 1).

[5] 221w To ascertain the underlying mechanisms responsible for the exacerbation of brain damage, this group studied alterations in hippocampal high-energy phosphate reserves and extracellular release of amino acids. 44 Hypoxia ischemia for both 15 and 30 minutes resulted in a decrease in adenosine triphosphate (ATP), which recovered significantly by 30 minutes, when seizures were induced. Seizures then led to both a decrease in ATP and phosphocreatine, with the ATP remaining significantly below homologous control animals In a recent study in 90 term human neonates with hypoxic-ischemic encephalopathy, Miller et al 45 showed that superimposed seizures further damage the brain. A seizure score was determined for each infant, based on clinical seizure frequency and onset, electroencephalogram (EEG) abnormalities on a routine recording, and number of antiepileptic drugs used, with higher scores correlating with more severe seizures. 1 H-MRS examinations were acquired at a median of 6 days of age (range 1-13 days). Seizure severity was associated with increased lactate/choline in both the intervascular boundary zone (P Ͻ .001) and the basal nuclei (P ϭ .01) and with diminished N-acetylaspartate/choline in the intervascular border zone (P ϭ .034) (Fig. 2). Although this study shows that increased severity of seizures in neonates with perinatal asphyxia is independently associated with brain injury, it is limited in that seizures were only defined clinically rather than confirmed electrographically.

[6] 45w Although the normal neonatal brain appears relatively resistant to seizure-induced neuronal death, this resistance is much reduced with a preexisting brain insult. The underlying etiology for the seizures and presence of coexisting brain injury must be considered when determining how aggressively to treat neonatal seizures.

[1] 91w Neonatal seizures are a common manifestation of a variety of serious neurologic disorders in the immature brain. Although many infants will suffer sequelae from the underlying cause of the seizures, there is now clear evidence that seizures per se lead to further injury, particularly in a "metabolically compromised" brain. As such, aggressive management of neonatal seizures appears warranted once a true epileptic condition is confirmed. However, conventional therapies are frequently ineffective in providing seizure control. Further work is needed to identify more successful acute therapies, as well as potential neuroprotective strategies.

[1] 187w Animal studies have shown significant variation in the amount and type of seizure-induced brain injury depending on age, species studied, and specific methodology used to induce seizures. The most commonly studied animal model is the rat, an animal whose brain maturation at 7 to 10 days of age is thought to resemble that of the term human neonate. 13,14 Most studies have shown that recurrent or prolonged seizures induced in otherwise healthy brains have minimal, if any, observable structural damage in rats less than 14 days of age, 9,10,[15][16][17][18][19] despite causing neuronal degeneration and sclerosis in the hippocampus and dentate hilus in older animals. 20 A small number of studies have documented injury with lithium-pilocarpine-induced status epilepticus in rat pups. Sankar et al 21 showed evidence of apoptotic cell death in CA1 hippocampal neurons in 2-week old rat pups who had suffered lithium-pilocarpine-induced status epilepticus. The CA1 region is the last part of Ammon's horn to mature, and it does so postnatally in the human. 22 Additionally, Kubova et al 23 found evidence of necrosis in the mediodorsal thalamic nucleus after lithium-pilocarpine-induced status epilepticus in 12-day-old pups.

[2] 127w Seizure-induced neuronal injury has also been documented in neonatal rabbits. However, the rabbit's development is advanced to that of the newborn. 14 Franck and Schwartzkroin 24 documented neuronal death in CA1 and the adjacent subiculum after kainic acid-induced seizures in P7 rabbits. A recent article comparing neuropathology in immature rabbits, both at age P6 to 7 and P10 to 12, after either kainic acid or pilocarpine-induced seizures suggested that although the brain appeared resistant to kainic acid-induced seizure damage, it was not resistant to pilocarpine. 25 Pilocarpine resulted in mild, infrequent damage in the CA1 region of the hippocampus in the younger group but more severe injury mainly of the CA1 region, and also in the subiculum, CA3, neocortex, and amygdala regions in P10 to 12 rabbits.

[3] 41w In summary, most studies have shown that despite induction of recurrent and prolonged seizures, the immature brain appears significantly more resistant than its mature counterpart to seizure-induced neuronal death and exhibits much less long-term structural alterations as a result of seizures.

[4] 64w Despite causing little to no neuronal loss, seizure activity early in postnatal development does lead to an alteration in development of the hippocampal circuitry, resulting in learning and cognitive problems, [26][27][28][29] enhanced susceptibility to further seizures 21,[27][28][29][30][31] and increased risk of brain injury with seizures later in life. [32][33][34] These changes can be summed up by the adage "cells that fire together, wire together."

[5] 292w Several hypotheses have been advanced to explain how seizures induce these changes. In the developing brain, neurotransmitters and neuropeptide modulators (eg, substance P, somatostatin, neuropeptide Y) also serve a trophic function, regulating neuronal survival, neuronal migration, outgrowth of neurites, and synapse elimination. [35][36][37] Early seizures induce sprouting of mossy fibers in the CA3 pyramidal cell layer and molecular layer of the dentate gyrus, as well as neurogenesis of newly formed dentate granule neurons. 28,30 Other cellular alterations reported include a reduction in dendritic spine density 38 and delayed neuronal loss. 39 A number of studies have shown impaired learning in mature animals exposed to repetitive seizures early in development. Rats subjected to flurothyl seizures show impaired performance on the water maze and on auditory location and have decreased activity levels. [26][27][28] Neonatal seizures also lower seizure threshold later in life. Sankar found an increased risk of spontaneous seizures later in life in rat pups subjected to lithium-pilocarpine status epilepticus. 21 The risk of spontaneous seizures increased markedly with age, being seen in 3 of 11 pups with status epilepticus at 2 weeks, 8 of 11 at 3 weeks, and 6 of 8 at 4 weeks. Neonatal seizures also have been shown to reduce seizure threshold in adulthood to flurothyl inhalation 28 and pentylenetetrazol. 27 In animal models, neonatal seizures also increase susceptibility to brain injury from seizures later in life. [32][33][34] Koh et al 32 induced status epilepticus in P15 rat pups using kainic acid and found no overt neuronal injury. If these animals then had another kainic acid-induced seizure at P45, much more severe hippocampal injury was seen than in animals who received kainic acid only at P45. Similar results have been obtained using a flurothyl model of neonatal seizures. 33

[6] 74w Making a confident, clinical diagnosis of seizures in the neonate can, at times, be difficult (Table 3). Immaturity of myelination leads to slow and unusual patterns of seizure propagation. At this age, seizures are often disorganized, multifocal, or subtle, likely reflecting the inability of the immature brain to sustain organized epileptiform activity. 46 Furthermore, many neonates manifest seizure-like behavior without electrographic change or, conversely, may show electrographic seizure activity without any clinical manifestation. 47

[7] 206w In any neonate with seizures, immediate attention must by paid to ensure adequate ventilation and perfusion and exclude hypoglycemia. Although 1 older study documented 42 that phenobarbital is the initial antiepileptic medication of choice for most neurologists followed by phenytoin, 48 only 1 randomized study has compared these 2 agents in neonates. 49 In this single-blinded study, 59 neonates with EEGconfirmed seizures were randomized to receive either phenobarbital or phenytoin to minimum levels of 22.5 g/mL and 2.5 g/mL, respectively; seizure control was assessed by EEG. Response to treatment was equivalent between these 2 drugs, with 43% of the phenobarbital group and 45% of the phenytoin group achieving control. Most striking was that 41% of neonates remained uncontrolled even after both drugs, with the biggest predictor of successful treatment being severity of the seizures before treatment. Although experience with fosphenytoin is limited in the newborn, it possesses a number of benefits over phenytoin. First, it is soluble in glucose-containing solutions, and, secondly, it will not cause soft-tissue necrosis if intravenous administration invades the interstitial space. Although fosphenytoin will likely supplant phenytoin in this age group because of improved tolerance, it is unlikely to result in improved efficacy because fosphenytoin is simply a phosphate-ester prodrug of phenytoin.

[8] 307w Another study of 14 infants treated with high-dose phenobarbital showed that electrographic and electroclinical seizures were controlled in only 4 (29%), whereas the remainder showed a decrease in their electroclinical seizures but an increase in electrographic seizures. 50 A recent report by the same group showed that 11 of 22 (50%) neonates with seizures responded to phenobarbital (20-40 mg/kg) when used as a first-line agent. 51 Several studies have examined whether "prophylactic" high-dose barbiturates given shortly after birth in infants with severe asphyxia reduce seizures and improve long-term outcome but with conflicting results. In a randomized, controlled, prospective study, Hall et al 52 compared outborn, term, asphyxiated infants treated with phenobarbital (40 mg/ kg) shortly after birth (n ϭ 15) with those not receiving phenobarbital (n ϭ 16). Although this dose of phenobarbital was well tolerated, no significant difference in the incidence of seizures was seen. However, significantly more patients in the treatment group had a neurologically normal outcome at 3 years compared with controls (P ϭ .003). Although this study poses an interesting question, review of its methodol-ogy raises several concerns. First, the definition of "perinatal asphyxia" was arguable because cord gases were not considered. Rather, the diagnostic criteria consisted of an initial arterial pH Յ7 with a base deficit of Ն15 mEq/L, an Apgar score of Յ3 at 5 minutes, or failure to initiate spontaneous respirations by 10 minutes of age, and half of subjects qualified based on the last 2 criteria. This clinical picture can be seen in infants with causes other than asphyxia. Second, the presence of seizures was determined by clinical observation only, without electroencephalographic confirmation. In contrast, Goldberg et al 53 found that thiopental treatment in asphyxiated infants was associated with an increased need for blood pressure support and did not result in improved developmental outcome at 12 months of age.

[9] 167w Benzodiazepines have been tried in newborn infants. Diazepam and lorazepam can both be given as single doses but are not suitable for infusion given their long half-life. Lorazepam is probably preferred over diazepam because it has a smaller volume of distribution and is thus retained at high levels in the brain for longer periods. Diazepam may also elevate serum bilirubin levels in neonates. 54 Midazolam is shorter acting and has been proven safe when used for sedation in ventilated infants. Midazolam infusion has been used in children and adults with refractory seizures and also appears to be well tolerated in neonates although reported efficacy varies. In 1 small case series of 6 neonates with seizures refractory to high-dose phenobarbital Ϯ phenytoin, 4 (67%) achieved complete electrographic and clinical control and none experienced serious adverse effects. 55 However, Boylan et al 51 reported that no infant achieved seizure control with either midazolam (n ϭ 3) or clonazepam (n ϭ 3) when used as second-line agents in neonatal seizures.

[10] 178w Lidocaine treatment of refractory neonatal seizures is reported in a small number of studies and may prove to be more effective than benzodiazepines. In 1 study of 46 neonates with seizures refractory to phenobarbital Ϯ diazepam, seizure control was achieved in 83% of cases, as determined using a cerebral function monitor. 56 In a second study, Rey et al 57 achieved control in 11 of 13 infants with refractory seizures with lidocaine. In another small case series, 3 of 5 neonates with seizures refractory to phenobarbital responded to lidocaine. 51 However, lidocaine itself may lower seizure threshold, and there are rare reports of seizures being provoked even in the absence of high blood levels in neonates receiving this agent for elective circumcision or intubation. 58,59 Despite this potential risk, the encouraging results in neonates with refractory seizures suggest that lidocaine therapy deserves further study. There are scant data on the use of other antiepileptic agents in refractory neonatal seizures; however, cases treated successfully with lamotrigine, 60 zonisamide, 61 vigabatrin, 62 , and valproic acid 63 have been reported.

[11] 351w Pyridoxine Dependency/Folinic Acid-Responsive Seizures/Biotinidase Deficiency Pyridoxine dependency is a rare cause of seizures in the newborn [64][65][66] with an estimated annual incidence of 1 in 783,000 live births. 67 Infants are commonly encephalopathic and may develop systemic symptoms such as temperature instability, vomiting and abdominal distension, respiratory distress, and metabolic acidosis. 66,68 Some may also have low Apgar scores and/or abnormal cord gases, and Baxter noted that nearly one third of his neonatal cases presented with suspected hypoxicischemic encephalopathy. 67 Seizures have been reported in utero and are described as episodic, recurrent hammeringlike movements that begin as early as 20 weeks' gestation. 64 A recent French study suggests a characteristic electroclinical picture in this condition. 68 A neonate presenting with a large variety of different seizure types from the first day of life with associated hyperexcitability and unremitting agitation and/or a similarly affected sibling is indicative, and the EEG characteristically shows continuous, diffuse, high-voltage, rhythmic delta associated clinically with myoclonic jerks. 69 Although typically refractory to conventional antiepileptic drugs, seizures may respond promptly to a trial of pyridoxine, usually given as an intravenous dose of 100 mg. Failure to respond to intravenous pyridoxine, however, does not rule out the diagnosis, and an oral trial should still be given. Furthermore, electroclinical confirmation of a response is usually recommended. Seizures usually resolve within 1 to 2 weeks in patients who are truly pyridoxine dependent. Close monitoring and availability of resuscitation equipment is suggested with an intravenous trial because some infants have developed hypotonia and respiratory depression. 70,71 A trial of folinic acid should also be considered in unexplained, intractable neonatal seizures. In some cases, a transient response to phenobarbital may be seen, but breakthrough seizures ultimately recur, which respond promptly to folinic acid. High-performance liquid chromatography analysis of cerebrospinal fluid has shown a potential marker for this condition. 72 Although biotinidase deficiency usually presents at a later age, rare cases with neonatal onset are reported. 73 Associated features include hypotonia, skin rash, or stridor, and ketolactic acidosis and organic aciduria are usually found. Neonatal screening for this disorder is performed in many regions.

[12] 95w A number of physiological changes may affect drug levels in neonates and need to be considered in dose determination and interpretation of drug levels. [74][75][76] Oral absorption is generally reduced because of prolonged gastric emptying, relative achlorhydria, relatively low absorptive surface, and splanchnic blood flow. The proportion of the drug bound to protein is reduced because of low serum albumin and competition for binding from free fatty acids and unconjugated bilirubin. Immaturity of hepatic enzymes results in slower metabolism and renal excretion is lower. Therefore, antiepileptic drugs generally have longer half-lives in the neonatal period.

[13] 77w Although we presume that antiepileptic drugs have few side effects in the human neonate, several drugs have been shown to lead to apoptotic neurodegeneration in P7 rats at plasma concentrations relevant for seizure control in humans, possibly by reducing expression of neurotrophins and decreasing concentrations of survival-promoting proteins. 77 This effect was seen with many antiepileptic agents, including phenytoin, phenobarbital, diazepam, clonazepam, vigabatrin, and valproic acid. The relevance of this finding to human infants needs further study.

[14] 189w Most physicians initiate antiepileptic drug treatment in neonates exhibiting clinical seizure activity. More controversial is defining the "adequacy" of treatment. As previously mentioned, electroclinical dissociation is common in the neonate. 47 Should one aim to control all seizure-like behaviors, even those not associated with electrographic change? Should the goal of therapy be to eradicate all electrographic seizures, even if no clinical accompaniment is noted? At present, these questions remain difficult to answer with certainty. Neonatal seizures are clearly not benign; in animal models, such seizures result in impaired learning, [26][27][28][29] increased susceptibility to further seizures 21,[27][28][29][30][31] and enhanced brain damage with subsequent seizures later in life. [32][33][34] Although structural brain injury appears minimal in "healthy" brains, this is clearly not the case in "metabolically compromised" brains, with superimposed seizures exacerbating brain injury, as shown by both animal and human studies. 42,45 Two studies have documented that human neonates with greater numbers of brief electrographically confirmed seizures have poorer neurologic outcomes. 78,79 This work would support an aggressive approach to the management of both electroclinical and electrographic seizures, particularly in neonates with an underlying brain insult such as hypoxicischemic encephalopathy.

[15] 49w One rare syndrome in which an aggressive approach is probably futile is early myoclonic encephalopathy, which often presents in the neonatal period with myoclonus, partial seizures, and tonic spasms. The EEG shows a suppressionburst pattern. Although rare cases may respond to pyridoxine, most infants are refractory to antiepileptic medications.

[16] 39w Prognosis is dismal, and many affected infants die by age 1 to 2 years. In this syndrome, aggressive treatment with highdose barbiturate, benzodiazepine, or lidocaine to the point in which the child will require intubation is probably not justified.

[17] 73w However, our present conventional treatments are often inadequate with seizures persisting in a significant minority even after phenobarbital, phenytoin, and benzodiazepine treatment. [49][50][51]55 Furthermore, very high doses of barbiturates and benzodiazepine may result in cardiopulmonary depression and need for inotropic support and ventilation. If an aggressive approach is to be supported, we will need to look beyond phenobarbital, phenytoin, and benzodiazepines to find more effective treatments to stop seizures and potentially provide neuroprotection.

[18] 226w The potential role of neuroprotective strategies in neonates with seizures, particularly those caused by hypoxic-ischemic encephalopathy, needs further consideration. Because of evidence implicating glutamate excitotoxicity in the pathogenesis of seizure-induced damage in the immature brain, several investigators have examined the role of glutamate antagonists as neuroprotectants in animal studies with promising results. Koh et al 80 subjected P10 rat pups to 15 minutes of global hypoxia, which resulted in seizures in all animals studied. Animals were then given 2,3-dihydroxy-6-nitro-7sulfamovlbenzo (f) quinoxaline-2,3-dione (NBQX), topiramate (both AMPA antagonists), or placebo in 4 doses over the next 48 hours. Kainic acid was then administered at P21 and P28 to induce seizures, and animals were killed 48 to 72 hours later. These brains were compared with a group of control animals that were subject to kainic acid-induced seizures at the above ages but that did not sustain a preceding hypoxic insult or receive an AMPA antagonist. Animals previously rendered hypoxic on day 10 showed a significant increase in brain injury compared with control litter mates. However, both NBQX and topiramate markedly attenuated neuronal injury, such that animals treated with either of these agents showed similar degrees of injury to controls that were never hypoxic. This work suggests that AMPA activation and subsequent Ca 2ϩ influx plays a key role in epileptogenesis and susceptibility to neuronal injury in the developing brain.

[19] 167w Because excitatory amino acids are involved in learning and memory in the developing brain, there are potential concerns about long-term consequences of the use of excitatory amino acid antagonists. Tandon and coworkers 81 treated P10 rat pups with NBQX or MK-801 [N-methyl-D-aspartate (NMDA) antagonist], either as a direct infusion into the lateral ventricle or given intraperitoneally and performed a series of behavioral studies starting at P50. Although animals receiving either agent as a direct infusion into the lateral ventricle had a higher mortality than controls (and developed seizures), intraperitoneal administration of these agents was not associated with these complications and was well toler-ated. No deficits in learning, memory, or behavior were seen in animals previously treated with either NBQX or MK-801. Although these results are promising, it must be stressed that these agents were given for only a short time period within a narrow age window, and further work is needed in both normal animals and epilepsy models to investigate potential long-term consequences of administering these agents.