Treating the symptom or treating the disease in neonatal seizures: a systematic review of the literature

Pediatric patients develop seizures more frequently during the neonatal period than during any other age. The precise incidence of neonatal seizures can be difficult to define and depends on the population being studied and the criteria used for diagnosis. Neonatal seizures have been estimated to occur in up to3–5out of every 1000 births, and preterm newborns are estimated to develop seizures more frequently than full- term newborns, with an overall incidence of1 0–15 per1,000 preterm newborns, compared with 3–5 per 1000 full-term newborns and a prevalence of 22.2% among preterm newborns, compared with 0.5% among full term newborns [1, 2]. Moreover, because improved critical care has increased the survival rate following neonatal seizures, long-term neurological sequelae constitute a growing challenge for neonatologists. Greater than 50% of survivors, especially among preterm newborns, experience considerable disabilities across a range of developmental domains, with cerebral palsy and intellectual disability being the most frequently reported [2‐4]. Recent studies have shown a 17.6% over- all incidence in epilepsy among children with a history of neonatal seizures [5].

Neonatal seizures are often misdiagnosed, resulting in both the under- and overestimation of clinically- diagnosed seizure occurrence, due to electro-clinical dissociation phenomena and because neonatal seizures are often highly focal, with very little spread to other brain regions [6‐9]. Therefore, our review focused only on studies that described seizures confirmed by electroencephalography (EEG)- or amplitude-integrated EEG (aEEG).

In addition to making a correct diagnosis, making an early diagnosis of neonatal seizure is fundamental to the administration of proper treatment. Both animal and human studies have demonstrated that recurrent and prolonged seizures are harmful to the developing brain, emphasizing the importance of early seizure recognition and the availability of effective therapy options [10‐14].

One of the major challenges facing clinicians who treat neonates with seizures is the lack of effective antiepileptic drugs (AEDs).

Advances on this front have occurred during the last few decades; the anticonvulsant properties of therapeutic hypothermia, for example, have been demonstrated by both preclinical and clinical data. However, a proper, specific, effective, and safe pharmacological treatment for neonatal seizures remains lacking.

Currently, the World Health Organization (WHO) recommends the use of phenobarbital and phenytoin as first-line treatment [15] options for neonatal seizures, despite the low-quality evidence available to support their efficacy and the number of studies highlighting their potential side effects, which include increasing neuronal apoptosis and, consequently, contributing to long-term neurological damage and adverse neurocognitive outcomes [16, 17].

Here, we systematically review the available evidence for the treatment of electrographic and electroclinical neonatal seizures caused by specific neurologic disorders in newborns and evaluate the efficacy of both first-line and add-on anticonvulsants. Data on the populations studied, the seizure etiologies, treatment protocols, and study strengths and limitations were collected [18, 19].

For this systematic review, we searched the PubMed database using search terms related to neonatal seizures (see below). The search period was from August 1949 to November 2020 (last update 30/11/2020). The only filters applied were publication in the English language and human studies.

A further search of ClinicalTrials.​gov was conducted, and a list of ongoing clinical trials is provided.

In the 67 articles included in this review (4 RCT, 11 prospective studies, 27 retrospective studies and 25 case reports), HIE, stroke and genetic channelopathies were the most frequent etiologies of seizures. Despite the number of patients described in this review, performing statistical analyses of the data and providing precise descriptions for how the considered anticonvulsants work was challenging. In an attempt to standardize the results, we grouped all neonates with specific seizure etiologies as though they belonged to a single study. This decision was made because most studies analyzed small populations, which were too small for statistical analysis; however, this method provided us with the opportunity to analyze the whole dataset as 1 large cohort of patients. Overall, 556 patients whit HIE, 45 patients whit stroke and 76 patients whit genetic channelopathies were considered.

The therapeutic management of seizures in the newborns has remained unchanged for decades, despite almost 20 years evidence that commonly-used medications are not only ineffective but also potentially neurotoxic for newborns.

This systematic review aimed to collect all of the available data from existing studies published in the literature that have examined the currently available pharmacological treatments of electrically-confirmed neonatal seizures, describing the real-world effectiveness and side-effects associated with drug administration.

Our paper illustrates the limited available evidence regarding the best pharmacological treatments for neonatal seizures and serves as a reference for future studies.

International surveys among neonatologists, worldwide, have confirmed the historical trend toward the use of phenobarbital (in up to 70% of cases), as a first-line AED, and phenytoin (in up to 40% of cases), as a second-line AED, regardless of the seizure etiology or gestational age [90, 91].

However, several studies have demonstrated that phenobarbital may have potential long-term side-effects on neurodevelopment, which is not often considered when making treatment decisions [92‐96]. In addition, the overall efficacy of phenobarbital varies widely across reports, and in line with previous data from the literature, our review found that the overall efficacy of phenobarbital does not exceed 66% among all patients treated. Several preclinical studies have explored the poor efficacy of GABAergic drugs, such as phenobarbital and benzodiazepines, by demonstrating that inhibitory mechanisms are underdeveloped in the immature brain, in a manner that is directly proportional to gestational age [97‐99]. Animal studies in P7 mice, a post-natal age that grossly corresponds with 30–32 weeks of human gestational age, have confirmed that GABA receptors and the enzymes involved in GABA synthesis are expressed at low levels at birth and increase with time, during the first weeks of life [100]. In particular, the poor efficacy of phenobarbital may represent a developmental consequence of the persistence of the immature form of the sodium-potassium-chloride transporter, NKCC1, which may compromise the chloride-concentration gradient that is essential to phenobarbital’s mechanism of action [101]. Furthermore, GABA is known to act in an excitatory, rather than inhibitory [102], role during early stages of neurodevelopment, which may not only explain the ineffectiveness of GABA-enhancer drugs but also their potential roles during paradoxical seizure disruption.

Phenytoin and lidocaine appear to be potentially effective as second-line treatments for refractory seizures; however, to date, no strong evidence exists to recommend their use.

Phenytoin was shown to be effective in approximately 45% of patients during an RCT [23]. When added as a second-line treatment for seizures that were refractory to phenobarbital, phenytoin facilitated seizure control in an additional 10–15% of treated patients. Different studies have described higher risks of drug accumulation that reach toxic plasma concentrations when administered to preterm compared with full-term newborns. Because of its non-linear pharmacokinetic profile and hepatic metabolism, phenytoin administration also requires frequent blood-level monitoring, making it a slightly manageable medication.

Overall, the effectiveness of lidocaine ranged from 20 to 81% of patients treated [24, 34, 52, 53, 55, 56]; among patients with HIE, however, we observed that only milder phenotypes responded well to lidocaine, whereas a much lower effectiveness rate (30%) was reported for severely asphyxiated newborns [56]. Lidocaine, instead, appears to be more promising for the treatment of patients with stroke [52, 55], and among this population, functional studies demonstrated that lidocaine, in comparison with phenytoin, acted less strongly to suppress background activity and more strongly to suppressing electrical activity in specific ischemic areas of the brain. We observed that lidocaine administration in patients with stroke resulted in seizure control for 84% of patients treated, compared with much lower response rates for both midazolam and phenobarbital. In contrast, several other papers reported potential side-effects associated with lidocaine, including cardiac arrhythmias and hypotension. Therefore, lidocaine should not be used after phenytoin, due to the increased risk of cardio-depressive effects [103]. In addition, a seizure-inducing effect associated with high doses of lidocaine has been reported [104].

In the only RCT that compared midazolam and lidocaine for the treatment of neonatal seizures caused by various etiologies, a toward improved efficacy was observed for lidocaine, although both groups of patients had poor outcomes at 1 year of age [24]. Serious adverse side-effects, such as respiratory depression and sedation, have been reported and potential side-effects may also occur due to interactions between benzodiazepines and other pharmacological treatments. In addition, midazolam clearance correlates with gestational age, with reduced elimination observed among preterm infants, due to immature hepatic metabolism, which may result in a higher risk of side effects due to accumulation [105]. For these reasons, benzodiazepines should be considered second- or third-line treatments that are more suitable for already sedated and intubated newborns.

During the last few years, levetiracetam use has increased, due to the growing amount of literature regarding the safety and efficacy of both loading and maintenance doses and because several studies have reported that levetiracetam, in contrast with phenobarbital, is devoid of any pro-apoptotic properties that might affect the developing brain, even at exceptionally high doses [106, 107]. In addition, both intravenous and enteral preparations are available, making levetiracetam extremely manageable for clinical use. Although the exact mechanism of action for levetiracetam remains unknown, it has been hypothesized to target the synaptic vesicle glycoprotein 2A (SV2A). Talos et al. [108] estimated that neonatal neuronal SV2A protein levels reach 94% of adult values by 37 weeks post-conceptional age, suggesting that the target for levetiracetam may be abundantly expressed, even in the immature neonatal brain.

We have observed that the efficacy of levetiracetam varies across studies, ranging from 32 to 100% of treated patients, for both full-term and preterm newborns [27‐30, 38‐40, 42, 44‐47, 65]. Stratifying patients by etiology allowed us to observe that, to date, more data regarding the efficacy of levetiracetam are available for patients with seizures due to HIE than those due to other causes, and in this population, levetiracetam was effective, providing seizure freedom in up to 50% of patients after 40 h of treatment, when used as a first-line, monotherapy, and in up to 92% of patients in a longer-term follow-up, as both a first- and second-line anticonvulsant [46, 47]. In a population of 44 asphyxiated newborns, initial treatment with levetiracetam predicted a shorter interval to seizure freedom than treatment with phenobarbital in univariate analysis, even after adjusting for initial seizure frequency and unbiased HIE severity score [47]. In addition, comparison between levetiracetam and phenobarbital for the treatment of neonatal seizures caused by various etiologies showed a short-term better effect of levetiracetam on tone and posture of patients according to HNNE score [26]. Treatment doses ranged from 10 to 60 mg/kg for the loading dose, and from 10 to 80 mg/kg for the maintenance dose. No serious adverse events were reported associated with levetiracetam administration, except for mild somnolence and feeding difficulty, which were resolved by dose-adjustment. Several studies focused on the safe and predictable pharmacokinetic profile of levetiracetam, even in preterm and extremely sick full-term newborns, emphasizing that because levetiracetam does not require hepatic metabolism, it rarely interferes with other treatments [109]. Considering its safety profile and higher distribution volume in newborns (0.89 compared with 0.6–0.7 L/kg in children), we recommend the use of higher doses (30–60 mg/kg for the loading dose and 30–50 mg/kg/day, divided into 2–3 doses for maintenance, eventually titrated up to 80 mg/kg/day) [110].

Several papers have described the efficacy of sodium-channel blockers for the treatment of genetic channelopathies. Phenytoin, lidocaine, carbamazepine, and oxcarbazepine act to block the movement of sodium ions through ion channels during the propagation of action potentials to prevent seizure activity. Due to structural similarities, sodium channel blockers also act on potassium channels, resulting in seizure control in patients with genetic epilepsies due to KCNQ mutations. The modulation of one type of channel has also been hypothesized to affect the functions of the entire channel complex.

In agreement with the literature, KCNQ2 mutations represented the most common genetic anomaly identified in our review. We observed a good response to treatment using sodium channel blockers in patients with these mutations, with an overall 63% efficacy. A better response was observed for carbamazepine (77% among responders to treatment), which was also the most commonly used medication because it has few to no reported side-effects and an oral, extremely manageable formulation is available [61, 62, 72‐74]. A few patients were treated with lidocaine or phenytoin, who responded to drug administration, were later dismissed on oral carbamazepine for the maintenance of seizure freedom [61, 71, 75, 76].

Interestingly, based on clinical features, familial anamnesis, and EEG patterns, 4 patients were treated early with oral carbamazepine as a first-line anticonvulsant and responded with seizure cessation within hours after the initial first administration [62]. We also observed that the patients who did not respond to carbamazepine were those who displayed features of severe KCNQ2 encephalopathy. Some of these patients responded to combinations of medications that included sodium channel blockers.

Less is known about other genetic encephalopathies, such as the KCNT1-related epilepsy of infancy with migrating focal seizures. Quinidine may effectively block the pathogenic constitutive activation of the KCNT1 channel at the molecular level, but no data regarding its administration for the neonatal population are available, to date [111].

When a genetic channelopathy is suspected, based on clinical features, familial anamnesis, and EEG patterns, the response to treatment with sodium-channel blockers not only represents the best treatment option available but may also be an ex juvantibus criteria to obtain a diagnosis when waiting for genetic test results, encouraging their early use and administration.

A change to the current “one size fits all” treatment model, in which treatment protocols do not account for etiology as a factor, is necessary to accommodate the possibility of customized, patient-specific, precision medicine.

Unfortunately, current data do not yet allow current treatment protocols to be replaced because the populations described, worldwide, are too heterogeneous, both in terms of etiology and treatment. The evaluation of each drug’s efficacy for the treatment of specific etiologies is difficult when the populations described include both preterm and full-term newborns with seizures caused by a variety of etiologies.

Current knowledge, however, allows us to highlight the good clinical and electrographic responses of genetic early-onset epilepsies to sodium channel blockers and the overall good response to levetiracetam, whose administration has also been demonstrated to be safe in both full-term and preterm newborns.

Future investigations should identify methods to better identify and distinguish, as early as possible, between acute seizures and neonatal-onset epilepsies, to facilitate patient-specific, minimally dangerous treatment options, which will offer newborns, especially preterm newborns, higher survival rates, better neurological outcomes, and a better long-term quality of life.

After more than 20 years of experience, limited evidence exists regarding the best pharmacologic treatments for neonatal seizures. Treatment, too often, remains guided by experience, because few RCTs have been performed and the data available from those that have been performed have not been significant.

Additional controlled trials and large prospective studies are urgently necessary to determine the correct drug choices, dosing regimens, and treatment durations for newborns that will result in better futures, in terms of both seizure freedom and neurocognitive outcome.

This systematic review of neonatal seizure treatment underlines the pitfalls in current neonatology practice and serves as a reference to guide future investigations.