Etiology of neonatal seizures and maintenance therapy use: a 10-year retrospective study at Toulouse Children’s hospital

Seizures are the most frequent neurological symptom during the neonatal period [1]. The neonatal brain is characterized by a high level of synaptogenesis and neuronal plasticity that explains a physiological hyperexcitability, and thus a vulnerability to seizure [2, 3]. The occurrence is between 1 and 3 per 1000 term newborns [4‐7].

Etiologies are mainly symptomatic. They are divided into: vascular (hypoxic-ischemic encephalopathy (HIE), ischemic infarction, intracranial hemorrhage); infectious (intracranial infections); metabolic (metabolic or electrolyte disorder, inborn error of metabolism), and malformation (congenital malformations of the central nervous system). The incidence of epileptic syndromes is less than 10% [8]. On the one hand, benign familial neonatal convulsions (BFNC) and benign idiopathic neonatal convulsions are associated with a favorable outcome. On the other hand, early myoclonic encephalopathy and early infantile epileptic encephalopathy have poor prognoses. The risk of developing epilepsy is about 17.9% according to a meta-analysis [9]. Etiology and electroencephalographic profile are important risk factors. However, designing a universal scoring system capable of providing early prognostic information on epilepsy development seems difficult because of the uncertainty related to etiology and gestational age [10].

The World Health Organization recommends treatment of a crisis lasting more than 3 min or repeated clinical or subclinical crises [11]. There is no strong recommendation about maintenance therapy: indication criteria, drug, or duration. The administration of maintenance therapy should be reserved for newborns at risk of seizure recurrence. This risk is less than 10% when seizure control is achieved and both the neurological examination and electroencephalogram (EEG) are normal. Phenobarbital is the acute treatment of choice in neonatal seizures [12, 13]. However, phenobarbital has been found to increase neuronal apoptosis in newborn rats and to have cognitive side effects in infants [13, 14]. This leads to the prescription of a different antiepileptic drug in maintenance therapy [15]. The duration should be as short as possible [16]. The criteria for discontinuing treatment should be both clinical and electroencephalographic [17].

The aim of this study was twofold: to investigate the incidence of etiologies of neonatal seizures in full-term neonates at the University Children’s Hospital of Toulouse, France and to report our practices concerning the maintenance antiepileptic therapy used.

This study reports the experience of our center concerning the long-term drug management of neonatal seizures depending on etiologies.

The main etiology of full-term neonates admitted to our center because of neonatal seizures was HIE (37%). Compared to a recent American prospective cohort of 426 term and preterm neonates based on EEG diagnosis, the distribution of the various etiologies is similar [18]. In our population, the incidence of each etiology was: 37% (versus 38% in the Glass cohort study) hypoxic-ischemic encephalopathy, 12% (11%) ischemic infarction, 15% (18%) intracranial hemorrhage, 8% (4%) intracranial infections, 3% (4%) metabolic or electrolyte disorder, 2% (3%) inborn errors of metabolism, 5% (4%) congenital malformations of the central nervous system, and 11% (6%) epileptic syndromes. The etiology was unknown in 16 patients. The etiology was not always found in the neonatal period and 7% (9%) of our cohort did not have a diagnosis at the end of the following period.

In this study, the overall incidence of seizures was 0.9 per 1000 full-term live births per year in the Midi-Pyrénées region. In other studies, this incidence was between 1 and 3 per 1000 live births [4‐7]. The diagnosis of seizures was based mainly on clinical observation, with a low rate (48%) of electroencephalographic confirmation. Consequently, the incidence of neonatal seizures in our population could have been underestimated. [19] This study highlights the need for our center to improve the use of prolonged EEG video to better diagnose neonatal seizures, especially in newborns at risk.

During neonatal admission, 18.5% of patients died. This is in line with previous publications of Mastrangelo et al. [20] (21%) and Glass et al. (17%) [12]. In our cohort, 15% developed epilepsy. A meta-analysis reported a similar rate of 17.9% in the literature. Etiologies with a poor prognosis were HIE, congenital malformations of the central nervous system, inborn error of metabolism and epileptic syndromes.

Regarding the acute treatment of neonatal seizures, we note that diazepam administered as a first-line treatment usually shows low efficacy. Despite its limited efficacy, diazepam is still used because of the ease of its intra-rectal administration in newborns that still do not have intravenous access. Administration of diazepam should be avoided because it leads to polymedication that will increase the risk of side effects in the neonate and may delay the establishment of an emergency venous route for the administration of phenobarbital.

A maintenance therapy was prescribed for 180 newborns (72%). In a previous study, Bartha et al. [21] reported a rate of 75% of newborns discharged with an antiepileptic therapy. The type of drug used, and duration were not reported in this study. Factors that determined the use of a maintenance therapy were abnormal EEG, neuro-imaging and second-line or further acute antiepileptic treatment in Bartha’s study.

Valproic acid was commonly used in our cohort whatever the etiology of seizures. While other centers use phenobarbital in maintenance treatment, the choice to use other molecules (valproic acid, carbamazepine ...) has been done in our center to avoid neurodevelopmentally related adverse effects related to long-term use of phenobarbital [15]. However, the use of valproic acid does not seem completely safe since serious adverse effects such as hyperammonemic encephalopathy are reported in neonates free of any metabolic disease, apart from overdose [22]. Since 2010, we have used new antiepileptic drugs such as carbamazepine and levetiracetam. Carbamazepine [23, 24] has been used in partial seizures due to stroke, infection, or malformation. Levetiracetam [25‐30] has been used in other types of seizures, especially when valproic acid [31‐34] was contraindicated, i.e. when liver enzymes were disturbed or when a metabolic disease was suspected. Although, levetiracetam is more and more used, we recall that there is still no marketing authorization for this drug in newborns. Phenobarbital has not been used since 2009 because of its potential cognitive side effects in infants and newborns and the difficulty of finding the correct dose (seizure-free without drowsiness). Vigabatrin has rarely been used in our practice because of the incidence of visual side effects [35‐39]. Because of those ophthalmologic side effects, it was only indicated when seizures were difficult to control. [40, 41] Clonazepam was used in similar indications. The small numbers of patients receiving levetiracetam, carbamazepine, vigabatrin, phenobarbital or clonazepam does not allow us to make statistical comparisons about the efficacy and the tolerance of different anti-epileptic treatments, especially because the etiologies of neonatal seizures are different in this group and it is the most important prognostic factor. The subgroup analysis comparing patients treated with levetiracetam, which is now widely used, with a group of patients treated with valproic acid, did not show any difference in the outcome (death and epilepsy) of the patient, maybe because of low effectives. However, it is interesting to note that the number of anticonvulsants used urgently in the treatment of newborns for whom long-term treatment with levetiracetam is set up is significantly lower. This is all the more important since it is known that the risk of adverse effects increases when several treatments are combined.

Recent studies support a targeted therapeutic approach for genetic epileptic encephalopathies based on the molecular dysfunction. For example, KCNQ2 and SCN2A genes are involved in neonatal epilepsy. Their mutations result in sodium and potassium channel dysfunctions. Carbamazepine, is a sodium channel blocker, that also modulates potassium channels, co-localized at the neuronal membrane. Low dose of carbamazepine are effective in this indication and it could be considered as first-line treatment [42‐44]. A better understanding of the physiopathology of neonatal epilepsies will help us to determine the more effective antiepileptic drugs to use.

In our cohort, the treatment was discontinued after a control EEG, on the average after 5 months of treatment. The duration depended on the severity of the initial profile, seizure recurrence, etiology and the presence of paroxysmal elements in the electroencephalogram. French guidelines published in 2016 about the treatment of neonatal cerebral arterial infarction, recommend to stop antiepileptic drug 72 h after the last seizure if the clinical examination and EEG are normal or at discharge if there are moderate abnormalities [45]. Achieving earlier discontinuance of maintenance therapy seems to be a major challenge in our practices to limit side effects of antiepileptic drugs.

This study has some biases. The lack of systematic confirmation of seizures at EEG, due to a low use of aEEG, may have led to the inclusion of patients with abnormal non-epileptic movements. The support in different services with the absence of a common protocol of care leads to a great variability of practices. Finally, the retrospective nature induces some missing data and the size of our cohort did not allow us to evaluate the efficacy and tolerance of the different treatments. However, this study allows us to improve the management of neonatal seizures in our center, with the increase of the use of prolonged video EEG, the cessation of diazepam use in the acute treatment of seizures, and the limitation of the duration of the maintenance therapy. We report pratices that have changed during the study period and since the end of data collection and that can’t be used as current practices now.

In conclusion, despite the lack of systematic electroencephalographic diagnoses, we report a neonatal cohort of full-term newborns comparable with other studies. Although neonatal seizures are often occasional events and the risk of developing epilepsy is about 15%, depending on the etiology, we frequently used a maintenance antiepileptic treatment. No current guidelines allow us to determine the best choice of drug to use in this indication. In our practice, valproic acid was the most commonly prescribed when liver function is normal and metabolic disease excluded. However, the arrival on the market of new antiepileptic drug, and a better understanding of the physiopathology of genetic encephalopathies is changing our practice. Additional studies are necessary to establish recommendations concerning the long-term management of neonatal seizures according to their etiology.