Epilepsy in pregnancy

It is estimated that, globally, approximately 15 million women with epilepsy are of childbearing age. Most of these women are in need of effective and safe treatment for their epilepsy also during pregnancy. For them, as well as for their partners, the possible risks to the unborn induced by use of antiepileptic drugs (AEDs) during pregnancy is a major concern. While it is important to understand that the vast majority of women with epilepsy can have uneventful pregnancies and give birth to perfectly healthy children, there are also fetal risks associated with treatment. These risks include negative effects on fetal growth, increased risks of major birth defects, as well as adverse effects on neurocognitive and behavioral development. However, uncontrolled maternal seizures can also be harmful not only to the pregnant woman but also to the fetus. The aim of this article is to present the facts on these risks and, most importantly, based on this, to provide practical recommendations on how risks can be balanced and reduced and suggest how epilepsy may be managed before and during pregnancy and after delivery in order to optimize outcomes. The recommendations are based on review of the relevant literature carried out by the International League Against Epilepsy (ILAE) Task Force on Women and Pregnancy. The search strategy and criteria for selection are summarized in Box 1. Emphasis was focused on publications in the last 10 years, i.e. after the publication of the guidelines from the American Epilepsy Society. Recommendations should be considered as an expert opinion rather than evidence-based guidelines. This approach was taken due to limitations in the present evidence base, which leave clinicians and their patients without adequate information. This report is intended to fill this gap, at least in part.

The challenge in the management of epilepsy during pregnancy is to balance the foetal and maternal risks associated with seizures against the teratogenic risks associated with exposure to AEDs in utero. Addressing issues related to pregnancy should begin well before conception in order to maximize pregnancy outcomes. It is clear that AEDs differ in their teratogenic potential. Valproate is associated with the greatest risk of malformations as well as adverse cognitive and behavioural outcomes, and should, whenever possible, be avoided for the treatment of patients who may become pregnant. Lamotrigine and levetiracetam are associated with the lowest risk of malformations, but data on neurodevelopment for levetiracetam is based on a small sample and evidence on the effects of prenatal exposure on neurodevelopment is lacking or insufficient for other newer-generation AEDs. Teratogenic risks should be considered at the time of initiation of AED treatment in young women. Pre-pregnancy counselling is essential to ensure that the woman enters pregnancy on the most appropriate AED treatment, taking efficacy as well as foetal safety into account, and that she is also taking folate supplementation. Once pregnancy is established, close monitoring is warranted and ideally in collaboration between epilepsy and obstetric care.

It should finally be emphasized that the vast majority of women with epilepsy will have uneventful pregnancies and give birth to healthy children. The aim of the recommendations in this review is to further facilitate such positive pregnancy outcomes.

Risks of seizures

The risk of seizures during pregnancy and the consequences that they might have on the developing foetus as well as the mother are fundamental reasons for the use of AEDs, and this should be no different for epileptic women with childbearing potential as they may become pregnant (Sveberg et al., 2015).

Focal seizures that do not evolve to bilateral tonic-clonic seizures are unlikely to have a major impact on the foetus, although there are a few case reports indicating brief foetal distress, expressed as deceleration of foetal heart rate for 2.5-3.5 minutes, during focal seizures with impaired awareness (Nei et al., 1998; Sahoo and Klein, 2005). Generalized tonic-clonic seizures (GTCS), including focal to bilateral tonic-clonic seizures according to the current classification (Fisher et al., 2017), are associated with hypoxia and lactic acidosis, which during pregnancy are transferred to the foetus through the placenta and may lead to asphyxia (Hiilesmaa and Teramo, 2013). Seizure-related falls can also cause blunt trauma to the uterus and thus affect the foetus.

A nation-wide register-based study from Taiwan found an association between the occurrence of seizures of all types during pregnancy and small foetal size for gestational age, moreover, seizures during pregnancy were more likely to be associated with preterm delivery and lower birth weight (Chen et al., 2009). In this study, occurrence of seizures was defined as those in women being hospitalized or treated in the emergency department for epilepsy during pregnancy, and no distinction was made between different seizure types. A smaller hospital-based retrospective study from Norway found no differences in obstetrical complications between women with seizures during the last five years, based on medical records (type of seizures not specified), and those without (Borthen et al., 2011). However, differences in methodology make it difficult to compare between the results of these two studies.

There are no indications of an association between maternal seizures and risk of MCM, but one retrospective study has indicated lower verbal IQ in children of mothers with five or more GTCS during pregnancy (Adab et al., 2004). The retrospective study design, however, precludes definitive conclusions, and this observation has not been confirmed in prospective studies (Meador et al., 2013; Baker et al., 2015), although these were not designed to address the question of possible impact of maternal GTCS during pregnancy on the child's neurodevelopment.

Epilepsy and uncontrolled seizures are also associated with maternal risks. Up to a 10-fold increased risk of maternal mortality (Edey et al., 2014) or mortality during hospitalised delivery (MacDonald et al., 2015) has been reported for women with epilepsy. When the causes of death were analysed, the majority were seizure related and most due to sudden unexpected death in epilepsy (SUDEP) (Edey et al., 2014).

Epilepsy is not considered as a reason for Caesarean delivery, unless a seizure occurs during labour, making the patient unable to cooperate (Donaldson, 2002).

Generalized tonic-clonic seizures are associated with risks to the foetus as well as to the pregnant woman. Other seizures are probably less harmful, but may be associated with injury, intrauterine growth retardation and premature delivery.

Risks of medications

Intrauterine growth

The possibility that AED use during pregnancy is associated with offspring that are small for gestational age (SGA) and have small head circumference has been discussed for many years, mainly based on outcomes of selected cohorts of pregnant women with epilepsy. Particular interest has been paid to the risk of microcephaly as this could be associated with functional deficits. SGA newborns are at risk of stillbirth and other long-term sequelae. Earlier hospital-based cohort studies (Battino et al., 1999; Hvas et al., 2000) as well as population-based register studies (Wide et al., 2000a; Almgren et al., 2009; Veiby et al., 2009) have indicated an increased risk of small head circumference among children exposed to AED polytherapy or monotherapy with primidone, phenobarbital (Hiilesmaa et al., 1981; Battino et al., 1999), carbamazepine (Hiilesmaa et al., 1981; Wide et al., 2000a; Almgren et al., 2009; Veiby et al., 2009) or valproate (Wide et al., 2000a). Although these studies report smaller head circumference among those exposed, they have not shown increased rates of microcephaly. The prospective NEAD study found increased rates of microcephaly (defined as <3rd percentile) with carbamazepine and valproate, but normalization by two years of age (Pennell et al., 2012).

For obvious reasons, the impact of the newer-generation AEDs was not assessed in the earlier studies. However, a more recent population-based study from Norway reported that, in general, children exposed to AEDs had a moderate risk of growth restriction. However, compared to children of healthy mothers, those exposed to topiramate had a considerable risk of microcephaly (11.4% vs. 2.4 %; odds ratio [OR]: 4.8) and SGA (24.4 % vs. 8.9 %; OR: 3.1) (Veiby et al., 2014). Increased risk of microcephaly (defined as head circumference less than the 2.5th percentile), although to a lesser extent, was also reported to be associated with carbamazepine (OR: 2.0) and AED polytherapy (OR: 2.0). The same study found a significantly increased risk for SGA with carbamazepine (OR: 1.4) and AED polytherapy (OR: 1.6). A Danish population-based study confirmed a more than two-fold increased risk of SGA with topiramate exposure, and a less, but significantly increased, risk associated with valproate and carbamazepine (Kilic et al., 2014). Based on a recent population-based Swedish register study, relative to lamotrigine exposure, infants exposed to carbamazepine and valproic acid had head circumference that was smaller by 0.2 SDs, those exposed to levetiracetam had smaller head circumference by 0.1 SD, while pregabalin-exposed infants had the same head circumference as those exposed to lamotrigine (Margulis et al., 2019). Other AEDs were not included in the report. Data from the North American Antiepileptic Drug Pregnancy Registry (NAAPR), based on a selected cohort of women with epilepsy, indicated that the prevalence of SGA was increased in infants exposed to AEDs compared with unexposed infants (relative risk [RR]: 2.0, 95% CI), and that the prevalence of SGA was particularly high (18.5%) for topiramate, but also increased with exposure to phenobarbital or zonisamide (Hernandez-Diaz et al., 2017).

An association between exposure to an AED and intrauterine growth restriction does not necessarily mean that there is a direct causal relationship. For example, an early study from Finland demonstrated a genetic influence, indicating that head circumference was smaller among fathers of the group of AED-exposed children with small head circumference (Gaily et al., 1990).

The most consistent finding in these studies is the growth restriction associated with maternal use of topiramate. While the potential functional consequences of this remain to be assessed, the findings call for some caution in the use of topiramate during pregnancy when reasonable treatment alternatives are available.

Treatment with certain AEDs is associated with an increased risk of intrauterine growth restriction.

The effect on growth varies between different AEDs and appears to be most pronounced with topiramate.

Major birth defects

The first reports of an association between AEDs and congenital abnormalities were published more than 50 years ago (Meadow, 1968). Research over the years since 1968 has revealed that AEDs differ in their potential to cause MCMs. Two recent systematic reviews agree that, for monotherapy, the highest risk is associated with valproate and the lowest with lamotrigine and levetiracetam exposure (Weston et al., 2016; Veroniki et al., 2017). Based on a network meta-analysis (Veroniki et al., 2017), the OR for MCM, relative to unexposed patients, was 2.93 for valproate (95% CI: 2.36-3.69), 1.90 for topiramate (95% CI: 1.17-2.97), 1.83 for phenobarbital (95% CI: 1.35-2.47), 1.67 for phenytoin (95% CI: 1.30-2.17), and 1.37 for carbamazepine (95% CI: 1.10-1.71), while no increase was noted with lamotrigine (OR: 0.96; 95% CI: 0.72-1.25) or levetiracetam (OR: 0.72; 95% CI: 0.43-1.16). A Cochrane review (Weston et al., 2016) reported similar results. Compared with offspring of women without epilepsy, the RR for valproate was 5.69 (95% CI: 3.33-9.73), topiramate was 3.69 (95% CI: 1.36-10.07), phenobarbital was 2.84 (95% CI: 1.57-5.13), phenytoin was 2.38 (95% CI: 1.12-5.03), and carbamazepine was 2.01 (95% CI: 1.20-3.36). There was no increased risk of MCMs with lamotrigine (Weston et al., 2016). Gabapentin, levetiracetam, oxcarbazepine, primidone or zonisamide were not associated with an increased risk, however, there were substantially fewer data for these medications (Weston et al., 2016). While meta-analysis is a method to increase statistical power, an important trade-off is the inclusion of data from heterogeneous studies, with different populations and even different criteria for MCMs (Veroniki et al., 2017), both of which can significantly impact the results (Tomson et al., 2010). A comparison between treatments should therefore ideally be made within individual studies. In this regard, prospective AED and pregnancy registries that have documented pregnancies over 20 years have provided very useful data (Hunt et al., 2008; Hernandez-Diaz et al., 2012; Mawhinney et al., 2013; Campbell et al., 2014; Tomson et al., 2018). The prevalence of MCMs for the most frequently used AEDs as monotherapy observed in the three major registries, the NAAPR, UK and Ireland Register, and EURAP, are summarized in table 1. All three registries confirm the greater risk with valproate and the comparatively low risk with lamotrigine and levetiracetam. Based on the individual pregnancy registries, the risk of MCM was analysed relative to dose. All registries revealed increasing risks with higher doses of valproate, with cut-offs for higher risks ranging from 500 mg/day based on the NAAPR, to 600 mg/day and 650 mg/day based on the UK-Ireland and EURAP registries, respectively. Based on the EURAP, a dose-dependent effect was also identified for carbamazepine, lamotrigine and phenobarbital, whereas the UK-Ireland registry confirmed dose-dependency for carbamazepine (Hernandez-Diaz et al., 2012; Campbell et al., 2014; Tomson et al., 2018). Based on the EURAP, the lowest risk was associated with lamotrigine at ≤325 mg/day at conception. In comparison, the prevalence of MCMs was significantly higher at all doses of carbamazepine and valproate. Valproate at doses as low as ≤650 mg/day was also associated with an increased risk compared with levetiracetam (OR: 2.43; 95% CI: 1.30-4.55) (Tomson et al., 2018).

Polytherapy has traditionally been considered to be associated with a higher risk of MCMs than monotherapy (Veroniki et al., 2017). However, more recent studies indicate that the type of AED included as polytherapy is more important than the number of AEDs. When data from different registries on specific drug combinations were compared, it was clear that inclusion of valproate as polytherapy was the main reason for higher prevalence of MCMs (Morrow et al., 2006; Holmes et al., 2011). A recent report from the prospective Kerala Registry of Epilepsy and Pregnancy indicates that the excess MCM risk associated with combination therapy with two AEDs (duotherapy) was largely dependent on the inclusion of valproate or topiramate, in combination with one other AED (Keni et al., 2018). Based on the prospective EURAP, a comparison of MCM rates relative to valproate monotherapy and duotherapy at different dose levels of valproate revealed that a low dose of valproate given in combination with another AED was associated with lower MCM rates than a higher dose of valproate as monotherapy (Tomson et al., 2015).

AEDs also differ according to the type of MCM they are associated with. A pooled analysis of 32 prospective cohort studies revealed a particularly high prevalence of cardiac malformations with barbiturates, and high prevalence of neural tube defects and hypospadia with valproate (Tomson et al., 2016), which was confirmed in two recent meta-analyses (Weston et al., 2016; Veroniki et al., 2017). Topiramate monotherapy has been associated with increased risk of cleft lip/palate (de Jong et al., 2016; Veroniki et al., 2017, Blotière et al., 2019), a risk that appears to be dose-dependent (Hernandez-Diaz et al., 2018). Based on a recent nationwide cohort study utilizing French health care databases, valproate was associated with eight different MCMs, whereas no significant association with any specific MCM was identified for lamotrigine, levetiracetam, carbamazepine, oxcarbazepine, or gabapentin (Blotière et al., 2019).

The risk of MCM is also affected by other factors. Based on the EURAP, parental history of MCMs was associated with a three-fold increase in risk (Tomson et al., 2018). Earlier studies indicate a high risk of MCMs in the offspring from subsequent pregnancies in women who had previously given birth to a child with MCM, suggesting individual susceptibility or a genetic predisposition (Campbell et al., 2013; Vajda et al., 2013).

Based on the EURAP, changes in AED prescription were recently reported during pregnancy over a 14-year period (Tomson et al., 2019). In parallel with decreased use of valproate and carbamazepine and increased prescription of lamotrigine and levetiracetam, the prevalence of MCMs declined by 27% over the time period.

Valproate is associated with the highest risk of MCMs, phenobarbital and topiramate with an intermediate risk, and lamotrigine and levetiracetam with the lowest risk.

The risk of MCMs is dose-dependent for valproate and probably also for other AEDs such as carbamazepine, phenobarbital and lamotrigine

For polytherapy, the type of AED included is at least as important as the number of AEDs.

Developmental and behavioral outcomes

Both animal and human studies demonstrate that prenatal exposure to valproate adversely affects foetal brain development. Children exposed to valproate in utero are at an increased risk of poorer development in infancy (Bromley et al., 2010; Veiby et al., 2013a) and reduced IQ and other cognitive functions in preschool years (Meador et al., 2009; Cohen et al., 2011; Meador et al., 2011). At school age, IQ is reduced by 7-11 IQ points with abilities falling below average range in 20-40% of valproate-exposed children (Nadebaum et al., 2011; Meador et al., 2013; Bromley et al., 2014; Baker et al., 2015). Other key school-aged cognitive skills, such as memory, attention and language skills, have also been found to be poorer in comparison to control children and children exposed to other AEDs (Nadebaum et al., 2011; Meador et al., 2013; Baker et al., 2015; Bromley et al., 2016). Not surprisingly, such cognitive deficits have been reported to lead to increased rates of educational intervention, with reported levels of need for 29-48% of exposed children across studies for valproate monotherapy (Adab et al., 2001, 2004; Baker et al., 2015), with a dose-dependent influence (Baker et al., 2015). For children with clinically diagnosed physical features of valproate embryopathy, the need for educational support is much higher, at 74% (Moore et al., 2000; Bromley et al., 2019). Recent data from the Danish National Birth Cohort has demonstrated poorer educational examination outcomes in late primary and early secondary school (Elkjaer et al., 2018) and an increase in learning disabilities (Bech et al., 2018), highlighting the longer term and likely permanent valproate effects on cognition. Additionally, prenatal exposure to valproate is associated with an increased risk of autistic spectrum disorder diagnoses (Bromley et al., 2013; Christensen et al., 2013), attention deficit hyperactivity disorder (Adab et al., 2004; Cohen et al., 2011; Bromley et al., 2013; Christensen et al., 2019), and other parent-rated behavioural problems (Huber-Mollema et al, 2019). These risks are clearly dose related although no dose of valproate has been proven to be devoid of neurodevelopmental risks, and doses of Adab et al., 2004; Meador et al., 2009, 2013; Bromley et al., 2014; Baker et al., 2015).

Investigations of foetal carbamazepine exposure and early infant development have highlighted that development levels are comparable overall to expected trajectories for early infant development and school-aged IQ (Bromley et al., 2014). School-age IQ is higher in children with foetal carbamazepine exposure relative to foetal valproate exposure (Adab et al., 2004; Gaily et al., 2004; Meador et al., 2011, 2013), and is comparable to foetal lamotrigine exposure (Meador et al., 2011, 2013) and that of control children (Rihtman et al., 2013; Baker et al., 2015). However, there may be risks associated with reduced verbal reasoning skills (Meador et al., 2014; Baker et al., 2015). Children with foetal carbamazepine exposure were worse at mathematics in sixth grade compared with unexposed controls, but the actual difference was small (Elkjaer et al., 2018). Lower grades are consistent with an earlier study in which a smaller number of children obtaining high-level passes was documented (Forsberg et al., 2011), however, the mechanism through which such an impact on education occurs remains unclear. An earlier association with an increased risk of autistic spectrum disorder among carbamazepine-exposed individuals (Rasalam et al., 2005) has not been replicated by a large national cohort (Christensen et al., 2013) or observational study (Bromley et al., 2013), or based on parental ratings of symptoms of autistic behaviours (Huber-Mollema et al., 2019). Increased behavioural problems have recently been reported in middle childhood (Huber-Mollema et al, 2019), however, only 14% of carbamazepine-exposed children reached this clinical threshold which was based on parental ratings.

Children exposed to phenytoin in utero do not appear to have poorer early developmental skills in infancy (Wide et al., 2000b; Thomas et al., 2008; Bromley et al., 2014), although this is in contrast to one early study (Scolnik et al., 1994). School-aged cohorts exhibit expected IQ levels (Gaily et al., 1988; Meador et al., 2009), but the data available are limited due to the cohort size exposed to monotherapy. Children exposed in utero to phenytoin are reported to have higher IQ than those exposed to valproate and have comparable IQ to groups exposed to carbamazepine or lamotrigine (Meador et al., 2013; Bromley et al., 2014). In terms of specific developmental outcomes, one study highlighted poorer language skills (Rovet et al., 1995) whilst another noted poorer motor development (Wide et al., 2002), however, based on the more recent NEAD study, poorer levels of language or motor functioning were not documented (Meador et al., 2009, 2011, 2013; Cohen et al., 2011, 2019). The relationship between phenytoin dose and poorer neurodevelopmental outcome was investigated, and no dose response was reported (Meador et al., 2013).

Early development of infants exposed in utero to lamotrigine has not been found to deviate from expected trajectories (Bromley et al., 2010; Cummings et al., 2011) and has been demonstrated to be superior to children exposed in utero to valproate (Meador et al., 2009). At school age, the NEAD study found that children exposed in utero to lamotrigine were superior in their IQ, memory and verbal skills compared to foetal valproate exposure (Meador et al., 2013). IQ levels comparable to controls have been demonstrated in two studies, but they were only powered to detect large levels of difference (Rihtman et al., 2013; Baker et al., 2015). Children exposed in utero to lamotrigine had comparable mathematics and Danish language examination results relative to control children (Elkjaer et al., 2018). In contrast, the Norwegian MoBa study reported poorer language and social skills at 36 months based on a parental report for children exposed to lamotrigine (Veiby et al., 2013a). This has not been found by others based on direct child assessment (Bromley et al., 2010), however, relative verbal weakness vs. non-verbal skills was identified in the NEAD study, although verbal skills were within normal range (Meador et al., 2013). Children exposed to lamotrigine have not been found to be at higher risk of autistic spectrum diagnoses (Bromley et al., 2013; Christensen et al., 2013), but a recent study reported an increase in certain behavioural outcomes (Huber-Mollema et al., 2019), however, a comparison to an unexposed population was not undertaken in this later study.

The available data on the longer-term outcomes of children exposed to levetiracetam in utero are provided by a UK study with a small sample size. In this study, the children exposed to levetiracetam did not differ from control children in terms of their early development and performed more strongly than the group exposed to valproate, with the same noted for school-aged IQ (Shallcross et al., 2011, 2014; Bromley et al., 2016). A dose-dependent effect was not observed at any age, but extension of these results into larger prospective cohorts is required. In one study on behavioural outcomes, an increase in certain behavioural problems was reported for children (mean age: 6.5 years), as rated by parents, (Huber-Mollema et al, 2019), however, this requires further investigation as there was no direct comparison to a control group.

Lamotrigine and levetiracetam have become the most commonly used AEDs in pregnancy in tertiary epilepsy centres in the USA (Meador et al., 2018) and other countries (Kinney et al., 2018). However, data for even these two AEDs are inadequate. The size of cohorts and the outcomes investigated are small, and consideration should be given to the variable relationship between prescribed dose and serum blood levels (Johannessen Landmark et al., 2017). Both lamotrigine and levetiracetam exhibit marked changes in metabolism during pregnancy, which can alter foetal exposure, and no study has adjusted for changes in AED blood levels during pregnancy.

Our understanding of the potential impact of other AEDs on the developing brain is almost non-existent. Whilst two studies have investigated topiramate exposure in utero, both were small with conflicting findings (Rihtman et al., 2012; Bromley et al., 2016). For oxcarbazepine, no association with an increased risk of autistic spectrum diagnosis was found (Christensen et al., 2013) and examination performance did not differ relative to controls, with the exception of sixth-grade mathematics, although the magnitude of difference was reportedly small (Elkjaer et al., 2018). Other studies contained sample sizes that were too small to provide reliable information on other neurodevelopmental outcomes associated with oxcarbazepine. No reliable data are currently available for other AEDs.

Breastfeeding has proven beneficial for both the mother and child (Ip et al., 2009). Some have raised concerns over breastfeeding when taking an AED. However, this concern has not been upheld. The NEAD study and a Norwegian study found no adverse neurodevelopmental effects in children at age three years old who were breastfed by mothers on AEDs, and in the NEAD study, children who were breastfed had higher IQ levels at six years of age compared to children of mothers with epilepsy who were not breastfed (Meador et al., 2010, 2014; Veiby et al., 2013b).

Despite the potential detrimental and lifelong impact on brain development, clear replicated data for the majority of currently prescribed AEDs are lacking. Larger cohorts with investigation of a dose-dependent, or even better, blood concentration-dependent influence on outcomes are required. Care should be taken to ensure that a lack of evidence of harm is not taken as evidence of safety, and risk-benefit discussions with patients about their treatment should reflect what is known and additionally, what is not known, about neurodevelopmental outcomes associated with individual AED treatments.

• Exposure in the womb to valproate carries a significant dose-dependent risk associated with child cognition and neurodevelopmental disorders (e.g. autistic spectrum disorder).

• Carbamazepine does not appear to be a major neurobehavioural teratogen.

• 88Current data regarding lamotrigine suggest comparable IQ to control children.

• There are limited data pertaining to levetiracetam, topiramate and other AEDs in terms of later child cognition.

Seizure control

The seizure burden remains unchanged during pregnancy for about two thirds of women (Thomas et al., 2012; Battino et al., 2013). In the prospective observational EURAP study, 67% of women were seizure-free throughout pregnancy (Battino et al., 2013). The period of pregnancy with the highest incidence of seizures is labour and delivery, but this occurs in no more than 1-2% of pregnancies in women with epilepsy (Battino et al., 2013).

The occurrence of seizures before pregnancy is the most important predictor of seizures during pregnancy (Thomas et al., 2012). Women who experienced seizures in the month prior to pregnancy had a 15-fold greater risk of seizures during pregnancy (Thomas et al., 2012). Conversely, more than 80% of women who were seizure-free the year before conception, remained so throughout pregnancy (Vajda et al., 2008).

The impact of a specific pattern of seizures before pregnancy on the course during pregnancy has been investigated in a prospective follow-up of seizure course in women with catamenial epilepsy and women with non-catamenial epilepsy (Cagnetti et al., 2014). Seizure control was improved during pregnancy in the catamenial group; 44.1% experienced a ≥50% reduction in seizures, whereas only 6.5% of those with non-catamenial epilepsy had a similar reduction in seizures.

Other general predictors of seizure worsening during pregnancy are focal epilepsy syndromes, the need for polytherapy, and decreased serum levels of AEDs compared to preconception baseline (Reisinger et al., 2013). The clearance of newer AEDs, such as lamotrigine, levetiracetam, and oxcarbazepine, is increased significantly during pregnancy which may result in break-through seizures related to lower serum levels if dosages are not adjusted (Pennell et al., 2008; Voinescu et al., 2018). This is discussed in more detail in the following section.

Other factors that may predispose to seizure aggravation during pregnancy include anxiety, non-adherence to medication, sleep deprivation, and difficulty in retaining orally administered AEDs due to vomiting.

Appropriate counselling can help to alleviate the stress factors and improve drug therapy adherence, which is in line with the observation that women with planned pregnancies have a lower frequency of seizures during pregnancy (Abe et al., 2014).

Status epilepticus was reported in only 0.6% of all pregnancies in the EURAP study (Battino et al., 2013); of these, 10 were convulsive, and cases were evenly distributed over the three trimesters. Perinatal death occurred in one of the pregnancies of women experiencing a convulsive status epilepticus, and none of the mothers died.

Most women with epilepsy maintain their seizure control during pregnancy. Pre-pregnancy seizure control is the most important predictor of seizure control during pregnancy. Non-adherence to AED medication and alterations in AED clearance are major causes of break-through seizures.

Changes in pharmacokinetics

A challenge in managing epilepsy during pregnancy is the pronounced pharmacokinetic alterations, including altered absorption, increased volume of distribution, elevated renal excretion, and induction of hepatic metabolism. AED clearance is a term that can account for all these changes, and is calculated as follows:

Clearance = AED dose (mg/kg/day) / AED concentration

Knowledge about the pattern of gestational age-dependent clearance changes can help guide the timing and range of AED dose adjustments and contribute to maintaining seizure stability during pregnancy.

Lamotrigine is the most studied AED with regards to clearance changes during pregnancy, with findings of markedly increased clearance during pregnancy (Pennell et al., 2008; Pennell, 2013; Tomson et al., 2013), likely mainly due to oestrogen-driven enhanced glucuronidation during pregnancy. Additionally, substantial inter-individual variability was noted in most of these studies. A formal pharmacokinetic modelling analysis demonstrated two subpopulations, one with 23% of women who had only a 20% increase in lamotrigine clearance and another with 77% of women who had a 220% increase in clearance, hypothesized to be due to pharmacogenetic differences (Polepally et al., 2014). A recent study with frequent sampling, beginning prior to pregnancy and through the first trimester, highlighted that clearance changes are measurable as early as the third week after conception and clearance increases by an average of 50% by the end of the first trimester (Karanam et al., 2018). These findings underscore the importance of performing therapeutic drug monitoring when available and beginning early in pregnancy.

Studies of other AEDs have demonstrated changes in clearance during pregnancy which vary according to the pathway of drug elimination (table 2). For example, early and substantial glomerular filtration rate and renal blood flow increase during pregnancy as well as studies on levetiracetam have demonstrated similar changes in clearance (Tomson et al., 2007; Westin et al., 2008; Lopez-Fraile et al., 2009). A larger and prospective study demonstrated that the increase in clearance is maximal in the first trimester, with a 71% increase above non-pregnant baseline (n=18 pregnancies) (Voinescu et al., 2018). In contrast, studies on carbamazepine report little change in total carbamazepine clearance during pregnancy and no significant change in free carbamazepine or free carbamazepine-10, 11-epoxide concentrations (Johnson et al., 2014).

The clinical importance of the changes in AED clearance has been demonstrated in a few studies. In one early study with lamotrigine (Pennell et al., 2008); preconception clinical data were used to determine an individualized target concentration; the ratio-to-target concentration (RTC) was calculated for each blood draw during pregnancy and an RTC threshold of 0.65 was identified as a significant predictor of seizure worsening in the second trimester. Two other studies have reported that this same rule applies to other AEDs, demonstrating that when the AED concentration decreases to 65% or less of the individual target concentration, the risk of seizure worsening increases (Reisinger et al., 2013; Voinescu et al., 2018).

During the postpartum period, AED clearances return to non-pregnant baseline, but studies delineating the rate of these changes are limited. An empiric taper of lamotrigine over 10 days to pre-pregnancy baseline plus 50 mg resulted in low rates of lamotrigine side effects and seizure worsening (Pennell et al., 2008). However, based on the only formal pharmacokinetic modelling study of lamotrigine to date, clearance was reported at non-pregnant baseline level over three weeks (Polepally et al., 2014).

• Pregnancy can have a major impact on the pharmacokinetics of antiepileptic drugs.
• The most marked decline in serum concentration during pregnancy is seen with lamotrigine, levetiracetam, and oxcarbazepine, but phenobarbital, phenytoin, topiramate, and zonisamide also undergo a clinically relevant increase in elimination.
• A decline in serum concentration by >35% from a pre-pregnancy optimal concentration is associated with increased risk of deterioration in seizure control.
• The extent to which pregnancy affects AED blood levels varies between individual women and is best controlled by blood level sampling.

Management before pregnancy

Preconception care

Preconception care and planning the pregnancy ahead can help to improve the outcome of pregnancy in women with epilepsy. In a study from the UK, women who had proactive counselling prior to pregnancy were more likely to be on monotherapy and on AEDs other than valproate, and the prevalence of MCM in their offspring was lower (Betts and Fox, 1999). Preconception planning and care has also been shown to be associated with improved seizure control and reduced AED burden during pregnancy (Abe et al., 2014). Surveys among women with epilepsy who were considering pregnancy revealed that they did not receive sufficient information on pregnancy and epilepsy (Crawford and Hudson, 2003; McGrath et al., 2014). One major challenge in provision of adequate pre-conception counselling is unplanned pregnancies. In a recent study from the US, approximately 65% of pregnancies in women with epilepsy were unplanned (Herzog et al., 2017). The importance of planning the pregnancy should therefore be brought up regularly in each routine consultation with women with epilepsy with childbearing potential.

The standards of preconception care for women with epilepsy have been discussed and reviewed in the past (Kinney and Morrow, 2016). A protocol for preconception care is provided in table 3. This is the period in which to reconsider the indication for, and the choice of AED treatment. If a change in medication is considered, it should be completed early enough to allow sufficient time to assess the effectiveness of the new regime before conception. This could mean that an attempt to change may need to be initiated as early as one year before a planned pregnancy. The objective of the treatment review and possible revision is to establish, before conception, the lowest effective dose of the appropriate AED for an individual woman, and to document the associated drug serum concentration, when possible.

Folate

It is also important that adequate folate supplementation is initiated in the preconception stage. The US regulatory authorities recommend 0.4 mg of folic acid for all women with child bearing potential and a higher dose of 4 mg daily for women with a higher risk, such as those with past pregnancies involving neural tube defect or anencephaly (Centers for Disease Control, 1992). A higher dose of folate at 4 mg was associated with a lower risk of recurrent neural tube defects based on a randomized controlled trial, in which women with epilepsy were excluded (MRC Vitamin Study Research Group, 1991). In the general population, folate supplementation has been associated with reduced cardiac malformations (Czeizel et al., 2013). There are also several studies in the general population showing positive effects of folate supplementation on neurodevelopmental and behavioural outcomes, but this effect remains controversial (Wehby and Murray, 2008; Julvez et al., 2009; Roza et al., 2010; Schlotz et al., 2010; Roth et al., 2011; Chatzi et al., 2012; Skorka et al., 2012; Villamor et al., 2012). Despite the fact that some AEDs interfere with folate, data on the effects of folate supplementation on pregnancy outcomes in women treated for epilepsy are not conclusive. Reports from the prospective epilepsy pregnancy registries have failed to demonstrate that periconceptional use of folate is associated with a lower risk of MCMs (Morrow et al., 2009; Tomson et al., 2018). The NEAD study found improved IQ scores in six-year-old children of women with epilepsy who began folate prior to conception and in early pregnancy (Meador et al., 2013). In a related UK study, folate supplementation was not found to be associated with increased IQ scores (Baker et al., 2015). In the Norwegian MoBa study, folate supplementation and higher plasma concentration of folate early in pregnancy were associated with a reduced risk of autism symptoms at three years of age in children of women taking AEDs, as rated by the mothers (Bjork et al., 2018). Further, data from the Norwegian MoBa study also highlighted a reduction in language delay in children receiving folate supplementation in comparison to those not taking folate (Husebye et al., 2018). It should be noted that the observational cohort studies (Morrow et al., 2009; Meador et al., 2013; Baker et al., 2015; Tomson et al., 2018) were not primarily designed to assess the effects of folate supplementation. Hence, presence or absence of an association does not prove or exclude a beneficial effect of folate supplementation on AED-induced adverse pregnancy outcomes. Unfortunately, the available data from women with epilepsy provide limited guidance regarding adequate dose of folate. Clearly, women with epilepsy should be prescribed at least 0.4 mg/day since this is the recommended dose for women in general, as well as women with epilepsy according to some guidelines. The fact that some AEDs can interfere with folate suggests that higher doses of folate might be required, but there have been concerns that folate, in particular at high dose, may increase the risk of cancer, cognitive impairment, and oral clefts (Frankenburg, 2009; Rozendaal et al., 2013; Morris et al., 2010; Murray et al., 2018). Some guidelines recommended that all women with epilepsy who are trying for pregnancy should start on folic acid at 5 mg daily, at least three months prior to pregnancy, and continue with the same dose throughout pregnancy (Wilson et al., 2007; NICE, 2012), however, evidence for an optimal periconceptional dose of folate in women with epilepsy taking AEDs remains inadequate. Given the incidence of unplanned pregnancies, women with childbearing potential taking AEDs should also be on folate, at least 0.4 mg/day.

Management during pregnancy and postpartum

For recommended management strategies during pregnancy, delivery, and early postpartum, see table 3. Once pregnancy occurs, it is important to quickly establish detailed coordinated care between the neurologist, obstetrician, and the patient. Since such a large percentage of women will have unplanned pregnancies, the first visit often needs to incorporate elements of preconception care, if not already done so. Regardless, it is ideal to schedule the first pregnancy visit early within the first trimester to ensure that supplementary folic acid is being taken. The physician should re-evaluate the AED dose if not done recently. Counselling of the patient should reinforce the need for AEDs, and any potential AED risk to the foetus should be balanced against the risk of increased seizures to both the mother and the developing foetus (see Maternal and foetal risks associated with seizures). If the woman is taking an AED that undergoes substantial clearance changes (table 2) and if drug levels are obtainable, it is ideal to determine a blood level by the mid first trimester given the early gestational changes during clearance that occur for many of the common AEDs used during pregnancy (Karanam et al., 2018; Voinescu et al., 2018). The individualized target concentration should be reassessed and maintained during pregnancy with blood levels throughout pregnancy. The need for follow-up drug level monitoring will depend on the type of AED (less important for some AEDs with minor changes during pregnancy, see table 2) as well as each woman's sensitivity to alterations in drug levels before pregnancy and her type of epilepsy. Many experts in the field obtain levels every four weeks for AEDs when major changes occur (e.g. for lamotrigine, levetiracetam, oxcarbazepine, phenobarbital, phenytoin, and zonisamide), with additional checks if clinically indicated by seizure frequency changes or medication side effects or concerns about medication adherence. The goal is to adjust the AED dosing to maintain the specific target concentration(s) for each woman with epilepsy, given that the risk of seizure worsening increases if the level decreases by 35% or more from the baseline, non-pregnant target concentration (Pennell et al., 2008; Reisinger et al., 2013; Voinescu et al., 2018).

If AED blood levels are not available, then it is reasonable to consider an increase in dose after the first trimester, at least in women whose epilepsy includes GTCS, that were sensitive to changes in AED levels before pregnancy, and who entered pregnancy on the lowest effective dose of their medication, providing they are treated with AEDs that are known to undergo marked changes during clearance (lamotrigine, levetiracetam, and oxcarbazepine). There is no evidence to guide how this should be done best, and the issue is further complicated by the fact that the timing and extent to which AED levels decline varies considerably between women. A precautionary approach would be to increase the dose by 30-50%, which in most cases is less than required to maintain the target concentration, while avoiding exposure of the foetus to unnecessarily high AED levels in the few mothers whose serum levels decline less on average. If a woman has a break-through GTCS, a dose increase should be strongly considered, especially if convulsions were previously controlled and if the woman is taking an AED with known marked clearance changes during pregnancy. An increase in other kinds of seizures (e.g. focal aware seizures, focal impaired awareness seizures, or JME myoclonic seizures) should also prompt consideration of dose increase.

If the woman is experiencing nausea and emesis, counselling should include strategies to re-dose if emesis occurs shortly after ingestion of her dose. It can be helpful to communicate with the obstetrician for recommendations to reduce emesis if it is interfering with maintaining AED steady-state concentrations.

Patients with epilepsy have higher rates of depression and anxiety than the general population, and this is also the case during pregnancy and the postpartum period. Studies have indicated especially higher rates of postpartum depression in women with epilepsy, and these symptoms often begin intrapartum (Turner et al., 2006; Galanti et al., 2009). Ideally, management of women with epilepsy should incorporate screening for depression and anxiety, especially during pregnancy and postpartum.

In settings where prenatal testing is available and acceptable, usual testing for all women will include first-trimester Down syndrome screening in the late first trimester, between 9-14 weeks gestational age, with combined measurements of serum beta-hCG and serum pregnancy associated plasma protein (PAPP-A), an ultrasound measurement of nuchal translucency, and consideration of maternal age to provide patient-specific risk (Pennell and McElrath, 2019). These tests also provide confirmation of gestational age and the potential to detect some severe MCMs early. However, a more definitive structural ultrasound will be performed at 18-22 weeks of gestation; in women with epilepsy, it is ideal to obtain a detailed (specialized) foetal structural survey by a trained specialist. Foetal echocardiography is not necessary as a routine measure, but only if there is a clinical indication. If there is a concern for intra-uterine growth retardation, then serial ultrasounds may be performed.

The third trimester is a critical time to coordinate recommendations for labour and delivery and early postpartum care between the neurologist and obstetrician, with consideration of the patient's desired birth plan. If prenatal screening has indicated anticipated newborn problems, the neonatology team should also be included in the planning phase. The diagnosis of epilepsy itself is not an indication for Caesarean section. Seizures and their treatment during labour and delivery may interfere with the patient's ability to participate in active labour, however, this occurs only rarely. Seizures during labour and delivery are best treated with the usual rescue therapy using a low dose of a quick-acting benzodiazepine. Vaginal deliveries are the norm. The benefits of pain management with epidural anaesthesia are the same for women with epilepsy as for any other women, and include reducing the duration of intense pain and maximal stress, and facilitating some rest prior to the active stage of labour. At delivery, all children should routinely receive 1 mg vitamin K IM.

Discussion during the late third trimester should also include possible increased seizure worsening peripartum, desires and strategies for breastfeeding, strategies to lessen prolonged sleep deprivation, newborn safety, and a plan to adjust AED dosing postpartum if increases have been made during pregnancy (Voinescu and Pennell, 2017).

If AED dosing has been increased during pregnancy, the rate of taper of AEDs back to pre-pregnancy dose or slightly above depends mainly on the primary route of elimination for each individual AED. The physiological changes of renal and some hepatic enzymatic functions (e.g. glucuronidation) associated with pregnancy will rapidly resolve over the first two to three weeks postpartum, while other hepatic enzymes (many of the cytochrome P450 enzymes) may take one to two months to return to baseline clearance rates (Yerby et al., 1990, 1992). Postpartum AED tapers are prescribed empirically, as a steady-state level is not obtainable with the rapid changes during clearance, and it often takes a few days to obtain results for most of the second- and third-generation AEDs at most clinical centres (see section on Pharmacokinetic changes during pregnancy).

Given the benefits of breastfeeding with regards to both short- and long-term neonatal health in the general population (http://www.who.int/topics/breastfeeding/en/ Archived 2016-02-20 at the Wayback Machine), and the data from studies showing no adverse neuropsychological effects in children of mothers taking AEDs, breastfeeding should in general be encouraged. Treatment should be adapted according to how sensitive their seizures are to sleep deprivation based upon their history and their epilepsy syndrome. However, the risk of seizures in the postpartum period relative to specific sleep patterns has not been addressed in high-quality studies.

The risk of seizures may be increased in the postpartum period, sometimes for several months, due to sleep deprivation. Couples and other family members should be counselled, ideally as a function of prenatal care, to make arrangements in order to allow adequate sleep while caring for a newborn. Some families adopt a “shift” approach so the mother can reliably obtain uninterrupted and regular nightly sleep. Additionally, even if the mother has been seizure-free for a long time, she should take a more conservative safety approach until she achieves regular sleep again, given that sleep deprivation is a strong provoker of many seizure types.

Common sense safety considerations during the newborn period should be discussed and reinforced; these include no driving, no bathing of the baby with the mother alone, and no co-sleeping with the mother in the parents’ bed. If the mother is at risk of myoclonic seizures, then a baby carrier (sling or harness) should be used when walking around with the baby. In the early postpartum period, the mother should also be discouraged from taking a bath by herself behind a closed, locked door or when no other adult is around.

References

• This article incorporates creative commons material from Tomson, T; Battino, D; Bromley, R; Kochen, S; Meador, K; Pennell, P; Thomas, SV (1 December 2019). "Management of epilepsy in pregnancy: a report from the International League Against Epilepsy Task Force on Women and Pregnancy" (PDF). Epileptic disorders : international epilepsy journal with videotape. 21 (6): 497–517. doi:10.1684/epd.2019.1105. PMID 31782407. Archived from the original (PDF) on 24 September 2021. Retrieved 25 March 2021.