Abstract
Most pregnant women with epilepsy need pharmacological treatment during pregnancy. Children exposed to antiepileptic drugs have an increased risk of being born with major malformations. Some antiepileptic drugs seem to have negative effects on psychomotor or cognitive development in children exposed during foetal life. Neither carbamazepine nor lamotrigine in monotherapy seem to affect the cognition of exposed children. Several studies have shown negative effects on the long-term neurodevelopment of children prenatally exposed to valproic acid or polytherapy (two or more antiepileptic drugs during pregnancy). For most of the newer antiepileptic drugs there are insufficient data regarding long-term outcome.
Epilepsy and pregnancy
Pregnant women with epilepsy are a group of patients who challenge the healthcare system. The safety and health of both the woman and the foetus must be taken into consideration. A pregnant woman in good health and with adequate symptom control has a greater chance of giving birth to a healthy baby. However, although pharmacological treatment may be beneficial for both mother and baby, it is also potentially harmful to the developing foetus.
Of all pregnant women, 0.3–0.4% have epilepsy [Olafsson et al. 1998]. Women with epilepsy are a heterogeneous group, both with regard to type of seizures, the pathogenesis of the condition, comorbid conditions, and medication before conception. Few studies have observed increased mortality during pregnancy in women with epilepsy, however, a study by Adab and colleagues showed that British women with epilepsy had tenfold increased mortality compared with the general population [Adab et al. 2004].
A large number of studies have concluded that first trimester exposure to most of the older types of antiepileptic drugs (AEDs) is associated with an increased risk of congenital malformations [Meador et al. 2008]. Valproic acid (VPA) is a well-known teratogen for which is reported an increased risk of several major malformations such as neural tube defects, hypospadia and orofacial cleft [Jentink et al. 2010b; Samren et al. 1997]. Prenatal folate is recommended to reduce the risk of neural tube defects in women medicating with AEDs during pregnancy [Harden et al. 2009b]. Polytherapy is related to an increased risk of congenital malformations compared with monotherapy [Wide et al. 2004; Samren et al. 1997]. Carbamazepine (CBZ) is associated with an increased risk of malformations, especially spina bifida, compared with the general population, but the risk is lower than that of VPA exposure [Jentink et al. 2010a]. Although the relative risk is increased with exposure to CBZ compared with the general population, the absolute risk is still small [Nulman, 2010]. Lamotrigine (LTG) is considered a safer option than VPA with regard to malformations, but symptom control may be less effective [Vajda et al. 2010]. Several studies have shown a markedly increased clearance of LTG during pregnancy, leading to decreased plasma levels. It is recommended that drug levels are measured repeatedly in pregnant patients who are medicating with LTG [Harden et al. 2009c; Tomson and Battino, 2007]. A Danish population-based cohort study of children exposed to oxcarbazepine, levetiracetam, topiramate, lamotrigine or gabapentin could not find any association between these AEDs and major birth defects. For some of the drugs, the number of exposed infants was low [Mølgaard-Nielsen and Hviid, 2011].
There have been numerous reports of a specific appearance or combination of different minor malformations attributable to intrauterine exposure to AEDs. These include, among others, hypoplasia of the fingers, midface hypoplasia, nail hypoplasia and joint stiffness. Some studies have found a correlation between intelligence quotient (IQ) and minor anomalies in children [Holmes et al. 2005], whereas others have not [Wide et al. 2000].
Long-term neurodevelopment in children exposed to maternal epilepsy and antiepileptic drugs
During the 1970s, reports were published on children exposed to anticonvulsants in utero, who displayed delayed psychomotor development, minor anomalies and congenital malformations [Hanson and Smith, 1976]. Since then, efforts have been made to establish whether AEDs during pregnancy lead to an increased risk not only of congenital malformations but also of impaired development of neurological and cognitive functions. Long-term neurodevelopment can be assessed in several ways, but IQ measured by a well-known standardized test is a common one. Test results at preschool age may not reflect later cognitive and motor function when tested at school age [Roze et al. 2010]. Longitudinal studies that follow children up to school age and beyond are preferred. Other types of outcome have also been studied, for example parental reports of maladaptive behaviour, school performance or neuropsychiatric conditions [Forsberg et al. 2011; Bromley et al. 2008; Rasalam et al. 2005; Adab et al. 2001].
Animal studies have shown behavioural deficits in rats or mice prenatally exposed to phenobarbital (PB), phenytoin (PHT) or VPA [Fisher and Vorhees, 1992]. The combination of several AEDs seems to lead to a higher apoptotic response in the immature rat brain [Bittigau et al. 2002].
Cognitive and psychomotor development are, of course, very complex processes that are influenced by many internal and external factors such as parental IQ, parental education and socioeconomic status [Meador et al. 2011a; Eriksson et al. 2005; Losche et al. 1994]. Factors related to the child’s health, such as gestational age at birth, perinatal complications and subsequent health, are also of importance for the mental development.
There may be negative effects of maternal epilepsy that are not related to prenatal exposure to AEDs but to some genetic or environmental effect of the disease [Karouni et al. 2010]. Holmes and colleagues found no significant IQ difference between 57 children born to women with epilepsy who had not been medicating with AEDs during pregnancy and matched controls [Holmes et al. 2000]. A German group investigated possible genetic effects by comparing children born to women with epilepsy not taking AEDs during pregnancy, children exposed to AEDs during pregnancy and children who had fathers with epilepsy. They found more speech difficulties in the group of children exposed to AEDs than in the two other groups [Steinhausen et al. 1994].
There is also a possibility that generalized tonic–clonic seizures can have a negative effect on neurodevelopment. Most studies have failed to find such a correlation [Bromley et al. 2010; Meador et al. 2009; Titze et al. 2008; Gaily et al. 1988], but Adab and colleagues showed a negative association between five or more tonic–clonicseizures during pregnancy and cognitive outcome [Adab et al. 2004].
Long-term neurodevelopment following exposure to specific antiepileptic drugs during pregnancy
Phenobarbital
A Danish study investigated adult men prenatally exposed to PB. These men were born between 1959 and 1961 and their mothers had been treated for other reasons than epilepsy. Men exposed to PB in utero had lower verbal IQ scores than their matched controls. Third trimester exposure affected the results more than exposure earlier in the pregnancy. The strengths of this study are long follow-up time and consistent results between a group of men tested with the Wechsler intelligence scale and a more simplified test performed at conscription. The diversity of underlying diseases in the mothers makes the results of this study difficult to interpret [Reinisch et al. 1995].
A follow-up study of children born to women who received a single prenatal dose of PB to prevent intracranial haemorrhage in the infant revealed no significant differences between children prenatally exposed to PB and a control group [Shankaran et al. 2002].
Phenytoin
Hanson and Smith first reported foetal hydantoin syndrome. Five children exposed to PHT in utero displayed a similar pattern of abnormalities: craniofacial anomalies, nail and digit hypoplasia, prenatal onset growth deficiency, and mental deficiency. However, four out of five children reported were also exposed to other AEDs in utero [Hanson and Smith, 1976].
Several studies have shown lower scores in various cognitive or psychomotor tests in children exposed to PHT in utero. Vanoverloop and colleagues showed that 20 children exposed to PHT (polytherapy or monotherapy) scored significantly lower in the Wechsler Preschool and Primary Scale of Intelligence (WPSSI) or the Wechsler Intelligence Scale for Children (WISC) compared with children whose mothers did not have epilepsy. This study was not population based, information on maternal medication was collected post partum and there was no information on parental IQ, only on socioeconomic status [Vanoverloop et al. 1992]. However, similar results were seen in a prospective cohort study with higher methodological quality, in 34 PHT-exposed toddlers who were compared with children exposed to nonteratogens, such as penicillins, during pregnancy. The mean IQ in the exposed group was about 10 points lower than in the unexposed group [Scolnik et al. 1994]. Significantly lower scores in the locomotor subtest of the Griffiths’ Mental Development Scale (GMDS) in a prospective cohort of AED-exposed children were found in 16 children exposed to PHT monotherapy. No differences were seen in the total score in this small but population-based cohort [Wide et al. 2002].
None of these small cohort studies showed an increased risk of mental retardation in children exposed to PHT in utero. A large, register-based study did not reveal an increased risk of not receiving a final grade from compulsory school in children exposed to PHT in monotherapy [Forsberg et al. 2011]. Several other studies have reported no differences between children exposed to PHT in monotherapy and children without AED exposure [Adab et al. 2004; Gaily et al. 1988].
Carbamazepine
CBZ has, in most studies, shown no negative effects on neurodevelopment. A well-designed Finnish prospective cohort study, a Canadian study of preschool children and a recent British study of toddlers exposed to CBZ all concluded that children prenatally exposed to CBZ in monotherapy displayed no significant differences in cognitive or psychomotor testing compared with nonexposed peers [Bromley et al. 2010; Gaily et al. 2004; Scolnik et al. 1994]. In our register-based study of children born to women with epilepsy, CBZ-exposed children did not have an increased risk of not receiving a final grade at age 16 compared with the general population. We found that children exposed to CBZ monotherapy had a significantly increased risk of not receiving a high grade in mathematics, English and Swedish. However, these results are difficult to interpret in a register study with limited access to data regarding confounders [Forsberg et al. 2011]. Meador and colleagues found a significant negative association between dose of CBZ during pregnancy and verbal index when prenatally exposed children were tested at 3 years of age [Meador et al. 2011b].
Valproic acid
Recent studies have increasingly focused on the effects of VPA on the developing brain. Adab and colleagues found, in a retrospective cohort study, that 22% of school-aged children exposed to VPA in utero had exceptionally low verbal IQ (69 or below), compared with 2% in a group of children unexposed to maternal epilepsy and AEDs, and 8% in children prenatally exposed to CBZ [Adab et al. 2004]. The same cohort study also showed more additional educational needs in the VPA-exposed group [Adab et al. 2001] and a correlation between lower verbal IQ and dysmorphic facial features [Kini et al. 2006]. A Finnish prospective cohort study that included 13 children prenatally exposed to VPA also found a significantly lower verbal IQ compared with children exposed to other AEDs or children born to women without epilepsy [Gaily et al. 2004].
In 2009, a large prospective multicenter study from the Neurodevelopmental Effects of Antiepileptic Drugs (NEAD) study group compared four groups of children prenatally exposed to monotherapy: CBZ, VPA, PHT or LTG. The children were tested at age 2–3 years and large amounts of data were gathered regarding the mother, the course of pregnancy, and the child. Children exposed to VPA had a significantly lower full-scale IQ at age 3 years. However, no significant differences were seen between the therapy groups when the analyses were restricted to children exposed to doses below the median, in the case of VPA meaning doses below 1000 mg/day. In children exposed to LTG, CBZ or PHT there was a significant correlation between maternal IQ and the child’s IQ. This correlation was not seen in children exposed to VPA [Meador et al. 2011a; Meador et al. 2009]. A recently published paper from the same study reported a number of other neuropsychological tests on the same study population. Children exposed to VPA had significantly lower verbal scores than children exposed to other monotherapies. There was a negative association between dose of AED and verbal and nonverbal scores [Meador et al. 2011b]. An Australian register-based study also reported dose-dependent negative effects on verbal abilities in school-aged children prenatally exposed to VPA. This small study contained prospectively gathered data on maternal epilepsy, health history and socioeconomic status, but maternal or paternal IQ was not tested [Nadebaum et al.2010].
Valproic acid has also been suspected of increasing the risk of autism spectrum disorders (ASD) [Williams et al. 2001]. A population-based cohort study from the UK found a significantly higher prevalence of ASD in children prenatally exposed to AEDs. Of children exposed to VPA, 8.9% (95% CI 1.3–16.5%) had been diagnosed with autistic spectrum disorder or Asperger’s syndrome, compared with a prevalence of 0.25% in a UK survey of the population [Rasalam et al. 2005]. Preliminary data from a British cohort study reported that 6.3% of children exposed to VPA in utero had been diagnosed with ASD. This prevalence was higher than expected, compared with both a control group and the general population [Bromley et al. 2008]. Another study from the same group reported an increased risk of delayed early development in children exposed to VPA in utero [Bromley et al. 2010].
Lamotrigine
In a study that prospectively followed a group of preschool children, those exposed to LTG or CBZ showed no significant difference compared with control children when tested with the GMDS. A drawback of this study is the young age at testing (mean age 10 months) but it was otherwise well performed [Bromley et al. 2010]. There were no significant differences in psychomotor and cognitive testing at age 3 years between children exposed to LTG, CBZ or PHT reported in the paper from the NEAD study group [Meador et al. 2009].
Polytherapy
There are several studies reporting that polytherapy during pregnancy has more harmful effects on neurodevelopment than monotherapy [Forsberg et al. 2011; Thomas et al. 2008; Titze et al. 2008; Gaily et al. 2004; Losche et al. 1994]. However, an observational study of pregnant women with epilepsy may include women taking a large number of different drug combinations. The conclusions drawn from a ‘polytherapy group’ may be accurate for some of these drug combinations, but not necessarily for all of them. Women who need to medicate with two or more AEDs during pregnancy may have a more severe form of epilepsy than women who become seizure free from taking just one drug. This may have an impact on the cognitive and psychomotor outcome of the child.
A population-based Finnish cohort study found a lower verbal IQ (WPPSI-R or WISC) in children exposed to polytherapy in utero compared with children born to women without epilepsy and children exposed to monotherapy. These results were adjusted for maternal education but not maternal IQ [Gaily et al. 2004]. Dessens and colleagues retrospectively studied school performance and tested intelligence in 147 adults prenatally exposed to AEDs (PB, PHT and combinations, also including amphetamine) and found significantly lower scores in the polytherapy group, compared with matched controls. Retrospective design and the inclusion of amphetamine-exposed subjects are drawbacks of this study [Dessens et al. 2000]. In our register-based study of children exposed to maternal epilepsy with and without AEDs during pregnancy, we found that children exposed to two or more drugs had a higher risk (odds ratio 2.99, 95% CI 2.14–4.17) of not receiving a final grade from compulsory school than the overall population. No information on type of maternal epilepsy or seizures during pregnancy was available [Forsberg et al. 2011].
A recent well-performed prospective cohort study found no statistical differences in early overall development between young children exposed to polytherapy compared with monotherapy or polytherapy, compared with controls [Bromley et al. 2010].
Newer antiepileptic drugs
Most of the newer AEDs, such as topiramate, levetiracetam (LEV) or zonisamide, have not been systematically studied with regard to long-term neurodevelopment. A recently published study from the Liverpool Manchester Neurodevelopment Group assessed 51 children exposed to LEV during pregnancy. The children were tested with the GMDS at 3–24 months of age. Children exposed to LEV showed significantly higher developmental scores than children exposed to VPA in utero. There were no significant differences between children exposed to LEV and control children born to mothers without epilepsy [Shallcross et al. 2011]. This study was well designed, prospective and adjusted for many important confounders such as maternal IQ. Testing at an older age would, however, have provided more information.
Discussion
There are many factors that contribute to the quality of a study on prenatal risk factors and neurodevelopment. Population-based studies are generally considered to have a higher external validity. Some of the above-mentioned studies are population based [Forsberg et al. 2011; Rasalam et al. 2005; Gaily et al. 2004], whereas others have recruited patients from tertiary centres [Meador et al. 2009; Adab et al. 2004; Steinhausen et al. 1994] or from teratology information centres and tertiary units [Scolnik et al. 1994]. Retrospective studies may overestimate the risks attributable to intrauterine drug exposure, mainly due to recall bias. Some well-performed retrospective studies in this field have been very important in addressing the risks attributed to, for example, VPA [Adab et al. 2004]. However, prospective studies are generally considered to have a higher scientific value.
Randomized studies are not ethically possible in this field of research. Confounders must therefore be taken into consideration when studies are designed. There are many potential confounding factors, such as parental IQ or parental education, that may influence the outcome of children prenatally exposed to maternal epilepsy and AEDs. Some of the earlier studies [Wide et al. 2000; Losche et al. 1994] have not adjusted for maternal IQ but most of the larger, more recent ones have. Register studies have natural limitations when it comes to accessing certain baseline data and adjusting for them, for example maternal IQ. However, in our register-based study on school performance, information on maternal education was available. The NEAD study group has made a very ambitious effort to control and adjust for a large number of potential confounders [Meador et al. 2009]. Some information, such as paternal IQ, was not obtained and the authors received some criticism on this matter [Battino, 2009].
Most women with epilepsy have an uncomplicated pregnancy and give birth to a healthy child. However, a woman with epilepsy needs specialized care during pregnancy and there are certain considerations that need to be taken with regard to the child. One of the most important recommendations to women of childbearing age with epilepsy is that their pregnancies should be planned. A woman who is seizure free 9 months before pregnancy has a high chance of staying seizure free throughout pregnancy [Harden et al. 2009a]. Physicians who treat women with epilepsy should discuss the most efficient and safe treatment plan for the patient before pregnancy. It is also important that paediatricians who treat adolescents raise the issues of pregnancy and contraception. Depending on the type of epilepsy and the likelihood of recurrence while off treatment, an attempt to stop medication before a planned pregnancy in women who have been free from seizures during a long period of time may be considered. For certain types of epilepsy, such as epilepsy resulting from a structural lesion, or juvenile myoclonic epilepsy, this is not advisable. Since many pregnancies are, in reality, unplanned [Finer and Henshaw, 2006], pharmacological recommendations regarding pregnant women have strong implications for how epilepsy in women of childbearing age is managed.
An expert committee assembled by the American Academy of Neurology in 2009 published a systematic review along with recommendations regarding the care of women with epilepsy during pregnancy. The committee recommended that ‘avoidance of VPA and antiepileptic polytherapy throughout pregnancy should be considered and avoidance of PHT and PB may be considered to prevent reduced cognitive outcomes’ [Harden et al. 2009b].
The NEAD study added further information on the negative effects of VPA on neurodevelopment. In this paper, the authors conclude that ‘valproic acid should not be used as a first-line antiepileptic drug in pregnant women’ [Meador et al. 2009]. There are, however, different opinions on how to interpret the results, and voices are raised partly in the defence of VPA, stating that in lower doses it may be as safe as other AEDs [Battino, 2009].
It is a difficult task to balance the health of the mother against the possibly teratogenic effects of pharmacotherapy or uncontrolled seizures. Prenatal exposure to certain AEDs may be a rare, but preventable, cause of developmental delay in children. To adopt a principle of caution and make strong efforts to avoid VPA and polytherapy during pregnancy seems, to us, to be the most reasonable way to approach this problem.
Newer AEDs must be thoroughly evaluated with regard to the long-term outcome of exposed children before it can be established that they are safe to use in pregnant women. Women with psychiatric conditions may also be considered for treatment with AEDs during pregnancy. These groups need to be further investigated.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Conflict of interest statement
The authors declare no conflict of interest in preparing this manuscript.