In most women with epilepsy, it is necessary to continue antiepileptic drug (AED) treatment during pregnancy to reduce the risks caused by epileptic seizures for both mother and child.^1,2,3 The risk of maternal death during pregnancy has been estimated to be ten times higher in women with epilepsy compared to the general population,⁴ possibly due to poor compliance with AED treatment and seizure occurrence. Seizure-related accidents and convulsive status epilepticus during pregnancy are associated with an increased risk of fetal death.⁵ Even brief generalized tonic–clonic seizures during pregnancy may have an unfavorable effect on cognitive outcome based on retrospective data,⁴ although this has not been confirmed prospectively.^6,7

Foetal exposure to AEDs occurs in 0.3%–0.5% of all pregnancies; of them, 17%–47% are exposed to two or more drugs.^8,9 Since both seizures during pregnancy and the medication needed to prevent seizures may have adverse foetal effects, the treatment of maternal epilepsy is a matter of balance by weighing advantages and disadvantages.

AEDs are freely transported across the placenta, and the foetal serum levels for most drugs are roughly the same as the maternal levels; valproate and possibly gabapentin seem to accumulate in the fetus.^10,11 There is some evidence that infants born to women with epilepsy have low birth weight more commonly than infants of women with no chronic diseases.^12,13 One minute Apgar scores may also be lower and the need for care in the neonatal ward is increased in the infants of mothers with epilepsy.⁹ These effects do not seem to be very strong and are not readily explained by pregnancy complications, exposure to specific AEDs, or seizures during pregnancy.^9,12,14

STRUCTURAL AND FUNCTIONAL TERATOGENESIS

Structural teratogenesis resulting in malformations occurs during the first trimester of pregnancy. The vulnerable period for functional teratogenesis producing cognitive and behavioral disturbances covers the whole pregnancy as neuronal migration and organization are teratogen-sensitive processes continuing also in the second and third trimesters.¹⁵ Animal studies have demonstrated that AEDs have dose-dependent teratogenic effects. The doses required for functional deficits are generally lower than those producing structural anomalies.¹⁶

The prevalence of major malformations in the general population is approximately 2.2%–2.8% according to population-based studies.^8,17 It is well established that prenatal exposure to AEDs increases the prevalence of major malformations in humans (Fig. 51–1).¹⁸ The magnitude of the risk is approximately two- to threefold for most monotherapies compared to the baseline rate. The most common malformations are the same that are common in the general population (e.g., heart defects and facial clefts). The highest relative risk in children of mothers with epilepsy has been observed in spina bifida and congenital anomalies of genital organs.¹⁷ Comparisons between different studies and pooling of data are not always straightforward because of differences in definitions of major malformations, of sampling, observational periods, and in whether terminated pregnancies are included in the analysis.¹⁹ Recently, the EURAP study contributed new data on major malformations showing dose-dependency of the risk for four monotherapies in a prospective series of near 5000 pregnancies with follow-up until the first year of postnatal life (Table 51–1).²⁰

The risk of major malformations does not appear to be significantly increased by epilepsy per se.²¹

Maternal epilepsy is associated with mildly but significantly reduced intelligence in the offspring, based on a large population-based cohort study.¹³ IQ scores of 1207 young men whose mothers had epilepsy during pregnancy were compared with a control group of 316,554 males of similar age. No individual AED exposure data were available from that study. Differential effects of individual AEDs on cognitive development are difficult to estimate as there are many confounding factors and long follow-up periods carry the risk of selective loss to follow-up. Moderate to severe mental deficiency may be diagnosed in the first years of life, but milder cognitive deficits cannot be reliably studied before preschool or early school age and standardized neuropsychological testing are always needed. Good-quality studies on the effects of AEDs on cognitive development are scarce.²²

CONFOUNDING FACTORS

Epilepsy type and severity are associated with AED choice, dosage and whether monotherapy is sufficient or if polytherapy is indicated, leading to a risk of confounding by indication especially in studies investigating cognitive outcome. There are some data suggesting that genetic epilepsy-related factors could sometimes contribute to cognitive outcome in children of mothers with epilepsy. For example, centrotemporal EEG spikes may be genetically determined²³ and may also be associated with increased cognitive²⁴ or attentional dysfunction²⁵ even in the absence of epilepsy. Idiopathic epilepsies which may have familial predisposition are associated with mild cognitive and behavioral dysfunction which is not totally explained by epileptiform discharges²⁶ and which may be present before seizure onset.²⁷ The effects of epilepsy on male and female reproductive fitness may differ²⁸ which implies that comparing offspring of fathers and mothers with similar types of epilepsy does not solve the problem of confounding by indication.

Epilepsy is associated with psychosocial and socioeconomic difficulties, especially in patients who are not seizure-free.²⁹ In addition to possible environmental effects to the offspring, it is conceivable that epilepsy may cause a disadvantage in partner choice,³⁰ expressed in one study by smaller head circumferences in fathers of children whose mothers had epilepsy.³¹

GENETIC VARIABILITY IN SUSCEPTIBILITY TO TERATOGENIC EFFECTS

Pharmacogenetic factors may influence vulnerability for teratogenic effects.³² Low epoxide hydrolase activity is associated with an increased risk of developmental defects after prenatal phenytoin exposure.³³ Dizygotic or heteropaternal twins or triplets from phenytoin-exposed multiple pregnancies have shown different outcomes, some infants being healthy while others with identical exposure have congenital defects.^34,35 Repeated occurrence of developmental defects in siblings after prenatal valproate^36,37,38 or phenytoin³⁹ exposure has been described. These observations suggest that AEDs exert their teratogenic effects in interaction with a specific genetic susceptibility.

The nature of genetic susceptibility factors is not well studied in human maternal epilepsy but probably subject to extensive polygenetic heterogeneity, both at the level of metabolism and pharmacodynamics. In addition, one has to take into account that genetic factors related to the maternal phenotype as well as independently segregating genetic factors may play a role in the occurrence of abnormalities in the offspring.

PHENYTOIN

Recent prospective cohort and registry studies have reported 2.4%–3.7% prevalence of major malformations after prenatal exposure to phenytoin monotherapy.^14,40 A meta-analysis of five prospective European studies found malformations in 9 of 141 monotherapy exposed pregnancies (6%),¹⁹ not significantly different from nonexposed controls.

The foetal hydantoin syndrome was first described in 1975.⁴¹ Five unrelated children of mothers with epilepsy had been prenatally exposed to either 100–400 mg of phenytoin (four children, only one monotherapy) or 300 mg of another hydantoin, mephenytoin. The characteristic pattern of minor anomalies included a short nose with low nasal bridge, hypertelorism, and hypoplasia of nails and distal phalanges. Other syndrome features were postnatal growth deficiency and motor or mental deficiency. The frequency of the foetal hydantoin syndrome among phenytoin-exposed children is unknown. The first estimation was as high as 11%,⁴² but this has not been confirmed in prospective population-based studies.^6,43

The facial features of the foetal hydantoin syndrome have been observed in controlled prospective and retrospective studies in exposed and nonexposed children of mothers with epilepsy and also as part of normal variation.^43,44 Distal digital hypoplasia, however, seems to be consistently associated with prenatal phenytoin exposure. This has been shown in prospective-controlled studies blinded to exposure using clinical^43,45 or anthropometric methods.^46,47 Prospective studies found no association between digital hypoplasia or deficient growth, and abnormal cognitive development.^6,47

Developmental data obtained with standardized methods in preschool or school-aged children with prenatal phenytoin exposure and maternal epilepsy are available from three studies—prospective, controlled, population-based—with evaluation blinded for prenatal exposure.^6,48,49 Two studies^6,48 controlled for socioeconomic class or maternal educational level. A total of 364 children of women with epilepsy were included, and approximately 50%–60% of the population were covered in each study. Out of the total, 222 children were exposed to phenytoin, 96 of them to monotherapy. IQ was assessed at age 4 (Stanford–Binet)⁴⁸ or 5.5 years (age-appropriate Wechsler scale),⁶ or a developmental quotient was obtained in six domains by the Griffiths test at age 2–8 years.⁴⁹ The results were compared with 40 nonexposed children of women with epilepsy and over 27,000 control children of mothers without epilepsy. The mean phenytoin dose was reported in only one study (253 mg/day).⁴⁹ Maternal phenytoin levels during pregnancy were reported in another study;⁶ most were within the reference range. Two studies^6,48 reported lower IQ values in children of women with epilepsy compared to control children of mothers without epilepsy, but no significant associations were observed to phenytoin or other drug exposure. The 15 children exposed to phenytoin monotherapy assessed by the Griffiths test had significantly lower scores in locomotor function; no difference was found in the other domains.⁴⁹

Prospective IQ data at 7 years were also reported by Hanson et al⁴² from the same database as Shapiro et al.⁴⁸ The results were controlled for socioeconomic status. The mean IQ of 83 children exposed to phenytoin (number with monotherapy not stated, but 25% at most) was five points lower than control children of mothers without epilepsy. All children of women with epilepsy were exposed to phenytoin; thus the independent drug effect cannot be estimated.

The prevalence of mental deficiency (IQ below 70) can be estimated from a prospective population-based study comprising 103 phenytoin-exposed children (54 monotherapy).⁶ One child exposed to phenytoin, carbamazepine, and alcohol had mental deficiency (IQ below 70). Of the children exposed to phenytoin monotherapy, one had borderline IQ (below 85). For comparison, the prevalence of mental deficiency was found to be 1.4% among 8–9-year-old children in a population-based study from the same country.⁵⁰

PHENOBARBITAL AND PRIMIDONE

A prospective registry study⁵¹ observed an increased risk of major malformations (5/77 or 6.5%) in pregnancies exposed to phenobarbitone monotherapy (1.6%, RR 4.2, 95% CI 1.5–9.4). Based on another prospective registry study, the risk is dose-dependent (Table 51–1).²⁰ A large retrospective study comprising 172 monotherapy exposed pregnancies⁵² observed five malformed infants (3%), which was not increased compared to nonepileptic controls. However, the study revealed the unexpected finding that the combination of phenobarbitone with caffeine was associated with a significantly elevated malformation risk.

Prospective-controlled data on cognitive outcome at 4 years of age after prenatal exposure to phenobarbital monotherapy have been reported in 35 exposed children of women with epilepsy and 4705 children of mothers without epilepsy but with phenobarbital exposure.⁴⁸ There was no IQ difference compared to unexposed control children.

Another study of male subjects whose mothers did not have epilepsy but were treated with phenobarbital for at least 10 days during pregnancy for variable obstetric indications⁵³ showed different results. The data was acquired from the Danish Perinatal Cohort, comprising the offspring of 9006 deliveries that took place at one hospital in Copenhagen from 1959 to 1961. An attempt was made to match for maternal phenobarbital indication among other potential confounders. Drugs used during pregnancy were recorded prospectively. Total phenobarbital dosages during pregnancy ranged from 225 mg to 22,500 mg. On standardized IQ testing at a mean age around 20 years, men exposed to phenobarbital showed significantly lower IQ scores (approximately 0.5 SD) than controls. The IQ impairment was greatest after exposure in the third trimester, especially in subjects of low socioeconomic background who were offspring of unwanted pregnancy.

Data on primidone-exposed pregnancies are scarce. A pattern of minor anomalies similar to phenytoin combined with developmental delay has been described in association with prenatal primidone exposure⁵⁴ but not confirmed as a separate syndrome. The only prospective results on cognitive outcome after prenatal exposure to primidone monotherapy are based on nine children who were tested at age 11–18 years.⁵⁵ There were no differences compared to 49 control children of mothers without epilepsy.

CARBAMAZEPINE

A meta-analysis of prospective studies included 795 exposures to carbamazepine monotherapy;⁵⁶ major malformations were observed in 5.3% that was significantly increased compared to controls without epilepsy (2.3%, OR 2.4, 95% CI 1.6–3.4). A recent UK registry study⁴⁰ found major malformations in 20/900 pregnancies exposed to carbamazepine monotherapy (2.2%, 95% CI 1.4–3.4) that was not significantly increased compared to unexposed pregnancies of women with epilepsy. Like valproate, maternal carbamazepine use is also associated with an increased risk of neural tube defects, although the risk is lower (0.5%–1.0%).⁵⁷ The risk of malformations is dose-dependent (Table 51–1).²⁰

A pattern of minor anomalies together with developmental delay has been described in a case series of children with prenatal carbamazepine exposure.⁵⁸ Many of the typical features overlap with dysmorphic features described after other exposures or with no exposure,^43,44 and the existence of a specific carbamazepine syndrome has not been confirmed.

Two recent population-based prospective controlled evaluator-blinded studies have reported cognitive outcome measured by standardized methods at preschool to school age in children with prenatal exposure to carbamazepine monotherapy. One study⁴⁹ found no difference in the results of the Griffiths test at 2–8 years between 35 carbamazepine-exposed children and 66 control children of mothers without epilepsy. The other study⁷ included 86 children exposed to carbamazepine, 45 nonexposed children, and 141 control children of mothers without epilepsy; there were no differences between these groups in the IQ scores obtained by the age-appropriate Wechsler scale at 5–11 years.