Clinical manifestations

Presentation and course

Most women with active epilepsy need treatment with antiseizure drugs during pregnancy. Antiseizure drugs are also frequently used for other indications, such as migraine, pain syndromes, and psychiatric disorders, which are prevalent among women of childbearing age.

It is generally accepted that antiepileptic drug treatment during the first trimester of pregnancy is associated with a small, but significant, increase in the risk of major congenital malformations. This risk is:

	• Probably no different or only slightly higher than the background (around 1% to 2%) in women with epilepsy who are not taking antiseizure drugs.
	• Probably less than twice the background rate with commonly used antiseizure drugs (other than valproate and topiramate) as monotherapy, although the relative risk may vary with individual antiseizure drugs.
	• Certainly increased with valproate monotherapy to three to five times the background rate.
	• Likely to be high with topiramate, though the relevant studies are still inadequate.
	• Certainly higher with polytherapy than monotherapy; valproate is a significant contributor to the high risk of major congenital malformations in polytherapy, particularly in combination with lamotrigine (10%).
	• Likely to be dose dependent, at least for valproate and probably for lamotrigine (ie, the higher the plasma antiseizure drug concentration, the higher the relative risk of major congenital malformations) (157).

The American Academy of Neurology (AAN) and the American Epilepsy Society (AES) published in 2009 three practice-parameter documents, each focusing on different aspects of pregnancy (61; 62; 63). The conclusions on Teratogenesis and perinatal outcomes are:

	• Antiseizure drug polytherapy probably contributes to the development of major congenital malformations and reduced cognitive outcomes compared to monotherapy.
	• Intrauterine first-trimester valproate exposure has the higher risk of major congenital malformations compared to most other older antiseizure drugs and this may be increased in polytherapy.
	• Intrauterine exposure to valproate monotherapy probably reduces cognitive outcomes and monotherapy exposure to phenytoin or phenobarbital possibly reduces cognitive outcomes.
	• Neonates of women with epilepsy taking antiseizure drugs probably have an increased risk of being small for gestational age and possibly have an increased risk of a 1-minute Apgar score of less than 7 (62).

The recommendations, based on these findings were: if possible, avoidance of valproate and antiseizure drug polytherapy during the first trimester of pregnancy should be considered to decrease the risk of major congenital malformations. If possible, avoidance of valproate and antiseizure drug polytherapy throughout pregnancy should be considered and avoidance of phenytoin and phenobarbital throughout pregnancy may be considered to prevent reduced cognitive outcomes.

An updated position statement by the AES in 2021 emphasized the importance of informing women of childbearing age of teratogenicity before initiating treatment with valproic acid.

The fetal anticonvulsant syndrome. Fetal anticonvulsant syndrome is a cluster of a variety of congenital malformations in infants exposed to antiseizure drugs in utero. These include major congenital malformations, minor malformations, cognitive and behavioral disturbances and intrauterine growth retardation. Many attempts have been made to characterize the different patterns of malformations that are typical or diagnostic for in utero antiseizure drug exposure. Nonetheless, now it is generally accepted that many of the conventional antiseizure drugs may produce a similar pattern of malformations (82; 03; 182; 62; 156; 164; 133; 159).

Major congenital malformations from in utero exposure to antiseizure drugs. Major congenital malformations are generally defined as abnormalities of an essential anatomical structure present at birth that interfere significantly with function and/or require major intervention. Such structural abnormalities are established during embryogenesis, up to 10 weeks of gestation, often before the woman is aware that she is pregnant. Malformations among offspring of women with epilepsy are not unique but generally follow a pattern similar to what is seen in the general population, with cardiac defects being the most common followed by facial clefts and hypospadias, but with some variation between different antiseizure drugs. Findings from several studies have confirmed greater risks of such malformations in children exposed to antiseizure drugs in utero; the risk was about three times that in the children of healthy women. A pooled analysis of data from 21 prospective studies looked at four different groups of major congenital malformations (cardiac, neural tube defects, orofacial clefts, and hypospadias) associated with monotherapy exposure to the five most commonly used antiseizure drugs in these studies (156). Cardiac malformations were the most frequent of the four major congenital malformations for carbamazepine, lamotrigine, barbiturates, and phenytoin, whereas neural tube defects were the most common for valproate. Cardiac malformations appeared more frequently with barbiturates than with any of the other antiseizure drugs, whereas neural tube defects and hypospadias were more prevalent with valproate than with the other antiseizure drugs. Phenytoin possibly contributes to the risk of cleft palate; carbamazepine possibly contributes to the risk of posterior cleft palate; valproate probably contributes to neural tube defects and facial clefts and possibly contributes to hypospadias; and phenobarbital exposure in utero possibly contributes to cardiac malformations (156).

Rates of major congenital malformations from the current major epilepsy and pregnancy registries associated with any specific antiseizure drug vary across the different registers, which can be explained by their methodological differences. However, some consistent patterns are seen within each register and have been synthesized in a 2023 Cochrane review of malformations in pregnancies complicated by antiseizure medication monotherapy (11). First, with few exceptions the major congenital malformation prevalence is highest in association with valproate exposure in all registers, ranging from 4.7% to 13.8%. Major congenital malformation rates with the two most frequently used antiseizure drugs, carbamazepine and lamotrigine, are lower than with valproate in all registers, and comparable between themselves in most studies. The major congenital malformation rate with phenobarbital is reported in three registers and appears to be in between rates reported for lamotrigine/carbamazepine and valproate. Major congenital malformation rates with phenytoin are highly variable, from 2.4% to 6.7%, and based on few exposures in most registers. The prevalence of major congenital malformations with levetiracetam appears to be low and so far in a similar range as lamotrigine/carbamazepine, although the number of exposures is still lower than for these older antiseizure drugs. Exposures to topiramate are too few to draw firm conclusions but the data from most registers indicate a higher prevalence than with lamotrigine/carbamazepine, although in general not as high as with valproate. It should be emphasized that the precision of the estimates with newer antiseizure drugs such as levetiracetam, oxcarbazepine, and topiramate is still unsatisfactory due to the limited number of exposed pregnancies. This is even more the case with other newer generation antiseizure drugs such as gabapentin, pregabalin, zonisamide, perampanel, and lacosamide (156; 159). See pregnancy and epilepsy.

Minor congenital malformations from in utero exposure to antiseizure drugs. Minor anomalies are defined as structural deviations from the norm that do not constitute a threat to health (133). They affect 6% to 20% of infants born to women with epilepsy, approximately 2.5-fold the rate of the general population (133), although there is little agreement about their frequency (149). The problem is complicated by the confounding influences of socioeconomic and genetic factors. Minor anomalies are usually subtle defects of appearance and structure evaluated subjectively or by measurement. Whereas malformations arise during blastogenesis and organogenesis, minor anomalies are defined as arising during phenogenesis (attainment of final form, between days 57 and 266 of development). Most of these features, though, have minor overlap with normal variations seen in children born to healthy mothers. Prospective and blinded studies have shown that only hypertelorism and digital hypoplasia occurred at any great frequency and even these associations are weak (149).

Minor congenital abnormalities include facial dysmorphisms, digital, skin, and genital anomalies, as well as digital anomalies (137; 49; 03; 174).

Characteristic dysmorphic features have been described with fetal exposure to phenytoin, valproate, and carbamazepine (114). The facial features of the fetal valproate syndrome include epicanthic folds, an infraorbital groove, medial deficiency of the eyebrows, flat nasal bridge, short nose with anteverted nares, smooth or shallow philtrum, a long thin upper lip, a thick lower lip, and a small, downturned mouth. Fetal phenytoin is said to be associated with hypertelorism, broad nasal bridge, short nose, and facial hirsutism. Fetal carbamazepine face includes epicanthic folds, short nose, long philtrum, and upward slanting palpebral fissures. However, these fetal anticonvulsant syndromes appear to have some features in common, such as a smooth or shallow philtrum and thin upper lip. Similar features give rise to the facial gestalt of fetal alcohol syndrome, of which the characteristic facial findings are short palpebral fissures, smooth philtrum, and a thin upper lip. It has also been reported that untreated maternal epilepsy is associated with minor dysmorphic features in the child, such as high forehead, frontal bossing, and epicanthus (114).

It has been suggested that there may be a fetal teratogen face arising from exposure to substances that affect the central nervous system and which includes epicanthic folds, short palpebral fissures, smooth philtrum, and thin upper lip (114). These features were seen with similar frequency in children exposed to any of the antiseizure drugs. Other facial features may be associated with specific teratogens, for example, a high, broad forehead, infraorbital grooves, broad nasal root with short nose, anteverted nares, and long philtrum were seen more commonly in valproate exposure. Telecanthus is more prominent with carbamazepine and phenytoin exposure (114).

The main anomaly of the fingers seen in this population is the hypoplasia of the distal phalanges or nails (49). The digital anomalies and transverse palmar creases may be more common with phenytoin exposure (49; 03).

Fetal hydantoin exposure and syndrome

Current findings. In studies major congenital malformation rates with phenytoin are highly variable, from 2.4% to 6.7%, and based on few exposures in most registers (159). Cardiac malformations and cleft palate are the most common major congenital malformation attributed to phenytoin (156).

Previous findings. The initial descriptions of the phenytoin teratogenicity in the humans were published in the 1960s and early 1970s (131; 110; 69; 97). In 1966, Massey reported the teratogenic effect of phenytoin in mice (101). Many of those early reports (69; 97; 60) were clouded by the in utero exposure to other antiseizure drugs.

Hanson and Smith observed a combination of craniofacial and digital dysmorphic features associated with growth deficiency, microcephaly, and intellectual disability in phenytoin-exposed children of epileptic mothers and called it fetal hydantoin syndrome. Phenytoin was considered to be the most probable cause of these symptoms, and the risk of developing the full syndrome was estimated to be about 11% (59).

Nine craniofacial and digital features have been considered as typical of fetal hydantoin syndrome: epicanthus, hypertelorism, typical nose, long philtrum, abnormal ears, low hairline, nail hypoplasia, distal phalangeal hypoplasia, and three or more dermal arches (49). Epicanthus is considered present if there is a distinct vertical skin fold at the medial end of the palpebral fissure. Hypertelorism is scored by inspection to determine whether the eyes appear unusually wide set in relation to other facial structures. A typical nose is considered present if nares are anteverted and the nose is small or the nasal bridge is low. Philtrum length is assessed by clinical impression in relation to other structures. Ears are considered abnormal if the auricles are either dysmorphic, low, or posteriorly rotated. Low hairline is considered present if growth of hair on the neck extends lower than usual. Nails that are smaller or more brittle than normal and distal phalanges that are either unusually short or tapering are considered hypoplastic. Of the nine typical features investigated by Gaily and colleagues, only hypertelorism, nail hypoplasia, distal phalangeal hypoplasia, and dermal arches in the offspring were associated with maternal phenytoin treatment (49). Other cranio-facial anomalies include wide anterior fontanelle, ocular hypertelorism, flat nasal bridge, short nose, and bowed upper lip (79).

Though digital and nail hypoplasia is often described as being typical of the fetal hydantoin syndrome (69; 59; 49; 16), this pattern is also seen with exposure to phenobarbital (147), primidone (145), and carbamazepine (80). In a study of the offspring of 98 women who took phenytoin during the pregnancy, only 30% were found to have exclusively digital hypoplasia (83).

The hypoplastic nails may improve or resolve over time (69). Polydactyly without digital-nail hypoplasia has been reported in a patient exposed to phenytoin in utero (178). A case report of fetal hydantoin syndrome with unilateral atypical cleft hand was indicative for a vascular disruption sequence (34).

Cleft lip and palate have been linked with the gestational exposure to phenytoin since early on (69). In a Norwegian study, the incidence of cleft lip and palate in infants of mothers with epilepsy declined in direct proportion with the decrease of phenobarbital and phenytoin use (84).

Infants exposed to phenytoin in utero may have delayed psychomotor development or mental retardation (69). Mental retardation is often mild, or the intelligence may be borderline (79). The association between mental retardation and in utero phenytoin exposure has been questioned. In a study of 103 infants of mothers with epilepsy exposed to phenytoin during pregnancy, only 1.4% had this problem, a frequency similar to that reported in the general population (49). Learning disabilities may be present in children who are not frankly retarded (170).

On these aspects of phenytoin teratogenicity see more reports below in cognitive effects of fetal exposure to antiseizure drugs. Failure to thrive has also been noticed in infants exposed to phenytoin in utero (69). The growth problems in these patients are prenatal and postnatal (60; 16).

Other malformations occasionally seen after the gestational exposures to phenytoin include hypospadias, congenital heart disease (tetralogy of Fallot, septal defects, coarctation of aorta, etc.), dislocation of the hip, dural epidermal cyst, pyloric stenosis, diaphragmatic and inguinal hernias, hirsutism, short neck, and colobomata (69; 109; 79). A rare malignancy associated with in utero phenytoin exposure is lymphoblastic lymphoma (119).

Fetal carbamazepine exposure and syndrome

Current findings. From studies, carbamazepine does not appear to be as teratogenic as previously believed. The overall rates of major congenital malformation found in current major epilepsy and pregnancy registries for monotherapy exposure to carbamazepine are: 0.8% to 1.57% for cardiovascular, 0.14% to 0.48% for orofacial clefts, 0.19% to 0.64% for hypospadias, and 0.2% to 0.36%for neural tube defects (159). These values are comparable to those with lamotrigine exposure (159; 134). In the latest report using data from The Health Improvement Network, the authors identified women who have given live birth and their offspring. Four subgroups were selected based on the antiseizure drug treatment in early pregnancy: valproate, carbamazepine, lamotrigine, and women not receiving antiseizure drug treatment. A total of 229 women were prescribed valproate in pregnancy, 357 were prescribed lamotrigine, 334 were prescribed carbamazepine, and 239,151 women were not prescribed antiseizure drugs. Fifteen out of 229 (6.6%) women prescribed valproate gave birth to a child with a major congenital malformation. The figures for lamotrigine, carbamazepine, and women not prescribed antiseizure drugs were 2.7%, 3.3%, and 2.2%, respectively. For women prescribed valproate in polytherapy, the prevalence was 4-fold higher (134).

According to an assessment, lamotrigine (2.31% in 4195 pregnancies) and levetiracetam (1.77% in 817 pregnancies) were associated with the lowest risk and valproate was associated with the highest risk (10.93% in 2565 pregnancies) compared with the offspring of women without epilepsy (2.51% in 2154 pregnancies) (13).

Previous findings. Maternal exposure to carbamazepine has been implicated in the genesis of spina bifida in infants of mothers with epilepsy. Spina bifida was mentioned in 8 of 20 published and unpublished studies of maternal carbamazepine exposures (140). The latter surveys found 11 cases of spina bifida in 1457 maternal exposures to carbamazepine (without valproate exposure), which implies an incidence of 1 in 132 cases. That is about 10 times the expected ratio in the general population. However, this was an overexaggeration because the estimation gave a rate of 0.2% to 0.36% for neural tube defects (159). The association between encephalocele and in utero carbamazepine exposure has been reported both alone and in combination with valproate exposure (140).

Jones and colleagues described a carbamazepine-related syndrome characterized by dysmorphic facial features, microcephaly, digital anomalies, and developmental delay (80). The facial dysmorphic features include up slanting palpebral fissures, epicanthal folds, short nose, and long philtrum. The study revealed that 20% of exposed children had developmental delay. These motor and cognitive deficits have been questioned (179; 182). In one of the newer reports the more commonly noted features with exposure to carbamazepine were full cheeks with a small chin and everted lower lip, giving a doll-like appearance (85).

Fetal barbiturate exposure and syndrome

Current findings. From reports, the overall rates of major congenital malformation found in current major epilepsy and pregnancy registries for monotherapy exposure to phenobarbital exposure are: 2.5% to 2.76% for cardiovascular, 0.46% to 2.0% for orofacial clefts, 0.46% to 0.97% for hypospadias, and 0.0% to 0.46%for neural tube (159).

Previous findings. Although phenobarbital had the reputation of being safe for use during pregnancy, Seip described children with pre- and postnatal growth retardation, developmental delay, dysmorphic facies, and minor abnormalities in association with phenobarbital use during the pregnancy (147). The facial anomalies included short nose, broad nasal bridge, hypertelorism, epicanthal folds, ptosis, low-set ears, and wide mouth with protruding lips. He recommended that a new fetal barbiturate syndrome not be created due to the similarities among the effects of phenobarbital, phenytoin, or alcohol exposures in utero. Phenobarbital, phenytoin, and alcohol can cause folate deficiency and are common mechanisms for these malformations (147).

Data are controversial as far as the true extent of the teratogenic effects of phenobarbital in humans. In animal studies, an increased frequency of cleft lip and palate, congenital heart disease, and other minor congenital anomalies were found in the offspring of pregnant mice or rats treated with phenobarbital (46). The extent that these studies can be compared to human situations is unknown because the phenobarbital dose used in mice was several times higher (in relation to body weight) than the one normally used in humans (48; 124; 30).

The frequency of congenital anomalies including major and minor malformations was not shown to be higher in 1415 women treated with phenobarbital (not necessarily for seizures) in the first 4 months of pregnancy in the National Collaborative Perinatal Project (65). Other studies have shown an increased frequency of congenital anomalies in infants born to mothers taking phenobarbital during the pregnancy for seizure disorders (148). Similar findings were seen in other series that reported an increased incidence of cleft palate and congenital heart disease in the fetuses born to epileptic women taking phenobarbital when compared to controls (55; 144; 121). Fetuses chronically exposed to phenobarbital in utero have been shown to have a 10% to 20% incidence of malformations, including serious problems such as congenital heart disease (151; 04). The incidence of fetal malformation may be increased in patients taking phenobarbital in combination with other antiseizure drugs (140). The exception for that rule is the lowering of the risk of spina bifida in patients exposed to valproate in combination with phenobarbital. This effect is not unexpected because phenobarbital has an accelerating effect on the catabolism of valproate, and lower antiseizure drug levels have been associated with a lower risk of fetal malformation (29; 140). One report mentioned the association of aortic stenosis with in utero exposure to high dose barbiturate (secobarbital and amobarbital) (139).

In the National Collaborative Perinatal Project, the intelligence quotient of children exposed and unexposed to phenobarbital in utero was compared (148). No statistically significant differences could be demonstrated in that study.

Phenobarbital exposure in utero has been associated with transient neonatal sedation and withdrawal symptoms in infants (35; 87). The withdrawal symptoms are similar to the ones seen with other addictive drug exposures in utero such as cocaine and opiates. The newborns tend to be irritable, have altered sleeping patterns, are difficult to feed, have increased startle reflex, and have hyperreflexia. The timing of the withdrawal symptoms appears to be different, with the mean onset around the seventh day and an average of 3-month duration (35). A similar syndrome, but with early onset of irritability in the first hours of life, was described after mephobarbital exposure in utero (69). The irritability continued until the fourth month of life.

Hemorrhagic disease of the newborn due to vitamin K deficiency has been described after phenobarbital exposure during the pregnancy (31; 69). Vitamin K should be administrated to all newborns exposed to barbiturates in utero (116).

Primidone exposure and embryopathy. Knowledge about the effects of primidone exposure in utero is based mostly on case reports. Primidone exposure in utero has been implicated in the development of specific facial dysmorphology. The latter includes increased facial hair (especially on the forehead), abnormal noses with anteverted nostrils, long philtrums, and thick nasal roots (145; 56). Cleft lip and palate have also been reported (140). Some of these patients appear to have features similar to Goldenhar syndrome (56). Patients with the primidone embryopathy pattern may also have cognitive and motor delays, retardation, and congenital heart disease. They are often small for gestational age at birth (145; 56). These reports also mention finger and nail abnormalities similar to ones seen with phenytoin exposure. Congenital heart disease has also been reported in cases of primidone exposure, including aortic coarctation and dextrocardia (140). Other rare associations include anencephaly and oculo-auriculo-vertebral anomaly (140). Two cases of spina bifida have been reported, one after primidone monotherapy exposure and the other in polytherapy with valproate and carbamazepine (140). Hemorrhagic disease of the newborn due to vitamin K deficiency has been described after primidone exposure in utero. Vitamin K should be administrated to all newborns exposed to primidone in utero.

Fetal trimethadione syndrome (fetal dione syndrome). This syndrome is now rare due to the decreased use of oxazolidine derivatives such as trimethadione and paramethadione. German and collaborators reported a case of an epileptic woman treated with trimethadione who had four unsuccessful pregnancies followed by the delivery of two healthy infants after she stopped taking the drug (52). Zackai and colleagues described a clinical phenotype of the fetal trimethadione syndrome (183).

This syndrome consists of intrauterine growth retardation, microcephaly, facial malformations, congenital heart disease, and genitourinary anomalies. Facial malformations include v-shaped eyebrows, epicanthal folds, low-set ears, and irregular teeth. Central nervous system malformations may also be seen. Mental retardation is present in up to 29% of the patients, speech delay in 47%, and developmental delay in 53% (143).

Congenital heart diseases include hypoplastic left heart syndrome, tetralogy of Fallot, and transposition of the great vessels (143). Malformations can be seen in the kidneys, trachea, larynx, and esophagus. The high rate of malformations, many of them severe, has been associated with a high neonatal mortality rate that may go close to 40%. In a study of 53 pregnancies in which trimethadione was used, 46 cases of fetal loss or congenital anomalies were recorded (42).

One follow-up study of patients who were exposed to trimethadione in utero demonstrated a 29% rate of mental retardation, microcephaly in 50% of the cases, and short stature in 53% (143).

Fetal valproate syndrome. Valproate is the most teratogenic of all other antiseizure drugs. It is definitely associated with an elevated risk for major congenital malformation (Tables 1 and 2), including a 10­fold increase in spina bifida aperta (1%-2% of infants exposed). The overall rates of major congenital malformations found in current major epilepsy and pregnancy registries for monotherapy exposure to valproate are: 1.1% to 2.18% for cardiovascular, 0.4% to 1.2% for orofacial clefts, 1.2% to 3.1% for hypospadias, and 1.09% to 1.2% for neural tube (159). The risk is dose­related, particularly at doses of more than 1000 mg/day. Polytherapy with valproate and any other antiseizure drug is highly teratogenic and appears to be even worse in combination with lamotrigine (1 of 10 infants exposed) (26; 117). In regard to minor congenital abnormalities, in one of the newer reports facial features often noted in children exposed to valproate were medial deficiency of eyebrows, infraorbital grooves, broad nasal bridge, anteverted nose, abnormal philtrum, and a thin upper lip. Trigonocephaly (triangular shaped skull) was not a common finding (85). Considering that valproate is possibly also associated with cognitive impairment of infants exposed to this drug during pregnancy (104; 22; 105), it should be avoided in women with epilepsy of childbearing age.

Table 1. Major Congenital Malformation Rates After Exposure to Antiseizure Drug Monotherapy

AED	Range (%) 4.0%-4.7% 2.7%-3.5% 2.6%-2.8% 2.8%-4.8% 6.3%-8.8% 5.4%-6.8% 3.9%-4.1% 9.7%-9.8%
Carbamazepine Lamotrigine Levetiracetam Oxcarbazepine Phenobarbital Phenytoin Topiramate Valproate
Background risk for malformations: 2% to 3% (11)

Previous findings. The first reports of a teratogenic effect of valproic acid appeared in the early 1980s (27; 20; 54). Subsequently, Robert and Guiband described the association between valproate and neural tube defects (138). DiLiberti and colleagues consolidated the idea of the specific constellation of malformations of the fetal valproate syndrome (36). The malformations considered to be characteristic of this syndrome were the typical facial dysmorphism and neural tube defects (07; 79). These malformations appear to be dose related because mothers using more than 1000 mg of valproate per day were more likely to have malformed infants than the ones taking 600 mg or less per day (146). Pre-pregnancy dose reduction resulted in fewer infants with malformations, and cessation prior to pregnancy lowered malformation prevalence to population baseline in an Australian registry of 580 pregnancies (162). The frequency of facial and minor malformations after valproate exposure in utero has been estimated to be between 50% to 75% (07; 91).

The neural tube defects that are often referred to loosely as spina bifida are probably the most conspicuous malformation associated with valproate exposure in utero. Ardinger and colleagues found both lumbosacral meningomyelocele and talipes equinovarus (with intact spine) to be typical of the fetal valproate syndrome (06). Neural tube defects have also been reported in patients exposed to carbamazepine in utero (140), and two patients (patient 9 and patient 16) with lumbosacral meningomyeloceles were also exposed to carbamazepine (06). The prevalence of neural tube defects is 10 to 20 times higher in fetuses exposed to valproate in utero when compared with the general population (125). The risk of neural tube defect after in utero exposure to valproate has been estimated to be 1% to 5% (07).

The facial appearance of these patients is rather typical, with epicanthal folds forming a crease under the orbits, telecanthus, and broad and low nasal bridge with short nose and anteverted nostrils (36; 06; 07; 79). Long philtrum, thin vermilion border, mid-face hypoplasia, and small mouth (microstomia) are often also present in the syndrome (36; 06). Cleft lip and palate have also been reported (06; 79; 91) but are infrequent, occurring in 1 out of 19 patients (06). Ardinger and colleagues found the outer orbital ridge deficiency and bifrontal narrowing to be typical of the facial appearance of patients with fetal valproate syndrome.

The fingers and toes of fetal patients with valproate syndrome are thin and overlapping, and the nails are hyperconvex and may be also hypoplastic (36; 07). Postaxial polydactyly, triphalangeal thumbs, thumb aplasia, syndactyly, and radial aplasia may also be seen (79; 91).

Mental retardation is usually mild to moderate (91). Cognitive deficits with or without other neurologic abnormalities are present in 71% of the monotherapy in utero exposures and in 90% of the polytherapy exposures (06).

Ardinger and colleagues evaluated 19 children exposed to valproate in utero in an attempt to verify the phenotype of the fetal valproate syndrome (06). In spite of the total absence of prenatal or postnatal growth deficiency in the children exposed to valproate monotherapy in utero, two thirds of the children also exposed to other antiseizure drugs had postnatal growth deficiency and microcephaly.

Signs of newborn encephalopathy appear to be more frequent than expected in infants of mothers taking valproate for seizures (74). This problem has been described with low Apgar scores (28% of the cases) and perinatal distress that are often followed by postnatal growth problems and decreased head circumference. Jeavons reports that 19.8% of the deliveries were abnormal after valproate exposure in utero, but no evidence of a dose response effect exists (77). Jeavons and Jager-Roman and colleagues data should be interpreted with caution due to the difficulties in diagnosing perinatal depression distress or newborn encephalopathy after in utero exposure to valproate. These reports (77; 74) may in fact reflect, at least in part, the sedative and teratogenic effect of valproate in the fetus. This view is supported by the almost universal hyperammonemia and sedation seen when valproate is used in the treatment of neonatal seizures (50).

High-dose valproate in utero has been associated with drug withdrawal and hypotonia in the neonatal period and subsequent motor and language delay (74).

Cardiovascular defects include aortic coarctation or valvular stenosis, hypoplastic left heart, atrial or ventricular septal defect, patent ductus arteriosus, and tetralogy of Fallot (06; 07; 79).

Other less frequently associated malformations include growth failure, hypospadias, cryptorchidism, bifid rib, broad chest, inguinal and umbilical hernia, supernumerary nipples, and lung hypoplasia (79; 91; 75). Studies have reported associations with craniosynostosis, autism, and ophthalmologic findings including dry eyes, myopia, and newborn nuchal edema (92; 175; 53; 177). In addition, three families have been described in whom the occurrence of fetal valproate syndrome occurred in all the siblings, strongly suggesting hereditary susceptibility to valproate-induced outcome (98).

In a study from Italy, no significant differences were observed in the risk of congenital malformations between 526 cases exposed to antiseizure drugs and 3682 controls, except for valproic acid (odds ratio (OR): 2.29; 95% confidence interval (CI): 1.24-4.22), where cases were more likely to be small for gestational age (chi(2)=7.66; p=0.006) (136).

Cognitive effects of fetal exposure to antiseizure drugs. There is significant uncertainty about the effect of epilepsy and antiseizure drug therapy on cognition in children born to mothers with epilepsy (08; 09). Results from the Maternal Outcomes and Neurodevelopmental Effects of Antiepileptic Drugs (MONEAD) study of 351 women with epilepsy and 105 without epilepsy demonstrated no difference in neurodevelopmental outcomes at 3 years of age in children with fetal exposure to newer antiseizure medications (lamotrigine and levetiracetam) (106). In contrast, a Nordic registrar study of antiepileptic drugs in pregnancy (SCAN-AED) found an increased risk of neurodevelopmental disorders, such as autism and intellectual disability, in children exposed to topiramate, valproic, and several duotherapies in utero (10). This finding supports prior studies demonstrating the relative safety of newer antiseizure medications.

In the AAN/AES practice parameter, the outcome measure was an assessment of the childs intelligence quotient (IQ) at age 2 years or older (62). Because maternal IQ is an important influence on child IQ, studies were downgraded if they did not control for maternal IQ. It was also assumed that the cognitive risk was related to antiseizure drug exposure throughout pregnancy and was not confined to the first trimester. The main conclusions of this AAN/AES report are:

(1) Cognition is probably not reduced in children of women with epilepsy who are not exposed to antiseizure drugs in utero.

(2) There is mounting evidence that children of women with epilepsy taking older antiseizure drugs are at increased risk for reduced cognition, but not children of women with epilepsy taking newer drugs.

	• Cognitive outcomes are probably reduced in children exposed to antiseizure drug polytherapy compared with monotherapy in utero.
	• Carbamazepine probably does not increase poor cognitive outcomes compared with unexposed controls.
	• Valproate is probably associated with poor cognitive outcomes compared with unexposed controls.
	• Phenytoin is possibly associated with poor cognitive outcomes compared with unexposed controls.
	• Phenobarbital is possibly associated with poor cognitive outcomes in the male offspring of women with epilepsy compared with unexposed controls.
	• Cognitive outcomes are probably reduced in children exposed to valproate during pregnancy compared with carbamazepine and possibly phenytoin.

It is difficult to assess how severely cognition is impaired in these children, but the differences, though significant, may not be extreme. According to a prospective, observational, multicenter study in the United States and UK, children with intrauterine exposure to high doses of valproate may suffer cognitive impairment (104; 105). The interim report detailed the cognitive outcomes in 309 children at 3 years of age (104). Contrary to the conclusions of the AAN/AES practice parameters, there were no significant differences between children exposed to phenytoin (IQ 98), carbamazepine (IQ 99), or lamotrigine (IQ 101). However, valproate was associated with a significantly lower IQ (92). The association between valproate use and IQ was dose dependent. Childrens IQs were significantly related to maternal IQs among children exposed to carbamazepine, lamotrigine, or phenytoin, but not among those exposed to valproate. A subsequent report of follow-up at age 6 years confirmed that children exposed to valproate did poorly on measures of verbal and memory abilities compared with those exposed to the other antiseizure drugs and on nonverbal and executive functions compared with lamotrigine (but not carbamazepine or phenytoin) (105). High doses of valproate were negatively associated with IQ, verbal ability, nonverbal ability, memory, and executive function, but other antiseizure drugs were not. Age 6 IQ correlated with IQs at younger ages, and IQ improved with age for infants exposed to any antiseizure drug. Mean IQs were higher in children exposed to periconceptional folate (108, 95% CI: 106-111) than they were in unexposed children (101, 98-104; p = 0.0009).

Furthermore, children of mothers who took valproate during their pregnancy were at a significantly greater risk for a diagnosis of attention deficit hyperactivity disorder (ADHD), difficulty with adaptive functioning, and atypical behaviors than those in the lamotrigine and phenytoin groups (22).

A relevant Cochrane review on neurodevelopmental outcomes in the child of women with epilepsy has been published (12). The most important finding is the reduction in IQ in the valproate exposed group, which is sufficient to affect education and occupational outcomes in later life.

Velez-Ruiz and Meador present a comprehensive review on the neurodevelopmental effects of fetal antiseizure drug exposure (167).

A systematic review and network meta-analysis assessed cognitive development and autism or dyspraxia in children exposed to antiseizure drugs during pregnancy and breastfeeding (168). Among all antiseizure drugs, only valproate was statistically significantly associated with more children experiencing cognitive developmental delay compared with control (OR=7.40, 95% credible interval (CrI) 3.00 to 18.46). Furthermore, compared with control, only valproate (OR 17.29, 95% CrI 2.40 to 217.60), oxcarbazepine (OR 13.51, 95% CrI 1.28 to 221.40), lamotrigine (OR 8.88, 95% CrI 1.28 to 112.00), and lamotrigine combined with valproate (OR 132.70, 95% CrI 7.41 to 3851.00) were significantly associated with increased occurrence of autism or dyspraxia. For the cognitive developmental delay and psychomotor developmental delay outcomes, children exposed to the combination of carbamazepine, phenobarbital, and valproate were at greater odds of harm than those who were not exposed to antiseizure drugs (168).

Childhood- and adolescent-onset psychiatric disorders in children with prenatal exposure to antiseizure medications were explored in a large cohort study of more than 38,000 children born during 1996 to 2017 (38). The study demonstrated increased risk with valproic acid, mostly driven by neurodevelopmental disorders, and no increased risk with exposure to lamotrigine, carbamazepine, or oxcarbazepine. There was an increased risk of ADHD with topiramate exposure and of both anxiety and ADHD with levetiracetam exposure.

Table 2. Major Congenital Malformations in Two Large Pregnancy Registries

	Number of women with epilepsy		MCM (%)		95% CI
	UK Epilepsy Pregnancy Register	North American Epilepsy Pregnancy Register	UK Epilepsy Pregnancy Register	North American Epilepsy Pregnancy Register	UK Epilepsy Pregnancy Register	North American Epilepsy Pregnancy Register
No exposure to AEDS	445	442	2.2	1.1	1.2-4.1	0.37-2.6
Exposure to AEDS	5475		3.9		3.4-7.2
Monotherapy	4276		3.4		2.9-4.9
Monotherapy with valproate	1290	323	6.7	9.3	5.5-8.3	6.4-13.0
Monotherapy with carbamazepine	1718	1033	3	2.6	1.9-3.5	2.1-4.2
Monotherapy with lamotrigine	2198	1562	2.3	2	1.8-3.1	1.4-2.8
Monotherapy with levetiracetam	367	450	0.7	2.4	0.2-2.5	1.2-4.3
Monotherapy with topiramate	359	197	4.8	4.2	1.9-7.6	2.4-6.8
Polytherapy (> 130 AED combinations)	1199		5.8		4.6-7.2
Polytherapy with valproate	451		8.6		6.4-11.6
Polytherapy with carbamazepine	526		4.9		3.4-7.1
Polytherapy with lamotrigine	644		5.3		3.8-7.3
Polytherapy with levetiracetam	367		5.6		3.5-8.6
Polytherapy with topiramate	162		8.6		5.2-14.0
(117b; 72; 67; 102; 17) Note: For monotherapy data from the North American antiseizure drug register please see The North American Antiepileptic Drug Pregnancy Registry.

Commonly used new generation of broad spectrum antiseizure drugs and major congenital malformations. Based on existing evidence, the risks for major congenital malformations associated with three of the more widely used broad spectrum antiseizure drugs lamotrigine, levetiracetam, and topiramate (Table 2) are as follows:

Lamotrigine. The overall rates of major congenital malformations found in current major epilepsy and pregnancy registries for monotherapy exposure to lamotrigine are: 0.19% to 0.63% for cardiovascular, 0.1% to 0.45% for orofacial clefts, 0.0% to 0.5% for hypospadias, and 0.0% to 0.17% for neural tube (159).

Significant differences in the risk for major congenital malformations have been reported by the UK and the North American pregnancy registries, though the rate of major congenital malformations was similar (2.3%-2.4%). The UK Epilepsy and Pregnancy Registry noted a positive dose­response relationship for major congenital malformations with lamotrigine (p = 0.006) with a major congenital malformation rate of 5.4% (95% CI 3.3%-8.7%) for total daily doses of more than 200 mg (117). However, in the report A non-significant trend towards higher MCM rate with increasing dose was found with lamotrigine. MCM rate for high-dose lamotrigine (> 400 mg daily) was lower than the MCM rate for pregnancies exposed to less than 600 mg daily of valproate, but this was not significant (3.4% vs. 5.0%, p = 0.31) (17).

The North American antiseizure drug pregnancy registry reported a 10.4­fold increase (95% CI 4.3-24.9) in isolated cleft palate or cleft lip deformity (70), but this has not been confirmed in the UK (73) and the European Surveillance of Congenital Anomalies (EUROCAT) registries (37).

The initial report of the international lamotrigine pregnancy registry concluded that the risk of all major birth defects after first trimester exposure to lamotrigine monotherapy (2.9%) was similar to that in the general population (26), but this was considered to be an overestimate by other authorities, who suggested that a more accurate conclusion may be the risk for major malformations associated with first trimester exposure to lamotrigine is only about twice that of the general population... when similar definitions, inclusions and case identification strategies are used (64). Furthermore, it would be even more difficult to assess the risks of major congenital malformations with lamotrigine precisely without therapeutic drug monitoring if the finding of a dose­related effect is replicated because of the significant decrease in plasma levels of lamotrigine (nearly by 50%) in the first trimester of pregnancy.

In polytherapy with lamotrigine and other antiseizure drugs, the risk of major congenital malformations is 5.6% (Table 2), and this doubles to around 10.7% in combination with valproate.

Alarmingly, lamotrigine plus valproate is the most frequently used antiseizure drug combination in pregnancy according to a EURAP report (41), despite the fact that it may harm 1 in 10 exposed babies. This is twice the frequency seen with any other combination and indicates a lack of information reaching those prescribing antiseizure drugs for women; the messages either do not get through or are unclear.

The conclusions of the Lamotrigine Pregnancy Registry Final Report, Issue Date: July 2010 are:

Lamotrigine monotherapy. There were 35 outcomes with major defects among 1558 outcomes (2.2%) involving a first trimester monotherapy exposure (95% CI: 1.6%-3.1%). There were four outcomes with a major defect among 95 outcomes following a second trimester monotherapy exposure and one outcome with a major defect among the 18 outcomes following a third trimester monotherapy exposure.

Polytherapy including valproate. There were 16 outcomes with major defects among 150 total outcomes (10.7%) involving first trimester exposure to lamotrigine and valproate, with or without one or more additional antiseizure drugs (95% CI: 6.4%-17.0%). There was one outcome with a major defect among the seven outcomes following a second trimester exposure to lamotrigine and valproate, with or without one or more additional antiseizure drugs. This exposure group had the highest proportion with major defects observed among first trimester exposures in the registry.

Polytherapy not including valproate. There were 12 outcomes with major birth defects among 430 total outcomes (2.8%) involving first trimester exposure to lamotrigine and at least one other antiseizure drug, excluding valproate (95% CI: 1.5%-5.0%). There was one outcome with a major birth defect among the three total outcomes involving third trimester exposure to lamotrigine and at least one other antiseizure drug, excluding valproate. There was no consistent pattern among the major birth defects reported prospectively to the registry.

Because Morrow and colleagues noted a positive dose-response effect for major congenital malformations with lamotrigine use, the Lamotrigine Pregnancy Registry Advisory Committee continuously examined the registry data related to dose and included the data in this report (117). The committee considered the data as reassuring, providing no evidence of a dose effect. The available data are insufficient to make a definitive conclusion, but they do suggest that any dose effect that might exist is likely to be small. The report can be accessed at: PregnancyRegistry.gsk.com.

The EURAP epilepsy and pregnancy registry found an increase in malformation rates with increasing dose at the time of conception for carbamazepine, lamotrigine, valproate, and phenobarbital (157).

Levetiracetam. The number of women treated with levetiracetam in pregnancy registries is still relatively small, but the data so far are encouraging (Table 2), with the major congenital malformations ranging from 0.7% to 2.4% for levetiracetam monotherapy (67; 102).

The overall rates of major congenital malformations found in current major epilepsy and pregnancy registries for monotherapy exposure to levetiracetam are: 0.0% to 0.22% for cardiovascular, 0.0% to 0.0% for orofacial clefts, 0.0% to 0.0% for hypospadias, and 0.0% to 0.22%for neural tube (159).

Topiramate. There is a growing body of evidence that this drug should not be given to women of childbearing age unless absolutely necessary (160). This is because topiramate (a) is teratogenic in animals, even at subtoxic doses (equivalent to 0.2 to 10 times the therapeutic doses recommended in humans) (130), (b) has serious adverse drug reactions that are likely to affect the fetus because of the extensive transplacental transfer of the drug, and (c) relevant results from pregnancy registries indicate that such precautions are justified (see Table 2) with an increased rate of major congenital malformations between 4.2% to 4.8%. In the UK Epilepsy and Pregnancy Register (Table 2), the major congenital malformations rate in 79 women receiving topiramate monotherapy was 3.8% (95% CI: 1.3-10.6) and in 162 women receiving polytherapy was 8.6% (95% CI: 5.2-14) (72). The rate of major congenital malformations was even higher (9.8%) in a report of 41 live births from 52 pregnancies during topiramate treatment (29 on monotherapy and 23 on polytherapy) (126); this finding has, however, been undermined by statistical deficiencies (44). Veiby and associates found that infants exposed to topiramate had a considerable risk of microcephaly and small for gestational age birth weight (166). Furthermore, topiramate malformation rates are higher in polytherapy pregnancies involving topiramate even when valproate is not used (163; 160).

Oxcarbazepine. There is practically no evidence of the teratogenic potential with oxcarbazepine, though preliminary results indicate that this may not be significant (25). However, oxcarbazepine is associated with an increase in seizure frequency during pregnancy, probably because of the significant 26% to 38% decrease in levels of its active monohydroxy derivative (135).

Teratogenesis of the newer antiseizure drugs. Data concerning the risk for congenital malformations associated with other newer antiseizure drugs (levetiracetam, gabapentin, felbamate, oxcarbazepine, tiagabine, topiramate, zonisamide, and vigabatrin) are still limited. In a review of the teratogenic effects of the newer antiseizure drugs in animal studies, felbamate, gabapentin, lamotrigine, oxcarbazepine, topiramate, tiagabine, and vigabatrin can cause intrauterine growth retardation (115). Felbamate, gabapentin, lamotrigine, and topiramate cause skeletal malformations, but only high doses of topiramate can cause limb agenesis. Vigabatrin can cause cleft lip and palate in animals. Teratogenic effects of vigabatrin were found in mouse fetuses; the exencephaly, mandibular and maxillary hypoplasia, arched palate, cleft palate, limb defects, omphalocele, and exomphalos were observed in the fetuses exposed to vigabatrin in utero (01). It is also possible that clobazam may be teratogenic, as indicated from very limited preliminary data on MCM rate for nine clobazam monotherapy women (22.2%; 95% CI 6.2-54.7) (154).

Gabapentin. A large postmarketing surveillance study of 3100 English patients receiving the drug identified only 11 pregnancies without any congenital anomalies (176). In another study of 39 women and 51 pregnancies receiving gabapentin, the rate of major malformations was 4.5%. Eighty-seven percent of the pregnancies resulted in live births, 11.3% in miscarriages, and 2% in therapeutic abortions (113).

Zonisamide. The largest reported cohort of pregnant women exposed to zonisamide was 112, with major congenital malformations in 3 out of 26 (12%) on monotherapy and 5 out of 86 (6%) with polytherapy exposure (103). A different registry of 98 live births exposed to prenatal zonisamide monotherapy found 0 out of 98 major congenital malformations (66).

Prognosis and complications

The true prognosis of fetal anticonvulsant syndrome depends greatly on the degree of involvement and the type of malformation seen (16; 140; 140). Major congenital malformations and particularly some neural tube and cardiac defects are the most serious. Patients with neural tube defects may develop hydrocephalus and tethered cord as complications; both require neurosurgical treatment.

Fetal trimethadione syndrome has a bad prognosis due to the high rate of serious malformations and neonatal death (143). The same is true for the fetal valproate syndrome due to its association with neural tube defects.

Individuals with fetal valproate exposure are now having their own families. A survey of 23 men and 85 women with complications due to fetal valproate exposure found that in their 187 cumulative children there were 43 (23%) with malformations and 82% (44%) with neurodevelopmental disorders, suggesting possible epigenetic inheritance (99). Further rigorous studies are required.

Clinical vignette

A neurologic consultation was requested for a 7-day-old infant who had neonatal seizures. The patient was born to a 34-year-old mother with a history of generalized tonic-clonic seizures since adolescence that had been treated with phenobarbital 90 mg per day. The treatment allowed the mother to be seizure-free, although she occasionally complained of feeling sad. The pregnancy was otherwise uncomplicated, and the mother was seizure-free through the gestation. A male baby was delivered without problems after a 39-week gestation; the birth weight was 2200 gm (small for gestational age), and the Apgar scores were 7 (1 minute) and 8 (5 minutes). Difficulty feeding and sedation were noticed from day 1. The feeding difficulties improved slowly, but formula and breast milk had to be given through gavage. A pediatric neurology consultation was requested on the seventh day of life due to poor feeding and spontaneous shaking movements involving the upper more than the lower extremities, lasting a few seconds, and occurring intermittently. The patients physical exam showed short nose, broad nasal bridge, hypertelorism, epicanthal folds, and wide mouth. A small palate cleft could be seen but no cleft lip was noted. The neurologic exam was remarkable for irritability alternating with sleepiness. The patient was almost never in the quiet alert state. Mild to moderate axial hypotonia was noticed. A few episodes of spontaneous fast (several cycles per second) shaking of the extremities were seen, which could be suppressed by restraining. Otherwise, the neurologic exam was normal for the patients chronological and gestational age.

A lumbar puncture was done, and the patient was started on ampicillin and gentamicin due to presumed sepsis. The cerebrospinal fluid was clear and had zero red cells and one white cell; protein and glucose were within the normal range for the patients age. An EEG was mildly abnormal due to dysmature features, but no electroencephalographic seizures were seen.

The hospital course was characterized by slow improvement of the patient irritability. The feeding improved with upright positioning and special nipple use for bottle-feeding. The patient was discharged on day 21 of life, feeding by mouth and gaining weight (10 to 20 g per day). The patient was referred to the cranio-facial clinic for management of the cleft palate. The patient was seen on follow-up (from 3 to 12 months old), and the neurologic exam was completely normal.

Biological basis

Etiology and pathogenesis

The reasons for the increased risks are multifactorial and may include genetic factors, maternal epilepsy and seizures during pregnancy, and socio-economic status, but accumulating data strongly suggest that antiseizure drugs are the main reason for the increased risk (62). Fetal anticonvulsant syndrome has been linked mainly to exposure to antiseizure drugs and effect independent of epilepsy (181). There are various proposed mechanisms of teratogenicity caused by antiseizure drugs. Oxidative stress causing release of free radicals, formation of toxic epoxide intermediates, altered folate metabolism, and histone deacetylase inhibition are thought to be some of the mechanisms, though enough evidence does not exist. In addition, not much is known about the mechanisms of teratogenicity of newer antiseizure drugs in particular.

In general, the biological basis of the fetal anticonvulsant syndrome remains largely speculative, although some evidence points to epoxide and free-radical toxicity playing a role in the teratogenesis associated with these drugs. Many drugs are primarily metabolized into epoxides (78). Among the epoxides, the arene oxides have been associated with a carcinogenic and mutagenic effect due to their electrophilic properties and covalent binding to macromolecules (78; 122; 150). One of the pathways of epoxide detoxification (epoxide hydrolase enzyme) is much less active in fetuses when compared to adults (128). Another contributing factor may be the fact that at least one third of the fetal circulation bypasses the liver, which is one of the main sites of drug detoxification (129).

Animal models suggest that antiseizure drug-induced apoptosis, altered neurotransmitter environment, and impaired synaptogenesis are some of the mechanisms responsible for cognitive and behavioral teratogenesis (167).

A study from Finland showed that signs of adverse effects of prenatal antiseizure drug exposure on brain function could already be detected during the first 2 weeks of life (169). The authors studied prospectively 56 full-term newborns with prenatal exposure to antiseizure drugs and 67 unexposed newborns for the following characteristics: background information, antiseizure drugs exposure data, pregnancy outcome, neuropsychological evaluation of the mothers, clinical neurologic status with Hammersmith Neonatal Neurological Examination, and early cortical activity using EEG. For EEG assessment, the authors developed and provided automated quantitation algorithms of several earlier described features: oscillatory bouts at theta and alpha frequencies, frequency spectra, interhemispheric synchrony, and interburst intervals. It was found that the antiseizure drug-exposed newborns had lower limb and axial tone and were less irritable than the unexposed newborns. EEG assessment disclosed significant differences in alpha bouts, in the frequency spectra, as well as in the spatial distributions of interhemispheric synchrony and interburst intervals. The authors concluded their results indicate that fetal antiseizure drug exposure may affect early neonatal neurologic status and several features of early cortical activity, and the results suggest that interference of activity-dependent network development may be a possible mechanism to explain the link from fetal antiseizure drug exposure to later neurocognitive sequelae (169).

Phenytoin forms a transdihydrodiol metabolite that may be detected in the human newborn exposed to that drug in utero (71). Phenytoin binds irreversibly in vitro to rat liver, and this binding is increased by inhibition of the epoxide hydrolase but decreased by glutathione (100). Spielberg and colleagues demonstrated that phenytoin toxicity in vitro is increased by inhibition of epoxide hydrolase (152).

Few human studies have been done that explore the role of epoxides in the fetal anticonvulsant syndrome. Strickler and colleagues incubated human lymphocytes and phenytoin with the microsomal system of mice (153). An increased death rate in the cells of the system was noticed when the lymphocytes came from children with major malformations. The lymphocytes from family members of these children often yield a similar response. Strickler and colleagues suspected that a low arene oxide detoxification capability was responsible for the increased risk of malformations after phenytoin exposure (153). One element that was missing in the latter study was the lack of measurement of the epoxide hydrolase activity in these patients. In 1985, Buehler found that a higher epoxide hydrolase activity correlated with an increased number of features of fetal hydantoin syndrome in a pair of dizygotic twins (14). In another study, the activity of the epoxide hydrolase was measured by thin-layer chromatography in amniotic cells of 100 pregnant women to establish normal levels; a trimodal distribution was found, suggesting that the enzyme is regulated by a single gene with two allelic forms (15). Fetuses homozygous for the recessive allele have low epoxide hydrolase activity and, therefore, would be at high risk if exposed to phenytoin or other epoxide producing drugs during gestation. In 19 cases of pregnant mothers receiving phenytoin monotherapy, in which the pregnancies were monitored by amniocentesis, the 15 mothers at low risk for fetal hydantoin syndrome on the basis of normal epoxide hydrolase activity delivered infants lacking the phenotype of the fetal phenytoin syndrome, but four fetuses considered at high risk on this basis all showed features of the syndrome at birth (15). These findings allow for some prenatal prediction of the risk of fetal malformation in mothers taking phenytoin. Even though all the evidence for peroxide teratogenicity above is impressive, ethotoin, mephenytoin, and trimethadione have been shown to have teratogenic effects but do not have arene oxide metabolites (179). Genetically determined patterns of detoxification and metabolization of anticonvulsants have an impact on teratogenic risk (94). Wells and colleagues reviewed the role of oxidative damage in chemical teratogenesis (171). Studies in pregnant mice that demonstrated sensitivity to phenytoin-induced malformations and that involved chronic phenytoin exposure have shown significantly altered expression of several genes at different times of morphogenesis; in particular, retinoic acid receptors alpha, beta, and gamma were significantly increased (51).

Many antiseizure drugs have folic acid antagonistic properties, and similar patterns of malformations can be seen after in utero exposure to folic acid antagonists (unsuccessful attempts to induce abortion) (110). Hill and colleagues noticed that some mothers taking phenytoin who delivered malformed fetuses had borderline or low folate levels (69). Although the utility of folic acid supplementation for the general female reproductive population is clearly established, whether or not it reduces teratogenic risk for women with epilepsy taking antiseizure drugs is unclear. There have been three case reports of women who took good doses (3.5 to 5 mg) of folic acid supplementation months before conception and throughout pregnancy who were also on valproate with resultant neural tube defects (1 with other malformations also) (24; 40). These cases are not that surprising as there has been demonstrated failure of folate to reduce valproate-induced neural tube defects and embryotoxicity both in vitro and in vivo rodent models (57; 58).

Other factors have been implicated in the genesis of fetal anticonvulsant syndrome, especially in association with phenytoin exposure. Acetylsalicylic acid inhibits prostaglandin synthase that oxidates phenytoin (90). The former reaction produces free radicals that may bind to proteins (90). Pretreatment of pregnant mice taking phenytoin with acetylsalicylic acid reduces the risk of cleft lip and palate in their offspring (172). The same authors found a similar effect with pretreatment with caffeic acid, an antioxidant agent.

A Swedish study in rats showed good correlation between pharmacokinetic data and pregnancy outcome data, ie, the degree of developmental toxicity was related to free maternal plasma phenytoin levels. The results suggest that the development of toxicity of phenytoin is caused by concentration-dependent induction of embryonic dysrhythmia and hypoxia-related damage during a restricted sensitive period (28).

Epidemiology

The reported incidence of major congenital malformations varies significantly by around 20­fold, mainly because of methodological differences and deficiencies. Earlier studies usually rely on small numbers of recruited patients and lack statistical power. To understand the extent of the difficulties, a total of 722 drug­exposed pregnancies is needed to identify a 7­fold increase in the rate of occurrence of a specific abnormality, such as spina bifida, with a frequency of 1 in 1000 (70), or if drug A has a 3% risk for major congenital malformations and drug B doubles the risk to 6%, then 750 patients on monotherapy are needed in each group to reach p < 0.05 at 80% power. Several large prospective pregnancy registries throughout the world are collecting data on antiseizure drug­related major congenital malformations and other pregnancy­ related outcomes in women with epilepsy (158). However, even these registries have important methodological differences in recruitment, ascertainment, inclusion/ exclusion criteria, malformation classification, and follow­up that may influence the results and prevent meaningful pooling of data (108).

Background rate of major congenital malformations is generally considered to be 1% to 2%, but reports vary. The Active Malformation Surveillance Program at Brigham and Women’s Hospital in Boston, United States, estimates the background rate to be 1.6% after exclusion of genetic and chromosomal anomalies (123). Some use the higher rate of 3.2% determined by the Metropolitan Atlanta Congenital Defects Program (23), but this population­based registry uses active case identification from multiple sources, undertakes direct chart review of potential cases, and includes all malformations identified up to the age of 5 years (64).

A pooled analysis of data from 26 studies reported a major congenital malformation rate of 6.1% in offspring that had been exposed to antiseizure drugs compared to 2.8% in children of untreated women with epilepsy, and 2.2% in offspring of mothers without epilepsy (09). Similar results were reported in a formal meta-analysis of 10 studies. The offspring of women with epilepsy who received antiseizure drugs had higher prevalence of major congenital malformation than controls (odds ratio (OR) 3.26; 95% CI 2.15-4.93), whereas the risk for major congenital malformations in the offspring of women with untreated epilepsy was not significantly higher than among nonepilepsy controls (OR 1.92; 95% CI 0.92-4.00) (45).

Previous studies. The rates of malformation in infants of mothers with epilepsy ranges from 2.3% (76) to 18.6% (112). The rate of malformations of the general population in various reports ranges from as low as 1.4% to as high as 50%. The most commonly accepted values are 1.5% to 3% (81; 82; 182). This wide range is caused by differences in the threshold for diagnosing minor facial or forefinger anomalies and the variability in methodology. In a review, 39 studies of malformation rates in the offspring of epileptic and control mothers were analyzed (182). The analysis demonstrated a combined estimate of malformation of 4% to 6% for an individual epileptic pregnancy. The same authors point out that most studies show that infants of mothers with epilepsy have about 2.5 to 10.5 times more malformations than the general control populations. A group studied the relative risks of particular major malformations with fetal exposure to antiseizure drugs. They found the relative risk for cardiovascular defects was 2.2, oral clefts 2.5, and urinary tract defects 2.5 (68). Fetal mortality is another risk for women with epilepsy. Reported rates of stillbirth (defined as greater than 20 weeks’ gestational age) vary between 1.3% and 14% compared with rates of 1.2% to 7.8% for women without epilepsy. Perinatal death rates are also up to two times higher for women with epilepsy (1.3% to 7.8%) compared to controls (1.0% to 3.9%) (180). Spontaneous abortions (less than 20 weeks’ gestation) may also occur more frequently although the figures vary considerably between different studies (132).

Several factors remain difficult to explain: women with epilepsy often will have a lower social economic status and decreased ability to hold high profile jobs, and they are more prone to accidents, falls, and serious injuries related to seizures. Epileptic patients are also less likely to be driving and may have more difficulty having access to prenatal care in areas where public transportation is not efficient. These factors may contribute in part to intrinsic differences in these populations, which are difficult to control. Genetic influences probably also play a role in the genesis of malformations in infants of mothers with epilepsy. The rate of malformations in the relatives of the affected children is higher than in the relatives of the ones without anomalies (30). Malformation rates appear to be higher in infants of untreated women with epilepsy than in those born of mothers without epilepsy (148; 88). A family history of spina bifida was present in 25% of the children who developed neural tube defects after antiseizure drug exposure in utero (95; 96). Spina bifida occulta is more common in the genetic epilepsies than in the symptomatic ones (86). Epilepsy is also more common among the first and second-degree relatives of patients with cleft lip and palate (39). Nonetheless, the authors still believe that malformations in infants of mothers with epilepsy are caused mostly by teratogenic effects of the antiseizure drugs (76; 151; 05; 120). When a child has fetal hydantoin syndrome, a 90% risk exists that the next sibling will be affected, in contrast to a 2% of risk if the first child is normal (165).

The rate of some specific malformations is especially increased in infants of mothers with epilepsy. For example, the rate of cleft lip and palate is 4.7 times higher when compared with the general population (47). The frequency of neural tube defects is 10 to 20 times higher in children exposed to valproate in utero (125) and about 10 times higher with carbamazepine exposure (140).

Most studies of cognitive outcomes of infants exposed to antiseizure drugs in utero report an increased risk of mental deficiency, affecting between 1.4% and 6% of children of women with epilepsy compared with 1% of controls (93; 89). Two studies from the United Kingdom involving a total of 350 children have shown that prenatal antiseizure drug exposure in the setting of maternal epilepsy is associated with developmental delay, particularly speech delay, and later childhood morbidity, in addition to congenital malformation (114; 32).

For more studies see above “cognitive effects of fetal exposure to antiseizure drugs”. The plasma concentrations of antiseizure drugs appear to be higher in the mothers of infants with epilepsy when compared to epileptic mothers who gave birth to normal children (29). A mother using more than 1000 mg of valproate per day is more likely to have malformed offspring than a mother taking 600 mg or less per day (146). The epileptic mother treated with antiseizure drug polytherapy is more likely to give birth to an infant with malformations when compared to the one treated with monotherapy (120).

The effect of maternal seizures in the incidence of malformation is unclear at this point. Annegers and colleagues and Nakane and associates found no impact of seizures in the risk of fetal malformations (05; 121). Lindhout and collaborators found a 3-fold increase in malformations in the fetuses exposed to seizures during the first trimester (95; 96).

Prevention

In many cases, the exposure to antiseizure drug medication is unavoidable during pregnancy. Consensus articles (33; 116), a practice parameter summary statement from the American Academy of Neurology (127; 63), and a review by Whelehan and Delanty (173) were published that outlined some guidelines for antiseizure drug use during pregnancy. The preconception management is the cornerstone for care of women with epilepsy who wish to become pregnant. Prepregnancy consultation provides an opportunity to review the clinical status of the woman, assess the risk of seizure recurrence during pregnancy and of fetal malformations and, thus, optimize therapy without undue risk to the mother or baby. The aim is, if possible, to treat with the least epileptogenic of the appropriate antiseizure drug, reduce the dosage, and switch some patients from polytherapy to monotherapy. Start with folic acid, provide detailed screening for fetal malformations in the third or fourth month of pregnancy, and monitor antiseizure drug blood levels for drug dosage corrections, particularly in the latter half of pregnancy and postpartum (155). Impact of planning of pregnancy in women with epilepsy is associated with good seizure control during pregnancy and less fetal exposure to antiseizure drugs (02).

Despite advances in the past decade, our knowledge of the teratogenic risks for most antiseizure drugs and the underlying mechanisms remain inadequate. Further, the long-term effects of antiseizure drugs in neonates and older children remain uncertain. The pace of progress is slow given the lifelong consequences of diminished developmental outcomes, exposing children unnecessarily to potential adverse effects. It is imperative that new approaches be employed to determine risks more expediently (107). Recommendations include a national reporting system for congenital malformations, federal funding of the North American AED Pregnancy Registry, routine meta-analyses of cohort studies to detect teratogenic signals, monitoring of antiseizure drugs prescription practices for women, routine preclinical testing of all new antiseizure drugs for neurodevelopmental effects, more specific Food and Drug Administration requirements to establish differential antiseizure drugs cognitive effects in children, and improved funding of basic and clinical research to fully delineate risks and underlying mechanisms for antiseizure drug-induced anatomical and behavioral teratogenesis (107).

Antiseizure drug-treated women with either a malformed fetus or a spontaneous abortion in their previous pregnancy had a statistically significant 2-fold to 3-fold increased risk of fetal malformation in their next pregnancy, compared with similarly treated women with normal offspring in their previous pregnancy (161). Thus, in assessing the hazard of an antiseizure drug-treated woman having a malformed fetus, it is important to know both the antiseizure drugs being taken and, if there had been a previous pregnancy, whether a fetal malformation or a spontaneous abortion occurred in it (161).

Differential diagnosis

Both the facial appearance and the history of antiseizure drug exposure during early pregnancy help make the diagnosis of fetal anticonvulsant syndrome. Some of the facial features are similar to the ones seen in fetal alcohol syndrome. Among the former are microcephaly, growth retardation, upturned nostrils, hypoplastic philtrum, and irritability (19). Micro- or retrognathia and hypoplastic maxilla are not common after fetal anticonvulsant exposure but are common in fetal alcohol syndrome (19). One of the features that tends to be helpful in supporting the diagnosis of fetal anticonvulsant syndrome is the hypoplasia of the nails and distal phalanges. Accurate physical diagnosis may be important because mothers who suffer from chemical dependency may not always volunteer the information about the drugs and alcohol during early pregnancy. On the other hand, phenytoin and other antiseizure drugs are commonly prescribed for patients with alcohol withdrawal seizures.

Diagnostic workup

The diagnosis of the fetal anticonvulsant syndromes, effects, and exposures is done by physical examination and observation of the patients.

Patients with suspected malformations of specific organs may require specialized radiological studies. When the neurologic examination of the lower extremities is abnormal, a spinal MRI is required to delineate neural tube defects. Patients exposed to antiseizure drugs in utero presenting with a heart murmur, cyanosis, and heart failure need thorough cardiological assessment including an echocardiogram.

Diagnostic criteria to establish a diagnosis of fetal valproate spectrum disorder were published by the European Reference Network for Congenital Malformations and Intellectual Disability in 2019 (21). The essential criteria include confirmed exposure, no other diagnosis to explain the phenotype, normal microarray-CGH, and fragile X studies, and other overlapping teratogenic disorders excluded. The specific malformations are listed as suggestive supporting features, such as classic facial dysmorphism.

Management

Counseling for women with epilepsy of childbearing potential surrounding pregnancy issues is of the utmost importance and should be done when antiseizure medications are prescribed and reviewed regularly. Physicians must be familiar with risks associated with antiseizure medication and should endeavor to minimize risks to a fetus while selecting best medications for epilepsy type (173).

No specific treatments exist for many of the antiseizure drug-specific effects other than the symptomatic effects. Obviously, detailed neurologic assessment is required. Physical, occupational, and speech therapy are indicated for patients with motor, cognitive, and language deficits. Patients with cleft lip and palate and head dysmorphisms (trigonocephaly) should undergo multidisciplinary evaluation and surgical treatment in centers familiarized with craniofacial pathology. Cardiac and other affected systems need to be evaluated and repaired accordingly. Uncommonly, newborns may have withdrawal symptoms, particularly if the mother was on large doses of barbiturates. This can be treated with phenobarbital, morphine, or other opiates.