Neurologic Disorders in Pregnancy

Valproate

Rates of MCMs with first-trimester exposure to valproate monotherapy range from 4.7% to 13.8%. In the two largest prospective cohorts from the United Kingdom and Ireland (1290 valproate exposures) and the European and International Registry of Antiepileptic Drugs in Pregnancy (EURAP, 1381 valproate exposures), the malformation rates were 6.7% and 10.3%, respectively.

In the European Surveillance of Congenital Anomalies (EUROCAT) database, a population-based database of 14 European countries, valproate exposure was associated with an increased risk of several specific defects. Compared with control pregnancies, those exposed to valproate monotherapy were at statistically significant increased risk for spina bifida (OR, 12.7), craniosynostosis (OR, 6.8), cleft palate (OR, 5.2), hypospadias (OR, 4.8), atrial septal defects (OR, 2.5), and polydactyly (OR, 2.2). These numbers, however, describe only relative risk and can be hard for a patient to understand. Tomson and Battino compiled the data of 22 prospective studies that reported on specific ASD-associated malformations and reported the absolute risks of neural tube defects (NTDs, 1.8%), cardiac malformations (1.7%), hypospadias (1.4%), and oral clefts (0.9%).

In addition to significantly increasing the risk of birth defects, valproate exposure during pregnancy has also been associated with cognitive and behavioral teratogenesis. Two prospective studies of children exposed to ASDs in utero have been published: the Neurodevelopmental Effects of Antiepileptic Drugs (NEAD) study and a study by the Liverpool and Manchester Neurodevelopmental Group (LMNG). Both recruited women with epilepsy in the first trimester of pregnancy and followed the development of their children until age 6. In contrast to many earlier studies of the cognitive effects of ASDs, both of these investigations controlled for several important confounding variables, including maternal intelligence quotient (IQ)—an important predictor of a child's cognitive performance. Of note, the two studies did overlap: 92 children from the LMNG study were also enrolled in the NEAD study. The NEAD study ultimately evaluated 224 children of age 6 years who had been exposed to carbamazepine, lamotrigine, phenytoin, or valproate monotherapy. The LMNG study assessed 198 6-year-old children born to women with epilepsy who took ASD monotherapy ( n = 143) or polytherapy ( n = 30) or no medication ( n = 25) during pregnancy and a control group of 210 children of the same age. In the NEAD study, exposure to valproate monotherapy was associated with a significant decrease in mean full-scale IQ (FSIQ) by 7 to 10 points compared with children exposed to carbamazepine, lamotrigine, or phenytoin. The LMNG found that exposure to first-trimester doses of valproate greater than 800 mg/day was associated with a significant decrease in FSIQ by 9.7 points when compared with a control group of children born to mothers without epilepsy. The mean FSIQ of children exposed to low valproate doses (≤800 mg/day) was also lower than that of controls, but this difference did not meet statistical significance. The low-dose group, did, however, have significantly lower verbal IQ scores and an increased need for educational intervention. The findings in LMNG and NEAD were bolstered by a large observational study in Denmark of 1865 sixth- to tenth-grade children, which demonstrated that children exposed to valproate in utero did worse on national exams than controls.

Another population study that utilized the National Psychiatric Registry and birth registries in Denmark found that school-age children whose mothers were prescribed valproate monotherapy during pregnancy had a significantly increased risk of receiving a formal diagnosis of autism or autism spectrum disorder. In the valproate-exposed cohort, the absolute risk of autism was 2.5%, whereas the rate in the general population was 0.48%, and the risk of autism spectrum disorder was 4.42%, with a baseline risk of autism spectrum disorder of 1.53%. The rates of autism and autism spectrum disorder in children born to mothers with epilepsy who did not take valproate during pregnancy did not differ from baseline population rates.

Carbamazepine

In its 2009 guidelines, the American Academy of Neurology (AAN) stated, “Carbamazepine probably does not substantially increase the risk of MCMs in the offspring of women with epilepsy.” This conclusion was based on one class I study from the United Kingdom and the Ireland Pregnancy Registry that did not find a difference between the rate of malformations in carbamazepine-exposed pregnancies and those of an internal control group. At the time, carbamazepine was the only medication that the authors believed had strong enough evidence to support this conclusion. Across seven pregnancy registries, rates of major malformations in pregnancies exposed to carbamazepine monotherapy have ranged from 2.6% to 5.5%. The two largest studies, the United Kingdom and Ireland Pregnancy Registries ( n = 1657) and the EURAP registry ( n = 1402) reported rates of 2.6% and 5.5%, respectively. Of note, the two registries that reported higher rates of major anomalies with carbamazepine exposure—the Australian and EURAP registries—both follow the exposed infants to 1 year and beyond, whereas the other registries performed the last assessment for malformations at birth or 3 months. In the EURAP registry, malformations that were most likely to be picked up between 2 and 12 months were cardiac, hip, and renal malformations. The rates of anomalies increased for several drugs at the later assessment, but rates with carbamazepine were most affected by the timing of the assessment.

In the EUROCAT database, carbamazepine exposure was specifically associated with an increased risk of NTDs compared with unexposed controls (OR, 2.6; 95% CI, 1.2 to 5.3). However, the risk of spina bifida with carbamazepine exposure was still significantly lower than the risk with valproate (OR, 0.2; 95% CI, 0.1 to 0.6) and was not different from the risk of exposure to other ASDs when valproic acid was excluded. The EUROCAT study did not find a specific association between carbamazepine exposure and other major malformations. In the compiled registry data prepared by Tomson and Battino, the absolute risks of certain anomalies with exposure to carbamazepine monotherapy were reported for NTDs (0.3%), cardiac malformations (0.8%), hypospadias (0.4%), and oral clefts (0.36%).

Early studies of carbamazepine's effect on cognitive development were conflicting, and many were limited by retrospective design or did not control for important confounders. A Cochrane review of prospective studies published prior to 2014 concluded that the reported effects of carbamazepine on developmental scores were largely accounted for by variability between studies and identified no clear risk of delayed development in infants and toddlers exposed to carbamazepine. This meta-analysis also reported no evident adverse effect of carbamazepine exposure on the IQ of school-age children. In the NEAD study, no specific effects of carbamazepine exposure on IQ were identified when this group of children was compared with the lamotrigine- and phenytoin-exposed cohorts at age 6. The recent LMNG study also found no difference in the adjusted mean IQ scores between the 6-year-old carbamazepine-exposed children and controls, but mean verbal IQ was 4.2 points lower in the carbamazepine-exposed children. In addition, the relative risk of having an IQ below 85 was significantly increased in the carbamazepine cohort. Both the NEAD and LMNG studies demonstrated that, compared with valproate exposure, prenatal carbamazepine exposure was less likely to be associated with adverse cognitive effects.

The LMNG found no increased risk for formally diagnosed NDDs at 6 years in the carbamazepine-exposed children when compared with controls. The large Danish population study by Christensen and colleagues also found no increased risk of autism or autism spectrum disorder in teenagers and children with prenatal carbamazepine exposure. A recent study of autistic traits from the Australian Registry also recruited mothers retrospectively from the prospectively identified cohort (63% enrollment). This study reported scores consistent with autism in one of 34 children exposed to carbamazepine and scores that raised “concern for autism” in another child based on a standardized assessment.

Lamotrigine

Rates of MCMs with lamotrigine exposure have been consistently low and range from 2% to 4.6%, across eight prospective registries. Initially, the North American AED Pregnancy Registry (NAAPR) reported a 10-fold increased risk in oral clefts with lamotrigine monotherapy exposure. However, with a larger sample size, this risk was reevaluated and reported as a fourfold increased risk (absolute risk with lamotrigine, 0.45%). Other registries, however, have reported much lower rates of clefting with lamotrigine exposure (0.1% to 0.25%), and a case-control study found no specific increased risk of oral clefting with lamotrigine. The absolute risks of clefting reported by Tomson and Battino's compiled registry data was 0.15%. The composite lamotrigine-associated risk of other specific malformations in this review was 0.6% for cardiac defects, 0.12% for NTDs, and 0.36% for hypospadias.

In two independent cohorts from the United Kingdom, developmental scores of infants prenatally exposed to lamotrigine did not differ from those of controls. In the LMNG cohort, at 6 years of age, the IQ scores of the lamotrigine-exposed children did not differ from those of controls. In addition, in the NEAD study, FSIQ scores in children exposed to lamotrigine were significantly higher than those of valproate-exposed children and did not differ from those of carbamazepine- or phenytoin-exposed children. However, in the NEAD study, both valproate and lamotrigine exposure were associated with decreased verbal IQ relative to nonverbal IQ. In a Norwegian population-based mail survey, parents of lamotrigine-exposed infants also reported impaired language functioning and an increase in autistic traits observed in their children. Language functioning was better among lamotrigine exposed children whose mothers also took periconceptual folic acid. In contrast to these parental observations, the LMNG found no increased risk of formally diagnosed NDDs in lamotrigine-exposed children, and the population study by Christensen and colleagues found no increased risk of autism or autism spectrum disorder.

Levetiracetam

The MCM rates associated with levetiracetam across eight registries range from 0% to 2.8%. A recent Cochrane review of three studies reported an MCM rate of 1.77% for levetiracetam across 817 pregnancies. Developmental effects of levetiracetam have been assessed in one study of 51 levetiracetam-exposed children recruited from pregnancies identified in the UK Epilepsy and Pregnancy Register (UKEPR). At 36 to 54 months, the developmental scores of the exposed children did not differ from those of controls but were better than a group exposed to valproate. A later study that recruited from the UKEPR (with some overlap in patients) conducted neuropsychologic testing at school age (mean age 6.5 years old) and showed no differences in FSIQ, verbal abilities, and nonverbal abilities in children exposed to levetiracetam monotherapy in utero compared with controls.

Phenytoin

Despite the fact that phenytoin is one of the oldest ASDs still in use, little certainty exists in regard to its teratogenic implications. In 1975, Hanson and Smith described a specific fetal hydantoin syndrome associated with in utero phenytoin exposure. They noted growth and performance delays and craniofacial abnormalities that included clefting and limb anomalies, including hypoplasia of nails and distal phalanges. They later reported that this was present in 11% of 35 exposed infants and that 31% of exposed infants had some aspects of the syndrome. Yet other studies have not substantiated this. In 1988, Gaily and colleagues reported no evidence of the hydantoin syndrome in 82 women exposed in utero to phenytoin. Some of the patients had hypertelorism and hypoplasia of the distal phalanges, but none had the full hydantoin syndrome. The true prevalence of this syndrome and contributing factors has not been established, and it has largely fallen out of current literature.

The more recent pregnancy registries have not focused on the description of syndromes and do not include many of the skeletal abnormalities included in the fetal hydantoin syndrome. Rates of MCMs associated with phenytoin in these registries have ranged from 2.4% to 6.7% across five registries, but individual cohorts are small. The largest cohort studied in the NAAPR published a major malformation rate of 2.9% among 416 phenytoin-exposed pregnancies. Tomson and Battino reported the rates of specific malformations with phenytoin exposure: 0.4% for cardiac malformations, 0% for NTDs, 0.2% for oral clefts, and 0.5% for hypospadias.

The cognitive implications of phenytoin exposure have only been evaluated in a few prospective studies. The 2014 Cochrane review found that the methodologies of these studies were too disparate to perform a meta-analysis. The review concluded that phenytoin exposure was associated with better developmental and cognitive outcomes than valproate exposure and that no discernable differences between phenytoin and carbamazepine exposure were present in terms of development and IQ. In the NEAD study, average FSIQ and verbal IQ scores of the phenytoin-exposed children were significantly higher than those of the valproate-exposed cohort and were not different from those of children exposed to carbamazepine or lamotrigine. Because the study did not include an unexposed control group, it is unknown whether the phenytoin group would differ from unexposed children. In terms of behavioral effects, Vinten and associates reported no difference between parentally assessed adaptive behaviors in the phenytoin-exposed Norwegian children when compared with unexposed controls born to mothers with epilepsy.

Phenobarbital

Phenobarbital is rarely used as a first-line ASD in developed countries given its adverse cognitive and metabolic side effects and the availability of alternative medications with fewer adverse effects. It is very difficult to wean patients from phenobarbital, however, and this process often leads to worsened seizure control. Thus unless pregnancy is planned well in advance, many women previously taking phenobarbital may remain on it. In the NAAPR, phenobarbital was associated with a risk of major malformations of 5.5% in 199 pregnancies, and cardiac malformations were the most frequent malformation reported. In a pooled analysis of 765 barbiturate-exposed pregnancies, Tomson and Battino reported a rate of 3.5% for cardiac malformations and a 1% risk or oral clefts. The absolute risk of NTDs and hypospadias in this analysis was 0.2% for each.

Retrospective studies of the effect of phenobarbital on cognitive and educational outcomes of exposed children have reported mixed results. The largest prospective study of phenobarbital and cognitive outcomes evaluated a cohort of 114 Danish men who had been exposed to phenobarbital in utero between 1959 and 1961. In this cohort, the most common indication for phenobarbital was pregnancy-related hypertension, and mothers with epilepsy were not evaluated. Thus the exposure to phenobarbital was likely shorter in duration than in the children of mothers with epilepsy and there may have been different exposures related to hypertension. The phenobarbital-exposed group had significantly lower IQ scores compared with controls, and men who had been exposed in the third trimester were most affected. In a more recent prospective study, neurologic testing of a cohort of Indian sixth graders recruited from Kerala Registry of Epilepsy and Pregnancy showed that phenobarbital exposed sixth graders had a lower IQ than those exposed to other ASD monotherapies. Unlike other previous studies, this study controlled for maternal IQ, an important predictor of child IQ.

Topiramate

Recently, concern has been raised that topiramate is a significant teratogen, although sample sizes of topiramate exposed cohorts are still small. The US Food and Drug Administration (FDA) reclassification of topiramate from pregnancy category C to category D is based on a specific association between exposure and oral clefts reported by the NAAPR. Several studies have corroborated this finding. A meta-analysis of six studies found that first-trimester topiramate exposure was associated with a sixfold increased risk of oral clefts compared with unexposed controls (OR 6.26; 95% CI 3.13 to 12.51). This corresponded to a summary incidence rate of 0.36% that is greater than the reported rate of 0.07% among infants not exposed to seizure medications. In terms of overall malformation rates, the risk of major congenital anomalies was 4.2% in 359 pregnancies exposed to topiramate monotherapy in the NAAPR and 3.9% in 152 monotherapy exposures in the EURAP registry. Similar to valproate, topiramate also appears to have a significant and dose-related risk of structural teratogenesis when used in polytherapy.

In addition to an increased risk of structural teratogenesis, topiramate has been associated with lower birthweights and an increased relative risk (2.4%) of being SGA when compared with lamotrigine-exposed infants. In this study from the NAAPR, the absolute risk of SGA with topiramate exposure was 18.5%. The risk of SGA was significantly greater with doses greater than 50 mg/day (19.9% vs. 8.5%) and with exposure during the third trimester (20.2% vs. 8.2%).

The neurodevelopmental outcomes in children exposed to topiramate in utero are still being clarified; neuropsychologic outcomes have only been published for two small cohorts. In the largest study from the UKEPR, 27 children exposed to topiramate performed similarly to a control group on blinded neuropsychologic testing. The control group consisted of children born to mothers with unmedicated epilepsy. The authors recommend interpreting these results with caution given the small sample size.

Other Antiseizure Drugs

A paucity of data is available to describe the teratogenic risks of other ASDs commonly used to treat epilepsy. The rate of MCMs with oxcarbazepine exposure among 393 prospective cases in the Danish birth registry was 2.8% and 3% in 333 cases within EURAP. Gabapentin was reported to have a malformation rate of 0.7% in a cohort of 145 exposed infants in the NAAPR. In another NAAPR study, no major anomalies were reported in a cohort of 98 pregnancies exposed to zonisamide monotherapy, but this was interpreted with caution given the small sample size. Similar to topiramate, zonisamide has been associated with lower birthweights and an increased risk of SGA. Studies of the effect of oxcarbazepine and zonisamide on cognitive and behavioral development are limited.

Little useful information is available on the effect of human in utero exposure to other ASDs, including benzodiazepines, eslicarbazepine, ethosuximide, ezogabine, felbamate, lacosamide, perampanel, pregabalin, rufinamide, and vigabatrin. The Kerala Registry suggested that clobazam exposure in utero may lead to a high rate of malformations, but this was based on finding of two malformations in a small sample size of nine exposures. The manufacturers of lacosamide caution that it is known to antagonize the collapsin response mediator protein 2, which is involved in axonal growth and neuronal differentiation, and appears to have adverse effects on brain development in rodents.

Effect of Antiseizure Drug Dose

The risk of major malformations has been shown to be dose related for several ASDs across different registries. In the EURAP registry, for example, valproate monotherapy was associated with a malformation risk of 6.3% with preconception doses of less than 650 mg/day and a risk of 25.2% with doses greater than 1450 mg/day. Exposure to typical doses used to treat epilepsy (650 mg to 1450 mg/day) were associated with a risk of 11.3%. Similar correlations between the risk of birth defects and preconception ASD dose were also noted for carbamazepine, lamotrigine, and phenobarbital. In addition, dose-related effects on cognitive and behavioral development have been noted for valproate, More data on the relationship between cognitive teratogenesis and dose of both valproate and other ASDs are needed. Further research is also required on the relevance of serum concentrations, instead of dose, because of the substantial differences in ASD metabolism among individuals. For now, preparing a woman with epilepsy for pregnancy involves trying to identify the minimum therapeutic dose and corresponding drug level to control her seizures.

Polytherapy

It was previously assumed that ASD polytherapy always posed more of a teratogenic risk than monotherapy and that polytherapy should be avoided whenever possible. This conclusion was based on several prior studies that demonstrated a higher rate of major malformations with polytherapy. However, it now seems that the results of these prior studies may have been driven by polytherapy combinations that included valproate. Within the NAAPR, Holmes and colleagues reported that the risk of major anomalies with lamotrigine and valproate therapy was 9.1%, whereas it was 2.9% for the combination of lamotrigine and any other ASD. Similarly, they reported that carbamazepine and valproate polytherapy was associated with a major malformation risk of 15.4%, which was much higher than the 2.5% risk seen with the combination of carbamazepine and any other ASD. The authors also highlighted similar findings that had been reported in the United Kingdom Epilepsy and Pregnancy Registry and the International Lamotrigine Pregnancy Registry. The Australian Register of Antiepileptic Drugs in Pregnancy (APR) also reported similar findings and noted that polytherapy combinations that included levetiracetam had a rate of malformations similar to polytherapy combinations that did not include levetiracetam (7.1% vs. 8.4%).

Although valproate seems to contribute significantly to the teratogenesis associated with polytherapy, the APR polytherapy combinations, excluding valproate, still have a significantly higher risk of malformations compared to monotherapies (8.18% vs. 4.77%). In an earlier study by the same group, polytherapy, including topiramate, was specifically associated with an increased risk of congenital malformations. In 87 pregnancies exposed to topiramate as part of polytherapy, there was a significantly higher rate of malformations when compared with polytherapies that did not include topiramate (14.9% vs. 6.6%). It should be noted that valproic acid was not excluded from this topiramate polytherapy cohort and 3 cases of hypospadias among the 13 malformations occurred in the same woman, which suggests genetic background played a role. The Kerala registry also found a significantly increased risk of teratogenesis associated with topiramate polytherapy when compared to other polytherapies.

Recent studies have suggested that the increased risk of teratogenesis in polytherapy combination may be largely driven by higher doses of certain ASDs in these polytherapy combinations . Patients taking polytherapy often have epilepsy that is harder to control and may be on higher doses of medications in addition to being on combination therapy. EURAP demonstrated that within polytherapy combinations that included valproate, much higher rates of malformations were present with higher dose of valproate compared to lower doses. This effect was statistically significant for valproate and lamotrigine in combination and a trend for other valproate polytherapies. In the APR polytherapy study, significant dose-response rates of malformations were associated with topiramate polytherapy when controlling for the doses of the combined ASDs.

Studies on cognitive development have suggested an adverse effect of some polytherapies. In an Australian cohort, in utero exposure to valproate polytherapy was associated with significantly lower FSIQ and verbal comprehension scores than exposure to valproate monotherapy or polytherapy combinations without valproate. The LMNG also reported that only polytherapies that included valproate were linked to decreased mean FSIQ and verbal IQ in school-age children.

The mainstay of epilepsy therapy, especially in women of childbearing age, is still to try to find the one ASD that best controls a patient's seizures at the minimum therapeutic dose or level. Polytherapy combinations that include valproic acid or topiramate should be avoided if possible. If required the doses of these two ASDs in particular should be kept as low as possible. In certain cases, polytherapy may be preferable to monotherapy. For example, women with genetic generalized epilepsies have a limited number of ASDs that are appropriate for their condition. Valproate is an effective option for this type of epilepsy but is a poor choice for these women. When one ASD—such as levetiracetam or lamotrigine—is ineffective for these patients, the combination of the two may sometimes be effective and likely carries a reduced risk of teratogenesis when compared with valproate monotherapy. Given the emerging information on the relationship between ASD dose and malformation risk for most ASDs, more research is needed to determine whether polytherapy combinations that involve low doses of two ASDs are ever preferable to a single nonvalproate ASD at a high dose.

Other Therapies

Most epilepsy is well controlled by ASDs; however, some patients continue to have seizures despite treatment with two or more appropriately selected ASDs. These patients are designated as having drug-resistant epilepsy (DRE). In these patients, other therapies may be considered. Patients with DRE should be referred to an epilepsy surgery center and evaluated for epilepsy surgery prior to planning pregnancy. If epilepsy surgery is deemed inappropriate, other options for DRE include responsive nerve stimulation, deep brain stimulation (SANTE), vagal nerve stimulation (VNS), and dietary therapies. There is a paucity of data on these therapies in pregnancy. The VNS is a device that is surgically implanted below the clavicle that delivers electrical stimulation to the left vagus nerve, and may reduce seizure severity and frequency. Sabers and colleagues reported on 26 pregnancies in which the mother had a VNS device. The majority of these patients were also on ASD polytherapy. Of these, only one child was born with a malformation. Obstetric intervention was needed in 53.9% of women, which the authors point out is higher than the 48% in the EURAP registry; however, women on VNS therapy represent a subset of epilepsy patients with particularly severe disease and 90% continued to have seizures during pregnancy.

The ketogenic diet (KD) and other such dietary therapies have mostly been efficacious in refractory pediatric populations but are being increasingly used in adults with epilepsy. The KD is a high-fat, low-carbohydrate diet that causes an acidosis in the patient, raising theoretical concerns for pregnancy. A single case series of two pregnant women with intractable epilepsy using the KD has been reported. One woman used KD with no ASDs and was not associated with any MCMs or adverse growth or developmental outcomes. The other woman used a modified Atkins diet and had a male neonate with bilateral ear deformities but was otherwise neurodevelopmentally normal at 8 months of age. No conclusions can be drawn from such a small number of patients; however, patients should be advised not to attempt radical dietary therapy, particularly during pregnancy, without the guidance of a nutritionist and obstetrician.