and FJE Vajda2
(1)
Clinical Neurology and Neuropharmacology, University of Queensland, and Honorary Consultant Neurologist, Royal Brisbane and Women’s Hospital, Brisbane, QLD, Australia
(2)
Department of Medicine and Neurology Director of the Australian Epilepsy and Pregnancy Register, University of Melbourne and Royal Melbourne Hospital, Melbourne, Australia
Abstract
Antiepileptic drug exposure in utero may have adverse effects on the developing brain, as has been shown in experimental animals. Exposure in the first trimester of human pregnancy may be associated with the occurrence of physical malformations in the foetus but, in addition, foetal valproate exposure poses a significant risk for the cognitive development of the infant, an effect that is dose dependent. Carbamazepine and lamotrigine exposures appear to hold a lesser risk for cognitive development, while the risks associated with exposure to the other antiepileptic drugs are uncertain. Further, evidence has become available that intrauterine valproate exposure may be associated with the development of autism-spectrum disorder in childhood. Not enough data exists to clarify the situation in this regard for the other antiepileptic drugs.
The highly important possibility that antiepileptic drug exposure during pregnancy may be associated with the presence of adverse consequences for cognition and neurobehavioral development in the infant and child has been rather intensively investigated in recent years. The situation in this regard has been progressively clarified in a series of publications from the collaborative NEAD (Neurodevelopmental Effects of Antiepileptic Drugs) studies carried out by Meador and colleagues, who have been collecting data and presenting interim analyses of a well-designed, comprehensive prospective investigation.
The actual recognition of cognitive and developmental adverse effects of antiepileptic drugs presents a more considerable challenge than observations and analysis of data related to physical malformations in the foetus and infant. The relevant data concerning intellectual functioning need to be obtained from prospective quantitative observational studies carried out in an area where human experimentation is unethical and the examinations and follow-up that are required have many limitations. Babies are not easy subjects for intellectual testing. Numerous potential confounders exist that make the assessment task very complex. Such confounders include parental factors such as educational level, socioeconomic status, drug taking (including tobacco and alcohol), concurrent illnesses and disorders that occur during pregnancy, family history, inherited illnesses, mental disorders, nutrition, folate intake, as well as environmental factors other than the above, and in addition among other factors the maternal epilepsy syndrome that is being treated and the effects of multiple or uncontrolled seizures during pregnancy.
Although many anecdotal reports had existed about impaired neurodevelopmental outcomes in the offspring of women with antiepileptic drug-treated epilepsies, it was not until relatively recently that prospectively recruited collections of in utero antiepileptic drug-exposed infants have been systematically explored and the findings about intellectual impairment followed up in a cohort of antiepileptic drug-exposed infants at progressively increasing ages during childhood (Meador et al. 2013).
Adverse Effects of Antiepileptic Drugs on the Developing Brain
Foetuses exposed to antiepileptic drugs during the first trimester of pregnancy, the primary period of organogenesis, have an elevated risk of congenital physical malformations, including defects that can affect the brain. The most obvious and serious of these are neural tube defects . The neural tube closes by the end of the third week after conception. This is the most vulnerable period for the development of neural tube malformations induced by antiepileptic drugs. It is also a time that often proves to be too early for the prospective mother to realise that she is pregnant. In the past, it was recommended that, after the first few weeks of pregnancy had passed, during which there was intense concern for the adverse effects of antiepileptic drugs on the brain, attention should shift to focus on seizure control throughout the remainder of pregnancy. This recommendation will now need to be revised, because it appears that the drug-related cognitive and developmental issues affecting the baby may originate throughout the whole length of pregnancy. Human brain development is a long, highly complex process that involves neurogenesis , synaptogenesis, apoptosis , synaptic pruning, glial development and myelination. It probably extends not only throughout gestation but also continues postnatally, through infancy, childhood, the teenage years and possibly beyond. The cellular processes involved in brain development are multiple, occurring in a parallel but interrelated manner, occurring in different brain regions and in different cellular types at different times. Antiepileptic drugs have been shown in experimental animals to have effects on all of these processes. The results of an insult to brain structure and function will depend on the processes which are most active in the time window when the insult occurs. The duration of the insult from antiepileptic drugs may extend throughout the entire duration of gestation. Kaindl et al. (2006) and Ikonomidou et al. (2007) have reported findings which lead to a better understanding of the potential adverse neurobiological effects that antiepileptic drugs can produce in the developing brains of experimental animals.
Obtaining objective clinical evidence of intrauterine brain damage in humans is easiest for the anatomical abnormalities that develop early in pregnancy. These produce structural malformations such as abnormal limbs or bones, or spina bifida , which are readily visible objective endpoints that are recognisable at birth. It is more difficult to identify abnormalities that develop later in pregnancy. Such disturbances of cognition , personality, behaviour and motor development may be due to multiple possible causes (Blom et al. 2006) and tend not to become obvious until well after birth.
Mechanisms of Behavioural Teratogenesis
Earlier animal experimental studies provided evidence of the adverse effects of postnatal phenobarbitone treatment on brain development in infant rats, resulting in reduced brain weights (Schain and Watanabe 1975), reduction in neuronal numbers in the hippocampus (Bergman et al. 1982) and cerebellum (Fishman et al. 1983), neurocognitive deficits (Rogel-Fuchs et al. 1992) and behavioural abnormalities that comprised increased aggression and overactivity (Diaz and Schain 1978). Phenobarbitone treatment in young animals with post-kainic acid-induced status epilepticus resulted in fewer seizures, as compared with saline-treated rats (Mikati et al. 1994), but also resulted in a worse neurocognitive performance (Bolanos et al. 1998).
Valproic acid is a histone deacetylase inhibitor, and it is possible that differential teratogenesis exists in B6 and D2 strain mice because of strain differences in histone acetylation . B6 mice foetuses are more susceptible than D2 foetuses to digit and vertebral malformations, but D2 mice are more susceptible to rib malformations. Downing et al. (2010) observed strain differences in acetylation of histones H3 and H4 in both the embryo and placenta following in utero valproate exposure. However, additional studies are needed to determine the roles of these changes in mediating teratogenesis, though the findings suggest that genetic factors , both maternal and foetal, may play a part in causing valproate-associated foetal malformations. Clonazepam , diazepam , phenobarbitone , phenytoin and valproate, as well as vigabatrin , have been observed to produce widespread neuronal apoptosis in neonatal rat brains (Bittigau et al. 2002, 2003; Asimiadou et al. 2005; Meador et al. 2007; Stefovska et al. 2008; Ikonomidou and Turski 2010). This effect is dose dependent and may occur at blood concentrations that would be therapeutically relevant in humans. Two antiepileptic drugs, each given at below apoptosis threshold dosages, can still trigger the full apoptotic response. Antiepileptics which do not produce apoptosis in monotherapy, including carbamazepine , lamotrigine and topiramate (Glier et al. 2004; Manthey et al. 2005; Kim et al. 2007), can enhance apoptosis induced by another agent (Katz et al. 2007). This suggests that antiepileptic drug polytherapy could increase the risk of neurodevelopmental problems in antiepileptic drug-exposed foetuses. Levetiracetam is the only antiepileptic drug among those tested thus far that does not produce apoptosis in monotherapy or enhance apoptosis produced by other antiepileptic drugs (Manthey et al. 2005). This information provides another indication that the availability of levetiracetam may offer a major advance in foetal safety in pregnancy (Shallcross et al. 2011; Vajda et al. 2014).
Human Cognitive Development
The impact on the cognitive and behavioural development of the offspring produced by foetal exposure to maternal antiepileptic drugs has emerged in recent times as an area of major concern in relation to the use of these drugs in pregnant women with epilepsy (Palac and Meador 2011). Pregnancy registers have contributed prospective data on women with epilepsy and their offspring in regard to many factors, including the intelligence of the offspring, though the different registers have employed different timings for their postpartum data collections (Vajda et al. 2007). Each of the individual registers has other weaknesses (see Chap. 8 and Meador et al. 2008). Nevertheless, enough information is now available from the registers to warrant making women with epilepsy who are planning a pregnancy aware of the possible neurobehavioral consequences of foetal antiepileptic drug exposure. In humans, foetal exposure, particularly to valproate or to antiepileptic drug polytherapy , may be associated with an increased risk of cognitive impairment and diminished verbal abilities in the offspring (Nadebaum et al. 2011a; Meador et al. 2013).
Cognitive Effects of Maternal Epilepsy per se
In a prospective study (Gaily et al. 2004), and in a blinded retrospective study (Holmes et al. 2000), no IQ differences were found between children of women with epilepsy that was not treated with antiepileptic drugs and healthy control children (Gedzelman and Meador 2012). Two prospective population-based studies (Gaily et al. 1988, 2004) found that there was no IQ impairment in children exposed in utero to self-limiting generalised tonic–clonic seizures (i.e. no status epilepticus was involved). However, a retrospective study (Adab et al. 2004) showed that the verbal IQ was significantly reduced in children who had been exposed to more than four generalised tonic–clonic maternal seizures during pregnancy. Based on case reports, prolonged seizures and status epilepticus are considered to provide a serious threat to both mother and foetus (Hiilesmaa 1996).
Other Factors Possibly Affecting Cognitive and Behavioural Development in Humans
Heritability accounts for 30–50 % of phenotypic variance in human IQ . The correlation coefficient between the IQs of monozygotic twins reared together is 0.85 and for those reared apart 0.67 (Sattler 1992). The correlation coefficient between parental and child IQ is 0.42 (Kaufman 1990). Maternal education and socioeconomic status correlate less closely with the child’s IQ (correlation coefficient = 0.298). The paternal IQ co-varies with the maternal IQ. Children’s IQs may also be affected by complications of pregnancy or severe childhood illnesses. Obstetrical complications and malnutrition do not generally appear to have an overall substantial effect on IQs, though individual children may be severely affected. Drug abuse in pregnancy influences foetal development, socioeconomic status is correlated with behavioural adaptation and social variables are known to predispose to biomedical risks. Other possible factors to be considered include gender, ethnic origin, geography, maternal age and parity. Approximately half of the patients with low IQs carry more than one risk factor (Meador et al. 2001). Although the majority of children born to women with epilepsy are normal, the somatic and functional development of these children, as a group, is reduced (Delgado-Escueta and Janz 1992).
Reported Antiepileptic Drug-Associated Effects on Human Cognitive Development
Antiepileptic drugs may decrease cell membrane excitability, increase postsynaptic inhibition or alter synchronisation of neural networks to decrease the excessive neuronal excitability that is associated with seizure development. Common side effects of decreasing neuronal excitability include slowed motor and psychomotor speed, poorer attention spans and mild memory impairment (Meador 2005). Treatment decisions made in childhood may have lifelong implications for cognitive ability (Loring and Meador 2001; Meador 2005).
Prenatal exposure to phenobarbitone , phenytoin , primidone and carbamazepine has been associated with reduced head circumferences (Hanson et al. 1976; Majewski et al. 1981; Gaily et al. 1990) and reduced psychomotor development of infants (Deblay et al. 1982). However, other studies did not support the correlation with phenytoin exposure (Smith et al 1986, 1994; Dodrill and Wilensky 1992; Pulliainen and Jokelainen 1994, 1995). Unfortunately, neurocognitive tests in young children appear to have a relatively poor reliability so that a long follow-up period may be required to obtain a valid assessment of this parameter. Selection biases may have been present in some studies concerning the matter, and multiple confounders such as those listed above may have been operative. The interpretation of the numerous studies on the differential effects of antiepileptic drugs on neurodevelopment remained controversial, until the publication of the multicentre, prospective series of NEAD studies.
Farwell et al. (1990) had studied 217 children (at 8 and 36 months) randomly assigned to 2 years treatment with phenobarbitone (4–5 mg/kg/day) or placebo. After 2 years (on treatment), the mean corrected IQ was 7.03 lower in the phenobarbitone-treated children. Six months after the drug had been discontinued, the mean IQ remained 5.2 points lower in the group that had been assigned to phenobarbitone. In the follow-up study, 139 children were re-tested 3–5 years later: the phenobarbitone-treated group performed significantly lower than the placebo group on WRAT-R, a reading achievement standard score.
Longer-term outcomes of children born to women with epilepsy were reported by Adab et al. (2004) from the UK. These workers performed a retrospective study of the prevalence of cognitive delay and possible associated dysmorphic features in children exposed to antiepileptic drugs in utero. The participants were aged between 6 months and 16 years. Structured interviews, hospital records, clinical examinations and psychometric tests (Wechsler) were used to assess the extents of drug exposure and the IQs. Of 249 children, 41 had been exposed to sodium valproate , 52 to carbamazepine , 21 to phenytoin and 49 to antiepileptic drug polytherapy , while 80 were unexposed to antiepileptic drugs. The mean verbal IQ was significantly lower in the valproate group compared with the unexposed and the other monotherapy groups. Both valproate exposure and frequent tonic–clonic seizures in pregnancy were significantly associated with a lower verbal IQ despite adjusting for other confounding factors. This study identified valproate as posing potential risks for developmental delay and cognitive impairment and was the first to suggest that frequent tonic–clonic seizures in the pregnant woman had a similar effect. Like many others in this field, this study may be criticised for some methodological flaws, perhaps in particular, the potential for ascertainment bias. Nevertheless, the study heightened existing concerns regarding the risks of in utero exposure to valproate during pregnancy.
A Finnish study (Gaily et al. 2004) of the effects of foetal antiepileptic drug exposure on cognition involved the intakes of carbamazepine (N = 73), phenytoin (N = 48), lamotrigine (N = 84) and valproate (N = 53). At 4 years of age, the IQs of the children showed statistically significant differences in favour of the first three drugs compared with valproate. The mean IQ values were, for carbamazepine 98, for lamotrigine 101, for phenytoin 99 and for valproate 92.
Neurocognitive effects of exposure to carbamazepine in utero have been reported in three further smaller population-based studies that showed no significant associations between prenatal exposure to the drug and poorer cognitive outcomes (Scolnik et al. 1994; Wide et al. 2002; Eriksson et al. 2005). A number of small population and clinic-based studies have found associations between intrauterine phenobarbitone and phenytoin exposures and lower IQs than those in unexposed controls, but these agents have not been established as posing a risk independent of maternal IQ (Vanoverloop et al. 1992; Reinisch et al. 1995). Although animal studies, referred to earlier, had shown that foetal exposure to antiepileptic drugs, at doses lower than the thresholds that are required to produce physical congenital malformations, can be associated with cognitive and behavioural abnormalities, the cognitive effects of human foetal exposure to antiepileptic drugs were not convincingly demonstrated until the NEAD data became available.
The NEAD Study
The NEAD study was a prospective, observational, investigation carried out in the United States and Britain. For the study, between 1999 and 2004, Meador and colleagues enrolled pregnant women with epilepsy who were taking a single antiepileptic agent (carbamazepine , lamotrigine , phenytoin or valproate ). The primary analysis was planned to involve a comparison of neurodevelopmental outcomes 6 years after exposure to the different antiepileptic drugs in utero. However, a planned interim analysis of cognitive outcomes at 3 years of age showed that children who had been exposed to valproate in utero had significantly lower IQ scores than those who had been exposed to the other antiepileptic drugs. The degree of neurodevelopmental impairment was valproate dose dependent (Meador et al. 2009). This interim report had compared 258 children exposed to antiepileptic drug monotherapy in utero (73 exposed to carbamazepine, 84 to lamotrigine, 48 to phenytoin and 53 to valproate). IQ adjustments were made to compensate for effects of maternal IQ, maternal age, antiepileptic drug dose, gestational age at birth and maternal use of folate before conception. The mean adjusted IQ was 101 for children exposed to lamotrigine, 99 for those exposed to phenytoin, 98 for those exposed to carbamazepine but 92 for those exposed to valproate. The children’s IQs were significantly related to maternal IQs in those exposed to carbamazepine, lamotrigine and phenytoin, but not in those exposed to valproate.
These findings, together with the obvious concern about high-dose intrauterine valproate exposure being related to the development of physical foetal malformations (see Chaps. 8 and 9), would support a recommendation that valproate should not be used as a first-choice antiepileptic drug in women of childbearing potential. These Meador et al. (2009) findings also indicate that foetal exposure to carbamazepine and lamotrigine is relatively safe in respect to adversely altering cognition in children exposed to them in utero. This is a very significant and welcome finding from the standpoint of counselling women about risks of antiepileptic drug therapy in pregnancy.
Results from the NEAD cohort at 6 years of age were published in 2013, as planned (Meador et al. 2013). At this stage, 305 mothers and 311 children were included in the primary analysis. The children’s IQs at this age were lower after exposure to valproate (mean 97; 95 % C.I. = 94–101), than after exposure to the other three drugs (lamotrigine mean IQ 108, 95 % C.I. = 105–111; carbamazepine mean IQ 105, 95 % C.I. = 102–108; phenytoin mean IQ 108, 95 % C.I. = 104–112). The children exposed to valproate did poorly in verbal and memory measures, as well as in measures of nonverbal and executive functions and in a dose-related manner that was statistically significant. Verbal abilities were more affected than nonverbal ones and there was evidence of benefit being derived from peri-conception folate administration. The latter findings were consistent with some, but not all, other published studies. The authors mentioned that a limitation of this landmark study may have resulted from the loss of a number of participants to follow-up. Nevertheless, the IQ values at 6 years correlated strongly with those in the earlier report, at 3 years, described above.
Other Recent Investigations
Cummings et al. (2011) carried out a blinded cohort study which employed control subjects. The study involved 186 children aged 8 years or less (of whom 142 had been exposed to antiepileptic drugs in utero). The children were prospectively ascertained through the UK Epilepsy and Pregnancy Register . It was found that those exposed to sodium valproate in utero were more likely to have evidence of neurodevelopmental delay than the controls. In the control population (N = 42), two children (4.8 %) showed delay (one significant, one mild). In the valproate group (N = 58), 23 children (39.6 %) showed delay, in five significant, in 18 mild. Ten children (20.4 %) exposed to carbamazepine in utero (N = 49) also demonstrated some degree of neurodevelopmental delay , of significant degree in two and mild in eight. However, those children exposed to lamotrigine in utero appeared to have similar neurodevelopmental outcomes to the control subjects, only one of the 35 (2.9 %) exhibiting delay.
Findings from the Australian Pregnancy Register also suggest that an association exists between exposure to valproate in pregnancy and lower IQs in the offspring, the IQ decrease being dose dependent. This finding, and that of the Meador et al. (2013) study described above, raise the possibility that lower valproate doses may be relatively safer from the neurodevelopmental point of view (Nadebaum et al. 2011a). However, it is recognised that there is considerable individual variability in the IQs among children exposed to similar valproate doses. The Australian study assessed the offspring of pregnancies in women with epilepsy who had participated in the Register and attempted to test the children when they were between 2 and 7 years of age. Children in the younger age groups are regarded as difficult test subjects, whose assessment may require the development of special methods. Children born to women with epilepsy that were exposed in utero to levetiracetam (N = 51) were assessed for early cognitive development and compared with children exposed in utero to valproate (N = 44) and also with a group of 97 children not exposed to antiepileptic drugs. The children were recruited prospectively and were assessed using the Griffiths Mental Development Scale (1996) when they were aged less than 24 months. The children exposed to levetiracetam obtained higher mean developmental scores than children exposed to valproate (P < 0.001). Those exposed to levetiracetam did not differ from control children in regard to the overall developmental quotient (P = 0.62). There thus was evidence that children under the age of 24 months exposed to levetiracetam in utero are not at an increased risk of delayed early cognitive development. The importance of this study lies in the possible future increased employment of levetiracetam in pregnant women who do not respond to or cannot tolerate lamotrigine and who are reluctant to take valproate or in whom that drug is contraindicated. In the future, levetiracetam may come to be considered an alternative medication to valproate in treating focal and also generalised epilepsy syndromes in women capable of pregnancy.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
