Epilepsy has special implications for women, especially during their childbearing years, and particularly when pregnancy is contemplated. Being aware of the relevant issues is therefore essential. For anyone with juvenile myoclonic epilepsy and other idiopathic generalised epilepsies, valproate is generally regarded as the gold standard treatment against which other antiepileptic drugs are compared. However, information from pregnancy registers consistently show valproate to be associated with the highest risk for major congenital malformations, and studies exploring cognitive and behavioural development of offspring also show that valproate has an adverse impact. This creates a significant headache when trying to choose the “right” drug for most women with a generalised epilepsy syndrome.
Statistics from Altmetric.com
So you’ve made a diagnosis of epilepsy and may even know the seizure type(s) and epilepsy syndrome. The decision to treat with antiepileptic drugs (AEDs) has been made and the patient has agreed to take what you have prescribed. Fortunately, irrespective of which AED you choose, there is a good chance that your patient will quickly become seizure free. But as you reach for your prescription pad, what factors should be guiding you to the correct choice from the now 20 or so licensed AEDs in the UK?
Consider this typical straightforward case. M is a 16-year-old with a 6 month history of myoclonic jerks on awakening, rare absences and one recent generalised tonic–clonic seizure. M has no other medical history and an EEG typical for an idiopathic epilepsy syndrome such as juvenile myoclonic epilepsy (JME). What is the best treatment for M, and does your choice really depend on whether M is Megan or Michael?
When choosing an AED for any patient with epilepsy the usual factors to consider are efficacy, tolerability, safety and ease of use. While quality of life for people with epilepsy is likely determined by more than just seizure control,1 no one would surely doubt that for new onset cases achieving full control of seizures is the main consideration. Sustained control also reduces the risk of injury and the (small) risk of increased mortality from all causes.
In considering M, the question arises why Michael is likely to be sent swiftly from the consultation with a prescription for valproate while Megan will likely be held back, probably after a lengthy discussion about “the future”, with a high chance, in the UK at least, that she will be given a prescription for lamotrigine or less likely another “broad spectrum AED”. Considering the SANAD study showed that the chances of achieving seizure freedom is significantly lower for lamotrigine than valproate for those with generalised or unclassified epilepsy syndromes, including JME,2 we need to ask why females and males with epilepsy are so often treated differently. And, in particular, why the potential for future pregnancy, anticipated or not, appears to be given so much weighting when choosing an AED for women with epilepsy.
Choosing the correct AED for women with epilepsy
It is well recognised that epilepsy has special implications for women. These are particularly obvious when she is of childbearing age and in particular when she is contemplating pregnancy.
Areas for consideration for women with epilepsy of child bearing age
The effects of AEDs on appearance.
The effect of female hormones on seizure control.
The effects of epilepsy and seizures, and AEDs, on fertility.
The effects of AEDs on contraception and vice versa.
The effects of epilepsy and AEDs on pregnancy.
The effects of pregnancy on AEDs and seizure control.
The effects of epilepsy, and in particular seizures, on the fetus.
The effects of AEDs on the fetus.
Getting it right for women with epilepsy is not an insubstantial problem. Women with epilepsy of childbearing age account for approximately 25% of people with epilepsy, or about 125 000 individuals in the UK. Since 3–4 pregnancies in every 1000 occur in women with active epilepsy,3 and women with epilepsy are considered a high risk group in pregnancy, the stakes are high. The confidential enquiries into maternal deaths in the UK, from 1985 to 1999, showed that epilepsy was overrepresented with more than a 10-fold increased risk of maternal death. While the reasons were not initially explored, a follow-up report covering the period 2000–2002 found that in 10 out of the 13 epilepsy related deaths that had a post mortem, from a total of 256 maternal deaths, four were felt to have died from sudden unexpected death in epilepsy, with another five probably having died from sudden unexpected death in epilepsy.4 Furthermore, in two out of the seven cases where blood levels of AEDs were measured, there was no trace of them.5 While poor compliance with AEDs in pregnancy is likely to have many determinants, fear of damaging the unborn child from AEDs is probably one of the most important. That compliance with medication is poor in this group is well recognised, with one study showing that up to 15% of women taking carbamazepine or lamotrigine had little or no AED in their proximal compared with distal hair samples.6
On the other hand, the current focus on the fetus might have resulted in harm to many women with epilepsy, in particular because they may have been denied an AED that offered the best chance of seizure control. It is the woman in front of us who is our patient and to whom our main duty of care applies. There is also clearly more to our female patients than their reproductive potential. Treatment choices should be individualised as for all patients with epilepsy and our approach in managing women with epilepsy of childbearing years must be more sophisticated than drug A bad and to be avoided at all costs, with drugs B, C, D etc, being alternatives of choice.
Risks to the fetus (and mother)
The fetus seems relatively resistant to the effects of maternal seizures although anecdotal evidence has shown that tonic–clonic seizures may cause fetal bradycardia. Studies have not suggested an association between any type of seizure and major congenital malformations although one did find that poorly controlled tonic–clonic seizures in pregnancy had an adverse effect on cognitive development.7
Status epilepticus in pregnancy has generally been felt to be particularly dangerous to both mother and baby. Teramo and Hiilesmaa compiled 29 cases from the literature and found that nine of the mothers and 14 of the fetuses died.8 In contrast, the prospective EURAP study did not find such an effect; of 1956 pregnancies there were 36 cases of status epilepticus, of which one-third were convulsive in type. There was just one stillbirth around the time of status, one spontaneous abortion long after the period of status and no maternal mortality.9
While seizures obviously pose a risk to the health and even the life of the mother, they clearly can have more widespread effects—on her quality of life including, for example, loss of driving rights. As a result most women who are seizure free and contemplating pregnancy appear reluctant to consider treatment withdrawal and the possibility of seizure relapse.
The risks of uncontrolled epilepsy to the mother must be weighed against the risks to the fetus from AEDs, usually taken to be of:
major congenital malformations (ie, of such significance they require medical or surgical intervention);
minor abnormalities which while occasionally disfiguring do not necessarily require intervention;
cognitive and behavioural delay or problems; and
intrauterine growth retardation.
Most of the literature has concentrated on the risk of major congenital malformations which, by definition, cause significant problems and are often a source of great distress to parents. Setting aside ethical or religious considerations, identification of such malformations in utero can present mothers with a choice, including termination of pregnancy. Some malformations can be rectified at birth or thereafter. But for offspring with adverse effects on cognitive development there is no effective corrective therapy.
Risks to the fetus from AEDs
There is now fairly robust evidence that AEDs can have an adverse effect on fetal development. It is a consistent finding that women with epilepsy who do not take AEDs have a lower risk of having a child with a major congenital malformation compared with those who do.10 Animal studies have demonstrated teratogenicity with all of the older AEDs and even some of the newer ones.
Women who take a single AED have at least a 2–3 times increased risk over the background population of having a child with a major congenital malformation, equivalent to a 4–9% chance (table 1).3, 10, 11 12 13 14 15 16 The risk is higher for women taking more than one AED3, 10 11 12 14 15 12, 14–16 (table 2) and higher still the more AEDs that are taken (fig 1).
Risks for AEDs taken as monotherapy to treat juvenile myoclonic epilepsy
For individual AEDs taken as monotherapy, there is convincing evidence that different AEDs carry different risks. Until recently most data were for the older AEDs (phenobarbital, phenytoin, carbamazepine and valproate). While there is now a substantial amount of information for lamotrigine, information on the other more recently available AEDs is either non-existent or preliminary (oxcarbazepine,17 levetiracetam and topiramate).
To better assess pregnancy outcomes in women with epilepsy, registries were set up towards the end of the 1990s with the aim of enrolling large numbers of pregnant women. Since information is never likely to be available from randomised controlled trials, these registries offer the only chance of better understanding the risks of developmental toxicity associated with AEDs, of identifying differences between them, and of assessing the impact of other important risk factors for adverse pregnancy outcomes.
Risks with valproate
That in utero exposure to valproate results in developmental toxicity was first recognised about 10 years after the drug’s introduction.18 Increasingly, valproate has been implicated as the AED associated with the highest risk for major congenital malformations. But what is the actual risk and why is there such a discrepancy between the risks reported in different studies—for example, 6.2% in the UK Epilepsy and Pregnancy Register16 and 10.7% (n = 149) in the North American AED Pregnancy Register?19 Other studies demonstrating an even higher risk for valproate include the Neurodevelopmental Effects of Antiepileptic Drugs study (17.4%)20 and one from Australia (16.8%).21 Possible reasons are different methods of recruitment, different definitions of malformations and different times for follow-up of the children.
For specific malformations, valproate has been associated with a significantly increased risk for spina bifida aperta (1–2%), with the greatest risk in those exposed to more than 1000 mg/day. Cardiovascular and urogenital malformations, skeletal defects and rib and vertebral anomalies and orofacial clefts have also been found to occur significantly above the background risk. There is some evidence of possible pharmacogenetic susceptibility to the teratogenic effects of valproate both from human reports22 and preclinical studies.23 There is also some information that high peak plasma concentrations are associated with an increased risk of malformations.
In addition to major congenital malformations there is information that minor anomalies may also be related to valproate exposure. A combination of facial dysmorphic features, which has been called the fetal valproate syndrome,24 and which is felt by some to be distinct from that seen with other AEDs such as phenytoin, has been described (similar dysmorphic features such as epicanthal folds, long philtrum, flat nasal bridge and hypertelorism do occur with other AEDs). The significance of these dysmorphic features in terms of whether they are markers for other problems, including long term development, is unknown. A few recent studies have shown that there is a significant negative correlation between dysmorphic features and verbal IQ, especially for valproate.7, 25
Features of the fetal valproate syndrome
Medial eyebrow deficiency
Flat nasal bridge
Broad nasal root
Long upper lip
Thin upper lip
The greatest worry about valproate however is probably developmental delay. For example, a study from the Merseyside Regional Epilepsy Centre found that 16% of 224 children who had been exposed to AEDs prenatally had additional educational needs compared with 11% of 176 exposed to no drugs.26 Thirty per cent of those exposed solely to valproate, and 20% exposed to polytherapy containing valproate, had additional educational needs. This compared with 3.2% and 6.5% for carbamazepine and other monotherapy regimens, respectively. In a follow-up of the same group, verbal IQ was significantly lower in children exposed to valproate monotherapy (mean 84) compared with unexposed children (mean 91) or children exposed to carbamazepine (mean 94) or phenytoin (mean 98).7 Multiple regression analysis revealed that as well as exposure to valproate, five or more tonic–clonic seizures in pregnancy and low maternal IQ were associated with lower verbal IQ. Doses over 800 mg/day were associated with lower verbal IQ than lower doses. In a study from Finland, full scale IQ was low (<80) in four of 21 infants that had been exposed to valproate (19%) and exceptionally low (<70) in two (10%). However, the mothers of children exposed to valproate had lower IQs and significantly lower educational levels.27
Assessing the effects of AEDs on cognitive and behavioural development is a complicated business. There are many potential confounding variables and these have generally not been fully studied. In fact, the Cochrane group had previously concluded that most studies in this area were of limited quality with little evidence to implicate one drug over another with respect to a detrimental effect on development.28
A recent study from India of 395 infants born to women with epilepsy attempted to address some of the above issues although it did not take account of the effect of maternal IQ.29 While valproate was associated with significantly lower mental and motor developmental scores compared with carbamazepine, it was no worse than the other AEDs studied in monotherapy. The importance of including all confounding variables had been shown in a prior study from the same group where low maternal IQ and maternal education as well as AED exposure were associated with significant impairment of intellectual and language functions in children of mothers with epilepsy.30
The Neurodevelopmental Effects of Antiepileptic Drugs (NEAD) study, recruiting from the USA and the UK, has very recently reported on developmental outcomes to 3 years of age in 258 children exposed to various AEDs as monotherapy. After adjusting for maternal IQ and other important confounding variables, children exposed to valproate in utero were significantly more impaired than infants exposed to carbamazepine, phenytoin and lamotrigine. Thirteen per cent of infants exposed to valproate had IQs in the learning disabled range compared with 5%, 3% and 2% for phenytoin, carbamazepine and lamotrigine, respectively.31 A dose–response was found for valproate but not for the other AEDs studied. One criticism of the study has been that there was no control group and therefore while providing important new data it is clear that we still urgently require further methodologically sound prospective observational studies of children exposed to all AEDs.
That valproate may be associated with an increased risk for autistic spectrum disorders and other problems with general behaviour has also been expressed in some preliminary reports.32
Risks with lamotrigine
There is now a good amount of safety data on lamotrigine for major congenital malformations. The International Lamotrigine Pregnancy Registry reported on the outcomes of 414 first trimester lamotrigine exposed pregnancies; 3% of the children had major birth defects with no distinctive pattern of malformations.33 The UK Epilepsy and Pregnancy Register revealed a similar risk for pregnancies exposed to lamotrigine alone, with 21 of 647 (3.2%) infants having a major congenital malformation, 5.4% if exposed to more than 200 mg a day.16 This dose effect has not been reported by other registers, including the International Lamotrigine Registry.34 With regards to the risks of specific malformations, the North American Registry found that of 16 (2.3%) of 684 infants exposed to lamotrigine with a major congenital malformation there was a 10-fold increased risk of cleft abnormalities.35 Neither the UK Epilepsy and Pregnancy Register16 nor the European Surveillance of Congenital Anomalies found an increased risk of isolated orofacial clefts.36
There is no information on the effect of lamotrigine on cognitive development although data from the NEAD study are encouraging. Further work is clearly needed before we can pronounce on whether or not lamotrigine has a deleterious effect on cognitive development and behaviour.
Risks with other “broad spectrum” AEDs
For other broad spectrum AEDs information on pregnancy is limited:
The UK Epilepsy and Pregnancy Register reported on 203 pregnancies exposed to topiramate. Of the 70 cases that had received only topiramate, three (4.8%) had a major congenital malformation of which two were clefting abnormalities and one was hypospadias.37 In contrast, another study of 52 exposed topiramate pregnancies did not raise any obvious concerns.38
No major congenital malformations from 39 pregnancies exposed to levetiracetam alone were reported by the UK Epilepsy and Pregnancy Register.39 The North American AED Pregnancy Registry have published data in abstract for 197 monotherapy exposures to both topiramate and levetiracetam where 4.1% and 2.0% risks were noted, respectively. Of interest and in keeping with results from the UK, two with topiramate were cleft lip deformities.40
For zonisamide, a drug which has been licensed for almost 20 years in some parts of the world, data for exposed pregnancies are embarrassingly scarce. The only report published to date included 26 pregnancies and did not raise any concerns.
Summary of data for AEDs used to treat juvenile myoclonic epilepsy
Based on available information it would seem a consistent finding that infants exposed to valproate in utero have a significantly greater risk of a major congenital malformation compared with those exposed to lamotrigine, especially if exposed to over 1000 mg/day. Some concerns have also been raised with high doses of lamotrigine and clearly whether this is a real effect needs to be clarified. The role of confounding factors, including parental effects, epilepsy type and seizure frequency has yet to be properly resolved.
The absolute risk of a major congenital abnormality for an infant exposed to valproate ranges from 6.2% to 17.4%, with the range quoted for lamotrigine being much tighter at between 2.3% and 3.2%. When counselling women it would therefore seem reasonable to state that based on current information the risk for a major congenital malformation with valproate is at least 2–3 times that for lamotrigine. But in some ways the actual risks may be of more academic than practical significance because our experience suggests it is not common for women who are contemplating pregnancy to take risks with the health of their offspring, no matter how low or inexact the risk.
For neurodevelopmental delay, while recent studies all point to outcomes being worse with valproate compared with lamotrigine, as yet we are only starting to understand the frequency of the problem, or the magnitude of effect. As a result, presenting this information to women is fraught with difficulties. Do we not mention preliminary results to our patients and later be accused of withholding information, or do we present all that is available, problematic as the data are, and in doing so potentially frighten women with information that can at times be uninterpretable even for the scientifically initiated and statistically aware?
Other issues to be covered
This should be available to all women with epilepsy contemplating pregnancy and given by someone who has a full understanding of epilepsy. It should start at the time of diagnosis and be repeated regularly. While it may not always be appropriate to discuss the many relevant issues (eg, in paediatric practice) it should certainly be considered in female adolescents, including those whose care is being transferred from a paediatrician to an adult physician. The fact that the relevant issues have been discussed should be clearly recorded in the notes. Recent work has shown that women with epilepsy of childbearing years do not always recall being given relevant information.41 Clear documentation of the issues discussed is good practice, not least for practical defensive purposes in the event of a bad outcome. Such verbal advice can be supplemented with written information and/or copying letters to patients.
The AEDs that can reduce the effectiveness of contraceptive preparations are shown in table 3. The situation for lamotrigine is not clear. While initially hailed as not interfering with the combined oral contraceptive pill, one report showed it caused a small decrease in the level of levonorgesterol. The manufacturer therefore recommends that women are informed of the potential for pill failure and that additional contraceptive methods are used. Since lamotrigine is not an inducer of the cytochrome P450 isoenzyme system, increasing the dose of ethinylestradiol is not indicated. Probably of more clinical relevance, oral contraceptives can reduce the level of lamotrigine, and to a clinically significant extent. For those who initially titrate up to a low daily lamotrigine dose (100–200 mg) some consideration might therefore be given to increasing this.
Much has been written about the effects of AEDs on female fertility. A full review is beyond the scope of this article. In summary, the situation is far from resolved. While valproate has been implicated more than other AEDs, study results are conflicting. At present it is therefore not possible to properly advise women, apart possibly from mentioning the association between weight gain and polycystic ovarian syndrome.
Large community based studies have demonstrated a reduction in the risk of neural tube defects in women taking folic acid periconceptionally. It has been inferred from this that folic acid may also be protective in women with epilepsy who have a higher than usual risk of neural tube defects. Despite the recommendations that high dose folate is prescribed for women with epilepsy,42, 43 there is no good evidence that this will protect against the neural tube defects, or other defects, seen in association with AEDs; indeed, there are some data to suggest that it does not.44 The effect of folate supplementation was reported for 4680 cases from the UK Epilepsy and Pregnancy Register.45 Those patients who received preconceptual folic acid, approximately 75% of whom received 5 mg each day, appeared more likely to have a child with a major congenital malformation than those who did not (3.9% vs 2.2%). While these results clearly do not mean that we should stop prescribing folate periconceptionally to women with epilepsy, they do question the wisdom of information contained in current guidelines.
General health measures
All women contemplating pregnancy, irrespective of whether or not they have epilepsy, should be reminded to look after their general health, taking measures to manage comorbid conditions, and to avoid other known teratogens such as cigarette smoking, alcohol and other drugs where possible.
Managing women with epilepsy syndromes such as JME is far from straightforward and how best to do it is still not entirely clear, despite the mountain of data that is now available. In some ways the results from the studies published, most of which can be criticised methodologically, have raised more questions than they have provided answers. What should be clear is that there is more to preventing adverse pregnancy outcomes than the effects of AEDs.
There do appear to be differences between AEDs, with valproate consistently being associated with a higher risk of major congenital malformations than lamotrigine. Likewise, for neurodevelopmental delay, there is some information suggesting that valproate is associated with worse outcomes than other AEDs, although most studies implicating it have had methodological flaws. Information for lamotrigine seems less concerning but much larger numbers of exposures are required to pronounce on its safety. What the actual difference in risks is between valproate and lamotrigine for any of the outcome measures is still unknown although for the woman contemplating pregnancy any increased risk may well be unacceptable. On the other hand, valproate is more effective than lamotrigine for idiopathic generalised epilepsies, including JME, and many women may be denying themselves seizure freedom in an attempt to minimise the risks to potential offspring which through choice or circumstance may never arise. The adage that women with epilepsy should never be prescribed valproate has the potential to do harm and a more sophisticated approach to managing women with epilepsy and all they represent as individuals should therefore be usual practice. With regards to the other newer AEDs considered appropriate for the management of JME, while some may feel that they offer a satisfactory alternative choice, efficacy data as yet do not support this approach and pregnancy data are quite frankly insufficient to pronounce on their safety.
Treatment with monotherapy in women of childbearing potential should be the aim where possible.
We need to clearly convey to patients that the vast majority of pregnancies in women with epilepsy are uneventful for both mother and fetus.
Opportunities to counsel women on other general health measures such as smoking should not be missed.
In some circumstances judicious use of valproate is justified in women of childbearing potential; this requires clear communication and documentation of issues relating to major congenital malformations.
More studies of the effect of AEDs on cognitive and behavioural development of exposed offspring are urgently required.
This article was reviewed by Susan Duncan, Edinburgh, UK.
Competing interests None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.