The role of CYP450 system on metabolism of psychotropic drugs

The role of the CYP450 enzyme system and glucuronosyltransferases in clinical psychopharmacology is being increasingly recognized. Among antidepressants, Tricyclic antidepressant drugs, such as amitriptyline, clomipramine and imipramine, are extensively metabolized by CYP1A2,2D6 and 3A4 (Table CYP enzymes involved in psychotropic drug metabolism). Nortriptyline and desipramine are, respectively, the active metabolites of amitriptyline and imipramine and are metabolized mainly by CYP2D6. Moclobemide is primarily metabolized by CYP2C subfamily, of which it is probably an inhibitor, while the atypical antidepressants mianserin and trazodone are metabolized by CYP2D6.

Table CYP enzymes involved in psychotropic drug metabolism

CYP1A2	CYP3A4	CYP2C9/10	CYP2C19	CYP2D6
Anti depressants	Antidepressants	Anticonvulsants	Antidepressants	Antidepressants
Amitriptyline	Amitriptyline	Phenytoin	Amitriptyline	Fluoxetine
Clomipramine	Clomipramine	Antipsychotics	Citalopram	Paroxetine
Imipramine	Desipramine	Thioridazine	Clomipramine	Mianserin
Trazodone	Imipramine	Olanzapine	Imipramine	Venlafaxine
Fluvoxamine	Norclomipramine		Moclobemide	Trazodone
Antipsychotics	Nortriptyline		Anticonvulsants	Nefazodone
Chlorpromazine	Trimipramine		Mephenytoin	Amitriptyline
Haloperidol	Nefazodone		Esobarbital	Clomipramine
Clozapine	Sertraline		Mephobarbital	Desipramine
Olanzapine	Venlafaxine			Imipramine
Ziprasidone	Antipsychotics			Norclomipramine
	Haloperidol			Nortryptiline
	Clozapine			Trimipramine
	Risperidone			Maprotiline
	Ziprasidone			Antipsychotics
	Iloperidone			Chlorpromazine
	Quetiapine			Thioridazine
	Anticonvulsants			Haloperidol
	Carbamazepine			Olanzapine
				Risperidone
				Iloperidone
				Quetiapine

The selective serotonin re-uptake inhibitors, fluoxetine and paroxetine are metabolized by CYP2D6, while sertraline, fluvoxamine and citalopram are respectively metabolized by CYP3A4, 1A2 and 2C. Paroxetine and fluvoxamine are, respectively, inhibitors of CYP2D6 and 1A2. In vitro and in vivo data demonstrated a moderate inhibition activity of fluoxetine on CYP2D6 and 3A4, probably mediated by its metabolites. No clinically significant induction-inhibition properties have been demonstrated for sertraline and citalopram.

Among the new generation of antidepressant drugs, venlafaxine is primarily metabolized by CYP2D6, while CYP3A4 metabolizes nefazodone and reboxetine. Nefazodone is a potent inhibitor of this enzymatic pathway.

Table CYP enzymes inhibited by different psychotropic drugs

CYP isoenzyme	Antidepressants	Antipsychotics
CYP1A2	Fluvoxamine
CYP3A4	Fluoxetine	Chlorpromazine
	Sertraline	Thioridazine
	Nefazodone	Haloperidol
		Risperidone
CYP2C9/10/19	Fluoxetine	Thioridazine
	Sertraline	Clozapine
	Fluvoxamine
	Moclobemide
CYP2D6	Fluoxetine	Thioridazine
	Paroxetine	Haloperidol
	Sertraline	Clozapine
		Olanzapine
		Risperidone

Neuroleptics, such as phenothiazines, are metabolized by intestinal sulfoxidases, although CYP2D6 plays an important role in chlorpromazine and thioridazine metabolism. They are also partially metabolized by CYP1A2 and 2C, respectively, and partially inhibit CYP3A4. Haloperidol’s metabolism has been studied for more than 30 years. It is metabolized by CYP3A4 and 1A2 and only partially by 2D6.

Among the atypical antipsychotics, clozapine undergoes extensive hepatic metabolism and multiple CYP enzymes are involved, however the two prominent ones are CYP1A2 and CYP3A4.

New antipsychotic drugs usually have better pharmacokinetic profiles. Risperidone is primarily metabolized by CYP2D6, although a correlation study using a panel of human microsomes suggest that CYP3A4 may also be involved. Olanzapine undergoes extensive hepatic metabolism and shares some of its metabolic routes with the structurally and pharmacologically related clozapine, but glucuronosyltransferases appear to be major metabolic pathways. Quetiapine shares some pharmacologic and structural characteristics with clozapine and olanzapine. In vitro studies using human microsomes showed that CYP3A4 is the main isoenzyme involved in quetiapine metabolism.

Interactions between anticonvulsants and antidepressants

SSRI-serotonin-noradrenergic re-uptake inhibitor

Data about fluoxetine-carbamazepine interactions are contradictory. Spina etal. (1993) found no modification in carbamazepine plasma levels before and after fluoxetine administration, although in a small group of patients. Grimsley et al. (1991) observed a slight increase in carbamazepine area under curve (area under the curve) levels and a decrease in 10,11-carbamazepine-epoxide area under the curve.

Nelson etal. (2001) studied the inhibition properties of several selective serotonin re-uptake inhibitors on phenytoin (combination of phenytoin) metabolism in an in vitro study with human liver microsomes. They suggested that the risk for a combination of phenytoin-SSRI interaction is highest with fluoxetine and less likely with the others (paroxetine and sertraline).

Andersen et al. (1991) investigated possible kinetic interaction between paroxetine and carbamazepine, valproate and combination of phenytoin in a single-blind, placebo-controlled, crossover trial. Paroxetine caused no change in plasma concentrations and protein binding of the anticonvulsants. Studies of paroxetine plasma concentrations are lacking, but the major enzymatic pathway is a non-inducible enzyme (CYP2D6), therefore modifications in plasma levels are unlikely, when co-administrated with antiepileptic drugs with inducer properties.

Leinonen et al. (1996) observed an increase in citalopram levels when carbamazepine was substituted with oxcarbazepine in two patients, demonstrating a significant induction effect of carbamazepine on citalopram metabolism.

Spina etal. (1993) studied the potential interaction between carbamazepine and fluvoxam-ine in eight epileptic patients in steady state for carbamazepine. No significant changes in carbamazepine and carbamazepine-10,ll-epoxide occurred.

Mamiya et al. (2001) described a single case of combination of phenytoin intoxication (from 16.6 to 49.1 µg/ml) after fluvoxamine administration. There are no studies of valproate-fluvoxamine interactions.

Not clear is the possibility of an interaction between sertraline and combination of phenytoin. Haselberger et al. (1997) described an elevation in combination of phenytoin plasma levels in two elderly patients, but without any symptoms of toxicity, while Rapeport et al. (1996a) demonstrated the absence of any pharmacokinetic interaction in a double-blind, randomized, placebo-controlled study with 30 healthy volunteers.

Kaufman and Gerner (1998) reported two cases of lamotrigine-sertraline interaction, leading to high lamotrigine plasma levels (doubled in the first case and 33% increase in the second one). Rapeport et al. (1996b), in a double-blind, randomized, placebo-controlled study on 14 healthy volunteers, observed no significant effects of sertraline on carbamazepine pharmacokinetics. Bonate et al. (2000) demonstrated the absence of drug interaction between clonazepam and sertraline in a randomized, double-blind, placebo-controlled, crossover study with 13 subjects.

No clinical studies are available about potential interactions between venlafaxine and antiepileptic drugs. Toy et al. (1995) demonstrated no pharmacokinetic interactions between venlafaxine and diazepam in a randomized, crossover study with 18 male subjects.

Roth and Bertschy (2001) reported three cases of increased carbamazepine plasma levels (from 20% to 100%) after nefazodone introduction. Laroudie et al. (2000) investigated kinetic interactions between nefazodone and carbamazepine in 12 healthy subjects. They observed a significant decrease in nefazodone area under the curve and an increase in carbamazepine area under the curve, demonstrating a potential inhibition property of nefazodone on carbamazepine metabolism.

TCA

Generally, phenobarbital, carbamazepine and combination of phenytoin stimulate the metabolism of Tricyclic antidepressant drugs, while valproate can increase their plasma levels. Wong etal. (1996) investigated the effect of valproate on amitriptyline and its active metabolite (nortriptyline) in an open-label study. The mean area under the curve and the peak plasma concentration, for the sum of nortriptyline and amitriptyline, were 42% and 19% higher. Fehr etal. (2000) reported the increase in serum clomipramine levels when coprescribed with valproate.

Szymura et al. (2001) investigated the effect of carbamazepine on imipramine and desipramine serum concentrations in 13 patients with major depression. They demonstrated that carbamazepine affects not only the metabolism of both Tricyclic antidepressant drugs but also their protein binding, leading to a significant increase in the free fraction. Because of this phenomenon, a modification in imipramine dosage regimen does not seem to be necessary in practice. Conversely, Van Belle et al. (1995) demonstrated a significant inhibition in carbamazepine metabolism by clomipramine in rats.

Others

Ketter et al. (1995) investigated the safety and efficacy of carbamazepine-moclobemide cotreatment in a double-blind study. The combination was well tolerated with no modifications in carbamazepine kinetics, but they did not assess moclobemide plasma concentrations.

Nawishy et al. (1981) investigated the presence of kinetic interactions between mianserin and three commonly prescribed anticonvulsants (combination of phenytoin, carbamazepine and phenobarbital). All of them are inducers of the CYP450 enzyme system. They observed a significant reduction in mianserin plasma concentrations.

The use of bupropion is limited by the high seizure risk. Carbamazepine is a potent inducer of its metabolism, taking the antidepressant plasma concentrations to undetectable levels. On the other hand, bupropion has shown marked inhibition properties, increasing valproate levels when prescribed in cotherapy, and Tekle and al-Kamis (1990) suggested a potential inhibition property of bupropion on combination of phenytoin metabolism. Odishaw and Chen (2000) investigated the effect of steady state slow release bupropion on the pharmacokinetics of lamotrigine in a randomized, open-label, crossover study with 12 healthy subjects. The kinetic parameters of a single 100-mg lamotrigine dose were not modified significantly.

Interactions between anticonvulsants and antipsychotic drugs

Phenothiazines-butyrophenones

Thioridazine is metabolized by intestinal sulfoxidases that are induced only partially by antiepileptic drug inducers such as carbamazepine, combination of phenytoin and phenobarbital but some authors have reported an increased clearance of thioridazine and a relevant decrease of mesoridazine (the active metabolite of thioridazine) in patients taking carbamazepine and/or combination of phenytoin. On the other hand, thioridazine, as chlorpromazine and prochlorperazine, inhibits combination of phenytoin, phenobarbital and valproate metabolism.

Several studies have shown that haloperidol plasma levels decrease by 50-60% after carbamazepine co-administration, with concomitant worsening of the psychiatric clinical features. Hirokane et al. (1999) evaluated haloperidol levels in patients comedicated with carbamazepine or phenobarbital. In the first group plasma levels were 37% lower; in patients treated with phenobarbital they were 22% lower. Interestingly, Iwahashi et al. (1995) observed that serum carbamazepine concentrations in patients treated without haloperidol were significantly decreased (on average 40%), compared to those treated with both carbamazepine and haloperidol. Hesslinger et al. (1999) compared the effects of carbamazepine and valproate cotreatment on the plasma levels of haloperidol and on the psychopathologic outcome in schizophrenic patients. Valproate had no significant effects on haloperidol plasma levels and it was associated with a better psychopathologic outcome. Doose etal. (1999) investigated the effect of topira-mate on haloperidol pharmacokinetics in healthy volunteers, observing no clinically significant interactions.

Benzisoxazoles and benzisothiazoles

Preliminary evidence from drug monitoring studies and case reports demonstrated that carbamazepine might cause a prominent decrease in plasma concentrations of risperidone. Spina et al. (2000) compared the risperi-done total active moiety (risperidone plus its active metabolite – TAM) steady state plasma concentrations in patients treated with risperidone alone and in patients comedicated with carbamazepine or valproate. Unlike carbamazepine, valproate (at dosages up to 1200-1500 mg/day) had minimal and clinically insignificant effects on plasma levels of risperidone TAM, suggesting that valproate could be added safely to an existing treatment with risperidone. Ono et al. (2002) evaluated the relationship between CYP2D6 genotype and the pharmaco kinetic interaction with carbamazepine, suggesting that the decrease in risperidone concentration is dependent on the CYP2D6 activity. Recently, an open study described a mild increase in carbamazepine plasma levels in eight patients with epilepsy after addition of risperidone 1 mg, suggesting that the antipsychotic, or more likely its metabolites, could modulate CYP3A4 activity. Interestingly, Furukory et al. (2001) demonstrated a different enantioselective 9-hydroxylation of risperidone by CYP2D6 and CYP3A4. In the literature, there is no information about differences in pharmaco-logic activity of these two enantiomers.

Ziprasidone and perospirone are newly available antipsychotic drugs and there are few clinical studies about their interactions. Miceli et al. (2000) studied the effect of carbamazepine on steady-state ziprasidone in healthy volunteers in an open, randomized, parallel-group study. They observed a clinically insignificant reduction (<36%) in steady-state ziprasidone levels.

Thienobenzodiazepine, dibenzothiazepine and dibenzothiazepine derivatives

Generally, combination of phenytoin, phenobarbital and carbamazepine cause a decrease in clozapine plasma concentrations. However, carbamazepine is rarely used in combination with clozapine because of the high risk of haematologic side effects. Existing data on the effect of valproate co-administration are contradictory. According to some authors, valproate has a moderate inhibiting effect on the demethylation of clozapine (catalysed by CYP1A2 and 3A4) but, in two small studies serum concentrations of clozapine and norclozapine (one of clozapine’s metabolites) were found to decrease respectively by 15% and 65%, suggesting induction of clozapine metabolism. Moreover, clozapine disposition is characterized by large interindividual variability, being affected by age, gender, body weight, dose per kg, smoking habits and ethnicity.

Olanzapine plasma concentrations are decreased by carbamazepine, but the authors did not consider this interaction clinically relevant because of the wide therapeutic index of the antipsychotic. In the literature, there are no controlled studies assessing drug interactions between olanzapine and new antiepileptic drugs in humans.

Quetiapine is a newly introduced atypical antipsychotic, and clinical data about pharmacokinetic interactions are lacking. Wong et al. (2001) demonstrated that combination of phenytoin has a marked effect on the metabolism of quetiapine, suggesting that dosage adjustment of quetiapine may be necessary when quetiapine is coprescribed with other antiepileptic drug inducers such as carbamazepine or phenobarbital.

Interactions between anticonvulsants and anxiolytics

Generally, anxiolytics have a wide therapeutic index; therefore the clinical relevance of pharmacokinetic interactions is very limited. Antiepileptic drugs with enzyme-inducing properties may stimulate the biotrasformation of many benzodiazepines. Carbamazepine has been reported to induce clobazam and diazepam metabolism. Carbamazepine has also been demonstrated to enhance the clearance of clonazepam and alprazolam. A clinically relevant interaction occurs between antiepileptic drug inducers and midazolam that is extensively metabolized by CYP3A4.

Anticonvulsants and antidepressants

Antidepressants have been extensively evaluated in relation to the general problem of their proconvulsant activity. But the definition of pharmacodynamic interaction implies that the typical pharmacologic properties of a drug are modified by another drug, without any change in the drug concentration. This definition comprises also side effects such as sedation, confusion, psychomotor impairment and others.

The risk of antidepressant-induced seizures is well known, particularly in people with epilepsy. Most of the data arise from studies using in vitro technique, animal studies and clinical observations. Among selective serotonin re-uptake inhibitors, fluoxetine is the most studied drug. It is interesting to note that several studies emphasized the role of serotoninergic transmission in enhancing the anticonvulsant effects of antiepileptic drugs. Leander (1992) demonstrated, in an animal model of epilepsy, that the selective inhibition of serotonin uptake by fluoxetine can enhance the anticonvulsant potency of combination of phenytoin and carbamazepine. Therefore, a favourable pharmacodynamic interaction maybe suggested.

As far as other interactions are concerned, Dursun et al. (1993) reported a single severe case of the serotonin syndrome after fluoxetine was added to carbamazepine. The occurrence of an extrapyramidal syndrome within fluoxetine and antiepileptic drug cotreatment are described, but clinical studies are lacking.

Two different studies of Rapeport et al. (1996a, b) investigated pharmacodynamic interactions between sertraline and combination of phenytoin or carbamazepine. Both of them showed no clinically significant interactions.

Anticonvulsants and antipsychotic drugs

Historically, antipsychotic drugs have been considered proconvulsants possibly because of their D2-receptor blocking activity. One of the most important issues in prescribing these two types of drug at the same time is about the effect of antipsy-chotics on the anticonvulsant effect of antiepileptic drugs.

To determine the risk for drug-induced seizures we can use different approaches: observational studies (case-control studies and case reports), drug-induced electroencephalograph changes, animal models and in vitro techniques in isolated tissue samples.

Table Pharmacokinetic interactions between antiepileptic and antidepressant

	carbamazepine	valproate	combination of phenytoin	lamotrigine	topiramate	phenobarbital
Fluoxetine	( = ↑)	(↓)	(↑)
Paroxetine	( = )	( = )	( = )
Citalopram	↓ ( = )
Sertraline	↓ ( = )		(↑=)
Fluvoxamine	( = )		(↑)
Venlafaxine	( = )
Reboxetine	↓
Amitryptiline	↓	↑
Clomipramine	↓ (↑)	↑	↓			↓
Imipramine	↓	↑	↓			↓
Desipramine	↓	↑	↓			↓
Nortriptyline	↓	↑	↓			↓
Moclobemide	( = )
Mianserin	↓		↓			↓
Trazodone			(↑)
Mirtazapine	↓ ( = )
Nefazodone	↓ (↑)
Bupropion	↓	(↑)	(↑)	( = )
Viloxazine			(↑)	(↑)

Symbols on the left are referred to antidepressant drug and within brackets to anticonvulsant drug, when prescribed in combination (in blank fields data are not available).

T, Increased plasma concentration; -l, decreased plasma concentration; =, unchanged plasma concentration.

*Dosage adjustments are not necessary.

Performed on psychiatric patients and, although theoretically correct, it is not known if drug-related seizures in non-epileptic patients predict risk in patients with epilepsy, and if different syndromes of epilepsy have different risks for psychotropic-induced seizures.

Generally chlorpromazine and clozapine are considered proconvulsant in epileptic patients. The former only at high doses (1000mg/day) and the latter at medium and high doses (>600mg/day). Clozapine frequently causes epileptiform electroencephalograph changes and seizures in 3-5% of patients treated, even at therapeutic doses. Devinsky etal. (1991) observed a mean prevalence of seizures of 2.9% with clozapine, and considering different doses, the prevalence is respectively 1, 2.7 and 4.4% for doses <300mg, 300-600 mg or 600-900 mg/daily.

Table Risk for seizures exhibited by some antidepressant and antipsychotic drugs

High risk	Intermediate risk	Low risk
Antidepressant drugs
Buproprion	Amitriptyline	selective serotonin re-uptake inhibitors
Clomipramine	Imipramine	Trazodone
Maprotiline		Venlafaxine IMAO
Antipsychotic drugs
Chlorpromazine (dose related)	Olanzapine	Fluphenazine
Clozapine (titration and dose related)	Quetiapine	Pimozide
	Haloperidol	Trifluoperazine Risperidone

Table Pharmacokinetic interactions between antiepileptic and antipsychotic drugs

	carbamazepine	phenobarbital	combination of phenytoin	valproate	lamotrigine	topiramate
Chlorpromazine	↓ (↑)	(↑)	(↑)	(↑)
Thioridazine	↓	↓ (↑)	↓ (↑)	(↑)
Mesoridazine	↓		↓
Haloperidol	↓ (↑)	↓	↓	=	=	=
Clozapine	↓	↓	↓	= ↑
Olanzapine	↓	↓	↓	↑	=
Risperidone	↓ (↑)	↓	↓	=
Ziprasidone	↓	↓	↓
Iloperidone	↓ (↑)	↓	↓	=
Quetiapine	↓	↓	↓

Symbols on the left are referred to antipsychotic drug and within brackets to anticonvulsant drug, when prescribed in combination (in blank fields data are not available).

T, Increased plasma concentration; -l, Decreased plasma concentration; =, Unchanged plasma concentration;

Theoretical data, no clinical studies available.

Devinsky (1994) analysed only patients without a previous history of seizures and the prevalence of seizures was respectively 0.9, 0.8 and 1.5% for the same range of doses of the previous study. Thus, with clozapine this seems to be a dose-related phenomenon but probably the role of the titration time and increase of dose is more important.

Olanzapine, quetiapine and risperidone demonstrated an extremely low risk of seizures when compared with haloperidol and can be considered safer.

Anticonvulsants and lithium

Lithium carbonate is frequently used for manic episodes in bipolar disorder, in association with valproate and carbamazepine. Carbamazepine also demonstrates antimanic properties, and a possible favourable pharmacodynamic interaction could be suggested, but carbamazepine can increase lithium toxicity as well. Shukla et al. (1984) suggested that carbamazepine enhanced the development of a lithium neurotoxic syndrome in patients with underlying central nervous system  disease or metabolic disease. This syndrome is characterized by symptoms such as confusion, drowsiness, lethargy, tremor and cerebellar signs that are typical of both lithium and carbamazepine toxicity. Therefore, a pharmacodynamic synergic interaction is probable. Kramlinger and Post (1990) studied the effects of this combination in 23 patients with affective disorders. They observed a significant increase in many haematologic parameters (mainly the mean white blood cell count, probably lithium counteracts the neutropenic properties of carbamazepine) and a significant modification in thyroid function with decrease in T4 and freeT4. Another well-known issue is the opposing effects of carbamazepine and lithium on electrolyte regulation, with the occurrence of severe hyponatremia when lithium alone is stopped.

The combination of lithium and valproate is widely used in rapid cycling, manic, depressive and mixed episodes in bipolar disorder. This combination has a higher tolerability than the co-administered carbamazepine and a pharmacodynamic synergistic interaction has been suggested.

Chen et al. (2000) investigated lithium pharmacokinetics when co-prescribed with lamotrigine in 20 healthy subjects. There were no significant differences in lithium pharmacokinetic parameters.

Background

In 1972 Kenyon sent a letter to the British Medical Journal describing a patient with epilepsy treated with phenytoin (combination of phenytoin) who became pregnant despite taking usual amounts of oral contraceptive pills. She astutely attributed the contraceptive failure to an inductive effect of the combination of phenytoin on the metabolism of the sex steroid hormones. This observation was soon confirmed by others and the underlying mechanisms were further elucidated. All the older antiepileptic drugs, carbamazepine, phenobarbital, combination of phenytoin and primidone except valproate were found to have similar effects. In contrast most of the new antiepileptic drugs with the exception of felbamate, oxcarbazepine and topiramate do not change the metabolism of the oral contraceptives. Parenteral formulations (intramuscular (i.m.) depot, subcutaneous implant and dermal patch) of contraceptive female sex hormones have also been reported to be subject to increased clearance. The effect of the antiepileptic drugs on testosterone metabolism has also indicated changes occur although the evidence of clinical effects is less easily assessed than an unplanned pregnancy. Conversely, except for lamotrigine the oral contraceptives do not appear to change the pharmacokinetics of antiepileptic drugs.

Frequency and importance of interactions

Following the initial case report by Kenyon, three other cases of oral contraceptive failure were cited by Janz and Schmidt (1974). By 1983 Sonnen found that 52 cases had been reported. He concluded that the incidence was probably much higher because 12 women in his own small population had experienced an unplanned pregnancy while using oral contraceptive pills when taking antiepileptic drugs. Although the effect of carbamazepine, phenobarbital and combination of phenytoin became established, the exact risk of an unplanned likelihood could only be estimated. The probability approximates that of condom use, a five-to ten-fold increase relative to use of oral contraceptives in women not receiving enzyme-inducing drugs. Considering the very large usage of oral contraceptives, the occurrence of an unplanned pregnancy may be relatively low but the consequences can be of great importance.

The increased clearance of lamotrigine by oral contraceptives can result in loss of seizure control or conversely, when oral contraceptives are discontinued, overdose of lamotrigine may occur. Antiepileptic drug-induced clearance and increased sex-hormone-binding globulin may result in lower free testosterone with resultant decrease in libido, potency and spermatogenesis.

Awareness of the issues

On the basis of accumulating reports of oral contraceptive drug failure the Epilepsy Foundation of America and the American College of Obstetricians and Gynocologists invited us to write a position paper on the problem of unplanned pregnancy associated with the use of antiepileptic drugs. The review with recommendations for management was published in the Journal of the American Medical Association, a publication with the widest physician distribution in the USA. It was estimated that failure rates of oral contraceptive in patients on carbamazepine, phenobarbital, phenobarbital and primidone were approximately five-fold the expected numbers. In contrast valproate use, a non-enzyme inducer, was not associated with increased risk of pregnancy. Breakthrough bleeding, an effect of low estrogen levels, was advised to be a warning sign of insufficient steroid effect. Increasing the strength of the oral steroid was recommended if continuation of an enzyme-inducing antiepileptic drug was deemed clinically advisable. A decade later Krauss etal. (1996) conducted a large survey of licensed neurologists and obstetricians and found approximately a quarter of those surveyed had a patient who had an unplanned pregnancy due to presumed oral contraceptive failure. A majority did not know which specific antiepileptic drugs were involved in interactions and did not make an effort to change the dose of the oral contraceptive. Only 4% of neurologists and none of the obstetricians knew the interactions of all of the antiepileptic drugs available at that time. Four years later Morrell et al. (1996) in a survey of health care professionals found only a small majority was aware of increased failure rates of oral contraceptives with antiepileptic drug use and only 27% could correctly identify the responsible drugs. At about the same time a survey in the UK revealed about half of women receiving antiepileptic drugs and oral contraceptive did not receive education about a possible interaction. It may be that the efforts to educate health care professionals and women about these interactions, the consequences and the options are now being heard. Aggressive marketing and educational efforts have been made by a number of organizations and especially by pharmaceutical companies that have introduced new antiepileptic drugs not having interactions with sex steroid hormones.

Mechanism of interactions and contraceptive failure

Specific antiepileptic drug interactions with oral contraceptives

Specific antiepileptic drug interactions with parenteral sex steroid administration

Parenteral administration of synthetic progestins and in particular, medroxyprogesterone (Depo-Provera-Ortho), has been available for decades. The route of administration has the advantage of steady release of hormone and minimizes the risk of non-compliance. We conducted studies giving medroxyprogesterone 10 mg three times daily in an effort to achieve amenorrhea to adequately assess any antiepileptic effect. Blood mycophenolic acid levels (determined by Upjohn Co.) were only about one-half (3-15 ng/ml) compared to normal controls (5-30 ng/ml). To assure compliance and avoid first-pass effect six patients were given 120 mg or 150 mg i.m. medroxyprogesterone (Depo-Provera, Upjohn Co.). Again concentrations in the blood ranged from 1-9 ng/ml (mean 2.6 µg/ml) compared to 5-10 µg/ml for controls. This suggested increased clearance secondary to use of enzyme-inducing antiepileptic drugs. Although these lower mycophenolic acid blood levels were found, the dose still sufficed to produce amenorrhea and inhibition of a rise in either estrogen or progesterone. A combination of medroxyprogesterone combined with a pro-drug of estradiol cypionate (Lunelle, Upjohn) has recently become available for contraceptive use. A dermal patch containing norelgestromin and ethinylestradiol (Ortho-Evra) has also become available. Although specific reports are not available, it can be inferred that increased clearance will occur with concomitant use of enzyme-inducing antiepileptic drugs with an increased risk of contraceptive failure. A subcutaneous implant slow-release formulation of levonorgesterol (Norplant) has been used with excellent contraceptive effects but numerous failures have been reported in women on enzyme-inducing antiepileptic drugs. Wyeth no longer manufactures this product. It can be predicted that increased clearance of all these parenteral products can be expected along with decreased efficacy despite avoiding first-pass effect.

Testosterone

Testosterone is produced in the Leydig cells of the testis. Testosterone is highly bound to proteins in the circulation primarily to sex hormone-binding globulin. Testosterone is metabolized to dihydroxytestosteroe that is physiologically active. Conversion to estrogen occurs in tissues in the body by an aromatase. Many investigators have reported that total and free testosterone levels are below normal in men with epilepsy. The enzyme-inducing antiepileptic drugs can not only increase clearance of testosterone, but increase sex hormone-binding globulin resulting in lower free testosterone levels. The effect of antiepileptic drug use on testosterone is more difficult to characterize clinically. Testosterone affects libido, potency and spermatogenesis and can lead to disturbances in these functions if amounts are deficient. However, determinants of libido and potency are multifactorial so attributing dysfunction to low testosterone associated with antiepileptic drug use is more difficult to establish. This is in contrast to the obvious endpoint of unplanned pregnancy with use of oral contraceptives. However, Fenwick et al. (1986) were able to correlate erectile dysfunction with low testosterone levels using penile tumescence measurements.

Oral contraceptive effect on lamotrigine

In a study of lamotrigine levels during delivery, in the neonate and during lactation, Ohman et al. (2000) found that lamotrigine levels at delivery were markedly lower than pre-pregnancy and 2-3 weeks post partum. Although many reasons can be found for a reduction in antiepileptic drug levels during pregnancy, such changes were not seen in patients also taking carbamazepine or combination of phenytoin. They concluded that, glucuronidation was induced by the elevated sex steroid hormones present during pregnancy. This finding was confirmed by Tran et al. (2002), who did lamotrigine clearance studies before, during and after delivery. They found a 65% increase in clearance during the first trimester of pregnancy. Sabers et al. (2001, 2003) reported a marked decrease (mean 49%) in lamotrigine levels after initiating oral contraceptive treatment in seven epilepsy patients and return after discontinuing the oral contraceptives. They concluded that the oral contraceptives act on the glucosonyltransferases which catalyzed the conjugation of lamotrigine with glucuronic acid. This initial observation was confirmed in a larger series of 56 women receiving oral contraceptives and lamotrigine. A two- to three-fold change in levels, was associated with adding or discontinuing oral contraceptives. In these two reports the changes resulting in a drop in lamotrigine levels were sometimes associated with an increase or breakthrough in seizures or adverse effects when lamotrigine levels rose with oral contraceptive discontinuation.

Management of women on oral contraceptive

The safest way of dealing with the problem of unwanted loss of oral contraceptive effectiveness is to avoid antiepileptic drugs that affect the clearance of sex steroids. Before the introduction of the newer antiepileptic drugs, valproate was the obvious choice. Unfortunately, if a woman elected to become or accidentally became pregnant, the concern about possible teratogenic-ity was of critical importance. Knowing the effect of an antiepileptic drug on sex hormone clearance allows some ability to predict the likelihood of success or failure of a contraceptive therapy. A second, but undependable, indicator of inadequate hormone effect is breakthrough bleeding. Sonnen (1983) observed 60-90% of 133 women taking oral contraceptives containing 30 or 50 mg of ethinylestradiol had breakthrough bleeding, whereas this bleeding occurred in only 6% of those taking valproate. A better measure of contraceptive effect is the absence of a rise is progesterone above 5 ng/ml during the luteal phase of the menstrual cycle. The limitation to this method of detection of a contraceptive steroid effect is timing of the day of the blood sample for analysis, unless samples are drawn every few days.

A change is the strength of the oral contraceptive pill may compensate for increased clearance by the enzyme-inducing antiepileptic drugs and provide a sufficient amount in the blood to allow adequate protection.

It is possible that increasing the quantity of ethinylestradiol from 20 or 30 µg to 50 µg (and the combined progestin) is insufficient. Krauss et al. (1996) pointed out that two of five unplanned pregnancies they observed were taking an oral contraceptive containing 50 µg of ethinylestradiol. Sonnen (1983) observed that increasing the dose to 75 µg corrected breakthrough bleeding in his patients on enzyme-inducing antiepileptic drugs. Since the bioavailability of the oral contraceptives is so variable, it may be difficult to predict the dose needed to provide protection and patients need to be advised of this uncertainty.

Summary

Interactions occur between enzyme-inducing antiepileptic drugs and synthetic sex hormones used for contraception whether given orally or parenterally. The decrease in available hormones is sufficient to lead to contraceptive failure. Antiepileptic drugs also lower free testosterone levels in men and may contribute to problems with libido, potency and fertility. These effects are not seen with use of valproate and the newer antiepileptic drugs, Gabapentin, levetiracetam, tiagabine and vigabatrin. Increased clearance and possible loss of contraceptive effect is found with felbamate and Oxcarbazepine. No effect is seen with use of topiramate at or below 200 mg/day. A reverse interaction is found with use of oral contraceptives. These hormones cause increased clearance and loss of effect of lamotrigine. Surveys indicate that a widespread lack of awareness of these issues persists despite original observations made more than three decades ago.

Most pharmacokinetic studies have suggested the estrogens and progestins in the oral contraceptive pill were cleared approximately twice as rapidly in women patients receiving enzyme-inducing antiepileptic drugs compared to normal controls.

The primary mechanism of action of the contraceptive sex hormones is thought to be due to inhibition of release of luteinizing hormone by the progestin and prevention of ovulation. The critical concentration of hormones needed to have this effect is not predictable, but the evidence that doubling of hormone clearance has been associated with contraceptive failure indicates a concentration below which failure is possible or likely.

The concentration of contraceptive sex hormones in the blood and brain is determined by a number of pharmacokinetic factors. After oral intake, a significant first-pass effect occurs especially for the estrogen component (usually ethinyl estradiol or mestranol). Some conjugation with sulfates and glu-curonides occurs in the gut as well as hepatic hydroxylation to inactive polar metabolites. Enterohepatic recirculation of these products may result in change back to ethinylestradiol and reabsorption into the blood. These multiple variables make uncertain the final quantity reaching the general circulation. The major hepatic biotransformation of the estrogen is by the CYP3A4 iso-enzyme system. Mestranol is converted to ethinylestradiol, the active drug, by demethylation thought to involve the CYP2C9 isoenzyme. Even in normal women from different populations bioavailability of ethynylestradiol (ethinyl) was found to vary up to ten-fold. (It is unclear if compliance could be assured.) The synthetic progestins used in older oral contraceptive combination pills are norethin-drone and levonorgestrel. More recent formulations have included other progestins, norgestimate (converted in part to levonorgesterol), degestrel and gestodene. The synthetic progestin, medroxyprogesterone, has been and continues to be used extensively although primarily in parenteral formulations. The progestin metabolism is less well defined than the estrogens but conjugation, oxidation and reduction all occur and can be induced by the antiepileptic drugs.

Carbamazepine, phenobarbital, primidone (that is metabolized to phenobarbital) and combination of phenytoin have inductive effects on the CYP isoenzymes as well as conjugation involved in sex steroid metabolism. Although having lesser inductive effect, felbamate, Oxcarbazepine and topiramate can all increase clearance of sex steroid hormones in contraceptive preparations. In contrast, valproate as well as many of the newer antiepileptic drugs, gabapentin, lamotrigine, levetiracetam , tiagabine and vigabatrin have no effect on sex steroid clearance.

The progestins undergo both oxidation and reduction as well as conjugation after entering the circulation. First-pass effect is much more extensive for norethindrone than levonorgesterol. Since the older enzyme-inducing antiepileptic drugs increase clearance of these sex steroids the amount circulating in the blood and brain may reach levels too low for the progestin to inhibit ovulation, especially in formulations containing 50 µg of ethinylestradiol or less (and comparable but higher doses of the pro-drug, mestranol) and 1 mg of levonorgesterol. After entering the circulation the sex steroids are protein bound. The progestins are extensively bound to sex hormone-binding globulin. This is relevant because the enzyme-inducing antiepileptic drugs increase the amount of sex hormone-binding globulin, which in turn effectively reduces the free, pharmacologically active, circulating progestin. This represents a second mechanism for oral contraceptive failure.

Among the older antiepileptic drugs, carbamazepine, phenobarbital, primidone and combination of phenytoin have specifically been found to increase clearance of the oral contraceptive sufficiently to reduce sex hormone levels by approximately 50% whereas valproate had no such effect. Among the new antiepileptic drugs introduced since the 1990s Gabapentin, lamotrigine, levetiracetam, tiagabine and vigabatrin have been studied and found to have no significant effect on clearance of the oral contraceptives. In addition progesterone levels did not rise during the luteal phase in the studies of lamotrigine, levetiracetam or tiagabine, a finding that confirmed the prevention of ovulation. Felbamate administration had modest effect on clearance of ethinylestradiol but lowered the area under the curve of gestodene, a newer synthetic progestin, by 42%. However, progesterone levels did not rise during the luteal phase suggesting ovulation had been blocked. Somewhat surprisingly Oxcarbazepine, a relatively non-CYP-inducing newer antiepileptic drug, had clear effect on clearance of oral contraceptives. Even in doses as low as 600 mg/day the area under the curve of both ethinylestradiol and levonorgestrol were reduced by 47%. The risk of contraceptive failure must be considered in view of the interaction.

The interaction between topiramate and oral contraceptives is more complex. In an initial study Rosenfeld et al. (1997) studied the effect of topiramate 200,400 or 800 on the metabolism of oral contraceptives containing 35 |xg of ethinylestradiol and 1 mg of norethindrone. Although norethindrone was not affected, ethinylestradiol clearance increased 15-33%. A follow up study was done using lower doses that are now more commonly used. Administration of 50,100 or 200 mg of topiramate in women using the same oral contraceptive did not significantly reduce concentrations of ethinylestradiol in the blood. In contrast a control group given 600 mg of carbamazepine exhibited an increase of oral clearance of norethindrone by 69% and ethinylestradiol by 127%. It was concluded a clinically significant interaction of topiramate with a ‘standard’ oral contraceptive did not occur at doses of 200 mg/day or less (Table Effect of antiepileptic drugs on oral contraceptive clearance and effectiveness).

Table Effect of antiepileptic drugs on oral contraceptive clearance and effectiveness

Increased	Equivocal	No effect
Carbamazepine	FLBa	Gabapentin
Phenobarbital	TPM6	lamotrigine
Combination of phenytoin		tiagabine
Primidone		ZNSC
Oxcarbazepine		valproate

Epilepsy in the mentally retarded differs from epilepsy in the mentally normal patient in relation to etiology, seizure types, epilepsy syndromes, choice of anti-epileptic drugs, identification of their side effects and treatment outcome. Consequently, a successful antiepileptic drug therapy is a demanding task in terms of choice of drug therapy, combinations of drugs and side effects in mentally retarded patients compared with mentally normal people. Adverse effects and interactions between different antiepileptic drugs are a potential risk in the presence of many and difficult-to-treat seizure types, leading to frequent polytherapy. There is also an increased risk of interactions between antiepileptic drugs and other drugs because of the increased incidence of co-morbidity among these patients.

In patients who are handicapped or mentally retarded, there is no evidence that pharmacokinetic drug interactions per se are quantitatively or qualitatively different from those seen in otherwise normal epilepsy patients. However, it is the context of the treatment of their epilepsy that puts a different emphasis on the potential for interactions. These patients are characterized by an increased incidence of co-morbidity that may require treatment with other medications. Their epilepsies are generally more refractory to treatment and antiepileptic drug combinations are more likely to be used. Also, central nervous system toxicity of drugs may be more prominent in mentally retarded patients, and this may include neurotoxic pharmacodynamic interactions between antiepileptic drugs as well as pharmacodynamic interactions between antiepileptic drugs and other psychotropic drugs. As a group, these patients maybe particularly vulnerable to the problems associated with polytherapy. The main purpose of this chapter is not to provide an exhaustive discussion of possible pharmacokinetic interactions that are discussed elsewhere in this book, but to emphasize the context in which pharmacokinetic and pharmacodynamic interactions are likely to occur during the treatment of epilepsy in handicapped and mentally retarded patients.

Epidemiology of epilepsy in the mentally retarded

Problems in diagnosing epilepsy

The diagnosis of epileptic seizures may be difficult in mentally retarded patients, because they cannot in many cases express themselves and therefore fail to tell about their perceived symptoms. Also, in these patients, motor automatisms are not easily distinguished from stereotypic movements, nor are nocturnal seizures easy to separate from parasomnias. Table Differential diagnosis of non-epileptic seizures in the mentally retarded lists the most important non-epileptic conditions which may lead to a misdiagnosis of epilepsy.

Intractability of seizures

The main groups of reasons for intractability of seizures are related to actions by the physician, to the patient, to the epilepsy itself and to the drugs (Table Intractability of epilepsy in the mentally ret). The type of epilepsy may be a priori intractable. Epileptic and non-epileptic seizures may be intermingled in the same patient. Certain antiepileptic medications, at therapeutic or at toxic doses, may cause or aggravate seizures. Remote symptomatic etiology, abnormal neurological status, occurrence of status epilepticus and poor short-term effect of drug therapy have been shown to be independent predictors of intractability.

Table Differential diagnosis of non-epileptic seizures in the mentally retarded

Cardiovascular mechanisms

Infantile syncope

•  Breath-holding spells

- Cyanotic infantile syncope

- Reflex anoxic seizures

• Syncope in older children

Paroxysmal movement disorders

Infantile jitteriness

Benign myoclonus of early infancy

Hyperekplexia

Gastroesophageal reflux

Paroxysmal dystonia/choreoathetosis

Shuddering attacks

Stereotypic movements

Alternating hemiplegia of childhood

Masturbation

Stool withholding activity and constipation

Psychological disorders

Psychogenic or pseudoseizures

Hyperventilation

Munchhausen by proxy

Migraine and migraine equivalents

Recurrent abdominal pain

Basilar migraine

Sleep disorders

Arousal disorders

REM sleep disorders

Drug interactions and adverse effects

Outcome of epilepsy in the mentally retarded

The outcome of drug therapy may be difficult to assess in the mentally retarded, for example in patients with infantile spasms. Video electroencephalograph monitoring may be needed for this purpose. Most of the few population-based studies dealing with the prognosis of epilepsy in the mentally retarded show a less favorable seizure outcome (seizure freedom in 38-46%) than in mentally normal patients (65-89%). In another recent long-term follow-up prospective study, 34% of patients with epilepsy and mental retardation and 67% of patients with uncomplicated epilepsy became seizure-free. The prognosis is better, the higher the intelligence level. Additional predictors of poor outcome are symptomatic etiology, association of cerebral palsy and perinatal brain injury.

Conclusion

Epilepsy is a common concomitant disorder in people with mental retardation. The diagnosis of epilepsy may be more difficult, because epilepsy in the mentally retarded often presents with several seizure types. The differential diagnosis between epileptic and non-epileptic events may also at times cause difficulties. In many patients, epileptic and non-epileptic seizures may co-occur. The effects of medication are difficult to evaluate, not least due to impaired abilities of these individuals to express themselves about perceived side effects. Video electroencephalograph monitoring may be needed. The responses to antiepileptic drugs may be different from that in mentally normal individuals. Numerous attempts in individual patients to attain seizure freedom or an acceptable level of seizure frequency mostly result in polytherapy and increasing adverse effects. These side effects may result from a different susceptibility of the brain to the drugs, to pharmacokinetic interactions, or to a greater susceptibility to pharmacodynamic interactions. To avoid or minimize these effects, the drugs should be as few as possible and a conversion to monother-apy with a broad-spectrum drug should be preferred when feasible. This seems to be particularly important in patients with mental retardation. The impact of the newer antiepileptic drugs may consist of a better tolerability with fewer interactions.

Epilepsy occurs in approximately 15% of patients with mild mental retardation (IQ 50-69) and 30% of those with severe mental retardation (IQ < 50). In institutionalized patients with mostly severe or profound mental retardation, the prevalence of epilepsy ranges from 35% to 60%. The age at the onset of the epilepsy does not differ from that in the general population. However, children with severe mental retardation were found to have a significantly earlier seizure onset than those with a mild mental retardation.

Table Occurrence of epilepsy in certain syndromes with MR shows several lesional, developmental, chromosomal and metabolic conditions in which epilepsy is associated in up to 100% of the cases. The etiology of severe mental retardation is reportedly prenatal in 55-78%, perinatal in 8-15%, postnatal in 1-12% and unknown in 13-22%. In patients who have a mild mental handicap, the corresponding figures are 23-43%, 7-18%, 4-5% and 43-55%, respectively. In many patients, the etiology is still unknown but probably prenatal.

Table Occurrence of epilepsy in certain syndromes with MR

Syndrome	Prevalence (%)
Cerebral palsy and MR	28-38
Mitochondrial disorders	96-100
Polymicrogyria	90
Tuberous sclerosis	90
Chromosomal anomalies
Angelman syndrome	84-90
Rett syndrome	75-80
Wolf-Hirschhorn syndrome	70
Fragile-X syndrome	25
Prader-Willi syndrome	15-20
Down syndrome	6-12
Klinefelter syndrome	2-10
Metabolic disorders
Peroxisomal diseases	80
Krabbe’s disease	50-75
Biotinidase deficiency	50-75
Disorders of urea cycle	60

Table Failure to recognize epileptic seizures in the mentally retarded

Seizures with vertigo

Seizures with paresthesias

Seizures with visceral disturbances

Seizures with headache

Seizures with loss of emotional control

Partial seizures with other clinical manifestations

Supplementary sensorimotor area seizures

Simple partial seizures

Absence seizures

Drop attacks

Automatisms

Most of the untoward effects are not as readily recognized in mentally retarded as in mentally normal patients. These patients may also be at higher risk for certain adverse effects of antiepileptic therapy, such as reduced bone density. Pharmacokinetics of antiepileptic drugs maybe affected in many ways. Administration of the drugs maybe complicated by the reluctance of the patient to take the pills, or decreased absorption due to slow bowel movements and constipation. Elimination of drugs metabolized by the liver may also be altered due to changes in genetic capacity, especially in inborn errors of neurometabolism involving the liver.

Table Intractability of epilepsy in the mentally retarded

Physician related

Incorrect diagnosis

Misclassification of epilepsy

Failure to recognize all seizure types

Failure in choice of drug

Failure to recognize seizure freedom

Patient related

Non-compliance

Epilepsy related

Severe early infantile encephalopathies

Minor motor seizures

Complex partial seizures

Atonic seizures

Multiple seizure types

Organic etiology of epilepsy

Progressive etiology of seizures

Non-epileptic seizures

Concomitant non-epileptic seizures

Drug related

Problems in ingestion of drug

Lack of good early effect of therapy

Side effects of single drug therapy

Side effects of polytherapy

Drug interactions

Deviating drug kinetics

Epilepsy in mental retardation commonly presents with several seizure types, drug resistance, concomitant psychiatric symptoms and syndromes with various enzyme abnormalities, which increase the risk of interactions. Often, polytherapy in mentally retarded patients with epilepsy can be reduced successfully. In a 10-year study in 244 institutionalized patients, the percentage of patients receiving monotherapy could be increased from 36.5% to 58.1% with no observed loss in seizure control. Whenever polytherapy is reduced, it is important to keep in mind that existing pharmacokinetic interactions are reversible upon removal of the drug responsible for the interaction.

Phenobarbital

Phenobarbital (and other barbiturates) has been used for almost one century for its good anticonvulsive efficacy. Phenobarbital (and primidone, the main active metabolite of which is phenobarbital) is considered to typically affect cognition, behavior and affect in mentally normal people. Combination of valproate with phenobarbital therapy results in elevated phenobarbital levels, due to inhibition of phenobarbital hydroxylation, with subsequent somnolence and even coma or hyperkinesis, aggressive bursts and insomnia. Inversely, phenobarbital accelerates the metabolism of valproate, thus lowering valproate levels in relation to the dose. The metabolism of cimetidine, used against peptic ulcer, which is not so uncommon in the mentally retarded, may be induced by phenobarbital with subsequent decreased blood levels. Because of its potential adverse effects, phenobarbital cannot be recommended as the first or second choice of drug for epileptic seizures associated with mental retardation.

Phenytoin

Along with phenobarbital, phenytoin was for decades the most important tool against seizures in the mentally retarded. Phenytoin therapy is not easily managed because of its saturation kinetics, marked differences in attaining steady-state levels in the blood and in other features of metabolism, and certain pharmacokinetic interactions which may in some cases result in toxic levels of phenytoin. Combined with primidone, phenytoin may cause phenobarbital intoxication by causing a marked rise in the ratio of phenobarbital to primidone.

The most serious groups of side effects include neurological adverse effects. Brain damage, which is commonly associated with mental retardation, and phenytoin in polytherapy further increase the risk for neurological adverse effects at therapeutic or even low levels of plasma phenytoin. A chronic and in the mentally retarded often irreversible syndrome of phenytoin encephalopathy was seen in 28%.

Phenytoin can no longer be recommended as the first or second drug of choice against epileptic seizures associated with mental retardation. This is particularly true when the patient has primary locomotion disorder or evidence of cerebellar disease.

Valproate

Valproate is a major antiepileptic drug with a broad spectrum, which is an advantage because it can cover several types of seizure so typical of the many mentally retarded. Seizure freedom is achieved by 20-70% of children with mental retardation and infantile spasms, and one-fifth of those with Lennox-Gastaut syndrome become seizure-free on a high-dosage valproate monotherapy. Valproate may have a clinically significant displacing effect on phenytoin and can cause phenytoin intoxication due to high free levels of phenytoin, even in the presence of therapeutic total levels. Valproate can significantly elevate levels of phenobarbital (also derived from primidone), ethosuximide and lamotrigine. The risk of death from liver failure is highest in children who are less than 2 years of age, especially among those with mental retardation, genetic metabolic disorders, brain injury or a family history of severe hepatic disease and/or who are receiving valproate in polytherapy.

Carbamazepine

Carbamazepine is effective against focal and generalized seizures. It is not effective against atypical absence, atonic and myoclonic seizures, and may even cause or increase these seizures, which are common in mentally retarded patients. Neurotoxicity is for the most part dose related. Though negative behavioral effects are in general fewer on carbamazepine than on phenytoin, phenobarbital or primidone, they may occur in mentally retarded patients and particularly in those with brain damage and those with pre-existing behavioral problems.

Carbamazepine levels are lower but carbamazepine-epoxide concentrations are higher in combination therapy with phenobarbital, phenytoin, primidone and valproate than in monotherapy. But carbamazepine and epoxide levels do not appear to be affected by newer anticonvulsants. Increasing displacement of carbamazepine from plasma proteins increases free fraction of carbamazepine during valproate co-medication. In case of co-medication with felbamate, lamotrigine, phenobarbitone, phenytoin, primidone, progabide and valnoctamide, carbamazepine-epoxide concentrations may reach toxic levels. Carbamazepine combined with valproate appears to have synergistic effects in frontal and temporal focal seizures.

Oxcarbazepine

Oxcarbazepine is similar to carbamazepine in its mode of action and efficacy against epileptic seizures. Few data are available on its efficacy in people with mental retardation. Given as adjunctive therapy for difficult-to-treat patients with mental retardation, a 50% or greater decrease in seizure frequency has been achieved in 50-60% of patients. The tolerability of oxcarbazepine is better and interactions are less frequent than those observed with carbamazepine, with the exception of higher frequency of hyponatremia. Electrical status epilepticus in sleep may occur during oxcarbazepine therapy in the mentally retarded. Oxcarbazepine has not shown any significant auto induction or interactions with other drugs, and may therefore be a useful drug for polytherapy in the treatment of difficult-to-treat seizures.

Benzodiazepines

Benzodiazepines are in most cases used as an adjunctive therapy, for example in children with Lennox-Gastaut syndrome or other epilepsy types with mental retardation. Clinically relevant interactions of benzodiazepines are rare, if any. In some patients, however, adjunctive therapy with clonazepam may cause toxic levels of phenytoin. The incidence of tolerance is higher in patients with clonazepam-treated West syndrome or Lennox-Gastaut syndrome than in epilepsy with typical absence seizures. Interactions with other drugs are based on pharmacodynamic influences. A combination with other central nervous system-depressant drugs may increase depression.

Vigabatrin

Vigabatrin proved to be an efficient drug against difficult-to-treat seizures in people with mental retardation and particularly in children with infantile spasms, with a 50% or greater decrease in seizure frequency in two-thirds. Vigabatrin does not cause excessive behavioral disturbances in mentally retarded patients. Hyperactive agitation or aggression, on the other hand, have been observed in up to 15-26% of pediatric patients. Myoclonic jerks may be provoked by vigabatrin, necessitating discontinuation of the drug.

The good efficacy of vigabatrin on seizures is shadowed by recent observations of visual field constriction, which occurs in one-third, is caused by accumulation of vigabatrin in the retina, and appears irreversible. The benefits, however, outweigh the risks and the therapy can be continued under strict clinical control. This is particularly true for infantile spasms due to tuberous sclerosis. Vigabatrin has not been found to be involved in any pharmacokinetic interaction.

Lamotrigine

The antiepileptic efficacy of lamotrigine is similar to that of other major antiepilep-tic drugs in placebo-controlled studies. In a retrospective evaluation of 44 institutionalized patients with mental retardation, lamotrigine, added to other antiepileptic drug therapy, decreased seizure frequency by 50% or more in 55% of the patients with mental retardation. Addition of lamotrigine to carbamazepine may accentuate or cause carbamazepine side effects, such as dizziness, diplopia and sedation which are subjective symptoms, and may present as behavioral disturbances in the mentally retarded. The most important effect of other antiepileptic drugs is inhibition of lamotrigine metabolism by valproate and the acceleration of lamotrigine metabolism by enzyme-inducing antiepileptic drugs. Methsuximide lowers lamotrigine to a clinically significant extent and this must be considered in the dosing of lamotrigine. Several other papers have reported favorable effects on seizure frequency, cognition and behavior, and quality of life and less successful involvement of behavior.

Gabapentin

Gabapentin has been shown to be effective as an adjunct on refractory partial-onset seizures. Eleven (42%) of 26 children with mental retardation experienced a 50% or greater decrease in seizure frequency on gabapentin add-on therapy. The response did not differ from that of mentally normal study subjects. Gabapentin has an effect on focal seizures but not on myoclonic, atonic or absence seizures. With regard to adverse effects, 16% of 110 mentally retarded people showed aggressiveness, 15% had increase in seizure frequency, and 9% had ataxia or lethargy. Mikati et al. (1998) reported behavioral adverse changes in 58% of 26 mentally retarded children. In one study, gabapentin was shown to extend the elimination half-life of felbamate by a 50%. No other interactions involving gabapentin have been described.

Tiagabine

Tiagabine, another GABAergic antiepileptic drug, is in many respects similar to gabapentin. According to a meta-analysis, the chance for at least 50% reduction in seizure decrease was three-fold with add-on tiagabine than without. No separate data on mentally retarded patients are so far available. Lack of clinically relevant cognitive adverse effects may encourage tiagabine trials in mentally retarded individuals. On the other hand, dizziness, asthenia, nervousness, abnormal thinking, depression, aphasia and abnormal abdominal pain are significantly more common compared with placebo. Three patients with tiagabine-associated encephalopathy have been reported. The elimination of tiagabine is accelerated by enzyme-inducing antiepileptic drugs, but tiagabine does not seem to affect the pharmacokinetics of any other drugs.

Topiramate

The meta-analysis of six pooled double-blind, placebo-controlled studies on the effectiveness of topiramate on partial-onset seizures showed that the seizure frequency had decreased by at least 50% in 43% of 527 patients, compared with 12% on placebo. In 98 patients with Lennox-Gastaut syndrome, topiramate decreased the frequency of both drop attacks and generalized tonic-clonic seizures in every third patient, whereas the same effect occurred in only every eighth patient on placebo. Interactions affecting other drugs are negligible due to predominantly renal excretion and low protein binding, but the half-life of topiramate is shortened by enzyme-inducing antiepileptic drugs such as carbamazepine, phenytoin, phenobarbital and primidone. Side effects during topiramate therapy include dizziness, fatigue, visual disturbance, diplopia, ataxia, psychomotor slowing, weight decrease and, in rare cases, renal stones, and hypohidrosis.

Felbamate

When felbamate was launched, it soon appeared effective in patients with, among other conditions, the Lennox-Gastaut syndrome and infantile spasms. Dangerous adverse effects, mainly aplastic anemia and liver failure, have greatly restricted its use. Felbamate has also several interactions with other drugs. It increases significantly carbamazepine epoxide, phenobarbital, phenytoin and valproate plasma levels and decreases total carbamazepine concentrations. Both carbamazepine and phenytoin induce felbamate metabolism and hence increase its clearance. For the moment, felbamate can only be used in well-selected patients under strict and individualized control.

Zonisamide

Few data are available on the efficacy of zonisamide in patients with mental retardation and refractory seizures. Iinuma et al. (1998) reported a more than 50% decrease in seizure frequency in 67% of mentally normal and in 41% of retarded patients. Adverse effects were as common in the retarded as in the mentally normal children (27% vs. 30%). The most common untoward effect was aggravation of seizures, which was more common in the mentally normal than in the retarded (28% vs. 18%), and drowsiness. No data on antiepileptic drug interactions were reported. Zonisamide does not induce or inhibit other drugs but its half-life is shortened in humans by enzyme-inducing drugs such as carbamazepine, phenobarbital, phenytoin and valproate.

Levetiracetam

Levetiracetam, a novel broad-spectrum antiepileptic drug, is effective against focal and generalized seizures. In three multicenter, double-blind, placebo-controlled studies, about one-third of 904 patients with partial-onset, drug-refractory seizures achieved an at least 50% overall decrease in seizure frequency. The tolerability was good. According to existing data, no interactions can be anticipated in clinical use. The place of levetiracetam in the treatment of seizures in the handicapped and mentally retarded patient remains to be established.

Scope of the problem

Women with epilepsy require chronic antiepilepsy drugs to prevent seizures, maintain their function and health. Unlike most young women they are unable to discontinue their medications if they become pregnant, for to do so increases their risk of seizures, personal injury, miscarriage and developmental delay in the offspring. With a prevalence of between 0.6% and 1.0% and an estimated 40% of those with epilepsy being women of childbearing years one can see that the potential public health impact is significant. Most women with epilepsy have healthy children but there is an increased risk for congenital malformations, fetal loss, developmental delay and neonatal hemorrhage. Maternal epilepsy is a contributor but the use of antiepileptic drugs is a significant confounder. To make matters more complicated 86% of pregnant women take medications during pregnancy. A survey by the World Health Organization of 14 778 women in 22 countries reported that of the 86% of women taking medications during pregnancy the average number of prescriptions was 2.9 (range of 1-15). This study did not evaluate over-the-counter medications. The preponderance of prescriptions, 73%, were written by obstetricians.

When evaluating antiepileptic drug use in pregnancy one is hampered by the lack of knowledge of specific co-medications, even though it is clear that this is a common event. While monotherapy with antiepileptic drugs is a goal of epilepsy management, it is not always an obtainable one. Polytherapy is also more common with the “newer” post-1993 introduction antiepileptic drugs, because all were initially approved for use as adjunctive therapy. The pregnancy outcome of greatest interest is congenital malformations. While there are substantial data on this outcome, most are in the form of case series and case reports, and accurate rates and risks cannot be determined. Other adverse outcomes are at least as common in terms of incidence (developmental delay, fetal loss), but have received significantly less attention.

Let us review some of the clinically important issues surrounding pregnancy and antiepileptic drug exposure.

Antiepileptic drugs and hormonal contraceptives

Maternal complications associated with antiepileptic drug

Seizures not infrequently worsen during pregnancy. One-quarter to one-third of woman with epilepsy will have an increase in seizure frequency during pregnancy. This increase is unrelated to seizure type, duration of epilepsy, or seizure frequency in a previous pregnancy. In a recent series of 215 pregnancies in woman with epilepsy an increase in seizures during the first trimester occurred in 30% of monotherapy- and 43% of polytherapy-treated women. One in 8 or 12.5% had to be hospitalized for their seizures during the pregnancy.

Plasma concentrations of anticonvulsant drugs decline as pregnancy progresses, even in the face of constant and in some instances increasing doses. Although reduction of plasma drug concentration is not always accompanied by an increase in seizure frequency, virtually all women with increased seizures in pregnancy have subtherapeutic drug levels. The decline of anticonvulsant levels during pregnancy is largely a consequence of decreased plasma protein binding, reduced concentration of albumin and increased drug clearance. The clearance rates are greatest during the third trimester.

Kaarkuzhali and colleagues (2002) found that a majority of their pregnant patients on carbamazepine, phenytoin or phenobarbital required numerous dose adjustments during pregnancy to maintain therapeutic levels. Fifty percent of the pregnancies had breakthrough seizures when the levels fell below the therapeutic range. It is therefore imperative to monitor antiepileptic drug levels at least monthly and adjust dosage to maintain therapeutic levels. Table Pharmacokinetic data for first generation antiepileptic drugs summarizes some of the pharmacokinetics of anticonvulsant drugs during pregnancy.

Less is known about the kinetics of the newer antiepileptic drugs in pregnancy. A report demonstrates that lamotrigine clearance increases by >50% during pregnancy and that the clearance changes occur relatively early in pregnancy. Eleven of 12 pregnancies required increased doses of lamotrigine to maintain therapeutic levels during pregnancy.

Table  Pharmacokinetic data for first generation antiepileptic drugs

Anticonvulsant	Percent decrease
	Total level by third trimester	Percent free fraction
		Normal	Maternal	Neonatal
Carbamazepine	40	22	25	35
Ethosuximide	?	90	?	?
Phenobarbital	55	51	58	66
Phenytoin	56	9	11	13
Primidone	55	?	?	?
Derived phenobarbital	70	75	80	?
Valproate	50	9	15	19

Polycystic ovaries

A great deal of confusing literature has been written about the effect of antiepileptic drugs on the development of polycystic ovaries. Ovarian cysts are found in approximately 6.6% of women of childbearing age. Most of these cysts (over 80%) will disappear within 3 months. Multiple or polycystic ovaries are more commonly found in women taking hormonal contraceptives with progesterone, and women who are infertile. The rates vary but average between 10% and 20%.

The polycystic ovarian syndrome is a specific disturbance of neuro-endocrine function defined as no or irregular menses (oligomenorrhea), elevated levels of male sex steroid hormones (hyperandrogenism) without evidence of other disturbances such as hyperprolactinemia, thyroid dysfunction or 21-hydroxylase deficiency. It is uncommon as occurs in approximately 6.5% of women of reproductive age. It is associated with sustained release of gonadotropic-releasing hormone and lutenizing hormone, and affected women are often overweight, have elevated serum lipids and are less sensitive to insulin. Polycystic ovarian syndrome is also seen in higher than expected rates in mothers (24-52%) and sisters (32-66%) of women with this disorder leading some to believe that it is a genetic disorder.

Isojarvi and colleagues (1993) demonstrated an excess of menstrual abnormalities inWWE taking valproic acid (valproate) compared to other antiepileptic drugs. They also stated that 80% of women taking valproate before the age of 20 developed polycystic ovaries. Despite this observation other researchers have not found a consistent association between specific antiepileptic drug or epilepsy types and polycystic ovarian syndrome. Women with bipolar disorder are often treated with valproate and do not have an increase in polycystic ovarian syndrome.

Fetal complications associated with antiepileptic drug

Mechanisms of teratogenicity

A hypothesis that metabolites of antiepileptic drugs are responsible for malformations has been developed on the basis of the following observations:

1   an arene oxide metabolite of phenytoin or other antiepileptic drug is the ultimate teratogen;

2  a genetic defect in epoxide hydrolase (arene oxide detoxifying enzyme) system increases the risk of fetal toxicity;

3  free radicals produced by antiepileptic drug metabolism are cytotoxic;

4  a genetic defect in free radical scavenging enzyme activity increases the risk of fetal toxicity.

Epoxides

A large number of drugs can be converted into epoxides, in reactions that are catalyzed by the microsomal monoxygenase system. Arene oxides are unstable epoxides formed by aromatic compounds. Various epoxides are electrophilic and may elicit carcinogenic, mutagenic and other toxic effects by covalent binding to cell macromolecules. Epoxides are detoxified by two processes:

1   conversion to dihydrodiols catalyzed by epoxide hydrolase in the cytoplasm,

2  conjugation with glutathione in the microsomes.

Epoxide hydrolase activity has been found in the cytosol and the microsomal sub-cellular fraction of adult and fetal human hepatocytes. Epoxide hydrolase activity in fetal liver is lower than that of adults. One-third to one-half of fetal circulation bypasses the liver, resulting in higher direct exposure of extra-hepatic fetal organs to potential toxic metabolites.

Phenytoin teratogenicity

Free radical intermediates of antiepileptic drugs and teratogenicity

Some drugs are metabolized or bioactivated by co-oxidation during prostaglandin synthetase (PGS)-catalyzed synthesis of prostaglandins. Such drugs serve as electron donors to peroxidases, resulting in an electron-deficient drug molecule, which by definition, is called a free radical. In the search for additional electrons to complete their outer ring, free radicals can covalently bind to cell macromolecules, including nucleic acids (DNA, RNA), proteins, cell membranes and lipoproteins to produce cytotoxicity.

Phenytoin is co-oxidated by PGS, thyroid peroxidase and horseradish peroxidase producing reactive free radical intermediates that bind to proteins (Kubow and Wells, 1989). Phenytoin teratogenicity can be modulated by substances that reduce the formation of phenytoin-free radicals. Acetylsalicylic acid irreversibly inhibits PGS, caffeic acid is an antioxidant, alpha-phenyl-AT-t-butylnitrone (PBN) is a free radical spin-trapping agent. Pretreatment of pregnant mice with these compounds reduces the number of cleft lip or pathies secondary to phenytoin in their offspring.

Conjugation with glutathione is believed to detoxify free radical intermediates by forming a non-reactive conjugate. N-acetylcysteine (NAC1) a conjugation with glutathione precursor, decreases phenytoin-induced orofacial clefts and fetal weight loss in rodents. l,3-bis(2-chloroethyl)-l-nitrosourea (BCNU) inhibits conjugation with glutathione reductase, an enzyme necessary to maintain adequate cellular conjugation with glutathione concentrations, and increases phenytoin embryopathy at doses at which BCNU alone has no embryopathic effect. The metabolism of phenytoin or other antiepileptic drugs to free radical intermediates may be responsible for the teratogenicity seen in infants of mothers with epilepsy. Twenty-six children with myelomeningocele and their parents were studied by Graf and colleagues (1995). They were found to have significantly lower antioxidant enzymes, particularly conjugation with glutathione peroxidase, than controls.

Neural tube defects and antiepileptic drugs

Folate deficiency as a potential mechanism of antiepileptic drug teratogenicity

Folate is a coenzyme necessary for the development of white and red blood cells, and proper function of the central nervous system. Normal concentrations are typically measured in the serum (plasma folate = 6-20 ng/ml) and erythrocytes (red blood cell folate, RBCF = 160-640 ng/ml). Low levels of folate are associated with hyperhomocysteinemia and concentrations required to prevent this are 6.6 ng/ml for SF and 140 ng/ml for RBCF.

Deficiencies of folate have been implicated in the development of birth defects. Dansky et al. (1987) found significantly lower blood folate concentrations in women with epilepsy with abnormal pregnancy outcomes. Co-treatment of mice with folic acid, with or without vitamins and amino acids, reduced malformation rates, and increased fetal weight and length in mice pups exposed to phenytoin in utero. Biale and Lewenthal (1984) reported a 15% malformation rate in infants of mothers with epilepsy with no folate supplementation, whereas none of 33 folate-supplemented children had congenital abnormalities. Eight trials have demonstrated that pre-conceptual folate reduces the risk of recurrence of neural tube defects in women with a previous affected pregnancy.

Table Pre-conceptual folate, after Lewis etal. (in press)

Study type	N	Dose of folate	Results
Non-randomized, controlled	Fully supplemented = 454 Partially supplemented = 519 Unsupplemented =114	0.36 mg	86% risk reduction
Non-randomized	Unsupplemented = 543 Supplemented = 421	0.36 mg	Risk reduction
Case-control	Case = 181 Control = 1480	Multivitamins with folate	60% risk reduction
Cohort	23 491	Multivitamins with folate	71% risk reduction
Randomized, double blind, controlled	1195	4.0 mg	72% risk reduction
Randomized, controlled	Case = 2420 Control = 2333	0.8 mg	No defects with folate s upplementation
Case-control	Case = 436 Control = 2615	?	60% risk reduction
Case-control	Case = 604 Control = 1658	Folate supplements	Folate did not decrease rates in women >70kg

Unfortunately pre-conceptual folate supplementation may not be protective for women with epilepsy. Craig and colleagues (1999) reported a young woman whose seizures were controlled for 4 years by 2000 mg of valproate a day. Though she took 4.0 mg of folic acid a day for 18 months prior to her pregnancy she delivered a child with a lumbosacral neural tube defect, a ventricular and atrial septal defect, cleft palate and bilateral talipes. Two Canadian women delivered children with neural tube defect despite folate supplementation. One taking 3.5 mg folic acid for 3 months prior to conception and 1250 mg of valproate aborted a child with lumbosacral spina bifida, Arnold Chiari malformation and hydrocephalus. A second woman who took 5.0 mg of folic acid had one spontaneous abortion of a fetus with an encephalocele and two therapeutic abortions of fetuses with lumbosacral spina bifida. These cases might have been predicted given the demonstrated failure of folate to reduce neural tube defect and embryotoxicity in vitro and in vivo in rodent models. In fact not all research supports the association with folate deficiency and malformations. Mills et al. (1992) found no difference between serum folate levels in mothers of children with neural tube defect and controls. A number of other studies also failed to demonstrate a protective effect of pre-conceptual folate. These studies are problematic due to small sample sizes, failure to document folate supplementation and recall bias in the retrospective investigation.

There is evidence to suggest that women with similar folate intake may have difference serum concentrations due to differences in folate metabolism. Absorption does not account for the difference in plasma concentration between cases and controls.

New antiepileptic drug in pregnancy

Syndromes of anomalies

Newer antiepileptic drug and anomalies

Neonatal complications associated with antiepileptic drug

A unique neonatal hemorrhagic phenomenon has been described in the infants of mothers with epilepsy. It differs from other hemorrhagic disorders in infancy in that the bleeding tends to occur internally during the first 24 h of life. It was initially associated with exposure to phenobarbital or primidone, but has subsequently also been described in children exposed to phenytoin, carbamazepine, diazepam, mephobarbital, amobarbital and ethosuximide. One group of investigators suggests that vigabatrin may also increase the risk of neonatal hemorrhage. Prevalence figures are as high as 30% but appear to average 10%. Mortality is high, over 30%, because bleeding occurs within internal cavities and is often not noticed until the child is in shock.

The hemorrhage is the result of a deficiency of vitamin K-dependent clotting factors II, VII, IX and X. Anticonvulsants can act like warfarin, and inhibit vitamin K transport across the placenta. This results in the increase in an abnormal pro-thrombin induced by vitamin K absence of factor II (PIVKA-II). Maternal coagulation parameters are invariably normal. The fetus, however, will demonstrate increased levels of PIVKA, diminished clotting factors, and prolonged prothrom-bin and partial thromboplastin times. PIVKA-II has been demonstrated in 54% of infants exposed to antiepileptic drugs in utero compared to 20% of controls (P = 0.01), and maternal vitamin K concentrations are lower in woman with epilepsy than those untreated though PIVKA is rarely detectable in mothers.

This phenomenon can be prevented by maternal ingestion of oral vitamin K in the last month of gestation. I use 10 mg/day of oral vitamin K. Routine intramuscular administration of vitamin K at birth is not adequate to prevent hemorrhage within the first 24 h of life.

The prevalence of antiepileptic drug-associated neonatal hemorrhage is unclear. One report states it is 1.6 times as common in infants of mothers with epilepsy as controls. A more recent prospective study followed 667 infants of mothers with epilepsy and 1334 controls and found neonatal bleeding in 5 of 667 (0.7%) infants of mothers with epilepsy and 5 of 1334 (0.4%) of controls. While more prevalent there was no statistical difference between the groups. The authors concluded that there was no increased risk for neonatal bleeding in the infants of mothers with epilepsy. I would point out that the sample size for a low frequency outcome such as this may need to be larger and there was clearly a trend for more bleeding in the infants of mothers with epilepsy.

Developmental complications associated with antiepileptic drug

Developmental delay

Infants of mothers with epilepsy have been reported to have higher rates of mental retardation than controls. This risk is increased by a factor of two- to seven-fold according to various authors. None of these studies controlled for parental intelligence, although differences in IQ scores at age 7 between groups of children exposed (full-scale IQ, FSIQ = 91.7) or not exposed (FSIQ = 96.8) to phenytoin reached statistical significance, the clinical significance of such difference is unknown.

We have found that infants of mothers with epilepsy display lower scores in measures of verbal acquisition at both 2 and 3 years of age. Though there was no difference in physical growth parameters between infants of mothers with epilepsy and controls, infants of mothers with epilepsy scored significantly lower in the Bailey Scale of Infant Development’s mental developmental index at 2 and 3 years. They also performed significantly less well on the Bates Bretherton early language inventory (P =£ 0.02) and in the Peabody Picture Vocabulary’s scales of verbal reasoning (P =£ 0.001) and composite IQ (P =£ 0.01), and they displayed significantly shorter mean lengths of utterance (P =£ 0.001).

Polytherapy-exposed infants performed significantly less well on neuropsychometric testing than those exposed to monotherapy. Socioeconomic status had the strongest association with poor test scores, but maternal seizures during pregnancy was also a significant risk factor.

Leonard et al. (1997) has in part addressed the question of whether maternal seizures or in utero exposure to antiepileptic drugs are responsible for the developmental delay seen. A group of children of mothers with epilepsy followed to school age were found to have a rate of intellectual deficiency of 8.6%. The Wechsler Intelligence Scale for Children revealed significantly lower scores for children exposed to seizures during gestation (100.3), than for children whose mother’s seizures were controlled (104.1) or controls (112.9). All antiepileptic drugs are clearly not created equal and Koch and co-workers (1999) have demonstrated that primidone, particularly when used in polytherapy, is associated with lower Wechsler score of intelligence.

Conclusion

The potential interactions of antiepileptic drugs in pregnant women with epilepsy can be characterized by those effecting the mother, and those effecting the fetus. While pregnancy, maternal seizures and antiepileptic drugs pose risks for successful pregnancy outcome, the majority of patients can and do have healthy children. Physicians cannot eliminate risk, but can reduce it. Pre-conceptual folic acid is an approved intervention but may not prevent all malformations. Though there are no head to head studies of the safety of antiepileptic drugs in pregnancy some principles have been clearly established. Monotherapy is safer than polytherapy. Phenobarbital is no safer than, and probably more hazardous than, other antiepileptic drugs in monotherapy. Valproate has in addition to the underlying increased risk for malformations an additional risk for development of neural tube defects. The newer antiepileptic drugs have theoretical advantages over older ones in terms of malformations but the sample sizes collected to date are not adequate to determine relative safety. Malformations are not the only adverse outcome that one should be concerned about. Developmental delay is, in terms of magnitude, as significant as birth defects. There is no drug-specific syndrome of anomalies but a tendency for all antiepileptic drugs to cause facial dysmorphism, which is a relatively transient condition.

Given the nature of the data available to date clinical judgement in determining the most effective antiepileptic drug for the seizure type and using the lowest effective dose is still the best approach.

A discussion of pregnancy needs to be preceded by reviewing the problems of contraception. Oral contraceptives have not been associated with exacerbation of epilepsy. The effectiveness of hormonal contraceptives can, however, be reduced by enzyme-inducing antiepileptic drug (carbamazepine, phenytoin, phenobarbital, felbamate, topiramate). Hormonal contraceptives come in three formulations:

•  oral (estrogen-progesterone combinations or progesterone only);

•  subcutaneous (levonorgestrel) or intrauterine (progestasert) implants;

•  injectable (depoprovera).

All three forms can be adversely impacted by enzyme-inducing antiepileptic drug.

Antiepileptic drugs may lower concentrations of estrogens by 40-50%. They also increase sex hormone-binding globulin, which increases the binding of progesterone and reduces the unbound fraction. The result is that hormonal contraception is less reliable with enzyme-inducing antiepileptic drugs.

The low- or mini-dose oral contraceptives are therefore to be used with caution. As it is the progesterone not the estrogen that inhibits ovulation, using higher-dose estrogens alone may not be effective. The more rapid clearance of the oral contraceptive when used in conjunction with an enzyme-inducing antiepileptic drug will reduce the likelihood of unwanted side effects from higher-dose tablets.

Failure of implantable hormonal contraceptives has also occurred. Mid-cycle spotting or bleeding is a sign that ovulation is not suppressed. If this occurs alternative or supplementary methods of contraception are required. Contraceptive failure may not always be predictable, even when mid-cycle spotting does not occur. Failure of basal body temperature to rise at mid-cycle can be used to document ovulatory suppression.

Medroxyprogesterone injections should be given every 10 instead of 12 weeks to women on enzyme-inducing antiepileptic drug. This shorter cycle is less likely to result in unintended pregnancy.

For multiparous women with epilepsy, intrauterine devices may be an excellent contraceptive choice. Alternatively non-enzyme-inducing antiepileptic drugs may need to be considered (valproate, lamotrigine, gabapentin or zonisamide). A recent report suggests that topiramate at doses of <200 mg a day lacks enough enzyme induction to effect hormonal contraceptives. Higher doses however do reduce ethinyl estradiol concentrations by 18% on 200 mg, 21% with 400 mg and 30% with 800 mg of topiramate a day.

The importance of the potential impact of enzyme-inducing antiepileptic drugs cannot be underestimated. In a survey of 294 general practices in the General Practice Research Database, 16.7% of women aged 15-45 with epilepsy were taking an oral contraceptive. Two hundred were on an enzyme-inducing antiepileptic drug and 56% on low estrogen (<50 µg) hormonal contraceptives.

There has been at least one circumstance in which oral contraceptives effect antiepileptic drug concentration. Sabers and colleagues (2003) have demonstrated a marked reduction in lamotrigine concentrations when oral contraceptives are also taken. The average plasma concentration in 22 women on lamotrigine monotherapy and an oral contraceptive was 13µmol/l. In a similar group of women on lamotrigine monotherapy with no oral contraceptive use, the plasma concentrations averaged 28 µmol/l: a significant reduction in antiepileptic drug concentration of over 50%. It has been suggested that oral contraceptives may induce the metabolism of glucuronidated drugs such as lamotrigine.