The interactions between hepatic microsomal-inducing drugs [phenytoin, phenobarbital (phenobarbitone) and carbamazepine] and the combined oral contraceptive pill have been recognised for many years.[1,2]

An interaction between antiepileptic drugs (AEDs) and the combined oral contraceptive pill was first proposed when the dose of estradiol in the oral contraceptive pill was reduced from 100 to 50μg. There was a higher incidence of breakthrough bleeding and contraceptive failure among women with epilepsy compared with women in general.[1] The drugs implicated included phenobarbital, methylphenobarbital (methylphenobarbitone), phenytoin, primidone, carbamazepine and ethosuximide.[2] Phenytoin initially appeared to be the AED most commonly involved, but at this time it was also the most commonly prescribed AED. In the period from 1975 to 1984, 43 cases of contraceptive failure in women taking both the combined oral contraceptive pill and AEDs were reported to the UK Committee on Safety of Medicine. Phenytoin accounted for 25 cases, phenobarbital for 20 and carbamazepine for six (some patients were taking more than one AED). Prescriptions for carbamazepine at this date were about a third of those for phenytoin.[3]

This article reviews the literature on drug interactions observed between AEDs and oral contraceptives as well as other hormonal contraceptive methods. It also needs to be appreciated that AEDs are not only used to treat epilepsy; psychiatrists and consultants in pain relief commonly use them, and some AEDs are beginning to be used to treat migraine.

1. Mechanism of the Drug Interaction

There are two mechanisms that may explain contraceptive failure of hormonal contraception in women with epilepsy. Phenytoin, phenobarbital, primidone and carbamazepine are all potent inducers of hepatic microsomal drug-metabolising enzymes and induce the metabolism of estrogen and progestogens.[4] Phenobarbital, phenytoin and carbamazepine increase the capacity of sex hormone-binding globulin (SHBG) to bind progestogens, and therefore may reduce the free concentrations of progestogens in the plasma.[5]

The majority of AEDs are eliminated via hepatic metabolism. Medications that induce the hepatic cytochrome P450 (CYP) 3A4 isoenzyme accelerate the hydroxylation of estrogen to inactive metabolites. Hepatic microsomal enzyme induction can therefore decrease serum concentrations of the oral contraceptive pill. Enzyme induction has been demonstrated to be proportional to the dose of phenytoin, carbamazepine and phenobarbital.[6,7] The extent of enzyme induction in an individual patient is difficult to quantify as genetic and environmental factors as well as concurrent disease influence hepatic isoenzyme expression.[8–10]

Progestogens are highly bound to SHBG and albumin.[11] Phenobarbital, phenytoin and carbamazepine have been shown to increase concentrations of SHBG. It has been suggested that this increased binding capacity might lead to decreased free progesterone concentrations in plasma and contribute to contraceptive pill failure.[11,12] However, in theory the pharmacologically active free concentration of progestogen remains constant.[13]

2. Interactions Between the Combined Oral Contraceptive Pill and Antiepileptic Drugs (AEDs)

Table I provides a summary of the studies that have investigated the possibility of interactions between the combined oral contraceptive pill and AEDs.

Table I
figure Tab1

Summary of studies that have investigated interactions between antiepileptic drugs (AEDs) and the combined oral contraceptive pilla

Full size image

2.1 Older AEDs

2.1.1 Phenobarbital (Phenobarbitone)

In view of the anecdotal evidence of interactions between the combined oral contraceptive pill and AEDs,[14] a study was undertaken looking for possible interactions. This study involved four women who were taking the combined oral contraceptive pill before they developed epilepsy.[14] The concentrations of ethinylestradiol and norethisterone were studied prior to and during treatment with phenobarbital 30mg twice daily. Overall, statistically there was no change in ethinylestradiol or norethisterone concentrations. However, in two women there were significant falls in their ethinylestradiol concentration, from 104.8 ± 13.4 to 47.7 ± 13.9 ng/L, and norethisterone concentration, from 125.6 ± 2.38 ng/L to 50.3 ± 4.7 ng/L (p < 0.05). In one of the patients, this fall in ethinylestradiol concentration was accompanied by breakthrough bleeding.[14] Breakthrough bleeding occurring in the middle of a cycle of contraceptive use is usually due to a relative estrogen deficiency and is taken as a sign of incipient failure of contraception.[27] Plasma follicle-stimulating hormone concentrations were also measured, as it was anticipated that these would rise if there was an imminent loss of efficacy of the combined oral contraceptive pill. No rise was seen.[14]

An observational study by Sonnen[15] noted that 7 of 11 patients receiving phenobarbital monotherapy and the combined oral contraceptive pill had breakthrough bleeding.

It therefore is normally recommended that women taking phenobarbital take an oral contraceptive pill containing at least 50μg of estrogen.

2.1.2 Phenytoin, Carbamazepine and Valproic Acid (Sodium Valproate)

Further studies were undertaken looking at the pharmacokinetics of a single dose of Eugynon® 50Footnote 1 (50μg ethinylestradiol and 250μg levonorgestrel) before and during long-term treatment with carbamazepine, phenytoin and valproic acid in women with epilepsy.[16,17] Six patients received phenytoin (200 to 300 mg/day), four patients received carbamazepine (300 to 600 mg/day)[16] and six patients received valproic acid (400 mg/day).[17]

Treatment with phenytoin resulted in a significant reduction in the area under the plasma concentration-time curve (AUC) of both ethinylestradiol and levonorgestrel. On average there was a 50% reduction in the amount of the contraceptive steroids during phenytoin therapy.[16] All four patients receiving carbamazepine showed a significant fall in plasma concentrations of ethinylestradiol and levonorgestrel. The mean fall was about 50%, but there was a considerable variation; for example, the fall of ethinylestradiol concentrations ranged from 6 to 66%.[16]

There were no significant differences in the kinetics of ethinylestradiol and levonorgestrel before and during treatment with valproic acid.[17] Interestingly, peak concentrations of ethinylestradiol were higher during treatment with the drug (130.0 ± 17.5 ng/L compared with 85.5 ± 7.5 ng/L). Therefore, this theoretically should increase contraceptive efficacy.

In a study by Sonnen,[15] 32 women who were receiving a single AED and the combined oral contraceptive pill were observed with regard to breakthrough bleeding, a sign of incipient contraceptive failure. There were no episodes of breakthrough bleeding in seven patients on concurrent valproic acid therapy (dosage range from 600 to 1800 mg/day) over an observation period of 4 to 39 months. Four of 6 patients receiving carbamazepine and one of two patients receiving phenytoin had breakthrough bleeding.

Therefore, women taking phenytoin or carbamazepine need to take an oral contraceptive pill with at least 50μg of estrogen while those on valproic acid can take a normal-dose contraceptive pill.

2.1.3 Ethosuximide

Although five pregnancies had been reported in the literature in women taking ethosuximide and the combined oral contraceptive pill, these pregnancies are likely to be as a result of other AEDs that had been used concurrently with ethosuximide.[28] Suximides do not have enzyme-inducing properties so it is unlikely that there will be significant drug interactions with the combined oral contraceptive pill.[29,30] No formal interaction studies have been undertaken.

2.1.4 Clobazam, Clonazepam and Diazepam

There are no published studies looking at the effects of clobazam on the combined oral contraceptive pill. However, the current Summary of Product Characteristics for clobazam comments that drugs inhibiting the CYP enzyme system may enhance or prolong the effect of clobazam.[31] There is no mention of any interactions with the combined oral contraceptive pill.

No interactions between clonazepam or diazepam and the oral contraceptive pill are anticipated.[32]

2.2 Newer AEDs

2.2.1 Vigabatrin

A study to assess the potential for interaction between vigabatrin and the combined oral contraceptive pill was undertaken in 13 female volunteers.[18] They received - in random order with a wash-out interval of greater than 4 weeks - two oral 4-week treatments with vigabatrin (3000 mg/day) and placebo, respectively. The single-dose kinetics of a combined oral contraceptive pill containing ethinylestradiol 30μg and levonorgestrel 150μg were determined after 3 weeks’ treatment with either vigabatrin or placebo. No significant difference in pharmacokinetic parameters between vigabatrin and placebo was observed for ethinylestradiol (half life 12.5 ± 3.2 versus 13.9 ± 3.2 hours; AUC 874 ± 301 versus 939 ± 272 ng/L/h) and levonorgestrel (half-life 17.7 ± 5.2 versus 23.1 ± 9.8 hours; AUC 27.5 ± 9.6 versus 30.0 ± 12.0 μg/L/h). Two subjects did show a 50 and 39% reduction in ethinylestradiol AUC during vigabatrin therapy. However, in this study there were no effects on the clearance of phenazone or the excretion of 6-β-hydroxycortisol, a selective marker of hepatic enzyme activity. It was therefore concluded that there are no interactions between vigabatrin and the combined oral contraceptive pill.

2.2.2 Felbamate

In vitro studies and interaction studies between felbamate and other AEDs suggest that felbamate inhibits the CYP2C19 isoenzyme and may induce CYP3A4.[33] The interaction between felbamate and the combined oral contraceptive pill was assessed in 31 volunteers who were randomised to receive felbamate 1200mg twice daily or placebo for 29 days spanning two consecutive cycles. All volunteers had been taking ethinylestradiol 30μg and gestodene 75μg for at least 3 months prior to the study. There were significantly lower gestodene peak plasma concentrations (Cmax) of -16% (p = 0.001) and AUC24hof −42% (p = 0.02) in the group taking felbamate. The AUC24h of ethinylestradiol was also 13% lower (p = 0.02). None of the volunteers had an increase in luteinising hormone or progesterone concentrations, suggesting that ovulation did not occur.

Based on these results, women wishing to take the combined oral contraceptive and felbamate should use a preparation with a higher estrogen content.[19] However, use of felbamate is restricted due to the occurrence of serious adverse effects such as aplastic anaemia.

2.2.3 Lamotrigine

The interactions between lamotrigine and the combined oral contraceptive pill were studied in 12 volunteers taking either Microgynon 30® or Ovranette® (ethinylestradiol 30mg and levonorgestrel 150μg). This was an open, multiple-dose study involving three consecutive menstrual cycles. Cycle 1 included a predosing control phase. Lamotrigine 150 mg/day was administered for 14 days from day 22 of cycle 1 to day 7 of cycle 2; cycle 3 was a post-dosing control phase.

Plasma ethinylestradiol and levonorgestrel concentrations were measured on days 5 to 8, urinary 6-β-hydroxycortisol excretion (an index of hepatic microsomal enzyme induction) was measured on days 7 to 8, and plasma progesterone concentration was measured on days 15 to 16 of each cycle. After administration of lamotrigine, mean plasma ethinylestradiol concentration was 89% [95% confidence interval (CI) 78 to 102], levonorgestrel concentration was 101% (95% CI 90 to 112) and urinary 6-β-hydroxycortisol was 104% (95% CI 84 to 128) of control cycles. Late-cycle progesterone concentration (mean 0.9 nmol/L) remained below the postovulatory range. The mean steady state trough lamotrigine concentration at the end of the administration period was 2.56 mg/L.

Therefore, it can be concluded that lamotrigine does not affect the plasma concentrations of the combined contraceptive pill nor does it cause enzyme induction.[20]

2.2.4 Gabapentin

Gabapentin is not metabolised and is excreted unchanged in urine. It has no actions on hepatic microsomal enzymes, therefore an interaction with the combined oral contraceptive pill is unlikely.[34]

The effects of gabapentin on plasma concentrations of the combined oral contraceptive pill were studied in 13 volunteers in a nonblind, multiple-dose crossover study.[21] The women received ethinylestradiol 50μg/norethisterone (norethindrone) 2.5 mg/day for three consecutive cycles. Gabapentin 1200 mg/day was added in the last week of the study. Addition of gabapentin produced small, insignificant increases in ethinylestradiol Cmax (+ 9.4%) and AUC24h(+ 5.5%). Therefore, as predicted, gabapentin does not appear to reduce the efficacy of the combined oral contraceptive pill.

2.2.5 Topiramate

In initial in vitro studies, topiramate appeared to have no significant effects on hepatic microsomal enzymes.[35] However, a subsequent study using human liver microsomes demonstrated that topiramate significantly inhibits CYP2C19.[35]

The effects of topiramate on the combined oral contraceptive pill (ethinylestradiol 35μg/norethindrone 1mg) were studied in 12 women with epilepsy who were also being treated with valproic acid.[22] For cycle 1, the oral contraceptive and the patient’s usual dose of valproic acid were administered. In cycles 2, 3 and 4, topiramate was added in dosages of 100mg twice daily, 200mg twice daily and 400mg twice daily, respectively. Peak ethinylestradiol concentrations decreased by 18 and 25% (p < 0.05) with topiramate dosages of 200mg and 400mg twice daily, respectively. Bioavailability was lowered and clearance was increased by 30% with a dosage of topiramate 400mg twice daily (p < 0.05). The only significant change in norethindrone pharmacokinetic parameters was an increase in norethindrone clearance by 22%, with a corresponding decrease in half-life, at a dosage of topiramate 400mg twice daily.

Decreases in ethinylestradiol and norethindrone concentrations are modest with topiramate compared with more potent enzyme-inducing AEDs such as carbamazepine. However, these decreases are possibly sufficient to compromise the efficacy of the combined oral contraceptive pill and, in the absence of further evidence, a higher-dose combined oral contraceptive pill would be advisable.

2.2.6 Tiagabine

The influence of tiagabine on the efficacy of the combined oral contraceptive pill was studied over two 28-day pill-taking cycles in ten female volunteers who were taking nonsequential combined oral contraceptive pills containing ethinylestradiol (30μg) and either levonorgestrel or desogestrel 150μg for at least 6 months.[23] Tiagabine (2mg once daily) was administered from day 24 of cycle 1 to day 7 of cycle 2. Analyses of variance revealed no statistical differences between the plasma concentrations of progesterone, ethinylestradiol, levonorgestrel and desogestrel, and urinary excretion of 6-β-hydroxycortisol. Plasma progesterone concentrations remained in the non-ovulatory range in all volunteers throughout the study, indicating no effect on the degree of ovulation suppression maintained by the oral contraceptive pill.

Tiagabine, therefore, does not induce hepatic microsomal enzymes and does not interact with the combined oral contraceptive pill.

2.2.7 Oxcarbazepine

Initially it was thought unlikely that oxcarbazepine would have a significant interaction with the combined oral contraceptive pill, as oxcarbazepine has little or no inducing effect on CYP isozymes. However, two studies demonstrated that oxcarbazepine caused a marked reduction in ethinylestradiol and levonorgestrel concentration, leading to the conclusion that the actions of oxcarbazepine on CYP3A-mediated metabolism of these sex steroids are comparable with those of carbamazepine.[24,25]

In the initial study,[24] oxcarbazepine 900 mg/day was administered to 13 healthy women who had been receiving a triphasic oral contraceptive pill (ethinylestradiol 30μg/levonorgestrel 50μg days 1 to 6, ethinylestradiol 40μg/levonorgestrel 75μg days 7 to 11, and ethinylestradiol 30μg/levonorgestrel 125μg days 12 to 21). Mean peak levonorgestrel and ethinylestradiol concentrations did not change significantly with the addition of oxcarbazepine. However, the bioavailability of the sex steroids was significantly decreased by −32 and −48%, respectively (p < 0.006), a 12% mean.

In the second study,[25] 22 women were allocated to either oxcarbazepine (600 mg/day) or placebo in a randomised sequence, each for a period of 26 days in a double-blind, placebo-controlled, crossover design. Throughout the two study periods, the women received a combined oral contraceptive pill containing ethinylestradiol 50μg and levonorgestrel 250μg. The two treatment periods were separated by a washout of one to three cycles. Blood samples were taken on day 21 of each study cycle. Compared with placebo, oxcarbazepine treatment induced a marked reduction in the plasma concentrations of both sex steroids. Evaluation of pharmacokinetic parameters revealed that oxcarbazepine therapy produced a 47% reduction (p < 0.01) in mean AUC values of both ethinylestradiol and levonorgestrel. Peak plasma concentrations of ethinylestradiol concentrations were reduced by 30% during oxcarbazepine treatment while peak levonorgestrel concentrations were decreased to a lesser extent. Times to peak concentrations were not affected by oxcarbazepine, while elimination half-lives were significantly shortened for both sex steroids (p < 0.01). No surges in plasma progesterone concentrations were detected, but one woman experienced breakthrough bleeding, suggesting that the activity of the sex steroids was reduced.

Therefore, reduced efficacy of the combined oral contraceptive pill should be anticipated in women receiving oxcarbazepine.

2.2.8 Levetiracetam

A study used human liver microsomes to test the actions of levetiracetam on 11 drug-metabolising enzyme systems.[36] The enzyme activity was not affected by levetiracetam, suggesting indirectly that it was unlikely that there would be an interaction between levetiracetam and the combined oral contraceptive pill.

Coadministration of levetiracetam (500mg twice daily) with the oral contraceptive pill Microgynon 21® (ethinylestradiol 0.03mg and levonorgestrel 0.15mg) or placebo in 18 volunteers did not influence the pharmacokinetics of ethinylestradiol, levonorgestrel or levetiracetam. Consistently low concentrations of progesterone and luteinising hormone measured on days 13, 14, 15 and 21 of each of two consecutive menstrual cycles indicated that ovulation had not occurred.[26]

2.2.9 Zonisamide

Zonisamide is metabolised predominantly by the CYP3A subfamily of hepatic microsomal enzymes.[37] Further studies suggest that zonisamide is metabolised not only by CYP3A4 but also to a lesser extent by CYP2C19 and CYP3A5.[38] No evidence of CYP3A4 induction by zonisamide currently exists and, therefore, no studies have been undertaken to look at the pharmacokinetics of a concurrently administered combined oral contraceptive preparation.

3. Other Methods of Hormonal Contraception

3.1 Progesterone-Only Pill

Although no data are available on interactions between AEDs and the progesterone-only pill, this method of contraception is likely to be unreliable in women taking enzyme-inducing AEDs.[32]

3.2 Medroxyprogesterone Injection

There is no evidence as to whether hepatic microsomal enzyme-inducing AEDs reduce the efficacy of medroxyprogesterone injections (Depo-Provera®), but theoretically they might. It is therefore recommended that the frequency of injection for women taking such drugs be increased to every 10 weeks from the usual 12 weeks.[39]

3.3 Levonorgestrel Implants

There have been more than 30 contraceptive failures reported to the manufacturers of levonorgestrel implants (Norplant®) among women with epilepsy who had the implants.[40] A study has been undertaken to look at the effects of AEDs on Norplant®.[41]

Norplant® subdermal capsules were implanted into nine women with epilepsy and ten controls. Eight of the nine women were receiving enzyme-inducing AEDs (two taking phenytoin monotherapy, three taking phenytoin and carbamazepine, two taking carbamazepine monotherapy and one taking phenytoin and valproic acid). Venous blood samples for plasma levonorgestrel concentrations were taken at 0, 1, 3, 6, 9 and 12 months after insertion. At 3 to 6 months, the mean overall plasma levonorgestrel concentration was significantly lower in the six women with epilepsy who were taking phenytoin (203 ± 128 ng/L) than in the controls (325 ± 135 ng/L, p < 0.01). After 1 year, nine of ten control women continued to use Norplant® and no pregnancies had occurred. Two of the nine women with epilepsy had become pregnant during contraception with Norplant®. They were both taking phenytoin and their plasma concentrations of levonorgestrel were low at the time of conception. Levonorgestrel released from the capsules did not alter seizure frequency.

In conclusion, Norplant® should not be used as a method of contraception by women receiving enzyme-inducing AEDs.

3.4 Intrauterine Progestogen-Only Contraceptive

With the intrauterine progestogen-only contraceptive (Mirena®), levonorgestrel is released directly into the uterine cavity. Contraceptive effects are mainly local and therefore not affected by the presence or absence of enzyme-inducing AEDs. Therefore, all women with epilepsy can use this method of contraception.[32,39]

4. Knowledge About Contraception and AEDs

4.1 Doctors

Despite the knowledge for many years that the interactions between the combined oral contraceptive pill and AEDs can lead to contraceptive failure, many neurologists and obstetricians seem unaware of the need to alter the dose of the oral contraceptive pill when enzyme-inducing AEDs are prescribed. A survey in the US showed that fewer than half of the neurologists (41%) and obstetricians (43%) who replied knew that alterations in the dose of an oral contraceptive pill might be indicated when prescribing AEDs in addition to the oral contraceptive pill.[40] The knowledge of the doctors about the interactions of AEDs and the oral contraceptive pill was very poor. Only 4% of the neurologists and none of the obstetricians were correct in identifying the interactions between the combined oral contraceptive pill and phenytoin, carbamazepine, valproic acid, phenobarbital, primidone and ethosuximide. Not surprisingly, 27% of neurologists and 21% of obstetricians reported contraceptive failures among their patients taking AEDs.

4.2 Patients

A recent survey undertaken by the British Epilepsy Association (BEA) asked about information given to patients with epilepsy on important topics including contraception and pregnancy.[42] Fifty-two per cent of women aged between 16 and 55 years claimed not to have received any advice about contraception, epilepsy and AED therapy. In those using the oral contraceptive pill, 27% claimed not to have been given any advice, while 52% had been given information regarding reduced effectiveness of the contraceptive pill. Forty-three per cent of these women were taking an increased dose of the oral contraceptive pill. Only 50% of the women taking the oral contraceptive pill and an enzyme-inducing AED had been informed regarding interactions and the reduced effectiveness the contraceptive pill. Forty-eight per cent of women taking the oral contraceptive pill felt that they had not been given enough information about the contraceptive pill and their medication. Forty-three per cent of the women taking an increased dose of the oral contraceptive pill were concerned about taking a higher dose.[42]

5. Conclusions

A number of AEDs do not appear to interact with the combined oral contraceptive pill (valproic acid, vigabatrin, lamotrigine, gabapentin, tiagabine, levetiracetam, zonisamide, ethosuximide and the benzodiazepines). However, it is recommended that women taking enzyme-inducing AEDs (phenobarbital, primidone, phenytoin, carbamazepine, felbamate, oxcarbazepine and topiramate) increase their ethinylestradiol dose from between 20 and 35μg to 50μg. If breakthrough bleeding occurs, ethinylestradiol doses may need to be increased to 75 or 100μg. Women also need to be advised that even if they are taking a higher-dose combined contraceptive pill, efficacy cannot be guaranteed. However, the pregnancy rates still appear to be lower (approximately 7%) compared with barrier methods, which have a failure rate of between 15 and 20%.[39,43]

If additional protection is needed, a spermicidal gel or a barrier method could be used in addition to the combined oral contraceptive pill. Other contraceptive alternatives include intrauterine devices or medroxyprogesterone injections (Depo-Provera®). It is important that unplanned pregnancies are avoided in women with epilepsy, particularly as many of the AEDs are teratogenic.[39]

Women with epilepsy state that they need more information about various methods of contraception and interactions with AEDs. In particular, they want adequate counselling about the need for a combined oral contraceptive pill with a high dose of estrogen. Unfortunately, these higher doses are likely to expose some women to risks due to individual variations in sex hormone induction. It is also clear that many doctors need more education about contraception and epilepsy so that they are able to counsel their patients appropriately.