Address correspondence and reprint requests to Dr. R.H. Levy at University of Washington, H-272N Health Sciences, Box 357610, Seattle, WA 98195, U.S.A. E-mail: firstname.lastname@example.org
Summary: Purpose: This study was designed to evaluate whether levetiracetam, a novel antiepileptic drug (AED), influences the pharmacokinetics of steroid oral contraceptives.
Methods: During a run-in phase, 18 healthy female patients received an oral contraceptive containing ethinyl estradiol, 0.03 mg, and levonorgestrel, 0.15 mg, for the first 21 days of two consecutive menstrual cycles. In a subsequent double-blind, randomized, two-way crossover treatment phase, subjects received either levetiracetam, 500 mg, or placebo twice daily concomitant with the oral contraceptive. Plasma concentrations of ethinyl estradiol and levonorgestrel were measured on days 14 and 15 of the two treatment periods for the evaluation of the 24-h kinetic parameters, and an additional sample was collected on day 21 to determine the trough plasma concentrations. Serum progesterone and luteinizing hormone (LH) levels were determined on days 13, 14, 15, and 21 of each cycle of the treatment phase.
Results: The plasma concentration–time curves and pharmacokinetic parameters of ethinyl estradiol and levonorgestrel were not statistically different during concomitant treatment with either levetiracetam or placebo. The ratios of the log-transformed geometric mean areas under the plasma concentration–time curves (AUCs), maximal (Cmax) and minimal (Cmin) plasma concentrations, and trough concentrations on day 21 (C21) ranged from 99.12 to 99.96% for ethinyl estradiol and from 97.13 to 99.41% for levonorgestrel. The 90% confidence intervals of these ratios were well within the 80 to 125% acceptance range for lack of interaction. Serum progesterone and LH concentrations were fairly constant during the run-in and treatment phases and remained markedly below their respective physiologic levels. Safety and menstrual-bleeding patterns were comparable during levetiracetam and placebo administration.
Conclusions: Levetiracetam does not affect the pharmacokinetics of an oral contraceptive containing ethinyl estradiol and levonorgestrel, and on the basis of serum progesterone and LH levels, it does not affect the contraceptive efficacy.
Drug–drug interactions are a major concern during therapy with antiepileptic drugs (AEDs). Some AEDs, such as phenytoin (PHT), phenobarbital (PB), and carbamazepine (CBZ), are potent inducers of specific cytochrome P (CYP) 450 isoenzymes, notably CYP 3A4 (1). As a result, these AEDs increase the clearance and thereby reduce the efficacy of concomitantly administered drugs that are metabolized by these CYP isoenzymes. CYP 3A4 is involved in the metabolism of oral contraceptives (OCs) (2).
Contraceptive failure and unplanned pregnancy have been well documented in women with epilepsy treated concomitantly with CYP-inducing AEDs and OCs (3–7). Newer AEDs, including topiramate (TPM), felbamate (FBM), and oxcarbazepine (OCBZ), are less potent inducers of CYP, but they also may interfere with the metabolism of OCs, thereby altering plasma concentrations of the contraceptive steroids and potentially decreasing their efficacy (8–12). A third group of AEDs, which includes gabapentin (GBP) and vigabatrin (VGB), is not metabolized in the liver and appears to be devoid of significant CYP enzyme-inducing or -inhibiting properties (13,14), in particular with contraceptive steroids (15,16).
Levetiracetam (LEV; Keppra; UCB, S.A.) is a novel AED that is indicated as adjunctive therapy for partial-onset seizures in adults. LEV has a useful pharmacokinetic profile, with almost complete oral absorption, linear and time-independent pharmacokinetics, minimal metabolism, an elimination half-life of 6–8 h (supporting a twice-daily dose regimen), a low volume of distribution (0.5–0.7 L/kg), and very limited protein binding (<10%). Its metabolism is not liver P450 dependent (17). In human liver microsomes, LEV and its inactive carboxylic metabolite did not inhibit 11 phase I and phase II drug-metabolizing enzymes, including CYP 3A4, and it did not induce CYP activity (18). Moreover, in controlled clinical trials, adjunctive therapy with LEV twice daily did not significantly affect the plasma concentrations of concomitantly administered AEDs (19–21). These in vitro and in vivo results suggest that LEV has a low potential for causing significant drug interactions.
The present study was designed to determine whether LEV influences the pharmacokinetics of a low-dose OC-containing ethinyl estradiol and levonorgestrel (Microgynon 21). In addition, the study evaluated the risk of ovulation as well as the safety and menstrual bleeding patterns during concomitant administration of LEV and the OC.
This study was performed in accordance with the Declaration of Helsinki and its amendments and in compliance with the guidelines of Good Clinical Practice. The study protocol and informed-consent form were approved by an independent ethics committee before recruitment of any subjects.
Healthy female nonsmokers aged 18 to 40 years were eligible if they had negative blood tests for alcohol and drug abuse, human immunodeficiency virus (HIV) and hepatitis B, and pregnancy. All subjects provided written informed consent before participating in the study.
Subjects entered a run-in phase, during which they received an OC containing ethinyl estradiol, 0.03 mg, and levonorgestrel, 0.15 mg (Microgynon 21), on days 1 to 21 of two consecutive menstrual cycles. The subsequent treatment phase had a randomized, double-blind, placebo-controlled, two-way crossover design. Subjects received the OC concomitant with LEV in one menstrual cycle and with placebo in the other menstrual cycle, with the sequence of treatments determined according to a randomization schedule. LEV, 500 mg, or placebo was administered every 12 h from day 1 to day 20 and once in the morning of day 21. The morning doses of study medication were taken on an empty stomach concomitant with the OC, whereas the evening doses were taken ≥2 h after dinner. Treatment was not administered during the final 7 days of each menstrual cycle. Plasma concentrations of ethinyl estradiol and levonorgestrel during a 24-h dosing interval were determined on days 14 to 15 of each cycle in the treatment phase. Trough concentrations of ethinyl estradiol and levonorgestrel were determined from a blood sample collected on day 21 before drug administration, and serum progesterone and luteinizing hormone (LH) levels were determined in samples collected before drug administration on days 13, 14, 15, and 21.
Subjects were admitted to the study site before the first dose of study medication in each cycle of the treatment phase and were discharged on day 3 after psychomotor tests showed no evidence of impairment. These procedures included the critical flicker fusion test, reaction-time test, steadiness test, and signal-detection test. Subjects were again admitted in the evening of day 13 and remained at the study site until the 24-h blood sample for the pharmacokinetic analysis had been collected on day 15. All other drug-administration and study activities were conducted on an ambulatory basis.
Plasma concentrations of ethinyl estradiol and levonorgestrel were determined from blood samples collected on day 14 of each cycle in the treatment phase. Two 7-ml blood samples were collected by venous puncture or indwelling catheter into plasma Li-heparinized Vacutainer tubes (Becton-Dickinson) before and 0.5, 1, 2, 4, 6, 8, 12, and 24 h after OC administration. A blood sample also was collected on day 21 before drug administration for measurement of trough levels. Blood samples were centrifuged within 20 min of collection at 1,600 g at 4°C for 10 min. The separated plasma was transferred to glass tubes and stored frozen until assay.
Serum concentrations of progesterone and LH were determined on days 13, 14, 15, and 21 of each cycle. On each day, a 4.5-ml blood sample was withdrawn before drug administration, collected into serum Monovettes, and allowed to stand at room temperature for 30 min. The sample was then centrifuged at 1,600 g at 4°C for 10 min, and the serum transferred into polystyrene tubes and stored frozen until assay.
Adverse events were monitored throughout the study, including the run-in phase, and routine clinical laboratory testing and 12-lead ECGs were conducted before and after the study. In addition, subjects recorded the onset, duration, and intensity of menstrual bleeding in a diary during each cycle.
The intake of OCs and LEV/placebo during the two randomized periods was supervised, and the administration times were recorded. Oral contraceptives during the run-in phase were administered on an ambulatory basis, and a pill count was performed for compliance check.
Plasma samples were assayed for ethinyl estradiol and levonorgestrel by using gas chromatography/mass spectrometry (GC/MS) methods. The methods were described in detail previously (22,23). In brief, the plasma samples for the estimation of ethinyl estradiol were extracted from acidified plasma into toluene and, after two further clean-up steps, back-extracted into dichloromethane. Extraction was followed by a derivatization step to obtain suitable derivatives for GC/MS analysis. Two microliters of the derivatized extract were injected into the GC/MS system. GC/MS measurements were performed in the chemical ionization mode (negative ions) with ammonia as reagent gas on a Trio 2000 GC/MS system. Negative ions chemical ionization was used for mass selective detection. Selected mass/charge ratios were m/z 490 for ethinylestradiol (derivative) and m/z 462 for the internal standard (derivative).
The plasma samples for the estimation of levonorgestrel were spiked with internal standard, alkalized, and extracted into cyclohexane/2-butanol. The sample extracts were evaporated to dryness, and the residue dissolved in 20 μL of ethyl acetate. Two microliters of the resulting solution was injected into the GC/MS system. The GC/MS analysis was performed on a Finnigan MAT 4500 G/MS system. The gas chromatograph was run in the temperature-programmed mode. Positive ion chemical ionization was used for mass selective detection. Selected mass/charge ratios were m/z 313 for levonorgestrel and m/z 305 for the internal standard.
All reagents and solvents, including water, were of analytic grade or better quality. Levonorgestrel was obtained from Sigma (Heidelberg, Germany). Deuterated norethindrone (internal standard) was obtained from AdW (Halle, Germany). Ethinyl estradiol and equilin (internal standard) were obtained from USP.
Calibration samples for the determination of ethinyl estradiol samples were prepared in the range of 10 pg/ml (lower limit of quantitation) to 1,000 pg/ml (upper limit of quantitation). Calibration samples for the determination of levonorgestrel samples were prepared in the range of 0.25 ng/ml (lower limit of quantitation) to 20 ng/ml (upper limit of quantitation).
The interassay accuracy (bias %) of the assays was between –3.44% and +2.89% (ethinylestradiol) and between –0.90 and 2.09% (levonorgestrel). The interassay precision (coefficient of variation) varied between 2.73 and 7.80% (ethinylestradiol) and 2.81 and 9.17% (levonorgestrel).
Progesterone and LH concentrations were determined by radioimmunoassay with Amerlex-M Progesterone and LH kits, respectively. The quantification ranges for progesterone and LH were 0.4 to 40 ng/ml and 5 to 150 mIU/ml, respectively.
The maximal (Cmax) and minimal (Cmin) concentrations of ethinyl estradiol and levonorgestrel and the times of their occurrence (tmax and tmin, respectively) were determined directly from the plasma concentrations measured on days 14/15. The area under the plasma concentration–time curve for the 24-h dosing interval (AUC0-24) was calculated by linear trapezoidal rule. The average concentration (Cav) was calculated from the AUC0-24/24, and the peak–trough fluctuation (PTF) from (Cmax– Cmin)/Cav. The terminal rate constant (λZ) and elimination half-life (t1/2) were calculated from a two-exponential structure model after extravascular bolus with lag time by using the Powell minimization algorithm and a weighting factor of 1/y(calc)2. The real sampling times were used in this calculation.
Statistical analyses were conducted by using SAS version 6.08. The statistical evaluation of the concentration-dependent pharmacokinetic parameters [AUC0-24, Cmax, Cmin, and C21 (trough value on day 21)] of ethinyl estradiol and levonorgestrel was conducted after logarithmic transformation by using an analysis of variance (ANOVA) model. The ratio of geometric means and the 90% conventional confidence interval between LEV and placebo treatment were calculated. Bioequivalence was confirmed if the 90% interval was included within the range of 80–125%(24). Safety data and serum progesterone and LH concentrations were analyzed by using descriptive statistics.
The study was initiated with 22 subjects who were included for the run-in phase. Four subjects were excluded during this phase, one because of lack of menstrual bleeding, one because of elevated diastolic blood pressure, one because of pregnancy, and one because she was a stand-by (the target number of 18 evaluable cases had been reached). All the remaining 18 subjects randomized into the treatment phase completed the study. These 18 subjects ranged in age from 19 to 39 years, with a mean (±SD) of 29.9 (±6.1) years, and they weighed 48.2 to 69.0 kg, with an average of 56.5 (± 5.3) kg.
Pharmacokinetics of ethinyl estradiol and levonorgestrel
Mean plasma concentrations over time of ethinyl estradiol and levonorgestrel during treatment with LEV or placebo are presented in Figs. 1 and 2. The pharmacokinetics of ethinyl estradiol and levonorgestrel were not affected by concomitant treatment with LEV (Table 1). The t1/2 of ethinyl estradiol averaged 10.81 (±2.85) and 10.83 (±2.33) h, and the t1/2 of levonorgestrel averaged 28.25 (±11.10) and 24.87 (±6.38) h during treatment with LEV and placebo, respectively. The PTF of ethinyl estradiol averaged 263.3% (±117.1) and 263.2% (±103.2) during administration with LEV and placebo, respectively, whereas the PTF of levonorgestrel averaged 161.9% (±42.9) and 163.6% (±28.2) during these respective periods.
Table 1. Effect of levetiracetam on the pharmacokinetics of ethinyl estradiol and levonorgestrel
A/B ratio: % (90% CI)
CI, confidence interval, AUC; area under the time–concentration curve.
AUC0–24 (pg × h/ml)
AUC0–24 (ng × h/ml)
The bioequivalence of the OC during treatment with LEV and placebo was evaluated by using the geometric means of the log-transformed AUC0-24, Cmax, Cmin, and C21 values. The ratio of each of these parameters ranged from 99.12 to 99.96% for ethinyl estradiol and from 97.13 to 99.41% for levonorgestrel (Table 1). The 90% confidence intervals of these ratios were included within the range of 80–125%, demonstrating the lack of interaction.
Serum progesterone and LH concentrations
Serum progesterone and LH concentrations were fairly constant during the run-in and treatment phases of the study and markedly below their respective physiologic levels. Mean serum progesterone levels ranged from 0.82 to 1.04 ng/ml during the first cycle and from 0.81 to 1.06 ng/ml during the second cycle of the run-in phase. During the treatment phase, mean serum progesterone ranged from 0.74 to 0.88 ng/ml with concomitant LEV administration and 0.67 to 0.94 ng/ml with placebo.
Progesterone concentrations >2 ng/ml, the threshold below which contraceptive efficacy is most likely maintained, were found in one subject on day 21 during LEV treatment (25,26). This subject had relatively high progesterone concentrations during each of the four cycles (ranging from 0.76 to 2.14 ng/ml).
Mean LH concentrations ranged from 3.66 to 4.54 mIU/ml during the first cycle and 3.65 to 6.31 mIU/ml during the second cycle of the run-in phase, and from 3.19 to 5.94 mIU/ml and from 3.19 to 5.20 mIU/ml during concomitant LEV and placebo treatment, respectively. LH concentrations >12 mIU/ml, the threshold below which contraceptive efficacy is most likely maintained, were measured in two subjects during the second cycle of the run-in period (25,26). In one of these subjects, these elevated levels also were measured during both cycles of the treatment phase.
No serious adverse events or sustained idiosyncratic events were reported. Overall, 53 adverse events were reported during the cycle in which LEV was administered with OCs, and 40 events during the cycle in which placebo and OCs were given (Table 2). Somnolence occurred in three subjects receiving LEV, with two of these subjects reporting somnolence on each day of the treatment period. Dizziness occurred in five subjects receiving LEV (nine events) and in one subject receiving placebo (three events). Dizziness occurred transiently in all subjects, except for one subject in each treatment group. Myalgia occurred in two subjects taking LEV, and diarrhea occurred in two subjects receiving placebo. All other adverse events were reported at a comparable frequency with LEV and placebo, or they occurred only in one subject each.
Table 2. Number (%) of subjects with adverse events during concomitant administration of ethinyl estradiol/levonorgestrel with either levetiracetam or placebo
Levetiracetam (n = 18)
Placebo (n = 18)
Adverse events occurring in at least two subjects in either cycle of the treatment phase are shown.
Menstrual cycle disturbances
The bleeding pattern was unaffected in 11 of the 18 subjects during the course of the study. The remaining seven subjects experienced irregularities in their bleeding pattern, including three subjects who had these irregularities only during the run-in phase after switching from another OC to ethinyl estradiol/levonorgestrel. Three other subjects had disturbances during both the run-in and treatment phases, including two subjects with premature onset of menstruation with placebo and one subject with intermenstrual spotting and premature onset of menstruation with both placebo and LEV. One subject reported a bleeding abnormality during only one cycle of the treatment phase: menstruation of milder intensity than usual, which occurred on day 27 of the treatment phase with LEV.
In addition, vital signs, ECGs, psychomotor tests, and clinical laboratory tests, performed during and at the end of the study, did not reveal any clinically relevant change in the health status of the subjects.
The results of this study demonstrate that LEV, at a dose of 500 mg twice daily (recommended starting dose for adjunctive treatment of partial-onset seizures), does not affect the pharmacokinetics and pharmacodynamics of ethinyl estradiol or levonorgestrel in healthy female volunteers. Mean plasma concentration–time curves and pharmacokinetic parameters of both steroids were not statistically different between concomitant administration of the OC with LEV or placebo. Ratios of the log-transformed values for AUC0-24, Cmax, Cmin, and C21 were close to 100%, and the 90% confidence intervals were contained within the 80–125% acceptance range for lack of interaction. Serum progesterone and LH levels were constant during each cycle of the study and remained markedly below physiologic levels of these hormones. Peaks in LH were not observed at the expected time of ovulation, and peaks in progesterone were not observed in the latter part of the cycle. Finally, breakthrough bleeding was not observed in any subject during LEV administration.
Drug interactions between AEDs and OCs leading to unwanted pregnancies are a major concern during epilepsy treatment, and patients receiving low-dose OC regimens are at greater risk because they already have relatively low systemic hormone concentrations.
AEDs can be categorized into three groups on the basis of their potential to cause induction drug interactions. Older AEDs, such as PHT, PB, and CBZ, potently induce CYP–drug metabolizing enzymes, and they also increase sex hormone–binding globulin levels (1,27,28). Both mechanisms may contribute to reduced steroid contraceptive levels during concomitant administration of these older AEDs and OCs, resulting in unplanned pregnancy (3–7). A second group of AEDs includes TPM, FBM, and OCBZ, which are less potent inducers of CYP but have been shown to alter plasma concentrations of the contraceptive steroids (12,29). If these or the older AEDs are used in women of childbearing age, higher-dose OCs are needed to compensate for the reduction in contraceptive steroid levels. However, the risk of thromboembolic complications with higher-dose contraceptives should be considered, especially in women older than 35 years and in those who smoke (27).
Members of a third group of AEDs [GBP, lamotrigine (LTG), tiagabine (TGB), valproic acid (VPA), and VGB] appear not to possess significant CYP-inducing properties. These AEDs do not significantly alter the pharmacokinetics of contraceptive estrogens and progestins (15,16,30–32). Based on the findings in our study, LEV also may be included in this third group of AEDs. Its in vivo behavior is consistent with the fact that in vitro LEV did not induce CYP activity in primary cultures of rat hepatocytes, which is a predictive model of CYP induction under physiologic conditions. At concentrations that were >5 times the therapeutic drug concentrations, it did not inhibit phase I and phase II drug-metabolizing enzymes, including CYP 3A4 (18).
The results of the present study demonstrate that LEV does not alter the pharmacokinetics of the contraceptive steroids in a low-dose monophasic OC. This finding supports the notion that LEV does not induce CYP 3A4, the isozyme responsible for contraceptive steroid hydroxylation. Although the study was performed with the recommended starting dose of LEV, other controlled clinical trials have indicated that LEV at doses ≤3,000 mg daily did not alter the plasma concentrations of any concomitantly administered AEDs (19–21).
When prescribing an AED for women of childbearing age, it is important to select an agent that will not compromise the efficacy of an OC. Members of this third group of AEDs are especially well suited, because they do not reduce plasma concentrations of contraceptive steroids and are thereby unlikely to affect contraceptive efficacy, even when low-dose OCs are used.
Acknowledgment: This study was sponsored by a grant from UCB S.A.