Epilcyisio. 40(6):783-787, I999 Lippincorr Wiliiaim CG Wilkins, Inc.

, Philadelphia 0 Inlernalional League Againqt Epilepsy

Clinical Research

Induction of Ethinylestradiol and Levonorgestrel Metabolism by Oxcarbazepine in Healthy Women

C. Fattore, G. Cipolla, G. Gatti, *G. L. Limido, t Y . Sturm, $C. Bernasconi, and E. Perucca
Clinical Pharmacology Unit, University o Pavia, Pavia, and *Novartis Medical Department (CNS), Origgio, Italy; f 7Novarti.s ICWCNS, Basel, and if1.P.A.S.SA, Stubio, Switzerland

Summary: Purpose: To evaluate the effect of oxcarbazepine (OCBZ) on the pharmacokinetic profile of steroid oral contraceptives. Methods: Twenty-two healthy women aged 18-44 years were recruited, and 16 of them completed the study. By using a randomized double-blind crossover design, each woman was studied in two different menstrual cycles, during which placebo or OCBZ (maintenance dosage, 1,200 mglday) was given in randomized sequence for 26 consecutive days with a washout of at least one cycle in between. A steroid oral contraceptive containing 50 p,g ethinylestradiol (EE) and 250 tJ-g levonorgestrel (LN) was taken for the first 21 days of each cycle. Plasma concentrations of EE and LN were measured by gas chromatography-mass spectrometry in samples collected at regular intervals on days 21-23 of each cycle. Results: Compared with placebo, areas under the plasma congeometric means) decreased by centration curves (AUC,,,,, 47% for both EE (from 1,677 to 886 pgWml; p < 0.01) and LN

(from 137 to 73 ngWml; p < 0.01), during OCBZ treatment. Peak plasma EE concentrations decreased from 180 pg/ml during the placebo cycle to 117 pg/ml during the OCBZ cycle (p < O.Ol), whereas peak plasma LN concentrations decreased from 10.2 to 7.7 ng/ml (p < 0.01). The half-lives of EE and LN also decreased from 13.6 to 7.9 h (p < 0.01) and from 28.8 to 15.8 h, respectively (p i0.01). Conclusions: OCBZ reduces plasma concentrations of the estrogen and progestagen components of steroid oral contraceptives, presumably by stimulating their CYP3A-mediated metabolism in the liver or gastrointestinal tract or both. Because this may lead to a decreased efficacy of the contraceptive pill, women treated with OCBZ should receive preferentially a high-dosage contraceptive and should be monitored for signs of reduced hormonal cover. Key Words: Oxcarbazepine-Oral contraceptives-Ethinylestradiol-LevonorgestreI-Drug interaction-Enzyme induction.

Despite its structural similarity, oxcarbazepine (OCBZ) differs from carbamazepine (CBZ) in pharmacokinetic behaviour, biotransformation profile, and interaction potential (1-4). Unlike CBZ, which is oxidized to an active stable epoxide intermediate, OCBZ undergoes rapid reduction to the active monohydroxy derivative (MHD), which is considered to be primarily responsible for the pharmacologic effect (2). Only a small fraction of MHD is eliminated unchanged or is oxidized to the 10,ll -dihydroxy derivative, the major part being conjugated with glucuronic acid before excretion in urine (1-3). Because of differences in metabolic fate, OCBZ kinetics is less variable than that of CBZ (2), it is less
Accepted November 16, 1998. Addres:, correspondence and reprint requests to Prof. E. Perucca at Clinical Pharmacology Unit, Department of Internal Medicine and Therapeutics,University of Pavia, Piazza Botta 10, 27100 Pavia, Italy.

influenced by enzyme-inducing agents (S), and unaffected or negligibly affected by known inhibitors of CBZ metabolism such as erythromycin (6), dextropropoxyphene (7), verapamil (8), and viloxazine (9). Unlike CBZ, OCBZ does not undergo metabolic autoinduction (lo), and it has little or no inducing effect on the cytochrome P-450 (CYP)-mediated metabolism of a number of concurrently administered drugs, including valproic acid (VPA; 1l), certain neuroleptics (12), citalopram (13), and probably warfarin (14). Among CYP isoenzymes whose activity is increased markedly by CBZ, those belonging to the CYP3A family stand out prominently (lS,l6). Whether OCBZ also affects the activity of these isoenzymes is unclear. Unlike CBZ, OCBZ, 600 mg/day, does not affect the CYP3Amediated conversion of cortisol to 6-P-hydroxycortisol (lo), or the elimination clearance of antipyrine (lo), a compound metabolized partly by CYP3A (17). Although at higher dosages some enhancement in antipyrine clear783

784

C. FATTORE ET AL.
followed by 18-day maintenance and 4-day tapering off). OCBZ dosage was 600 mglday on days 1-2,900 mglday on days 3 4 , 1,200 mglday on days 5-22,600 mglday on days 23-24, and 300 mglday on days 25-26, each daily dose being divided into two separate administrations except for the last two doses, which were taken at bedtime. Throughout the two study periods, subjects received a steroid oral contraceptive (Evanor-D; Wyeth, Aprilia, Italy) containing 50 p g EE and 250 p g LN, taken once daily for the first 21 days of each 28-day menstrual cycle. The first dose of OCBZ or placebo was given on day 1 of each cycle. The two treatment periods were separated by a washout of one to three cycles. Blood samples (14 ml) for the determination of plasma EE and LN concentrations were collected just before the last dose of the oral contraceptive (day 2 1, time 0) and 1, 2, 3, 4, 6, 8, 10, 12, 24, 32, and 48 h thereafter. Serum progesterone levels, as indicator of possible ovulation, also were determined in the sample at time 0. As a measure of compliance, plasma MHD concentrations were determined in samples (10 ml) collected before the morning dose on days 1, 8, 15, and 21.

ance may be observed (1 8), OCBZ 900 mglday has been shown to have only a minor influence on the first-pass metabolism of felodipine (19), another CYP3A substrate whose biotransformation is dramatically enhanced by CBZ (20). Although these data suggest that any inducing effect on CYP3A activity is negligible or modest, in a separate study, OCBZ, 900 mglday, caused a marked reduction in the plasma concentration of ethinylestradiol (EE) and levonorgestrel (LN), leading to the conclusion that its effects on the CYP3A-mediated metabolism of these steroids is comparable to that of CBZ (21). However, only a relatively small number of subjects completed the latter study, and the kinetics of contraceptive steroids with and without OCBZ were evaluated in a nonrandomized sequence, thus raising concern about potential bias in the interpretation of the results. In view of the controversial data on the effect of OCBZ on CYP3A-mediated activity and the potential clinical importance of an interaction with oral contraceptives, a randomized double-blind placebo-controlled study was carried out in which the effect of OCBZ on the kinetics of EE and LN was reevaluated in healthy women.

METHODS Subjects Twenty-two women aged 18-44 years (mean age, 27 years) and weighing 42-76 kg (mean weight, 56 kg) were included in the study. Protocol was approved by the Ethics Committees of the two participating centers. All women were healthy according to their medical history, physical examination, and routine laboratory tests. Main exclusion criteria were history of thromboembolic disease, migraine, phlebitis, or hypertension or any other condition contraindicating the use of the study drugs, a history of drug or alcohol abuse, intake of any drug other than contraceptive steroids during the previous 3 weeks, pregnancy or lactation, and smoking status (>I0 cigaretteslday). All subjects were required to use a reliable nonhormonal contraceptive method throughout the study. According to the protocol, 16 subjects were needed to have an 80% chance of detecting, with a probability level of 0.05, a 40% difference between treatments in the area under the plasma concentration curve (AUC) of EE (assuming a 40% within-subject coefficient of variation). Study design The study was carried out according to a randomized, double-blind, placebo-controlled, crossover design. Subjects satisfying the entry criteria were randomly allocated to receive OCBZ (Trileptal; Novartis, Basel, Switzerland; 300-mg tablets) or placebo in randomized sequence, each for a period of 26 days (4-day titration

Analytic assays Plasma EE concentrations were determined by using a validated specific gas chromatography (GC)-mass spectrometry (MS) method after extraction into dichloromethane and derivatization. Determinations were carried out in the chemical ionization (negative ions) mode. The method had a precision (expressed as coefficient of variation) of 7-14%, an accuracy (expressed as bias) of 7%, and a lower limit of quantitation of 10 pg/ml. Plasma concentrations of LN were determined by a validated specific GC-MS method by using deuterated norethindrone as internal standard. After liquidliquid extraction, samples were analyzed by positive ion chemical ionization and selected ion monitoring. The method had a precision better than lo%, an accuracy better than 5%, and a lower limit of quantitation of 0.25 nglml. Plasma MHD concentrations were determined by a validated high-performance liquid chromatography (HPLC) assay. The method had a precision better than 13%, an accuracy better than 8%, and a limit of quantitation of 0.5 p M . Plasma progesterone levels were determined by a commercial enzyme immunoassay (SRI System; Serono Diagnostics, Coinsins, Switzerland). Pharmacokinetic and statistical analysis Peak plasma concentrations (Cmax)and time to peak (Tmax) of EE and LN were derived directly from the measured values. Half-lives (t,,J were calculated by linear regression, and areas under the curve (AUC), by the trapezoidal rule. Values given in the text are geometric means and 95% confidence intervals calculated on logtransformed data. Comparisons of kinetic parameters be-

OXCARBAZEPINE AND ORAL CONTRACEPTIVES

ETHINYLESTRADIOL

785
LEVONORGESTREL

IS0 160

17

''1
0 OCBZ session 0 PL session
8-

I Ib
0 OCBZ session 0 PL session

FIG. I . Plasma concentrations (geometric means) of ethinylestradiol and levonorgestrel in 16 healthy women taking a steroid oral contraceptive containing 50 pg of ethinylestradiol and 250 pg of levonorgestrel during comedication with oxcarbazepine (OCBZ, 1,200 mg/day) or placebo (PL). For calculations of means, concentrations below the lower limit of quantitation (LLQ) were taken as equal to the LLQ (10 pg/ml and 0.25 ng/rnl for ethinylestradiol and levonorgestrel, respectively). Means were not calculated if more than eight subjects had levels below the LLQ.

-140

2
B

120

2 ;;a
5 8 .c

.g 100
80

6 -

6 60
40

E V

E!
4-

2-

20
0 0-

10

20

30

40

50

10

Time (h)

20 30 Time (h)

40

50

tween placebo and OCBZ were made by paired Student's t test on log-transformed data.

RESULTS
Pharmacokinetic data Plasma concentrations (geometric means) of EE and LN during treatment with OCBZ and placebo in the 16 women who completed the study are shown in Fig. 1. Compared with placebo, OCBZ treatment induced a marked reduction in the plasma concentrations of both steroids. Evaluation of kinetic parameters revealed that OCBZ treatment produced a 47% reduction (p < 0.01) in mean AUC values of both EE and LN (Table 1). Peak plasma concentrations of EE were reduced by -30% during OCBZ treatment, whereas peak concentrations of LN decreased to a somewhat lesser extent. Times to peak concentrations were not affected by OCBZ, whereas elimination half-lives were shortened significantly for both steroids (Table 1). Plasma progesterone levels were below the detection limit (0.20 p h f ) in three subjects during treatment with OCBZ and in four subjects with placebo. Among subjects with detectable levels, plasma progesterone ranged from 0.26 to 1.1 1 p M (median, 0.54 pM) during treatment with OCBZ and from 0.30 to 1.06 pA4 (median, 0.67 phf) during treatment with placebo. The difference was not statistically significant. Plasma concentrations of MHD were reliably detected in all samples collected during OCBZ treatment. Trough MHD concentrations (geometric means and 95% confidence intervals) were 60 (52-69) pA4 on day 8 , 5 6 (48-66) p 4 on day 15, and 67 A (59-74) pA4 on day 21. Tolerability data Of the 22 subjects enrolled in the study, five (three with OCBZ and two with placebo) discontinued prematurely the randomized treatment because of adverse

events, whereas one withdrew her consent. Adverse events leading to treatment discontinuation during OCBZ in three subjects included bronchitis (after 14 days), fever, urticaria, and hepatosplenomegaly (after 11 days, due to serologically confirmed infectious mononucleosis), and muscle weakness, vertigo, ataxia, and headache (after 2 days), respectively. Of the two subjects discontinued because of adverse events with placebo, one complained of confusion, somnolence, and visual disturbances (after 2 days), and the other developed eyelid edema and headache on the first day. Adverse events in the population who completed the study were minor and occurred in 16 subjects with OCBZ and in eight subjects with placebo. Events recorded during OCBZ included nausea (n = 9), somnolence (n = 9), headache (n = 7), dizziness (n = 6), asthenia (n = 4), retching (n = 3), vomiting (n = 3), lymphadenopathy (n = 2), muscle weakness (n = I ) ,
TABLE 1. Pharmacokinetic parameters derived from plasma ethinylestradiol and levonorgestrel concentrations in 16 healthy women during comedication with oxcarbazepine or placebo
OCBZ (1,200 mglday)
Ethinylestradiol C",,, (pg/ml) T,",, AUCcO-Z4h) (pg.h/ml) t l l 2 (h) Levonorgestrel c,,,, (ng/ml) T",,, (Wh AUC(,-24,, (ng,h/ml) tIl2 (h)
1 I7 (97-142)"

PL I80 ( 1 55-208) 2 (1,3) 1,677 (1,355-2,086) 13.6 (10.4-17.9)

2(1,6) 886 (741-1,059)o 7.9 (5.6-1 1.1)" 7.7 (6.5-9.1 )" 1 (1,6) 73 (61-88)" 15.8 (12.6-20.0)"

10.2 (8.7-1 1.9) 2~ , 4 ) 137 ( I 16-162) 28.8 (23.1-36.0)

Geometric means and 95% confidence intervals calculated on logtransformed data. OCBZ, oxcarbazepine; PL, placebo. " p < 0.01 (vs. placebo). Median (range).

Epilepyia, Vol. 40, No. 6, I999

786

C. FATTORE ET AL.

pain in breast (n = 2), and breakthrough bleeding (n = 1). Events recorded during placebo included nausea (n = 5 ) , somnolence (n = 3), headache (n = l), dizziness (n = 2), vomiting (n = 2), pain in breast (n = 3), eyelid edema (n = l), and fatigue (n = 3).

DISCUSSION
This study demonstrates that OCBZ influences the pharmacokinetic profile of both EE and LN, reducing the AUC of both compounds by -50%. For EE, the effect is most likely to be mediated by induction of the CYP3A isoenzymes responsible for 2-hydroxylation, the main route of elimination of the steroid (22,23). Because a significant proportion of EE oxidation occurs on firstpass in the gastrointestinal tract and in the liver, induction of metabolism provides an explanation not only for the shortened half-life of the steroid, but also for its reduced peak plasma concentration after absorption (23). LN disposition is more complex than that of EE, because it involves reduction of the unsaturated ketone ring A as well as 2- and 6-hydroxylation (24,25). The precise isoenzymes involved in LN metabolism are not fully clarified, but it is likely that enzyme induction, possibly involving CYP3A-mediated oxidation pathways, also accounts for the reduced plasma levels and shorter half-life of LN during OCBZ treatment (21). LN does not undergo very extensive first-pass metabolism (25), which explains why its peak plasma concentration was less affected by OCBZ. Overall, our results are in agreement with those reported in a previous smaller non-randomized study, in which OCBZ reduced the AUC of EE and LN by 48 and 32%, respectively (21). Compared with that study, our investigation also differs in that a larger maintenance dosage of OCBZ was given ( I ,200 vs. 900 mg/day) and a high-dosage, fixed-ratio combination of EE (50 pg) and LN (250 pg) was used instead of a low-dose triphasic preparation (EE, 30-40 kg; LN, 50-125 pg). In our study, a high-dosage preparation was chosen based on the expectation that the plasma concentration of both steroids would be reduced by OCBZ, thus making results more representative for the type of formulation usually selected in patients with epilepsy treated with enzymeinducing drugs (23). The marked reduction in plasma EE and LN is in contrast to the reported inability of OCBZ to affect the CYP3A-mediated production of 6-j3-hydroxycortisol (lo), its absent or minor effect on antipyrine clearance (10,18), and its modest influence on the kinetics of another CYP3A substrate, felodipine ( 1 9). This apparent discrepancy might be explained by differences in dosages (a 600-mg OCBZ dose was tested in the 6-phydroxycortisol study) or by the heterogeneous nature of CYP3A isoenzymes. In fact, CYP3A includes different

enzyme subfamilies (15,26), and it is plausible that OCBZ, unlike CBZ, selectively stimulates a family of isoenzymes that metabolizes contraceptive steroids preferentially to cortisol or felodipine. Alternatively, OCBZ may have stimulated other pathways of EE and LN rnetabolism that do not involve CYP3A enzymes. Irrespective of the mechanism involved, this interaction is likely to have important clinical implications. Because the reduction in EE and LN levels is comparable to that reported before for CBZ, phenytoin (PHT), and barbiturates (27,28), OCBZ is expected to reduce the efficacy of the contraceptive pill in a similar way. Although a surge in plasma progesterone levels (an index of ovulation) was not detected, one of the women experienced breakthrough bleeding during OCBZ treatment, suggesting that the activity of the steroids was reduced. In two previous studies (15,29), breakthrough bleeding during intake of OCBZ and contraceptive steroids was observed in two of I3 and in four of six women, respectively: the higher incidence compared with our study may be due to use of a low-dose contraceptive, as also suggested by the observation that in three of the four women with breakthrough bleeding surveyed by Sonnen (29), the bleeding resolved after increasing the dose of EE. In conclusion, based on these results, reduced efficacy of the contraceptive pill should be anticipated in women receiving OCBZ. These patients should be given preferentially a high-dosage steroid preparation and should be monitored carefully for possible signs of insufficient contraceptive cover such as breakthrough bleeding. Additional or alternative contraceptive measures also may be considered in these patients.

REFERENCES
1. Faigle JW, Menge GP. Pharmacokinetic and metabolic features of

2. 3.

4.

5.

6.
7.

8.

9.

oxcarbazepine and their clinical significance: comparison with carbamazepine. In? Cfin Pvychopharmacot 1990;5:73-82. Grant SM, Faulds D. Oxcarbazepine: a review of its pharmacology and therapeutic potential in epilepsy, trigeminal neuralgia and affective disorders. Drugs 1992;43:873-88. Lloyd P, Flesch G, Dieterle W. Clinical pharmacology and pharmacokinetics of oxcarbazepine. Epilepsia 1994;35(suppl 3): 10-3. Baruzzi A, Albani F, Riva R. Oxcarbazepine: pharmacokinetic interactions and their clinical significance. Epilepsia 1994;35 (suppl 3):14-9. Tartara A, Galimberti CA, Manni R, et al. The pharmacokinetics of oxcarbazepine and its active metabolite 10-hydroxy-carbazepine in healthy subjects and in epileptic patients treated with phenobarbitone or valproic acid. B r J Clin Pharmacol 1993;36:366-8. Keranen T , Jolkkonen J, Klosterkov Jensen P, Menge GP, Andersson P. Absence of interaction between oxcarbazepine and erythromycin. Acta Neurol Scund 1992;86: 120-3. Mogensen PH, Jorgensen L, Boas J, et al. Effects of dextropropoxyphene on the steady-state kinetics of oxcarbazepine and its metabolites. Actu Neurol Scand 1992;85:14-7. Kramer G, Tettenbom B, Flesch G. Oxcarbazepine-verapamil drug interaction in healthy volunteers. Epilepsia 1991 ;32(suppl 1):70-1. Pisani F, Fazio A, Oteri G, Artesi C, Xiao B, Perucca E, Di Perri R. Effects of the antidepressant drug viloxazine on oxcarbazepine and its hydroxylated metabolites in patients with epilepsy. Actu Neurol Scnnd 1994;90: 130-2.

OXCARBAZEPINE AND ORAL CONTRACEPTIVES

10. Larkin JG, McKee PJ, Forrest G, et al. Lack of enzyme induction with oxcarbazepine (600 mg daily) in healthy subjects. Br J Clin Pharmacol 1991;31:65-71. 1 I . Houtkooper MA, Lammertsma A, Meijer JWA, et al. Oxcarbazepine (CGP47680): a possible alternative to carbamazepine? Epilepsia 1987;28:693-8. 12. Raitasuo V, Lehtovaara R, Huttunen MO. Effect of switching carbamazepine to oxcarbazepine on the plasma levels of neuroleptics. Psychophurmacology 1994;116:115-6. 13. Leinonen E, Lepola U, Koponen H. Substituting carbamazepine with oxcarbazepine increases citalopram levels: a report on two cases. Pharmacopsychiatry 1996;29: 156-8. 14. Kramer G, Tettenborn B, Klosterkov Jensen P, Menge GP, Stoll KD. Oxcarbazepine does not affect the anticoagulant activity of warfarin. Epilepsia 1992;33: 1145-8. 15. Perucca E, Richens A. Biotransformation. In: Levy RH, Mattson RH, Meldrum B, eds. Antiepileptic drugs. 4th ed. New York: Raven Press, 19953-50. 16. Spina E, Pisani F, Perucca E. Clinically significant pharmacokinetic drug interactions with carbamazepine: an update. Clin Pharmacokinet 199631: 198-214. 17. Brockmoller . , IRoots I. Assessment of liver metabolic function: clinical implications. Clin Pharmacokinet 1994;27:21648. 18. Patsalos PN, Zakrzewska JM, Elyas AA. Dose-dependent enzyme induction by oxcarbazepine? Eur J Clin Pharmacol 1990;39: 187-8. 19. Zaccara G, Gangemi PF, Bendoni L, et al. Influence of single and

787

20.
21,

22. 23. 24. 25. 26. 27. 28. 29.

repeated doses of oxcarbazepine on the pharmacokinetic profile of felodipine. Ther Drug Monit 1993;15:3942. Capewell S, Freestone S, Critchley JAJH, Pottage A, Prescott LF. Reduced felodipine bioavailability in patients taking anticonvulsants. Luncet 1988;2:480-2. Klosterkov Jensen P, Saano V, Haring P, Svenstrup B, Menge GP. Possible interaction between oxcarbazepine and an oral contraceptive. Epilepsia 1992;33:1149-52. Guengerich FP. Oxidation of l7P-ethinyloestradiol by human liver cytochrome P450. Mol Pharmacol 1988;33:500-8. Back DJ, Orme MLE. Pharmacokinetic drug interactions with oral contraceptives. C i Pharmucokinet 1990;18:472-84. ln Fotherby K. Pharmacokinetics of gestagens: some problems. Am J Obstet Gynecol 1990;163:323-8. Fotherby K. Levonorgestrel: clinical pharmacokinetics. Clin Pharmacokinet 1995;28:203-15. Bertz RJ, Grauneman GR. Use of in vitro and in vivo data to estimate the likelihood of metabolic pharmacokinetic interactions. Clin Pharmacokinet 1997;32:210-58. Crawford P, Chadwick DJ, Martin C, et al. The interaction of phenytoin and carbamazepine with combined oral contraceptive ln steroids. Br J C i Pharmacol 1990;30:892-6. Back DJ, Bates M, Bowden A, et al. The interaction of phenobarbital and other anticonvulsants with oral contraceptive steroid therapy. Contraception 1980;22:495-503. Sonnen AEH. Oxcarbazepine and oral contraceptives. Actu Neurol Scand 1990;82(suppl 133):37.

Epilepsia, Vol. 40, No. 6, 1999