Colleen J. Wurden MD*
René H. Levy PhD*
* Department of Pharmaceutics, University of Washington School of Pharmacy and Medicine, Seattle, Washington
** Professor and Chair, Department of Pharmaceutics, Professor of Neurophysical Surgery, University of Washington School of Pharmacy and Medicine, Seattle, Washington
Carbamazepine (CBZ) is associated with clinically relevant drug interactions with a variety of other therapeutic agents. CBZ is almost completely cleared by metabolism, with less than 5% of a dose excreted unchanged in urine (1). The main pathway of metabolism is conversion to the active 10,11-epoxide (CBZE). Cytochrome P450 (CYP) 3A4 (CYP3A4) has been identified as the primary isoform that catalyzes the oxidation to the epoxide, whereas CYP2C8 is a minor contributor to this pathway in human liver (2). CYP3A4 also is constitutively expressed in the human intestine. Microsomes prepared from intestinal tissue can catalyze the epoxidation of CBZ, although the in vivo contribution of the intestine to the total clearance of CBZ probably is small (3). The epoxide subsequently is hydrated to a trans-dihydrodiol by microsomal epoxide hydrolase (4). In patients receiving CBZ chronically, 30% to 50% of a dose is excreted as the diol in urine (5). Because 90% of CBZ epoxide is converted to CBZ diol (6), it can be concluded that 30% to 50% of a dose of CBZ goes through the epoxidation pathway and a significant fraction of the primary metabolism is mediated by CYP3A4.
Several other characteristics of CBZ increase its susceptibility to interactions with other drugs. CBZ has a narrow therapeutic range, and, in an effort to maintain seizure control with a single anticonvulsant agent, plasma CBZ concentrations often are maximized to the upper limit of tolerance. Also, CBZ is a low-clearance drug, and thus sensitive to enzyme induction or enzyme inhibition, especially because a large number of drugs inhibit or induce CYP3A4.
Because CBZ is an inducer of CYP3A4 and other metabolic enzymes, another type of drug interaction resulting in subtherapeutic plasma concentrations of other agents can occur. Last, because CBZE can contribute to both therapeutic effect and toxicity, its interactions must be taken into consideration (7).
EFFECTS OF OTHER DRUGS ON CARBAMAZEPINE
Inhibition of Carbamazepine Metabolism
Inhibition of CBZ metabolism can occur when a coadministered drug competes with CBZ for binding to metabolic enzymes. Inhibitory drug interactions occur when the coadministered drug binds to CYP3A4 and reaches concentrations at the site of the enzyme that are high enough relative to the Ki (affinity of the inhibitor for binding to the enzyme) to inhibit a significant fraction of the CYP3A4 population. Inhibition of isoforms other than CYP3A4 has not been implicated in drug interactions resulting in decreased clearance of CBZ. Conversion to CBZE accounts for 30% to 50% of total CBZ clearance (5); thus, inhibition of this pathway can result in 1.5- to 2-fold increases in steady-state plasma levels of CBZ. Because CBZE is pharmacologically active, interactions leading to changes in the plasma concentration of this metabolite (e.g., due to inhibition of epoxide hydrolase) also can be clinically relevant, and they also are described in this section.
The binding affinity (Km) of CBZ to CYP3A4 determined in expressed CYP3A4 is 442 µmol/L (10-fold higher than therapeutic concentrations). Therefore, CBZ concentrations in vivo are far below those required to cause significant inhibition of CYP3A4 (2). Accordingly, there have been no reported interactions with CBZ resulting in decreased clearance of another CYP3A4-metabolized drug.
Although lamotrigine has been reported occasionally to increase the plasma concentration of CBZE, possibly by inhibiting epoxide hydrolase, most studies found no changes in plasma CBZ or CBZE concentration after addition of
lamotrigine (8), and one study in healthy volunteers showed no effect of lamotrigine on the kinetic parameters of orally administered CBZE (9). Based on these findings, lamotrigine's ability to potentiate the adverse effects of CBZ probably can be attributed to a pharmacodynamic interaction (10).
Remacemide (300 mg twice daily) increased plasma CBZ concentrations by 20% to 30% without consistent changes in the concentrations of CBZE, suggesting concomitant inhibition of the epoxidation pathway (11). In addition, the areas under the under the plasma concentration versus time curves (AUCs) of remacemide and its active metabolite were 60% and 30%, respectively, in patients who received CBZ compared with values observed in healthy volunteers treated only with remacemide, suggesting that CBZ induces the clearance of remacemide (11).
Stiripentol is a broad-spectrum inhibitor of oxidative metabolic enzymes, including CYP3A4 (12). The formation clearance of CBZ to CBZE was inhibited by stiripentol in seven epileptic patients by 50% (0.212 to 0.091 mL/min/kg) (13). In agreement, in human liver microsomal incubations with therapeutic concentrations of CBZ (14 µg/mL) and stiripentol (7 µg/mL), stiripentol inhibited CBZE formation velocity by 40% to 50%. Additional studies in our laboratory found that stiripentol also inhibited the metabolism of CBZ to two minor metabolites, 2-OH-CBZ and 3-OH-CBZ, that account for approximately 20% of CBZ clearance in vivo and appear to be formed by multiple CYP isoforms in microsomal incubations (unpublished results).
Administration of the anticonvulsant valproic acid is widely reported to increase the ratio of CBZE to CBZ. Also, valproic acid coadministration increased the CBZE half-life and decreased CBZE clearance in six patients (14). Although in theory, increased CBZE plasma concentrations could contribute to symptoms of CBZ toxicity, this has not been clearly demonstrated, and the interaction may be clinically insignificant (15,16).
Valpromide, the amide derivative of valproic acid, is used in many countries as an antiepileptic and as a mood-stabilizing agent. Compared with valproic acid, valpromide is a much more potent inhibitor of epoxide hydrolase, and by this mechanism it can lead to a very marked increase in plasma CBZE concentration (16). The interaction is clinically relevant because it often results in signs of CBZ intoxication.
Vigabatrin is a newer anticonvulsant that is not metabolized by liver enzymes; however, addition of vigabatrin to CBZ monotherapy causes elevated CBZ plasma levels. In most studies, vigabatrin has not been found to affect plasma CBZ levels (17). In a recent investigation in 66 epileptic patients, however, mean CBZ concentrations were increased by an average of 14% (9.41 versus 11.31 µg/mL, p < .001) after the addition of vigabatrin (average dose, 31.1 mg/kg/day; average CBZ dose, 16.7 mg/kg/day) (18).
The addition of fluoxetine to maintenance regimens of CBZ in patients and healthy volunteers resulted in increased plasma CBZ concentrations in some but not all studies (19, 20, 21,22). There are two case reports describing patients who received CBZ chronically. The addition of fluoxetine to their drug regimens resulted in CBZ plasma level increases of 33% to 63% (22). Fluoxetine was reported significantly to decrease both oral and intrinsic clearance of CBZ in a clinical study with six patients (21). Conversely, at least one clinical trial with CBZ and fluoxetine in eight subjects found that steady-state plasma concentrations of CBZ and CBZE were not affected by coadministration of fluoxetine (23). In human liver microsomes, we found that fluoxetine could mildly inhibit conversion of CBZ to CBZE only at concentrations 25 times therapeutic (unpublished results). Fluoxetine also was a weak inhibitor of the CYP3A4 probe activity, 10-hydroxylation of (R)-warfarin. Gidal et al. obtained results similar to ours in human liver microsomes (20). In vitro kinetic studies determined the Ki for fluoxetine inhibition of the CYP3A4-mediated 1′-hydroxylation of midazolam to be 65.7 µmol/L (100-fold higher than therapeutic), which also suggests that fluoxetine is not a reversible CYP3A4 inhibitor at clinical concentrations (24).
Other research indicates that fluoxetine coadministration has the potential to decrease the clearance of alprazolam, a compound that is metabolized by CYP3A4 (25, 26, 27). The impairment was attributed mainly to inhibition of alprazolam clearance by norfluoxetine, a primary circulating metabolite of fluoxetine, because the decrease in alprazolam clearance persisted for 12 days after discontinuation of fluoxetine, when norfluoxetine still was detectable in plasma but fluoxetine was not (27). Norfluoxetine appears to be a more potent inhibitor of CYP3A4 than fluoxetine and is known to accumulate in the plasma after multiple dosing of fluoxetine. In human liver microsomes, the Ki for inhibition of conversion of midazolam to 1′-hydroxymidazolam by norfluoxetine was 19.1 µmol/L (24). Because this metabolite has a lower Ki for CYP3A4 than its parent, we investigated the possibility that inhibition of CBZ metabolism was the result of inhibition by this metabolite. In vitro, norfluoxetine demonstrated an ability to inhibit the CYP3A4-mediated epoxidation of CBZ and 10-hydroxylation of (R)-warfarin only at norfluoxetine concentrations that were 25 times higher than expected in plasma in vivo
(unpublished results). Also, norfluoxetine did not significantly inhibit formation of either of the phenolic metabolites of CBZ. Our in vitro results suggest that fluoxetine and its metabolite, norfluoxetine, are not potent reversible inhibitors of CBZ metabolism. In agreement with these results, concurrent fluoxetine administration did not result in decreased clearance of the CYP3A4 substrates, cyclosporine, triazolam, and terfenadine (28, 29, 30). Recently, however, fluoxetine was shown to bind irreversibly to and inactivate CYP3A4 in vitro at rates sufficient to affect in vivo concentrations of CYP3A4, resulting in inhibition of clearance (31).
Fluvoxamine is a moderate inhibitor of CYP3A4 and inhibited the in vitro metabolism of alprazolam, terfenadine, and triazolam, which are metabolized predominantly by CYP3A4 (32,33, 34). In addition, fluvoxamine decreased the clearance of alprazolam by 55% in 10 healthy men (35). Interactions between CBZ and fluvoxamine have been suggested in case reports (36,37). Addition of fluvoxamine 600 mg/day to the regimen of a patient previously maintained on CBZ increased plasma concentrations of CBZ from 7.3 to 12.4 µg/mL (38). However, no interaction was observed in a controlled study of fluvoxamine addition to steady-state CBZ therapy in seven subjects (23).
Viloxazine has been suggested to be less epileptogenic than conventional antidepressants and therefore can be used in depressed epileptic patients. Viloxazine was added to the therapeutic regimen of six epileptic patients with symptoms of depression, and plasma levels of CBZ increased 55%, from 7.5 ± 3.2 to 11.6 ± 4.8 µg/mL (39). CBZ epoxide levels also increased from 0.76 ± 0.32 to 0.88 ± 0.40 µg/mL. These results suggest that viloxazine is an inhibitor of oxidative metabolism to the epoxide, as well as inhibiting epoxide clearance to the trans-diol. However, further work to determine the effect of viloxazine on CBZE clearance after administration of CBZE demonstrated no decrease in clearance to the trans-diol metabolite (40).
The steady-state AUC of CBZ increased from 60.77 ± 8.44 to 74.98 ± 12.88 µg/hr/mL (p < .001) in 12 subjects after the addition of nefazodone (200 mg/day for 5 days). Also, during the combined administration period, the steady-state AUC of nefazodone decreased from 7,326 ± 3,768 to 542 ± 191 ng /hr/mL, suggesting that CBZ induces the clearance of nefazodone (41).
Other Psychoactive Agents
Elevation of the CBZE:CBZ ratio in plasma, possibly related to inhibition of epoxide hydrolase, has been reported in patients receiving CBZ and the antipsychotic loxapine; however, no adverse effects were noted (42).
Valnoctamide, a valproic acid derivative used in some countries as an over-the-counter tranquilizer, has been shown to inhibit epoxide hydrolase both in vitro and in vivo, and by this mechanism to increase the plasma concentration of CBZE in CBZ-treated patients (43). The interaction can result in clinical signs of CBZ intoxication.
Calcium Channel Blockers
Coadministration of CBZ and diltiazem has been reported to increase CBZ plasma concentrations up to 50% (44, 45, 46, 47). Two diltiazem metabolites (N-desmethyl-diltiazem andN,N-didesmethyl-diltiazem) are more potent inhibitors (11 and 200 times, respectively) of CYP3A4-mediated testos-terone-6-β-hydroxylation than diltiazem, suggesting that these metabolites are most likely the major contributors to this interaction (48).
Verapamil (120 mg three times a day) was given as adjunctive therapy to six patients with refractory partial epilepsy receiving CBZ. Within a few days, symptoms of CBZ neurotoxicity developed in all six patients. There was a mean rise of 46% in total and 33% in free plasma CBZ concentrations in five of these patients (49). Verapamil is extensively metabolized in the liver by several CYP isoenzymes, including CYP3A4, CYP2C8, and CYP1A2, to several metabolites (50,51). However, the in vitro Ki values determined for the inhibition of CYP3A4 by verapamil are significantly higher than levels attained therapeutically. In addition, Km values determined for verapamil metabolism by CYP3A4 and CYP2C8 also are significantly higher than therapeutic verapamil plasma concentrations, suggesting that simple competitive inhibition by verapamil in vivo is unlikely (50). In vitro, inhibition of CBZ metabolism was found to be minor at therapeutic concentrations of verapamil (unpublished results from our laboratory). One metabolite of verapamil, norverapamil, accumulates in vivo at plasma levels comparable with the parent drug, and could contribute to inhibition of CBZ metabolism (52). In addition, verapamil concentrations at the active site of the enzyme may be significantly higher than measured in plasma because mouse liver concentrations are 1,000-fold higher than mouse plasma concentrations (53).
For the purposes of drug interactions, macrolide antibiotics can be classified into three groups (54). The first includes troleandomycin and erythromycin, which are metabolized
by CYP3A4 to nitrosoalkanes that form stable complexes with the heme of CYP and render it inactive (55). CBZ interactions with erythromycin and troleandomycin have been reported on multiple occasions since the late 1970s (56). Mesdjian et al. described 17 patients on CBZ who became intoxicated within 24 to 48 hours after receiving troleandomycin (8 to 33 mg/kg/day) (57). Symptoms of toxicity disappeared 2 to 3 days after troleandomycin was withdrawn. Erythromycin interaction reports typically note a twofold to fourfold increase in serum CBZ concentrations within 24 to 72 hours after the addition of erythromycin.
Josamycin, flurithromycin, roxithromycin, clarithromycin, miocamycin, and midecamycin comprise the second group, which form complexes to a lesser extent, are less potent inhibitors, and sometimes cause drug interactions. Slight but statistically significant decreases in plasma clearance of CBZ were reported after single or multiple doses of josamycin in healthy volunteers and in patients with epilepsy, although none of the subjects had symptoms of CBZ toxicity (58,59). In normal volunteers, coadministration of flurithromycin (500 mg three times daily for 10 days) resulted in a slight but statistically significant increase in CBZ AUC and a 25% decrease in the AUC of the CBZE metabolite, suggesting inhibition of this pathway of CBZ clearance (60). Coadministration of roxithromycin did not affect the clearance of CBZ (61). Administration of clarithromycin (500 mg/day) to a patient maintained on CBZ resulted in increased plasma concentration of CBZ and decreased plasma concentration of CBZE, indicating inhibition of the epoxide pathway of CBZ elimination (62). Miocamycin caused a 13% increase in the CBZ AUC and a 26% decrease in the AUC of CBZE (63).
Members of the third group—spiramycin, rokitamycin, dirithromycin, and azithromycin—do not inactivate CYP and do not appear to modify the clearance of other compounds cleared by CYP3A4 (54,64).
Addition of the antitubercular agent isoniazid to the therapeutic regimen of 13 patients receiving CBZ resulted in elevated plasma CBZ concentrations and signs of CBZ toxicity in 10 of those patients (65). In one patient, CBZ clearance decreased 45% when isoniazid (300 mg for 3 days) was coadministered (66). Another patient demonstrated elevated CBZ serum levels and symptoms of CBZ toxicity 5 days after the addition of isoniazid to his regimen (67). In vitro experiments demonstrated that isoniazid is a potent inhibitor of CBZ epoxidation and of CYP3A4-mediated (R)-10-warfarin hydroxylation at a therapeutically relevant concentration (unpublished results). These findings suggest that isoniazid is a CYP3A4 inhibitor. In support of this hypothesis, the involvement of CYP3A4 in the metabolism of vincristine is well established, and isoniazid is suspected of inhibiting vincristine clearance in at least one published case report (68).
Ketoconazole is a potent CYP3A4 inhibitor that causes clinically significant drug interactions with CYP3A4 substrates such as cyclosporine, tacrolimus, triazolam, midazolam, alprazolam, and terfenadine (69). In one study, addition of ketoconazole (200 mg/day orally for 10 days) to eight patients stabilized on CBZ resulted in a 29% increase in mean plasma CBZ levels (70). This effect was less pronounced than expected based on ketoconazole interactions reported with other CYP3A4 substrates.
There is at least one report of a patient who experienced CBZ toxicity after the addition of fluconazole to his drug regimen. Plasma CBZ levels increased from 11.1 to 24.5 µg/mL after taking 150 mg of fluconazole for 3 days. CBZ plasma levels returned to normal after discontinuation of fluconazole (71). This, in relation to the finding with ketoconazole, contradicts the rank order of inhibition potencies found with other CYP3A4 substrates (69).
Ritonavir, an antiretroviral agent, has been reported to be a potent mechanism-based inhibitor of CYP3A4, and there are at least two case reports of elevated CBZ plasma levels after comedication with ritonavir (72). A patient receiving CBZ (350 mg/day for 8 years) and zonisamide (140 mg/day) had a CBZ plasma concentration of 9.5 µg/mL before receiving ritonavir. Twelve hours after a single 200-mg dose of ritonavir, the serum CBZ concentration was increased markedly to 17.8 µg/mL (73). Another patient became toxic on CBZ when his antiviral drug regimen was changed from stavudine (40 mg twice daily), lamivudine (150 mg twice daily), and indinavir (800 mg three times daily) to lamivudine (150 mg twice daily), didanosine (400 mg/day), ritonavir (600 mg twice daily), and saquinavir (400 mg twice daily). At the time of the antiviral drug change, CBZ plasma concentrations were 6.5 µg/mL. The patient developed dizziness and a progressive gait disorder that initially was attributed to left hemiparesis. CBZ plasma levels were checked 2 months later and found to be 18 µg/mL; after CBZ therapy was discontinued, the gait disorder disappeared (74).
Metronidazole has been reported to increase plasma CBZ levels, possible by inhibiting its metabolism (75).
CBZ toxicity resulting from the coadministration of CBZ and propoxyphene has been the subject of a multitude of
case reports in the literature (76, 77, 78). Also, in two clinical trials, propoxyphene (65 mg three times per day for 3 to 7 days) added to drug regimens of patients stabilized on CBZ resulted in increased plasma levels of CBZ by 45% to 77% (79,80). In a population kinetics study, the doses of CBZ and propoxyphene were found to be lower among patients who used a combination of the two drugs than among those who used only one. However, the mean level of CBZ in the serum was significantly higher and the serum CBZE level was significantly lower among the patients who used the combination of CBZ and propoxyphene, indicating an inhibition of the metabolism of CBZ (81). Propoxyphene (65 mg/day) was found to inhibit the clearance of alprazolam (a drug that appears to be metabolized exclusively by CYP3A4) by 40% (1.3 to 0.8 mL/min/kg) in eight patients, further suggesting that propoxyphene is an inhibitor of CYP3A4 metabolism (82).
In human liver microsomes, concentration-dependent inhibition of CBZ-10,11-epoxidation by propoxyphene was observed, although inhibition was mild. Formation of the 10-hydroxylated metabolite of (R)-warfarin, a probe of CYP3A4 activity, paralleled the findings with CBZ, suggesting that propoxyphene is a weak inhibitor of CBZ metabolism by CYP3A4 (unpublished data from our laboratory).
The inhibition noted clinically may result from contribution of the propoxyphene metabolite, norpropoxyphene, which has been shown to inhibit microsomal oxidative metabolism and to accumulate in the plasma at levels up to 13 times those of its parent compound (83,84). Alternatively, tissue distribution studies in the rat demonstrated that propoxyphene distributed into tissues at concentrations that were 10 to 20 times greater than in blood (85). Highest concentrations were observed in liver. These results, in conjunction with our in vitro study results, suggest that the observed clinical drug interaction may be the result of inhibition of the CYP3A4-mediated metabolism of CBZ by propoxyphene or its metabolite. In addition, the N-dealkylation of propoxyphene has been suggested to result in irreversible inhibition of CYP3A4 (31).
Although CBZ intoxication has been reported after addition of cimetidine, the interaction is not identified consistently and probably is of limited clinical relevance (8). Any increase in CBZ concentration appears to be relatively small and may be transient. This was demonstrated in a study of eight subjects who received CBZ (300 mg twice daily) for 42 days (days 1 to 42) and cimetidine (400 mg three times a day) for 7 days (days 29 to 35). CBZ plasma concentrations increased 17% after 2 days of cimetidine treatment but returned to the precimetidine level by the seventh day of cimetidine administration (86).
Danazol, a synthetic estrogen used in the management of endometriosis, inhibits the epoxide-trans-diol elimination of CBZ, resulting in 50% to 100% increases in steady-state plasma CBZ levels. A study of one patient maintained on CBZ (600 mg/day for 5 years) demonstrated an increase in CBZ plasma steady-state concentrations (41.6 to 68.4 µmol/L) after 33 days of danazol comedication (600 mg/day) (87). Another patient's previously stable CBZ plasma level increased from 38 to 76 µmol/L after beginning danazol administration (88). A study of six women with epilepsy and fibrocystic breast disease reported a 91% mean increase in CBZ concentration (from 7.55 to 14.45 µg/mL) after danazol was added (89).
Nicotinamide caused elevation of CBZ levels in two patients, with high correlations between CBZ clearance and nicotinamide doses (90).
Induction of Carbamazepine Metabolism
Metabolism of CBZ in vivo is induced by some other anticonvulsants, resulting in decreased plasma concentrations of CBZ. In vitro evidence obtained from incubation of CBZ in human liver microsomes demonstrated that CBZE formation velocity in livers obtained from donors who received CYP enzyme-inducing agents (e.g., rifampin, dexamethasone, and phenytoin) was 1.9-fold higher than that in livers from uninduced donors (2). Because most CBZE formation is mediated by CYP3A4, induction of CYP3A4 by these agents is suggested.
Phenytoin, Phenobarbital, and Primidone
Several studies have consistently demonstrated that CBZ metabolism is highly inducible by other antiepileptic drugs, including phenytoin, phenobarbital, and primidone (8). In patients receiving multiple drugs, the inducing effects of multiple anticonvulsants can be additive in terms of reducing CBZ concentrations (91). Christiansen and Dam (92,93) showed in 123 patients that the slope of the relationship between plasma level and dose of CBZ is reduced separately by phenytoin, phenobarbital, and the combination of phenytoin and phenobarbital. Schneider (94) reported similar findings in 184 patients receiving 8 different combinations of CBZ and other drugs. This study showed that primidone also has a significant effect in decreasing the slope of the relationship between plasma CBZ level and dose, presumably as a consequence of enzyme induction. In 142 patients, Johannessen and Strandjord (95) also found that phenytoin or phenobarbital, or a combination of both, could reduce the plasma
level-to-dose ratio of CBZ. They emphasized the need to monitor CBZ levels during polytherapy to achieve the proper dosage increments. Rane et al. (96) compared the mean CBZ levels in children on monotherapy and polytherapy and found that the ratio of plasma level to dose was lower in the group on polytherapy, but the concentration of CBZE metabolite was significantly higher.
Several other studies have found that the ratio of CBZE to CBZ steady-state plasma levels was higher in patients taking CBZ with other antiepileptic drugs than in those taking CBZ only (93,94,96, 97, 98). There is some evidence that epoxide hydrolase also is subject to induction by pretreatment with inducing agents such as phenytoin and phenobarbital (99,100).
Korczyn et al. studied the plasma levels of CBZ and the epoxide and dihydrodiol metabolites in two groups of patients (monotherapy versus polytherapy) and concluded that other anticonvulsant agents induce the epoxidation of CBZ as well as the conversion of the epoxide to the dihydrodiol (101). Induction of CBZE clearance by phenobarbital was demonstrated in six epileptic patients stabilized on phenobarbital and in six drug-free, healthy volunteers (7). The plasma clearance of a single dose of CBZ epoxide was significantly higher in the patient group than in the control (220.2 ± 63.5 versus 112.5 ± 46). Because CBZ epoxide is almost completely converted to trans-CBZ diol by microsomal epoxide hydrolase, it appears that phenobarbital induces this enzyme (4,5). The clinical significance of these interactions has not been shown.
Coadministration of CBZ and phenytoin appears to result in a simultaneous dual effect, inhibition or induction of phenytoin metabolism by CBZ (see page 253, under “Phenytoin”) and induction of CBZ metabolism by phenytoin. The mean serum concentration of CBZ was 59% lower in patients receiving CBZ in combination with phenytoin, compared with patients receiving only CBZ (102). Conversely, reduction of phenytoin dosage resulted in increased CBZ concentrations (103). In another study that examined serum concentrations of CBZ and its major metabolites, CBZE and CBZ-diol, phenytoin comedication decreased CBZ serum levels and increased the concentrations of both CBZE and CBZ-diol (104). These results suggest that phenytoin is an inducer of CYP3A4 and CBZ epoxidation.
Felbamate has been identified as an inducer of CYP3A4 (105). Several studies show a decrease in serum concentrations of CBZ with felbamate comedication. Addition of felbamate to 22 patients on stable CBZ monotherapy resulted in a decrease of 10% to 42% in CBZ levels (106). A double-blind, crossover, placebo-controlled study of 32 patients stabilized on CBZ and phenytoin demonstrated that felbamate decreased mean CBZ concentrations (7.5 to 6.1 µg/mL) and increased CBZE concentrations (1.8 to 2.4 µg/mL) (107,108). Because of the increase in plasma CBZE levels and a concomitant pharmacodynamic interaction, felbamate often potentiates the adverse effects of CBZ.
Oxcarbazepine has been found to produce a modest reduction in the plasma levels of CBZ at steady state (109).
St. John's Wort
St. John's Wort, a herbal medicine used as an antidepressant and for other indications, is an inducer of CYP3A4. At a dosage of 900 mg/day, St. John's Wort has been found to reduce plasma CBZ levels after a single dose of CBZ (110), but in a separate study, steady-state plasma CBZ levels were not affected by St. John's wort given at the same dosage (111).
Plasma Protein Binding Interactions
CBZ is only 75% bound in human plasma, and interactions resulting from displacement of CBZ by other drugs or from displacement of coadministered drugs by CBZ are likely to be free from clinical consequences. The influence of several drugs on the plasma binding of CBZ at a concentration of 5 µg/mL was studied by ultrafiltration. It was found that phenobarbital, phenytoin, and nortriptyline had no effect on the protein binding of CBZ, whereas ethosuximide showed a very slight increase in the percentage of CBZ bound to protein (112). Valproic acid was found to cause a 25% increase in the free fraction of CBZ (23.5% to 29.5%), although wide fluctuations in the CBZ dose-toplasma level relationship are likely to obscure any clinical effect (113,114).
EFFECT OF CARBAMAZEPINE ON OTHER DRUGS: INDUCTION
The number of drug interaction reports resulting from CYP induction by CBZ has been increasing. CBZ autoinduces its own metabolism through the CYP3A4-mediated pathway during long-term administration (115, 116, 117). In humans, the induction half-life of CYP3A4 appears to be approximately 4 days, and therefore a period of 3 to 4 weeks of CBZ administration is required to achieve steady state (118). The increased clearance is associated with a shortened half-life and a reduction in the total serum CBZ concentration at steady state compared with single dose. One study in four patients found that the average steady-state concentration of CBZ was reduced by 50% after 3 weeks of drug administration (115). Induction of CYP3A4 by CBZ has been confirmed in human hepatocyte cultures (119). This inductive effect on CYP3A4 also decreases the plasma
concentrations of other coadministered drugs that are CYP3A4 substrates. It is likely, however, that CBZ induces other CYP isoforms because in vivo, CBZ appears to increase the metabolic clearances of olanzapine, bupropion, the active (S) enantiomer of warfarin, and phenytoin, which appear to be substrates of CYP1A2, CYP2B6, CYP2C9, and CYP2C9/CYP2C19, respectively. The hypothesis of induction of CYP1A2 is supported by another study that found that CBZ treatment increased the percentage of labeled caffeine exhaled as carbon dioxide, a method used to assess CYP1A2 activity in vivo (120).
CBZ (300 mg/day for 10 days) significantly decreased the plasma concentration of a single oral dose of alprazolam by increasing the oral clearance (0.90 versus 2.13 mL/min/kg) and shortening the elimination half-life (17.1 versus 7.7 hours) (121) (121a). Alprazolam is cleared primarily by CYP3A4, and the induction of this enzyme by CBZ is the likely cause for the increased clearance of this drug.
Clobazam is a benzodiazepine used as adjuvant therapy for intractable seizures. A pharmacokinetic study in healthy volunteers found that CBZ (200 mg twice daily for 2 weeks) decreased clobazam plasma steady-state concentrations by 61% and increased concentrations of its major metabolite, N-desmethylclobazam, by 44% (122). The results of a study in epileptic children also showed that CBZ decreased clobazam serum concentrations and increased the concentrations of N-desmethylclobazam.
Clonazepam plasma concentrations may be reduced by CBZ (123). CBZ (200 mg/day) decreased clonazepam steady-state concentrations by 19% to 37% after a 5- to 15-day administration period (118).
Enzyme-inducing anticonvulsants reduce the plasma levels of the active metabolite of clorazepate, N-desmethyldiazepam (124).
Enzyme-inducing anticonvulsants, including CBZ, have been found to cause a reduction in the plasma concentration of diazepam and concomitantly to increase the plasma levels of its active metabolite, N-desmethyldiazepam (125).
The AUC of an oral 15-mg dose of midazolam in patients on chronic CBZ therapy was reduced to 5.7% of the AUC in noninduced control subjects (126). This dramatic effect was attributed to the significant first-pass metabolism of midazolam and the induction of both intestinal and hepatic CY3A4 by CBZ. Because after intravenous (i.v.) administration, midazolam clearance depends more on liver blood flow than on microsomal enzyme activity, the reduction of plasma midazolam levels in patients treated with CBZ would be expected to be far less significant when midazolam is given i.v., such as in the treatment of status epilepticus.
When CBZ (200 mg/day) was added to the regimen of healthy subjects receiving 250 mg/day of ethosuximide, ethosuximide clearance increased significantly after 10 days of CBZ therapy, with a synchronous decrease in ethosuximide half-life from 54 to 45 hours (127).
Felbamate clearance is increased significantly by enzyme-inducing anticonvulsants, including CBZ (128).
CBZ also is an inducer of uridine diphosphate-glucuronosyl transferase, and this appears to be the basis for increased rate of lamotrigine clearance when these drugs are used together. The half-life of lamotrigine is reduced from 24 to 15 hours in patients receiving enzyme-inducing anticonvulsants, including CBZ (129).
Phenobarbital and Primidone
Phenobarbital is a product of primidone metabolism, and serum phenobarbital concentrations increased significantly when CBZ was added to the regimen of four patients receiving primidone (130). The increase in phenobarbital levels appears to be due to increased conversion from primidone rather than inhibition of phenobarbital clearance, because one study demonstrated no change in phenobarbital clearance in 25 patients taking both phenobarbital and CBZ (131).
The effect of CBZ on phenytoin pharmacokinetics appears to be variable. CBZ has been reported to shorten phenytoin half-life and reduce phenytoin plasma concentrations at steady state (132, 133, 134, 135, 136). However, there also have been reports of CBZ causing prolonged phenytoin half-life and increased plasma phenytoin concentrations (137, 138, 139). In particular, one study found that CBZ caused an increase in steady-state phenytoin concentrations in half of the patients studied and no change in the remainder (138). These data suggest that CBZ may both induce and inhibit the enzymes
responsible for phenytoin biotransformation, and that the prevailing effect may vary from patient to patient.
Phenytoin is metabolized predominantly by CYP2C9 (140) and to a smaller extent by CYP2C19 (141). Induction of CYP2C9 is suggested as the cause of decreased anticoagulation reported in patients stabilized on warfarin and then prescribed CBZ (142). Inducibility of CYP2C19 by rifampicin has been demonstrated, but there are no reports regarding inducibility of CYP2C19 by CBZ (143). In vitro, CBZ has been shown to inhibit the CYP2C19-mediated 4′-hydroxylation of (S)-mephenytoin, with a Ki value of 35 µmol/L (therapeutic range of CBZ is 25 to 50 µmol/L) (unpublished data from our laboratory). The available evidence suggests that the effect of CBZ on phenytoin pharmacokinetics could be the result of a balance between induction of CYP2C9 and inhibition of CYP2C19.
Tiagabine is extensively metabolized in the liver, largely by CYP3A4, with less than 1% excreted unchanged (129). The monotherapy half-life of tiagabine is 5 to 8 hours and is shortened to 2 to 3 hours when coadministered with CYP3A4 inducers, including CBZ. Tiagabine does not alter the clearance of CBZ.
Topiramate metabolic clearance is increased by concomitant CBZ administration, whereas CBZ clearance is unchanged (144). When patients were changed from CBZ plus topiramate to topiramate monotherapy, topiramate clearance was reduced by approximately 50%, suggesting that an adjustment in topiramate dosage may be required when CBZ is discontinued (145).
Several studies indicate that CBZ increases the plasma clearance of valproic acid, probably through induction of the metabolic processes responsible for valproic acid elimination. Six healthy subjects received 250 mg of valproic acid twice daily for 4 weeks; a very small dose of CBZ (200 mg once daily) was added after 4 days of valproic acid therapy (146). A significant decrease in steady-state blood valproic acid levels was apparent after 2 weeks of CBZ therapy.
The effects of CBZ on specific metabolic pathways of valproic acid were examined in epileptic patients (147). The formation of Δ4-valproate (hepatotoxic metabolite of valproic acid) was increased by 105% in the presence of CBZ. The clearances by several other pathways, ω and ω-1 oxidation (both CYP mediated) as well as glucuronidation, also were increased.
Zonisamide is metabolized partly by CYP3A4 (148). The plasma half-life of zonisamide is reduced from 60 hours in healthy subjects to 36.4 hours in patients receiving CBZ (149,150). In another study, CBZ was found to decrease the zonisamide steady-state plasma concentration-to-dose ratio, indicating that zonisamide clearance is increased by CBZ coadministration (151).
In addition to CYP2D6 (a noninducible enzyme) and CYP1A2, CYP3A4 is partially responsible for a minor fraction of demethylation of the tertiary amine tricyclic antidepressants (amitriptyline, clomipramine, and imipramine), and CBZ may induce the metabolism of these drugs (152). One study found that children receiving CBZ required higher doses of imipramine than control subjects (153). Despite the higher imipramine dosages, the CBZ group had lower plasma concentrations of imipramine and its active metabolite, desipramine. CBZ has been associated with a decrease in amitriptyline serum levels (42%) and in the levels of its active metabolite, nortriptyline (40%) (154). Because the metabolites of imipramine and amitriptyline are active, induction of metabolite clearances may play a role in the diminished effects of these drugs.
The clearance of nortriptyline appeared to be increased by the addition of CBZ in a 73-year-old woman (155). Average serum concentrations decreased from 355 ± 49 nmol/L (13 samples over 2 years) to 140 and 134 nmol/L (at two separate measurements) 9 weeks after beginning CBZ (600 mg/day). The main metabolic pathway of nortriptyline is mediated by CYP2D6 (high affinity) and CYP3A4 (low affinity) (156).
Patients with mood disorders received a single oral dose of bupropion (150 mg) while receiving placebo (n = 17) or chronic blind CBZ (n = 12). The AUC of bupropion was 90% lower in patients receiving CBZ, suggesting that CBZ induces the clearance of bupropion (157). Bupropion undergoes extensive biotransformation primarily to hydroxybupropion by CYP2B6 (158,159).
The metabolism of many other antidepressants, including mianserin (160) and nefazodone (41), is accelerated by concomitant treatment with enzyme-inducing anticonvulsants.
Clozapine is an atypical antipsychotic drug that is used mainly for the treatment of refractory schizophrenia. Clozapine is eliminated by oxidation in the liver, predominantly by CYP1A2. CBZ was found to decrease the plasma levels of clozapine by 47% in a study of 12 patients (161). Another study compared patients receiving clozapine alone with those also receiving CBZ, and found that patients on CBZ had a mean 50% lower concentration-to-dose ratio than the monotherapy group (p < .001), indicating that CBZ is an inducer of the metabolism of clozapine (162).
Comedication with CBZ and haloperidol is associated with lower haloperidol plasma levels and worse clinical outcome than haloperidol alone (163). In a study of 231 schizophrenic patients, patients who received CBZ and haloperidol concomitantly had a mean haloperidol concentration-to-dose ratio that was 37% less than that of subjects who received haloperidol alone (164). Several other studies found that haloperidol plasma levels are reduced by more than 50% after the addition of CBZ (165).
Olanzapine is an antipsychotic agent cleared mainly by CYP1A2. Olanzapine clearance was compared before and after a 2-week treatment with CBZ (200 mg twice daily) in 12 healthy volunteers. The olanzapine AUC was significantly decreased after the addition of CBZ (336 ± 103 hr·mg/L versus 223 ± 59 hr·mg/L, p < .001), suggesting that CBZ is an inducer of CYP1A2 (166).
Risperidone is an antipsychotic agent cleared mainly by CYP2D6, although CYP3A4 is a contributor. In five patients assessed with and without CBZ comedication, dose-normalized plasma risperidone and 9-hydroxyrisperidone levels were significantly lower when the patients received combination therapy, suggesting that CBZ can induce the clearance of this drug (167). This interaction seems to have considerable clinical significance, particularly in patients with deficient CYP2D6 activity (168).
Breakthrough bleeding and contraceptive failure have been reported in women taking oral contraceptives (169,170). In a pharmacokinetic study of four women who received a single dose of oral contraceptive containing ethinyl estradiol (50 µg) and levonorgestrel (250 µg) before and 8 to 12 weeks after beginning therapy with CBZ (171), the AUCs of ethinyl estradiol and levonorgestrel were decreased by 40% when CBZ was added, suggesting that CBZ induced the metabolism of both hormone components of the oral contraceptive preparation. CYP3A4 is the major enzyme responsible for the 2-hydroxylation of ethinyl estradiol (172,173).
Enzyme-inducing anticonvulsants, including CBZ, have been found to increase the metabolic clearance of prednisolone, methylprednisolone, dexamethasone, and many other steroids (8). These interactions can result in inadequate response when these steroids are used therapeutically or diagnostically.
CBZ has been reported to induce the metabolism of the immunosuppressive drug, cyclosporine, which is metabolized by CYP3A4 (174, 175, 176). Cyclosporine pharmacokinetics were studied in three pediatric renal transplant recipients receiving CBZ doses of 16.4 to 20.8 mg/kg/day and compared with control patients. Steady-state trough concentrations of cyclosporine were significantly lower in the CBZ group (57 ± 14 ng/mL versus 162 ± 22 ng/mL), even though the CBZ patients received higher cyclosporine doses (16.2 ± 8.8 mg/kg/day versus 10.8 ± 5.2 mg/kg/day) (177).
Patients stabilized on warfarin exhibit increased prothrombin times after the addition of CBZ to their drug regimens, suggesting that CBZ induces warfarin metabolism (142). Also, a patient maintained on CBZ and warfarin discontinued her CBZ without consulting her physician and suffered excessive hyperprothrombinemia and hemorrhage (178). Up to twofold reductions in warfarin dosage have been required when CBZ therapy was discontinued (179). Warfarin is an enantiomeric compound, with the clearance of the more active (S)-warfarin mediated by CYP2C9 and the less active (R)-warfarin by CYP3A4 (180,181). Although it is likely that induction of CYP3A4 increases the biotransformation of the (R) enantiomer, the degree of the clinical interaction between CBZ and warfarin suggests that CYP2C9 also is induced by CBZ. The metabolism of other coumarin anticoagulants also has been found to be stimulated to a clinically important extent by CBZ (182).
Dihydropyridine Calcium Antagonists
Nimodipine is a calcium channel blocking agent administered as a racemic mixture to prevent cerebral vasospasm in
patients with subarachnoid hemorrhage, and to improve cerebral function in elderly patients. In epileptic patients taking CBZ, there was a sevenfold decrease in the AUC of nimodipine (183). The plasma levels of other dihydropyridine calcium antagonists, including nisoldipine, nifedipine and felodipine, are markedly reduced by enzyme-inducing anticonvulsants (8).
In five patients on long-term CBZ therapy, the half-life of doxycycline (8.4 hours) was significantly shorter than the mean half-life of 15.1 hours in nine control patients (184).
CBZ is prescribed for treatment of seizures or postherpetic neuralgia in human immunodeficiency virus-infected patients. In one case report, the plasma concentration of the antiretroviral drug indinavir (800 mg every 8 hours) was decreased up to 16 times after the addition of CBZ (200 mg/day) (185).
Plasma concentrations of the antifungal itraconazole are markedly reduced by concomitant administration of CBZ. Induction of itraconazole metabolism is of clinical relevance in systemic mycoses because treatment failures have occurred in patients receiving CBZ and itraconazole. Bonay et al. described a patient who had undetectable plasma itraconazole concentrations during coadministration with CBZ (186). Three weeks after withdrawal of CBZ, plasma itraconazole concentrations increased and reached the therapeutic range without modification of the antifungal dosage (186).
The antineoplastic agent, vincristine, is at least partially metabolized by CYP3A4. One clinical study found vincristine clearance was 63% higher in a patient group receiving CBZ (eight patients) or phenytoin (one patient) compared with a control patient group, suggesting that CBZ is an inducer of vincristine clearance (187).
The primary route of fentanyl clearance is N-dealkylation to norfentanyl, and this step is catalyzed predominantly by CYP3A4 (188). The effect of chronic anticonvulsant therapy on the minimum dose of fentanyl required during cranial surgery was studied in four groups of patients receiving either no anticonvulsants; CBZ alone; CBZ with either valproic acid or phenytoin; and CBZ, valproic acid, and either phenytoin or primidone. The patients receiving any of the anticonvulsant therapies required significantly more fentanyl during anesthesia (189).
Neuromuscular Blocking Agents
Patients receiving CBZ have been found to exhibit a reduced sensitivity to some neuromuscular blocking agents. Both enzyme induction and pharmacodynamic factors may be involved in these interactions (8).
Because CYP3A4 and other enzymes that are induced by CBZ metabolize a large number of drugs, the clearance of many other compounds is likely to be increased in patients taking CBZ. In particular, because the enzyme-inducing spectrum of CBZ is similar to that of phenytoin and phenobarbital, it is likely that many of the interactions mediated by enzyme induction with the latter drugs (Chapters 53 and 60) also are seen with CBZ.
Identification of CYP3A4 as the primary catalytic enzyme for the main clearance pathway of CBZ allows an understanding of the effects of several drugs on plasma CBZ concentrations. Drugs that inhibit CYP3A4 increase plasma CBZ. CYP3A4 is inducible, and this explains the decreases in CBZ levels during coadministration of CYP3A4 inducers. Also, CBZ appears to induce CYP3A4, CYP2C9, CYP2C19, and CYP1A2, resulting in decreased plasma concentrations of substrates of these isoforms when they are prescribed with CBZ.
Work on this chapter was supported in part by National Institutes of Health grant P01 GM 32165.
121a. Furukori H, Otani K, Yasui N, et al. Effect of carbamazepine on the single oral dose pharmacokinetics of alprazolam. Neuropsychopharmacology 1998; 18:364-369.