Steven J. Leeder PharmD, PhD
Associate Professor, Department of Pediatrics and Pharmacology, University of Missouri, Kansas City; and Chief, Section of Developmental Pharmacology and Experimental Therapeutics, Children's Mercy Hospital and Clinics, Kansas City, Missouri
The potential for drug-drug interactions continues to be an important consideration when phenobarbital is included in a given patient's pharmacotherapeutic regimen. Although alterations in drug pharmacokinetics and pharmacodynamics are theoretically possible, almost all documented, clinically relevant drug-drug interactions involving phenobarbital result in pharmacokinetic changes in the concomitantly administered medication. The nature and extent of the pharmacokinetic interactions are best documented through evaluation of parameters such as total body clearance, area under the blood, plasma, or serum concentration-time curves (AUC), distribution volume, elimination half-life, and peak or steady-state plasma concentrations.
Because most drug interactions involving phenobarbital occur at the level of drug biotransformation, the potential for a clinically significant interaction increases if drug elimination is highly dependent on a single metabolic pathway that is subject to modulation by phenobarbital. Furthermore, each individual patient has his or her own unique complement of individual cytochrome P450 (CYP) or glucuronosyl transferase isoforms, and as a consequence, the combination of phenobarbital and drug X may result in lower concentrations of drug X in some patients but may not be associated with any discernible effect in others. Likewise, additional factors such as drug dosages, diet, smoking or alcohol intake, and disease conditions all contribute to the variability in clinical manifestations of a particular drug interaction with phenobarbital.
Considerable progress has been made in evaluating potential drug-drug interactions in humans both in vitro and in vivo. This chapter is largely restricted to information that has been generated by clinical studies conducted in human subjects or through the use of human-derived experimental systems.
PHARMACOKINETIC DRUG INTERACTIONS
Effects of Other Drugs on Phenobarbital Kinetics
Absorption.
Phenobarbital is essentially completely absorbed after oral administration (1), and few, if any, interactions leading to impaired absorption have been reported. Conversely, several studies have documented the utility of activated charcoal in limiting systemic exposure after a phenobarbital overdose. Although absorption of orally administered phenobarbital is reduced by activated charcoal administered in close temporal proximity to the phenobarbital dose (2), several studies also report that the clearance of intravenous phenobarbital is enhanced 60% to 270% by charcoal administered in repeated doses over 36 to 96 hours while half-lives are decreased 2.5- to eightfold (3, 4, 5). Studies in rabbits attribute this effect to the ability of the activated charcoal to disrupt phenobarbital enterohepatic recirculation (6).
Distribution.
Binding of phenobarbital to plasma proteins is approximately 50% (7), and so clinically significant interactions resulting from displacement from protein binding sites are not anticipated.
Biotransformation.
Preliminary data attribute the p-hydroxylation of phenobarbital primarily to the CYPs CYP2C9 and CYP2C19 (8). As a result, coadministration of drugs that either induce or inhibit these CYP isoforms has the potential to modulate phenobarbital pharmacokinetics to a clinically significant extent. For example, addition of chloramphenicol to phenobarbital therapy has been reported to reduce phenobarbital clearance by 40% (9). Because polypharmacy is a common feature of antiepileptic drug therapy, there is an increased potential for drug-drug interactions. Some of the more clinically important interactions are discussed in the following paragraph and are summarized in Table 52.1.
P.505
|
TABLE 52.1. EFFECT OF CONCURRENTLY ADMINISTERED DRUGS OR TREATMENTS ON THE CLEARANCE OF PHENOBARBITAL |
||||||||||||||||||
|
Carbamazepine.
The addition of carbamazepine to patients treated with primidone is reported to result in increased concentrations of the active metabolite, phenobarbital (10). Because concurrent carbamazepine therapy apparently does not affect phenobarbital clearance per se (11), the increased phenobarbital concentrations can be attributed to formation from primidone rather than to reduced phenobarbital clearance. The enzymes responsible for primidone biotransformation to phenobarbital have not been identified.
Felbamate.
Felbamate, an inhibitor of CYP2C19 but not of CYP2C9 in vitro (12), also has been reported to decrease phenobarbital clearance in vivo. The addition of a 9-day course of felbamate (2,400 mg/day) to healthy volunteers receiving phenobarbital, 100 mg daily for 28 days, reduced phenobarbital clearance by 24% from baseline values; no such effect was observed in a parallel group receiving placebo instead of felbamate. The change in phenobarbital clearance was largely (~55%) attributed to a reduction in the formation of p-hydroxyphenobarbital, a finding consistent with a role for CYP2C19 in mediating this pathway of phenobarbital elimination (13).
Valproic Acid.
Valproic acid is associated with a relatively predictable decrease in phenobarbital clearance that manifests initially as drowsiness and increasing somnolence and other signs of concentration-dependent toxicity. Based on single-dose pharmacokinetic studies and studies conducted at steady-state phenobarbital concentrations, valproic acid treatment is associated with an approximately 25% reduction in phenobarbital clearance and a 50% longer half-life (14, 15, 16, 17), attributed to decreased biotransformation to p-hydroxyphenobarbital (15,18). Increases in phenobarbital serum concentrations appear to be greater in pediatric patients (112.5%) compared with adults (50.9%) (19), and the extent of the interaction is characterized by considerable interindividual variability; reports indicate that between 50% and 80% of patients may require a reduction in phenobarbital dose (14,19). Concurrent administration of valproic acid and primidone is associated with an increase in the phenobarbital:primidone ratio consistent with an inhibitory effect on further biotransformation of phenobarbital (20).
Phenytoin.
Phenytoin has been observed to have no effect or to significantly increase plasma phenobarbital concentrations (10,11, 21). When plasma phenobarbital concentrations are derived from primidone, it appears that coadministration of phenytoin and carbamazepine can be expected ultimately to increase phenobarbital levels (10,22). In one study, derived phenobarbital concentrations were 30.6±2.9 mg/L in the presence of phenytoin versus 14.4±3.3 mg/L in its absence, but whether this finding represents increased conversion of primidone to phenobarbital or inhibition of phenobarbital biotransformation has not been unequivocally resolved (22). Overall, the data suggest that phenytoin is less likely to increase phenobarbital concentrations than either valproic acid or felbamate.
Effects of Phenobarbital on the Kinetics of Other Drugs
Almost all clinically significant drug-drug interactions involving phenobarbital are the result of induction, and most involve the CYPs. The effects of phenobarbital on the pharmacokinetics of concomitantly administered medications are summarized in Table 52.2.
Phenobarbital Induction of Drug-Metabolizing Enzymes in Humans
Phenobarbital, a pleiotropic agent that, among other effects, causes proliferation of hepatic smooth endoplasmic reticulum, has been recognized as a prototypical inducer for a group of structurally diverse compounds since the 1960s (23). Extensively studied in rodents and rodent-based systems,
P.506
induction of members of the CYP2A, CYP2B, CYP2C, and CYP3A subfamilies by phenobarbital is now well characterized (24).
|
TABLE 52.2. EFFECT OF PHENOBARBITAL ON THE APPARENT ORAL CLEARANCE OF CONCURRENTLY ADMINISTERED DRUGS |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Studies in human-based experimental systems are consistent with induction of CYP2B6, CYP2C8, CYP2C9, and CYP3A4. For example, 48-hour exposure of cultured human hepatocytes to 3.2 mmol/L phenobarbital produced a modest (1.4- to 1.8-fold) increase in CYP2C8/9 immunoreactive protein and no change in CYP3A4 protein (25,26), whereas 2.0- to 2.5-fold increases in testosterone hydroxylation in the 16α- and 16β-positions with smaller increases in 2β- and 6β-hydroxylated metabolites suggest moderate induction of the human CYP3A and CYP2C subfamilies (27). Chang et al. (27a) observed increases in immunoreactive CYP2B6, CYP2C8, CYP2C9, and CYP3A4 proteins after phenobarbital treatment (2 mmol/L, 96 hours), whereas CYP1A1 and CYP1A2 expression was not affected. Studies with replicating human hepatoma cell lines indicate that phenobarbital also induces CYP3A7, but not CYP3A5, messenger RNA, and immunoreactive protein (28). Nevertheless, phenobarbital is less potent as an inducer of CYP activity than dexamethasone or rifampin. Induction of glucuronosyl transferase (29,30), glutathione S-transferase α (31), and epoxide hydrolase (32) activities by phenobarbital in vitro has also been reported.
The degree of CYP induction observed in primary cultures of human hepatocytes is highly variable. Although some variability is a consequence of the range of culture conditions employed in different laboratories, the use of standardized conditions reveals that induction of CYP activities by phenobarbital and other inducers is highly reproducible in multiple laboratories (33). A more important issue is considerable intersubject variability in the extent of induction observed in vitro, consistent with the variability observed in vivo.
In vivo, phenobarbital and phenytoin treatment of epileptic children resulted in four- to sevenfold greater 6β-hydroxycortisol:cortisol ratios in urine, a marker of CYP3A activity (34), relative to age-matched controls (35,36). These and other studies (37, 38, 39) are consistent with induction of CYP3A activity by phenobarbital in humans, and most drug interactions in which it is involved can be attributed to this factor. Because of the long half-life of the drug, however, maximum effect may not be observed until after 10 to 14 days of therapy.
Mechanism of Induction
Phenobarbital has moderate activity as an activator of the pregnane X receptor (PXR), the orphan nuclear receptor that plays an important role in the transcriptional regulation of theCYP3A4 gene by prototypical inducers such as dexamethasone, rifampin, and clotrimazole (40). The heterodimer formed between PXR and the 9-cis retinoic acid receptor (RXR) binds to specific nucleotide sequences that are organized as an everted repeat separated by six nucleotides (ER-6; TGAACT-N6-AGGTCA). These binding sites are located ≤8,000 base pairs upstream of the CYP3A4 gene transcription start site (41).
Induction of the CYP2B6 gene by phenobarbital appears to be mediated by an additional orphan nuclear receptor, the constitutive androstane receptor (CAR), which also heterodimerizes with RXR. CAR-mediated transcriptional activity is stereospecifically inhibited by 5α-reduced steroids with a 3α-hydroxyl group, specifically androstanol and androstenol, by promoting the release of coactivators (e.g., steroid receptor coactivator-1; SRC-1) from the ligand-binding domain of the receptor (42). Induction by phenobarbital is thought to represent release (or derepression) of CAR from the usual state of inhibition by endogenous steroids (Figure 52.1) (43). Although helpful, this model likely is an oversimplification because supraphysiologic concentrations of androstanol are required for inhibition of CAR, and more recent data implicate either a phenobarbital metabolite that can bind directly to CAR or an indirect mechanism involving phosphorylation to account for the induction process (44).
Interactions with Other Antiepileptic Drugs
Carbamazepine.
In general, phenobarbital increases carbamazepine clearance, and higher doses are required to maintain carbamazepine concentrations at prephenobarbital levels (45). Epileptic patients receiving both agents display lower steady-state carbamazepine concentrations and higher carbamazepine-epoxide concentrations relative to patients receiving carbamazepine monotherapy (46), although the differences (10% to 15% changes) are generally modest (47). Pediatric patients concurrently treated with phenobarbital had 15% lower carbamazepine concentrations (25.2±5.9 µmol/L versus 29.8±8.8 µmol/L) despite receiving higher total carbamazepine doses (21.1±8.1 versus 18.4±7.2 mg/kg/day) than children treated with carbamazepine alone (48). More extensive reviews of the phenobarbital-carbamazepine interaction indicate that steady-state carbamazepine concentrations may decrease as much as 50%, whereas carbamazepine-epoxide concentrations remain unchanged during combination therapy with phenobarbital (49).
Phenytoin.
Because phenobarbital and phenytoin have similar induction profiles and are subject to biotransformation by the same enzymes (CYP2C9 and glucuronosyl transferases), the possibility of induction and competitive inhibition exists. The net outcome of an interaction between the two drugs in a given individual will therefore reflect which, if either, mechanism predominates. For example, phenytoin clearance, half-life, or volume of distribution was not significantly altered at 4 and 12 weeks after the initiation of phenobarbital in one study (50). Furthermore,
P.507
there were no significant differences in the urinary recovery of unchanged phenytoin, its major p-hydroxylated metabolite, or the dihydrodiol metabolite in the presence and absence of phenobarbital. Diamond and Buchanan (51) used single point determinations of phenytoin concentration to determine that the addition of phenobarbital to, or the removal of phenobarbital from, a stable regimen of phenytoin produced no change in phenytoin concentrations. However, because phenytoin is subject to saturable metabolism, inhibition by phenobarbital may be more likely to occur under conditions in which maximum induction has been achieved and phenytoin concentrations are in the high therapeutic range (49).
|
FIGURE 52.1. Proposed mechanism for phenobarbital induction. A: The constitutive androstane receptor (CAR) forms a heterodimer with the retinoic acid X receptor (RXR) and bids to a direct repeat separated by four nucleotides (DR-4 motif). Addition of a coactivator (SRC-1) completes the functional transcription unit. B: Endogenous 5α-androstanes (androstenol) inhibit CAR transcriptional activity by dissociation of the CAR-RXR-DNA complex from the nuclear receptor activator SRC-1. The degree to which this situation exists determines “normal” constitutive activity. C: In the presence of phenobarbital-type inducers, the binding of inhibitory androstanes to CAR-RXR is abolished. The CAR-RXR-DNA complex is then able to interact with the coactivator SRC-1 and reconstitute the functional transcription unit. Thus, phenobarbital induction represents “de-repression” or a reversal of androstane repression. (Adapted from Forman BM, Tzameli I, Choi H-S, et al. Androstane metabolites bind to and deactivate the nuclear receptor CAR-b. Nature 1998;395:612-615; and Sueyoshi T, Kawamoto T, Zelko I, et al. The repressed nuclear receptor CAR responds to phenobarbital in activating the human CYP2B6 gene. J Biol Chem1999;274:6043-6046.) |
Lamotrigine.
N-glucuronidation accounts for most (~80%) lamotrigine clearance in humans (52). Several pharmacokinetic studies (53, 54, 55) indicate that the lamotrigine half-life is shorter (~14 hours) in the presence of a CYP-inducing anticonvulsant such as phenobarbital, phenytoin, or carbamazepine compared with the combination of lamotrigine and valproic acid (40 to 45 hours) or triple therapy of lamotrigine plus an inducing anticonvulsant and valproic acid (~30 hours). Data from a therapeutic drug monitoring service (56) provide further indirect evidence of induction of lamotrigine clearance. In this study, the lamotrigine serum concentration:dose ratio reportedly was lower in patients receiving the drug in combination with phenobarbital (ratio = 0.52), phenytoin (0.32), or carbamazepine (0.57), compared with monotherapy (0.98, p < .05). In the absence of metabolite data, however, alternative mechanisms for lower lamotrigine concentrations may also be operative.
Valproic Acid.
In addition to the inhibitory effects of valproic acid on phenobarbital clearance, phenobarbital increases valproic acid clearance approximately 10% (57). Population-based kinetic studies also indicate that, compared with valproic acid monotherapy, serum concentrations of valproic acid are 25% lower when the drug is given with phenobarbital (58). Studies in children reveal similar results, although the effect of phenobarbital and other CYP inducers on valproic acid clearance is reported to be more pronounced and variable relative to that observed in adults (59).
An additional consideration with respect to phenobarbital interactions with valproic acid therapy is the role of induction by phenobarbital and other CYP-inducing
P.508
antiepileptic drugs (carbamazepine and phenytoin) as risk factors for valproic acid-associated hepatotoxicity, primarily in young children. Retrospective studies of valproic acid-associated hepatotoxicity in the United States identified polytherapy with enzyme inducing antiepileptic medications as well as patient age <2 years, developmental delay, and coincident metabolic disorders as important risk factors for developing this adverse event (60, 61, 62). Although valproic acid-associated hepatotoxicity may occur at any age, data collected between 1978 and 1986 indicate that the risk of fatal hepatotoxicity is highest in children <2 years of age receiving concurrent anticonvulsant therapy; in these children, the incidence of this complication was estimated to be approximately 1:500 (60,61). This represents a 16-fold increase in risk relative to children of the same age receiving valproic acid monotherapy (1:8,000). Comparative estimates of risk for older children aged 3 to 10 years were 1:11,000 during monotherapy and 1:6,000 during polytherapy. Other studies confirmed polytherapy as a risk factor but found little difference in risk between younger (<3 years of age) and older children (3 to 6 years of age) (63,64).
CYP isoforms are responsible for the formation of the potentially reactive metabolite, 4-ene-valproic acid, and this activity is inducible by phenobarbital (65). Further work has specifically implicated CYP2C9 and, to a lesser extent, CYP2A6 in the formation of 4-ene-valproic acid in humans (66). Induction of CYP2A6 and CYP2C9 activities by phenobarbital has not been as rigorously evaluated as has CYP3A4 activity, although modest increases in CYP2C immunoreactive proteins have been observed after phenobarbital treatment of cultured primary human hepatocytes (27,67). Given the comparative variabilities of CYP2A6 activity (30-fold) and CYP2C9 activity (less than fivefold) in human liver microsomes (68), it has been proposed that CYP2C9 may be responsible for most constitutive VPA 4-ene-desaturation, whereas CYP2A6 plays a greater role during polytherapy with anticonvulsants (66). Nevertheless, phenobarbital treatment represents a significant risk factor for the development of valproic acid-associated hepatotoxicity, particularly in children <2 years of age.
Other Antiepileptic Agents.
Approximately 50% to 100% increases in dose are recommended for ethosuximide, felbamate, topiramate, zonisamide, and tiagabine when phenobarbital is added to the treatment regimen (49).
Interactions with Other Medications
Calcium Channel Blockers.
The apparent clearance of a single oral dose of verapamil was reported to be increased fivefold after a 21-day course of phenobarbital. When verapamil was administered intravenously, total drug clearance was increased 200%, whereas free drug clearance was unchanged (69). In a study that considered the effect of the CYP-inducing antiepileptic drugs carbamazepine, phenytoin, and phenobarbital collectively on the disposition of nimodipine, significantly (90%) lower peak concentration and AUC values and a 57% shorter half-life were observed in the anticonvulsant-treated group (70). A similar evaluation of felodipine suggested a reduction in peak concentration and AUC by 82% and 94%, respectively (71). In a study in which a single dose of nifedipine was used as a phenotyping probe for CYP3A4 activity, its clearance was increased by 270% (72). Collectively, the data indicate that bioavailability of calcium channel antagonists is considerably compromised by concurrent administration of phenobarbital, possibly because of induction of intestinal CYP3A isoforms.
Cimetidine.
Long-term phenobarbital administration (100 mg/day for 21 days) increased total body clearance of intravenous cimetidine by an average of 18%, an effect that was largely the result of a 37% increase in nonrenal clearance that was accompanied by increased renal excretion of the sulfoxide metabolite of cimetidine. In the same study, the AUC of orally administered cimetidine was reduced by 15% by the same phenobarbital treatment regimen. The amount of cimetidine and its sulfoxide metabolite excreted in urine were reduced by ~30%, a finding consistent either with impaired absorption or, more likely, with induction of intestinal drug biotransformation (73).
Clozapine.
Phenobarbital treatment was associated with 33% lower plasma clozapine concentrations in a group of patients with schizophrenia compared with a control group matched for sex, age, and body weight who were treated with clozapine alone. Concentrations of clozapine N-oxide were increased more than twofold in the phenobarbital-treated group compared with the control group, whereas norclozapine concentrations did not differ between the two groups (74). These results are consistent with induction of CYP3A4, the CYP isoform primarily responsible for N-oxidation of clozapine (75,76).
Corticosteroids.
In a study comparing the pharmacokinetics of prednisolone in subjects receiving phenobarbital and phenytoin, alone or in combination, versus controls receiving no concurrent medications, prednisolone half-life was significantly shorter (32%) and total body clearance was significantly greater (44%) in the anticonvulsant-treated group, whereas volume of distribution and protein binding were unchanged. Additionally, patients receiving anticonvulsants had more than a twofold mean elevation in their early morning hydrocortisone peak concentrations, a finding suggesting that the concurrent administration of anticonvulsant enzyme inducer with an exogenously administered corticosteroid resulted in less suppression of the hypothalamic-pituitary-axis and a more normal circadian rhythm of endogenous steroid production (77). Similarly, prednisolone half-life was 25% shorter in patients with
P.509
rheumatoid arthritis and was accompanied by symptoms of disease exacerbation (e.g., articular index worsened by 31%, pain score doubled, duration of morning stiffness increased 117%) with the addition of phenobarbital (78).
In 16 asthmatic patients before and 3 weeks after the initiation of phenobarbital, the half-life of intravenously administered dexamethasone decreased significantly (45%) and clearance increased significantly (87%) over prephenobarbital values (79). The authors also described three subjects in whom pulmonary function, eosinophilia, and clinical degree of bronchospasm deteriorated while they received phenobarbital, with subsequent improvement in these measurements on discontinuation of the drug. Similar changes in pharmacokinetic parameters have been reported for intravenously administered methylprednisolone, although changes in the pharmacokinetics of the more water-soluble methylprednisolone sodium hemisuccinate were less dramatic (80). The clinical significance of enzyme induction is most evident in transplant recipients in whom evidence of decreased graft survival and increased risk of graft failure have been demonstrated in patients receiving anticonvulsant enzyme inducers compared with control transplant recipients receiving no such agents (81).
Cyclosporine.
Phenobarbital therapy has the potential to produce clinically significant changes in cyclosporine pharmacokinetics. Subtherapeutic cyclosporine concentrations were observed in a pediatric transplant recipient who was receiving concurrent phenobarbital treatment, but these concentrations increased as the dose of phenobarbital was reduced (82). Similarly, a 70% reduction in the clearance of cyclosporine (from 12.6 to 3.8 mL/min/kg) was observed when phenobarbital was discontinued from the regimen of a pediatric renal transplant recipient (83).
Metronidazole.
Several authors report failure of metronidazole treatment in cases of vaginal trichomoniasis and giardiasis when phenobarbital is prescribed concurrently. Consequently, an increase in dose is necessary to effect a microbiologic cure (84,85). In a crossover study of six patients with chronic disease, metronidazole AUC and half-life were significantly decreased by 30% and 23%, respectively. No significant difference in apparent volume of distribution was observed, although the AUC of the hydroxy metabolite was increased by 29% (86).
Oral Contraceptives.
The potential for decreased efficacy of oral contraceptive therapy secondary to the use of enzyme-inducing anticonvulsants has been recognized since 1980 (87,88). This interaction has frequently been attributed to enhanced metabolism of both estrogenic and progestin components, but alterations in protein binding, specifically serum hormone binding globulin, also appear to be involved. In a prospective evaluation of plasma ethynyl estradiol concentrations in the presence and absence of phenobarbital, two of four subjects demonstrated 64% and 72% reductions from baseline values. However, the overall change for all subjects was not statistically significant, given only a moderate decrease in the third subject and an increase in ethynyl estradiol concentrations in the fourth. In contrast, serum hormone binding globulin capacity increased 15% to 49% in all subjects after the administration of phenobarbital. Of clinical interest, breakthrough bleeding developed in the women who demonstrated reductions in plasma ethynyl estradiol concentrations (87). Thus, a pharmacodynamic interaction can be expected with the concurrent administration of oral contraceptives and phenobarbital; however, whether this is a result of alterations in metabolism or a consequence of altered protein binding remains to be elucidated.
Teniposide.
Long-term treatment with CYP-inducing antiepileptic agents, including phenobarbital, increases the systemic clearance of the epipodophyllotoxin chemotherapeutic agent teniposide in pediatric patients with leukemia. Compared with control patients matched for age at diagnosis, sex, and race but not receiving antiepileptic drug therapy, teniposide clearance was two- to threefold higher in the treated patients (89). The dramatic reduction in systemic teniposide exposure has been associated with reduced efficacy as measured by significantly worse event-free survival, hematologic relapse, and CNS relapse with hazard ratios ranging from 2.67 to 3.4 (90). The mechanism of this interaction likely involves induction of CYP3A4 (possibly CYP3A5 as well), the CYP primarily responsible for O-demethylation of teniposide and etoposide to form their respective catechol metabolites (91). Thus, a similar interaction between phenobarbital, phenytoin, or carbamazepine and etoposide may also be anticipated.
Theophylline.
Although theophylline disposition is generally considered to be the result of CYP1A2-mediated biotransformation (92,93), evidence suggests that phenobarbital increases its clearance in older children and adults. In one crossover study with six subjects, theophylline clearance increased 34% after phenobarbital coadministration (94). In seven pediatric patients with asthma, theophylline clearance increased by 42%, and average steady-state concentrations decreased by 30% (95). In contrast, no change in the clearance and dose requirements of aminophylline were observed in premature neonates receiving the agent alone or in combination with phenobarbital (96).
Warfarin.
In three subjects administered R-warfarin or S-warfarin on separate occasions before and concurrently with phenobarbital, the clearance of each enantiomer increased by 65% and 50%, respectively, and was accompanied by approximately 40% decreases in half-lives (97). The increased warfarin clearance has also been observed to result in an approximately 25% decrease in prothrombin time over
P.510
a 3-week period (98). CYP2C9 appears to be the principal form of human hepatic CYP modulating levels of the pharmacologically more active S-enantiomer with a minor contribution from CYP3A4, whereas CYP3A4 and CYP1A2 are primary determinants of R-warfarin biotransformation (99). The lack of stereospecificity suggests that multiple CYP pathways are induced, whereas the decreased pharmacologic effects are consistent with CYP2C9 induction. Regardless of CYP isoforms induced, studies with phenobarbital and other barbiturates indicate that induction may persist for 3 or 4 weeks after drug discontinuation (100).
Interactions between Phenobarbital and Nutrients
Folic Acid.
The role of reduced folic acid levels and adverse fetal outcomes, such as neural tube defects, is well documented (101,102). Mean serum folate concentrations were found to be significantly lower in patients treated with phenobarbital (3.91 ± 1.73 ng/mL, p < .01) and carbamazepine (3.85±1.02 ng/mL, p < .01) compared with age-matched controls not treated with antiepileptic agents (5.14±1.88 ng/mL). Neither valproic acid nor zonisamide was associated with folate depletion (103). In addition to maintaining pregnant patients seizure free with the lowest possible dose of antiepileptic agents, folic acid supplementation (5 mg/day) should be implemented 3 months before conception (101,102).
SUMMARY
As a general rule, phenobarbital has the potential to increase the clearance of any drug that is primarily dependent on CYP3A4 activity (and possibly CYP2C9 activity) for most of its elimination from the body. However, there are also cases in which these anticonvulsants have been reported to induce the metabolism of drugs primarily metabolized by CYP isoforms other than CYP3A4. For example, phenobarbital has been reported to increase desipramine clearance by 30% and to produce a comparable increase in 2-hydroxydesipramine formation, a CYP2D6-dependent metabolite (104). However, no convincing evidence of CYP2D6 induction by “enzyme-inducing” anticonvulsants is available that would allow the “induction” to be attributed to CYP2D6. Therefore, one must recognize that each individual patient has his or her own unique complement of CYP isoforms expressed in their liver and other tissues, and the potential consequences of induction in that patient will depend on the CYP isoforms that are quantitatively most important for the disposition of the drug in question. Recognition that additional factors such as the therapeutic index of the drug in question, the presence and contribution of numerous competing drug biotransformation pathways, and the pharmacologic or toxicologic potential of the metabolites produced also contribute to the overall clinical significance of a drug-drug interaction will help to minimize unexpected adverse responses in patients when enzyme-inducing anticonvulsants such as phenobarbital are added to existing treatment regimens.
REFERENCES
P.511
27a. Chang TKH, Yu L, Maurel O, Waxman DJ. Enhanced cyclophosphamide and ifosfamide activation in primary human hepatocyte cultures: response to cytochrome P-450 inducers and autoinduction by oxazaphosphorines. Cancer Res 1997; 57:1946-1954.
P.512
P.513