Richard D. Scheyer MD
Director, Clinical Discovery and Human Pharmacology, Aventis Pharmaceuticals, Bridgewater, New Jersey
Valproic acid (VPA) frequently interacts with other drugs, especially other antiepileptic drugs (AEDs). The extent of such interactions is often sufficient to alter the pharmacokinetics of both VPA and the interacting drug in clinically significant ways. Endogenous substances and disease states also may alter VPA pharmacokinetics.
MECHANISMS
An understanding of the mechanisms of drug-drug interactions is gradually replacing our reliance on a laundry list of interactions to an understanding of mechanistically based principles.
Metabolism
VPA is metabolized through multiple metabolic pathways, including cytochrome P450 (CYP) CYP2A6, CYP2C9, and CYP2C19. In addition, glucuronidation by uridine diphosphate glucoronosyltransferase (UGT) plays a major role in VPA metabolism. Details may be found in Chapter 84.
Metabolism of VPA is subject to induction by drugs that induce these pathways, including many AEDs. As described later, this enzyme induction may take days or weeks to manifest itself. VPA causes little or no induction of the metabolism of itself or of other drugs (1, 2, 3).
VPA may also affect the metabolism of other compounds. Although effects may be seen on CYP-mediated reactions, the most prominent effects have been noted on compounds whose metabolism is through UGT-mediated glucuronidation or through epoxide hydrolase. These inhibitory effects are typically faster than enzyme induction, but they may take time to manifest, especially if the half-life of the affected drug (e.g., phenobarbital) is quite long.
Protein Binding
VPA is highly protein bound, and thus small changes in protein binding may cause large changes in unbound VPA fraction. Binding is subject to displacement not only by other drugs, but also by endogenous substances or even by VPA itself.
Transport
VPA exists in the body primarily ionized as valproate. In the ionized form, its lipophilicity is relatively low, and its ability to enter the brain by diffusion across the blood-brain barrier is limited. To enter the brain, VPA depends, at least in part, on transport processes. These transporters may be inhibited by other drugs or by endogenous substances.
SPECIFIC DRUG INTERACTIONS
Changes in Valproic Acid Pharmacokinetics Induced by Other Drugs
Enzyme-Inducing Antiepileptic Drugs.
Carbamazepine, phenytoin, and phenobarbital are capable of inducing VPA clearance. Coulthard noted increasing serum levels of VPA and improved seizure control in one patient after discontinuation of phenytoin (4). When administered with VPA, carbamazepine, phenobarbital, and phenytoin lower VPA concentrations (5, 6, 7, 8). Very large VPA doses may be needed to achieve therapeutic levels and efficacy with concomitant use of enzyme-inducting drugs.
The timing of induction or deinduction of VPA metabolism has received little study. Miller showed evidence of deinduction beginning within 2 days of carbamazepine discontinuation (9). Mattson and coworkers found the rise in VPA levels to begin in some patients while they were still receiving 100 to 200 mg phenytoin. In others, no change occurred for 1 to 2 weeks after phenytoin was discontinued (10,11). Battino found that when carbamazepine was replaced by oxcarbazepine, an antiepileptic prodrug with less propensity for drug interactions, unbound VPA concentrations rose, followed by a rise in total VPA concentrations
P.802
(12). Maximal deinduction was not apparent for ≥2 weeks after completion of the crossover.
As mentioned earlier, VPA binding is saturable. Hence a reduction in VPA concentration by enzyme-inducing drugs can result in a decreased unbound fraction of VPA and an even greater decrease in unbound VPA concentration. There is no evidence of alteration in the number of VPA binding sites or in the binding dissociation constant in patients receiving concomitant AEDs (13).
The coadministration of enzyme-inducing drugs not only increases the clearance of VPA but also may change metabolic pathways (14). Increases in the putative hepatotoxic 4-en-VPA and 2-4-en-VPA may be one reason for greater incidence of VPA hepatotoxicity in patients treated with polytherapy (15, 16, 17, 18).
Ethosuximide.
Ethosuximide may lower VPA levels, but the effect is modest, with a decrease in VPA concentration of ~28% when ethosuximide was added to VPA in two children (19).
Felbamate.
The addition of felbamate (2,400 mg/day) increases total VPA area under the curve by approximately 50% (20). A reduction in VPA dosage may be necessary.
Lamotrigine.
Lamotrigine may cause a modest increase in VPA clearance. Anderson and coworkers reported a 25% decrease in VPA concentration with the addition of lamotrigine (21).
Tiagabine.
The addition of tiagabine results in a 10% to 12% decrease in VPA area under the curve. This is unlikely to be clinically significant (22).
Topiramate.
Topiramate may modestly reduce VPA levels. This effect is also unlikely to be clinically significant (23).
Investigational Antiepileptic Drugs.
Stiripentol may inhibit ω-oxidation of VPA and may decrease the formation clearance of 4-en-VPA, a hepatotoxic metabolite of VPA (24).
Nonantiepileptic Drugs.
Salicylic acid displaces VPA from protein and results in higher unbound VPA levels (25). Schobben (26) noted a significant rise in urinary excretion of VPA after administration of 1 to 2 g of aspirin daily. Aspirin also competes with VPA for mitochondrial oxidation. This leads to an increase in microsomal metabolism with production of 4-en-VPA (27). With the exception of aspirin, changes in VPA protein binding caused by exogenous displacers are relatively small.
The enzyme-inducing antibiotic rifampin increases VPA clearance by 40% (28). This effect may be expected to be clinically significant. Conversely, isoniazid may increase VPA concentration, with resulting clinical VPA toxicity (29). Probenecid increases systemic and especially central nervous system concentrations by blockade of the organic anion transporter in rats (30); the effect in humans is unknown. Case reports suggest that fluoxetine may significantly increase VPA concentrations, possibly through inhibition of CYP2C19 (31, 32, 33).
Valproic Acid-Induced Changes in the Pharmacokinetics of Other Drugs
Benzodiazepines.
The combined use of VPA with benzodiazepines is not unusual. VPA inhibits the clearance of lorazepam, a drug eliminated primarily through glucuronide conjugation (34). A total daily dose of VPA of 1,000 mg raised lorazepam area under the curve 20% in healthy male volunteers but did not result in increased sedation (35). VPA displaces diazepam from plasma protein binding sites in a manner similar to its effects on phenytoin, as detailed later (36). With these exceptions, there is little evidence of pharmacokinetic interactions of benzodiazepines with VPA.
Carbamazepine.
Some patients develop sedation, nausea, diplopia, or a confusional state when VPA is added to carbamazepine therapy (37, 38, 39, 40). These case reports suggest an interaction, but has been proven with blood level changes. The considerable variation in carbamazepine levels between doses makes analysis difficult unless determinations are consistently made both before drug administration and at times of clinical toxicity.
Inhibition of clearance of carbamazepine-10,11-epoxide may contribute to toxicity (41, 42, 43, 44). However, direct administration of carbamazepine epoxide to patients with seizures, with drug levels in excess of those usually achieved with the combination of VPA and carbamazepine, did not produce significant neurotoxicity (45).
When VPA is added in vitro to plasma samples containing carbamazepine, the pharmacologically active unbound fraction of carbamazepine increases 25% (46). Because carbamazepine is only moderately bound, with a baseline unbound fraction of about 25%, this binding interaction is unlikely to be of major clinical significance (47, 48, 49, 50, 51). Carbamazepine toxicity tends to occur at times of peak concentration. Ideally, samples of total and unbound carbamazepine and epoxide should be obtained both at trough and at times of clinical toxicity (52). A more marked inhibition of carbamazepine epoxide clearance is seen with coadministration of valpromide, a prodrug of VPA (53).
Ethosuximide.
Ethosuximide levels may increase as much as 53% when VPA is added, but there appears to be substantial interindividual variability (54,55).
Felbamate.
Felbamate concentrations are increased in patients receiving VPA compared with those receiving
P.803
enzyme-inducing AEDs. This effect appears to be the result of a felbamate-mediated inhibition of VPA clearance as well as the effect of the val of enzyme-inducing AEDS (20).
Lamotrigine.
VPA blocks the glucuronidation of lamotrigine and results in increased concentrations of lamotrigine (56). This effect appears to be maximal even at subclinical concentrations of VPA (57,58). Perhaps as a result of the more rapid rise in lamotrigine concentrations in patients receiving VPA, or by increased formation of reactive metabolites (e.g., arene oxides) through other pathways (59), patients receiving VPA are at greater risk of lamotrigine-related rashes and possibly hepatic toxicity (60).
Phenobarbital.
Serum phenobarbital levels rise when VPA is added (61,62). Loiseau noted an elevation of serum phenobarbital levels and an increase in phenobarbital half-life from a mean of 83 to 105 hours after administering VPA. At the same time, urinary phenobarbital excretion is unchanged or increases, but urinary hydroxyphenobarbital decreases (62,63).
High concentrations of VPA may compete with phenobarbital for microsomal oxidation. At low to moderate levels of VPA, the major metabolite is 3 OXO-VPA created by oxidation in the mitochondria. Higher levels of VPA saturate this pathway and lead to microsomal smooth endoplasmic reticulum production of metabolites 4-OH-VPA, 5-OH VPA, and conjugates to D-glucuronide (14). These latter actions compete with phenobarbital metabolism.
Other Barbiturates.
Primidone concentrations, as well as the derived phenobarbital concentrations, increase when VPA is added. The magnitude is unclear, ranging from a 17% to a >100% increase in primidone (37,64).
Phenytoin.
Most reports note an initial fall in phenytoin levels after initiation of VPA. VPA is highly bound to plasma albumin and displaces phenytoin from binding sites (1). The acute displacement of phenytoin from plasma proteins produces an increase in the concentration of unbound phenytoin. Redistribution of unbound drug to tissues (including brain) may lead to the paradox of neurotoxicity despite a lower total serum phenytoin concentration (65,66).
As the liver removes the increased unbound phenytoin, unbound levels decrease to their initial levels, with a corresponding further decrease in total levels (2,67). Changes in phenytoin dosage to bring total levels back to the usual therapeutic range may raise unbound phenytoin to toxic levels. Even in the absence of dose adjustment, total phenytoin concentrations may return to pre-VPA baseline, and unbound phenytoin levels exceed baseline (68,69). This may be caused by a noncompetitive metabolic inhibition of phenytoin metabolism (70,71), possibly resulting from inhibition of CYP2C9 or CYP2C19.
Interdose fluctuations in VPA concentration may cause fluctuation in phenytoin protein binding and transient phenytoin toxicity. This effect may be estimated using one of several regression equations that permit calculation of unbound phenytoin concentrations from measured total phenytoin and VPA levels. These equations include those described by Haidukewych—% unbound phenytoin = 9.5 + .1 × VPA (µg/mL)—and Scheyer—% unbound phenytoin = 8.0 + .07 × VPA (µg/mL) (49,66). These equations were based on studies in which unbound phenytoin concentrations were determined after ultrafiltration at 20° to 25°C. The unbound fraction of phenytoin would be expected to be greater at 37°C (72).
Investigational Antiepileptic Drugs.
Unlike carbamazepine and phenytoin, VPA does not appear to decrease concentrations of the investigational drug remacemide (73). The extensive glucuronidation of retigabine suggests a potential for VPA to inhibit its metabolism (74).
Nonantiepileptic Drugs.
VPA may be a weak displacer of warfarin binding (75). VPA has been reported to inhibit the clearance of the antiviral drug zidovudine, presumably by inhibition of glucuronidation (76).
Unlike enzyme-inducing AEDs such as carbamazepine, VPA does not appear to decrease concentration of the antipsychotic drug haloperidol (77), and it has no clinically significant effects on clozapine concentration (78). Concentrations of amitriptyline (and its active metabolite nortriptyline) increase by about one-third with the addition of VPA 1,000 mg/day (79).
Unlike enzyme-inducing AEDs, VPA does not decrease cyclosporine concentration, a property that may be useful in patients with seizures after organ transplantation (80). Likewise, VPA does not increase clearance of female sex hormones used in oral contraceptives. Consequently, birth control failures are less likely in women taking VPA (81).
Antiepileptic Drugs Believed Not to Interact with Valproic Acid.
VPA has been reported not to interact with the renally eliminated drugs vigabatrin (82) or gabapentin (83). VPA also does not appear to have a significant interaction with clobazam (84) or zonisamide (85). In vitro data suggest that there is no significant interaction with levetiracetam (86). As noted earlier, oxcarbazepine has less propensity to interact with VPA than does carbamazepine.
INTERACTIONS WITH ENDOGENOUS COMPOUNDS
Endogenous free fatty acids can displace VPA binding (87). Because free fatty acids are acutely affected by food intake, this may confound assessments of unbound VPA concentration (13). Free fatty acids may also compete with VPA
P.804
for transport through the medium-chain fatty acid transporter across the blood-brain barrier (88). The clinical relevance of this interaction is unknown.
|
TABLE 85.1. EFFECTS OF OTHER DRUGS ON VALPROIC ACID |
||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||
Decreased binding is found in uremia and can only partly be attributed to decreased protein concentration. Allosteric changes in plasma protein may explain increased unbound fraction (71). Endogenous “toxic” uremic displacers also may be responsible (89). The result is a lower total serum VPA concentration and an increase percentage of unbound VPA with no change in the concentration of unbound drug.
PHARMACODYNAMIC INTERACTIONS
Few data indicate a supraadditive (synergistic) or infraadditive (antagonistic) pharmacodynamic interaction when VPA is coadministered with other drugs (10) (Tables 85.1 and 85.2). A combination of drugs, such as phenytoin or carbamazepine with VPA, may provide additive efficacy with infraadditive toxicity resulting from different doselimiting neurotoxicity and may yield an improved therapeutic index for the combination. This has not been established in clinical trials. Reports have described instances of profound sedation with addition of VPA to a regimen of phenobarbital in excess of what may be expected from the pharmacokinetic interaction and the additive effect of their sedative properties (90). Some cases may be explained by increased ammonia concentration caused by hepatic dysfunction (91).
|
TABLE 85.2. EFFECTS OF VALPROIC ACID ON OTHER DRUGS: INTERACTIONS WITH PROBABLY CLINICAL RELEVANCE |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Jeavons and Clark reported absence status in five of 12 patients who were given the two drugs (92). The observation is unexplained, and it appears to be uncommon.
CONCLUSIONS AND CLINICAL IMPLICATIONS
Pharmacokinetic interactions between VPA and other drugs are frequent. Many AEDs induce hepatic enzymatic activity to accelerate VPA metabolism. This results in lower plasma VPA concentrations and a shorter VPA half-life. The effect of metabolic interactions may be underestimated when total VPA concentrations are measured, because the saturable nature of VPA binding means that modest changes in total VPA concentration may be associated with greater changes in unbound VPA concentration, and the short half-life makes a consistent sampling schedule critical. Newer AEDs may be associated with fewer interactions, but any advantages of combining these agents with VPA remains to be demonstrated clinically.
P.805
VPA is metabolized almost completely in the liver and can interfere with the biotransformation of other AEDs. VPA slows hydroxylation and glucuronidation of some drugs and thus causes a rise of serum levels of, for example, carbamazepine epoxide, phenobarbital, and lamotrigine. A similar effect on phenytoin metabolism may be masked by changes in protein binding that produce opposite effects on the usually measured total concentration. These interactions may lead to toxicity and may require adjustment of dosage.
The frequency of VPA interactions calls for clinical awareness and appropriate determinations of blood levels of both VPA and other drugs, at times including unbound levels. Although the conversion to VPA monotherapy requires special effort, the benefits of avoiding interactions with other drugs are worthwhile (93).
ACKNOWLEDGMENT
This work was supported in part by the United States Department of Veterans Affairs.
REFERENCES
P.806
P.807