René H. Levy PhD*
Danny D. Shen PhD**
Frank S. Abbott BSP, MS, PhD***
Wayne K. Riggs PhD****
Houda Hachad MD*****
* Professor and Chair, Department of Pharmaceutics, Professor of Neurological Surgery, University of Washington School of Pharmacy and Medicine, Seattle, Washington
** Professor, Department of Pharmacy and Pharmaceutics, University of Washington, Seattle, Washington
*** Professor and Dean, Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver, British Columbia, Canada
**** Associate Professor of Pharmaceutical Sciences, University of British Columbia, Vancouver, British Columbia, Canada
***** Research Associate, Department of Pharmaceutics, University of Washington, Seattle, Washington
The branched-chain fatty acid structure of valproic acid (VPA) explains some of its disposition characteristics. VPA is highly bound to plasma albumin, and this property tends to keep most of the drug within the vascular compartment. In common with endogenous fatty acids, VPA undergoes phase I biotransformation mediated by both β-oxidative enzymes in the mitochondria and cytochrome P450 enzymes in the smooth endoplasmic reticulum. Early on, it was assumed that VPA, being lipophilic like the endogenous long-chain fatty acids, readily permeates the blood-brain barrier. However, more recent studies suggest that specialized transport mechanisms may be involved in the distribution of VPA into the brain, as well as its transfer across the placenta to reach the fetus. These unique features of VPA disposition represent an important aspect of the clinical pharmacology of this widely used and versatile antiepileptic and mood-stabilizing agent.
CHEMISTRY AND METABOLIC SCHEME
VPA (2-propylpentanoic acid, dipropylacetic acid) (1) (Figures 84.1 and 84.2) is an achiral C-8 branched-chain fatty acid having a molecular weight of 144.2 g mol-1. The pure acid, a colorless liquid (boiling point, 221 to 222°C) with a characteristic odor, is only slightly soluble in water but is highly soluble in organic solvents (log p = 2.72 - 2.75) (1,2). The white crystalline sodium salt is very soluble in water and in some organic solvents (e.g., methanol and acetone), whereas the calcium and magnesium salts are insoluble in water. A common therapeutic form of the drug is divalproex sodium (Depakote, Epival), a stable coordination compound derived from sodium VPA and VPA in a 1:1 molar ratio. The free acid (negative log of dissociation constant = 4.56 - 4.8) (2) and sodium salt forms are also stable compounds. Further details of physical constants and spectroscopic details of VPA can be found in the monograph by Chang (3). Key references on assays of VPA and its metabolites can be found in the previous edition of this book (4) and in a book on VPA (5). Of the chemical forms of VPA, the amide analog valpromide is an active anticonvulsant and is readily converted in vivo to VPA. As such, it can be considered to be a prodrug form of VPA, although it has effects of its own, such as being a selective inhibitor of human microsomal epoxide hydrolase (6). Other amides and chemical analogs of VPA including metabolites of VPA have been studied for anticonvulsant activity in the search for an alternative drug to VPA. Much of that work has been reviewed (5).
The gastrointestinal absorption of VPA from all its oral formulations appears to be almost complete. The absolute bioavailability of a divalproex sodium extended-release
tablet administered as a single dose after a meal was ~90% (product information, Depakote ER, 2000). Several bioequivalence studies have shown comparable bioavailability (>90%) for all dose formulations (7,8). Controlled-released formulations of VPA given once daily were shown to be bioequivalent to enteric-coated formulations given twice daily, with respect to area under the curve (AUC) (9). The sprinkle formulation of sodium hydrogen divalproex (coated particles in capsules) exhibited the same extent of absorption as enteric-coated tablets (10).
FIGURE 84.1. Valproic acid (1) and metabolites involved in phase I metabolic pathways observed in humans: 2, 2-ene-VPA (E- and Z-isomers); 3, 3-OH-VPA; 4, 3-keto-VPA; 5, 3-ene-VPA (E- and Z-isomers); 6, 2,3′-diene-VPA (E,E- and E,Z-isomers); 7, 4-OH-VPA; 8, 4-keto-VPA; 9, 2-PSA; 10, 5-OH-VPA; 11, 2-PGA; 12, 4-ene-VPA; 13, 2,4-diene-VPA (E- and Z-isomers). The putative characterized enzymatic pathways are as follows: (a) β-oxidation; (b) P450-dependent desaturation; (c) P450-dependent ω-hydroxylation; (d) P450-dependent (ω-1)-hydroxylation; (e) P450-dependent (ω-2)-hydroxylation. The broken lines indicate a metabolic route in which the details are not yet confirmed.
FIGURE 84.2. Valproate (1) and phase I metabolites (12, 13) involved in phase II metabolic pathways: 12, 4-ene-VPA; 13, 2,4-diene-VPA; 14, 3-keto-4-ene-VPA; 15, 4,5-epoxy-VPA; 16, 5-GS-4-OH-VPA-γ-lactone; 17, 4-GS-5-OH-VPA; 18, 5-NAC-4-OH-VPA-γ-lactone; 19, 5-GS-2-ene-VPA; 20, 5-GS-3-ene-VPA; 21, 5-NAC-2-ene-VPA; 22, 5-NAC-3-ene-VPA;23, 5-GS-3-keto-VPA; 24, 5-NAC-3-keto-VPA; 25, VPA glucuronide. The putative enzymatic pathways are as follows: (a) glucuronidation; (b) glutathione conjugation; (c)mercapturic acid pathway. Postulated intermediate compounds are shown in square brackets.
The rate of absorption of VPA depends on the dosage form; the rapid-release formulations (syrup, capsule, and uncoated tablet) are absorbed with peak times (Tmax) between 0.5 and 2 hours. Tmax increases in the following order: dragee > tablet > solution (11). Enteric-coated tablets exhibit variable absorption rates (12, 13, 14, 15, 16), and peak times range between 3 and 8 hours. Compared with enteric-coated tablets, sprinkle capsules have a slightly slower absorption rate, with a Tmax value of 4.0 versus 3.4 hours (10).
The AUC and maximum concentration resulting from intravenous administration of VPA (500 mg as a single 1-hour infusion) and a single oral (500-mg) dose of VPA syrup to healthy volunteers were equivalent. Tmax occurred at the end of the infusion, whereas Tmax after oral administration
in this study occurred at -4 hours (product information, Depacon, 2000).
Food intake had no significant effect on the bioavailability of VPA in soft gelatin capsules (17) or in a sustained-release preparation (18). However, multiple-dose administration of enteric-coated VPA tablets with meals in patients was associated with a delay in absorption; Tmax was increased from 2.0 to 5.8 hours. Meals had no effect on the extent of absorption (19).
The plasma VPA concentration time course after rectal administration of VPA syrup is comparable to that observed after the oral capsule (20). The absorption of VPA from suppositories is generally slower than that from rectal solutions (21, 22, 23, 24, 25).
Three different chemical forms of VPA are most commonly used: the free acid, the sodium salt, and divalproex sodium, which is a coordinated 1:1 complex of sodium VPA and VPA. VPA is available as a capsule (Deproic) or as soft gelatin capsules (Depakene) of 250 mg. It is also available as a syrup containing the equivalent of 250 mg/5 mL as the sodium salt.
An intravenous formulation of VPA sodium is available in 5-mL single-dose vials available in trays of 10 vials (Depacon). Each milliliter contains VPA sodium equivalent to 100 mg VPA.
Divalproex sodium tablets are supplied in three dosage strengths containing divalproex sodium equivalent to 125 mg, 250 mg, or 500 mg of VPA. It is also available as coated particles in capsules (Depakote Sprinkle capsules), delayed-release tablets, or extended-release tablets. The Springle capsule form contains specially coated particles of divalproex sodium equivalent to 125 mg of VPA in a hard gelatin capsule. Depakote delayed-release tablets are supplied in three dosage strengths containing divalproex sodium equivalent to 125 mg, 250 mg, or 500 mg of VPA. Depakote extended-release tablets contain divalproex sodium in a once-a-day extended release formulation equivalent to 500 mg of VPA.
Routes of Administration
The rapid-release, enteric-coated, and slow-release formulations of VPA are administered orally with food to minimize gastrointestinal irritation.
The intravenous formulation of VPA should be administered as 60-minute infusion at a rate not exceeding 20 mg/min. Because the use of VPA sodium injection for more than 14 days has not been studied, patients should be switched to oral formulations as soon as clinically feasible.
Rectal administration of VPA has been successful in the treatment of intractable status epilepticus in children (26). Commercially available VPA syrup (250 mg/5 mL) was diluted 1:1 with tap water and was given as a retention enema in a loading dose of 10 to 20 mg/kg. Maintenance doses (10 to 15 mg/kg every 8 hours) were started 8 hours after the initial loading dose.
Plasma Protein Binding
VPA has a rather small distribution volume (slightly larger than extracellular fluid volume) that reflects its high binding affinity for plasma proteins relative to its binding or sequestration at extravascular sites. VPA binds mainly to albumin in human plasma (27). Early studies established that ~90% of VPA in plasma is bound to albumin at therapeutic concentrations (28).
A significant feature of VPA binding to plasma albumin is the dependence of free fraction on drug concentration. Detailed studies on VPA binding in plasma obtained from patients with epilepsy and receiving long-term VPA therapy have shown that the equilibrium dissociation constant of VPA for albumin is in the range of 90 to 200 µmol/L or 13 to 29 µg/mL (29,30). Because the therapeutic concentrations of free VPA in plasma (i.e., 5 to 15 µg/mL) are close to the dissociation constant for the VPA-albumin complex, the serum free fraction of VPA is expected to vary with total drug concentration. Cramer et al. (31) reported that the average plasma unbound fraction in adult patients receiving VPA monotherapy ranged between 7% and 9% at total concentrations <75 µg/mL; it increased to 15% at 100 µg/mL, 22% at 125 µg/mL, and 30% at 150 µg/mL. Similar concentration-dependent changes in the plasma free fraction of VPA have also been observed in children and infants with seizure disorders (32, 33, 34, 35).
Concentration dependence in the plasma free fraction of VPA leads to differing degrees of fluctuation between total and unbound VPA concentrations at steady state. Numerous studies have shown that fluctuations in unbound concentration were generally 50% to 100% greater than fluctuations in total concentrations (35, 36, 37, 38, 39, 40). Aside from concentration dependence in plasma protein binding, elevated free fatty acid levels in early morning samples and true diurnal differences in metabolic clearance can also explain the apparent difference in total and free VPA kinetics.
Diminished plasma protein binding of VPA has been observed in numerous pathophysiologic states, notably those associated with hypoalbuminemia, for example, pregnancy (41,42), aging (37,43), head trauma, liver diseases (44), uremia (45,46), and advanced disease with human immunodeficiency virus (47). In uremia, the increase in free fraction is the result, at least in part, of displacement from protein binding sites by endogenous compounds. The concentration dependence in plasma free fraction of VPA becomes more pronounced at low serum albumin (48). The
effects of lowering of plasma protein binding on VPA clearance kinetics are discussed later, in the section on elimination and half-life.
As expected, other drugs bound to albumin can competitively reduce the plasma protein binding of VPA. Such is the case for sulfamethoxazole (47), salicylates (49), and naproxen (50). A similar binding displacement is observed when plasma endogenous fatty acid levels are raised (51,52). An elevation in free fatty acid level also explains the decrease in VPA binding to plasma proteins in the sera of patients with insulin-dependent diabetes mellitus (53).
The merits of monitoring free, rather than total, VPA concentrations, an approach that assumes that only free drug is available for distribution into the brain, have been debated in the literature with no clear resolution (42,54, 55, 56). Lenn and Robertson (57) reported the results of a retrospective analysis of clinical records and unbound and total plasma concentrations of VPA that were collected from 395 epileptic patients over a 13-month period. Unbound VPA concentration was considered clinically meaningful when (a) an unbound fraction fell outside the normal range of 5% to 15%, (b) unbound and total concentrations appeared discordant (not both high or both low), and (c) current seizures or side effects such that a change in antiepileptic drug regimen was indicated. About 15% of unbound values from 18% of patients fit clinically significant criteria. Most of these patients had unsatisfactory control of current seizures. The authors advocated monitoring of plasma free VPA regardless of total plasma VPA concentration when there is an unresolved clinical problem. Anecdotal reports attest to the usefulness of unbound VPA levels in diagnosing side effect problems in patients with hypoalbuminemia (58,59), and during removal of an enzyme-inducing anticonvulsant from VPA polytherapy when monitoring total VPA concentration may lead to underestimate the rise in free VPA concentration, because of concentration-dependent change in plasma free fraction (60).
Central Nervous System
VPA enters the central nervous system very rapidly. Tissue distribution studies in mice and rats (61,62) showed that peak concentration in brain was reached within minutes after either intravenous or intraperitoneal injection of VPA. The subsequent decline of drug concentration in brain paralleled that in plasma, a finding indicating a facile equilibration of VPA between brain and capillary blood. In rhesus monkeys equipped with a chronically implanted ventricular catheter, the upswing and decline of VPA in cerebrospinal fluid (CSF) followed closely the time course of plasma concentration during and after the cessation of intravenous VPA infusion (63). A reasonably rapid penetration of VPA into brain also occurs in humans, as evidenced by the effective treatment of status epilepticus, acute manic episodes, and severe refractory migraine with rectal or intravenous VPA (26,64, 65, 66, 67, 68).
Much of the early information on the extent of VPA distribution into the central nervous system of humans was obtained indirectly through CSF sampling studies in patients with epilepsy (69). In general, a good correlation was observed between lumbar CSF and total plasma concentration of VPA. The CSF:total plasma concentration ratio averaged about 0.1 to 0.15, with notable variation among subjects within any given study. VPA concentrations in CSF tended to be lower than unbound VPA concentrations in plasma. The reported CSF:free plasma concentration ratio varied from 0.6 to near 1.
There are three available reports of VPA concentrations in the human brain (70, 71, 72). All three studies reveal an exceptionally low and variable presence of VPA in the brain. In the most recent study, by Shen et al. (70), cortical (gray matter) samples were obtained in 13 patients receiving long-term VPA therapy who underwent surgical treatment for intractable seizures. The respective mean brain:plasma concentration ratios based on total and free drug in serum were 0.11 and 0.54. This finding stands in contrast to the older aromatic or heterocyclic antiepileptics such as phenytoin and phenobarbital, which typically concentrate in brain tissue and exhibit brain:CSF concentration ratios well exceeding unity. The brain:total serum concentration ratio varied over a fourfold range. Some of this variability was related to interpatient variability in serum protein binding, as indicated by a modest correlation between the distribution ratio and serum free fraction (r2 = 0.47, p < .01). However, the brain-to-unbound concentration ratio still showed a threefold variation. Two conclusions were reached. First, the unusually low distribution ratio of VPA explains the relatively high plasma levels of VPA (usually >350 µmol/L or 50 µg/mL) that are needed for effective seizure control. A brain:unbound concentration ratio less than the brain water content indicates a more rapid removal rate of VPA from the brain than its entry rate into the brain at steady state. Using the tissue-sampling, carotid artery-injection method (i.e., the Oldendorf technique), Cornford et al. (73) showed that the kinetic rate constant for the efflux of VPA from rat brain exceeded the rate constant for uptake into the brain, a finding that confirms the apparent asymmetry in the transport kinetics of VPA between blood and brain. Second, the variability in distribution of VPA between brain and blood may be one underlying factor for the lack of a clearly definable therapeutic range of serum VPA concentrations in epileptic patients. Hence, there is a need to elucidate the mechanisms for the transport of VPA into and out of the brain as well as the physiologic factors that regulate those transport processes.
Animal studies have revealed that the bidirectional movement of VPA across the blood-brain barrier is mediated jointly by passive diffusion and carrier transport. The entry of VPA from blood into the brain is mediated by a medium- and
long-chain fatty acid selective anion exchanger at the brain capillary epithelium, which accounts for two-thirds of the barrier permeability. The mechanism governing the efficient clearance of VPA from the brain into blood appears to involve a probenecid-sensitive, active transport system at the brain capillary endothelium.
A brain microdialysis study in rabbits suggests that another set of transporters exists within the brain parenchyma that shuttles VPA between the interstitial fluid and intracellular compartments. The putative parenchymal cell transport system appears to concentrate VPA within the cellular compartments (i.e., those of neurons and glia). This, coupled with the efficient removal process at the blood-brain barrier, results in a very low interstitial concentration of VPA. During intravenous VPA infusion, the average steady-state concentration in the interstitial fluid was 36% of that in the intracellular compartment and 17% of that in plasma water. The high intracellular:interstitial concentration gradient suggests that intraneuronal mechanisms of VPA may be more important than its actions at the plasma membrane with respect to anticonvulsant and neuropsychiatric effects.
One other important aspect of the blood-brain barrier transport mechanism is the effect of plasma protein binding on the uptake of VPA. The traditional notion that drug molecules bound to plasma proteins and blood cells do not diffuse readily across the capillary endothelium appears to hold true in the blood:brain distribution behavior of certain anticonvulsants. It was previously mentioned that Shen et al. (70) found that the interpatient variation in brain:serum distribution ratio of VPA is in part related to variation in serum free fraction, a finding consistent with limitation of brain uptake by serum protein binding. However, Cornford et al. (73) showed that VPA uptake into rat brain is not entirely restricted to the equilibrium free fraction. These investigators estimated that as much as 18% of VPA that was extracted during a single pass through the capillaries is derived from protein-bound VPA in serum-containing injectate. It was also shown that although brain extraction increased with increasing serum concentration of VPA as a result of saturation in serum drug binding, the extraction of the bound moiety remained constant. The mechanism by which protein-bound drug is released for uptake across the capillary endothelium is not understood.
Aside from the obvious need for studies relating to the brain distribution of VPA, questions have been raised on the central nervous system uptake and accumulation of VPA metabolites (discussed in the section on biotransformation). Some of the mono-ene and di-ene metabolites of VPA have been shown to possess anticonvulsant activity at a dose potency near that of VPA (74,75). Earlier attention had focused on the role of the predominant unsaturated metabolite in circulation -(E)-2-ene-VPA in the pharmacodynamics of VPA. However, the circulating levels of the unsaturated metabolites are typically lower than those of the parent drug (see the later section on biliary and renal excretion). Moreover, studies have shown that, in epileptic patients, the brain cortical and CSF concentrations of these pharmacologically active metabolites are much lower than their plasma concentrations and are low relative to VPA concentrations at the respective sites (76,77). Although sequestration of active metabolites at the critical target sites in the brain has been proposed (78), there is no firm evidence that the unsaturated metabolites play a quantitatively significant role in the pharmacodynamics of VPA.
VPA has been implicated as a teratogen causing neural tube defects, especially spina bifida aperta. Consequently, transplacental transfer and fetal accumulation of VPA in pregnant mouse and rats have been investigated extensively, as reviewed by Nau (79). More recently, maternal-to-fetal distribution of VPA was studied in a chronically catheterized late-term pregnant sheep model (80,81). Collectively, these studies found that VPA crosses the placental barrier readily, resulting in significant fetal exposure. A study by Utoguchi and Audus (82) in a human trophoblast cell line (BeWo) suggested that the transplacental transfer of VPA from the maternal to the fetal side may be mediated by the proton-coupled monocarboxylic acid transporters. Nau and Scott (83) observed higher VPA concentrations in early mouse embryos than the corresponding free VPA concentration in maternal serum. These investigators attributed the accumulation of VPA to “ion trapping” resulting from the remarkably high intracellular pH of the rodent embryo during organogenesis. The same investigative team also reported preferential accumulation of VPA in embryonic neuroepithelium of the mouse during early stages of organogenesis (84).
Information on distribution of VPA into human fetus has been gathered through amniocentesis in pregnant women and by sampling of umbilical and maternal blood at birth. Omtzigt et al. (85) gathered maternal serum and amniotic fluid data from 52 pregnant women during the late first trimester and early second trimester of pregnancy. VPA concentrations in the amniotic fluid correlated better with total than unbound concentrations in maternal serum. The median amniotic fluid:total serum concentration ratio was 0.09. All unsaturated and hydroxylated metabolites of VPA present in the serum were detected in the amniotic fluid, although at very low concentrations.
Some studies have been performed during delivery of newborns from mothers who received VPA throughout pregnancy. VPA concentration is 1.5 to 2 times higher in umbilical cord serum than in maternal serum (86, 87, 88, 89). Several investigators examined the possible contribution of protein binding to this phenomenon and consistently found that the unbound fraction of VPA is lower in umbilical cord
serum than in maternal serum: 9.1% versus 15% in one study (86) and 6% versus 12% in another study (90).
A longitudinal study in fetus-mother pairs throughout pregnancy was performed by Nau and Krauer (41). In fetal serum, unbound fractions were 40% to 80% during weeks 13 through 16 of gestation, they decreased to 20% around week 20, and they continued to decrease to 10% at the end of gestation. This behavior of unbound fraction was explained by the increase in fetal serum albumin concentrations from 3 to 12 g/L during early gestation to 30 to 40 g/L at term. Maternal unbound fractions, conversely, increased during pregnancy from 10% to 20%, a finding that correlates with the decline in maternal serum albumin (91). Thus, whereas the extent of serum protein binding is higher in fetus than in mother at term, the opposite is true during gestation.
Literature reports on excretion of VPA into breast milk and consumption by the suckling infant were summarized in two reviews (92,93). The concentrations of VPA in breast milk were found to range from > 1% to 10% of the maternal serum concentration. Serum concentrations of VPA in the nursing infants were mostly between 4% and 12% of the corresponding maternal serum level. The estimated dose to the suckling infant is <6% of the recommended initial pediatric therapeutic dose of 20 mg/kg/day. Therefore, exposure of newborn to VPA through ingestion breast milk is generally not an issue. Stahl et al. (94) reported an incidence of thrombocytopenia purpura and anemia in a breast-fed infant whose mother was treated with VPA. These authors suggested that significant transfer of VPA metabolites, some of which may be toxic, could be a cause of this adverse event.
ROUTES OF ELIMINATION
Figures 84.1 and 84.2 summarize the metabolic scheme for VPA. Figure 84.1 is not inclusive but for the most part describes the phase I metabolites of VPA most commonly found in human plasma and urine (4,95, 96, 97). Five metabolic pathways of VPA are illustrated, with at least three—β-oxidation, cytochrome P450 (P450)-dependent (ω)- and (ω-1)-hydroxylation—being shared by endogenous fatty acids (98). Complexity in the metabolism of VPA arises because several of the metabolites are formed by more than one pathway. For example, mitochondrial β-oxidation transforms VPA to 2-ene-VPA (2-n-propyl-2-pentenoic acid; 2 in Figure 84.1), 3-OH-VPA (2-n-propyl-3-hydroxypentanoic acid; 3 in Figure 84.1), and 3-keto-VPA (2-n-propyl-3-oxopentanoic acid; 4 in Figure 84.1) (99), whereas P450-dependent (ω-, (ω-1)-, and (ω-2)-hydroxylation produces 5-OH-VPA (2-n-propyl-5-hydroxypentanoic acid; 10 in Figure 84.1), 4-OH-VPA (2-n-propyl-4-hydroxypentanoic acid; 7 in Figure 84.1), and 3-OH-VPA (2-n-propyl-3-hydroxypentanoic acid; 3 in Figure 84.1), respectively (100, 101, 102). Evidence suggests that the 3-OH-VPA detected in serum is largely of P450 origin (103,104). Of the unsaturated metabolites, 2-ene-VPA (2 in Figure 84.1), 3-ene-VPA (2-n-propyl-3-pentenoic acid; 5 in Figure 84.1), 2,3′-diene-VPA (2-[1′-propenyl]-2-pentenoic acid; 6 in Figure 84.1), and 2,4-diene-VPA (2-n-propyl-2,4-pentadienoic acid; 13in Figure 84.1) are products of β-oxidation (103, 104, 105). The 2,4-diene-VPA (13 in Figure 84.1), a metabolite implicated in the hepatotoxicity of VPA (105, 106, 107, 108, 109, 110,111), is a product of the mitochondrial β-oxidation pathway that converts 4-ene-VPA (2-n-propyl-4-pentenoic acid; 12 in Figure 84.1) as the coenzyme A (CoA) ester to the corresponding 2,4-diene-VPA-CoA (105). The 2,4-diene metabolite is also the product of endoplasmic reticulum P450 transformation of the primary serum metabolite 2-ene-VPA (2 inFigure 84.1) (110). The route by which 2,4-diene-VPA (13 in Figure 84.1) is formed is thought to have significance with respect to the hepatotoxicity of VPA (105).
A novel discovery of VPA biotransformation was the finding that terminal desaturation of VPA to 4-ene-VPA (12 in Figure 84.1) is catalyzed by P450 (112,113) and is common to certain animal species including human (113). This reaction is unique to VPA metabolism in showing a high degree of specificity, with the (R)-isomer of 4-ene-VPA (12 in Figure 84.1) being the preferred product (114). Mechanistic studies using stable isotopes and mass spectrometry illustrated that an intermediate carbon-centered radical at C-4 is an essential common rate-limiting step for the formation of both 4-ene-VPA (12 in Figure 84.1) and 4-OH-VPA (7 in Figure 84.1) (113). Initial studies on subcellular fractions demonstrated that rat (112) and rabbit (113) CYP2B (cytochrome P450, CYP) isoforms as well as rabbit lung CYP4B1 significantly catalyze the biotransformation of VPA to 4-ene-VPA (115). More recent experiments with human cDNA-expressed P450 isoforms have demonstrated that multiple human P450 enzymes are involved in the desaturation of VPA. These include CYP2C9 and CYP2A6 (116) as well as CYP2B6 (117). In an elegant experiment using stable isotopes, this type of P450 desaturation was also shown to be true for the formation of 3-ene-VPA (5 inFigure 84.1) (102). In female rat microsomes induced by triacetyloleandomycin and pregnenolone-16α-carbonitrile and in baculovirus expressed P450, CYP3A1 was determined to mediate the direct dehydrogenation of VPA to E (trans) and Z (cis)-3-ene-VPA. Based on the deuterium isotope effects, the metabolite most likely arises from partitioning of the intermediate C-3 or C-4 radicals between olefin formation and hydroxyl radical rebound to form 3-OH-VPA and 4-OH-VPA, respectively. The significance of this pathway to the formation of 3-ene-VPA is uncertain because this route is highly specific to the rat in that neither human CYP3A4 nor rabbit CYP3A6
forms significant quantities of the metabolite. Control microsomes (noninduced) did not produce detectable levels of the metabolite (102). In vivo in rats, 3-ene-VPA is a secondary metabolite of VPA that is formed by isomerization of the primary mitochondrial product 2-ene-VPA (2 in Figure 84.1) (103,104). Because (E,E)-2,3′-diene-VPA (6 in Figure 84.1) is a β-oxidation product of 3-ene VPA (5 in Figure 84.1) (103) and is also a major serum metabolite of VPA in rats and human (4,95), it is likely that most of the 3-ene-VPA seen in the human arises from 2-ene-VPA.
Of the ketone metabolites, 3-keto-VPA (4 in Figure 84.1) is a major metabolite of VPA in most species (4,118) and is derived largely from (E)-2-ene-VPA (2 in Figure 84.1) in the β-oxidation pathway (99,104). Enzymes involved in the β-oxidation of VPA may differ from those reported for endogenous straight chain fatty acids (99,119). For example, in mitochondrial studies, 3-OH-VPA-CoA was not a substrate for mitochondrial or peroxisomal L-3-hydroxyacyl-CoA dehydrogenases but was oxidized to 3-keto-VPA (4 in Figure 84.1) by a novel membrane oxidized nicotinamide-adenine dinucleotide (NAD+)-specific 3-hydroxyacyl-CoA dehydrogenase (99). The resulting 3-keto-VPA-CoA appeared to be resistant to hydrolysis by 3-keto-acyl-CoA thiolase, an enzyme that catalyzes the final step of fatty acid CoA hydrolysis in the β-oxidation pathway (99). The corresponding 4-keto-VPA (2-n-propyl-4-oxopentanoic acid; 8 in Figure 84.1) is formed from 4-OH-VPA (7 in Figure 84.1) and is subsequently oxidized to the dicarboxylic acid 2-PSA (2-n-propylsuccinic acid; 9 in Figure 84.1) (120). The 2-PGA (2-n-propylglutaric acid; 11 in Figure 84.1) metabolite is the primary dicarboxylic acid derived from VPA biotransformation that is found in the urine of patients on VPA therapy (4) and is thought to be an oxidation product of 5-OH-VPA (10 in Figure 84.1) (120).
The phase II metabolism of VPA is described in Figure 84.2. Glucuronidation is the major route of VPA metabolism and results in the formation of the 1-O-acyl-β-D-esterlinked glucuronide in most animals (121) and humans (122, 123, 124). Because of continuing interest in reactive metabolites of VPA that may contribute to the hepatotoxicity and teratogenicity of VPA, the glutathione (GSH) and mercapturic acid (NAC, N-acetylcysteine) conjugates arising from reactive intermediates are also summarized in Figure 84.2. Other acyl conjugates of VPA such as valproyl-Lcarnitine, VPA-CoA, VPA adenosine monophosphate (VPA-AMP) and amino acid conjugates of VPA are known and have been described previously (4).
Conjugation of VPA with glucuronic acid is mediated by hepatic microsomal UDPGT (uridine diphosphate glucuronosyltransferase) enzymes (125), although the specific UDPGT isoforms involved in VPA conjugation have yet to be reported. A high metabolic capacity was observed in this in vitro system for VPA glucuronidation, which, in fact, inhibited the conjugation of other drugs in a competitive fashion (125). Like other acyl glucuronide conjugates, VPA glucuronide (25 in Figure 84.2) is intrinsically reactive and is capable of undergoing numerous reactions including hydrolysis, rearrangement, and covalent binding to proteins (126). Rearrangement of the acyl moiety of VPA glucuronide occurs in a pH-dependent fashion to yield the β-glucuronidase-resistant 2-, 3-, and 4-O positional isomers (127). Although VPA glucuronides may give rise to covalent adducts with proteins (128), these conjugates appear to be only weakly immunogenic in humans (129). One example of a reactive VPA glucuronide comes from liquid chromatography-tandem mass spectrometry (LC/MS/MS) analysis of (E)-2,4-diene-VPA (13 in Figure 84.2) metabolites in rats that indicated the presence of novel diconjugates characterized as 5-GS-3-ene-VPA glucuronide and the corresponding 5-NAC-3-ene-VPA glucuronide (106). This study provided direct evidence for the reactivity of a VPA acyl glucuronide with GSH through a Michael addition mechanism, a result that may partially reflect the large propensity of the rat for the glucuronidation of VPA (121). Conversely, no 5-NAC-3-ene-VPA glucuronide or its 2-ene isomer could be found in the urine of patients receiving VPA therapy (130). Other metabolites of VPA also undergo phase II glucuronidation (4,5).
Numerous investigators have sought evidence for the bioactivation of VPA to reactive metabolites as a mechanism of VPA-induced hepatotoxicity. The detection of GSH conjugates (mostly in bile) and their corresponding mercapturate conjugates in urine is strong evidence of the reactivity of metabolic intermediates in the biotransformation of VPA. Figure 84.2summarizes some of the key findings. Not surprisingly, 4-ene-VPA (12 in Figure 84.2) and (E)-2,4-diene-VPA (13 in Figure 84.2), the two metabolites shown to be hepatotoxic in rats (44), give rise to GSH conjugates 16, 17, 19, 20, 23 seen in rat bile and NAC conjugates 18, 21, 22, and 24 isolated from rat urine (105,106, 109). The NAC conjugates 21 and 22 derived from (E)-2,4-diene-VPA (13 in Figure 84.2) have also been isolated from patient urine (109,130). When 4-ene-VPA (12 in Figure 84.2) was administered to rats, the predominant GSH conjugate was that of 4-OH-VPA-γ-lactone (16 in Figure 84.2), a product of the reactive epoxide intermediate (15 in Figure 84.2) (105). Further β-oxidation of 4-ene-VPA (12 in Figure 84.2) generates the metabolites (E)-2,4-diene-VPA (13 in Figure 84.2) and 3-keto-4-ene-VPA (14 in Figure 84.2) that in mitochondria are present as their CoA thioester derivatives, a necessary form to react with GSH (105,109). In support of this, when 4-ene-VPA was substituted with a fluoro atom at the α position to block β-oxidation, no thiol conjugates from the biotransformation of α-fluoro-4-ene-VPA could be detected in either bile or urine of rats (108). Significantly reduced levels of thiol conjugates in urine were evident when either VPA or (E)-2-ene-VPA was administered to rats with only the NAC conjugate 22 (in Figure 84.2) being detected (105). This finding is
consistent with VPA patient urine samples in which NAC conjugates 21 and 22 (in Figure 84.2) were the only GSH-derived conjugates of VPA to be identified (130,131). The enzyme glutathione-S-transferase is implicated in these conjugation reactions (132), although the specific isozymes to catalyze the reactions involving VPA metabolites have yet to be characterized. The case to link the formation of GSH conjugates of VPA metabolites with the toxicity of the drug can be found in several references (4,105,109,132), but other mechanisms may equally apply.
Plasma concentrations of VPA metabolites exhibit extremely high interindividual variability. The major VPA metabolites in plasma are generally (E)-2-ene-VPA, (E,E)-2,3′-diene-VPA, and 3-keto-VPA (4,95, 96, 97,124), with plasma concentrations of these metabolites usually within the range of 1 to 10 µg/mL, that is, well less than 20% of parent drug. Other metabolites such as 3-ene-VPA, 4-keto-VPA, and the hydroxylated metabolites are present at concentrations of 0.5 to 2 µg/mL (4). The 4-ene-VPA, (E)-2,4-diene-VPA, 2-PSA, and 2-PGA metabolites in plasma are usually present only in trace amounts; however, plasma concentrations of 4-ene-VPA may be elevated in patients taking enzyme-inducing comedication (133).
Genetic Aspects and Isoenzymes
Conjugation of VPA with glucuronic acid is catalyzed by the action of hepatic microsomal UDPGT enzymes (125) that may be responsible for the wide variation in the excretion of VPA glucuronide. Although much of this variability likely arises from the dose-dependent increase in glucuronide formation (124,134), pharmacogenetic factors could also be a contributor because some of these enzymes are polymorphically expressed (135). Specific UDGPT isoforms responsible for VPA conjugation have yet to be identified. Similarly, the specific P450 isozymes involved in phase I metabolism of VPA have not been fully characterized. As described previously, studies using subcellular fractions have shown that rat and rabbit liver CYP2 isoforms (112,113), as well as rabbit lung CYP4B1 (115), are capable of catalyzing the formation of 4-ene-VPA. In contrast, the lauric acid ω-hydroxylases, CYP4A1 and CYP4A3, do not result in 4-ene-VPA formation (115). More recent work using individual human cDNA-expressed isoforms indicate that CYP2C9, CYP2A6, and CYP2B6 (116,117) contribute to the terminal desaturation of VPA, and rat microsomal CYP3A1 catalyzes the oxidative formation of 3-, 4-, and 5-hydroxy-VPA, plus 3-and 4-ene-VPA (102). Based on known interactions of VPA with other antiepileptic drugs, CYP2C19 may also be involved in the oxidative metabolism of VPA (136). Of these P450 isozymes, CYP2A6, CYP2C9, and CYP2C19 are polymorphically expressed (137,138) and may therefore contribute to patient-to-patient variability in the metabolism of VPA. In the case of VPA biotransformation to 4-ene-VPA, genetic polymorphisms may contribute to the occurrence of hepatotoxicity associated with this metabolite in certain individuals.
In humans, VPA undergoes extensive hepatic metabolism ,and on average only 1% to 3% of the total dose is excreted unchanged in urine (122, 134). The major urinary metabolites of VPA are VPA glucuronide (10% to 70% of the dose) and 3-keto-VPA (6% to 60% of the dose) (97,122,124,133). Other metabolites such as (E,E)-2,3′-diene-VPA, 3-OH-VPA, 4-OH-VPA, 5-OH-VPA, 4-keto-VPA, and 2-PGA may account for 1% to 5% of the administered VPA dose (122,124,133,134). The 4-ene-VPA, (E)-2,4-diene-VPA, (E)-2-ene-VPA, 3-ene-VPA, and 2-PSA metabolites are very minor urinary products, and each accounts for 0% to 0.5% of the total VPA dose (122,133). The urinary excretion of 2-ene-VPA and 3-keto-VPA have been observed to increase after 200 mg twice daily administration of VPA over a 3-week period, apparently as a consequence of autoinduction of the β-oxidation pathway (118). This is accompanied by a small (~18%) decrease in the plasma AUC of VPA. As in humans, an increase in the urinary excretion of VPA and products of β-oxidation has also been observed in long-term VPA administration studies in rats, again apparently the result of autoinduction (139). Contribution of the glucuronide pathway to total VPA metabolism increases as a function of increasing VPA dose because of saturation of the β-oxidation pathway (120,122,124,134,140); the elimination of (ω) and (ω-1) cytochrome P450 oxidation products, however, appears to be relatively independent of dose. The percentage of VPA dose recovered in urine as 4-ene-VPA and its sequential metabolites has been observed to increase with increasing doses of VPA (141).
CLEARANCE AND HALF-LIFE
It has long been recognized that the level-dose relationship for VPA is highly variable among patients. Results from therapeutic monitoring studies (142, 143, 144) have shown that, at a given daily dose of VPA, the plasma drug level could vary as much as six- to eightfold among individual patients. Variation in VPA level:dose ratio largely reflects variability in the clearance characteristics of the drug.
The reported plasma (or metabolic) clearance in healthy volunteers (Table 84.1) is in the range of 6 to 8 mL/hr/kg, which is much lower than (≤0.03) the average hepatic blood flow (1,500 mL/min). Thus, VPA may be classified as a lowextraction drug, whose clearance is independent of blood flow and is positively dependent on its plasma free fraction (145). Consequently, factors affecting drug plasma protein binding and hepatic drug-metabolizing enzyme activities are important determinants of VPA clearance. A review of the relevant factors is presented in the following sections.
TABLE 84.1. PHARMACOKINETIC PARAMETERS OF VALPROIC ACID IN ADULT VOLUNTEERS AND PATIENTS WITH EPILEPSY
One complicating factor that contributes to the reported interindividual variability in the elimination kinetics of VPA is its dependence on dose. During long-term drug administration, a curvilinear relationship between plasma VPA level and dose has been noted in numerous studies (142,146, 147, 148, 149, 150, 151). Above a daily dose of 500 mg/kg, the steady-state plasma concentration of VPA increased less than proportionately with an increase in dose (i.e., a convex plot of concentration versus dose). A population kinetic study (150) in Japanese patients found that dose dependence in clearance was more pronounced in patients receiving carbamazepine than in patients receiving monotherapy or polytherapy with phenobarbital.
The mechanism of nonlinearity in VPA clearance was examined earlier by Bowdle et al. (152) in a multiple-dose study in healthy volunteers. Each volunteer received 500, 1,000, and 1,500 mg/day of oral VPA in three consecutive steps. The nonlinearity in clearance was attributed principally to an increase in free fraction (as plasma VPA concentration increased). No consistent change in the clearance of unbound VPA was observed. Comparable findings were reported by Gómez-Bellver et al. (153) in a single-dose study at 1,000, 2,000, and 3,000 mg of orally administered VPA. These investigators noted a trend toward a decrease in the clearance of unbound VPA with the increase in dose, a finding consistent with the hypothesis of a saturation in β-oxidation as mentioned earlier in the discussion of urinary excretion.
Hussein et al. (154) investigated the effect of infusion duration on the pharmacokinetics of intravenous VPA in a group of healthy volunteers. Each subject received an intravenous infusion of 1,000 mg sodium VPA over 5, 10, 30, and 60 minutes on four separate occasions. A biphasic postinfusion decline in plasma VPA concentration was observed. A more rapid initial decline occurred as the infusion duration was shortened; that is, initial half-life varied from 48 minutes for the 60-minute infusion to 8 minutes for the 5-minute infusion. The plasma clearance of VPA for the 5-minute infusion study was consistently lower than those for the longer infusion durations (0.56 L/hr versus ≥0.64 L/hr). These investigators proposed that the more rapid extravascular distribution of VPA with the 5-minute infusion duration was the result of saturation of plasma protein binding. The lower plasma clearance, conversely, reflected a partial saturation of intrinsic metabolic clearance that more than compensated for the effect from the increase in plasma free fraction.
Gender and Body Size
In a population pharmacokinetic study in a large cohort of Japanese patients, Yukawa et al. (150) reported that female patients on average had a slightly (~10%) lower weight-normalized clearance than male patients. This and another study by Suemaru et al. (155) in moderately obese Japanese patients also showed that apparent oral clearance correlated better with ideal body weight than either total body weight or body mass index. Hence, VPA should be given according to ideal body weight, rather than total body weight, in overweight patients.
Concomitant Antiepileptic Medications
VPA clearance is on average twofold higher in adult epileptic patients who are taking enzyme-inducing comedication than in patients receiving VPA monotherapy or in nonepileptic, healthy volunteers (Table 84.1). This increase in VPA clearance is attributed to induction of liver drug-metabolizing enzymes caused by concurrent antiepileptic medications (Chapter 85). An approximate 40% to 100% increase in VPA clearance has also been observed in epileptic children receiving multiple enzyme-inducing antiepileptic medications as compared with children receiving VPA monotherapy (Table 84.2). The increase in VPA clearance is seen in patients taking the older antiepileptic drugs, including phenobarbital, phenytoin, carbamazepine, or a combination thereof. Thus, higher doses of VPA may be required to maintain therapeutic concentrations in patients receiving polytherapy. Some reports indicate that, in some patients, these interactions are so pronounced that therapeutic levels of VPA are barely maintained or are not achievable even at extraordinary high doses (156, 157,158). There is also evidence that the magnitude of metabolic interaction may depend on the VPA dose (150).
The reported plasma half-life of VPA in healthy adult volunteers ranged from 12 to 16 hours (Table 84.1). The half-life of VPA is typically shorter in comedicated epileptic patients, and this is attributable to induction of VPA metabolism by other antiepileptic drugs. The mean half-life in epileptic adults taking enzyme-inducing comedication is ~9 hours (Table 84.1). However, elimination half-life values as short as 5 hours have been documented (34).
The half-life values in Tables 84.1 and 84.2 are all based on total VPA in plasma. When the plasma concentration of VPA ranges over 75 µg/mL during a dosage interval, the half-life of unbound drug will be shorter than that of total drug, as a result of the diminishing free fraction as the total drug concentration declines (see the earlier section on plasma protein binding). For example, in a group of young adult epileptic patients, Herngren and Nergardh (34) found that the mean terminal half-life of free VPA was 6.4±3.9 hours as compared with 11.9 ± 5.9 hours for the mean half-life of total VPA. The relatively short plasma half-life of free VPA argues for the use of sustained-release formulations.
Physiologic changes occurring during childhood development are known to affect the disposition of many antiepileptic drugs (159,160). A compilation from available literature of data on VPA pharmacokinetics in pediatric patients (Table 84.2) revealed marked changes in drug clearance throughout the early stages of development.
TABLE 84.2. PHARMACOKINETIC PARAMETERS FOR VALPROIC ACID IN PEDIATRIC PATIENTS
VPA is not used routinely in the treatment of neonatal seizures; therefore, pharmacokinetic data on newborns are scanty. Most of the literature reports were anecdotal observations in infants exposed to VPA in utero, which provide information on elimination half-life but do not afford clearance estimates. Pronounced changes in VPA half-life were observed during the postnatal period (Table 84.2). Within the first 10 days after birth, half-lives ranging between 10 and 67 hours were observed (161). Longer half-lives appeared to be associated with low birth weight (<1,000 g) and prematurity. Clearance data from a study of six neonates with intractable seizures were reported by Gal et al. (162). VPA was added to an existing regimen of anticonvulsants (mainly phenobarbital), and clearance kinetics were studied at steady state. Although highly variable, the steady-state clearance per kilogram of body weight for total drug in serum appeared to be within the range of values reported for adult epileptic patients receiving multiple anticonvulsants. The estimates were also consistent with results from two earlier case studies (161,163). The free fractions of VPA in neonatal serum are significantly higher than those reported for adult serum at comparable drug levels (an area average free fraction of 14.4±9.3% in the neonates versus ≤10% in adults). Consequently, clearance for unbound VPA was in the low range of expected values for adult epileptic patients, a finding suggesting that the intrinsic metabolic clearance of VPA may be low because of immature drug-metabolizing enzyme activities.
A remarkable increase in plasma clearance, a decrease in extravascular distribution volume, and a corresponding shortening in half-life of VPA occur from 10 days to 2 months of infancy. The increase in clearance presumably reflects maturation in drug-metabolizing function.
Data from the studies by Hall et al. (164) showed that in older infants between the ages of 3 to 36 months, clearance values exceeding 30 mL/min/kg are often observed in patients receiving enzyme-inducing comedication. In such patients, daily doses much higher than the usually recommended range of 15 to 30 mg/kg/day would be required to achieve plasma VPA levels >50 µg/mL. During this middle to late period of infancy, mean elimination half-lives of 8 to 12 hours and 6 hours have been reported for patients receiving monotherapy and for patients receiving enzyme-inducing comedication, respectively (Table 84.2).
Studies detailing age-related changes in VPA glucuronidation and β-oxidation, which appear to represent the primary pathways for VPA elimination in all species, have been described in postnatal lambs (165,166). Similar to human newborns and children <2 years of age, plasma concentrations of β-oxidation metabolites, 2-ene-VPA and 3-keto-VPA, and the P450-mediated hydroxylation metabolite, 4-OH-VPA were approximately five- to 10-fold higher in newborn lambs (1 day, 10 day, and 1 month of age) compared with the ewe, whereas glucuronide formation was considerably reduced (approximately threefold) (165,166). Urinary excretion of the major urinary metabolite, VPA glucuronide, accounted for ~28% of the dose in 1- and 10-day-old lambs but increased to adult levels (~75% of the dose) by 2 months of age. In agreement with very limited data available in infants and children, these studies suggest that β-oxidation activity is substantially developed early in life, whereas glucuronidation activity increases significantly with increasing age.
Ample data are available on VPA clearance in school-age children (3 to 16 years). Overall, the mean clearance estimates from numerous studies (Table 84.2) are in the range of 13 to 18 and 19 to 28 mL/hr/kg for patients receiving monotherapy and for patients taking enzyme-inducing comedication, respectively. These estimates are intermediate between those reported for the infants and young adults. It appears that VPA clearance normalized to per kilogram of body weight begins to decline after infancy and continues on throughout childhood, reaching adult values by adolescence. This pattern of continual decline in clearance relative to body weight over the first decade or so of childhood has been verified in more recent large-scale plasma VPA monitoring studies coupled with sophisticated population pharmacokinetic modeling (150,151,167,168). The pattern of development in the metabolic clearance of VPA is similar to the general trend observed with other antiepileptic drugs that are subject to oxidative metabolism, such as phenytoin and phenobarbital (159,160), and it can be attributed to prepubertal changes in cytochrome P450-mediated drug metabolism in the liver.
The elimination half-life of VPA begins to assume adult values during the early years of childhood. The average half-lives for school-age children and young adolescents are within the range of adult values (Table 84.2).
The pharmacokinetics of VPA in elderly subjects has been studied by several groups of investigators (Table 84.1). Both Perucca et al. (43) and Bauer et al. (37) reported that the clearance of total VPA in serum did not appear to differ between young and elderly volunteers. However, a decrease in serum protein binding of VPA associated with hypoalbuminemia was observed in the elderly group. Consequently, the mean clearance of free VPA was 40% lower in the elderly than in the young control subjects. This decrease in free drug clearance is consistent with the generally recognized age-related decline in hepatic function, notably with respect to oxidative drug metabolism (169). The results of these studies suggest that monitoring free, rather than total, serum VPA concentration may be more meaningful in elderly patients.
Alterations in VPA clearance may be observed during pregnancy and after parturition. Plasse et al. (170) reported the level:dose ratio of VPA in one pregnant mother. The level:dose ratio began to decline in the latter part of the second trimester and continued through the early part of the third trimester; it finally reached nadir within 3 weeks of delivery. After parturition, VPA levels rose rapidly and regained prepregnancy values within 2 to 3 weeks. In a review article, Philbert and Dam (171) cited similar experience in five pregnant patients. Other studies confirmed that a gradual and marked decrease in total plasma VPA concentration occurs during pregnancy, followed by a rapid rise after delivery; however, the concentration of free VPA in plasma was found not to be markedly altered during pregnancy compared with preconception levels, and therefore a need for dosage adjustments would not be anticipated (171a). Overall, these data suggest that the apparent increase in total VPA clearance during late gestation is largely caused by the previously recognized decrease in maternal serum protein binding of VPA, as a result of elevated nonesterified fatty acids and hypoalbuminemia (41). Although the question remains whether there is an actual change in the intrinsic metabolic clearance of VPA during pregnancy, monitoring of free, rather than total, serum VPA concentration may be more meaningful throughout the entire period of pregnancy and postpartum.
Orr et al. (46) reported the disposition kinetics of VPA in a 9-year-old uremic epileptic child. Total serum clearance was 23.6 mL/hr/kg after the first dose and increased to 40.8 mL/hr/kg after 5 months of therapy. The observed steady-state clearance was higher than reported estimates in polytherapy patients of comparable age (19 to 28 mL/hr/kg). Serum free fractions were higher than normal, at 22.4% and 27.2%, respectively, for the single-dose and steady-state studies. The corresponding free serum clearances of VPA were calculated to be 149 and 152 mL/hr/kg, in line with estimates for children with normal renal function. Thus, the primary effect of uremia is a decrease in serum protein binding resulting in an apparent increase in total drug clearance. Because free drug clearance and, therefore, the average free drug concentration at steady state is not altered, adjustment in VPA dosage may not be necessary in uremic patients. The elimination half-life of VPA after 5 months of treatment was 10.2 hours, a finding that agrees with expected values for epileptic children of this age. The rise in serum free fraction induced by uremia had no apparent effect on half-life, because there was a comparable increase in both clearance and apparent volume of distribution.
Only a limited fraction of the VPA dose (<20%) is removed by either hemodialysis (172) or peritoneal dialysis (46), probably because of the significant degree of binding of VPA to plasma protein. Therefore, there is no need to supplement the VPA dose in uremic patients who receive maintenance dialysis treatment. Hemodialysis for detoxification in VPA overdose, however, should be considered, because the plasma free fraction is elevated (>30%) at toxic concentrations (173).
VPA pharmacokinetics has been examined in patients with alcoholic liver cirrhosis and in patients recovering from acute hepatitis (44). VPA free fraction in serum was increased by more than twofold in patients with liver disease. However, the clearance of total VPA in serum was not significantly different from that in healthy volunteers because intrinsic clearance (reflecting drug-metabolizing activity) of VPA was also reduced, presumably as a consequence of hepatocellular damage. Thus, hepatic disease causes two opposing effects resulting in no apparent change in total clearance. Accordingly, AUC of plasma total VPA at steady state would not change, whereas that of plasma free VPA would increase in such a situation. The increase in unbound concentrations of VPA may warrant a downward adjustment in daily dose. The mean half-life in seven patients with alcoholic cirrhosis was 18.9±5.1 hours, and in patients recovering from acute hepatitis it was 17.0±3.7 hours. The prolongation in half-life reflects the decrease in intrinsic metabolic clearance of VPA in hepatic diseases.
Metabolic clearance of some drugs is known to increase in critically ill patients suffering from acute head injury (174,175). Anderson et al. (176) reported pharmacokinetic data on intravenously administered VPA in 35 patients with head trauma as part of a clinical trial to evaluate the efficacy of VPA in preventing posttraumatic seizures. The clearance of total VPA in plasma showed a steady increase over the next 2 weeks of treatment. By 1 month, the VPA clearance had returned to baseline level. Part of the apparent increase in plasma clearance was attributed to hypoalbuminemia in response to trauma. In addition, clearance of unbound VPA also showed a significant rise. The mechanism underlying the apparent induction in VPA metabolism is not known.
RELATIONSHIP BETWEEN SERUM CONCENTRATION AND DOSE
In patients receiving VPA monotherapy, doses between 10 and 20 mg/kg/day will usually achieve a good clinical response and will result in concentrations within the therapeutic range (50 to 100 µg/mL) (177, 178, 179, 180). The dose may be increased as necessary and as tolerated by 5 to 10 mg/kg/day at weekly intervals. Because of age-dependent kinetics, younger children may require a higher dose (181,182). However the level-dose relationship is highly variable among patients.
During long-term drug administration, a nonlinear relationship between plasma VPA levels and dose has been
observed in numerous studies, as mentioned earlier in the section on clearance and half-life. Above a daily dose of 500 mg, the steady-state plasma concentration of VPA increased less than proportionately with an increase in dose. The nonlinearity in clearance was attributed principally to an increase in the unbound fraction as plasma VPA concentration increased (152). However, dose proportionality was observed for doses >500 mg/day in other studies; Davis et al. (183) reported that peak plasma concentrations after oral administration of VPA (capsule, uncoated tablet, liquid formulation) increased dose proportionally, ranging from 24.5 µg/mL after a 250-mg dose to 108.5 µg/mL after a 1,000-mg dose. In another study, Wangemann et al. (184), investigated the proportionality of low doses, 100 to 300 mg, and the pharmacokinetics of sustained-release sodium VPA in healthy persons. Parameters determining the extent and rate of absorption, AUC, and maximum concentration increased proportionally with the dose.
Loading and maintenance intravenous doses necessary to achieve and maintain therapeutic serum VPA concentrations were determined by Hovinga et al. in a study in children (185). A 20 mg/kg loading dose and maintenance infusions of 4 and 6 mg/kg/hr produced steady-state total concentrations of 66 and 92.4 mg/L, respectively. In neonates, Alfonso et al. reported that each 1 mg/kg loading dose of intravenous VPA increased the 45-minute and 3-hour postinfusion serum VPA concentrations by approximately 4 and 3 µg/mL, respectively (186).
RELATIONSHIP BETWEEN SERUM CONCENTRATION AND EFFECT
Interpretation of VPA levels obtained in patients is difficult given that VPA levels fluctuate considerably during the 24-hour period because of a short half-life. No clear correlation between VPA levels and clinical effects at any given time has been demonstrated so far.
In epilepsy, several studies found clinical response, in term of seizure control or reduction in electroencephalographic seizure discharge, at VPA concentrations ranging from 43 to 109.5 µg/mL (146,179,181,187, 188, 189, 190, 191). In children, complete seizure control was obtained at lower VPA levels, between 20.2 and 50.5 µg/mL (192). The therapeutic range in epilepsy is considered to be between 50 and 100 µg/mL of total VPA. High concentration (80 to 150 µg/mL) may be needed in patients with epilepsy that is difficult to control (193).
Although no relationship could be established between the incidence of side effects and plasma VPA levels, patients with side effects had usually received significantly higher doses and exhibited higher serum VPA concentrations (194, 195). Lethargy and drowsiness were observed at mean levels of 80.4 and 94.5 µg/mL (195).
VPA is also currently used for treatment of psychiatric disorders, including acute mania, bipolar disorders, and schizophrenic disorders. A marked antimanic response may require plasma levels to be >50 µg/mL (196). However, Grunze et al. reported a drastic remission of mania in depressed patients with bipolar I disorder with VPA levels at or only slightly >50 µg/mL, when blood was drawn 12 hours after VPA intravenous infusion (66). In other psychiatric disorders, the therapeutic range is unknown, although 60 to 90 mg/L was cited on the basis of retrospective studies in patients with mania (197). In uncontrolled studies of VPA in bipolar and schizophrenic disorders, significant response occurred within 1 to 2 weeks of achieving serum VPA concentrations >50 mg/L (198).
VPA has also been used in the prophylaxis of migraine headaches. Lenaerts et al. assessed the prophylactic effect of VPA (mean daily dose, 928.5 mg) over 6 months in 56 patients with migraine or tension type-headaches. In the migraine group, 60% of the patients had a 75% to 100% improvement. This clinical improvement correlated with VPA levels, and the authors suggested aiming for VPA plasma levels between 70 and 90 µg/mL (199). Another clinical trial studied VPA in migraine prophylaxis and found no correlation between VPA levels and the therapeutic response (200). Erdemoglu et al. showed that a mean average daily dose of 1,250 mg VPA produced improvement in headache frequency in 67% of patients, with plasma levels ranging from 27 to 128 µg/mL (average, 74 µg/mL). In that study, no clear correlation between VPA levels and either treatment efficacy or side effects was found (201).
171a. Yerby MS, Friel PN, McCormick K. Antiepileptic drug disposition during pregnancy. Neurology 1992;42[Suppl 5]:12-16.