Levetiracetam, a New Antiepileptic Agent: Lack of In Vitro and In Vivo Pharmacokinetic Interaction with Valproic Acid

Authors


Address correspondence and reprint requests to Dr. T. R. Browne at 36 Riddle Hill Rd., Falmouth, MA 02540, U.S.A. E-mail: tbrowne@capecod.net

Abstract

Summary:  Purpose: The novel antiepileptic drug (AED) levetiracetam (LEV; Keppra) has a wide therapeutic index and pharmacokinetic characteristics predicting limited drug-interaction potential. It is indicated as an add-on treatment in patients with epilepsy, and thus coadministration with valproic acid (VPA) is likely. These studies were performed to determine whether coadministration of LEV with VPA might result in pharmacokinetic interactions.

Methods: In vitro assays were performed to characterize the transformation of LEV into its main in vivo metabolite UCB L057. The reaction was examined for its sensitivity to clinically relevant concentrations of VPA. An open-label, one-way, one-sequence crossover clinical trial was conducted in 16 healthy volunteers to assess further the possibility of any relevant pharmacokinetic interaction.

Results: Human whole blood and, to a lesser extent, human liver homogenates were demonstrated to hydrolyze LEV to UCB L057, its main metabolite. The reaction possibly involves type-B esterases and is not affected by 1 mM VPA (i.e., 166 μg/ml). Pharmacokinetic parameters of a single dose of LEV (1,500 mg) coadministered with steady-state concentrations of VPA (8 days of 500 mg, b.i.d.) did not differ significantly from the pharmacokinetics of LEV administered alone [area under the curve (AUC) of 397 and 400 μg/h/ml, respectively]. Furthermore, LEV did not affect the steady-state pharmacokinetics of VPA.

Conclusions: These findings suggest the absence of a pharmacokinetic interaction between VPA and LEV during short-term administration, and suggest that dose adjustment is not required when these two drugs are given together.

Whereas the majority of epilepsy patients receive single-drug treatment regimens, failure to control seizures adequately often necessitates the administration of multiple antiepileptic drugs (AEDs). Pharmacokinetic drug interactions among AEDs are often experienced, which can result in undesirable drug concentrations, with concomitant effects on seizure control and side effects (1).

VPA is a broad-spectrum inhibitor of drug-metabolizing enzymes, including uridine diphosphate glucuronosyltransferase (UGT), epoxide hydrolase, and cytochrome P-450 2C (CYP2C) enzymes (2,3). As a consequence, VPA can reduce the metabolic elimination of the AEDs that are substrates for these enzymes, such as carbamazepine (CBZ), ethosuximide (ESM), lamotrigine LTG), and phenobarbital (PB) (2,4,5). In addition, VPA is highly bound to plasma albumin and hence has the potential to displace drugs from their plasma protein (6). Numerous articles have reported the ability of VPA to increase free phenytoin (PHT) concentrations simultaneously through metabolism inhibition and to decrease protein-bound PHT through displacement from protein-binding sites (7,8).

These two conflicting mechanisms result in a complex and almost unpredictable effect on total PHT concentrations. Conversely, because of its extensive transformation, the disposition of VPA can be affected by coadministered drugs; particularly the AEDs that induce UGTs may accelerate the metabolic elimination of VPA. PB, PHT, and CBZ can double the clearance of VPA (2,9,10). Conversely, felbamate (FBM) inhibits the β-oxidation pathway of VPA, causing a significant increase in its plasma concentrations (11).

When compared with VPA and the previously mentioned AEDs, the new AED levetiracetam (LEV) has a more favorable pharmacokinetic profile, with low risks for drug interactions. LEV is not significantly bound to proteins (<10%) and shows limited metabolism (12). In humans, 66% of the dose is eliminated unchanged in urine, with 24% excreted in the urine as its carboxylic derivative, UCB L057 (12). Furthermore, in vitro assays using human liver fractions and specific marker activities have shown that LEV does not inhibit the most common drug-metabolizing enzymes, including CYPs (1A2, 2A6, 2C9, 2C19, 2D6, 2E1, 3A4), UGTs, and epoxide hydrolase (13). The low potential of LEV for drug interactions was noted in several clinical trials in which LEV was administered to patients with refractory seizures who were receiving other AEDs, including CBZ, LTG, PHT, VPA, PB, gabapentin (GBP), and primidone (PRM (14–16). Although these were not specifically drug-interaction studies, LEV appeared to have no significant effect on the serum concentrations of other AEDs.

LEV and VPA are likely to be coadministered for an extended period in patients with refractory partial complex seizures. Because VPA has been associated with numerous drug interactions, in vitro and in vivo approaches were combined to evaluate the possibility of pharmacokinetic interactions when the two drugs are coadministered. In vitro assays were first conducted to characterize the hydrolysis of LEV to UCB L057 and to determine the ability of VPA to impair the reaction. The potential for a pharmacokinetic interaction was further assessed in an open-label, one-way, one-sequence crossover clinical study in healthy volunteers.

METHODS

In vitro hydrolysis of levetiracetam

Human samples for in vitro incubations

Whole blood was drawn from healthy male volunteers who had taken no medications during the previous 7 days. Blood was collected into tubes containing lithium heparin as an anticoagulant and was used immediately after collection. Human liver samples were obtained under strict ethical conditions from organ donors at the Department of Pediatric Hepatology, University of Louvain, Brussels, Belgium. Liver samples were homogenized in 50 mM ammonium hydrogen carbonate buffer (pH 7.4) to give ∼20% wt/vol homogenates. Samples were then centrifuged at 600 g for 10 min at 4°C, and the resulting supernatants were decanted off and stored at ca.–80°C until subsequent use. Protein content was measured as described elsewhere (17) with a ready-to-use Bicinchoninic Acid Protein Assay kit (Pierce, Rockford, IL, U.S.A.).

In vitro assay for levetiracetam hydrolysis by using human liver homogenates

The incubation mixture (final volume of 1 ml) consisted of liver homogenates (2 mg protein/ml) and 200 μM LEV in 50 mM ammonium hydrogen carbonate buffer (pH 7.4). Control incubates were run with heat-denaturated liver homogenates (10 min at 100°C). Incubation was conducted at 37°C for 0.5 to 6 h in a shaking water bath. At the end of the incubation, samples were mixed with 1 ml acetonitrile and centrifuged at 3,000 g for 10 min. The supernatant was spiked with 50 μl internal standard (UCB 24167, 2.5 μg/ml), evaporated to dryness by using a concentrator evaporator, and the dry residue reconstituted with 500 μl acetonitrile. The sample was mixed thoroughly, centrifuged at 10,000 g for 10 min, the supernatant collected, evaporated to dryness, and reconstituted in 150 μl mobile phase (composition given in the section, Analysis of UCB L057 in the in vitro incubates).

In vitro assay for levetiracetam hydrolysis using human whole blood

LEV hydrolysis in human whole blood was investigated first for its linearity with respect to incubation time. Incubation mixture consisted of a 990-μl blood sample spiked with 10 μl LEV (solution in water) to achieve the final concentration of 200 μM. Incubation was conducted at 37°C from 0.5 to 6 h in a shaking water bath. At the end of the incubation, the incubated mixture was centrifuged at 3,000 g for 10 min. Three hundred microliters of the plasma fraction was mixed with 300 μl acetonitrile and centrifuged at 3,000 g for 10 min. The supernatant was evaporated to dryness by using a concentrator evaporator, the dry residue reconstituted with 100 μl mobile phase and spiked with 50 μl internal standard (UCB 24167, 2.5 μg/ml).

The reaction was then characterized for its kinetic parameters. The assay was performed as described earlier by using 2-h incubation and LEV concentrations ranging from 20 to 1,000 μM. The incubation time was selected because it was within the linear region of the reaction. KM (Michaelis-Menten constant) and maximal reaction velocity (Vmax) were calculated by nonlinear regression analysis (Statistica 5.0; StaSoft Inc, Tulsa, OK, U.S.A.).

Subsequent assays were conducted to measure the hydrolysis of LEV in the presence of VPA. Incubation mixtures contained whole blood (980 μl), 200 μM LEV (added as 10 μl in water), and 1 mM VPA (added as 10 μl in water). Control incubations were performed in which VPA was replaced by an equivalent volume of its vehicle. Mixtures were incubated for 2 h at 37°C and were processed as described earlier. The VPA concentration in the assay was selected as encompassing the therapeutic plasma concentrations, whereas the LEV concentration was selected as being around the KM of the reaction (i.e., 439 μM).

A similar inhibition assay was performed by using reference inhibitors of hydrolytic enzymes. They were selected, as was their final concentration in the assay, based on the literature: edetic acid (EDTA) and chloromercuribenzoate (1 mM and 100 μM, respectively; inhibitors of A-esterases), paraoxon, physostigmine, metoclopramide, sodium fluoride (NaF), and bis-p-nitrophenyl phosphate (BNPP) (all at 100 μM; inhibitors of B-esterases), acetazolamide (100 μM, inhibitor of carboxylanhydrase), and acetylsalicylic acid (100 μM;aspecific inhibitor of esterases) (18–23). Incubation mixtures contained whole blood (980 μl), 700 μM LEV (added as 10 μl in water), and the inhibitor (added as 10 μl in either water or methanol). Mixtures were incubated and were processed as described earlier. Control incubations were conducted with the vehicles.

Analysis of UCB L057 in the in vitro incubates

Concentrations of UCB L057 in the in vitro incubates were measured by using high-performance liquid chromatography (HPLC) with electrospray ionization mass spectrometry detection (LC/ESI/MS). The method was essentially the same as that described in the drug-assay section. In brief, a 10-μL aliquot was injected into the HPLC fitted with an analytic column (Inertsil 5, ODS 250 × 2-mm ID) and protected by a guard column of the same material. The oven temperature was kept at 30°C. The flow rate was set at 0.2 ml/min. The mobile phase consisted of acetonitrile/water, 5:95 vol/vol, containing 5 mM of ammonium acetate. The flow was split to deliver 0.125 ml/min to the triple quadrupole mass spectrometer equipped with electrospray ionization source. The spectrometer was operated in the negative ion mode with a source temperature of 150°C, a capillary voltage of 3 kV, and a cone voltage of 30 V. Single ion recording was used with a dwell time and an interchannel delay fixed at 0.5 and 0.02 s, respectively. Chromatographic traces corresponding to ions with m/z of 170 a.m.u. [(M-H) of UCB L057] and 156 a.m.u. [(M-H) of internal standard] were recorded in function of time. Calibration curves for UCB L057 were constructed in human whole blood and human liver homogenate. The method was found to have a lower limit of quantitation of 0.1 μM.

In vivo study of levetiracetam and valproic acid pharmacokinetic interactions

Sixteen healthy volunteers (10 men and six women between the ages of 22 and 52 years, with a body mass index from 18.5 to 30.3 kg/m2) were included in this open-label, one-way, one-sequence crossover study. Subjects were excluded if they had a history of serious or relevant organ dysfunction, illness, or medication, or if they were pregnant or lactating. Subjects also were phenotyped for genetic polymorphic CYP 450 isoforms by using dextromethorphan and S-mephenytoin before drug dosing. Subjects with abnormal isoforms were excluded. The use of an open-label design was justified by the fact that objective parameters such as pharmacokinetic measurements were the main end points. Healthy subjects were used because of the high risk of withdrawal seizures when VPA treatment is interrupted in patients with epilepsy.

Study design

LEV doses were administered during two confinement periods, each lasting 3 days. A single dose of 1,500 mg was taken on the morning of day 1. Over the next 3 days, a multipoint pharmacokinetic profile was performed for LEV and its major metabolite, UCB L057. From day 3 to day 11, 500 mg of delayed-release VPA (Depakine Enteric) was administered b.i.d. Achievement of steady-state conditions of VPA was checked by trough plasma level measurements. A second dose of LEV (1,500 mg) was given on the morning of day 10 during the second 3-day confinement period. A multipoint pharmacokinetic profile for VPA, LEV, and UCB L057 was conducted over a 3-day period.

Drug assays

Blood samples for plasma trough concentrations of VPA were taken in the morning before VPA dose administration. Blood samples were taken just before LEV dosing and then at 20 and 40 min and 1, 1.5, 2, 3, 4, 6, 9, 13, 24, 36, and 48 h after drug administration. On the third day after coadministration of LEV with VPA, additional blood samples were taken for VPA plasma analysis. Urine was collected for LEV and UCB L057 analysis before each LEV administration and at 3, 6, 9, 13, 24, 36, and 48 h after each dose.

Plasma and urine samples were analyzed for their concentrations of VPA, LEV, and UCB L057 following the recommendations of the Conference of Washington (24).

Extraction and analysis of valproic acid

VPA concentrations in plasma were measured by a validated liquid chromatography–mass spectroscopy (LCMS) method. This method has a limit of quantification of ∼2 μg/ml. In brief, 50 μl of plasma was mixed with a methanolic solution of the internal standard (250 μl cyclohexanecarboxylic acid). After centrifugation, the organic phase was evaporated to dryness at 40°C under nitrogen. The dry residue was taken up in 100 μl of an acetonitrile/10 mM ammonium acetate solution containing 0.1% formic acid (20:80, vol/vol), and 20 μl was injected in the HPLC system. Chromatographic separation was achieved on a Nucleosil C18, 5 μm, 70 × 2-mm I.D. column, by using a mobile phase made of acetonitrile/10 mM ammonium acetate solution containing 0.1% formic acid (50:50), vol/vol at a flow rate of 0.02 ml/min. The samples were injected in the ion-spray interface. Detection was achieved by a single quadrupole mass spectrometer in the negative ion mode (for VPA: Q1, m/z = 143.1; for internal standard: Q1, m/z = 127.1).

Calibration curves ranging from 2 to 200 μg/ml were prepared from human plasma. The measurement errors and the coefficients of variation for within-run and between-run precision and accuracy did not exceed 10.3% throughout the range. Samples were properly diluted before analysis if required.

Extraction and analysis of levetiracetam and UCB L057

LEV concentrations in plasma and in urine were measured by validated gas chromatographic assays with nitrogen–phosphorus detection. This method has a limit of quantification of ∼0.5 μg/ml for plasma and 5 μg/mL for urine. In brief, 50 μl of plasma spiked with a methanolic solution of the internal standard (50 μl) was added with methanol (200 μl). The solution was stirred and then stored for 30 min at –20°C. After centrifugation, portions of the clear supernatant (3 μl) were injected into the gas chromatograph fitted with a 15 m × 0.31-mm ID fused silica column coated with a FFAP stationary phase (0.52 μm). The oven temperature was kept at 230°C.

Calibration curves ranging from 0.5 to 40.5 μg/ml were prepared from human plasma. Samples were diluted before analysis, if required. The mean of daily adjusted recoveries of LEV plasma QC samples analyzed within the study were 105.8, 101.7, and 99.0% at nominal concentrations of 1.0, 8.0, and 30 μg/ml, respectively. The intraassay imprecision (RSD) assessed from QC samples was ≤5.7%.

For urine, 20 μl of samples spiked with 20 μl of the internal standard solution and added with 960 μl of methanol were analyzed with the same chromatographic conditions. Calibration curves ranging from 5 to 303 μg/ml were used. The mean of daily adjusted recoveries of LEV urine QC samples analyzed in the study were 104.7, 99.9, and 101.1% at 10, 40, and 200 μg/ml, respectively. The intraassay imprecision (RSD) assessed from QC samples was ≤4.5%.

UCB L057 concentrations in plasma and in urine were measured by validated HPLC assays with electrospray ionization mass spectrometry detection (LC/ESI/MS). This method has a limit of quantification of ∼0.02 μg equivalent LEV per milliliter for plasma and 10 μg equivalent LEV per milliliter for urine.

In brief, 100 μl of plasma spiked with an aqueous solution of the internal standard (50 μl) was extracted with ethyl acetate (500 μl). After centrifugation, the aqueous phase was added with acetonitrile (500 μl), mixed, and then centrifuged. The supernatant was evaporated to dryness by using a concentrator evaporator, and the dry residue reconstituted with the mobile phase. Portions of 20 μl were injected into the liquid chromatograph fitted with an analytic column (Inertsil 5 ODS 250 × 2-mm ID) protected by a guard column (Hypersil ODS 5 μm, 20 × 2.1-mm ID). The oven temperature was kept at 40°C. The mobile phase consisted of acetonitrile/water 5:95 vol/vol, containing 5 mM ammonium acetate. Chromatographic traces corresponding to ions with m/z of 170 a.m.u. [(M-H) of ucb L057] and 156 a.m.u. [(M-H) of internal standard] were recorded in function of time. Calibration curves ranging from 0.02 to 2.02 μg equivalent LEV/ml were prepared from human plasma. Samples were diluted before analysis if required. The mean of daily adjusted recoveries of UCB L057 plasma QC samples analyzed in the study was 110.6, 106.7, and 106.9% at the nominal concentrations of 0.060, 0.200, and 1.00 μg equivalent LEV/ml, respectively. The intraassay imprecision (RSD) assessed from QC samples was ≤3.3%.

For urine samples, an aliquot (10 μl) spiked with the internal standard prepared in the mobile phase (50 μl) was added with mobile phase (940 μl). Fractions of 10 μl were injected into the LC/ESI/MS system. The chromatographic conditions were the same as for plasma. The calibration curve prepared in human urine ranged from 10 to 1,000 μg equivalent LEV/ml.

The mean of daily adjusted recoveries of UCB L057 urine QC samples analyzed in the study was 97.9, 104.4, and 99.4% at nominal concentrations of 20, 100, and 600 μg equivalent LEV/ml, respectively. The intraassay imprecision (RSD) assessed from QC samples was ≤6.6%.

Pharmacokinetic parameters

The pharmacokinetic parameters evaluated for LEV were maximal plasma concentration (Cmax), time after dose at which Cmax occurred (tmax), terminal elimination rate constant (λZ), plasma elimination half-life (t1/2), area under the plasma concentration–time curve (AUC), apparent total body clearance (CL/f), volume of distribution (VZ/f), amount of drug (mg) excreted in urine (Ae), amount of drug (percentage of dose) excreted in urine (fe), renal clearance (CLR), and nonrenal clearance (CLNR). The parameters calculated from VPA plasma concentrations included Cmax, minimalm plasma concentration (Cmin), AUC(0-12h), average concentration (Cav), peak trough fluctuation (PTF), tmax, and trough concentrations. The terminal t1/2 was calculated as ln(2)/λZ. The CL/f was calculated as dose/AUC and adjusted for a body surface area of 1.73 m2.

Statistical analysis

Descriptive statistics were conducted for plasma concentrations and pharmacokinetic parameters for LEV, UCB L057, and VPA. Absence of drug interaction between LEV and VPA was assessed by means of analysis of variance and the calculation of the 90% confidence interval of the ratio of geometric least squares mean, as recommended by the Food and Drug Administration (FDA). This method is operationally equivalent to Schuirmann's two one-sided t test procedure at the 5% significance level. Bioequivalence, or lack of interaction, was concluded if the 90% confidence interval for the ratio of geometric least squares mean of AUC for LEV with and without VPA was fully within the 80 to 125% acceptance range (25).

Adverse events

Adverse events were evaluated throughout the study at subject visits and at blood sampling. A complete physical examination was performed on each subject before study initiation and within 12 h after the last pharmacokinetic sampling.

RESULTS

In vitro hydrolysis of levetiracetam

When incubated in vitro with human whole blood, and to a lesser extent with human liver homogenate, LEV (200 μM, i.e., 34 μg/ml) was hydrolyzed to UCB L057. The reaction followed a linear course up to 6 h (Table 1). No UCB L057 was generated after incubation with buffer alone or with heat-inactivated liver homogenates, providing evidence that the reaction is enzymatic.

Table 1. Formation of ucb L057 (nmol/ml incubate) after incubation of levetiracetam with human whole blood and liver homogenate
Incubation
time (h)
Whole
blood
Liver
homogenate
Heat-denatured
liver homogenate
  1. The incubation was conducted by using 200 μM levetiracetam (i.e., 34 μg/ml). Whole blood was derived from a single donor, whereas liver homogenate was a pool from three donors.

0.50.700.000.00
11.690.090.00
24.350.200.00
410.210.270.00
614.000.400.00

The kinetic parameters of the reaction were determined in whole blood after a 2-h incubation with LEV concentrations ranging from 20 μM to 1 mM(Fig. 1). The reaction followed monophasic Michaelis–Menten kinetics with a KM of 439 μM and a Vmax of 129 pmol/min/ml blood, consistent with a low-affinity and low-turnover-rate process.

Figure 1.

Influence of levetiracetam (LEV) concentration of the hydrolysis rate in human whole blood. Blood sample from a single donor was incubated for 2 h at 37°C with the indicated final concentration of LEV. Michaelis–Menten and Lineweaver–Burke (insert) plots are presented. In the Michaelis–Menten curve, data are given as mean ± standard deviation of triplicate incubations. For the Lineweaver–Burke plot, only the means are presented.

In an attempt to characterize the enzymes involved, some selected prototypic inhibitors of hydrolytic enzymes were tested for their ability to impair LEV hydrolysis in whole blood (Table 2). Control incubates contained an equal volume of vehicle (i.e., water or methanol depending on the inhibitor tested). Methanol (1% vol/vol) was found to decrease the activity by 35%. In line with this finding, methanol was previously reported to inhibit hepatic carboxylesterase (22). Production of UCB L057 was reduced by 92% by 100 μM paraoxon, a broad inhibitor of B-esterases. In the present conditions, no relevant effect was experienced with the other B-esterase type inhibitors specific toward cholinesterase and/or carboxylesterase (physostigmine, metoclopramide, NaF, BNPP). Esterase-A inhibitors (EDTA, chloromercuribenzoate), acetazolamide (carboxylanhydrase), and acetylsalicylic acid (various esterases) also were without relevant effects.

Table 2. Effect of reference inhibitors on the hydrolysis of levetiracetam by human whole blood
Inhibitorucb L057 formed
(pmol/min/ml blood)
Mean %
of control
  1. The incubation was conducted for 2 h with 700 μM levetiracetam (i.e., 119 μg/ml). A single donor was evaluated in triplicate.

  2. EDTA, ethylenediaminetetraacetic acid; BNPP, bis-p-nitrophenylphosphate; NaF, sodium fluoride.

Control (1% water)84.4 ± 12.8100
 1 mM EDTA87.2 ± 2.7103
 100 μM BNPP93.2 ± 2.5110
 100 μM physostigmine83.3 ± 8.899
 100 μM NaF91.4 ± 4.1108
 100 μM acetazolamide72.5 ± 10.586
Control (1% methanol)54.9 ± 6.0100
 100 μM paraoxon4.5 ± 0.78
 100 μM acetylsalicylic acid59.6 ± 3.9109
 100 μM chloromercuribenzoate68.7 ± 4.5125
 100 μM metoclopramide57.8 ± 1.5105

Finally, the hydrolysis of 200 μM LEV by whole blood (2-h incubation; three separate donors) was found to be unaffected by 1 mM VPA (166 μg/ml; Table 3).

Table 3. Effect of valproic acid on the hydrolysis of levetiracetam by human whole blood
Inhibitorucb L057 formed
(pmol/min/ml blood)
Mean %
of control
  1. The incubation was conducted for 2 h with 200 μM levetiracetam (i.e., 34 μg/ml). Three donors were evaluated separately; each was tested as a single replicate.

Control (1% water)42.2 ± 5.1100
 1 mM valproic acid41.7 ± 6.099

In vivo

All 16 subjects completed the study. Steady-state plasma concentrations of VPA were achieved after 7 days of twice-daily administration of 500 mg of a delayed-release formulation (Depakine Enteric; Table 4). The similarity in mean trough plasma concentrations on days 9 to 12 demonstrates that steady state was reached on day 10 at the time of coadministration with LEV. Plasma concentrations measured on day 10 remained fairly constant during one dose interval, as expected after repeated administration of a delayed-release formulation (26). At steady state on day 10, Cmax of VPA was 93.7 ± 19.9 μg/ml (mean ± SD) and was achieved ∼3 h after dosing.

Table 4. Pharmacokinetic parameters of valproic acid in healthy volunteers after repeated administration of valproic acid, 500 mg b.i.d. (mean ± SD; n = 10 men and six women) and a single dose of levetiracetam (1,500 mg) on the morning of day 10
  Study day
Pharmacokinetic
parameter
UnitDay 3Day 4Day 5Day 7Day 9Day 10Day 11Day 12
  1. Cmax, peak plasma concentration; tmax, time to reach Cmax; AUC(0–12h), area under the plasma concentration–time curve over the 12-h administration interval; Cav, average plasma concentration; PTF, peak-to-trough concentration.

Morning troughμg/ml56.7 ± 19.953.8 ± 17.660.1 ± 17.075.6 ± 25.084.5 ± 21.278.7 ± 26.074.6 ± 26.4
Evening troughμg/ml32.1 ± 11.565.7 ± 20.072.5 ± 25.2
Cmaxμg/ml93.7 ± 19.9
tmaxh2.90 ± 3.62
AUC (0–12 h)μg.h/ml870 ± 199
Cavμg/ml72.5 ± 16.6
PTF morning0.12 ± 0.15
PTF evening0.40 ± 0.22

The pharmacokinetic parameters of LEV administered alone are shown in Table 5, and plasma concentrations are shown in Fig. 2. The pharmacokinetic parameters of LEV and UCB L057 were similar when LEV was administered with VPA (Table 5, Fig. 3). Multiple-dose coadministration of VPA did not modify the extent and rate of exposure to UCB L057 (Table 5, Figs. 2 and 3). The cumulative urinary excretion of UCB L057 also was not noticeably modified compared with that when LEV was given alone. A lack of interaction between LEV and VPA was clearly demonstrated by the bioequivalence of the pharmacokinetics of LEV and UCB L057 with and without VPA cotreatment (Table 5). The 90% confidence interval for all pharmacokinetic parameters for LEV and its metabolite were well within the 80–125% range. The coefficient of variability of LEV for AUC and Cmax were equal to 4.5 and 14%, respectively, for within-subject variation and 17.5 and 27%, respectively, for between-subject variation.

Table 5. Pharmacokinetic parameters of levetiracetam and ucb L057 after administration of a 1,500-mg single oral dose of levetiracetam either alone or with valproic acid
  Levetiracetam administered
Pharmacokinetic
parameter
UnitAlone
(mean ± SD)
With valproic acid
(mean ± SD)
Ratio of geometrica
means in % (90% CI)
  • Cmax, peak plasma concentration, tmax, time to reach Cmax; AUC, area under the plasma concentration–time curve extrapolated at infinite time; λz, terminal rate constant; CL/f, apparent body clearance; Vz/f, volume of distribution; Ae and fe, cumulative amount excreted in urine over 48 h in mg and % of the dose, respectively; LEV, levetiracetam.

  • a

     Computed from back-transformation of the difference of least square means between test (co-administration with VPA) and reference (levetiracetam alone) treatment.

  • b

     Median (min-max).

  • c

     Median difference between test and reference and 90% confidence interval of difference.

Levetiracetam    
 Cmaxμg/ml39.6 ± 9.941.4 ± 9.7104.7 (95.9:114.4)
 tmaxbh0.67 (0.33–2.00)0.67 (0.33–2.00)0.00 (−0.29:0.13)c
 λzh−10.095 ± 0.0140.094 ± 0.01299.2 (97.6:100.7)
 AUCμg/h/ml400 ± 47397 ± 6098.9 (96.2:101.6)
 CL/fml/min/kg0.89 ± 0.110.90 ± 0.13101.1 (98.4:103.9)
 Vz/f1/kg0.57 ± 0.070.58 ± 0.08101.8 (99.4:104.3)
 Aemg939 ± 85871 ± 11392.9 (89.1:97.0)
 fe%62.6 ± 5.758.1 ± 7.6
ucb L057    
 Cmaxμg eq./ml1.08 ± 0.201.02 ± 0.2294.1 (89.9:98.6)
 tmaxbh6.00 (3.00–9.00)6.00 (3.00–9.00)1.00 (0.00:1.5)‡
 λzh−10.082 ± 0.0110.082 ± 0.011100.0 (97.4:102.8)
 AUCμg eq.h/ml23.0 ± 6.521.9 ± 6.395.0 (90.3:99.9)
 Aemg eq LEV308 ± 75275 ± 5887.9 (84.7:91.3)
 fe%20.5 ± 5.018.3 ± 3.9
Figure 2.

Mean plasma concentration–time profile of levetiracetam (LEV) and UCB L057 (mean ± SD) in healthy volunteers (10 men, six women) after oral administration of LEV (1,500 mg) given alone. Data are presented in semi-log scale.

Figure 3.

Mean plasma concentrations–time profile of levetiracetam (LEV) and UCB L057 (mean ± SD) in healthy volunteers (10 men, six women) after oral administration of LEV (1,500 mg) given with valproic acid (500 mg). Data are presented in semi-log scale.

All adverse events were mild. The most frequently reported adverse events included dizziness, asthenia, headache, and dry mouth. The majority of events occurred during the administration of VPA alone. There were no reported clinically relevant changes in vital signs, physical status, ECG, or laboratory parameters.

DISCUSSION

Pharmacokinetics and the risk for drug interactions are key elements in the selection of AEDs (27). In this respect, LEV shows an optimal profile with absolute oral bioavailability of nearly 100%; linear kinetics; an elimination half-life of 6 to 8 h in healthy, young volunteers; no evidence for autoinduction; and no propensity for drug interactions (28).

In humans, LEV shows limited metabolism; 66% of the dose is eliminated unchanged in urine. Previous pilot in vitro studies suggested that the hydrolysis of LEV to its main in vivo metabolite, UCB L057, is catalyzed by blood cells (with little activity found in plasma) as well as a wide range of diverse tissues (liver, kidney, lung, brain, and intestinal homogenates derived from male rats). Because the reaction does not require nicotinamide adenine dinucleotide phosphate, contribution of CYP29 can be ruled out. The present in vitro data further show that the hydrolysis of LEV in human whole blood is a low-clearance process in line with the in vivo findings. The reaction is inhibited by paraoxon, a broad inhibitor of type B esterases, but not by the other investigated prototypic esterase inhibitors including physostigmine and BNPP. This inhibition profile closely resembles that reported for the hydrolysis of tazarotene and suggests the involvement of a type B esterase distinct from the classic cholinesterases and carboxylesterases (21).

The in vitro production of UCB L057 in whole blood is not affected by 1 mM VPA (166 μg/ml). This is an important finding considering the low affinity of the reaction, and hence its potential sensitivity to competitive inhibition, and the broad-spectrum inhibition profile of VPA. The assay was designed to satisfy basic kinetic requirements (i.e., incubation time comprised within the linear region of the reaction and a LEV concentration chosen around the KM). At therapeutic doses, the total plasma level of VPA is in the range of 50–100 μg/ml. As expected from its low volume of distribution (0.16 L/kg) and its high protein binding (27), VPA shows limited tissue uptake (animal biodistribution studies) (30) and restricted partitioning into blood cells (red blood cell/plasma partition ratio of 0.2 in human) (31). Thus it can be deduced that the VPA concentration tested in the present in vitro interaction assay largely encompasses the concentration expected at the site of LEV hydrolysis in the clinical situation.

Previous in vitro studies reported that LEV does not inhibit the major human liver CYPs nor VPA glucuronidation at 1 mM (170 μg/ml), a multiple of the Cmax measured after recommended dosage (13). Although all of the available in vitro data are reassuring, they do not preclude pharmacokinetic interactions between LEV and VPA through alternate mechanisms. Particularly, the in vitro interaction assays did not investigate a potential inhibitory effect of LEV on mitochondrial β-oxidation, an important pathway in VPA elimination. Furthermore, it has been reported that recommended doses of VPA may induce some esterase activities in epilepsy patients (32), raising the issue of a potential acceleration of LEV elimination after coadministration with VPA. These considerations led us to conduct a one-way, one-sequence crossover drug-interaction clinical trial in healthy volunteers.

The pharmacokinetics of LEV and UCB L057, after an oral dose of 1,500 mg, are unaffected by VPA (500 mg b.i.d.). Multiple-dose coadministration of VPA, 500 mg b.i.d. (Depakine Enteric) did not modify the pharmacokinetic behavior of UCB L059, nor were the rate and extent of exposure to UCB L059 and UCB L057, CL/f, λz, Vz/f, and urinary excretion modified when compared with the values obtained when LEV was given alone. These pharmacokinetic parameters are consistent with those previously reported (26).

Conversely, steady-state VPA pharmacokinetics are not modified by LEV (26). The single dose of LEV used in this study is at the upper end of the recommended therapeutic dose for epilepsy. VPA was administered at the recommended dose of 500 mg b.i.d. for 8 days. This dosing regimen is required to achieve steady-state plasma concentrations and is sufficient to evaluate VPA drug-interaction potential (26).

The results from the one-way, one-sequence crossover study performed in healthy adults suggest that VPA does not interact with the novel AED LEV, corroborating the in vitro findings.

Ancillary