- Email: [email protected]
intravenous maintenance dosing of valproate following intravenous loading: a simulation
Epilepsy Research 53 (2003) 29–38
Oral/intravenous maintenance dosing of valproate following intravenous loading: a simulation Sandeep Dutta a,∗ , James C. Cloyd b , G. Richard Granneman a , Stephen D. Collins a,1 a
Abbott Laboratories, Department R4PK, Bldg. AP13A-3, 100 Abbott Park Road, Abbott Park, IL 60064-6104, USA b University of Minnesota, College of Pharmacy, Minneapolis, MN, USA Received 9 May 2002; received in revised form 2 October 2002; accepted 11 October 2002
Abstract Valproic acid (VPA) has a narrow therapeutic range (50–100 mg/l) and exhibits nonlinear protein binding. Additionally, VPA pharmacokinetics are dependent on age, induction status, and formulation; so titration and dosing vary between individuals. The aim of these simulations was to determine optimal intravenous (i.v.) loading dose, and i.v. and oral VPA maintenance regimens. A 5-min 15 mg/kg loading dose resulted in total and free plasma VPA concentrations of ∼65 and 7.5 mg/l in children, and ∼80 and 11 mg/l in adults, 1 h after the infusion; induction status had little effect. For uninduced children and adults, 7.5 and 3.5 mg/kg q6h i.v. valproate sodium, initiated 6 h after loading dose maintains therapeutic plasma VPA concentrations. The rapid decline of plasma VPA concentrations following an i.v. loading dose in combination with the delayed initial absorption of drug from delayedrelease divalproex sodium tablets warrant beginning q12h oral maintenance regimens of delayed-release divalproex sodium within 2 h of a loading dose in the uninduced population. Plasma VPA concentrations can be sustained in the therapeutic range using once-daily maintenance regimens of extended-release divalproex sodium tablets if initiated concurrently with i.v. loading dose in the uninduced population. A two-fold higher i.v. and oral maintenance regimen dose may be required in induced patients. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Divalproex; Plasma; Pharmacokinetics; Concentration
1. Introduction Valproic acid (VPA) and related salt formulations are widely used in a variety of conditions, including seizures, mania associated with bipolar disorder, as well as for prophylaxis of migraine headaches. The recent approval2 of an intravenous (i.v.) formulation ∗ Corresponding author. Tel.: +1-847-937-8502; fax: +1-847-938-5193. E-mail address: [email protected] (S. Dutta). 1 Present address: RW Johnson PRI, Raritan, NJ, USA. 2 DEPACON is now approved for infusions up to 15 mg/kg at 5–10 mg/kg/min.
(valproate sodium injection, DEPACON® PI, 2001) now makes it possible to rapidly attain targeted plasma VPA concentrations when the clinical situation warrants such an intervention. Following an i.v. loading dose, maintenance regimens can be employed using either the i.v. formulation or one of several oral formulations, each of which displays different absorption characteristics. Sustaining VPA concentrations following loading doses is a function of both the formulation and patient characteristics (i.e. age, weight, metabolic enzyme induction status). After an i.v. loading dose of valproate sodium, selection of the proper dose and dosing interval (either
0920-1211/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0920-1211(02)00252-8
30
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
oral or i.v.) is important to ensure that plasma VPA concentrations remain within the therapeutic range. It is now possible to draw upon the pharmacokinetic information that has been collected regarding rapid infusion of valproate sodium (companion paper by Cloyd et al., 2003) in order to devise rational dosing regimens of valproate for maintenance therapy following these loading infusions. The oral formulations of VPA (and related salts) include a delayed-release enteric-coated tablet of divalproex sodium (DEPAKOTE® Tablets PI, 2001) and an extended-release tablet of divalproex sodium (DEPAKOTE® ER PI, 2001). Following administration of enteric-coated divalproex sodium, there is a delay of 1–2 h before drug is released, after which Cmax occurs within an hour (Carrigan et al., 1990; DEPAKOTE® Tablets PI, 2001; Fischer et al., 1988). Absorption of the extended-release divalproex sodium tablet begins shortly after administration and continues in a slow, sustained manner, with Cmax occurring in 7–14 h (DEPAKOTE® ER PI, 2001). Patient characteristics that are known to affect VPA pharmacokinetics must also be considered when designing a maintenance regimen. Young children have a significantly higher VPA clearance and shorter half-life than older children and adults (Chiba et al., 1985; Cloyd et al., 1993). Enzyme-inducing comedication also increases VPA clearance and shortens the elimination half-life. The present simulations were designed to evaluate the impact of various loading and maintenance dosage combinations on VPA concentration–time profiles for typical induced and uninduced adults and children. The simulations provide the clinician with dosing strategies for a variety of clinical situations requiring rapid infusion of i.v. valproate, followed by initiation of maintenance therapy.
relationship between total and unbound concentrations can be characterized using a 2-binding site model as described by Eqs. (1) and (2) (Gibaldi and Perrier, 1982): N1 K1 CU ALB N2 K2 CU ALB CB = + (1) 1 + K 1 CU 1 + K 2 CU CTOTAL = CU + CB
(2)
where, N1 and N2 represent the number of binding sites per class of binding site; K1 and K2 are the binding affinity constants for the two binding sites, CU and CB are the unbound and bound VPA concentrations, respectively, and ALB is the albumin concentration. For the simulations, a one-compartment model was used to describe the concentration–time profile of unbound plasma VPA concentrations. For all simulations, the unbound VPA concentrations were initially simulated using unbound clearance and volume of distribution values. Subsequently, total VPA concentrations were simulated for each unbound concentration using Eqs. (1) and (2). For the simulations, N1 , K1 , N2 and K2 were assumed to be 1.54 ± 0.108, 11.9 ± 1.99, 0.194 ± 0.0783, and 164 ± 141 mM−1 , respectively (companion paper by Cloyd et al., 2003). The simulations were performed using ADAPT II (D’Argenio and Schumitzky, 1979). ADAPT II is a regression program that can accommodate linear and nonlinear models with multiple inputs and outputs. The pharmacokinetic model can be defined using differential equations or in the integrated form for single dose and, uniform and nonuniform multiple dosing regimens. The program employs a variable-step, variable-order integration routine to solve model differential equations, and Nelder–Mead simplex procedure to determine the parameter values which minimize a weighted least squares criterion. 2.2. VPA pharmacokinetic parameters
2. Methods 2.1. Pharmacokinetic model VPA exhibits nonlinear protein binding (Levy et al., 2002). As a result, total VPA concentrations increase less than proportionally with increasing dose; however, unbound concentrations of VPA increase proportionally with dose (Levy et al., 2002). The nonlinear
VPA pharmacokinetic parameter values from children and adults with epilepsy and some healthy adult volunteers were used for the simulations (Cloyd et al., 1993; DEPAKOTE® Tablets PI, 2001). VPA pharmacokinetic parameters may differ depending on the age of the patient; in particular, clearance per kilogram body weight is relatively large in young children but decreases with maturation, and adult values are reached at approximately age 10–16 years (Chiba
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
et al., 1985; Cloyd et al., 1993; DEPAKOTE® Tablets PI, 2001). For the purpose of the simulations, the typical child was represented by a 6 year old. The apparent volume of distribution of unbound VPA was assumed to be the same in the induced and uninduced populations, with children (1.8 l/kg) having larger values than adults (1.3 l/kg). For the simulations, the clearance of unbound VPA for an uninduced child and adult were assumed to be 130 and 65 ml/h/kg, and for the induced child and adult were assumed to be 220 and 110 ml/h/kg, respectively. A lag time of 2 h (Fischer et al., 1988) and a firstorder absorption rate constant of 0.1 h−1 were used for simulations of all delayed-release divalproex sodium regimens. For extended-release divalproex sodium regimen simulations, 90% (absolute bioavailability) of the extended-release dose was assumed to be delivered at a constant zero-order rate over 18 h. The peak-to-trough fluctuation of the plasma VPA concentration–time profiles generated from the simulations was quantified using the pharmacokinetic parameter, degree of fluctuation (DFL). DFL was calculated as (Cmax −Cmin )/Cave ; Cave was calculated as the ratio of the area under the plasma concentration–time curve during a dosing interval to the dosing interval. 2.3. Simulations Three sets of simulations were performed using the aforementioned pharmacokinetic model and parame-
31
Table 1 Maintenance doses used for simulations Age
Formulation Intravenousa (mg/kg q6h)
Uninduced-optimal regimen Child (6 years) 7.5 Adult 3.5 Induced-optimal regimen Child (6 years) 15 Adult 7
DR (mg q12h)
ER (mg qd)
375 500
NS 1000
625 1000
NS 2000
DR, delayed-release; ER, extended-release; NS, no simulations were performed. a Intravenous doses were administered as 5-min infusions.
ters. The first set of simulations was performed to illustrate the effect of nonlinear protein binding on total and unbound VPA concentrations. These simulations of unbound and total plasma VPA concentrations were performed for a typical uninduced adult. For these simulations, loading doses of 5, 15, 30, and 45 mg/kg of i.v. valproate sodium were administered over 5 min. The second set of simulations explored the effect of age, concomitant enzyme-inducing medications, and valproate formulation on i.v. (valproate sodium injection) loading, and i.v. and oral (enteric-coated delayed-release divalproex sodium or extended-release divalproex sodium) maintenance regimens. Plasma VPA concentration–time profiles were simulated for clinical scenarios of a typical child (6 years and 22 kg)
Fig. 1. Maximum plasma total and unbound VPA concentrations achieved following 5-min loading infusions of 5, 15, 30, and 45 mg/kg.
32
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
and adult (70 kg), with and without an enzyme-inducing comedication. Intravenous loading (15 mg/kg) and maintenance doses were administered over 5 min. The i.v. and delayed-release divalproex sodium maintenance regimens were initiated 6 and 2 h after the start of the loading infusion, respectively. Extended-release divalproex sodium maintenance regimens were initiated simultaneously with the loading infusion. The maintenance dosing regimens used in these simulations (Table 1) were selected such that an average total VPA concentration of 75 mg/l would be maintained
over a dosing interval, with minimum and maximum total VPA concentrations being maintained within the therapeutic range of 50–100 mg/l. For delayed-release and extended-release divalproex sodium, the calculated theoretical doses that would result in average total VPA concentration of approximately 75 mg/l were rounded to the closest available dose strengths (DEPAKOTE® ER PI, 2001; DEPAKOTE® Tablets PI, 2001). A third set of simulations was performed to examine the effect of withdrawal of an enzyme-inducing
Fig. 2. Simulated total (top panels) and unbound (bottom panels) VPA concentrations for a typical uninduced and induced child (left panels) and adult (right panels) receiving a loading infusion followed by various i.v. maintenance regimens. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance i.v. dosing regimens for the typical uninduced child (7.5 mg/kg q6h) and adult (3.5 mg/kg q6h) are shown as a continuous line. The dashed and dotted lines represent the 7.5 mg/kg (child) and 3.5 mg/kg (adult) q6h regimens, which are suboptimal, in the induced population. Increasing the dose to 15 mg/kg q6h for a typical child and 7 mg/kg q6h for a typical adult can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced population (dotted line). The horizontal dashed lines represent the accepted minimum effective plasma total VPA concentration of 50 mg/l (top panels) and corresponding plasma unbound VPA concentrations of 5 mg/l (assuming a plasma protein binding of 90% at a total VPA concentration of 50 mg/l; bottom panel).
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
comedication on plasma VPA concentration–time profiles. Valproate dose-tapering regimens were also simulated to illustrate safe withdrawal of enzymeinducing comedication. The simulations were performed only for adults undergoing de-induction following withdrawal of enzyme-inducing comedications. For these simulations, the typical induced adult patient was assumed to be at steady state with respect to his/her plasma VPA concentration prior to withdrawal of the enzyme-inducing comedications. The half-life for de-induction of liver enzymes following withdrawal of a concomitant enzyme inducer was assumed to be 3 days (Mikati et al., 1989).
33
3. Results The results from the first set of simulations, which illustrate the effect of nonlinear protein binding on total and unbound VPA concentrations, are presented in Fig. 1. This figure shows the relationship between maximum total and unbound concentrations attained following various loading doses. Because of nonlinear protein binding, total VPA concentrations increase less than proportionally with dose, while unbound VPA concentrations increase linearly with dose. The results of the second set of simulations, which explored the combinations of an i.v. loading dose
Fig. 3. Simulated total (top panels) and unbound (bottom panels) VPA concentrations for a typical uninduced and induced child (left panels) and adult (right panels) receiving a loading infusion of valproate sodium, followed by various delayed-release divalproex sodium regimens. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance delayed-release dosing regimens for the typical uninduced child (375 mg q12h) and adult (500 mg q12h) are shown as continuous lines. The dashed and dotted lines represent the 375 mg (child) and 500 mg (adult) q12h regimens, which are suboptimal, in the induced population. Increasing the dose to 625 mg q12h for a typical child and 1000 mg q12h for a typical adult can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced population (dotted line). The horizontal dashed lines are same as in Fig. 2.
34
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
followed by various maintenance regimens are presented in Figs. 2–4. A 5-min 15 mg/kg loading dose resulted in total and free plasma VPA concentrations of ∼65 and 7.5 mg/l in children, and ∼80 and 11 mg/l in adults, 1 h after the infusion; induction status had little effect. The optimal maintenance regimens of the i.v., delayed-release, and extended-release valproate products used in the simulations resulted in plasma total VPA concentrations that were within 50–100 mg/l and corresponding unbound VPA concentrations that
Fig. 4. Simulated total (top panel) and unbound (bottom panel) VPA concentrations for a typical uninduced and induced adult receiving a loading infusion of valproate sodium and various extended-release divalproex maintenance regimens initiated simultaneously. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance extended-release dosing regimen for the typical uninduced adult (1000 mg qd) is shown as a continuous line. The dashed and dotted line represents the 1000 mg qd regimen, which is suboptimal, in the induced adult. Increasing the dose to 2000 mg qd can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced adult population (dotted line). The horizontal dashed lines are same as in Fig. 2.
were within 5–15 mg/l. For uninduced children and adults, 7.5 and 3.5 mg/kg q6h i.v. valproate sodium, initiated 6 h after loading dose maintains therapeutic plasma VPA concentrations (Fig. 2). The rapid decline of plasma VPA concentrations following an i.v. loading dose in combination with the delayed initial absorption of drug from delayed-release divalproex sodium tablets warrants beginning q12h oral maintenance regimens of delayed-release divalproex sodium within 2 h of a loading dose in the uninduced population (Fig. 3). Plasma VPA concentrations can be sustained in the therapeutic range using once-daily maintenance regimens of extended-release divalproex sodium tablets if initiated concurrently with i.v. loading dose in the uninduced population (Fig. 4). A two-fold higher i.v. and oral maintenance regimen dose may be required in induced patients (Figs. 2–4). The DFL for the simulated concentration–time profiles from Figs. 2–4 are presented in Fig. 5. Several patterns are apparent from Fig. 5. First, the DFL was higher for the unbound concentrations compared with the total VPA concentrations. Second, the DFL for both unbound and total VPA concentrations was higher in children compared with adults. Third, the DFL for both unbound and total VPA concentrations was higher in the induced population compared with the uninduced population. Fourth, within the induced population, increasing the dose for the optimal regimen compared with the suboptimal regimen resulted in no change in DFL of unbound concentrations but decreased DFL of total VPA concentrations. Fifth, the DFL of the extended-release formulation was lower than that of the delayed-release formulation. The third set of simulations explored the effect of withdrawal of a concomitant enzyme-inducing agent on plasma VPA concentrations for patients at steady state (Fig. 6). Maintenance of therapeutic VPA concentrations required tapering of the divalproex sodium doses in three stages for the delayed-release product but only two stages for the extended-release product. The divalproex sodium doses were tapered 3, 6, and 9 days after withdrawal of the enzyme-inducing comedications for the optimal regimen of the delayed-release product. For the optimal regimen of the extended-release product, the divalproex sodium doses were tapered 3 and 6 days after withdrawal of the enzyme-inducing comedications.
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
35
Fig. 5. DFL of plasma total and unbound VPA concentrations calculated for simulated concentration–time profiles presented in Figs. 2–4.
4. Discussion Valproate pharmacokinetics have been extensively reviewed (Levy et al., 2002). However, various combinations of a loading dose and maintenance regimens have not, to date, been systematically explored. With the demonstration of the safety of loading doses of i.v. valproate (companion paper by Cloyd et al., 2003), and knowledge of the pharmacokinetics of various older and newer (e.g. extended-release) formulations of valproate in diverse populations, it is now possible to design rational and optimal dosing regimens for different combinations of valproate formulations in different populations.
Fig. 2 presents the dosing regimens that would optimally produce average total and free VPA concentrations of 75 and 10 mg/l, respectively, in the uninduced child and adult. In the induced population, these dosing regimens result in VPA concentrations that are approximately one-half of the VPA concentrations achieved in uninduced patients because of increased VPA clearance. Doses that are approximately double the values appropriate for the uninduced population produce VPA concentrations in the induced child and adult comparable to those of the uninduced population. Similarly, increasing the dose of delayed-release divalproex sodium (Fig. 3) and extended-release divalproex sodium (Fig. 4) in induced children and adults
36
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
Fig. 6. Simulated total (top panel) and unbound (bottom panel) VPA concentrations for a typical induced adult undergoing de-induction following withdrawal of an enzyme-inducing agent. The left panels show VPA concentrations in patients on the delayed-release product, and the right panels show VPA concentrations in patients on the extended-release product. Negative time indicates steady-state VPA levels prior to withdrawal of the enzyme inducer, and time zero indicates withdrawal of enzyme-inducing comedications. The arrows indicate reductions in the valproate dose. The dotted line represents the expected rise in plasma VPA levels on withdrawal of the enzyme inducer if no adjustment is made to the valproate dosage. The continuous line represents the optimal regimen for a typical adult undergoing de-induction. The horizontal dashed lines are same as in Fig. 2.
result in mean plasma VPA concentrations that are comparable to those of the corresponding uninduced populations. However, the DFL of the plasma VPA concentrations in induced patients is much higher than in uninduced patients (Figs. 2–5) because of a more rapid clearance and the resulting shorter elimination half-life in the induced population. As might be expected for an extended-release product that delivers drug at a constant rate over a prolonged time period following ingestion, the DFL of the plasma concentrations is significantly lower than that achieved with
the i.v. and delayed-release products (Figs. 4 and 5). Therefore, with once-a-day dosing, it is possible to maintain plasma total VPA concentrations within the 50–100 mg/l range, and unbound VPA concentrations within the 5–15 mg/l range, in both induced and uninduced adults. Fig. 5 illustrates the following facts about maintenance regimens: (i) higher fluctuations of plasma concentrations can be expected for the induced population and for children; (ii) nonlinear protein binding of VPA attenuates fluctuations in total, but not unbound VPA
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
concentrations; and (iii) the extended-release product achieves a more desirable concentration–time profile (i.e. lower DFL) with once-a-day dosing than other valproate products with more frequent dosing regimens. The fluctuation in plasma VPA concentrations during a dosing interval (quantified as the DFL) is a function of the elimination half-life of the drug and its formulation. The patient population (i.e. induced versus uninduced and children versus adults) dictates the elimination clearance of VPA and therefore, the elimination half-life. Consequently, selection of appropriate maintenance regimens should consider the formulation to be used and the patient population of interest. There are several interesting features of the simulated concentration–time profiles (from Figs. 2–4) that can be gleaned from the DFL values presented in Fig. 5. First, the DFL of unbound concentrations is higher than that of total VPA concentrations because nonlinear protein binding of VPA attenuates the fluctuations of the total concentrations during a dosing interval. Second, the DFL for both unbound and total VPA concentrations is higher in children than in adults. As children clear valproate more rapidly than adults (i.e. they have shorter VPA half-lives), they have higher DFL values compared with adults. Third, the DFL for both unbound and total VPA concentrations is higher in the induced population than the uninduced population. Again, this is because higher clearance in the induced population results in shorter half-lives in the induced population compared to the uninduced population. Fourth, within the induced population, increasing the dose for the optimal regimen (compared with the suboptimal regimen) does not change the DFL of unbound VPA concentrations since the kinetics of unbound VPA are linear, but lowers DFL of total VPA concentrations which exhibits nonlinear pharmacokinetics. Again this is due to attenuation of fluctuations of total, but not unbound, VPA concentrations resulting from nonlinear protein binding. Finally, theoretical considerations indicate that when the total daily dose is administered as a divided dose multiple times per day, a lower DFL can be expected than with once-a-day administration. The extended-release product had a lower DFL than the delayed-release product despite the fact that the extended-release product was administered on a once-a-day regimen, compared with the twice-a-day
37
regimen of the delayed-release product. This suggests that a more desirable pharmacokinetic profile can be achieved with the extended-release product while at the same time allowing a once-a-day regimen, which may be more acceptable to the patient and may help increase compliance (Cramer et al., 1989). It should be noted that the lower DFL of the i.v. regimen is due to a more frequent q6h regimen that is feasible only in a hospital or nursing home setting. A q12h or a once-daily regimen of the i.v. product would result in much higher DFL values than that of the delayedrelease and the extended-release products. Fig. 6 illustrates the effect of withdrawal of a concomitant enzyme-inducing agent on plasma VPA concentrations. Since induction typically is associated with a doubling of clearance, de-induction would be expected to reduce clearance by half. As a result, plasma VPA levels would double without any change in valproate dosage. For an induced adult who was stabilized on a valproate dosage, this doubling of plasma concentrations typically occurs in approximately 2 weeks following complete de-induction. In order to maintain relatively constant and safe plasma VPA levels during de-induction, the valproate daily dose may have to be reduced, as illustrated in Fig. 6, or as appropriate, so that therapeutic levels are maintained without precipitating undesirable side effects. In the simulated concentration–time profiles presented in Fig. 6, the doses of the delayed-release divalproex sodium product were tapered three times at 3-day intervals, whereas for the extended-release product, this reduction in dosage was necessary only twice at 3-day intervals. This demonstrates that, with the extended-release product, an easier dose-tapering scheme is achievable for de-induction because of a concentration–time profile that has lower fluctuation. Relatively constant plasma VPA levels and dosing regimens can be achieved in about 1–2 weeks with the establishment of a new steady state in the de-induced state. Note that the fluctuations in plasma concentrations are lower in the de-induced state (e.g. days 12–14 in Fig. 6) than in the induced state (e.g. days −4–0 in Fig. 6) because of the decreased elimination clearance and longer half-life associated with de-induction. Simulations can serve as guides for designing clinical studies to test how well models predict observed values. Selection of the optimal VPA dosage regimen
38
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
must take into consideration differences in the various valproate formulations, pharmacokinetics, and the patient population. Loading doses using the i.v. formulation can be individualized for each clinical situation. The variability in distribution volume is relatively small and is primarily dependent on body size. In contrast, selection of an optimal maintenance regimen must take into consideration the valproate formulation and the patient age and induction status. Many patients receiving enzyme-inducing drugs, such as carbamazepine, phenobarbital, and phenytoin, realize an improvement in their condition when valproate is added. In such a circumstance, the clinician may determine that the enzyme-inducing drug can be discontinued, resulting in de-induction of drug-metabolizing enzymes and an increase in VPA concentrations. The simulations presented here indicate that reducing the valproate dosage by 40–50% while the inducing drug is withdrawn will maintain VPA concentrations within the therapeutic range. These adjustments reduce the risk of VPA concentration-dependent adverse effects and simplify the dosing regimen. Valproate has emerged as an important therapy in epilepsy, migraine headaches, and bipolar disorder. Patients with these conditions will respond earlier and sustain that response when desired VPA concentrations are rapidly attained and maintained. The use of individualized loading and maintenance dosage regimens of valproate sodium such as the ones presented in this report enables the clinician to more rationally design dosing regimens that are likely to produce optimal benefit while minimizing adverse effects. Acknowledgements This study was supported by Abbott Laboratories. Cloyd’s contributions to the manuscript were supported, in part, by a grant from the National Institutes of Health: NIH NINDS P50-NS16308.
References Carrigan, P.J., Brinker, D.R., Cavanaugh, J.H., Lamm, J.E., Cloyd, J.C., 1990. Absorption characteristics of a new valproate formulation: divalproex sodium-coated particles in capsules (Depakote Sprinkle). J. Clin. Pharmacol. 30, 743–747. Chiba, K., Suganuma, T., Ishizaki, T., Iriki, T., Shirai, Y., Naitoh, H., Hori, M., 1985. Comparison of steady-state pharmacokinetics of valproic acid in children between monotherapy and multiple antiepileptic drug treatment. J. Pediatr. 106, 653– 658. Cloyd, J.C., Dutta, S., Cao, G., Walch, J.K., Collins, S.D., Granneman, G.R., The Depacon Study Group, 2003. Valproate unbound fraction and distribution volume following rapid infusions in patients with epilepsy. Epilepsy Res. 53, 19–27. Cloyd, J.C., Fischer, J.H., Kriel, R.L., Kraus, D.M., 1993. Valproic acid pharmacokinetics in children. IV. Effects of age and antiepileptic drugs on protein binding and intrinsic clearance. Clin. Pharmacol. Ther. 53, 22–29. Cramer, J.A., Mattson, R.H., Prevey, M.L., Scheyer, R.D., Ouellette, V.L., 1989. How often is medication taken as prescribed? A novel assessment technique. JAMA 261, 3273–3277. D’Argenio, D.Z., Schumitzky, A., 1979. A program package for simulation and parameter estimation in pharmacokinetic systems. Comput. Programs Biomed. 9, 115–134. DEPACON® PI, 2001. Valproate Sodium Injection. Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp 417–422. DEPAKOTE® ER PI, 2001. Divalproex Sodium Extended-Release Tablets, Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp. 3470–3474. DEPAKOTE® Tablets PI, 2001. Divalproex Sodium DelayedRelease Tablets, Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp. 431–437. Fischer, J.H., Barr, A.N., Paloucek, F.P., Dorociak, J.V., Spunt, A.L., 1988. Effect of food on the serum concentration profile of enteric-coated valproic acid. Neurology 38, 1319–1322. Gibaldi, M., Perrier, D., 1982. Pharmacokinetics. Marcel Dekker, New York. Levy, R.H., Mattson, R.H., Meldrum, B.S., Perucca, E., 2002. Antiepileptic Drugs, 5th ed. Raven Press, New York. Mikati, M.A., Browne, T.R., Collins, J.F., 1989. Time course of carbamazepine autoinduction. The VA Cooperative Study No. 118 Group. Neurology 39, 592–594.
Oral/intravenous maintenance dosing of valproate following intravenous loading: a simulation Sandeep Dutta a,∗ , James C. Cloyd b , G. Richard Granneman a , Stephen D. Collins a,1 a
Abbott Laboratories, Department R4PK, Bldg. AP13A-3, 100 Abbott Park Road, Abbott Park, IL 60064-6104, USA b University of Minnesota, College of Pharmacy, Minneapolis, MN, USA Received 9 May 2002; received in revised form 2 October 2002; accepted 11 October 2002
Abstract Valproic acid (VPA) has a narrow therapeutic range (50–100 mg/l) and exhibits nonlinear protein binding. Additionally, VPA pharmacokinetics are dependent on age, induction status, and formulation; so titration and dosing vary between individuals. The aim of these simulations was to determine optimal intravenous (i.v.) loading dose, and i.v. and oral VPA maintenance regimens. A 5-min 15 mg/kg loading dose resulted in total and free plasma VPA concentrations of ∼65 and 7.5 mg/l in children, and ∼80 and 11 mg/l in adults, 1 h after the infusion; induction status had little effect. For uninduced children and adults, 7.5 and 3.5 mg/kg q6h i.v. valproate sodium, initiated 6 h after loading dose maintains therapeutic plasma VPA concentrations. The rapid decline of plasma VPA concentrations following an i.v. loading dose in combination with the delayed initial absorption of drug from delayedrelease divalproex sodium tablets warrant beginning q12h oral maintenance regimens of delayed-release divalproex sodium within 2 h of a loading dose in the uninduced population. Plasma VPA concentrations can be sustained in the therapeutic range using once-daily maintenance regimens of extended-release divalproex sodium tablets if initiated concurrently with i.v. loading dose in the uninduced population. A two-fold higher i.v. and oral maintenance regimen dose may be required in induced patients. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Divalproex; Plasma; Pharmacokinetics; Concentration
1. Introduction Valproic acid (VPA) and related salt formulations are widely used in a variety of conditions, including seizures, mania associated with bipolar disorder, as well as for prophylaxis of migraine headaches. The recent approval2 of an intravenous (i.v.) formulation ∗ Corresponding author. Tel.: +1-847-937-8502; fax: +1-847-938-5193. E-mail address: [email protected] (S. Dutta). 1 Present address: RW Johnson PRI, Raritan, NJ, USA. 2 DEPACON is now approved for infusions up to 15 mg/kg at 5–10 mg/kg/min.
(valproate sodium injection, DEPACON® PI, 2001) now makes it possible to rapidly attain targeted plasma VPA concentrations when the clinical situation warrants such an intervention. Following an i.v. loading dose, maintenance regimens can be employed using either the i.v. formulation or one of several oral formulations, each of which displays different absorption characteristics. Sustaining VPA concentrations following loading doses is a function of both the formulation and patient characteristics (i.e. age, weight, metabolic enzyme induction status). After an i.v. loading dose of valproate sodium, selection of the proper dose and dosing interval (either
0920-1211/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0920-1211(02)00252-8
30
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
oral or i.v.) is important to ensure that plasma VPA concentrations remain within the therapeutic range. It is now possible to draw upon the pharmacokinetic information that has been collected regarding rapid infusion of valproate sodium (companion paper by Cloyd et al., 2003) in order to devise rational dosing regimens of valproate for maintenance therapy following these loading infusions. The oral formulations of VPA (and related salts) include a delayed-release enteric-coated tablet of divalproex sodium (DEPAKOTE® Tablets PI, 2001) and an extended-release tablet of divalproex sodium (DEPAKOTE® ER PI, 2001). Following administration of enteric-coated divalproex sodium, there is a delay of 1–2 h before drug is released, after which Cmax occurs within an hour (Carrigan et al., 1990; DEPAKOTE® Tablets PI, 2001; Fischer et al., 1988). Absorption of the extended-release divalproex sodium tablet begins shortly after administration and continues in a slow, sustained manner, with Cmax occurring in 7–14 h (DEPAKOTE® ER PI, 2001). Patient characteristics that are known to affect VPA pharmacokinetics must also be considered when designing a maintenance regimen. Young children have a significantly higher VPA clearance and shorter half-life than older children and adults (Chiba et al., 1985; Cloyd et al., 1993). Enzyme-inducing comedication also increases VPA clearance and shortens the elimination half-life. The present simulations were designed to evaluate the impact of various loading and maintenance dosage combinations on VPA concentration–time profiles for typical induced and uninduced adults and children. The simulations provide the clinician with dosing strategies for a variety of clinical situations requiring rapid infusion of i.v. valproate, followed by initiation of maintenance therapy.
relationship between total and unbound concentrations can be characterized using a 2-binding site model as described by Eqs. (1) and (2) (Gibaldi and Perrier, 1982): N1 K1 CU ALB N2 K2 CU ALB CB = + (1) 1 + K 1 CU 1 + K 2 CU CTOTAL = CU + CB
(2)
where, N1 and N2 represent the number of binding sites per class of binding site; K1 and K2 are the binding affinity constants for the two binding sites, CU and CB are the unbound and bound VPA concentrations, respectively, and ALB is the albumin concentration. For the simulations, a one-compartment model was used to describe the concentration–time profile of unbound plasma VPA concentrations. For all simulations, the unbound VPA concentrations were initially simulated using unbound clearance and volume of distribution values. Subsequently, total VPA concentrations were simulated for each unbound concentration using Eqs. (1) and (2). For the simulations, N1 , K1 , N2 and K2 were assumed to be 1.54 ± 0.108, 11.9 ± 1.99, 0.194 ± 0.0783, and 164 ± 141 mM−1 , respectively (companion paper by Cloyd et al., 2003). The simulations were performed using ADAPT II (D’Argenio and Schumitzky, 1979). ADAPT II is a regression program that can accommodate linear and nonlinear models with multiple inputs and outputs. The pharmacokinetic model can be defined using differential equations or in the integrated form for single dose and, uniform and nonuniform multiple dosing regimens. The program employs a variable-step, variable-order integration routine to solve model differential equations, and Nelder–Mead simplex procedure to determine the parameter values which minimize a weighted least squares criterion. 2.2. VPA pharmacokinetic parameters
2. Methods 2.1. Pharmacokinetic model VPA exhibits nonlinear protein binding (Levy et al., 2002). As a result, total VPA concentrations increase less than proportionally with increasing dose; however, unbound concentrations of VPA increase proportionally with dose (Levy et al., 2002). The nonlinear
VPA pharmacokinetic parameter values from children and adults with epilepsy and some healthy adult volunteers were used for the simulations (Cloyd et al., 1993; DEPAKOTE® Tablets PI, 2001). VPA pharmacokinetic parameters may differ depending on the age of the patient; in particular, clearance per kilogram body weight is relatively large in young children but decreases with maturation, and adult values are reached at approximately age 10–16 years (Chiba
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
et al., 1985; Cloyd et al., 1993; DEPAKOTE® Tablets PI, 2001). For the purpose of the simulations, the typical child was represented by a 6 year old. The apparent volume of distribution of unbound VPA was assumed to be the same in the induced and uninduced populations, with children (1.8 l/kg) having larger values than adults (1.3 l/kg). For the simulations, the clearance of unbound VPA for an uninduced child and adult were assumed to be 130 and 65 ml/h/kg, and for the induced child and adult were assumed to be 220 and 110 ml/h/kg, respectively. A lag time of 2 h (Fischer et al., 1988) and a firstorder absorption rate constant of 0.1 h−1 were used for simulations of all delayed-release divalproex sodium regimens. For extended-release divalproex sodium regimen simulations, 90% (absolute bioavailability) of the extended-release dose was assumed to be delivered at a constant zero-order rate over 18 h. The peak-to-trough fluctuation of the plasma VPA concentration–time profiles generated from the simulations was quantified using the pharmacokinetic parameter, degree of fluctuation (DFL). DFL was calculated as (Cmax −Cmin )/Cave ; Cave was calculated as the ratio of the area under the plasma concentration–time curve during a dosing interval to the dosing interval. 2.3. Simulations Three sets of simulations were performed using the aforementioned pharmacokinetic model and parame-
31
Table 1 Maintenance doses used for simulations Age
Formulation Intravenousa (mg/kg q6h)
Uninduced-optimal regimen Child (6 years) 7.5 Adult 3.5 Induced-optimal regimen Child (6 years) 15 Adult 7
DR (mg q12h)
ER (mg qd)
375 500
NS 1000
625 1000
NS 2000
DR, delayed-release; ER, extended-release; NS, no simulations were performed. a Intravenous doses were administered as 5-min infusions.
ters. The first set of simulations was performed to illustrate the effect of nonlinear protein binding on total and unbound VPA concentrations. These simulations of unbound and total plasma VPA concentrations were performed for a typical uninduced adult. For these simulations, loading doses of 5, 15, 30, and 45 mg/kg of i.v. valproate sodium were administered over 5 min. The second set of simulations explored the effect of age, concomitant enzyme-inducing medications, and valproate formulation on i.v. (valproate sodium injection) loading, and i.v. and oral (enteric-coated delayed-release divalproex sodium or extended-release divalproex sodium) maintenance regimens. Plasma VPA concentration–time profiles were simulated for clinical scenarios of a typical child (6 years and 22 kg)
Fig. 1. Maximum plasma total and unbound VPA concentrations achieved following 5-min loading infusions of 5, 15, 30, and 45 mg/kg.
32
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
and adult (70 kg), with and without an enzyme-inducing comedication. Intravenous loading (15 mg/kg) and maintenance doses were administered over 5 min. The i.v. and delayed-release divalproex sodium maintenance regimens were initiated 6 and 2 h after the start of the loading infusion, respectively. Extended-release divalproex sodium maintenance regimens were initiated simultaneously with the loading infusion. The maintenance dosing regimens used in these simulations (Table 1) were selected such that an average total VPA concentration of 75 mg/l would be maintained
over a dosing interval, with minimum and maximum total VPA concentrations being maintained within the therapeutic range of 50–100 mg/l. For delayed-release and extended-release divalproex sodium, the calculated theoretical doses that would result in average total VPA concentration of approximately 75 mg/l were rounded to the closest available dose strengths (DEPAKOTE® ER PI, 2001; DEPAKOTE® Tablets PI, 2001). A third set of simulations was performed to examine the effect of withdrawal of an enzyme-inducing
Fig. 2. Simulated total (top panels) and unbound (bottom panels) VPA concentrations for a typical uninduced and induced child (left panels) and adult (right panels) receiving a loading infusion followed by various i.v. maintenance regimens. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance i.v. dosing regimens for the typical uninduced child (7.5 mg/kg q6h) and adult (3.5 mg/kg q6h) are shown as a continuous line. The dashed and dotted lines represent the 7.5 mg/kg (child) and 3.5 mg/kg (adult) q6h regimens, which are suboptimal, in the induced population. Increasing the dose to 15 mg/kg q6h for a typical child and 7 mg/kg q6h for a typical adult can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced population (dotted line). The horizontal dashed lines represent the accepted minimum effective plasma total VPA concentration of 50 mg/l (top panels) and corresponding plasma unbound VPA concentrations of 5 mg/l (assuming a plasma protein binding of 90% at a total VPA concentration of 50 mg/l; bottom panel).
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
comedication on plasma VPA concentration–time profiles. Valproate dose-tapering regimens were also simulated to illustrate safe withdrawal of enzymeinducing comedication. The simulations were performed only for adults undergoing de-induction following withdrawal of enzyme-inducing comedications. For these simulations, the typical induced adult patient was assumed to be at steady state with respect to his/her plasma VPA concentration prior to withdrawal of the enzyme-inducing comedications. The half-life for de-induction of liver enzymes following withdrawal of a concomitant enzyme inducer was assumed to be 3 days (Mikati et al., 1989).
33
3. Results The results from the first set of simulations, which illustrate the effect of nonlinear protein binding on total and unbound VPA concentrations, are presented in Fig. 1. This figure shows the relationship between maximum total and unbound concentrations attained following various loading doses. Because of nonlinear protein binding, total VPA concentrations increase less than proportionally with dose, while unbound VPA concentrations increase linearly with dose. The results of the second set of simulations, which explored the combinations of an i.v. loading dose
Fig. 3. Simulated total (top panels) and unbound (bottom panels) VPA concentrations for a typical uninduced and induced child (left panels) and adult (right panels) receiving a loading infusion of valproate sodium, followed by various delayed-release divalproex sodium regimens. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance delayed-release dosing regimens for the typical uninduced child (375 mg q12h) and adult (500 mg q12h) are shown as continuous lines. The dashed and dotted lines represent the 375 mg (child) and 500 mg (adult) q12h regimens, which are suboptimal, in the induced population. Increasing the dose to 625 mg q12h for a typical child and 1000 mg q12h for a typical adult can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced population (dotted line). The horizontal dashed lines are same as in Fig. 2.
34
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
followed by various maintenance regimens are presented in Figs. 2–4. A 5-min 15 mg/kg loading dose resulted in total and free plasma VPA concentrations of ∼65 and 7.5 mg/l in children, and ∼80 and 11 mg/l in adults, 1 h after the infusion; induction status had little effect. The optimal maintenance regimens of the i.v., delayed-release, and extended-release valproate products used in the simulations resulted in plasma total VPA concentrations that were within 50–100 mg/l and corresponding unbound VPA concentrations that
Fig. 4. Simulated total (top panel) and unbound (bottom panel) VPA concentrations for a typical uninduced and induced adult receiving a loading infusion of valproate sodium and various extended-release divalproex maintenance regimens initiated simultaneously. The arrows indicate the 5-min loading infusions of 15 mg/kg. The optimal maintenance extended-release dosing regimen for the typical uninduced adult (1000 mg qd) is shown as a continuous line. The dashed and dotted line represents the 1000 mg qd regimen, which is suboptimal, in the induced adult. Increasing the dose to 2000 mg qd can produce plasma total VPA concentrations that are maintained in the 50–100 mg/l range in the induced adult population (dotted line). The horizontal dashed lines are same as in Fig. 2.
were within 5–15 mg/l. For uninduced children and adults, 7.5 and 3.5 mg/kg q6h i.v. valproate sodium, initiated 6 h after loading dose maintains therapeutic plasma VPA concentrations (Fig. 2). The rapid decline of plasma VPA concentrations following an i.v. loading dose in combination with the delayed initial absorption of drug from delayed-release divalproex sodium tablets warrants beginning q12h oral maintenance regimens of delayed-release divalproex sodium within 2 h of a loading dose in the uninduced population (Fig. 3). Plasma VPA concentrations can be sustained in the therapeutic range using once-daily maintenance regimens of extended-release divalproex sodium tablets if initiated concurrently with i.v. loading dose in the uninduced population (Fig. 4). A two-fold higher i.v. and oral maintenance regimen dose may be required in induced patients (Figs. 2–4). The DFL for the simulated concentration–time profiles from Figs. 2–4 are presented in Fig. 5. Several patterns are apparent from Fig. 5. First, the DFL was higher for the unbound concentrations compared with the total VPA concentrations. Second, the DFL for both unbound and total VPA concentrations was higher in children compared with adults. Third, the DFL for both unbound and total VPA concentrations was higher in the induced population compared with the uninduced population. Fourth, within the induced population, increasing the dose for the optimal regimen compared with the suboptimal regimen resulted in no change in DFL of unbound concentrations but decreased DFL of total VPA concentrations. Fifth, the DFL of the extended-release formulation was lower than that of the delayed-release formulation. The third set of simulations explored the effect of withdrawal of a concomitant enzyme-inducing agent on plasma VPA concentrations for patients at steady state (Fig. 6). Maintenance of therapeutic VPA concentrations required tapering of the divalproex sodium doses in three stages for the delayed-release product but only two stages for the extended-release product. The divalproex sodium doses were tapered 3, 6, and 9 days after withdrawal of the enzyme-inducing comedications for the optimal regimen of the delayed-release product. For the optimal regimen of the extended-release product, the divalproex sodium doses were tapered 3 and 6 days after withdrawal of the enzyme-inducing comedications.
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
35
Fig. 5. DFL of plasma total and unbound VPA concentrations calculated for simulated concentration–time profiles presented in Figs. 2–4.
4. Discussion Valproate pharmacokinetics have been extensively reviewed (Levy et al., 2002). However, various combinations of a loading dose and maintenance regimens have not, to date, been systematically explored. With the demonstration of the safety of loading doses of i.v. valproate (companion paper by Cloyd et al., 2003), and knowledge of the pharmacokinetics of various older and newer (e.g. extended-release) formulations of valproate in diverse populations, it is now possible to design rational and optimal dosing regimens for different combinations of valproate formulations in different populations.
Fig. 2 presents the dosing regimens that would optimally produce average total and free VPA concentrations of 75 and 10 mg/l, respectively, in the uninduced child and adult. In the induced population, these dosing regimens result in VPA concentrations that are approximately one-half of the VPA concentrations achieved in uninduced patients because of increased VPA clearance. Doses that are approximately double the values appropriate for the uninduced population produce VPA concentrations in the induced child and adult comparable to those of the uninduced population. Similarly, increasing the dose of delayed-release divalproex sodium (Fig. 3) and extended-release divalproex sodium (Fig. 4) in induced children and adults
36
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
Fig. 6. Simulated total (top panel) and unbound (bottom panel) VPA concentrations for a typical induced adult undergoing de-induction following withdrawal of an enzyme-inducing agent. The left panels show VPA concentrations in patients on the delayed-release product, and the right panels show VPA concentrations in patients on the extended-release product. Negative time indicates steady-state VPA levels prior to withdrawal of the enzyme inducer, and time zero indicates withdrawal of enzyme-inducing comedications. The arrows indicate reductions in the valproate dose. The dotted line represents the expected rise in plasma VPA levels on withdrawal of the enzyme inducer if no adjustment is made to the valproate dosage. The continuous line represents the optimal regimen for a typical adult undergoing de-induction. The horizontal dashed lines are same as in Fig. 2.
result in mean plasma VPA concentrations that are comparable to those of the corresponding uninduced populations. However, the DFL of the plasma VPA concentrations in induced patients is much higher than in uninduced patients (Figs. 2–5) because of a more rapid clearance and the resulting shorter elimination half-life in the induced population. As might be expected for an extended-release product that delivers drug at a constant rate over a prolonged time period following ingestion, the DFL of the plasma concentrations is significantly lower than that achieved with
the i.v. and delayed-release products (Figs. 4 and 5). Therefore, with once-a-day dosing, it is possible to maintain plasma total VPA concentrations within the 50–100 mg/l range, and unbound VPA concentrations within the 5–15 mg/l range, in both induced and uninduced adults. Fig. 5 illustrates the following facts about maintenance regimens: (i) higher fluctuations of plasma concentrations can be expected for the induced population and for children; (ii) nonlinear protein binding of VPA attenuates fluctuations in total, but not unbound VPA
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
concentrations; and (iii) the extended-release product achieves a more desirable concentration–time profile (i.e. lower DFL) with once-a-day dosing than other valproate products with more frequent dosing regimens. The fluctuation in plasma VPA concentrations during a dosing interval (quantified as the DFL) is a function of the elimination half-life of the drug and its formulation. The patient population (i.e. induced versus uninduced and children versus adults) dictates the elimination clearance of VPA and therefore, the elimination half-life. Consequently, selection of appropriate maintenance regimens should consider the formulation to be used and the patient population of interest. There are several interesting features of the simulated concentration–time profiles (from Figs. 2–4) that can be gleaned from the DFL values presented in Fig. 5. First, the DFL of unbound concentrations is higher than that of total VPA concentrations because nonlinear protein binding of VPA attenuates the fluctuations of the total concentrations during a dosing interval. Second, the DFL for both unbound and total VPA concentrations is higher in children than in adults. As children clear valproate more rapidly than adults (i.e. they have shorter VPA half-lives), they have higher DFL values compared with adults. Third, the DFL for both unbound and total VPA concentrations is higher in the induced population than the uninduced population. Again, this is because higher clearance in the induced population results in shorter half-lives in the induced population compared to the uninduced population. Fourth, within the induced population, increasing the dose for the optimal regimen (compared with the suboptimal regimen) does not change the DFL of unbound VPA concentrations since the kinetics of unbound VPA are linear, but lowers DFL of total VPA concentrations which exhibits nonlinear pharmacokinetics. Again this is due to attenuation of fluctuations of total, but not unbound, VPA concentrations resulting from nonlinear protein binding. Finally, theoretical considerations indicate that when the total daily dose is administered as a divided dose multiple times per day, a lower DFL can be expected than with once-a-day administration. The extended-release product had a lower DFL than the delayed-release product despite the fact that the extended-release product was administered on a once-a-day regimen, compared with the twice-a-day
37
regimen of the delayed-release product. This suggests that a more desirable pharmacokinetic profile can be achieved with the extended-release product while at the same time allowing a once-a-day regimen, which may be more acceptable to the patient and may help increase compliance (Cramer et al., 1989). It should be noted that the lower DFL of the i.v. regimen is due to a more frequent q6h regimen that is feasible only in a hospital or nursing home setting. A q12h or a once-daily regimen of the i.v. product would result in much higher DFL values than that of the delayedrelease and the extended-release products. Fig. 6 illustrates the effect of withdrawal of a concomitant enzyme-inducing agent on plasma VPA concentrations. Since induction typically is associated with a doubling of clearance, de-induction would be expected to reduce clearance by half. As a result, plasma VPA levels would double without any change in valproate dosage. For an induced adult who was stabilized on a valproate dosage, this doubling of plasma concentrations typically occurs in approximately 2 weeks following complete de-induction. In order to maintain relatively constant and safe plasma VPA levels during de-induction, the valproate daily dose may have to be reduced, as illustrated in Fig. 6, or as appropriate, so that therapeutic levels are maintained without precipitating undesirable side effects. In the simulated concentration–time profiles presented in Fig. 6, the doses of the delayed-release divalproex sodium product were tapered three times at 3-day intervals, whereas for the extended-release product, this reduction in dosage was necessary only twice at 3-day intervals. This demonstrates that, with the extended-release product, an easier dose-tapering scheme is achievable for de-induction because of a concentration–time profile that has lower fluctuation. Relatively constant plasma VPA levels and dosing regimens can be achieved in about 1–2 weeks with the establishment of a new steady state in the de-induced state. Note that the fluctuations in plasma concentrations are lower in the de-induced state (e.g. days 12–14 in Fig. 6) than in the induced state (e.g. days −4–0 in Fig. 6) because of the decreased elimination clearance and longer half-life associated with de-induction. Simulations can serve as guides for designing clinical studies to test how well models predict observed values. Selection of the optimal VPA dosage regimen
38
S. Dutta et al. / Epilepsy Research 53 (2003) 29–38
must take into consideration differences in the various valproate formulations, pharmacokinetics, and the patient population. Loading doses using the i.v. formulation can be individualized for each clinical situation. The variability in distribution volume is relatively small and is primarily dependent on body size. In contrast, selection of an optimal maintenance regimen must take into consideration the valproate formulation and the patient age and induction status. Many patients receiving enzyme-inducing drugs, such as carbamazepine, phenobarbital, and phenytoin, realize an improvement in their condition when valproate is added. In such a circumstance, the clinician may determine that the enzyme-inducing drug can be discontinued, resulting in de-induction of drug-metabolizing enzymes and an increase in VPA concentrations. The simulations presented here indicate that reducing the valproate dosage by 40–50% while the inducing drug is withdrawn will maintain VPA concentrations within the therapeutic range. These adjustments reduce the risk of VPA concentration-dependent adverse effects and simplify the dosing regimen. Valproate has emerged as an important therapy in epilepsy, migraine headaches, and bipolar disorder. Patients with these conditions will respond earlier and sustain that response when desired VPA concentrations are rapidly attained and maintained. The use of individualized loading and maintenance dosage regimens of valproate sodium such as the ones presented in this report enables the clinician to more rationally design dosing regimens that are likely to produce optimal benefit while minimizing adverse effects. Acknowledgements This study was supported by Abbott Laboratories. Cloyd’s contributions to the manuscript were supported, in part, by a grant from the National Institutes of Health: NIH NINDS P50-NS16308.
References Carrigan, P.J., Brinker, D.R., Cavanaugh, J.H., Lamm, J.E., Cloyd, J.C., 1990. Absorption characteristics of a new valproate formulation: divalproex sodium-coated particles in capsules (Depakote Sprinkle). J. Clin. Pharmacol. 30, 743–747. Chiba, K., Suganuma, T., Ishizaki, T., Iriki, T., Shirai, Y., Naitoh, H., Hori, M., 1985. Comparison of steady-state pharmacokinetics of valproic acid in children between monotherapy and multiple antiepileptic drug treatment. J. Pediatr. 106, 653– 658. Cloyd, J.C., Dutta, S., Cao, G., Walch, J.K., Collins, S.D., Granneman, G.R., The Depacon Study Group, 2003. Valproate unbound fraction and distribution volume following rapid infusions in patients with epilepsy. Epilepsy Res. 53, 19–27. Cloyd, J.C., Fischer, J.H., Kriel, R.L., Kraus, D.M., 1993. Valproic acid pharmacokinetics in children. IV. Effects of age and antiepileptic drugs on protein binding and intrinsic clearance. Clin. Pharmacol. Ther. 53, 22–29. Cramer, J.A., Mattson, R.H., Prevey, M.L., Scheyer, R.D., Ouellette, V.L., 1989. How often is medication taken as prescribed? A novel assessment technique. JAMA 261, 3273–3277. D’Argenio, D.Z., Schumitzky, A., 1979. A program package for simulation and parameter estimation in pharmacokinetic systems. Comput. Programs Biomed. 9, 115–134. DEPACON® PI, 2001. Valproate Sodium Injection. Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp 417–422. DEPAKOTE® ER PI, 2001. Divalproex Sodium Extended-Release Tablets, Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp. 3470–3474. DEPAKOTE® Tablets PI, 2001. Divalproex Sodium DelayedRelease Tablets, Abbott Laboratories Inc., North Chicago, IL 60064, USA Physicians’ Desk Reference, 55th ed. Medical Economical Company, Inc., Montvale, NJ, pp. 431–437. Fischer, J.H., Barr, A.N., Paloucek, F.P., Dorociak, J.V., Spunt, A.L., 1988. Effect of food on the serum concentration profile of enteric-coated valproic acid. Neurology 38, 1319–1322. Gibaldi, M., Perrier, D., 1982. Pharmacokinetics. Marcel Dekker, New York. Levy, R.H., Mattson, R.H., Meldrum, B.S., Perucca, E., 2002. Antiepileptic Drugs, 5th ed. Raven Press, New York. Mikati, M.A., Browne, T.R., Collins, J.F., 1989. Time course of carbamazepine autoinduction. The VA Cooperative Study No. 118 Group. Neurology 39, 592–594.