What Is the Best Strategy for Converting from Twice-Daily Divalproex to a Once-Daily Divalproex ER Regimen?

Ronald C. Reed; Sandeep Dutta

Disclosures

Clin Drug Invest. 2004;24(9) 

In This Article

Methods

Four distinct divalproex q12h to once-daily divalproex ER conversion strategies for hypothetical adult patients were tested using population simulations for monotherapy (uninduced) and polytherapy patients (induced) on concomitant enzyme-inducing AEDs in order to examine the impact on plasma total VPA concentrations in the first 48 hours. These various conversion strategies were selected based on anecdotal observations that each has been attempted in clinical practice.

The four distinct divalproex q12h to once-daily divalproex ER formulation conversion strategies selected for study were as follows: immediate conversion 12 hours after the last divalproex dose; delayed conversion 24 hours after the last divalproex dose; stepwise conversion, i.e. half the divalproex ER daily dose in both the morning and evening for 1 day; and mixed conversion, i.e. half divalproex daily dose in the morning and half divalproex ER daily dose in the evening for 1 day, prior to converting to once-daily divalproex ER.

Uninduced Adult Patients

The uninduced hypothetical adult patients were converted from a baseline steady-state 625mg divalproex q12h regimen to (a) 1500mg divalproex ER once daily in the morning or evening, 12 hours after the last divalproex dose (immediate conversion); (b) 1500mg divalproex ER once daily in the morning or evening 24 hours after the last divalproex dose (delayed conversion); (c) 750mg divalproex ER in the morning and 750mg divalproex ER in the evening for 1 day, followed by 1500mg divalproex ER once daily (stepwise conversion); and (d) 625mg divalproex in the morning (i.e. one-half of the divalproex total daily dose in the morning) and 750mg divalproex ER in the evening (one-half of the intended total daily divalproex ER dose) for 1 day, followed by 1500mg divalproex ER once daily (mixed conversion).

Induced Adult Patients

Typical dose. The induced hypothetical adult patients (in whom higher divalproex doses are needed) were converted from a baseline steady-state 1250mg divalproex q12h regimen to (a) 3000mg divalproex ER once daily morning or evening, 12 hours after the last divalproex dose (immediate conversion); (b) 3000mg divalproex ER once daily morning or evening, 24 hours after the last divalproex dose (delayed conversion); (c) 1500mg divalproex ER in the morning and 1500mg divalproex ER in the evening for 1 day, followed by 3000mg divalproex ER once daily (stepwise conversion); and (d) 1250mg divalproex in the morning and 1500mg divalproex ER in the evening for 1 day, followed by 3000mg divalproex ER once daily (mixed conversion).

Higher dose. The induced hypothetical adult patients on a higher daily dose of divalproex (e.g. for high-risk patients with refractory epilepsy concomitantly taking multiple enzyme-inducing co-medications) were converted from a baseline steady-state 2000mg divalproex q12h regimen to (a) 4500mg divalproex ER once daily morning or evening, 12 hours after the last divalproex dose (immediate conversion); (b) 4500mg divalproex ER once daily morning or evening, 24 hours after the last divalproex dose (delayed conversion); (c) 2250mg divalproex ER in the morning and 2250mg divalproex ER in the evening for 1 day, followed by 4500mg divalproex ER once daily (stepwise conversion); and (d) 2000mg divalproex in the morning and 2250mg divalproex ER in the evening for 1 day, followed by 4500mg divalproex ER once daily (mixed conversion).

A previous report has discussed extensively the basis for selection of the pharmacokinetic model and the associated parameters.[19] Briefly, the population mean pharmacokinetic parameters used in the programming of our model for VPA[6,7,8] and the divalproex ER tablet[16,17,18] included the following: absolute bioavailability from the divalproex tablet of 100% and divalproex ER tablet of 89%; first-order absorption for divalproex (rate = 1.0h-1; lag time 2 hours); zero-order absorption from the divalproex ER tablet (total dose absorption duration 22 hours, with no absorption lag time); steady-state distribution volume of 1.3 L/kg; elimination half-life of 13.9 hours for uninduced and 8.2 hours for induced patients; and clearance of unbound VPA of 65 mL/h/kg for uninduced and 110 mL/h/kg for induced patients. Bodyweight was assumed to be 70kg.

Protein binding was assumed to be nonlinear. The nonlinear relationship between total and unbound plasma VPA concentrations was described using a 2-binding site model. The nonlinear protein-binding parameters,[20] i.e. the number of binding sites for two classes of binding sites (N1 and N2) and their binding association constants (K1 and K2), were N1 = 1.54, N2 = 0.194, K1 = 11.9 mmol-1, and K2 = 164 mmol-1. The protein (i.e. albumin) concentration was assumed to be 0.5279 mmol. All computer simulations incorporated 20% interpatient variability and 10% residual variability (sum of intrapatient, assay variability and all other unexplained sources of variability).

Conventional divalproex formulation doses commonly used in clinical practice were chosen for analysis. The total daily dose of divalproex ER was approximately 8-20% greater than the total daily divalproex dose, based on clinically derived evidence indicating that VPA exposure (area under the curve) is the same at the higher divalproex ER dose.[16,17] The bioavailability of divalproex ER was assumed to be similar when it was given as a daily dose either in the morning or the evening.[21] Plasma VPA concentration-time profiles were simulated for 1000 hypothetical adult patients (not paediatric or geriatric patients) for each scenario/conversion strategy using a one-compartment (rapid distribution assumed) population kinetic model with nonlinear protein binding (Pharsight® Clinical Trial Simulator).

For each stochastic simulation scenario, a normal distribution of each pharmacokinetic parameter (unbound clearance and volume of distribution, protein-binding parameters and albumin concentration) was generated with the above specified population mean values and 20% intersubject variability. Patients were generated by randomly selecting parameter values from these distributions. The concentrations generated by the randomly selected pharmacokinetic parameters (i.e. hypothetical patients) were then perturbed with 10% residual error to generate noisy ('real-world') concentration-time profiles for each virtual patient.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....