SEARCH

SEARCH BY CITATION

Keywords:

  • Levetiracetam;
  • Children;
  • Dosing recommendation;
  • Population pharmacokinetics;
  • Body weight

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Purpose: To develop a population pharmacokinetic model to evaluate the demographic and physiologic determinants of levetiracetam (LEV) pharmacokinetics (PK) and to suggest recommended doses of LEV in children.

Methods: LEV PK were investigated in a prospective open trial of LEV as adjunctive therapy using a population approach performed with NONMEM (Nonlinear Mixed Effects Model) on 170 LEV concentration–time records and covariate information from 44 children between 4 and 16 years of age. Possible associations between pharmacokinetic parameters and age, gender, body weight, creatinine clearance, and concomitant antiepileptic drugs (AEDs) were assessed. The final model was used to perform Monte Carlo simulations in order to identify the dosing regimens that should achieve the same nominal target concentration range as in adults.

Results: LEV PK were well described by a one-compartment model with first-order absorption and elimination. Both LEV apparent clearance and distribution volume were related to body weight, and no pharmacokinetic interaction was observed. Monte Carlo simulations showed that a 10mg/kg twice daily (b.i.d.) regimen provides a plasma concentration similar to that obtained in adults for the recommended 500 mg b.i.d. starting dose, and that a 20 mg/kg b.i.d. regimen would achieve the previously described 6–20 mg/L target range for the trough concentration.

Discussion: Our results support the use of a weight-based LEV dosing regimen and provide a basis for a recommended pediatric dosage regimen. The relationship between LEV plasma concentrations and clinical effect has not been evaluated fully and could differ between adults and children. Clinical studies should be able to validate these dosing recommendations.

Levetiracetam (LEV) is a newly developed antiepileptic drug (AED) that has been approved by the US Food and Drug Administration (FDA) and the European Medicines Agency for clinical use in adults since the year 2000. It has a unique mechanism of action different from that of classical AEDs, as it acts via the involvement of a novel binding site, the synaptic vesicle protein 2A (Lynch et al., 2004). In adults, adjunctive therapy with LEV was shown to be well tolerated and effective in controlling partial seizures (Cereghino et al., 2000; Shorvon et al., 2000). Similar results were shown in children between 4 and 16 years of age, as nearly half of the subjects on LEV had a more than 50% reduction in seizure frequency and 7% became seizure free with a LEV target dose of 60 mg/kg daily (Glauser et al., 2006).

After oral administration, LEV is completely absorbed and it is not bound to plasma proteins. Unlike other AEDs that are metabolized, LEV is primarily excreted unchanged by the kidney. Only one-third of LEV is metabolized into an inactive compound via a metabolic pathway involving enzymatic hydrolysis. LEV is not metabolized by the hepatic cytochrome P450 system (Patsalos, 2000).

The pharmacokinetics (PK) of LEV are well characterized in adults (Patsalos, 2000; Radtke, 2001; Pigeolet et al., 2007), but information in children is limited (Pellock et al., 2001; Fountain et al., 2007; Glauser et al., 2007; Snoeck et al., 2007), with some discrepancies among studies. On the one hand, three different LEV pharmacokinetic studies performed in either infants (ages ranging from 2.3 to 46.2 months) or children (ages ranging from 4 to 12 years) found similar weight-normalized clearances (approximately 1–1.5 ml/min/kg) and elimination half-lives (4–6 h) (Pellock et al., 2001; Fountain et al., 2007; Glauser et al., 2007). These results strongly suggested that both apparent clearance and distribution volume of LEV are directly proportional to body weight during childhood and that the pediatric LEV dosing regimen should be expressed in mg per kg and should furthermore be constant throughout a wide age range in order to provide similar plasma concentrations in pediatric patients of different ages, from a 2-month-old infant to a 12-year old child.

However, on the other hand, a population pharmacokinetic analysis performed in children from 3 months to 18 years revealed conflicting results, as the apparent clearance of LEV increased less rapidly than body weight, suggesting that the weight-normalized dose should decrease with increasing body weights (Snoeck et al., 2007).

The pharmacokinetic parameters of many drugs do not vary linearly with age or body weight in the pediatric population (Kearns et al., 2003; Perucca, 2005). In the case of LEV, children seem to have lower concentrations than adults for the same dose per kg body weight (Pellock et al., 2001; May et al., 2003), as the clearance of a single dose of LEV was shown to be about 30–40% higher in children than in adults (Pellock et al., 2001).

Considering all the results currently available, the relationship between the PK parameters of LEV and age and/or body weight, and, therefore, the need for adjusting the LEV dosing regimen with respect to age or body weight still appear to be under debate in some instances.

Population PK is the optimal methodology for investigating the relationships between the pharmacokinetic parameters of a given drug and the possibly relevant covariates. Furthermore, because a population pharmacokinetic model also takes into account the interindividual and intraindividual variabilities of pharmacokinetic parameters, a population approach makes possible the determination of the dosing regimen to achieve plasma concentrations within a given target range. Because of the lack of knowledge regarding the specific pharmacokinetic–pharmacodynamic relationship of LEV in children as a function of the epilepsy syndrome under investigation, an acceptable endpoint would be the achievement of the same LEV steady-state area under the curve (AUC) as in adults for the recommended adult doses. Another possibility could be the achievement of the same target range of trough concentrations (i.e., 6–20 mg/L) that was previously suggested in adults (Johannessen et al., 2003).

The purpose of the present study was to assess LEV PK in a pediatric population using a population approach to identify demographic and physiologic determinants of LEV concentration, including the effects of concomitant AEDs. The optimal pediatric dosing regimens were also defined to achieve plasma concentrations within a given target range.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Study design

This was an open-label, multicenter, prospective clinical study in children with epilepsy treated with LEV as adjunctive therapy for refractory epilepsy. The study involved the following phases: a 4-week baseline period, a 6-week titration period, and a 14-week evaluation period.

The study was approved by the local ethics committee Paris–Cochin and was carried out in accordance with the Declaration of Helsinki and guidelines for Good Clinical Practice. After comprehensive information on the clinical trial was provided, written informed consent was obtained before enrollment in the study from a parent or legal guardian for each patient and, when applicable, consent was obtained from the patient.

During the baseline period, patient eligibility was confirmed. During the titration period, LEV tablets were administered at dose of 10 mg/kg daily for 2 weeks, followed by 20 mg/kg daily for 2 weeks, and finally 40 mg/kg daily for 2 weeks. The LEV dose regimen could be adjusted in cases of inadequate seizure control or side effects.

Patients

Children with epilepsy aged 4–16 years were recruited by the local pediatrician at outpatient clinics in each investigation center if they satisfied the inclusion criteria and informed consent was obtained. In addition, patients had to be able to swallow the LEV medication without crushing the tablet.

They were eligible if they had a stable AED treatment (one or two drugs) considered as unsatisfactory by the investigator in terms of efficacy and/or adverse events, and if adjunctive treatment with LEV could be considered of potential benefit. Patients were required to be on a stable LEV dose treatment for at least one month. They could present with various types of epilepsy syndromes, including partial, generalized, and undetermined, whether partial or generalized epilepsy, such as infantile spasms, Dravet syndrome, Lennox-Gastaut syndrome, myoclonic-astatic epilepsy, or continuous spike-waves during sleep (ILAE Classification, 1989).

Subjects were not included if their monthly seizure frequencies varied significantly during the 3-month period before starting treatment with LEV, if they had any other progressive neurodegenerative disease, or if they had clinically significant laboratory abnormalities.

For each patient, the time between dosing and sampling, gender, age, body weight, serum creatinine, and comedication were recorded. Creatinine clearance was estimated with the Schwartz formula in children (Schwartz et al., 1976).

Sample collection

Patients were enrolled in a randomized protocol divided into three groups according to the sampling schemes scheduled on day 1, day 30, and day 180: on day 1 after the first administration 0.33 and 1.33 h (group A), or 0.5 and 1.5 h (group B), or 0.66 and 1.66 h (group C). On day 30, 3 h (group A), or 2 h (group B), or 4 h (group C) after the last LEV administration. On day 180, 2 h (group A), or 4 h (group B), or 3 h (group C) after the last LEV administration. Blood samples were collected before the morning dose of LEV at day 90. A maximum of five blood samples were drawn from each child. The volume of each blood sample was 1 ml.

Analytical method

The assay for LEV was performed by using high pressure liquid chromatography (HPLC) with ultraviolet (UV) detection. In summary, 50 μl of NaOH (5 M) and 5 ml of chloroform were added to 100 μl of plasma added with the internal standard. After 10 min of mixing, the supernatant was evaporated at 30°C under a stream of nitrogen. Dry residues were then reconstituted with 200 μl of ultrapure water, and 50 μl of this mixture was injected into the chromatographic system. The separation was performed on a Reverse Phase C18 AB Satisfaction Column (250 by 4.6 mm, 4.6 μm; Cluzeau, Sainte-Foy la Grande, France) at a flow rate of 0.5 ml/min. Detection was performed at 205 nm. The quantification limit of the method was 1 mg/L, with the interassay precision and bias being less than 8% in the calibration range of 1–30 mg/L.

Population Pharmacokinetic Modeling

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Method

Concentration–time data were analyzed by using of the first-order conditional estimation with interaction method of the Nonlinear Mixed Effects Model (NONMEM) program (version VI, 1.1, double precision, University of California, San Francisco, CA, U.S.A.). Several structural pharmacokinetic models, classical one- and two-compartment models with first- and zero-order absorption and an absorption lag-time, were evaluated. Several error models (i.e., proportional, exponential, and additive random effects models) were also investigated to describe interpatient and residual variabilities. Systematic testing for the influence of continuous covariates on the pharmacokinetic parameters was done by use of a generalized model, according to the following equation, by using, for example, apparent oral clearance (CL/F) and body weight (BW):

  • image

where TV(CL/F) was the typical value of the apparent clearance for a patient with the median covariate value, and θ was the influential factor for body weight. Other allometric models based on previous recommendations for analyzing pediatric data (Anderson et al., 1997), as well as a simple linear model (CL/F and V/F  =  θ x BW), were also investigated.

Binary covariates (gender, combined treatment) were investigated as follows:

  • image

where SEX was equal to 1 for male and to the influential factor for female.

Cotreatments were first integrated separately in the model and then were simultaneously investigated according to their possible effect on the LEV apparent oral clearance (inducers or inhibitors). The combination responsible for the greater objective function decrease was kept. The possible effect of a covariate on LEV bioavailability was investigated as described previously by estimation of the same influential factor on both CL/F and V/F.

The significance of a relationship between a pharmacokinetic parameter and a covariate was assessed by use of the chi-square test of the difference between the objective functions of the basic model (without the covariate) and the model with the covariate. A covariate was retained in the model if it produced a minimum decrease in the objective function of four units (p = 0.05, 1 degree of freedom) and if its effect was biologically plausible. An intermediate multivariate model that included all selected covariates was then obtained. A covariate was retained in the final multivariate model if its deletion from the intermediate model led to a 7-point increase in the objective function (p = 0.01, 1 degree of freedom). At each step, the goodness of fit was evaluated by use of a graph of the weighted residuals versus predicted concentrations (PRED) and versus the time after administration of the dose (time) or by use of a graph of the Normalized Prediction Errors (NPDE) versus time and PRED (Brendel et al., 2006). Individual Bayesian estimates of the pharmacokinetic parameters were used to calculate the individual Ctroughs.

Model validation

The accuracy and robustness of the final population model were assessed by a visual predictive check. The final population model parameters were used to perform 1,000 simulations of the database. The 2.5th and the 97.5th percentiles as well as the 50th (median) percentiles of simulated concentrations were plotted against observed concentrations.

Evaluation of levetiracetam dosing regimens

Monte Carlo simulations were used to estimate dose regimens that achieved LEV plasma steady-state AUCs similar to those reached in adults. Five different b.i.d. regimens were investigated (5 mg/kg, 10 mg/kg, 15 mg/kg, 20 mg/kg, and 30 mg/kg). The simulations (200 simulated subjects per body weight) were performed for body weights increasing by 1 kg from 20 to 60 kg, that is, for each dose regimen, 200 children weighing 20 kg, 200 children weighing 21 kg, 200 children weighing 22 kg, and so on until 200 children weighing 60 kg were simulated. The LEV starting dose allowing the achievement of the target trough concentration range of 6–20 mg/L was determined by the same methodology.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Patients demographics

A total of 170 samples obtained prospectively from 44 patients ranging in age from 4.6 to 16.6 years were available for pharmacokinetic evaluation. The population comprised 22 boys and 22 girls. The characteristics of the studied population are summarized in Table 1. In this population, 20.5% and 79.5% had one and two concomitant AED(s), respectively. The maximum LEV dose that could be used was 60 mg/kg daily. The most frequently used concomitant AEDs were valproic acid, lamotrigine, carbamazepine, and vigabatrin.

Table 1.   Baseline characteristics of 170 samples coming from 44 children with epilepsy treated by levetiracetam
 Number of samplesArithmetic mean ± SDMedian value (range)
  1. M, male; F, female; SD, standard deviation.

Gender (M/F)86/84  
Body weight (kg)17036.4 ± 13.233 (16–65)
Age (years)17010.7 ± 3.111 (4.6–16.6)
Creatinine clearance (ml/min/1.73m2)170170.6 ± 42.6169.0 (63.7–302.9)

Population pharmacokinetics

A classical one-compartment model with an absorption lag-time, a first-order absorption, and linear elimination (subroutines ADVAN2 and TRANS2) best described the data. The one-compartment model with zero-order absorption and the two-compartment model did not improve the fit. Interpatient variability and residual variability were described by a proportional error model. A significant covariance term was found between the etas (η) of CL/F and V/F; η is describing the deviation between the individual pharmacokinetic parameter and its typical population estimate. Interindividual variability of the absorption lag-time could not be estimated.

The use of body weight as a covariate was found to improve the fit. Although the different weight-based models provided relatively similar results, the general model with estimation of the exponents explaining the influence of BW on the variability of CL/F and V/F was chosen, since it provided the lowest values for the objective function and omega22); ω2 is describing the interindividual variability of the corresponding pharmacokinetic parameter (Table 2A). No significant interaction with the concomitantly administered AED was found. The effects of the different covariates tested on the objective function and the interindividual variability of the pharmacokinetic parameters are summarized in Table 2B.

Table 2A.   Comparison of the objective function for the different tested models
ModelObjective functioninline imageinline image
  1. CL/F, apparent oral clearance; V/F apparent volume of distribution; BW, body weight; θ, mean population estimate of the corresponding pharmacokinetic parameter; θcovariate, influential factor for the covariate; ω2, interindividual variability.

Base model5460.1670.162
Final model
inline image inline image4900.0590.0265
inline image inline image4940.06230.0281
inline image inline image4910.06010.0275
inline image inline image4930.06170.0289
Table 2B.   Effect of the tested covariates on the objective function
Tested covariatePharmacokinetic parameterΔFobj1Δη1ΔFobj2Δη2
  1. ΔFobj, observed change in the objective function induced by the corresponding covariate after its addition to the base model (ΔFobj1) or its deletion from the intermediate model (ΔFobj2); Δη, percent change in the interindividual variability of the corresponding pharmacokinetic parameter provided by the addition of the tested covariate in the base model (Δη1) or by its deletion from the intermediate model (Δη2); AED, antiepileptic drug; CL/F, apparent oral clearance; V/F apparent volume of distribution.

  2. [LEFT RIGHT ARROW] No significant change in the objective function.

Body weightCL/F−9.657−15.8%+12.935+6.0%
V/F−22.131−24.1%+18.204+10.1%
AgeCL/F− 3.856−7.7%[LEFT RIGHT ARROW] 
V/F− 10.623−12.8%[LEFT RIGHT ARROW] 
Creatinine clearanceCL/F−4.367−2.7%+2.412 
V/F[LEFT RIGHT ARROW]   
SexV/F and Cl/F[LEFT RIGHT ARROW]   
Concomitant AEDs
InducersCL/F−4.339−1.2%−6.203 
V/F[LEFT RIGHT ARROW]   
InhibitorsCL/F−5.262−3.3%−4.889 
V/F[LEFT RIGHT ARROW]   

The covariate final model was then as follows:

  • image
  • image
  • image
  • image

The coefficient variation of CL/F and V/F decreased from 40.9% (CL/F) and 40.2% (V/F) for the basic model to 25.1% (CL/F) and 16.2% (V/F), respectively, for the final model. Table 3 summarizes the population parameter estimates. The goodness of fit was evaluated graphically by the distribution of the points on the model predicted versus the observed plot of LEV concentrations and on the NPDE versus predicted concentration or time (Fig. 1A, B). The final model was characterized by a poor fit for some LEV concentrations (8 of 49) determined within 1.5 h after the first LEV intake. This phenomenon disappeared for data corresponding to the steady-state period. Different absorption models were unsuccessfully tested to improve this result (sequential zero-order and first-order absorption, Michaelis-Menten absorption, and gamma absorption).

Table 3.   Population pharmacokinetic parameters of Levetiracetam in 44 children with epilepsy (final model)
ParametersMeanSE
  1. SE, standard error of the estimate; TV, typical value of the corresponding pharmacokinetic parameter; θcovariate, influential factor for the covariate; ω2, interindividual variability; covCL,V, covariance between η values of CL/F and V/F.

TV (CL/F) (L/h)2.470.103
CL/F, θBW0.890.0923
V/F (L)21.90.954
V/F, θBW0.930.0932
Ka (h−1)3.830.451
Lag time (h)0.2830.023
inline image0.0590.026
inline image0.02650.009
inline image1.370.379
COVCL/F,V/F0.0280.0103
Proportional residual variability (σ12)0.03570.00742
image

Figure 1.  (A) Goodness of fit evaluated by the plots of model population observed versus predicted levetiracetam plasma concentrations. (B) Goodness of fit evaluated by the Normalized Prediction Errors (NPDE) versus time (hour). Goodness of fit evaluated by the Normalized Prediction Errors (NPDE) versus predicted levetiracetam plasma concentrations (PRED).

Download figure to PowerPoint

Visual predictive check

The nonparametric 95% confidence interval (CI) of the simulated concentrations and the observed concentrations are shown in Fig. 2. The LEV observed concentrations were symmetrically distributed around the median, and 3.68% of the observed concentrations were outside the CI.

image

Figure 2.  Visual Predictive Check obtained from 1,000 simulations of the database. Dotted lines indicate the nonparametric 95% confidence interval of the simulations, whereas the points indicate the observed concentrations.

Download figure to PowerPoint

Evaluation of levetiracetam dosing regimens

The population pharmacokinetic parameters of LEV estimated in this study are shown in Table 3. According to the simulations, twice-daily LEV doses of 10 mg/kg, 15 mg/kg, 20 mg/kg, and 30 mg/kg in children are equivalent to the following twice daily doses in adults: 500 mg, 1,000 mg, 1,500 mg, and 2,000 mg, respectively. A pediatric starting dose of 10 mg/kg LEV b.i.d. (20 mg/kg/day) leads to a plasma concentration of LEV equivalent to that in adults receiving a starting dose of 500 mg b.i.d. Moreover, the probabilities of achieving the suggested target plasma concentration of 6–20 mg/L for the dosing regimens of 10mg/kg, 20 mg/kg, and 30 mg/kg b.i.d. are 44%, 90%, and 66% respectively. A LEV pediatric starting dose of 20 mg/kg b.i.d. could be recommended to achieve the target concentration with a risk of 5.1% and 4.9% to reach a concentration under 6 mg/L and above 20 mg/L, respectively. A dose regimen greater than 20 mg/kg b.i.d. increased the risk of a trough concentration greater than 20mg/L, and, therefore, did not improve the probability of achieving the target range concentration. The question of the validity of the target range is raised in the discussion that follows.

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In the present study, we developed a one-compartment model with first-order absorption and elimination that characterized the pharmacokinetic profile of LEV oral administration in children. The model we developed describes the data adequately, except for a few (i.e., 8 of 49) concentrations determined <1.5 h after the first LEV dose. This alteration in the fit is probably not clinically significant because it does not involve steady-state data. The predictive ability of our model is furthermore evidenced by the close agreement between our results and the results of previous studies that are displayed in Table 4 (Pellock et al., 2001; Fountain et al., 2007). Moreover, our mean CL/F and V/F (2.47 L/h and 21.9 L) are also similar to the values found in the previous population pharmacokinetic analysis performed in children (2.17 L/h and 21.5 L) (Snoeck et al., 2007).

Table 4.   Levetiracetam pharmacokinetic parameters from literature and from our study
DoseParametersFountain et al., 2007 N = 21Pellock et al., 2001 N = 24Our study N = 44
  1. AUC, area under the curve; CL/F, apparent oral clearance; T1/2, elimination half life; V/F apparent volume of distribution.

10 mg/kg b.i.d.AUC0–12 (mg h/L)145 ± 44/142 ± 33
C12 h (mg/L)8.4 ± 3.8/6.4 ± 2.7
T1/2 (h)4.9 ± 0.6/6.8 ± 1.5
CL/F (ml/min/kg)1.14 ± 0.18/1.24 ± 0.29
20 mg/kg b.i.d.AUC0–12 (mg h/L)322 ± 71241 ± 76283 ± 65
C12 h (mg/L)15.6 ± 5.3/12.7 ± 4.7
T1/2 (h)4.9 ± 0.46.0 ± 1.16.8 ± 1.5
CL/F (ml/min/kg)1.10 ± 0.161.43 ± 0.361.24 ± 0.29
V/F (L/kg)/0.72 ± 0.120.72 ± 0.21
30 mg/kg b.i.d.AUC0–12 (mg h/L)433 ± 94/425 ± 98
C12 h (mg/L)20.6 ± 5.8/19.1 ± 7.2
T1/2 (h)4.9 ± 0.7/6.8 ± 1.5
CL/F (ml/min/kg)1.12 ± 0.119/1.24 ± 0.29

No significant drug–drug interaction was observed, suggesting that dose adjustment is unnecessary with the concomitant AED. This result is consistent with previously published studies in adults and in children (Perucca et al., 2003; Contin et al., 2004; Fountain et al., 2007; Otoul et al., 2007).

Although the interindividual variability of the absorption lag-time could not be estimated, the absorption profile defined by our model is in agreement with the data found in adults. For instance, a dosing regimen of 500 mg b.i.d. in adults leads to a mean Cmax of 17 mg/L (Patsalos, 2004). Considering the mean population parameters of our model, and the median BW in our population (33 kg), the pediatric dose of 10 mg/kg b.i.d., which is equivalent to the 500-mg b.i.d. adult dose, would provide a very consistent mean Cmax of 13.5 mg/L.

Body weight was identified as the sole covariate that explained interindividual variability on both LEV apparent plasma clearance and distribution volume. Although LEV is eliminated mainly unchanged via renal excretion, creatinine clearance was not found to be a significant covariate. This can probably be explained by the natural maturation of the renal function during childhood, so that creatinine clearance evolves in parallel with age and, therefore, with body weight. The fact that creatinine production increases with the muscular mass, which is reflected by body weight, is a supplemental argument explaining that all the information on the interindividual variability of LEV PK provided by creatinine clearance is included in the relationship between LEV pharmacokinetic parameters and body weight. Finally, the predictive ability of calculated creatinine clearance is probably hampered by this confusion between production and clearance. Indeed, when standard equations, which assume a normal creatinine production, are used for children with low creatinine value, nonphysiologic creatinine clearance was sometimes estimated, probably because the low creatinine concentration resulted from a low creatinine production rather than a high clearance. Furthermore, no child with impaired renal function was included in the study, which also explains the lack of relationship between apparent clearance and creatinine clearance in our model. Therefore, our model is not relevant for children with renal failure, and the dosing recommendations we defined cannot be applied to them.

In the allometric model, an exponent value close to 1 suggests that CL/F and V/F are directly proportional to body weight. For instance, considering CL/F, the relationship CL/F = 2.47x(BW/33)0.87 is close to 2.47x(BW/33), which means that CL/F is directly proportional to BW as it is close to 0.074xBW. This is also consistent with previously published studies (Pellock et al., 2001; Fountain et al., 2007; Glauser et al., 2007). Therefore, our results support the use of a weight-based LEV dosing regimen in children between 4.6 and 16.6 years of age and weighing between 16 and 65 kg. According to the results of the Monte Carlo simulations, two different therapeutic dosing strategies can be used in children with epilepsy. The first one confirmed the current recommendations that consist of starting LEV therapy at a 10 mg/kg b.i.d. dosing regimen, followed by further dose adjustment according to the clinical outcome. Indeed, from a pharmacokinetic point of view, a 10 mg/kg b.i.d. pediatric dose provides the same plasma steady-state AUC as the 500 mg b.i.d. dose in adults, which is the starting recommended dose of LEV for adults. Furthermore, our study provides evidence that children weighing up to 65 kg need a higher dose if a 500-mg dose is considered for a 70-kg adult. So, for an equivalent body weight of 60 kg, adolescents need higher LEV doses than adults in order to achieve the same plasma concentration. This result is also consistent with those of previous studies (Pellock et al., 2001; Fountain et al., 2007; Glauser et al., 2007). The second strategy would be to give an LEV starting dose at a 20 mg/kg b.i.d. dose in order to have the highest probability (90%) of reaching the 6–20 mg/L target range of trough concentrations that was defined in adults. However, clinically some patients would not tolerate this starting dose and would become excessively sleepy. But because trough concentrations comprised a minority of our data (i.e., 25%), our dosing proposal based on target trough concentrations should be validated on a larger population. Furthermore, the relationship between LEV plasma concentrations and clinical effect has not been evaluated and could differ between adults and children. This assumption also emphasizes the need for clinical studies to validate these dosing recommendations.

Acknowledgments

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The authors are grateful to the “Fondation Francaise pour la Recherche sur l’Epilepsie” (FFRE) for their financial support and are indebted to the investigators for their precious work throughout the entire study. We confirm that we have read the Journal’s position on issues involved in ethical publication and state that this report is consistent with those guidelines.

Disclosure of conflicts interest: Stephanie Chhun was a recipient of a fellowship from the “Fédération Hospitalière de France and from the “Chancellerie des Universités de Paris.”

The authors have no conflict interest directly relevant to the content of this study.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Population Pharmacokinetic Modeling
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References