Population pharmacokinetics and pharmacodynamics of edoxaban in pediatric patients

Abstract

Edoxaban is an orally active inhibitor of activated factor X (FXa). Population pharmacokinetic (PK) and pharmacodynamic (PD) analyses were performed to characterize the PK and PK–PD relationships of edoxaban in pediatric patients to identify the covariates that may contribute to inter‐subject variability in PK and PD of edoxaban in pediatric patients, and to compare the PK and PD data between pediatric and adult patients. The pediatric PK of edoxaban was best described by a two‐compartment model with transit compartments, first‐order oral absorption, and linear elimination. The estimated glomerular filtration rate (eGFR), body weight, and post‐menstrual age were the significant covariates explaining variability in edoxaban PK among pediatric patients. A function based on renal maturation was applied to edoxaban clearance. The clearance for a 70 kg patient with an eGFR of 110 mL/min/1.73 m² was estimated to be 42.9 L/h (CV ~ 31.8%). PK simulation showed that exposures across five pediatric age groups were comparable to that in adult patients receiving 60 mg once daily dose. The PK–PD relationship for anti‐factor Xa was best fit with an E_max (8.65 IU/mL) model with an EC₅₀ of 631 ng/mL. The PK–PD relationships for activated partial thromboplastin time and prothrombin time were best fit with linear models (slopes of 0.0467, and 0.0415 s mL/ng, respectively). In addition, due to the small number of efficacy and safety events, an exploratory analysis did not detect a correlation between efficacy events (recurrent venous thromboembolism) or safety events (clinically relevant bleeding) and edoxaban exposure.

---

Study Highlights

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Currently, the dose regimen, pharmacokinetics, and pharmacodynamics relationships of edoxaban in pediatric VTE patients are not available.

• WHAT QUESTION DID THIS STUDY ADDRESS?

Pediatric population pharmacokinetic (PopPK) analysis coupled with an exposure matching strategy and pediatric pharmacokinetics–pharmacodynamics (PK/PD) analyses were used to assess the appropriateness of pediatric dosing regimen tested in two phase III trials.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

The PopPK and PK/PD analyses showed that the PK exposures across five pediatric age groups from 0 to <18 years were slightly lower than adult reference exposure but still within pre-pecified exposure range. Further dose adjustment may not be needed for pediatric VTE patients.

• HOW MIGHT THIS CHANGE DRUG DISCOVERY, DEVELOPMENT, AND/OR THERAPEUTICS?

The study demonstrates that the selection of appropriate adult reference population is important for pediatric PK exposure matching and pediatric dose selection, especially for a drug that is substantially eliminated through renal excretion. The findings are useful for the extrapolation of the efficacy of other drugs from adult to pediatric populations.

---

INTRODUCTION

Edoxaban is an orally active, selective, direct, and reversible inhibitor of activated coagulation factor X (FXa).¹ Since 2015, edoxaban has been approved globally as an oral anticoagulant for the prevention of stroke and systemic embolism in adult patients with nonvalvular atrial fibrillation (NVAF), the prevention of venous thromboembolism (VTE) in adult patients undergoing orthopedic surgeries, and the treatment of VTE in adults. A 60 mg once daily (q.d.) dosing regimen was found to be safe and effective for the treatment of VTE in adult patients.¹ Although the general incidence of VTE in the pediatric population is low (0.07–0.14 per 10,000), it is increased 100–1000-fold in hospitalized children, particularly those with a central venous access device (CVAD).² As such, the development of safe and effective therapies for the prevention and treatment of VTE in pediatric patients is still an unmet medical need.

To extend the VTE treatment indication of edoxaban from adults to children, one phase I study and two phase III studies were conducted to evaluate the safety, efficacy, pharmacokinetics (PK), and pharmacodynamics (PD) of edoxaban in children.^3–5 In the single-dose phase I pediatric study, an age- and body weight-based dosing regimen was established to achieve edoxaban exposure in pediatric patients similar to that in adult VTE patients receiving edoxaban 60 mg once daily dose.³ The age- and body weight-based dosing regimen was further modified and investigated in the two phase III studies. Renal clearance accounts for ~50% of the total clearance of edoxaban in healthy subjects.¹ The dose of edoxaban is reduced by 50% for adult VTE patients with moderate renal impairment (creatinine clearance between 15 and 50 mL/min).¹ In addition, edoxaban is a substrate for the efflux transporter P-glycoprotein (P-gp). Concomitant use of edoxaban with certain P-gp inhibitors (cyclosporin, dronedarone, erythromycin, or ketoconazole) requires a 50% reduction of edoxaban dose in adult patients.¹ Thus, in the two pediatric phase III studies, the dose of edoxaban was reduced for pediatric patients with moderate renal impairment or pediatric patients who were concomitantly using certain P-gp inhibitors.⁴

The objective of the population PK and PD (PopPK/PD) analysis being reported here is to describe the PK and PD of edoxaban in pediatric VTE patients and support dosing recommendations for edoxaban in different pediatric age groups. Specifically, the aims of current PopPK/PD analysis include: (1) to characterize the population PK and PK–PD relationships of edoxaban in pediatric patients; (2) to identify the covariates that may contribute to inter-subject variability in PK and PD of edoxaban in pediatric patients; (3) to compare the PK and PD data between pediatric and adult patients; (4) to use PK simulation to evaluate different potential pediatric dosing regimens; and (5) to explore potential correlations between binary efficacy end point (recurrent VTE) or safety event (clinically relevant bleeding) and edoxaban exposure in pediatric patients.

METHODS

Clinical studies

Data for PopPK and PopPK/PD analyses were derived from three clinical studies in pediatric patients who required or were on anticoagulant therapy: phase I PK/PD study ( identifier, NCT02303431), Hokusai-VTE PEDIATRICS (NCT02798471), and ENNOBLE-ATE (NCT03395639). Descriptions of the clinical study designs are provided in Table S1; additional details have been published elsewhere.^3–5 All studies were conducted in accordance with the ethical principles derived from the Declaration of Helsinki and Council for International Organizations of Medical Sciences International Ethical Guidelines, applicable International Council for Harmonization Good Clinical Practice guidelines, and applicable laws and regulations. Ethics approval was obtained from institutional review boards and ethics committees. Written informed consent was provided prior to study entry; participants less than 18 years of age signed an assent form, and a parent/guardian provided written consent.

In the three studies, pediatric patients were divided into five groups depending on their age (0 to <6 months, 0.5 to <2 years, 2 to <6 years, 6 to <12 years, and 12 to <18 years). In the phase I PK/PD study, study participants of 12 to <18 years old received a single dose of 30 or 60 mg edoxaban tablets, and participants of 0 to <12 years received a single 30 mg-equivalent or 60 mg-equivalent dose of edoxaban oral granules in suspension. The 30 mg- and 60 mg-equivalent doses for each age cohort are listed in Table S1. In Hokusai-VTE PEDIATRICS and ENNOBLE-ATE, study participants in the analysis dataset received age-, weight-, and renal function-based once daily dose (Table S2). The dose for adolescent patients (12 to <18 years) was adjusted based on body weight (<30 kg, 30 to <60 kg, and ≥60 kg), while it was body weight normalized (mg/kg) for patients <12 years and weighing <60 kg. A subject with eGFR within ≥30% to ≤50% of normal for the subject's age at randomization or with concomitant administration of a P-gp inhibitor (excluding amiodarone) received a 25–50% dose reduction. No eGFR- or P-gp inhibitor-based dose reduction was applied to adolescent patients with body weight <30 kg and patients ≤28 days old. All participants who received edoxaban and had at least one measurable concentration of the drug were included in the PopPK analysis dataset. Participants who had individual PK parameter estimates from the PopPK analysis and at least one PD measurement with associated sampling time were included in the PD analysis dataset.

Renal function measure

In the three pediatric studies, based on current regulatory guidance,⁶ renal function in pediatric patients was estimated using eGFR, which was calculated using the bedside Schwartz formula (Equation 1).⁷ eGFR was used as a covariate in the PopPK and PopPK/PD analyses.1$\begin{array}{c}\text{eGFR}\left(\mathrm{mL}/\min /1.73{\mathrm{m}}^2\right)=0.413\times \left(\text{height}\left(\mathrm{cm}\right)/\text{serum}\right.\hfill \\ \left.\text{creatinine},,\left(\mathrm{mg}/\mathrm{dL}\right)\right)\hfill \end{array}$

In contrast, creatinine clearance (CrCl) was used to assess renal function in adult VTE patients in Hokusai-VTE trial.⁸ To compare the renal function of pediatric and adult subjects, the CrCl of pediatric subjects was calculated using the Schwartz equation (Equation 2).⁹2$\begin{array}{c}\text{CrCl}\left(\mathrm{mL}/\min /1.73{\mathrm{m}}^2\right)=\left[\text{height}\left(\mathrm{cm}\right)\times k\right]/\text{serum}\hfill \\ \text{creatinine}\left(\mathrm{mg}/\mathrm{dL}\right)\hfill \end{array}$where, k = 0.45 in term infants to 1 year of age; k = 0.55 in children to 13 years of age; k = 0.70 in adolescent males (females remain at 0.55 after age 13 years).

PK/PD sampling and assay

Venous blood PK and PD samples were obtained at different times prior to and after treatment administration (Table S1). Edoxaban plasma concentrations were determined using a validated liquid chromatography–tandem mass spectrometry (LC–MS/MS) method, with a lower limit of quantification (LLOQ) of 0.764 ng/mL.¹⁰ The mean inter-assay relative error for accuracy was ≤ ±5.1% and the coefficient of variation for precision was ≤5.6%. Three biomarkers, anti-Factor Xa (anti-FXa), activated partial thromboplastin time (aPTT), and prothrombin time (PT) were measured using validated coagulation tests.¹¹

Modeling software and dataset

PopPK and PopPK/PD modeling were performed using NONMEM, version 7.4.4 and version 7.3.0 (ICON, Hanover, MD, US), respectively. Xpose and Perl Speaks NONMEM (PsN; Department of Pharmacy, Uppsala University, Uppsala, Sweden) was also used for model diagnostics and automated model-building procedures, such as covariate testing. A pooled NONMEM-ready dataset was constructed using R (Version 4.1.1). R was used for graphical analysis, model diagnostics, and statistical summaries. R was also used for linear or nonlinear regression analysis.

PopPK model development

Edoxaban plasma concentration–time data were analyzed using a nonlinear mixed-effects modeling approach. The first-order conditional estimation method of NONMEM with interaction (FOCE INTER) was used for PK model development. Two- and three-compartment models with a first-order or zero-order absorption with or without transit compartments were applied to the data and assessed for their capacity to sufficiently characterize the PopPK of edoxaban. In the base model, apparent clearance (CL/F), intercompartment clearance (Q/F), volume of central (V_c/F), and peripheral (V_p/F) compartments were allometrically scaled with body weight relative to a body weight of 70 kg. A sigmoid hyperbolic model based on post-menstrual age (PMA) in weeks was incorporated as a maturation factor (F_PMA) on CL/F to describe the gradual maturation of drug elimination at an early age (Equation 3). The values of time to half maturation (TM50) and Hill coefficient were fixed as 47.7 weeks and 3.40, respectively.¹² The values of transit rate constant (K_tr) and oral absorption rate constant (K_a) in the final pediatric model were estimated by fitting pediatric PK data.3${\mathrm{F}}_{\mathrm{PMA}}=\frac{{\mathrm{PMA}}^{\text{Hill}}}{\mathrm{TM}{50}^{\text{Hill}}+{\mathrm{PMA}}^{\text{Hill}}}$

The inter-individual random effects on the model parameters (K_tr, K_a, CL/F, Q/F, and V_c/F) were evaluated assuming a log-normal distribution of model parameters. A combined additive and proportional error model was used to describe the residual variability. The inter-individual (IIV) and residual variability (RV) were assumed to be symmetrically distributed around 0, with variance ω² and σ², respectively.

Following the development of an appropriate base structural model, the influence of covariates on selected parameters was evaluated using a systematic univariate search followed by a stepwise forward inclusion and backward elimination approach based on the change in the objective function value (OFV). All the covariates that were planned to be evaluated are listed in Table S3. If covariates showed a correlation of >0.5, only one of the correlated covariates was included in the formal analysis. This was either the covariate with the strongest influence as determined by exploratory graphical analysis or the variable that was most meaningful from a clinical perspective. Only covariates available in at least 80% of subjects and categorical covariates with a minimum number of 15 subjects in each category, and with statistically significant parameter–covariate relationships at a significance level of 0.01 in the univariate search were included in the formal covariate analysis. For decisions on the inclusion of covariates, changes in the OFV with significance levels of 0.01 and 0.001 were employed for forward addition and backward elimination, respectively. Overall diagnostic evaluation of the model was based on goodness-of-fit (GOF) plots, and prediction-corrected visual predictive checks (VPCs). VPCs were based on 250 simulations and were stratified by age group.

The final PopPK model was used to simulate a population of at least 1000 subjects by sampling from a multivariate normal distribution informed by the original population PK dataset. The simulated edoxaban exposures in age-, body weight-, eGFR-based pediatric subgroups were compared with edoxaban exposures in adult VTE patients. An edoxaban adult median AUC_0–24h,ss of 1613 ng h/mL was used as the target for pediatric exposure matching. This target exposure was based on the previously reported adult PopPK analysis from 3106 adult VTE patients receiving the approved 60 mg q.d. edoxaban regimen in Hokusai-VTE trial.⁸

PopPK/PD model development

The PK–PD relationships for anti-FXa, aPTT, and PT were visually evaluated using correlation plots of observed PD end points vs. observed plasma concentrations of edoxaban. The PK–PD relationships in pediatric subjects were visually compared with those observed in adult subjects. The adult PD data used in this comparison included anti-FXa and PT data collected from 3461 and 3662 adult VTE patients enrolled in Study Hokusai-VTE,⁸ and aPTT from 403 healthy adult subjects receiving edoxaban at 10–150 mg doses in multiple phase I studies. To reduce uncertainty in the estimation of population PK/PD parameters, only PD data with corresponding time-matched edoxaban concentrations were used for PD modeling. Sequential PK/PD analysis was conducted. The relationships between observed edoxaban concentration and observed anti-FXa, aPTT, and PT were estimated using mixed-effect models in NONMEM. Various direct response models such as linear, power law, E_max, and sigmoidal ones were tested to describe the effect of edoxaban plasma concentrations on PD responses. An exponential IIV on PD model parameters was selected in the base model. A proportional or additive error model was used to describe RV. The covariates evaluated in the PopPK/PD model are listed in Table S4.

Exploratory exposure–response analyses

Due to the low rate of these events, formal model-based estimation of the exposure–response (ER) relationships for efficacy (symptomatic recurrent VTE) and safety [clinically relevant bleeding (CRB)] could not be pursued but were visually examined in pediatric patients using an exploratory data analysis.

RESULTS

Dataset and covariates

The original PopPK analysis dataset comprised 605 plasma concentration observations from a total of 208 subjects. Of these samples, 7 PK samples (1.1%) were excluded because of unrealistically high or low concentrations of edoxaban, and 9 PK samples (1.5%) were excluded because the concentration was below LLOQ. The final PopPK analysis dataset comprised 589 plasma concentration observations from 208 subjects. For the PopPK/PD analysis, 233 anti-FXa observations from 122 subjects, 431 aPTT observations from 197 subjects, and 432 PT observations from 198 subjects were available (Table S1). The categorical covariates and continuous covariates of the 208 subjects are summarized by age group in Tables S5 and S6, respectively.

PopPK analysis

The PK profile of edoxaban in pediatric patients was best described by a two-compartment disposition model, with transit compartments, first-order absorption, and linear elimination (Figure S1). Allometric scaling with body weight on CL/F, Q/F, V_c/F, and V_p/F terms was included in the base model. The scaling exponents of V_c/F and V_p/F with body weight were estimated to be not significantly different from 1; therefore, they were fixed to 1. The estimated scaling exponent of CL/F and Q/F did not substantially improve model fit and the exponent was thus fixed to 0.75. Transit compartments were added to capture the large variability in oral absorption in pediatric patients. The number of transits (NN) and K_tr were estimated as 15 and 47.5 h⁻¹, respectively (Table 1). The mean transit time (MTT) was calculated as 0.337 h. With the addition of the transit compartments, there was a statistically significant improvement in the fit, up to a decrease in OFV of 74 points. The incorporation of PMA or eGFR in clearance significantly improved the base model. The covariate effect of body weight, renal function, and maturation on edoxaban clearance was described using Equation 4 and visualized in Figure S2.4$\begin{array}{c}\mathrm{CL}/\text{F}=\mathrm{CL}/{\text{F}}_{\mathrm{TYP}}*{\left(\mathrm{WT}/70\right)}^{0.75}*{\left(\mathrm{\text{eGFR}}/110\right)}^{0.268}*\hfill \\ \left[{\mathrm{PMA}}^{3.4}/\left({47.7}^{3.4}+{\mathrm{PMA}}^{3.4}\right)\right]\hfill \end{array}$where CL/F_TYP is the estimated CL/F for a typical patient with eGFR = 110 mL/min/1.73 m² and body weight = 70 kg; and WT and PMA represent body weight in kg and post-menstrual age in weeks, respectively.

TABLE 1 Population PK/PD model parameter estimates for edoxaban for a typical pediatric subject (body weight = 70 kg and eGFR = 110 mL/min/1.73 m²).

PK parameter	Estimate	RSE (η-shrinkage)
Apparent clearance, CL/F (L/h)a	42.87	3%
Absorption rate constant, Ka (1/h)	3.71	0.6%
Apparent central compartment volume, Vc/F (L)	261	0.9%
Apparent inter-compartmental clearance, Q/F (L/h)	8.59	2.3%
Apparent peripheral compartment volume, Vp/F (L)	343.5	10.7%
Transit rate constant, Ktr (1/h)	47.5	1%
Fixed exponents for body weight-based scaling	0.75 (CL/F and Q/F); 1.0 (Vc/F and Vp/F)	–
Hill coefficient of maturation function, HILL	3.40 (Fixed)	–
Post-menstrual age (PMA) at half max of the maturation function, TM50	47.7 (Fixed)	–
eGFR effect on clearance	0.268	16%
Inter-individual variability (IIV)	Estimate	RSE
IIV Clearance	31.8%	10% (15%)
IIV Central compartment volume	35.6%	10% (35%)
IIV Inter-compartmental clearance	76.7%	7% (37%)
IIV Transit rate constant	79.6%	9% (60%)
IIV Absorption rate	144.9%	2% (49%)
Residual variability	Estimate	RSE
Proportional error	22.8%	8%
Additive error (ng/mL)	0.71 FIX	–
PD parameter (Anti-FXa)	Estimate	RSE (Shrinkage)
Baseline anti-Factor Xa activity (IU/mL)	0.10 (fixed)	N.A.
Maximum effect of edoxaban (IU/mL)	8.65	42.3%
Edoxaban concentration at half max (ng/mL)	631	50.0%
Inter-individual variability (IIV)
On maximum effect (Emax)	14.8	28.9% (42.5%)
Residual variability	Estimate	RSE
Additive error (IU/mL)	0.247	15.9%
PD parameter (aPTT)	Estimate	RSE (Shrinkage)
Baseline-activated partial thromboplastin time (s)	35.5	2.30%
The slope of the effect of edoxaban (s mL/ng)	0.0467	10.3%
Inter-individual variability (IIV)
On baseline aPTT	30.7	13.7% (13.9%)
Residual variability
Proportional error (CV%)	19.7%	15.3%
PD parameter (PT)	Estimate	RSE (Shrinkage)
Baseline prothrombin time (s)	14.9	1.47%
Slope of effect of edoxaban (s mL/ng)	0.0415	3.92%
Inter-individual variability (IIV)
On baseline PT	14.5	20.3% (29.8%)
Residual variability
Proportional error (CV%)	15.9%	28.1%

A systematic univariate covariate evaluation was performed initially to identify potentially significant covariates, with the covariates planned to be tested listed in Table S3. The effect of concomitant use of P-gp inhibitors or inducers on model parameters was not tested because only 7 subjects (3.4%) and 1 subject (0.5%) among the 208 subjects had recorded concomitant use of P-gp inhibitors or inducers, respectively. Similarly, covariate effects of proton pump inhibitors or nonsteroidal anti-inflammatory drugs (NSAIDs) were not tested. The fed status (food effect) was not tested because only 14 patients (6.7%) received edoxaban under fed state. In the univariate search, the covariate effects of formulation on K_a, ethnicity on V_c/F, and race on V_c/F reached pre-specified significance level of 0.01 but none of these covariates resulted in a statistically significant decrease in OFV in the full stepwise covariate modeling. Variability in edoxaban PK was additionally described by IIV on CL/F, Q/F, V_c/F, K_tr, and K_a with a diagonal matrix and proportional and additive RV models. The final PopPK parameter estimates, standard errors, and covariate effects are shown in Table 1. Most PK parameters were estimated with a good precision (RSE ≤20%). The magnitudes of the IIV were 32–77% for CL/F, Q/F, and V_c/F. The estimates of IIV on K_tr (79.6%) and K_a (144.9%) were large. The η-shrinkage for IIV on Q/F, V_c/F, K_tr, and K_a was higher than the threshold of 30%,¹³ which was likely due to the sparse PK sampling design in the pediatric studies.

Based on the final model and a reference subject of 70 kg and eGFR of 110 mL/min/1.73 m², the typical value for systemic CL/F was 42.9 L/h, K_a was 3.71 h⁻¹, V_c/F was 261 L, and V_p/F was 344 L (Table 1). GOF plots for the final PopPK model show the model adequately described the observed data (Figure S3). The higher residual trend at later time points was driven by very few data points at later times. Both the VPC plot of all observed data (Figure S4) and VPC plots stratified by age group (Figure S5) show that observed concentrations were mostly contained within the range of the 5th and 95th percentiles of the model-predicted concentration values. Overall, the final model captured the central tendency (median) and the extent of variability (5th and 95th percentiles) of observed PK data across all five age groups.

PopPK model-estimated individual steady-state PK exposures (AUC_0–24h,ss and C_max,ss) of 141 pediatric subjects in Hokusai-VTE PEDIATRICS and ENNOBLE-ATE are summarized by age cohort (Table 2). The median C_max,ss of edoxaban in adolescent patients (12 to <18 years) was comparable to that observed in adult VTE patients enrolled in study Hokusai-VTE. The median C_max,ss values in younger patients (0 to <12 years) were 30–50% higher than that in adult VTE patients. The median AUC_0–24h,ss values of edoxaban in 0 to <6 months, 6 months to <2 years, and 6 to <12 years cohorts were comparable to that observed in adult subjects enrolled in Hokusai-VTE. In contrast, the median AUC_0–24h,ss values in 2 to <6 years and 12 to <18 years cohorts were 18% and 22% respectively, lower than that observed in adult VTE patients (Table 2), although this might be due to sampling variability since the 6 to <12 years cohort was estimated to differ from the adult median by less than 3%.

TABLE 2 Summary of PopPK model-estimated pharmacokinetics exposures for pediatric patients in Hokusai-VTE PEDIATRICS and ENNOBLE-ATE trials by age group.

	0 to <6 months (N = 9)	6 months to <2 years (N = 19)	2 to <6 years (N = 36)	6 to <12 years (N = 38)	12 to <18 years (N = 39)	Pediatric overall (N = 141)	Adult VTE Patientsa (N = 3106)
Cmax (ng/mL)
Mean (SD)	309 (91.1)	321 (85.8)	275 (86.0)	289 (79.5)	202 (51.7)	267 (86.5)	212 (49.4)
Median	311 [197,	307 [176,	281 [25.3,	293 [127,	193 [109,	265 [25.3,	211 [68.8,
[Min, Max]	510]	497]	390]	520]	312]	520]	362]
AUC0-24 h,ss (ng h/mL)
Mean (SD)	1790 (667)	1690 (834)	1380 (578	1650 (576)	1310 (346)	1500 (591)	1652 (318)
Median	1820 [718,	1580 [712,	1330 [72.1,	1650 [828,	1260 [679,	1320 [72.1,	1613 [732,
[Min, Max]	2970]	3500]	2900]	3807]	2480]	3807]	3141]

To evaluate the age-, weight-, and eGFR-based doses, simulations were performed using the final PopPK model and following the dosing regimen used in Hokusai-VTE PEDIATRICS and ENNOBLE-ATE. The simulated AUC_0–24h,ss of the five age cohorts were compared with 0.5-fold and 1.5-fold median AUC_0–24h,ss of adult subjects receiving 60 mg q.d. dose in Hokusai-VTE trial (Figure 1). Across the five age groups, the simulated 1st quartile, median, and 3rd quartile of AUC_0–24h,ss values fell within the adult reference range. The median AUC_0–24h,ss values for 0 to <6 months, 6 months to <2 years, 2 to <6 years, 6 to <12 years, and 12 to <18 years cohorts are 1398, 1379, 1322, 1284, and 1174 ng h/mL, respectively, which are 13–27% lower than that observed in adult subjects taking 60 mg q.d. dose (median AUC_0–24h,ss = 1613 ng h/mL) (Figure 1). The body weight-based dose adjustments for adolescent patients were assessed by simulations (Table 3). The 30 mg q.d. dose generated a steady-state exposure (median AUC_0–24h,ss = 1428 ng h/mL) in adolescents <30 kg comparable to that of adult subjects receiving 60 mg q.d. dose. The proposed doses for adolescents with a body weight of 30 kg to <60 kg and ≥ 60 kg resulted in median AUC_0–24h,ss values, 26% and 29%, respectively, lower than the adult reference median exposure (Table 3). The eGFR-based dose adjustments for pediatric patients were also assessed by simulations (Table 4). The median AUC_0–24h,ss values in pediatric subjects with eGFR >50% of normal for age across five age groups were 11% to 27% lower than the adult reference median exposure. For pediatric subjects with eGFR between 30 and 50% of normal for age in 0 to <6 months, 6 months to <2 years, 2 to <6 years, 6 to <12 years, and 12 to <18 years cohorts, the eGFR-based dose reductions resulted in median AUC_0–24h,ss values 57%, 44%, 59%, 18%, and 48%, respectively, lower than the adult median exposure (Table 4).

[IMAGE OMITTED. SEE PDF]

TABLE 3 Simulated AUC_0–24,ss of edoxaban for adolescent patients by body weight.

Body weight	Adolescents (12 to <18 years)	Adult VTE patientsa
<30 kg	30 kg to <60 kg	≥60 kg
N	70	391	556	N = 3106
Mean ± SD (ng h/mL)	1701 ± 958	1259 ± 439	1196 ± 421	1652 (318)
Median [5th, 95th] (ng h/mL)	1428 [711, 3899]	1198 [670, 2047]	1150 [597, 1978]	1613 [1204, 2246]

TABLE 4 Simulated AUC_0–24,ss of edoxaban for pediatric patients by age and eGFR.

eGFR % of Normal	0 to <6 months	6 months to <2 years	2 to <6 years	6 to <12 years	12 to <18 years	Adult VTE patientsa
30–50	>50	30–50	>50	30–50	>50	30–50	>50	30–50	>50	N.A.
Dose	0.4 mg/kg	0.8 mg/kg	0.75 mg/kg	1.5 mg/kg	0.7 mg/kg	1.4 mg/kg	0.8 mg/kg	1.2 mg/kg	45 or 30 mgb	60, 45 or 30 mgb	60 mg
N	61	1184	19	1017	8	1014	10	1000	19	998	3106
Mean (SD) ng h/mL	723 (228)	1498 (577)	905 (330)	1491 (606)	735 (334)	1404 (519)	1236 (518)	1356 (502)	920 (267)	1261 (500)	1652 (318)
Median [5th, 95th] ng h/mL	687 [415, 1142]	1428 [697, 2555]	904 [420, 1330]	1395 [670, 2669	657 [408, 1274]	1331 [696, 2357]	1327 [487, 1957]	1284 [671, 2211]	833 [622, 1287]	1177 [646, 2092]	1613 [1204, 2246]

The dosing regimens tested in Hokusai-VTE PEDIATRICS and ENNOBLE-ATE appear inadequate to match adult exposure. To match the adult exposure (median AUC_0–24h,ss = 1613 ng h/mL), simulations were conducted. The dose for subjects in five age cohorts with normal renal function was increased by 20–50% and the dose for subjects with moderate renal impairment was increased by 25–100% across the five cohorts (Table S2). PopPK simulations showed the adjusted dosing regimens could better match adult exposures (Figure S6).

PopPK/PD analyses

The three PD biomarkers, PT, aPTT, and anti-FXa activity exhibited exposure-dependent responses to edoxaban in children of different ages in a manner that was visually analogous to adult subjects (Figure 2). Anti-FXa data were best fit with an E_max model with IIV on E_max (Table 1). Baseline anti-FXa (E₀) was fixed to 0.1 IU/mL because of the small number of pre-treatment anti-FXa measurements (N = 9). An additive error model was used to describe the RV for anti-FXa. aPTT data were best fit with a linear model with IIV on the intercept (E₀). A proportional error model was used to describe the RV for aPTT (Table 1). PT data were best fit with a linear model with IIV on the intercept (E₀). A proportional error model was used to describe the RV for PT (Table 1). GOF plots for the final anti-FXa (Figure S7), aPTT (Figure S8), and PT (Figure S9) models show the models adequately described observed data. VPC plots showed that observed PD data are mostly contained within the range of the 5th and 95th percentiles of the model-predicted anti-FXa, aPTT, and PT values (Figure S10).

[IMAGE OMITTED. SEE PDF]

Exploratory ER analyses

Three VTE events and five CRB events including major bleedings and clinically relevant non-major bleedings were reported in the 208 subjects in the PopPK dataset during the main or extension treatment period. Due to the small number of VTE and CRB events, only visual exploration of the E–R relationships was conducted. The model-estimated edoxaban exposures (AUC_0–24h,ss, C_max,ss and C_trough,ss) in the three subjects with VTE were higher than or comparable to the median exposures in subjects without VTE (Figure 3, upper row). The data indicated that the three VTE events were not associated with a low edoxaban exposure. The edoxaban exposures (AUC_0–24h,ss, C_max,ss and C_trough,ss) in the five subjects with CRB were generally within the 1st and 3rd quartiles of edoxaban exposures in subjects without CRB (Figure 3, lower row). The five CRB events were not associated with a high edoxaban exposure.

[IMAGE OMITTED. SEE PDF]

DISCUSSION

Pediatric dose selection for edoxaban depends on similar PK–PD relationships and PK exposures between the adult and the intended pediatric VTE population.^14,15 We used pediatric PopPK and PopPK/PD analyses to compare the PK–PD relationships and PK exposures of edoxaban between adult and pediatric populations. For PK exposure matching, appropriate selection of the adult reference population is critical, especially for drugs such as edoxaban, which are substantially eliminated through renal excretion. In the current study, adult subjects receiving 60 mg q.d. dose in Hokusai-VTE trial were selected as the reference population for PK exposure matching. The pediatric PopPK model analysis showed that the pediatric dosing regimen tested in phase III trials would result in 13–27% lower exposures than the adult reference exposure (Figure 1). The lower exposure may be explained by greater renal clearance in the pediatric VTE population. The mean (median) age of the adult reference population was 55.6 (57.0) years and the median (mean) CrCl for these adult patients was 89 (92) mL/min/1.73 m².⁸ In contrast, the median (mean) CrCl values calculated by the Schwartz equation⁹ for 0 to <6 months, 6 months to <2 years, 2 to <6 years, 6 to <12 years, and 12 to <18 years groups in the current PopPK analysis were 79 (86), 146 (161), 169 (176), 165 (180), and 152 (161) mL/min/1.73 m², respectively. The median body surface area (BSA) normalized CrCl values in pediatric subjects of 0.5–18 years old are ~ 64–90% higher than that in the adult reference population. Since renal clearance accounts for ~ 50% of the total clearance of edoxaban,¹⁶ the higher pediatric renal clearance is the primary reason for the lower edoxaban exposures in pediatric subjects. Similarly, the total clearance of dabigatran and rivaroxaban in adolescents was higher than that in older adult subjects. Renal clearance accounts for ~ 80% of the total clearance of dabigatran.¹⁷ The mean age and CrCl of the adult reference population for dabigatran pediatric PK exposure matching was 55.0 years and 106 mL/min.¹⁸ The approved dabigatran dose for adult VTE is 150 mg twice daily (b.i.d.). Due to the high clearance in adolescents, to match adult PK exposure, the approved doses for pediatric subjects with a body weight of 41 to ≤60 kg and 61 to ≤80 kg are 185 mg b.i.d. and 220 mg b.i.d., respectively.¹⁷ Renal clearance accounts for ~ 40% of the total clearance of rivaroxaban. Because of the high renal function in children, the same dose (20 mg q.d.) resulted in a lower rivaroxaban exposure in adolescents¹⁹ than that in the adult phase III population (mean age of 60.5 years).²⁰ Age substantially affects the renal function and PK exposure of rivaroxaban in adults and the AUC of rivaroxaban in subjects of 75–83 years old is ~ 1.4-fold higher than that in subjects of 18–43 years old.²¹ Overall, for a drug substantially eliminated through renal excretion, the age of adult reference population can substantially affect pediatric PK exposure matching criteria and pediatric dose selection.

The exposure–response analysis for VTE shows that adult patients with an edoxaban AUC ≥965 ng h/mL are expected to have a lower rate of VTE events than in the warfarin treatment arm.²² PK simulation with the pediatric dosing regimen tested in phase III trials showed approximately equal or more than 25% of subjects across five age cohorts would have AUC < 965 ng h/mL (Figure 1). Guided by PK simulation, the doses were increased by 25–50% across five age cohorts. With the adjusted dosing regimen, 85–91% of pediatric subjects across five cohorts could achieve AUC ≥965 ng h/mL (Figure S6).

In the current PopPK model, TM50 and Hill coefficient were fixed as 47.7 weeks and 3.4, respectively, to estimate the pediatric CL/F of edoxaban. The maturation equation was developed to describe the nonlinear relationship between GFR maturation and PMA in pediatric subjects from 0 to 18 years old.¹² Renal excretion accounts for 50% of CL/F of edoxaban in healthy adults. Edoxaban is metabolized mainly by carboxyesterase 1 (CES1) and cytochrome P450 3A (CYP3A).²² The plasma exposure of the major metabolite formed by CES1 was ~10% that of edoxaban in healthy subjects. The remaining metabolites together equal <5% of total plasma exposure to edoxaban. Ideally, different TM50 values should be used to describe the maturation of eGFR, CES1, and CYP3A separately. A fixed TM50 of 47.7 weeks and Hill coefficient of 3.4 were used in the current model because the maturation equation with these fixed parameters could reasonably fit observed PK data. The expression levels of hepatic CES1 and CYP3A4 in infants reach 50% of adult levels at 3 weeks (PMA = 43 weeks)²³ and 36 weeks (PMA = 76 weeks) after birth.²⁴ The TM50 of CES1 is comparable to that of GFR (TM50 = 47.7 weeks), which supports the use of a single TM50 of 47.7 weeks.

For pediatric subjects with moderate renal impairment, eGFR-based dose reductions of edoxaban are predicted to result in 18%–59% lower median AUC across five age groups than the adult reference median exposure (Table 4). For adult VTE patients with moderate renal impairment, the recommended edoxaban dose is reduced from 60 mg q.d. to 30 mg q.d.¹⁶ The dose reduction resulted in 21% lower edoxaban median AUC compared with subjects receiving 60 mg q.d. dose.⁸ Both pediatric and adult PopPK analyses suggest that a 20%–40% dose reduction would be more appropriate than a 50% dose reduction for subjects with moderate renal impairment. With the availability of oral suspension formulation, it is feasible to adjust the dose more accurately for pediatric subjects with moderate renal impairment.

Inter-individual random effects were tested on five PK parameters, K_tr, K_a, CL/F, Q/F, and V_c/F. The η-shrinkages related to IIV for K_tr, K_a, CL/F, Q/F, and V_c/F were 60%, 49%, 15%, 37%, and 35%, respectively. The low η-shrinkage for CL/F indicated that the IIV on CL/F was well characterized in the model. However, the high η-shrinkages for K_tr, K_a, Q/F, and V_c/F indicated that the individual parameter post hoc estimates were considerably shrunken toward the typical values and might not represent the true IIV. The high shrinkages were likely due to the sparse PK sampling designs in the pediatric clinical studies, combined with the relatively small sample sizes. The sparse PK data sampling during the initial absorption phase of edoxaban likely resulted in a poor estimation of C_max in pediatric subjects. This sparse sampling may have affected the precise estimation of K_tr, K_a, V_c/F, and Q/F as well. The E–R analysis in adult VTE patients showed that the probability of recurrent VTE decreased with increasing edoxaban C_avg or AUC.²⁵ Conversely, no statistically significant E–R relationships were found for CRB or major adverse cardiovascular events in adult VTE patients.²⁵ It is not evident that C_max is an adequate predictor of efficacy or safety based on the adult data. Thus, AUC instead of C_max was used as the PK end point for pediatric exposure matching.

Across all five pediatric age groups, high inter-individual variability was observed in the PD data, in particular for aPTT and PT data. At birth, the plasma levels of most coagulation proteins are approximately half of the adult levels.^26,27 The coagulation factor activities in children typically reach normal adult levels between a few months of age and up to above 16 years for specific coagulation measures such as aPTT.²⁷ Therefore, age may affect the pediatric baseline level of aPTT. However, these differences in baseline activity of coagulation factors are not likely to substantially affect drug treatment effect in children and adults. In vitro coagulation studies have demonstrated that edoxaban increased PT, aPTT, and anti-FXa levels in a similar concentration-dependent manner in adults and children across an age range of 0–18 years.²⁸ Unfortunately, baseline PD data normalization could not be conducted in the PK/PD analyses because the pre-treatment baseline PD data was collected from only a small fraction of pediatric subjects. The estimate of EC₅₀ for anti-FXa (631 ng/mL) in our model may be inaccurate due to the lack of observed PK data at >500 ng/mL. In an in vitro study of the anticoagulant effects of edoxaban, pooled citrated plasma from healthy subjects was spiked with edoxaban (10–2000 ng/mL). The EC₅₀ of inhibition on thrombin generation by edoxaban was 460 ng/mL.¹¹ Due to the small sample size and large variability of pediatric PK and PD data, our PopPK/PD model might overestimate EC₅₀ of anti-FXa by 30–40%.

In summary, in the current study, we developed pediatric PopPK and PopPK/PD models of edoxaban. Renal function (eGFR), body weight, and PMA were estimated to be significant covariates explaining the PK variability of edoxaban in pediatric subjects. PK simulation showed that the phase III pediatric doses resulted in exposures across all five pediatric age groups that were 13%–27% lower than that in adult VTE subjects receiving 60 mg q.d. dose. Guided by PopPK simulation, the pediatric dosing regimen was adjusted to better match adult exposure. Currently, neither the original phase III trial dosing regimen nor the adjusted dosing regimen of edoxaban has been approved for pediatric indication by regulatory agencies. The PK–PD relationship for anti-FXa was the best fit with an E_max model. The PK–PD relationships for aPTT and PT were best fit with linear models. In addition, due to the small number of efficacy and safety events, an exploratory analysis did not detect a correlation between efficacy events (recurrent VTE) or safety events (clinically relevant bleeding) and edoxaban exposure.

AUTHOR CONTRIBUTIONS

P.Z. and T.A.L. wrote the manuscript. P.Z. and T.A.L. designed the research. P.Z., A.A., and P.C. performed the research. P.Z., A.A., M.G., P.C., and T.A.L. analyzed the data.

FUNDING INFORMATION

This work was supported by Daiichi Sankyo, Inc.

CONFLICT OF INTEREST STATEMENT

Zou P and Leil TA are employees of Daiichi Sankyo Inc. Atluri A and Chang P are employees of Certara, Inc. Leil TA reports stock ownership in Daiichi Sankyo Inc. Atluri A and Chang P report stock ownership in Certara, Inc.

References

Daiichi‐Sankyo. Lixiana Summary of Product Characteristics. 2020. https://www.ema.europa.eu/en/documents/product‐information/lixiana‐epar‐product‐information_en.pdf. Accessed March 27, 2023.

Monagle P, Cuello CA, Augustine C, et al. American Society of Hematology 2018 guidelines for management of venous thromboembolism: treatment of pediatric venous thromboembolism. Blood Adv. 2018;2:3292‐3316.

Daiichi‐Sankyo‐Inc. Phase 1 Pediatric Pharmacokinetics/Pharmacodynamics (PK/PD) Study. 2022. https://clinicaltrials.gov/ct2/show/study/NCT02303431. Accessed March 25, 2023.

Portman MA, Jacobs JP, Newburger JW, et al. Edoxaban for thromboembolism prevention in pediatric patients with cardiac disease. J Am Coll Cardiol. 2022;80:2301‐2310.

van Ommen CH, Albisetti M, Chan AK, et al. The edoxaban Hokusai VTE PEDIATRICS study: an open‐label, multicenter, randomized study of edoxaban for pediatric venous thromboembolic disease. Res Pract Thromb Haemost. 2020;4:886‐892.

EMA/CHMP. Guideline on the evaluation of the pharmacokinetics of medicinal products in patients with decreased renal function. 2015. https://www.ema.europa.eu/en/documents/scientific‐guideline/guideline‐evaluation‐pharmacokinetics‐medicinal‐products‐patients‐decreased‐renal‐function_en.pdf

Schwartz GJ, Work DF. Measurement and estimation of GFR in children and adolescents. Clin J Am Soc Nephrol. 2009;4:1832‐1843.

Niebecker R, Jönsson S, Karlsson MO, et al. Population pharmacokinetics of edoxaban in patients with symptomatic deep‐vein thrombosis and/or pulmonary embolism—the Hokusai‐VTE phase 3 study. Br J Clin Pharmacol. 2015;80:1374‐1387.

Schwartz GJ, Haycock GB, Edelmann CM Jr, Spitzer A. A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics. 1976;58:259‐263.

He L, Kochan J, Lin M, Vandell A, Brown K, Depasse F. Determination of edoxaban equivalent concentrations in human plasma by an automated anti‐factor Xa chromogenic assay. Thromb Res. 2017;155:121‐127.

Samama MM, Mendell J, Guinet C, Le Flem L, Kunitada S. In vitro study of the anticoagulant effects of edoxaban and its effect on thrombin generation in comparison to fondaparinux. Thromb Res. 2012;129:e77‐e82.

Rhodin MM, Anderson BJ, Peters AM, et al. Human renal function maturation: a quantitative description using weight and postmenstrual age. Pediatr Nephrol. 2009;24:67‐76.

Savic RM, Karlsson MO. Importance of shrinkage in empirical bayes estimates for diagnostics: problems and solutions. AAPS J. 2009;11:558‐569.

EMA/CHMP. Paediatric Addendum on the guidelines on clinical investigation of medicinal products for the treatment and prophylaxis of venous thromboembolic disease 2023. https://www.ema.europa.eu/en/documents/scientific‐guideline/paediatric‐addendum‐guidelines‐clinical‐investigation‐medicinal‐products‐treatment‐prophylaxis_en‐0.pdf

FDA U. E11 Clinical Investigation of Medicinal Products in the Pediatric Population. 2000. https://www.fda.gov/media/71355/download. Accessed March 28, 2023.

Daiichi‐Sankyo‐Inc. SAVAYSA (edoxaban) tablets label approved by FDA. 2019. https://www.accessdata.fda.gov/drugsatfda_docs/label/2019/206316s015lbl.pdf. Accessed March 24, 2023.

Boehringer‐Ingelheim. FDA approved drug label for PRADAXA (Dabigatran capsules). 2021. https://www.accessdata.fda.gov/drugsatfda_docs/label/2021/022512s041lbl.pdf

Schulman S, Kearon C, Kakkar AK, et al. Dabigatran versus warfarin in the treatment of acute venous thromboembolism. N Engl J Med. 2009;361:2342‐2352.

Ahn J et al. FDA clinical pharmacology review of Xarelto‐pediatric indications. 2021. https://www.fda.gov/media/158802/download

Willmann S, Zhang L, Frede M, et al. Integrated population pharmacokinetic analysis of rivaroxaban across multiple patient populations. CPT Pharmacometrics Syst Pharmacol. 2018;7:309‐320.

Janssen‐Pharmaceuticals. FDA approved drug label for XARELTO (Rivaroxaban tablets). 2023. www.accessdata.fda.gov/drugsatfda_docs/label/2023/022406s041lbl.pdf

Andersen D et al. FDA clinical pharmacology review of edoxaban tosylate. 2014. https://www.accessdata.fda.gov/drugsatfda_docs/nda/2015/206316orig1orig2s000clinpharmr.pdf

Salem F, Johnson TN, Barter ZE, Leeder JS, Rostami‐Hodjegan A. Age related changes in fractional elimination pathways for drugs: assessing the impact of variable ontogeny on metabolic drug‐drug interactions. J Clin Pharmacol. 2013;53:857‐865.

de Wildt SN, Kearns GL, Leeder JS, van den Anker JN. Cytochrome P450 3A: ontogeny and drug disposition. Clin Pharmacokinet. 1999;37:485‐505.

Nyberg J, Karlsson KE, Jönsson S, et al. Edoxaban exposure‐response analysis and clinical utility index assessment in patients with symptomatic deep‐vein thrombosis or pulmonary embolism. CPT Pharmacometrics Syst Pharmacol. 2016;5:222‐232.

Monagle P, Ignjatovic V, Savoia H. Hemostasis in neonates and children: pitfalls and dilemmas. Blood Rev. 2010;24:63‐68.

Toulon P. Developmental hemostasis: laboratory and clinical implications. Int J Lab Hematol. 2016;38(Suppl 1):66‐77.

Sinegre T, Zlobecki M, Doré E, Pereira B, Grèze V, Lebreton A. In vitro assessment of edoxaban anticoagulant effect in pediatric plasma. Thromb Res. 2019;178:112‐118.