An integrated population pharmacokinetic model of febuxostat in pediatric patients with hyperuricemia including gout and adult population of healthy subjects and patients with renal dysfunction

Abstract

The study objective was to validate febuxostat dosage and administration in pediatric patients with hyperuricemia including gout, using an integrated population pharmacokinetic (PopPK) analysis in the Japanese population. Integrated PopPK analysis of febuxostat used a nonlinear mixed‐effects modeling (NONMEM) program on plasma febuxostat concentration data for 2611 samples from Japanese pediatric patients with hyperuricemia including gout (n = 29) and from adult subjects who are healthy or have renal dysfunction (n = 113). We described febuxostat pharmacokinetics using an integrated PopPK model applicable both to pediatric patients and to the adult population. The covariates of body weight and eGFR were identified for CL/F and the covariate of fasted/fed status for bioavailability. The range of steady‐state exposures (C _max,ss and AUC_τ,ss) for 5, 10, 20, and 30 mg of febuxostat in fed pediatric patients weighing 20 to 40 kg was within that for 10, 20, 40, and 60 mg of febuxostat in fed pediatric patients and adults weighing 40 to 120 kg. Post hoc estimates of CL/F, adjusted by body weight, differed little between pediatric patients and the adult population in the renal function categories of normal, mild dysfunction, and moderate dysfunction. We successfully validated the febuxostat dose that provided the same level of exposure in pediatric patients as in the adult population: half the adult dose for pediatric patients weighing <40 kg and the full adult dose for pediatric patients weighing ≥40 kg. As in adults, the results support the use of febuxostat without dose adjustment in pediatric patients who have mild to moderate renal dysfunction.

Our PopPK model simulation of febuxostat demonstrated that the doses used in the clinical study were appropriate for providing the same level of exposure in pediatric patients as in the adult population: half the adult dose for pediatric patients weighing <40 kg and the full adult dose for pediatric patients weighing ≥40 kg.

Abbreviations

ALAG1

absorption lag time

AUC_τ,ss

area under the concentration‐time curve during a dosage interval at steady state

CL/F

apparent total body clearance

CLCr

creatinine clearance

C _max,ss

maximum drug concentration at steady state

eGFR

estimated glomerular filtration rate

F

bioavailability

Ka

first‐order absorption rate constant

NCA

non‐compartmental analysis

OFV

objective function value

pc‐VPC

prediction‐corrected visual predictive check

PK

pharmacokinetic(s)

PopPK

population pharmacokinetic(s)

Q/F

apparent intercompartment clearance

sUA

serum uric acid

ULT

urate lowering therapy

V/F

apparent volume of distribution

WGT

body weight

1. INTRODUCTION

Febuxostat is an inhibitor of two isoforms of xanthine oxidoreductase involved in the catabolism of purines, specifically xanthine oxidase and xanthine dehydrogenase, ¹ and lowers serum uric acid levels (sUA) by inhibiting the formation of uric acid from hypoxanthine. ² , ³ , ⁴ , ⁵ , ⁶ This drug is used worldwide for urate lowering therapy (ULT) in adults with hyperuricemia resulting in gout or hyperuricemia associated with cancer chemotherapy. ⁷ , ⁸ , ⁹ , ¹⁰ In Japan, the starting daily dose of febuxostat for adult patients with hyperuricemia including gout is 10 mg and can be titrated up to 20, 40, or 60 mg while monitoring sUA levels. ¹¹ Febuxostat has a high plasma protein binding rate of approximately 99% and binds mainly to albumin. ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ Once absorbed, the drug is metabolized primarily by multiple uridine diphosphate‐glucuronosyltransferases to glucuronide conjugates, with a portion converted by multiple cytochrome P450s to the oxidized form and excreted in the urine and feces, and <5% eliminated as unchanged drug through the urine. ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Because uric acid is excreted by the kidneys, patients with kidney disease often experience elevated sUA levels. Febuxostat can provide ULT suitable for adult patients with hyperuricemia including gout, even those with mild to moderate renal dysfunction, as the drug requires no dose adjustment in these patients. ¹¹ , ¹⁹ , ²⁰ , ²¹

Although gout is rare in children, chronic hyperuricemia occurs more often as a complication of renal dysfunction due to congenital and hereditary underlying diseases. ²² , ²³ , ²⁴ However, there are no ULT drugs worldwide with pediatric indications. Therefore, specific dosage recommendations are needed for the pediatric patient population with renal dysfunction.

In recent years, population analysis methods, including population pharmacokinetic (PopPK) analysis, have been used in pediatric drug development because their statistical methods and quantitative approaches based on modeling and simulation allow quantitative use of a small number of sparse sampling data and support extrapolation from adult populations to pediatric populations. ²⁵ , ²⁶ For febuxostat pharmacokinetics (PK), several non‐compartmental model analyses, ¹² , ¹³ , ¹⁵ , ¹⁶ , ¹⁷ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ compartmental model analyses, ³⁴ , ³⁵ , ³⁶ population analyses, ³⁷ , ³⁸ , ³⁹ , ⁴⁰ and physiologically based PK analyses ⁴¹ have been reported. We are unaware of any reports that have conducted model analyses using pediatric plasma febuxostat concentration data to examine pediatric dosage. We recently conducted Phase 2 clinical studies to evaluate the long‐term efficacy and safety of febuxostat in pediatric patients with hyperuricemia including gout in Japan and obtained sparse pediatric plasma febuxostat concentration data. ⁴²

In the present study, we developed an integrated PopPK model of febuxostat using Japanese pediatric and adult febuxostat concentration data and examined the appropriateness of dosage for pediatric patients with hyperuricemia including gout.

2. METHODS

2.1. Study design and population

To develop the PopPK model for febuxostat and to examine pediatric dosing, we combined plasma febuxostat concentration data from the following six studies in Japanese subjects:

Phase 2 studies (evaluation phase as Study 1 and continuation phase as Study 2) ⁴² of febuxostat in 30 Japanese pediatric patients aged 6 to 18 years with hyperuricemia including gout, with the dosing category determined by body weight at the pre‐enrollment examination, using a dose range of 5, 10, 20, and 30 mg for the 10 patients weighing <40 kg and a dose range of 10, 20, 40, and 60 mg for the 20 patients weighing ≥40 kg, and febuxostat given orally once daily for 52 weeks, with up‐titration as needed; three Phase 1 studies (Studies 3, 4, and 5, with a total of 92 subjects in the three studies) ⁴³ in healthy Japanese adult males, with subjects receiving febuxostat 10–160 mg as a single or repeated dose under fasted or fed conditions; a Phase 1 study (Study 6) ¹⁶ of 20 mg febuxostat administered repeatedly to fed Japanese adults with normal renal function (10 subjects) or with renal dysfunction (mild in five patients, moderate in nine patients).

The design, dosage and administration, and time points of blood collection for each clinical trial are summarized in Table S1. All clinical trials were conducted in compliance with the ethical principles based on the Declaration of Helsinki, the Clinical Trial Protocol, the Pharmaceuticals and Medical Devices Law, and the Ministerial Ordinance on Good Clinical Practice (GCP) for Drugs.

Plasma febuxostat concentrations were determined using validated liquid chromatography–tandem mass spectrometry with the lower limit of quantification of 0.5 ng/mL for Studies 1 and 2, and validated high‐performance liquid chromatography with the lower limit of quantification of 4 ng/mL for Studies 3 to 6. Both analytical methods showed within‐run and between‐run accuracy within ±15%, or within ±20% at the lower limit of quantification. Within‐run and between‐run precision was also within 15%, or within 20% at the lower limit of quantification. Subject characteristics were tabulated in the categories of age, sex, height, body weight, body surface area (BSA), the liver function markers of alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, and gamma‐glutamyl transferase (GGT), the renal function markers of serum creatine, creatinine clearance (CLCr) and estimated glomerular filtration rate (eGFR), total cholesterol, triglycerides, albumin, and fasted‐fed status of subjects (fed: orally administered within 0.5 h after a meal; fasted: orally administered under fasting conditions or more than 0.5 h after a meal). BSA was calculated using the Du Bois formula, ⁴⁴ CLCr was assessed by the Cockcroft–Gault formula, ⁴⁵ and eGFR was calculated from the formula recommended by the Japanese Society of Nephrology and the Japanese Society for Pediatric Nephrology ⁴⁶ (Supporting method).

2.2. Population PK modeling

2.2.1. Software

Nonlinear mixed‐effects modeling (NONMEM [ver. 7.4.3; ICON plc]) was used in the PopPK analysis. First‐order conditional estimation with interaction (FOCEI) was used in parameter estimation. R (ver. 3.6.1), RStudio (ver. 1.2.1335), Perl‐speaks‐NONMEM (ver. 4.9.0), ⁴⁷ SAS (ver. 9.4; SAS Institute), and WinNonlin (ver. 8.1; Certara Company) were used for exploratory analysis, model diagnostics and assessment of adequacy, and simulation.

2.2.2. Modeling strategy

In the data set for development of the PopPK model, plasma febuxostat concentration data were excluded when they were below the lower limit of quantification (BLQ) or missing, and no imputation was performed. To improve the robustness of the PopPK model, physiologically implausible abnormal values and outliers were excluded. The criterion established for identifying outliers was “any data point for which the absolute value of the conditional weighted residuals (CWRES) exceeded 6.” ⁴⁸

In developing the PopPK model, we first determined the basic model by examining the structural models for PK and for the absorption phase and the error models for between‐subject variability and residual variability. We then considered the range of distribution of subject characteristics, relationships among those characteristics, relationships between those characteristics and post hoc estimated PK parameters, and physiological and pharmacokinetic findings. Based on our findings, we selected age, body weight, CLCr, eGFR, eGFR category, and fasted/fed status as candidate covariates. The final covariates were determined through stepwise covariate modeling, using a combination of forward selection and backward elimination methods, and then, the final model was developed. Model selection was based on the objective function value (OFV), Akaike information criterion, physiologic and pharmacokinetic validity of parameter estimates, relative standard error (RSE) of parameter estimates, shrinkage, condition number, and goodness‐of‐fit (GOF) plot results. The significance level for the likelihood ratio test of OFV was p < .05 for the forward selection method and p < .01 for the backward elimination method. The criterion for the condition number was set at ≤1000. In addition, as a part of our evaluation of the final PopPK model, we conducted a prospective internal assessment using the bootstrap method and a prediction‐corrected visual predictive check (pc‐VPC). ⁴⁷ , ⁴⁸

2.3. Population PK simulation

2.3.1. Sensitivity analysis of the impact of covariates on PK parameters

For the covariates incorporated into the PopPK final model, sensitivity analyses were performed using Monte Carlo simulations (1000 hypothetical subjects for each condition) to determine the effects on PK parameters and steady‐state exposures (maximum drug concentration at steady state [C _max,ss] and area under the concentration‐time curve during a dosage interval (τ) at steady state [AUC_τ,ss]). The values of covariates not included in the sensitivity analysis were set to their median values.

2.3.2. Comparison of PK parameters and the relationship between febuxostat exposure and body weight

The PopPK final model was used to compare the steady‐state PK parameters and exposures (C _max,ss and AUC_τ,ss) of the pediatric patients and the adult population in clinical studies by post hoc estimation. Monte Carlo simulations (1000 hypothetical subjects for each condition) were also performed to compare the relationship of steady‐state exposure and body weight in pediatric patients and that in adults when febuxostat was administered orally once daily after a meal.

2.4. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY, ⁴⁹ and are permanently archived in the Concise Guide to PHARMACOLOGY 2023/24. ⁵⁰

3. RESULTS

3.1. Subject characteristics

The subjects analyzed included 29 pediatric patients aged 8 to 18 years with hyperuricemia including gout (10 weighing <40 kg, 19 weighing ≥40 kg) and 113 adult subjects (76 with normal renal function and 37 with renal dysfunction based on eGFR: 23 mild, 12 moderate, and 2 severe). A summary of subject characteristics is presented in Table 1. The proportion of subjects with reduced renal function (eGFR) was greater among the pediatric patients than among the adult population.

TABLE 1.

Characteristics of subjects in the PopPK analysis.

	Pediatric patients			Adult population (n = 113)
	All (n = 29)	<40 kg (n = 10)	≥40 kg (n = 19)
Age, years	13.0 (8, 18)	11.5 (8, 16)	13.0 (9, 18)	24.0 (20, 72)
Male, n (%)	23 (79.3)	5 (50.0)	18 (94.7)	110 (97.3)
Height, cm	160.1 (117, 173)	142.7 (117, 154)	161.9 (129, 173)	171.0 (141, 187)
Body weight, kg	46.70 (26.7, 94.3)	36.85 (26.7, 39.9)	52.70 (41.1, 94.3)	61.30 (48.1, 86.5)
BSA, m2	1.420 (0.91, 2.07)	1.220 (0.91, 1.29)	1.610 (1.18, 2.07)	1.720 (1.47, 2.08)
eGFR, mL/min/1.73m2	76.65 (33.9, 145.3)	66.40 (41.4, 114.9)	84.07 (33.9, 145.3)	99.47 (20.9, 143.6)
eGFR category, n (%)
Normal (≥90 mL/min/1.73m2)	11 (37.9)	2 (20.0)	9 (47.4)	76 (67.3)
Mild (≥60–<90 mL/min/1.73m2)	11 (37.9)	4 (40.0)	7 (36.8)	23 (20.4)
Moderate (≥30–<60 mL/min/1.73m2)	7 (24.1)	4 (40.0)	3 (15.8)	12 (10.6)
Severe (≥15–<30 mL/min/1.73m2)	0 (0.0)	0 (0.0)	0 (0.0)	2 (1.8)
Serum creatinine, mg/dL	0.820 (0.37, 2.15)	0.825 (0.37, 1.21)	0.750 (0.47, 2.15)	0.800 (0.58, 2.26)
CLCr, mL/min	109.83 (38.5, 296.1)	75.24 (48.3, 140.0)	126.10 (38.5, 296.1)	121.02 (27.3, 199.4)
Albumin, g/dL	4.40 (3.0, 5.0)	4.20 (3.0, 4.5)	4.50 (4.2, 5.0)	4.30 (3.4, 5.1)
Total bilirubin, mg/dL	0.400 (0.20, 1.20)	0.400 (0.20, 0.50)	0.500 (0.20, 1.20)	0.700 (0.30, 1.30)
ALT (GPT), IU/L	13.0 (8, 55)	11.0 (9, 16)	22.0 (8, 55)	14.0 (6, 47)
AST (GOT), IU/L	21.0 (12, 37)	21.0 (17, 32)	21.0 (12, 37)	16.0 (10, 36)
GGT, IU/L	18.0 (9, 59)	14.0 (10, 19)	21.0 (9, 59)	16.0 (5, 380)
Triglycerides, mg/dL	91.0 (55, 538)	109.0 (68, 352)	91.0 (55, 538)	92.0 (42, 481)
Total cholesterol, mg/dL	155.0 (113, 260)	188.5 (143, 260)	142.0 (113, 207)	166.0 (109, 236)

Note: Data are presented as median (min, max) unless stated otherwise.

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BSA, body surface area; CLCr, creatinine clearance; eGFR, estimated glomerular filtration rate; GGT, gamma‐glutamyl transferase; GOT, glutamic oxaloacetic transaminase; GPT, glutamic pyruvic transaminase; PopPK, population pharmacokinetics.

3.2. Overview of plasma febuxostat concentration

The data set for the PopPK analysis included 2611 data points on plasma febuxostat concentration (110 data points for pediatric patients weighing <40 kg, 190 data points for pediatric patients weighing ≥40 kg, and 2311 data points for the adult population), using samples from a total of 142 subjects. Figure 1 shows a scatter plot of dose‐normalized plasma febuxostat concentrations against time after last dose. There was a certain amount of variability in plasma febuxostat concentrations for pediatric patients in the <40 kg and ≥ 40 kg categories, but those concentrations changed within a very similar range over time, just as in the adult population.

FIGURE 1.

Changes in dose‐normalized plasma febuxostat concentrations in pediatric patients and adult population.

3.3. Final PK model

Table 2 shows parameter estimates of the PopPK final model. The PK of febuxostat, in both the pediatric patients and the adult population, was described by a linear two‐compartment model, incorporating first‐order absorption with lag time and first‐order elimination. The population parameter estimates of the final model were 6.53 L/h for apparent total body clearance (CL/F), 19.4 L for apparent volume of distribution of the central compartment (V2/F), 1.80 L/h for apparent intercompartmental clearance (Q/F), 15.6 L for apparent volume of distribution of the peripheral compartment (V3/F), 3.43 h⁻¹ for the absorption rate constant (Ka), and 0.437 h for the absorption lag time (ALAG1). Between‐subject variability was described by an exponential error model and was estimated as coefficient of variation (CV)% of 26.2% for CL/F, 25.7% for V2/F, 71.3% for Q/F, 54.4% for V3/F, and 306.3% for Ka. Covariances were set for the between‐subject variability of CL/F‐V2/F and Q/F‐V3/F. Residual variability was described by an exponential error model and was estimated as CV% of 36.9%.

TABLE 2.

Estimated parameters of the final PopPK model of febuxostat in pediatric patients and adult population.

PK parameter	Estimate	CV (%)	RSE (%)	Shrinkage (%)	95% CI (Estimate ±1.96 × SE)	Bootstrap parameter estimate Mean ± SE	Median	95% CI (2.5th‐97.5th percentiles)
Fixed Effect
CL/F, L/h	6.53	–	3.0	–	6.14, 6.92	6.52 ± 0.197	6.52	6.13, 6.93
V2/F, L	19.4	–	4.3	–	17.8, 21.0	19.4 ± 0.792	19.4	17.9, 21.1
Q/F, L/h	1.80	–	7.2	–	1.55, 2.05	1.78 ± 0.127	1.79	1.52, 2.06
V3/F, L	15.6	–	5.3	–	14.0, 17.2	15.5 ± 0.822	15.6	13.8, 17.3
Ka, 1/h	3.43	–	24.6	–	1.78, 5.08	4.24 ± 2.78	3.51	2.47, 14.5
ALAG1, h	0.437	–	4.6	–	0.398, 0.476	0.441 ± 0.0210	0.438	0.406, 0.492
F1 (FED)	0.838	–	4.0	–	0.773, 0.903	0.838 ± 0.0274	0.839	0.780, 0.906
CLWGT	0.584	–	20.4	–	0.351, 0.817	0.589 ± 0.130	0.580	0.347, 0.853
CLEGFR	0.324	–	17.3	–	0.214, 0.434	0.326 ± 0.0586	0.326	0.207, 0.441
Between‐subject variability
ω CL/F 2	0.0662	26.2	17.5	6.0	0.0435, 0.0889	0.0655 ± 0.0106	0.0674	0.0430, 0.0894
ω CL/F‐V2/F 2	0.0483	–	21.5	–	0.0279, 0.0687	0.0484 ± 0.0112	0.0485	0.0274, 0.0719
(Correlation variability)
ω V2/F 2	0.0638	25.7	30.4	22.0	0.0258, 0.102	0.0648 ± 0.0188	0.0633	0.0312, 0.107
ω Q/F 2	0.411	71.3	17.7	14.5	0.268, 0.554	0.414 ± 0.0686	0.409	0.281, 0.575
ω Q/F‐V3/F 2	0.307	–	18.5	–	0.196, 0.418	0.304 ± 0.0507	0.302	0.209, 0.418
(Correlation variability)
ω V3/F 2	0.259	54.4	20.5	13.5	0.155, 0.363	0.254 ± 0.0511	0.251	0.165, 0.364
ω Ka 2	2.34	306.3	21.8	13.5	1.34, 3.34	2.60 ± 0.908	2.37	1.60, 5.80
ω ALAG1 2	0, FIX	–	–	–	–	–	–	–
ω F1 (FED) 2	0, FIX	–	–	–	–	–	–	–
Residual variability
σ 2 (exponential error)	0.136	36.9	6.2	6.5	0.120, 0.152	0.135 ± 0.00823	0.135	0.120, 0.153

Note: CV (%) for between‐subject variability = SQRT (EXP (ω²)–1) × 100. CV (%) for residual variability = σ × 100.

Abbreviations: ALAG1, absorption lag time; CI, confidence interval; CL/F, apparent total body clearance; CLEGFR, covariate for effect of eGFR on CL/F; CLWGT, covariate for effect of body weight on CL/F; CV, coefficient of variation; F1(FED), relative bioavailability at postprandial; Ka, first‐order absorption rate constant; PK, pharmacokinetics; PopPK, population pharmacokinetics; Q/F, apparent inter‐compartment clearance; SE, standard error; RSE, relative standard error; V2/F, apparent central volume of distribution; V3/F, apparent peripheral volume of distribution; ω _X ², between‐subject variability or covariance of parameter X; σ ², residual variability.

Among the candidate PK covariates for febuxostat, we identified body weight and eGFR on CL/F, and fasted/fed status on bioavailability (F), as significant covariates. Age was not included as a significant covariate. The effects of body weight and eGFR on CL/F were modeled using power functions that were centered around their respective medians:

$$\mathrm{CL}/F_i=6.53\times {\left({\mathrm{WGT}}_i/60.6\right)}^{0.584}\times {\left({\text{EGFR}}_{\mathrm{i}}/98.6\right)}^{0.324}$$

The effect of fasted/fed status on F was described as the ratio of F when fed (F1: relative bioavailability) to F when fasted (fixed value of 1):F (FASTED) = 1F1 (FED) = 0.838

3.4. Model evaluation

The parameters for the PopPK final model were estimated with good precision, with RSE <25%, except for the between‐subject variability in ω _V2/F ² (RSE = 30.4%). The η‐shrinkage and ε‐shrinkage were within acceptable limits, <30% for each (Table 2). The GOF plot (Figure S1) confirmed that the model‐predicted values were consistent with observed values and that the residuals were independent both of time after last dose and of predicted values. The success rate of bootstrap analysis (1000 simulations) was 94.1%, and the parameter estimates were within the 95% confidence interval (CI) for that analysis (Table 2). In the pc‐VPC, the median observed values were generally within the 95% CI of the median model prediction, and the majority of the observed values were distributed within the 95% CI of the 2.5th percentile and the 97.5th percentile in the model predictions. These findings demonstrated the good predictive performance of the final model (Figure 2).

FIGURE 2.

pc‐VPC for PopPk final model of febuxostat in pediatric patients and adult population. Blue circle: measured values; red solid line: median of measured values; red dashed line: 2.5th and 97.5th percentiles of measured values; red area: 95% CI for median of predicted values from 1000 simulations; blue area: 95% CI for 2.5th and 97.5th percentile points of predicted values from 1000 simulations. CI, confidence interval; pc‐VPC, prediction‐corrected visual predictive check; PopPK, population pharmacokinetics.

3.5. Sensitivity analysis of the effect of covariates on PK

Sensitivity analysis was performed for the effect of body weight and eGFR on the CL/F of febuxostat (Figure 3). Compared with subjects having median body weight (60.6 kg) and normal renal function (eGFR = 98.6 mL/min/1.73 m²: median), CL/F decreased with decreasing body weight, from 119.8% at 82.6 kg (95th percentile) to 76.6% at 38.4 kg (5th percentile). CL/F was reduced by 14.9% for mild renal dysfunction (eGFR = 60 mL/min/1.73 m²), 27.5% for moderate renal dysfunction (eGFR = 30 mL/min/1.73 m²), and 45.7% for severe renal dysfunction (eGFR = 15 mL/min/1.73 m²) (Figure 3).

FIGURE 3.

Effect of covariate on PK parameters of febuxostat (forest plot). Sensitivity analysis was conducted for 1000 hypothetical subjects for each condition. The reference value used the steady‐state PK parameters of febuxostat 20 mg administered orally once daily, set for fasted subjects weighing 60.6 kg with normal renal function (eGFR = 98.6 mL/min/1.73 m²). The mean and the 95% CI for the variability ratio compared to the PK parameters in the reference subjects are shown in this figure. CI, confidence interval; eGFR, estimated glomerular filtration rate; PK, pharmacokinetics.

After adjusting for body weight, CL/F showed no obvious differences between pediatric patients and the adult population in each of the renal function categories (normal, mild dysfunction, moderate dysfunction) for which comparison was possible (Figure 4).

FIGURE 4.

Relationship between renal function and CL/F adjusted by body weight (post hoc estimates) for febuxostat. (A) Pediatric patients, (B) Adult population. In this box plot, lines at the bottom, top, and middle of each box represent first quartile, third quartile, and median, respectively. The whiskers represent first quartile − 1.5 × IQR and third quartile + 1.5 × IQR, where IQR is the interquartile range (third quartile−first quartile). Renal function was grouped by eGFR classification. Normal: ≥90 mL/min/1.73 m² (Grade 1), mild: ≥60–<90 mL/min/1.73 m² (Grade 2), moderate: ≥30–<60 mL/min/1.73 m² (Grade 3), and severe: ≥15–<30 mL/min/1.73 m² (Grade 4). CL/F, apparent total body clearance; eGFR, estimated glomerular filtration rate; IQR, interquartile range.

3.6. PK parameters in pediatric patients and adult population

Post hoc estimates of steady‐state PK parameters for febuxostat in pediatric patients and the adult population are summarized in Table 3. The post hoc estimates for the pediatric patients in the present study were based on the maintenance doses that were originally used in that Phase 2 studies (20 mg for subjects weighing <40 kg and 40 mg for subjects weighing ≥40 kg), because the Phase 2 studies included dose escalation for those pediatric patients who required it.

TABLE 3.

PK parameters of febuxostat at steady state in pediatric patients and adult population (post hoc estimates).

	Pediatric patients		Adult population a
	<40 kg b (n = 10)	≥40 kg c (n = 19)	(n = 113)
Body weight, kg	36.9 (26.7, 39.9)	52.7 (41.1, 94.3)	61.3 (48.1, 86.5)
eGFR, mL/min/1.73m2	66.4 (41.4, 114.9)	84.1 (33.9, 145.3)	99.5 (20.9, 143.6)
Dose‐normalized C max,ss, ng/mL/mg	42.7 ± 7.6	35.7 ± 10.4	30.2 ± 10.6
Dose‐normalized AUCτ,ss, ng·h/mL/mg	251 ± 72	176 ± 75	128 ± 32
t max, h	1.43 ± 0.40	1.35 ± 0.51	1.46 ± 0.84
t 1/2, h	9.53 ± 0.58	8.71 ± 1.60	7.98 ± 0.97
CLss/F, L/h	4.28 ± 1.20	6.59 ± 2.45	8.27 ± 2.00
V z,ss/F, L	58.4 ± 15.0	79.3 ± 22.8	94.5 ± 22.4

Note: Data are presented as median (min, max) or mean ± SD. Data estimated for the following conditions: dose for each study administered once daily to fed pediatric patients and adults.

Abbreviations: AUC_τ,ss, area under the concentration‐time curve during a dosage interval (τ) at steady state; CL_ss/F, apparent total body clearance at steady state; C _max,ss, maximum drug concentration at steady state; eGFR, estimated glomerular filtration rate; PK, pharmacokinetics; SD, standard deviation; t _max, time‐to‐maximum drug concentration; t _1/2, elimination half‐life; V _z,ss/F, apparent volume of distribution during terminal phase at steady state.

^a Dose of febuxostat is 10–160 mg.

^b Dose of febuxostat is 20 mg.

^c Dose of febuxostat is 40 mg.

Dose‐normalized C _max,ss estimates in pediatric patients weighing <40 kg were higher than in the adult population (1.41‐fold) and in pediatric patients weighing ≥40 kg were similar to the adult population (1.18‐fold). Dose‐normalized AUC_τ,ss estimates were 1.96‐ and 1.38‐fold higher in pediatric patients weighing <40 kg and ≥40 kg, respectively, than in the adult population. Body weights and eGFR were lower in the pediatric patients than in the adult population, and CL_ss/F values in pediatric patients weighing <40 kg and ≥40 kg were 0.52‐fold and 0.80‐fold, respectively, that of the adult population.

3.7. Simulation of the relationship between exposure and body weight

For each dose of febuxostat in the Phase 2 studies in pediatric patients, the relationship between body weight and steady‐state exposure (C _max,ss and AUC_τ,ss) was examined by Monte Carlo simulations. Note that the PopPK final model was applicable to pediatric patients as well as to the adult population, since age was not included as a covariate, and that the clinical dose for pediatric patients weighing ≥40 kg was the same as the adult dose, so pediatric patients weighing ≥40 kg and the adult population were grouped together. To improve comparability, the renal function condition was standardized to the normal median eGFR value of 98.6 mL/min/1.73 m². Figure 5 showed simulated results obtained for the relationship between body weight and exposure (median and 90% prediction interval) at the maintenance dose (20 mg for body weight <40 kg and 40 mg for body weight ≥40 kg). In fed pediatric patients weighing 20–40 kg who received 20 mg of febuxostat once daily by oral administration, the 90% prediction interval for steady‐state exposure (C _max,ss and AUC_τ,ss) is generally included within that for fed pediatric patients and adults weighing 40–120 kg who received 40 mg febuxostat in the same manner. Similar exposure‐weight relationships were observed for other pediatric doses (5, 10, and 30 mg for body weight <40 kg; 10, 20, and 60 mg for body weight ≥40 kg) (data not shown).

FIGURE 5.

Relationship between exposure levels for maintenance dose of febuxostat at steady state (C _max,ss and AUC_τ,ss) and body weight. (A) Relationship between C _max,ss and body weight. (B) Relationship between AUC_τ,ss and body weight. Simulation results for febuxostat administered once daily to fed patients. Dose: 20 mg for pediatric patients <40 kg, 40 mg for pediatric patients ≥40 kg and adults. Normal renal function: eGFR = 98.6 mL/min/1.73 m² (median of PopPK final model). Plotted data: median of predicted values for C _max,ss and AUC_τ,ss; gray area: 90% prediction interval in 1000 hypothetical subjects. AUC_τ,ss, area under the concentration‐time curve during a dosage interval (τ) at steady state; C _max,ss, maximum drug concentration at steady state; eGFR, estimated glomerular filtration rate; PopPK, population pharmacokinetics.

4. DISCUSSION

We created an integrated PopPK model of febuxostat in pediatric patients with hyperuricemia including gout and an adult population (healthy or with renal dysfunction) in Japan. We could express it as a linear two‐compartment model that incorporated first‐order absorption and first‐order elimination with lag time. Statistically significant covariates for febuxostat PK were identified as body weight and eGFR on CL/F and fasted/fed status on bioavailability, as detailed in Table 2.

Previously reported PopPK models differed in the incorporation of body weight as a covariate for febuxostat CL/F. ³⁷ , ³⁸ , ³⁹ , ⁴⁰ This could be because the weight data from the analyzed subjects showed a biased distribution toward values above the standard of 60–70 kg, potentially leading to failure to detect the significance of the covariate effects of body weight. In this PopPK analysis, we were able to evaluate the effects of body weight on CL/F for a wide range of weight data by integrating data from pediatric patients.

The effects of renal function on febuxostat PK have been investigated in several studies, but with somewhat inconsistent results. ¹² , ²⁷ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ In our previous Phase 1 study of febuxostat (Study 6) ¹⁶ in Japanese subjects with normal renal function or renal dysfunction based on CLCr, the mild dysfunction group did not differ noticeably from the normal group in steady‐state plasma C _max,ss, but AUC_τ,ss increased by 52.5% compared with the normal group, and in the moderate renal dysfunction group C _max,ss increased by 25.5% and AUC_τ,ss by 67.6% compared with the normal group, based on findings from non‐compartmental analysis (NCA). In the present PopPK model, the impact of renal function on CL/F was represented by incorporating eGFR, which showed a greater reduction in the OFV than when using CLCr.

Sensitivity analysis in this PopPK model showed that the rate of increase in C _max,ss and AUC_τ,ss in patients with mild or moderate renal dysfunction was <50% of that in subjects with normal renal function. The somewhat reduced effect of renal function on exposure that was noted in the sensitivity analysis, as compared to the Phase 1 study results, could be attributed to two factors. First, when the criteria for renal function categories were changed from CLCr‐based for the Phase 1 study to eGFR‐based for PopPK analysis, some of the Phase 1 subjects were reclassified, for example from normal to mild or from moderate to severe dysfunction. Second, patients with mild or moderate renal dysfunction had lower body weights than those with normal renal function in the Phase 1 study. In addition, as shown in Figure 4, there were no clear differences in CL/F when adjusted by body weight between pediatric patients and the adult population in each of the renal function categories (normal, mild dysfunction, and moderate dysfunction based on eGFR) for which comparisons were possible. These findings suggest that the rate of increase in exposure in patients with renal dysfunction was similar between pediatric patients and the adult population. The results thus support the use of febuxostat without dose adjustment for mild or moderate renal dysfunction, in pediatric patients and adults. One reason that febuxostat CL/F may have been affected by renal function, despite the low urinary excretion rate of unchanged febuxostat (<5%), is because the acylglucuronide conjugate, which is the main metabolite of febuxostat and excreted primarily by the kidneys, may accumulate in patients with renal dysfunction as Kamel et al. ⁴⁰ , ⁵¹ suggested, and the subsequent hydrolysis may regenerate unchanged forms, reducing the total systemic clearance of febuxostat.

Previous reports on the effect of fasted/fed status on the PK of febuxostat have shown consistent results across studies. ²⁹ , ³⁰ In our previous clinical study evaluating the effect of fasted/fed status in healthy Japanese adult males (Study 3), ⁴³ NCA analysis confirmed that C _max and AUC from time zero to infinity (AUC_inf) in the fed group were 28% and 18% lower, respectively, than in the fasted group, and the present PopPK analysis suggests that relative bioavailability, AUC_τ,ss, and C _max,ss were each reduced by 16.2% in the fed group. In contrast, the estimated Ka exhibited substantial between‐subject variability, with a high CV% of 306.3%, and bootstrap analysis showed a wide nonparametric 95% CI, ranging from 2.47 to 14.5 h⁻¹. A possible reason for the low estimation accuracy of Ka is that fasted/fed status was not included as a covariate on Ka. In general, acidic drugs are known to be easily influenced by the gastric emptying rate, ⁵² , ⁵³ and febuxostat is a weakly acidic and poorly soluble class II drug in the biopharmaceutical classification system (BCS), which may have complicated the estimation of Ka and the effects of food (fasted/fed status).

In examining covariates for the PopPK model, sex was not considered as a candidate covariate because there were only 9 females (6 children and 3 adults) compared to 133 males (23 children and 110 adults). However, the post hoc estimate of CL/F from the PopPK final model, adjusted by body weight, showed no clear difference between males and females for either children or adults (Figure S2). Liver function was not considered as a covariate candidate because the distribution of liver function parameters (ALT, AST, total bilirubin, and GGT) was generally within the normal range and did not correlate with the post hoc estimated PK parameters. Previous reports on standard NCA analysis of febuxostat have reported that sex ¹⁵ , ³⁰ , ³³ , ³⁶ and mild (Child‐Pugh class A) or moderate (Child‐Pugh class B) hepatic dysfunction ¹³ have no notable effect on PK. Therefore, we believe that the exclusion of sex and liver function from the candidate covariates does not affect the predictive performance of the PopPK model.

As shown in Figure 5, simulations using the integrated PopPK model show that the range of steady‐state exposures (C _max,ss and AUC_τ,ss) for the maintenance dose of 20 mg febuxostat in fed pediatric patients weighing 20–40 kg are generally included in those for the maintenance dose of 40 mg febuxostat in fed pediatric patients and adult population weighing 40–120 kg. These findings suggest that dose setting of half the adult dose is appropriate for pediatric patients weighing <40 kg and the full adult dose for pediatric patients weighing ≥40 kg, to obtain the same level of exposure in pediatric patients as in the adult population. A recent Phase 2 studies of Japanese pediatric patients with hyperuricemia including gout showed that febuxostat was as effective and well tolerated in pediatric patients weighing <40 kg and ≥40 kg as in adults. Those findings indicate the validity of body weight‐based categories and a 40‐kg threshold for dosing pediatric patients, not only from the PK perspective, but also in terms of efficacy and safety. ⁴²

Japanese pediatric patients with hyperuricemia including gout have a high prevalence of complications associated with renal dysfunction. We hope that the results of this study will help to encourage the appropriate use of febuxostat in those patients.

5. CONCLUSION

In this study, we developed the first integrated PopPK model of febuxostat in a Japanese population including pediatric patients. We then used that model to validate the appropriateness of dosage and administration for febuxostat in pediatric patients with hyperuricemia including gout from a PK perspective. The findings confirmed that, for pediatric patients weighing <40 kg, half the adult dose is effective, while those weighing ≥40 kg will benefit from the standard adult dose, taken once daily. The results also substantiated the appropriateness of febuxostat administration without dose adjustment in the relatively high percentage of pediatric patients who have mild to moderate renal dysfunction, a finding that has previously been demonstrated in adults.

AUTHOR CONTRIBUTIONS

R. I. contributed conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, validation, visualization, writing—original draft and writing—review and editing. K. N. and D. I. contributed conceptualization, formal analysis, investigation, methodology, software, validation, visualization and writing–review and editing. T. I. contributed conceptualization, resources, supervision and writing—review and editing. All authors read and approved the final manuscript and agreed to be accountable for all aspects of the manuscript.

FUNDING INFORMATION

This study was supported by Teijin Pharma Limited.

CONFLICT OF INTEREST STATEMENT

R.I., K.N., D.I., and T.I. are currently employees of Teijin Pharma Limited.

ETHICS STATEMENT

Ethics approval is not required because this study uses PopPK analysis of data from several clinical trials.

CLINICAL TRIAL REGISTRATION

Not applicable because this study uses PopPK analysis of data from several clinical trials.

PRINCIPAL INVESTIGATOR STATEMENT

Principal investigators were not involved in this analysis.

Supporting information

Data S1.

Iwama R, Nishida K, Ishii D, Iijima T. An integrated population pharmacokinetic model of febuxostat in pediatric patients with hyperuricemia including gout and adult population of healthy subjects and patients with renal dysfunction. Pharmacol Res Perspect. 2024;12:e70032. doi: 10.1002/prp2.70032

DATA AVAILABILITY STATEMENT

The data set analyzed for this study is not currently available for data sharing.