Population pharmacokinetics of atazanavir/ritonavir in HIV-1-infected children and adolescents

Abstract

AIMS

To investigate atazanavir (ATV) population pharmacokinetics in children and adolescents, establish factors that influence ATV pharmacokinetics and investigate the ATV exposure after recommended doses.

METHODS

Atazanavir concentrations were measured in 51 children/adolescents during a mean therapeutic monitoring follow up of 6.6 months. A total of 151 ATV plasma concentrations were obtained, and a population pharmacokinetic model was developed with NONMEM. Patients received ATV alone or boosted with ritonavir.

RESULTS

Atazanavir pharmacokinetics was best described by a one-compartment model with first-order absorption and elimination. The effect of bodyweight was added on both apparent elimination clearance (CL/F) and volume of distribution using allometric scaling. Atazanavir CL/F was reduced by ritonavir by 45%. Tenofovir disoproxil fumarate (TDF) co-medication (300 mg) increased significantly by 25% the atazanavir/ritonavir (ATV/r) CL/F. Mean ATV/r CL/F values with or without TDF were 8.9 and 7.1 L h⁻¹ (70 kg)⁻¹, respectively. With the recommended 250/100 mg and 300/100 mg ATV/r doses, the exposure was higher than the mean adult steady-state exposure in the bodyweight range of 32–50 kg.

CONCLUSIONS

To target the mean adult exposure, children should receive the following once-daily ATV/r dose: 200/100 mg from 25 to 39 kg, 250/100 mg from 39 to 50 kg and 300/100 mg above 50 kg. When 300 mg TDF is co-administered, children should receive (ATV/r) at 250/100 mg between 35 and 39 kg, then 300/100 mg over 39 kg.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Population pharmacokinetics has been studied in adult patients. Tenofovir disoproxil fumarate (TDF) co-medication could decrease atazanavir/ritonavir (ATV/r) exposure.

WHAT THIS STUDY ADDS

• This first ATV/r population analysis in children and adolescents provides useful insights into the fate of this drug and shows that the allometric scaling of clearance and volume parameters based on bodyweight reduced the associated between-subject variability. Moreover, the effects of combined ritonavir and tenofovir on ATV clearance are precisely quantified.

Introduction

Protease inhibitors (PIs) represent a very important part of the highly active antiretroviral therapies in the treatment of HIV-infected adults [1]. Atazanavir (ATV) was approved by the US Food and Drug Administration (FDA) in 2003 [2] and in Europe in 2004 [3]. Atazanavir has demonstrated a long-term suppression of viral replication in HIV-experienced and naïve infected adults [4–7]. However, there is a lack of data on the children/adolescent population, while the number of children treated by highly active antiretroviral therapies is increasing [8]. Indeed, the dosage recommendation by bodyweight or age could be insufficient to take into account all the differences between adults and children, i.e. the metabolic pathways (expression of CYP 450 isoenzymes) or the effects of physiological modifications such as puberty on drug pharmacokinetics. Therefore, studies on atazanavir/ritonavir pharmacokinetics in children are necessary to ensure that dosage prescription on a bodyweight basis is safe and efficacious. At the present time, only one study (PACTG 1020A) is available on the atazanavir pharmacokinetics in paediatric patients [9]. The FDA dosing recommendations are derived from this study [10].

Atazanavir alone is prescribed at a dose of 400 mg once daily in adults. This PI is more frequently co-administered with low-dose ritonavir (RTV) to enhance atazanavir patient's exposure (atazanavir/ritonavir, ATV/r): 100 mg of RTV increases by two- to threefold the mean atazanavir minimal concentrations, relative to ATV 400 mg alone [11]. The recommended ATV/r adult dosage is 300/100 mg taken once daily with food for a better bioavailability [12]. The recommended dosage of ATV/r for paediatric patients (6 years to less than 18 years old) is based on bodyweight and should not exceed the recommend adult dosage. For treatment-naïve paediatric patients, the FDA recommends the following once-daily ATV/r doses: 150/80 mg (15–25 kg), 200/100 mg (25–32 kg), 250/100 mg (32–39 kg) and 300/100 mg (>39 kg). For treatment-experienced patients weighing at least 25 kg, the recommended ATV/r dosage is similar to the recommendations for treatment-naïve patients [10].

Atazanavir is rapidly absorbed, and the concentration peak is reached 2.5–3 h after oral administration. Atazanavir is highly bound to albumin (89%) and α1-acid glycoprotein (86%). Atazanavir is metabolized by the cytochrome P450 (CYP) isoenzymes 3A4 and 3A5. The mean ATV half-life at steady state is around 8.8 h for an adult drug dosage of 300/100 mg once daily [13].

Atazanavir administration is often associated with hyperbilirubinaemia; indeed, an indirect elevation of unconjugated bilirubin is related to the inhibition of UDP-glucuronosyltransferase 1A1 [12, 14, 15].

Tenofovir disoproxil fumarate (TDF) is known to decrease the exposure of atazanavir; thus, when combined with 300 mg TDF, ATV should be co-administered with 100 mg of ritonavir in children aged at least 12 years and weighing at least 35 kg. The mechanism of interaction between atazanavir and tenofovir remains unknown [16, 17].

This observational study is based on therapeutic drug monitoring in children and adolescents. The aims of this study were as follows: (i) to investigate the population pharmacokinetics of ATV in children and adolescents; (ii) to investigate the factors that influence ATV pharmacokinetics in this population; (iii) to compare ATV/r pharmacokinetic parameters with previous reports in adults; and (iv) to simulate ATV concentration–time profiles based on current dosing recommendations and compared the results with ATV exposure at steady state in adults.

Methods

Patients

Data were collected from children and adolescent patients experienced with antiretroviral (ARV) treatments. The ARV therapy was monitored on a routine basis for each sampling. The drug, time after the last dosing, bodyweight, age and combined treatments were recorded on the same day. The samples were not collected on the same occasion. The median (minimum – maximum) number of samples per patient was 2 (1–13).

The assay for atazanavir was performed using a high-pressure liquid chromatography method. The quantification limit of the method was 0.10 mg l⁻¹, with an interassay precision and a bias of less than 15% and 5%, respectively, in the 0.05–5 mg l⁻¹ calibration range [18].

Population pharmacokinetic modelling of ATV/r

The data were analysed using the nonlinear mixed effect modelling program NONMEM (version VI) [19] driven by Wings for Nonmem (http://wfn.sourceforge.net). The first-order method with the INTERACTION option was used. To handle the concentration below the limit of quantification (LOQ), the following three different approaches were tried: (i) replacement of the first LOQ observation with LOQ/2 and deletion of the following concentrations below the LOQ; (ii) use of the built-in Beal M2 method; or (iii) use of the Beal M3 method [20]. Different structural models for ATV/r pharmacokinetics were investigated: one or two compartments with linear elimination and first-order or zero-order absorption, with or without a lag time or a transit compartment for absorption. As ATV/r was exclusively given by the oral route, clearance (CL) and volume of distribution (V) are apparent parameters, V/F and CL/F, where F is the unknown bioavailability fraction.

Several error models (proportional, additive or mixed) were investigated to describe the residual variability (ε). The between-subject variabilities (η or BSVs, expressed as the square root of the ω estimate) were assumed to be exponential. Only significant BSVs on pharmacokinetic analysis were kept.

The main covariates of interest in the population were age, sex, bodyweight (BW), ritonavir and co-medication, such as tenofovir. Parameter estimates were standardized for a mean standard BW using an allometric model, as follows:

where P_STD is the standard value of parameter for a patient with the standard BW value, and P_i and BW_i are the parameter and bodyweight of the ith individual. The PWR exponents may be estimated from the data. However, from allometric scaling theory these are typically 0.75 for clearance and 1 for volume of distribution [21].

The effect of each patient covariate (sex, ritonavir and tenofovir co-medication) was systematically tested via the linear model on the typical value of a given parameter θ; e.g. θ = θ_a(1 +θ_b×X), where θ_a is the typical value and θ_b is the estimated influential factor for the categorical covariate X. Age was tested by using the Hill equation [21]. All the covariates were tested via an upward model building. A covariate was selected if it met the following criteria: (i) its effect was biologically plausible; (ii) it produced a minimum decrease of 6.63 units (χ², 1 d.o.f., 0.01) in the objective function value; and (iii) it produced a reduction in the variability of the pharmacokinetic parameter, assessed by the associated intersubject variability.

Graphical evaluation of the goodness of fit was mainly assessed by observed vs. predicted concentrations and weighted residuals vs. time and/or weighted residuals vs. predicted concentrations. The final population model was also appreciated by the normalized prediction distribution errors metrics [22] and the visual predictive check. The stability of the model and accuracy of the parameters were assessed by a bootstrap method implemented in Wings for Nonmem, and diagnostic graphics and distribution statistics were obtained using RfN (link on http://wfn.sourceforge.net) via the R program [23].

Individual Bayesian estimates of the pharmacokinetic parameters were used to calculate the individual area under the concentration–time curve from time 0 to 24 h (AUC_0–24) and the trough concentration (C_trough) of atazanavir.

Using the final population model, simulations were performed (1000 simulations of the database) in order to determine the ATV/r once-daily dosing according to bodyweight that provided the mean adult steady-state AUC_0–24 obtained with an administration of 300/100 mg. The mean adult AUC_0–24 value and its 90% confidence interval (CI) were determined by a weighted average of three different studies [15, 16, 24].

Results

Demographic data

The median (range) follow-up duration of therapeutic drug monitoring for the 51 children (25 girls and 26 boys) was 2.3 months (0–34 months), and 151 ATV concentrations were available for pharmacokinetic evaluation. The median age was 14 years (4.5–18 years), and the median bodyweight was 52 kg (17–72 kg; Table 1). The median ATV/r and ATV doses were 300/100 and 400 mg, respectively. Thirty-nine patients were treated with ATV/r, nine with ATV, and three subjects received both therapies successively. Twenty-one children treated with ATV/r also received tenofovir.

Table 1.

Characteristics of the 51 children (25 girls, 26 boys)

Baseline	Mean (SD)	Median	Range
Age (years)	14.3 (3.8)	14	3–18
Weight (kg)	51 (15)	52	13–79
Dose of ATV/r (mg)	283 (50)	300	100–400
Dose of ATV (mg)	417 (144)	400	150–600
Dose of TDF (mg)	285 (43)	300	150–300
Time of follow-up (months)	6.6 (9)	2.3	0–39

ATV/r, atazanavir/ritonavir; ATV, atazanavir; TDF, tenofovir disoproxil fumarate.

Population pharmacokinetics

Four concentrations were below the LOQ, corresponding to less than 3% of the total observations. The M3 method and the built-in M2 method did not modify the parameter estimates; thus, we kept the method of setting these concentrations to half of the LOQ.

A one-compartment model with first-order absorption and elimination adequately described the data. Between-subject variability was retained only for apparent clearance. The proportional model for the residual variability ensured a good adequacy between observed and predicted values.

The use of the allometric scaling on CL/F and V/F decreased the objective function value by 26.9 units and the CL/F BSV from 0.56 to 0.46.

Children received two different drug dosages, with or without ritonavir used as booster. The effect of the absence of ritonavir was tested on CL/F in the model and was statistically significant, decreasing the objective function value by 30.7 units and the BSV on CL/F to 0.189. A significant effect of TDF as a co-treatment was also found on CL/F, decreasing the objective function value by 7.2 units and the BSV on CL/F from 0.189 to 0.164. The empirical Bayesian estimate of CL/F shrinkage was 0.23. Finally, ritonavir decreased CL/F by 45% (from 12.8 to 7.1 l h⁻¹ (70 kg)⁻¹), and when associated with ATV/r, TDF increased CL/F by 25% (from 7.1 to 8.9 l h⁻¹ (70 kg)⁻¹). Based on 1000 simulations of the final model, TDF co-medication decreased ATV/r C_trough values by 37%.

Table 2 summarizes the final population pharmacokinetic estimates, including the bootstrap verification. All parameters were well estimated, and the bootstrap confidence intervals were reasonably narrow and did not include zero.

Table 2.

Population pharmacokinetic parameters of atazanavir in 51 children (3–18 years old) receiving atazanavir (ATV) with ritonavir (RTV) and bootstrap statistics^*

	Final model original dataset	Bootstrap*
Parameter	Mean (RSE %)	Median (90% CI)
Structural model
CL/F (l h−1 (70 kg)−1)†	7.1 (8)	7.2 (6.1–9.2)
θNO_RTV, CL/F	0.80 (24)	0.80 (0.52–1.4)
θTDF, CL/F	0.25 (37)	0.26 (0.07–0.49)
V/F (l (70 kg)−1)	103 (19)	98 (27–146)
Ka (h−1)	0.44 (26)	0.44 (0.1–0.8)
Statistical model
BSV (CL/F)	0.16 (34)	0.15 (0.04–0.51)
σproportional	0.53 (15)	0.51 (0.43–0.59)

Key: RSE%, relative standard error (standard error of estimate/estimate × 100); K_a, absorption rate constant; CL/F, apparent elimination clearance for ATV in presence of ritonavir; V/F, apparent central volume of distribution; σ, residual variability estimates; BSV, between-subject variability estimates; θ_{NO_RTV}, influential factor on ATV CL/F without RTV; and θ_TDF, influential factor on ATV CL/F with tenofovir.

^*Statistics from 1000 bootstrap analyses.

^†The typical parameters refer to a patient weighing 70 kg according to an allometric model: [Typical value] = [Typical parameter]× (bodyweight/70)^PWR, where PWR = 0.75 for CL and 1 for V terms. CL/F = CL/F× (BW/70)^0.75× (1 +θ_{NO_RTV}) × (1 +θ_TDF). For example, without RTV, mean CL/F is 7.1 × 1.80 = 12.8 l h⁻¹ (70 kg)⁻¹ and with TDF, mean CL/F = 7.1 × 1.25 = 8.9 l h⁻¹ (70 kg)⁻¹.

The visual predictive check performed on the final model showed that the average prediction obtained from 1000 Monte Carlo simulations matched the observed concentration time courses for the ATV/r and ATV/r + TDF regimens. Patients received different dosages, so the observed and simulated concentrations were normalized for a 300 mg median dosage. Accordingly, 12.2 and 3.5% (exact binomial test 95% confidence intervals of 6–22 and 0.4–12%) of the ATV/r and ATV/r + TDF observations were outside the 90% confidence limits (Figure 1). The mean and variance of the normalized prediction distribution errors metrics were not significantly different from 0 (P = 0.35) and 1 (P = 0.58), and their distribution was not different from a normal one (P = 0.08; global adjusted P value, P = 0.234). Table 3 summarizes AUC_0–24, C_trough from the present and previous studies for a median once-daily ATV/r dose of 300/100 mg, with or without 300 mg of TDF.

Figure 1.

Visual predictive check standardized to a 300-mg atazanavir/ritonavir with (a) and without tenofovir disoproxil fumarate dosage (b). Predicted percentiles are based upon 1000 Monte Carlo simulations of the final model. Continuous lines represent the median (red curve) and confidence interval limits: 5th and 95th percentiles (blue curves). Dotted lines include the 95% confidence interval for each curve

Table 3.

Comparison of atazanavir/ritonavir (ATV/r) (300/100 mg, median)-derived pharmacokinetic parameters with or without 300-mg tenofovir disoproxil fumarate (TDF), between our study and previous adult studies

	Present study standardized for 70 kg ATV/r 300/100 mg	Previous adult studies ATV/r 300/100 mg
Parameter	Geometric mean (95% CI)	Geometric mean (95% CI)	Study (reference)
ATV/r
AUC0–24 dosing (mg h l−1)	41.6 (31.8–51.3)	44.5 (15.1–73.9)	Burger et al. [24]
Ctrough (mg l−1)	0.75 (0.37–1.1)	0.71 (0–1.5)	Burger et al. [24]
ATV/r + TDF 300 mg
AUC0–24 dosing (mg h l−1)	32.8 (25.2–39.1)	28.6 (2.3–54.9)	Zhu et al. [34]
Ctrough (mg l−1)	0.47 (0.24–0.74)	0.49 (NA)	Taburet et al. [31]

NA, not applicable.

Figure 2a displays the median ATV/r AUC_0–24 values derived from 1000 simulations of the final model using the FDA-recommended once-daily doses of 200 mg (25–32 kg), 250 mg (32–39 kg) and 300 mg (>39 kg) according to bodyweight. The horizontal lines represent the mean adult exposure of 44.6 mg h l⁻¹ and the 90% confidence interval (23–66.2, dotted lines). Different dosing schemes were simulated to obtain exposures close to 44.6 mg h l⁻¹. As shown in Figure 2b, optimal ATV/r dosages were 200/100, 250/100 and 300/100 mg for children weighing 25–39, 39–50 and above 50 kg, respectively. These proposals were compared with the FDA recommendation (Table 4); they were different only in the group 32–50 kg, without TDF. When 300 mg TDF is used as a co-treatment, there is no FDA recommendation to modify the current ATV/r dose. Thus, 250/100 mg is recommended for the bodyweight range 35–39 kg and 300/100 mg over 39 kg. These doses were simulated, and the mean (CI) AUC_0–24 calculated was 45.9 (32.9–62.1) mg h l⁻¹ for ATV/r 250/100 mg with TDF and 41.7 (27.5–61.4) mg h l⁻¹ for ATVr 300/100 mg with TDF (Figure 2c). These values were close to the target mean adult AUC_0–24.

Figure 2.

Calculated atazanavir (ATV) AUC_0–24 (continuous line) and 90% CI (dashed lines) vs. bodyweight in patients with Food and Drug Administration (FDA)-recommended doses for atazanavir/ritonavir (ATV/r) without tenofovir disoproxil fumarate (TDF) (a), in patients with doses proposed in this study for ATV/r without TDF (b) and in patients with (FDA-recommended = proposed) doses for ATV/r with TDF (c). The horizontal continuous and dotted lines represent the mean adult exposure of 44.6 mg h l⁻¹ and its 90% CI (23–66.2 mg h l⁻¹), respectively

Table 4.

Comparison of atazanavir/ritonavir dosage between Food and Drug Administration (FDA) recommendations and our proposals, with or without tenofovir disoproxil fumarate (TDF)

	25–32 kg	32–35 kg	35–39 kg	39–50 kg	>50 kg
Without TDF
FDA recommendation	200/100	250/100	250/100	300/100	300/100
Our study proposal	200/100	200/100	200/100	250/100	300/100
With TDF
FDA recommendation	Not recommended: TDF not prescribed		250/100	300/100	300/100
Our study proposal			250/100	300/100	300/100

Discussion

The pharmacokinetics of ATV/r was satisfactorily described by a one-compartment model with linear absorption and elimination. This modelling was based on therapeutic drug monitoring data, with 95% of the children weighing between 17 and 72 kg, with a median of 52 kg. Bodyweight was included in the model, using a 1 and ¾ allometric scaling for V/F and CL/F, respectively. This reduced the variability associated with CL/F from 0.56 to 0.46. The bodyweight allometric scaling finally removed the age effect on pharmacokinetic parameters. The influences of RTV and TDF co-medication on CL/F were significant. The CL/F value without ritonavir was 12.8 l h⁻¹ (70 kg)⁻¹, similar to previously reported values: 12.9, 13.2 and 12.6 l h⁻¹ when standardized to 70 kg [13, 25, 26]. The clearance estimated for ATV/r (7.1 l h⁻¹ (70 kg)⁻¹) was close to other reported values in adult patients: 6.2, 7.7 and 7.6 l h⁻¹[27–29], supporting the use of the allometric model. As previously described in adults, the effect of TDF co-treatment increased by 25% the ATV/r CL/F value and decreased by 37% the C_trough values [30–32].

The FDA recommendations for ATV/r in paediatric ARV-naïve patients, according to bodyweight, are once-daily doses of 200/100 mg (25–32 kg), 250/100 mg (32–39 kg) and 300/100 mg (>39 kg) [10]. Simulations of these doses in our children–adolescent population have shown that the mean adult AUC_0–24 value is reached for each bodyweight interval. Patients treated with ATV present increased unconjugated bilirubin levels. According to the European Medicines Agency (EMEA) report, hyperbilirubinaemia (grades 2–4) was reported in 65% of subjects, and 39% of them developed grade 3–4 total bilirubin levels [33]. The exposure obtained with the FDA dosing recommendations from 32 to 50 kg was higher than the mean adult AUC_0–24 target. This potential overexposure could increase the risks to develop high grades of hyperbilirubinaemia. Thus, lower doses into this bodyweight interval should provide a comparable efficacy with a decrease in toxicity. The following once-daily ATV/r doses appeared to be an appropriate alternative to the FDA recommendations: 200/100 mg (25–39 kg), 250/100 mg (39–50 kg) and 300/100 mg (>50 kg). According to our simulations, when combined with 300 mg TDF, the usual ATV/r dosing recommendations are 250/100 mg for children weighing 35–39 kg, then 300/100 mg for children weighing over 39 kg [16, 17]. Although the median age (bodyweight) of children was 14 years old (52 kg), 24 children weighed less than 50 kg, which allowed us to make suggestions for younger or lighter children. These dosing regimens resulted in 24-h exposures close to the mean adult values for a 300/100 mg ATV/r dosage. These suggestions should be prospectively confirmed.

Competing Interests

CD has received fees for speaking at several conferences by BMS. There are no other competing interests to declare.