The current preferred first-line antiretroviral therapy regimen for previously untreated children with no evidence of antiretroviral resistance comprises 2 nucleoside reverse transcriptase inhibitors (NRTIs) with either a nonnucleoside reverse transcriptase inhibitor or a ritonavir-boosted protease inhibitor. According to the World Health Organization recommendations, the preferred NRTIs combinations are lamivudine (3TC) + zidovudine (AZT), 3TC + abacavir or 3TC + stavudine. These therapies which are essential to reach virological control of the patient need a twice-daily administration. A once-daily tablet combining tenofovir and emtricitabine is increasingly popular for use in adult patients and may be useful in adolescents and children. Thus several clinical trials have assessed the safety and efficacy of this combination, which could be administered once daily to maintain virologic suppression in HIV-1–infected virologically controlled children. Indeed, this switch should improve the adherence of the child and reduce the cost of the treatment. However, no specific pediatric doses of tenofovir disoproxil fumarate (TDF) are available for children less than 35 kg; Food and Drug Administration recommends the adult dose of 300 mg per day for children beyond 35 kg and 12 years of age.
TDF is a NRTI oral prodrug of tenofovir. Tenofovir is rapidly absorbed and mainly eliminated unchanged in the urine by glomerular filtration and active tubular secretion.1 Although the pharmacokinetics of tenofovir is quite well described in adults,2–5 very few studies have reported tenofovir pharmacokinetics in children.6 However, the many developmental changes occurring during the newborn period, infancy, childhood, and adolescence could affect drug disposition. In young children, renal elimination is decreased during the first years of life, thus predictions cannot be made from adult data.
In the present study, we have developed a population pharmacokinetic model for tenofovir in children to study the influence of covariates [body weight (BW), age, cotreatments] on pharmacokinetics. The main goal was then to suggest for the first time the dose of TDF to give in children.
Patients and Treatment
The population comprised 93 pediatric patients, ranging in age from 5 to 18 years with a median [95% confidence interval (CI)] of 14 years (6.3 to 18); 12 patients were aged younger than 9 years. BW ranged from 20 to 82 kg with a median (95% CI) of 46 kg (22.5 to 72.2); 19 children weighted less than 30 kg. Children received 1 tablet (40% of patients) or one-half tablet (54%) of TDF (300 mg/tablet) for the treatment of HIV infection once a day, and tenofovir concentrations were monitored on a routine basis. Six patients switched from one half to 1 tablet during the follow-up time. The mean duration of tenofovir therapy was 1 year. For each patient, time elapsed between administration and sampling times; gender, BW, and age were carefully recorded and combined antiretroviral drugs. Serum creatinine clearance was calculated using the Schwartz equation. Samples were obtained at different occasion at each patient visit.
Among all the patients, 84.5% were treated with a ritonavir-boosted protease inhibitor (33% lopinavir, 28% atazanavir, 16% amprenavir, 3% saquinavir, 3% darunavir, and 1.5% nelfinavir). And 15.5% of patients were treated with nonnucleoside reverse transcriptase inhibitor (12.5% efavirenz and 3% nelfinavir). Patients received as a second NRTI: emtricitabine (30% of patients), abacavir (25%), 3TC (25%), didanosine (10%), AZT (7%), or stavudine (3%).
The tenofovir assay was performed according to a previously published method7 with a limit of quantification, interassay precision, and bias of 0.01 mg/L, 9.6 and 11.4%, respectively.
Modeling Strategy and Population Pharmacokinetic Model
Data were analyzed using the nonlinear mixed effect modelling software program Monolix version 31s http://wfn.software.monolix.org).8 Parameters were estimated by computing the maximum likelihood estimator of the parameters without any linearization of the model using the stochastic approximation expectation maximization algorithm combined to a Markov Chain monte Carlo procedure. The number of Markov Chain monte Carlo chains was fixed to 10 for all estimations. Several structural pharmacokinetic models were investigated. Data were analyzed according to a 1-compartment or 2-compartment model. Between-subject variabilities (BSV or η) were ascribed to an exponential model. Parameter shrinkage was calculated as (1-sd(eta)/omega), where sd(eta) and omega are the standard deviation of individual eta parameters and the population model estimate of the BSV, respectively.9 The Likelihood Ratio, Test including the log-likelihood, the Akaike information criterion and the bayesian information criterion were used to test different hypotheses regarding the final model, covariate effect(s) on pharmacokinetic parameter(s), residual variability model (proportional versus proportional plus additive error model), and structure of the variance–covariance matrix for the BSV parameters.
Parameter estimates were standardized for a mean standard BW using an allometric model:
Where PSTD is the standard value of parameter for a patient with the standard BW value and Pi and BWi are the parameter and BW of the the individual. The PWR exponents are typically 0.75 for clearance parameters and 1 for volumes of distribution.10 The exponents were also tried to be estimated from our data.
For evaluation of the goodness-of-fit, the following graphs were performed for the final model: observed and predicted concentrations versus time, observed concentrations versus population predictions, weighted residuals versus time, and weighted residuals versus predictions. Similar graphs using individual predictive estimation were examined. Diagnostic graphics were obtained using the R program11 (see Figure, Supplemental Digital Content 1, http://links.lww.com/QAI/A217).
Model Validation: Visual Predictive Check and Normalized Prediction Distribution Errors
Tenofovir concentration profiles were simulated and compared with the observed data to evaluate the predictive performance of the model. Simulated concentrations were then compared with the observed data to evaluate the predictive performance of the model. The vector of pharmacokinetic parameters was simulated using the final model. Each vector parameter was drawn in a log-normal distribution with a variance corresponding to the BSV previously estimated. A simulated residual error was added to each simulated concentration. All observed and simulated concentrations were standardized for a tenofovir dose of 136 mg. The 5th, 50th, and 95th percentiles of the simulated concentrations at each time were then overlaid on the observed concentration data and a visual inspection was performed. The variability was reasonably estimated if the 95% CI for the proportion of observed data outside the bounds included the theoretical value of 10%. The model was also appreciated by the normalized prediction distribution errors metrics.12
The dose that produces a 24-hr exposure similar to that observed in adult4 as a function of BW was derived from our model. Recommended doses were then simulated in each group (1000 Monte Carlo simulations of the final model); AUC0–24 hrs and Cmin obtained from the simulations were compared with the previously reported values in adult following an administration of 300 mg of TDF.4
A total of 93 patients and 283 plasma concentrations were available for pharmacokinetic evaluation. The mean follow-up time was around 1 year.
Median (range) serum creatinine and creatinine clearance were 53 (31–96) μmol/L and 140.8 (79.3–214.6) mL·min·1.73 m−2, respectively, suggesting no patients were experiencing renal dysfunction.
A total of 8 concentrations were below the limit of quantification and were treated as left-censored data by the program. A 2-compartment model adequately described the data (Fig. 1A), thus the apparent parameters of the model were the clearance (CL/F), the central volume of distribution (Vc/F), the peripheral volume of distribution (Vp/F), the intercompartmental clearance (Q/F), and the absorption rate constant (Ka). F is the unknown bioavailability. Residual variability was best described by a proportional error model. BSVs were described by exponential error model and retained only for CL/F and Vc/F. A significant covariance of 0.73 was found between CL/F and Vc/F. BW has the most significant effect on TDF pharmacokinetics. The allometric exponents on CL/F and Vc/F were estimated and did not differed significantly from 0.75 and 1, respectively, thus they were fixed to these values. The allometric scaling of clearance (CL, Q) and volume terms (Vc, Vp) decreased the Akaike information criterion/bayesian information criterion critera, resulted in a 52-U decrease in the objective function value, and improved the goodness of fit. Figure.1B represents tenofovir clearance as a function of BW. Thirty-one percent of patients received lopinavir/ritonavir (LPV/r) as a cotreatment, the addition of LPV/r treatment decreased tenofovir clearance by 25% and resulted in a 7.46-U decrease in the objective function value, and BSV on CL/F decreased from 0.50 to 0.48. The median dose of lopinavir received was 266 mg twice a day. Thereby, the final covariate model on CL/F was:
where θCL is the typical value of CL/F for an adult of 70 kg, LPVCL/F is the influential factor of lopinavir treatment on tenofovir CL/F, and LOPI is a categorical covariable and equals 1 if the patient received LPV/r as a cotreatment or 0 if not present.
Table 1 summarizes the final population pharmacokinetic estimates. All the parameters were well estimated, with low relative standard error. The η-shrinkages for CL/F and Vc/F were respectively 0.062 and 0.26, indicating that the empirical Bayesian estimates for individual clearance and Vc/F parameters are reliable. Table 2 summarizes tenofovir AUC0–24 hr Cmin and CL/F in the present (normalized for an adult of 70 kg) and previous adult studies.
Evaluation and Validation
Visual predictive check shows that the average prediction matches the observed concentration time-courses and that the variability is reasonably estimated (see Figure, Supplemental Digital Content 2, http://links.lww.com/QAI/A218). The number (percentage) of observed points within the 90% prediction interval was 260/283 (91%) with a 95% CI of (86.8 to 93.6).
The mean and variance of the normalized prediction distribution errors metrics were not significantly different from 0 (P = 0.41) and 1 (P = 0.17), respectively, and their distribution was not different from a normal one (P = 0.52).
Figure 2 displays the theoretical dose needed to reach 24-hr exposure observed from previous adult studies4 in function of BW according to our model. The proposed doses and BW group have been determined as follows: the dose was increased each time AUC was lower than the median without exceeding the limit of the 97.5th percentile of adult AUC0–24 hr as shown in Figure 2. To reach the adult exposure, children without coadministration with LPV/r should receive the following: 150 mg TDF from 20 to 30 kg, 225 mg TDF from 30 to 40 kg, and the adult dosage of 300 mg TDF over 40 kg. However, children cotreated with LPV/r should receive the following: 150 mg TDF from 20 to 40 kg, 225 mg TDF from 40 to 55 kg, and the adult dosage of 300 mg TDF over 55 kg.
Table 3 and Table 4 displays the simulated AUC0–24 hr and Cmin obtained with our dosing recommendations in each group as they received or not LPV/r. According to follow-up time, a patient could be included in successive BW groups. In each group, AUC0–24 h and Cmin values were close to previous described values in adult.
This article describes tenofovir pharmacokinetics in 93 children aged from 5 to 18 years. Tenofovir concentrations were satisfactorily described by a 2-compartment model. The population model was used to investigate the effect of growth (BW) and maturation (age) on pharmacokinetic parameters. In our model, no effect of age on clearance was observed after BW allometric scaling of the parameters, probably due to the fact that no children younger than 5 years were involved in the study; the maturation effect additionally to the effect of growth may be seen in the very first years of life. Moreover, this modelling predicts fully mature adult values. Pharmacokinetics drug interaction data in adults have shown that coadministration of LPV/r was associated with a decrease of tenofovir clearance,1,2 probably via inhibition of Mrp-2–mediated transport of tenofovir outside renal tubules.14 However, the effect of lopinavir concomitant administration seems to be more important in children than in adult, indeed, Jullien et al.2 reported an effect of 14% on tenofovir adult clearance compared with 25% found in our pediatric study. No effect of creatinine clearance was found on tenofovir pharmacokinetic parameters as no patient with renal insufficiency were involved in the study.
The current Food and Drug Administration recommendation for TDF is the adult dose of 300 mg once a day from 12 years of age and from 35 kg. There is no pediatric dose available for children younger than 35 kg although TDF could be an appropriate alternative to AZT in maintaining virologic suppression in HIV-1–infected children. Thereby, specific pediatric doses are currently strongly needed and that is why we have proposed in this study to evaluate an optimized dosing scheme for TDF.
No relationship between concentration and efficacy of tenofovir in children has been demonstrated. Thus we considered the 24-hr exposure observed in adults after an administration of 300-mg once a day of TDF.4 BW had the most important effect on TDF pharmacokinetics, LPV/r coadministration had also a significant effect on TDF clearance, and thus our dosing recommendations are based on these 2 factors. The dose needed to reach adult exposure according to BW was calculated. The adult dose of 300 mg TDF seemed to be adapted for children over 40 kg not treated with LPV/r. For the youngest, the dose of 150 mg TDF for 20 to 30 kg, and 225 mg TDF for 30 to 40 kg was well adapted to reach adult exposure. The effect of body surface area on TDF was tested, but BW had a more significant effect than BSA, that is why we proposed dosing scheme according to BW, which is much easier in practical terms; however, we simulated the dose of 175 mg/m2 used in the pediatric phase I study6 and exposures were comparable to those obtained with our dosing propositions for the less than 40 kg. However, the use of 300mg from 35kg to 40 kg seems to overexpose the children.
The decrease of tenofovir clearance associated with the coadministration of LPV/r leads to increase tenofovir concentrations and may induce risk of renal dysfunction as reported in few studies.14–17 Riordan et al18 reported renal toxicity in children taking TDF associated with concurrent lopinavir–ritonavir use. Thereby the dose of tenofovir should be decreased for children who received LPV/r as cotreatment: the doses should be maintained longer in a BW category. Thus, they should receive 150 mg TDF for 20–40 kg children and 225 mg TDF for 40–55 kg children, and 300 mg TDF over 55 kg.
Nevertheless, new scaled down tablets of TDF are not yet available for children, thus 75 mg TDF tablets and/or scored tablets are strongly needed to adapt dosing in pediatric patients.
In conclusion, this study reports tenofovir pharmacokinetics in children. The pharmacokinetic parameters were consistent with previous studies. The tenofovir elimination clearance is related to BW and is influenced by LPV/r coadministration. According to this model, TDF dosing schemes were established for the time as a function of BW to reproduce adult exposure. These assumptions should be prospectively confirmed.
1. Kearney BP, Flaherty JF, Shah J. Tenofovir disoproxil fumarate: clinical pharmacology and pharmacokinetics. Clin Pharmacokinet. 2004;43:595–612
2. Jullien V, Treluyer JM, Rey E, et al. Population pharmacokinetics of tenofovir in human immunodeficiency virus-infected patients taking highly active antiretroviral therapy. Antimicrob Agents Chemother. 2005;49:3361–3366
3. Kiser JJ, Fletcher CV, Flynn PM, et al. Pharmacokinetics of antiretroviral regimens containing tenofovir disoproxil fumarate and atazanavir-ritonavir in adolescents and young adults with human immunodeficiency virus infection. Antimicrob Agents Chemother. 2008;52:631–637
4. Boffito M, Pozniak A, Kearney BP, et al. Lack of pharmacokinetic drug interaction between tenofovir disoproxil fumarate and nelfinavir mesylate. Antimicrob Agents Chemother. 2005;49:4386–4389
5. Ramanathan S, Shen G, Cheng A, et al. Pharmacokinetics of emtricitabine, tenofovir, and GS-9137 following coadministration of emtricitabine/tenofovir disoproxil fumarate and ritonavir-boosted GS-9137. J Acquir Immune Defic Syndr. 2007;45:274–279
6. Hazra R, Balis FM, Tullio AN, et al. Single-dose and steady-state pharmacokinetics of tenofovir disoproxil fumarate in human immunodeficiency virus-infected children. Antimicrob Agents Chemother. 2004;48:124–129
7. Jullien V, Treluyer JM, Pons G, et al. Determination of tenofovir in human plasma by high-performance liquid chromatography with spectrofluorimetric detection. J Chromatogr B Analyt Technol Biomed Life Sci. 2003;785:377–381
8. Lavielle M, Mentre F. Estimation of population pharmacokinetic parameters of saquinavir in HIV patients with the MONOLIX software. J Pharmacokinet Pharmacodyn. 2007;34:229–249
9. Savic RM, Karlsson MO. Importance of shrinkage in empirical bayes estimates for diagnostics: problems and solutions. AAPS J. 2009;11:558–569
10. Anderson BJ, Holford NH. Mechanism-based concepts of size and maturity in pharmacokinetics. Annu Rev Pharmacol Toxicol. 2008;48:303–332
11. Ihaka R, Gentleman R. R: a language for data analysis and graphics. J Comput Graph Stat. 1996;5:299
12. Comets E, Brendel K, Mentre F. Computing normalised prediction distribution errors to evaluate nonlinear mixed-effect models: the npde add-on package for R. Comput Methods Programs Biomed. 2008;90:154–166
13. Barditch-Crovo P, Deeks SG, Collier A, et al. Phase i/ii trial of the pharmacokinetics, safety, and antiretroviral activity of tenofovir disoproxil fumarate in human immunodeficiency virus-infected adults. Antimicrob Agents Chemother. 2001;45:2733–2739
14. Peyriere H, Reynes J, Rouanet I, et al. Renal tubular dysfunction associated with tenofovir therapy: report of 7 cases. J Acquir Immune Defic Syndr. 2004;35:269–273
15. Goicoechea M, Liu S, Best B, et al. Greater tenofovir-associated renal function decline with protease inhibitor-based versus nonnucleoside reverse-transcriptase inhibitor-based therapy. J Infect Dis. 2008;197:102–108
16. Verhelst D, Monge M, Meynard JL, et al. Fanconi syndrome and renal failure induced by tenofovir: a first case report. Am J Kidney Dis. 2002;40:1331–1333
17. Karras A, Lafaurie M, Furco A, et al. Tenofovir-related nephrotoxicity in human immunodeficiency virus-infected patients: three cases of renal failure, Fanconi syndrome, and nephrogenic diabetes insipidus. Clin Infect Dis. 2003;36:1070–1073
18. Riordan A, Judd A, Boyd K, et al. Tenofovir use in human immunodeficiency virus-1-infected children in the United kingdom and Ireland. Pediatr Infect Dis J. 2009;28:204–209
children; HIV; population pharmacokinetics; tenofovir
Supplemental Digital Content
© 2011 Lippincott Williams & Wilkins, Inc.