PHARMACOGENETICS

Population pharmacokinetics and pharmacogenetics of oncedaily prolonged-release formulation of tacrolimus in pediatricand adolescent kidney transplant recipients

Wei Zhao & May Fakhoury & Véronique Baudouin &

Thomas Storme & Anne Maisin & Georges Deschênes &

Evelyne Jacqz-Aigrain

Received: 26 March 2012 /Accepted: 30 May 2012# Springer-Verlag 2012

AbstractBackground and Objectives TacrolimusPR is a newprolonged-release once-daily formulation of the calci-neurin inhibitor tacrolimus, currently used in adult trans-plant patients. As there are no pharmacokinetic dataavailable in pediatric kidney transplant recipients, theaims of this study were to develop a population phar-macokinetic model of tacrolimusPR in pediatric and ad-olescent kidney transplant recipients and to identifycovariates that have a significant impacts on tacrolimusPR

pharmacokinetics, including CYP3A5 polymorphism.

Methods Pharmacokinetic samples were collected from22 pediatric kidney transplant patients. Population phar-macokinetic analysis was performed using NONMEM.Pharmacogenetic analysis was performed on the CYP3A5gene.Results The pharmacokinetic data were best described by aone-compartment model with first order absorption and lag-time. The weight normalized oral clearance CL/F [CL/F/(weight/70)0.75] was lower in patients with CYP3A5*3/*3as compared to patients with the CYP3A5*1/*3 (32.2±10.1vs. 53.5±20.2 L/h, p00.01).Conclusions The population pharmacokinetic model oftacrolimusPR was developed and validated in pediatric andadolescent kidney transplant patients. Body weight andCYP3A5 polymorphism were identified as significant fac-tors influencing pharmacokinetics. The developed modelcould be useful to optimize individual pediatric tacrolimusPR dosing regimen in routine clinical practice.

Keywords Tacrolimus . Once daily . Prolonged-releaseformulation . Pediatrics . Pediatric pharmacology .

Population pharmacokinetics . Pharmacogenetics .

CYP3A5 . Kidney transplantation

Introduction

Compliance and adherence to immunosuppressive therapyplay a major role in the long-term success of kidney trans-plantation, in particular for children and adolescents. It hasbeen demonstrated that about 44 % of graft losses and 23 %of late acute rejection episodes were associated with non-compliance in pediatric kidney transplant recipients [1].Approaches to improve this situation include reducing the

W. Zhao :M. Fakhoury : E. Jacqz-AigrainDepartment of Pediatric Pharmacology and Pharmacogenetics,Hôpital Robert Debré, Université Paris Diderot,Assistance Publique - Hôpitaux de Paris,Paris, France

W. Zhao : E. Jacqz-AigrainClinical Investigation Center CIC9202, INSERM,Paris, France

V. Baudouin :A. Maisin :G. DeschênesDepartment of Pediatric Nephrology, Hôpital Robert Debré,Université Paris Diderot,Assistance Publique - Hôpitaux de Paris,Paris, France

T. StormeDepartment of Pharmacy, Hôpital Robert Debré,Université Paris Descartes,Assistance Publique - Hôpitaux de Paris,Paris, France

E. Jacqz-Aigrain (*)Department of Pediatric Pharmacology and Pharmacogenetics,Clinical Investigation Center CIC9202, INSERM,Hôpital Robert Debré, 48 Boulevard Sérurier,75935 Paris Cedex 19, Francee-mail: evelyne.jacqz-aigrain@rdb.aphp.fr

Eur J Clin PharmacolDOI 10.1007/s00228-012-1330-6

complexity of dosing regimens by reducing the number ofadministrations per day and/or the number of pills. It hasbeen reported that a once-daily dose of immunosuppressantwas associated with increased adherence over a twice dailydose in adult kidney transplant recipients [2].

The once daily prolonged-release formulation of tacroli-mus (tacrolimusPR) holds potential to improve the compli-ance of tacrolimus therapy, and has been licensed recentlyfor the prophylaxis of transplant rejection in adult kidney orliver allograft recipients and for the treatment of allograftrejection resistant with other immunosuppressive medicinalproducts in adult patients [3]. The pharmacokinetics oftacrolimusPR have been widely evaluated in adults [4–6].

TacrolimusPR added ethylcellulose, hypromellose andlactose monohydrate to tacrolimus, modifying the drug re-lease by controlling water penetration and forming a protec-tive polymer gel layer around tacrolimus [7]. Due todifferent dissolution properties, the pharmacokinetics areconsiderably different between the two formulations: a re-duced maximum concentrations (Cmax) and a delayed timeto maximum concentrations (Tmax) have been reported to beassociated with tacrolimusPR in adults [7]. In addition, giventhe different drug release rates, the extent of influence ofcovariates may differ between the two formulations, inparticular if covariate effect is associated with tacrolimusabsorption [8]. These differences between two formulationshave been demonstrated in adults but tacrolimusPR pharma-cokinetic data in children are limited. Only one bioequiva-lence study was reported in pediatric liver transplantrecipients [9] and TacrolimusPR is still used off-label inchildren. Because both developmental maturation andCYP3A5 polymorphism have significant impacts on tacro-limus disposition [10–14], detailed pharmacokinetic infor-mation is mandatory to optimize tacrolimusPR treatment inchildren.

The objectives of the present study were to develop apopulation pharmacokinetic model of tacrolimusPR, includ-ing CYP3A5 polymorphism as a covariate, in pediatric andadolescent kidney transplant recipients.

Patients and methods

Patients

Pediatric kidney transplant recipients from the departmentof pediatric nephrology in Robert Debré Hospital wereincluded between 2009 and 2011. This clinical trial wasdesigned in accordance with the legal requirements and theDeclaration of Helsinki. The pharmacokinetic data wereobtained according to the local clinical practice. The parentsof our patients gave additional informed consent for phar-macogenetic testing.

In the present paper, the formulation of tacrolimus willrefer to tacrolimus (for Prograf®, Astellas, Levallois-Perret,France) and tacrolimusPR (for Advagraf®, Astellas,Levallois-Perret, France). Patients receiving stable treatmentwith tacrolimus twice daily were switched to tacrolimusPR

once daily, on a daily milligram-for-milligram basis. Dosageadjustment was based on biological follow-up and tacro-limus monitoring in order to maintain C0 in the recom-mended therapeutic range of 5-15 ng/mL [12]. A fullconcentration-time profile was determined during hospi-talization and/or routine follow-up visits when a steady-state condition was achieved. A limited number of bloodsamples were obtained before, 1, 2, 3, 6, 12, 16 and24 hours after drug intake. Mycophenolate mofetil, MMF(Cellcept®, Roche, Neuilly-sur-Seine, France) and myco-phenolic acid, MPA (Myfortic®, Novartis, Huningue,France) doses were adjusted to maintain AUC between30 and 60 h*mg/L [15].

Assay of tacrolimus

Blood concentrations were measured using an enzyme-multiplied immunoassay technique (Dada–Behring Diag-nostics, Milton Keynes, UK). In accordance with the man-ufacturer’s information, the lower limit of quantificationwas set at 2.0 ng/mL and the linearity ranged from 2.0 to30 ng/mL.

Pharmacokinetic analysis

Pharmacokinetic analysis was carried out using the nonlin-ear mixed effects modeling program NONMEM VI (V2.0;Icon Development Solutions, USA). The first order condi-tional estimation (FOCE) method with interaction optionwas used to estimate pharmacokinetic parameters and theirvariability.

Different one or two-compartment open models with firstorder elimination were compared, each with a different inputto describe the absorption phase: (i) a zero order input withor without a lag time, (ii) a first order input with or without alag time, (iii) an Erlang distribution [16, 17] and (iv) atransit compartment [18].

Inter-individual variability of the pharmacokineticparameters was estimated using an exponential modelexpressed as follows:

θi ¼ θ� exp ηið Þwhere θ is the typical population value of pharmacokineticparameter, ηi is the difference between the log-transformedindividual-specific parameter and the log-transformed typi-cal individual’s parameter, and represents the random effectfor parameter θ in patient i. The ηi values are independent,

Eur J Clin Pharmacol

identically distributed random variables and normally dis-tributed around 0 with variance ω2 and i is the variable forith individual.

Residual variability (additive, proportional, exponentialor mixed) model were selected according to improvement ofOFV and visual inspection of routine diagnostic plots.

Covariate analysis

On the basis of allometry principles, body weight was usedas a covariate to pharmacokinetic parameters. The relation-ship was modeled with the following equation:

Pi ¼ Pstd � WTi=WTstdð ÞPWR

where Pi is the pharmacokinetic parameter of the ith individ-ual, WTi is the weight of the ith individual and Pstd theparameter of an individual with a weight WTstd of 70 kg.The PWR exponent is the allometric coefficient: 0.75 forCL/F and 1 for V/F [19].

The additional effects of age, post transplantation time,hemoglobin and hematocrit were investigated as potentialvariables influencing pharmacokinetic parameters.

The effect of CYP3A5 genotype on tacrolimusPR CL/Fwas investigated using the following equation:

CL=F ¼ θ2� weight=70ð Þ0:75 � FCYP3A5

where If patient with CYP3A5 *1/*3, FCYP3A50θ3, whichwas estimated by the model. If patient with CYP3A5 *3/*3,FCYP3A501.

The likelihood ratio test was used to analyze the effect ofeach variable. The selection of variables was determinedusing a forward and backward selection process. Duringforward selection, a covariate was selected only if a signif-icant (p<0.05, χ2 distribution with one degree of freedom)decrease (reduction>3.84) in the OFV from the basic modelwas obtained. Then all the significant variables were includ-ed simultaneously into a ‘full’ model. The importance ofeach variable was then re-evaluated by backward selection.Each variable was independently removed from the fullmodel to confirm its relevance. An increase in the OFV ofmore than 6.635 (p<0.01, χ2 distribution) was required forconfirmation. The resulting model was called the"final"population pharmacokinetic model and included all signifi-cant variables

Model validation

The stability and performance of the final model wereassessed by means of an internal validation method involv-ing a nonparametric bootstrap with re-sampling and

replacement. Re-sampling was repeated 500 times andthe values of estimated parameters from the bootstrapprocedure were compared with those estimated from theoriginal data set. The entire procedure was performed inan automated fashion, using Perl-speaks-NONMEM(PsN) [20].

Simulation-based diagnostics was performed by usingnormalized prediction distribution errors (NPDE) [21]and prediction-corrected visual predictive check (VPC)[22]. The dataset was simulated 1000 times using thefinal population model parameters. For NPDE, a cumu-lative distribution was assembled for each observationwith the 1000 simulated concentrations. The NPDE isexpected to follow the N (0, 1) distribution [23]. Thefollowing graphs were plotted by using NPDE R pack-age (v1.2): (i) QQ-plot of the distribution of the NPDEversus the theoretical N (0,1) distribution; (ii) histogramof the NPDE. For prediction-corrected VPC, thesimulation-based 95 % confidence interval for 5 %,50 % and 95 % percentiles of prediction-corrected con-centrations was plotted by using Xpose 4.

Pharmacogenetic analysis

Total genomic DNA was extracted from blood samplesusing a QIAmp DNA Blood Mini Kit (Qiagen, Chats-worth, CA) and quantified using a Nanodrop spectropho-tometer (Labtech, Palaiseau, France). CYP3A5polymorphism was determined using the TaqMan allelicdiscrimination technique with 3′-minor groove binding(MGB) quencher probes (ABI Prism 7900; Applied Bio-systems, Foster City, CA), as previously reported [14].

Table 1 Characteristics of 22 children

Number Mean (SD) Range

Number of patients 22

M/F 14/8

Age (years) 15.2 (5.1) 5.6 – 22.8

Weight (kg) 45.2 (17.1) 16.7 – 70.0

Time posttransplantation (days)

1950 (1546) 193 – 4983

TacrolimusPR dose (mg) 5.3 (2.4) 1.5 – 12.0

TacrolimusPR dose (mg/kg) 0.12 (0.06) 0.05 – 0.26

Hematocrit (%) 35.5 (7.0) 25.8 – 61.2

Hemoglobin (g/dL) 11.9 (2.4) 8.2 – 20.2

Genotypes for CYP3A5

CYP3A5 *3/*3 16/22

CYP3A5 *1/*3 6/22

CYP3A5*1/*1 0/22

Eur J Clin Pharmacol

Statistical analysis

Results are expressed as mean and standard deviation (meanand SD). The distribution of continuous data was evaluatedand parametric or nonparametric tests were consequentlyapplied when appropriate. The patients’ genotypes wereused as categorical independent variables. These statisticalanalysis were performed using SPSS for Windows, version16.0 (Chicago, Illinois, USA). p<0.05 was considered sta-tistically significant.

Results

Twenty-two children (14 boys and 8 girls) were included inthe present study: age was 15.2 years (5.1) and weight was

45.2 kg (17.1) (mean (SD)). The characteristics of thepatients on the day of pharmacokinetic sampling are pre-sented in Table 1.

Population pharmacokinetic modeling

A total of 171 blood tacrolimusPR concentrations from 22pharmacokinetic profiles were available for populationmodeling. The concentrations ranged from <2.0 to55.4 ng/mL (only one pharmacokinetic profile had troughconcentrations below the limit of quantification).

Pharmacokinetic data was described by a one- com-partment model with first order elimination. The absorp-tion phase was characterized by a first order absorptionwith a lag-time, which described absorption better than

Table 2 Populationpharmacokineticparameters of tacrolimusand bootstrap results(n0500)

PK parameters SE(%) Bootstrap

Median 2.5th 97.5th

Lag-time (h)r 0.872 4.3 0.849 0.067 0.907

Absorption rate constant (h-1) Ka 8.34 43.8 6.48 2.05 35.10

Volume of distribution (L) V/F

V/F0θ1×(bodyweight /70)

θ1 1100 13.2 1080 846 1393

Oral clearance (L/h) CL/F

CL/F0θ2×(bodyweight/70)0.75× FCYP3A5θ2 30.6 8.9 30.6 25.2 36.6

If CYP3A5 *1/*3, FCYP3A50θ3

If CYP3A5 *3/*3 FCYP3A501

If CYP3A5 *3/*3, FCYP3A501

θ3 1.66 17.6 1.69 1.22 2.51

Inter-individual variability

Ka 150 42.3 138 76.5 238.9

V/F 52.1 34.5 49.6 28.8 67.9

CL/F 34.6 29.0 32.1 22.1 43.0

Residual variability (exponential) 22.1 16.3 22.0 18.3 25.9

Fig. 1 Diagnostic goodness-of-fit plots for the final populationpharmacokinetic model oftacrolimusPR, including ob-served (DV) versus populationpredicted concentrations(PRED), observed versus indi-vidual predicted concentrations(IPRED)

Eur J Clin Pharmacol

zero-order absorption and Erlang distribution, as judgedby OFV and routine goodness-of-fit plots. The transitcompartment to describe the absorption process gave thesimilar performance as first order absorption; howeverthe stand errors were not obtained for pharmacokineticparameters. Inter-individual variability was best de-scribed by an exponential model and was then estimatedfor the absorption rate constant (Ka), the apparent vol-ume of distribution (V/F) and the apparent oral clear-ance (CL/F). Residual variability was best described byan exponential model. A more complex distributionmodel did not improve model fit to data.

The effect of body weight was investigated as a potentialcovariate. Given the limited number of patients, the allome-tric coefficients were fixed at 0.75 for CL/F and 1 for V/F.

As extrapolated adult pharmacokinetic parameters weresimilar to those reported in published data [3, 24], theallometry-based physiological model was chosen for thesubsequent covariate analysis for 3A5 genotype, age, posttransplantation time, hemoglobin and hematocrit values.

The effect of 3A5 genotype (CYP3A5*1/*3 orCYP3A5*3/*3) was investigated as a categorical variable.The effects of age, post transplantation time, hemoglobinand hematocrit were investigated as continuous variables. Inthe forward selection processes, only inclusion of CYP3A5genotype on CL/F produced significant decreases of OFV(ΔOFV 7.5), which was retained in the final model. Allother covariates tested had no effect on tacrolimusPR phar-macokinetic parameters. Frequencies of CYP3A5 *1/*3 and*3/*3 genotype were 27 % and 73 %, respectively. TheBayesian individual estimated CL/F was obtained usingthe final model. For CYP3A5, the weight normalized CL/F [CL/F/ (weight/70)0.75] is lower in patients withCYP3A5*3/*3 as compared to patients with the CYP3A5*1/*3 (32.2±10.1 vs. 53.5±20.2 L/h, p00.01, Mann–Whitneytest).

Routine diagnostic plots, including observed versus indi-vidual prediction, observed versus population prediction,indicates sufficient goodness-of-fit (Fig. 1).

The mean parameter estimates from the bootstrap proce-dure very closely agreed with their respective values fromthe final population model, indicating that the estimates forthe population pharmacokinetic parameters in the final mod-el were accurate and that the model was stable. The resultsof 500 bootstrap replicates are summarized in Table 2.

The prediction-corrected VPC is presented in Fig. 2. Theprediction interval was obtained by simulating 1000 datasetswith the final model, and the prediction-corrected concen-trations were well predicted by the final model. The NPDEis presented in Fig. 3. The mean and variance of NPDEwere -0.08 (Wilcoxon signed rank test: 0.36) and 0.93(Fisher variance test: 0.51), respectively.

Fig. 3 Normalized predictiondistribution errors (NPDE)analysis for the tacrolimusPR

final model. NPDE: QQ-plot ofthe distribution of the NPDEversus the theoretical N (0,1)distribution (left). Histogram ofthe distribution of the NPDE,with the density of the standardGaussian distribution overlaid(right)

Fig. 2 Prediction-corrected visual Predictive Check (VPC) for thetacrolimusPR final model. VPC: prediction-corrected concentrationsare plotted using a circle (○). The dashed lines represent the 5th and95th percentiles of prediction-corrected concentrations. The solid linesrepresent 50th percentiles of prediction-corrected concentrations. Thesemitransparent field represents a simulation-based 95 % confidenceinterval

Eur J Clin Pharmacol

Discussion

This is the first population pharmacokinetic study ofprolonged-release formulation of tacrolimusPR in pediatrickidney transplant recipients aiming to investigate the impactof maturation and pharmacogenetics on drug disposition.Our results showed that a one-compartment model with firstorder absorption and lag-time well described data, and thatbodyweight and CYP3A5 genotype had significant impactson tacrolimusPR pharmacokinetics.

In the present study, the mean CL/F of tacrolimusPR was 0.63(range 0.20 -1.27) L/h/kg. This is in agreement with previouslyreported values of tacrolimus in pediatric kidney transplantpatients: Kim et al., reported that a mean dosage of 0.12 mg/kg resulted in a mean AUC of 192 h·ng/mL in 30 children [25];Montini et al., reported that a mean dosage of 0.15 mg/kgresulted in a mean AUC of 197 h·ng/mL in 18 children at2 weeks after transplantation [26]; Zhao et al., reported that amean CL/F of 0.76 L/h/kg in 50 children [14]. The inter-individual variability of tacrolimusPR CL/F was 34.6 %, whichwas slightly lower than that of tacrolimus (41.9 %) [14].

In accordance with the principles of allometry, and in linewith the approach previously adopted with tacrolimus inpediatric liver and kidney transplant recipients [14, 27], wedeveloped an allometry-based population pharmacokineticmodel for the new formulation of tacrolimusPR: Bodyweight was used as a covariate with allometric coefficientsof 0.75 for CL/F and 1 for V/F. These power models havetheoretical and observational bases in biology [19]. Thischoice allowed estimating tacrolimusPR CL/F of a patientwith standard body weight of 70 kg: the estimated CL/F of35 L/h was in agreement with previously reported CL/F of33.5 L/h (Mean clearance06.7 L/h; mean bioavailability020 %) in adults [3, 24].

As CYP3A5 plays an important role in tacrolimus dispo-sition [28], we investigated a possible association betweenCYP3A5 polymorphism and tacrolimusPR CL/F. Our resultsdemonstrated that the weight normalized CL/F of tacroli-musPR was significantly higher in expressers (CYP3A5*1allele) than in non-expressersnon-expressers (CYP3A5*3/*3). Similarly, the CL/F of tacrolimus was significantlyhigher in CYP3A5 expressers than in non-expressers, asshown in a study of 50 pediatric kidney recipients [14].

As the pharmacokinetic sampling was taken during hos-pitalization and/or routine follow-up visits, primarily formonitoring therapy due to the absence of pediatric data,we did not perform a bioequivalence study. Although bio-equivalence has been demonstrated according to regula-tory recommendation and a daily milligram-for-milligramconversion between tacrolimus and tacrolimusPR was rec-ommended by the manufacturer, therapeutic drug moni-toring (TDM) is still required to optimize tacrolimusPR

therapy in children.

Conclusion and perspectives

In this pediatric study, we have developed and validated apopulation pharmacokinetic model for once daily adminis-tration of prolonged-release formulation of tacrolimusPR.Body weight and CYP3A5 polymorphism were identifiedas significant predictors of tacrolimus pharmacokinetics.The present model would be an efficient tool to individual-ize pediatric tacrolimusPR dosing regimen in routine clinicalpractice. A pooled population pharmacokinetic analysiscould provide additional information on the pharmacokinet-ic difference between the two formulations in children [29].

Acknowledgments We thank all the children and their families par-ticipating in this study. We also acknowledge the local clinical inves-tigators (Marc Fila, Sonia Azib, Theresa Kwon, Anne Laure Leclerc,Marie-Alice Macher, Ferielle Louillet and Daolun Zhang) for conduct-ing the study and technicians (Christel Grondin, Michel Popon, SamiraBenakouche and Yves Médard) for technical support. This work wassupported by Global Research in Paediatrics – Network of Excellence(GRIP, EU-funded FP7 project, Grant Agreement number 261060).

Conflict of interest statement The authors declare no conflict ofinterest related to this work

References

1. Dobbels F, Ruppar T, De Geest S, Decorte A, Van Damme-Lombaerts R, Fine RN (2010) Adherence to the immunosuppres-sive regimen in pediatric kidney transplant recipients: a systematicreview. Pediatr Transplant 14(5):603–613

2. Weng FL, Israni AK, Joffe MM et al (2005) Race and electroni-cally measured adherence to immunosuppressive medications afterdeceased donor renal transplantation. J Am Soc Nephrol 16(6):1839–1848

3. EuropeanMedicines Agency. Advagraf 0.5mg prolonged-release hardcapsules: Summary of product characteristics [online]. Available fromURL: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Product_Information/human/000712/WC500022234.pdf[assessed in 22/03/2012]

4. Benkali K, Rostaing L, Premaud A et al (2010) Population phar-macokinetics and Bayesian estimation of tacrolimus exposure inrenal transplant recipients on a new once-daily formulation. ClinPharmacokinet 49(10):683–692

5. Wlodarczyk Z, Squifflet JP, Ostrowski M et al (2009) Pharmaco-kinetics for once- versus twice-daily tacrolimus formulations in denovo kidney transplantation: a randomized, open-label trial. Am JTransplant 9(11):2505–2513

6. Glowacki F, Lionet A, Hammelin JP et al (2011) Influence ofCytochrome P450 3A5 (CYP3A5) Genetic Polymorphism on thePharmacokinetics of the Prolonged-Release, Once-Daily Formula-tion of Tacrolimus in Stable Renal Transplant Recipients. ClinPharmacokinet 50(7):451–459

7. European Medicines Agency. Advagraf: scientific discussion[online]. Available from URL: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Scientific_Discussion/human/000712/WC500022237.pdf [assessed in 22/03/2012]

8. Barraclough KA, Isbel NM, Johnson DW, Campbell SB, Staatz CE(2011) Once- versus twice-daily tacrolimus: are the formulationstruly equivalent? Drugs 71(12):1561–1577

Eur J Clin Pharmacol

9. Heffron TG, Pescovitz MD, Florman S et al (2007) Once-dailytacrolimus extended-release formulation: 1-year post-conversionin stable pediatric liver transplant recipients. Am J Transplant 7(6):1609–1615

10. Staatz CE, Goodman LK, Tett SE (2010) Effect of CYP3A andABCB1 single nucleotide polymorphisms on the pharmacokineticsand pharmacodynamics of calcineurin inhibitors: Part I. Clin Phar-macokinet 49(3):141–175

11. Staatz CE, Goodman LK, Tett SE (2010) Effect of CYP3A andABCB1 single nucleotide polymorphisms on the pharmacokineticsand pharmacodynamics of calcineurin inhibitors: Part II. ClinPharmacokinet 49(4):207–221

12. Zhao W, Fakhoury M, Jacqz-Aigrain E (2010) Developmentalpharmacogenetics of immunosuppressants in pediatric organ trans-plantation. Ther Drug Monit 32(6):688–699

13. de Wildt SN, van Schaik RH, Soldin OP et al (2011) The inter-actions of age, genetics, and disease severity on tacrolimus dosingrequirements after pediatric kidney and liver transplantation. Eur JClin Pharmacol 67(12):1231–1241

14. Zhao W, Elie V, Roussey G et al (2009) Population pharmacoki-netics and pharmacogenetics of tacrolimus in de novo pediatrickidney transplant recipients. Clin Pharmacol Ther 86(6):609–618

15. Staatz CE, Tett SE (2007) Clinical pharmacokinetics and pharma-codynamics of mycophenolate in solid organ transplant recipients.Clin Pharmacokinet 46(1):13–58

16. Rousseau A, Léger F, Le Meur Y et al (2004) Population pharma-cokinetic modeling of oral cyclosporin using NONMEM: compar-ison of absorption pharmacokinetic models and design of aBayesian estimator. Ther Drug Monit 26(1):23–30

17. Zhao W, Fakhoury M, Deschênes G et al (2010) Population phar-macokinetics and pharmacogenetics of mycophenolic acid follow-ing administration of mycophenolate mofetil in de novo pediatricrenal-transplant patients. J Clin Pharmacol 50(11):1280–1291

18. Savic RM, Jonker DM, Kerbusch T, Karlsson MO (2007) Imple-mentation of a transit compartment model for describing drugabsorption in pharmacokinetic studies. J Pharmacokinet Pharma-codyn 34(5):711–726

19. Anderson BJ, Holford NH (2008) Mechanism-based concepts ofsize and maturity in pharmacokinetics. Annu Rev Pharmacol Tox-icol 48:303–332

20. Lindbom L, Ribbing J, Jonsson EN (2004) Perl-speaks-NONMEM(PsN)–a Perl module for NONMEM related programming. Com-put Methods Programs Biomed 75(2):85–94

21. Brendel K, Comets E, Laffont C, Laveille C, Mentré F (2006) Metricsfor external model evaluation with an application to the populationpharmacokinetics of gliclazide. Pharm Res 23(9):2036–2049

22. Bergstrand M, Hooker AC, Wallin JE, Karlsson MO (2011)Prediction-corrected visual predictive checks for diagnosing non-linear mixed-effects models. AAPS J 13(2):143–151

23. Comets E, Brendel K, Mentré F (2008) Computing normalisedprediction distribution errors to evaluate nonlinear mixed-effectmodels: the npde add-on package for R. Comput Methods Pro-grams Biomed 90(2):154–166

24. Cross SA, Perry CM (2007) Tacrolimus once-daily formulation: inthe prophylaxis of transplant rejection in renal or liver allograftrecipients. Drugs 67(13):1931–1943

25. Kim JS, Aviles DH, Silverstein DM, Leblanc PL, Matti VV (2005)Effect of age, ethnicity, and glucocorticoid use on tacrolimuspharmacokinetics in pediatric renal transplant patients. PediatrTransplant 9(2):162–169

26. Montini G, Ujka F, Varagnolo C et al (2006) The pharmacokineticsand immunosuppressive response of tacrolimus in paediatric renaltransplant recipients. Pediatr Nephrol 21(5):719–724

27. Fukudo M, Yano I, Masuda S et al (2006) Population pharmaco-kinetic and pharmacogenomic analysis of tacrolimus in pediatricliving-donor liver transplant recipients. Clin Pharmacol Ther 80(4):331–345

28. Barry A, Levine M (2010) A systematic review of the effect ofCYP3A5 genotype on the apparent oral clearance of tacrolimus inrenal transplant recipients. Ther Drug Monit 32(6):708–714

29. Woillard JB, de Winter BC, Kamar N, Marquet P, Rostaing L,Rousseau A (2011) Population pharmacokinetic model and Bayes-ian estimator for two tacrolimus formulations–twice daily Prografand once daily Advagraf. Br J Clin Pharmacol 71(3):391–402

Eur J Clin Pharmacol