Enliven: Surgery and Transplantation

Objectives:
Equoral® is a generic formulation of cyclosporine widely used in Tunisian renal transplant patients. The purposes of this study were to explore the
pharmacokinetic variability of Equoral® in Tunisian renal transplant, to develop a population pharmacokinetic model and a MAP-BE of Equoral®
PK parameters and global exposure.

Methods:
Full-PK profiles were obtained from 17 renal transplant patients given Equoral® twice-daily. Measurements were performed using an FPIA
technique. To estimate popPK parameters of cyclosporine, a non linear mixed effects approach was used and the popPK parameters were used as
priors to develop a MAP-BE for estimation of Equoral® PK parameters and AUC using three blood concentrations. Predictive performances were
tested by calculating mean predicted error between estimated and reference AUC. The validation of these PK tools was performed using jackknife
approach, visual predictive check and bootstrapping method.

Results:
Equoral® exhibited a high inter-patient variability: C0= 90±57 ng/mL, Cmax = 762±343 ng/mL and AUC0-12 = 2460±1240 ng.h/mL. A two compartment
model with Erlang distribution best described the data: residual proportional error was 26.8% and imprecision parameter estimate was less than 26%.
None of the tested covariates significantly affected cyclosporine pharmacokinetics. Using this model and cyclosporine concentrations measured at
0.30 and 180min post-dose, MAP-BE could accurately estimate cyclosporine AUC, mean bias between estimated and reference AUC was 5.47 ±
29% with 82% of the patients with AUC bias < 20%. The doses proposed by the BE were similar to those proposed using all concentrations in 14
out of 17 patients.

Conclusion:
A popPK model for Equoral® when given to Tunisian kidney recipients. A MAP-BE using only three blood concentrations that estimates accurately
Equoral® exposure in these patients was developed and could allow dose adjustments based on the AUC in clinical setting.

Abstract

Objectives

Equoral^® is a generic formulation of cyclosporine widely used in Tunisian renal transplant patients. The purposes of this study were to explore the pharmacokinetic variability of Equoral^® in Tunisian renal transplant, to develop a population pharmacokinetic model and a MAP-BE of Equoral^® PK parameters and global exposure.

Methods

Full-PK profiles were obtained from 17 renal transplant patients given Equoral^® twice-daily. Measurements were performed using an FPIA technique. To estimate popPK parameters of cyclosporine, a non linear mixed effects approach was used and the popPK parameters were used as priors to develop a MAP-BE for estimation of Equoral^® PK parameters and AUC using three blood concentrations. Predictive performances were tested by calculating mean predicted error between estimated and reference AUC. The validation of these PK tools was performed using jackknife approach, visual predictive check and bootstrapping method.

Results

Equoral^® exhibited a high inter-patient variability: C₀= 90±57 ng/mL, C_max = 762±343 ng/mL and AUC_0-12 = 2460±1240 ng.h/mL. A two compartment model with Erlang distribution best described the data: residual proportional error was 26.8% and imprecision parameter estimate was less than 26%. None of the tested covariates significantly affected cyclosporine pharmacokinetics. Using this model and cyclosporine concentrations measured at 0.30 and 180min post-dose, MAP-BE could accurately estimate cyclosporine AUC, mean bias between estimated and reference AUC was 5.47 ± 29% with 82% of the patients with AUC bias < 20%. The doses proposed by the BE were similar to those proposed using all concentrations in 14 out of 17 patients.

Conclusion

A popPK model for Equoral^® when given to Tunisian kidney recipients. A MAP-BE using only three blood concentrations that estimates accurately Equoral^® exposure in these patients was developed and could allow dose adjustments based on the AUC in clinical setting.

Introduction

Cyclosporine, an immunosuppressant introduced in 1980 for the management of organ transplantation. Its pharmacokinetics is characterized by a large inter and intra individual variability and a narrow therapeutic window. This can lead to an insufficient immunosupression or toxic side effects [1,2]. Various factors may explain variability of cyclosporine pharmacokinetic, such as drug absorption, gastrointestinal motility, and liver and pancreatic function. The development the micro emulsion formulation of cyclosporine Néoral^® ameliorates significantly the absorption of cyclosporine compared to the original oil-based formulation sandimmune [3]. However, pharmacokinetics of cyclosporine remains influenced by demographic factors, time after transplantation and drug interaction. Equoral^® is a generic formulation of cyclosporine widely used as an alternative treatment to Neoral^® in Tunisian renal transplant patients. Population pharmacokinetic data of cyclosporine Equoral^® was not reported elsewhere.

The purposes of this study were:
• To explore the pharmacokinetic variability of Equoral^® in Tunisian renal transplant patients.
• To develop a population pharmacokinetic (popPK) model and a Bayesian estimator (MAP-BE) for the estimation of Equoral® PK parameters and global exposure (ie, AUC).

Methods

Patients and Data Collection

Data were obtained from 17 adult patients who underwent renal transplantation from 2006 to 2010 at the department of Urology in the Charles Nicolle Hospital of Tunis, Tunisia. Data collection was approved by the hospital ethic comity. Patients received the generic formulation of cyclosporine (Equoral^®) twice a day with doses varying between 0.66 and 4.06 mg/kg/d. All patients received concomitantly mycophenolate mofetil.

Blood Collection

Full pharmacokinetic profiles were collected in EDTA tubes at pre-dose, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12 hours post-dose. Blood samples were stored at -20°C until analysis.

Cyclosporine Assay

All whole-blood samples were analyzed by fluorescence polarization immunoassay (FPIA-Axym) in Laboratory of Clinical Pharmacology of the National Center of Pharmacovigilance. Precision data of immunoassay was demonstrated previously by comparison of the two methods for CsA therapeutic monitoring HPLC and AxSYM. Result showed that the correlation coefficient is 0.99 [4]. The lower limit of quantification the assay was 23ng/ml. The method was linear between 0 and 800ng/ml.

Population Pharmacokinetic Analysis

The population pharmacokinetic analysis was conducted by the nonlinear mixed-effects modeling (NONMEM version VI) software (GloboMax LLC, Ellicott City, MD, USA) using Wings for NONMEM version 614 (developed by N. Holford, available from http://wfn.sourceforge.net/) [1]. All population pharmacokinetic analyses were carried out using the first-order conditional estimation method (FOCE).

One, two and three structural compartment models were tested. Models with Erlang distribution (ADVAN5 SS5), which is a particular case of the gamma distribution [5] and was previously used for cyclosporine in renal transplant recipients [6], were also tested. Erlang distribution explain the transition of the drug through a linear chain of n identical transit compartments placed upstream from the deposit to the central compartment and connected by an identical exiting rate constant (ktr).

Between subjects variability (BSV) was described using exponential models. The covariance of the parameters was studied during the modeling process. Additive, proportional and combined (i.e. additive and proportional) error models were tested to describe the residual variability.

Covariates Analysis

The screening and selection of covariates were performed as part of population pharmacokinetic analysis following a classic stepwise approach [7]. In the first step, a covariate free population pharmacokinetic model was analyzed. The potential covariates considered were the demographic characteristics [bodyweight (BW), age] and laboratory test results (bilirubin, AST, ALT). Gender was not considered as a potential covariate in our study because we have a sex ratio M/F of 13/4. The influence of continuous covariates on pharmacokinetic parameters was tested systematically via a generalized modeling approach according to an allometric function. For example, the effect of bodyweight on the central volume of distribution (VC) was tested using the following equation: VC = TV (VC) X (BW / medcov)^θBW

Where TV (VC) is the typical value of the central volume of distribution for a patient with the median covariate value (medcov) and θBW is the estimated influential factor for bodyweight.

Covariates were introduced individually and one by one in the base model. A covariate was considered as influencing the pharmacokinetic model when its adjunction to the base model induce (i) a decrease of the objective function value (OFV) by 6.6 between the two models, and / or (ii) an improvement of parameter estimation precision and / or (iii) a reduction in Between subject variability (BSV).

Evaluation of the Model

The final population pharmacokinetic model was evaluated using bootstrap, jackknife methods and visual predictive check (VPC). The bootstrap resampling was applied to assess the stability of the final parameter estimates and to confirm the robustness of the final model [8,9]. The bootstrap procedure was performed using Wings for NONMEM in order to compare the mean and median values of the pharmacokinetic parameters estimated using at least 1000 populations of 17 patients, with the population parameters obtained from the original data set [9]. The Jackknife method consists at a 20% at-a-time omission applying to the original data set and then reanalyzed by NONMEM. The 95% CI for parameter estimates obtained by Jackknife procedure were compared with those of the final model using the mean and SE of the NONMEM estimation [10].

Building of a Bayesian Estimator

PopPK parameters where then used as priors to develop a Bayesian estimator (MAP-BE) for the estimation of cyclosporine AUC using a limited sampling strategy (LSS). The best LSS among the combinations of a maximum of three sampling times was selected on the basis of the D-optimality criterion (implemented in Adapt II Pharmacokinetic / Pharmacodynamic Biomedical Simulations Resource, Los Angeles, CA, USA) computed Systems Analysis Software; on the population pharmacokinetics parameters. The predictive performance of the MAP-BE were tested (i) by means of the bias (mean error) and the root mean square error (RMSE) comparing the estimated AUC to that obtained by the trapezoidal rule [11]; and (ii) by its ability to provide similar dose adjustments to those obtained using all the available time-points.

AUC-Guided Dosing

After each AUC-assessment, dose adjustments were made to reach the predefined target.
AUC_0-12 which is 3800ng.h/mL. According to the model there was a linear correlation between dose and AUC. Dose adjustment was made by the model according to the formula D_new = D_current x AUC_target/AUC_current.

Results

A total of 17 full pharmacokinetics profiles were analyzed in this study. Patient characteristics were summarized in (Table 1).

Cyclosporine exhibited a high inter-patient variability (Figure 1): trough concentration= 90.38 ± 65.84ng/mL, C_max = 762 ± 343ng/mL, time to reach the C_max = 1.61 ± 0.37 and AUC_0-12 = 2457 ± 1.24ng.h/mL, CV values are 62.89%, 44.98%, 22.96 % and 50.68% respectively (Table 2).

Table 1: Characteristics and laboratory test results of patients

Figure 1: Concentration of Cyclosporine in 17 renal transplant patients participating in the full pharmacokinetics study based on ten blood samples.

Table 2: Pharmacokinetics parameters for the measured PK profiles

Population Pharmacokinetics

Cyclosporine pharmacokinetics was well described by a two-compartment model combined with an Erlang distribution to describe the absorption phase. Five delay compartments gave the best improvement to the model. This model showed a residual proportional error equal to 26.8% and imprecision parameter estimate less than 26% (8.7 - 25.6%). Models tested with one, two, three and four delay compartments and models with only a central compartment did not show improvement. Among the investigated covariates, no one affected significantly cyclosporine pharmacokinetics (Figure 2,4,5).

Figure 2: Correlation between AUC calculated based on trapezoidal rule and cyclosporine trough concentration.

Figure 4: Correlation of predicted inter dose area under the curve (AUC) of cyclosporine versus reference AUC obtained by the trapezoidal method.

Figure 5: Distribution of residual doses between MAP-BE suggested dose and those obtained by trapezoidal rules.

The scatter plot of predicted (PRED) and individual model-predicted (IPRED) concentrations versus observed concentrations showed no major bias (Figure 3-a,b,c,d), and 94.5% of weighted residuals were homogenously distributed over the sampling time period and within an acceptable range (-2 to 2). The population pharmacokinetic parameters obtained with the final model are summarized in (Table 3).

Figure 3a-d: Quality of pharmacokinetic fitting in the final model
Figure 3a: Individual predicted concentrations (IPRED) versus observed concentrations (DV)

Figure 3b: Predicted concentrations (PRED) versus observed concentration (DV)

Figure 3c: Individual residuals (RES) versus predicted concentrations (PRED)

Figure 3d: Individual weighted residuals (WRES) versus sampling time (TIME)

Table3: Finales estimates of population pharmacokinetics parameters (OFV= 1604)
ME: mean error; SE: Standard error; RES: Residual standard error; BSV: Between subject variability; WSV: Within subject variability; VC: central volume of distribution; VP: Peripheral volume of distribution; Q: Intercompartmental volume of distribution; CL: Total systemic clearance; Ktr: absorption rate; CI: Confidence Interval.

Evaluation of the Model

Evaluation of the final model using the Bootstrap analysis shows that 71.7% of the 1000 runs were successful and that the mean parameter estimates were not statistically different from those obtained using the original dataset. The mean, SE and 95% CI for each parameter estimated by the bootstrap resampling are shown in the (Table 3,4). Values for bootstrapping and the final model coincided well except SE for CL and VC. The mean and SE of jackknife estimates and 95% CI coincided also with parameter estimate of the final model, except SE for CL and VC. The final model selected was also evaluated using a visual predictive check (VPC) to check its accuracy and robustness. A total of 1000 replicates of the original dataset were simulated using the final model to generate expected concentrations and the 90% prediction intervals. The observed data were overlaid on the prediction intervals and compared visually. Because the cyclosporine dose was different in each patient and the pharmacokinetics for cyclosporine was linear, the VPC was based on dose-normalized concentration.

Table 4: Comparison of 95% confidence intervals estimate by SE of bootstrapping and Jackknife estimate

Bayesian Estimation

PopPK parameters where used as priors to develop a MAP-BE for the estimation of cyclosporine AUC using a limited sampling strategy. MAP-BE based on the three blood concentrations measured at 0, 60 and 180min post-dose provided a good estimation of cyclosporine AUC. The mean bias between trapezoidal and Bayesian AUCs was 5, 47 + 29% with 82% of the patients having AUC bias < 20%. This MAP-BE is also characterized by a good precision (RMSE = 28.6%). Using a target AUC of 3, 8 mg.h/L and comparing doses calculated by the MAP-BE and those by trapezoidal rules, we found that MAP-BE suggested similar doses to those proposed using all concentrations in 14 out of 17 patients.

Discussion

We have developed in this study a population pharmacokinetics model for cyclosporine in a Tunisian population of renal transplant patients. It is the first model developed for a generic formulation of cyclosporine and in Tunisian renal transplant patients. In fact, Equoral_® is equivalent to Neoral_® [12], however bioequivalence studies have some limits. They are done only in healthy young volunteers. Otherwise, cyclosporine absorption differs between healthy volunteers and transplant recipients [13] and varies depending on time after transplantation and on the type of organ graft [14-16] Moreover, patient characteristics such as age [17,18] ethnicity [19,20] or co-morbid disease [19,21,22] also affect cyclosporine absorption. In this case pharmacokinetics models allow a better comprehension of pharmacokinetics comportments of the generic formulation.

Our model was developed using full pk profiles (10 concentrations: from pre-dose to 12 h post-dose) and consisted on a two compartment model with Erlang distribution to describe the absorption phase. Cyclosporine pharmacokinetics models reported previously were essentially based on a one [23] two compartments [24] or three compartment models [25]. Models with transient compartments were also described [26]. Pharmacokinetics of cyclosporine in our population was well described by the Erlang distribution. In fact, the addition of transient compartments to the two compartment model showed an important decrease of the objective function value and a better estimation of pharmacokinetics parameters comparing to the other tested models. In this study all tested covariates are non significant and don’t influence the pharmacokinetics of cyclosporine. This may be explained by the little number of patients included in this study. Our results are suitable with those described previously. The CL and VC of this model were 33 l.h^-1 and 65.5 l respectively. In the literature, cyclosporine CL varies between 28 and 30. l.h^-1 [23, 27, 28]. The VC was about 79 [11].

Using PopPK parameters of the final model and on the basis of three blood concentrations measured at 0, 60 and 180 min post-dose, we developed a Bayesian estimator for the estimation of cyclosporine AUC_0-12. Our Bayesian estimator provided a good estimation of AUC with a mean bias equal to 5, 47 ± 29% and a precision (RMSE) of 28.6%. This Bayesian estimator is able to provide similar dose adjustments to those obtained using all the available time-points. This prediction error should not have important clinical consequences, with respect to the proposed therapeutic range for cyclosporine. In addition our bayesian estimation suggested similar doses in 14 out of 17 patients. Cyclosporine dose adjustment in renal transplant patients was described with Sandimun^® and Neoral^® formulation. An accuracy of 13.1% was found in the study of Hélène Bourgoin et al. [2] for a Bayesian estimation using three blood concentrations measured at 0, 60 and 120min.

In another hand, dose adjustment using Bayesian estimation seems to be specific of the analytical assay [24]. So it is not recommended to develop MAP-BE for a specific assay using results from a different one.

In summary, we report a popPK model for Equoral^® when given to Tunisian kidney recipients. A MAP-BE using only three blood concentrations that estimates accurately Equoral^® exposure in these patients was developed and could allow dose adjustments based on the AUC in clinical setting.

Conclusion

Therapeutic drug monitoring of cyclosporine using twelve concentrations to calculate area under the curve in order to adjust drug doses remains invasive and cannot be achieved frequently in renal transplant patients. We developed in a Tunisian population a pharmacokinetic model for cyclosporine.

References

1. Boekmann AJ, Sheiner LB, Beal SL (1992) NONMEM user’s guide, part V: introductory guide. NONMEM Project Group, University of California at San Francisco.


2. Bourgoin H, Paintaud G, Büchler M, Lebranchu Y, Autret-Leca E, et al. (2005) Bayesian estimation of cyclosporine exposure for routine therapeutic drug monitoring in kidney transplant patients. Br J of Clin Pharmacol 59: 18-27.


3. Kovarik JM, Mueller EA, van Bree JB, Tetzloff W, Kutz K (1994) Reduced inter and intra individual variability in cyclosporine pharmacokinetics from a microemulsion formulation. J Pharm Sci 83: 444-446.


4. Jebabli N, Klouz A, Bahlous A, Gaïes E, Trabelsi S, Lakhal M, Belkahia C (2007) Comparison of three methods for cyclosporine therapeutic monitoring. Transplant Proc 39: 2557-2559.


5. Rousseau A, Leger F, Le Meur Y, Saint-Marcoux F, Paintaud G, Buchler M, Marquet P (2004) Population pharmacokinetic modeling of oral cyclosporin using NONMEM: comparison of absorption pharmacokinetic models and design of a Bayesian estimator. Ther Drug Monit 26: 23-30.


6. Benkali K, Premaud A, Picard N, Rerolle JP, Toupance O, et al. (2009) Tacrolimus population pharmacokinetic- pharmacogenetic analysis and Bayesian estimation in renal transplant recipients. Clin Pharmacokinet 48: 805-816.


7. Wählby U, Jonsson EN, Karlsson MO (2002) Comparison of stepwise covariate model building strategies in population pharmacokinetic-pharmacodynamic analysis. AAPS PharmSci 4: E27.


8. Ette EI (1997) Stability and performance of a population pharmacokinetic model. J Clin Pharmacol 37: 486-495.


9. Parke J, Holford NH, Charles BG (1999) A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models. Comput Methods Programs Biomed 59:19-29.


10. Ette EI, Ludden TM (1995) Population pharmacokinetic modeling: the importance of informative graphics. Pharm Res 12: 1845-1855.


11. Sheiner LB, Beal SL (1981) Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm 9: 503-512.


12. Masri MA, Haberal M, Rizvi A, Stephan A, Bilgin N (2004) The pharmacokinetics of equoral versus neoral in stable renal transplant patients: A multinational multicenter study. Transplant Proc 36: 80-83.


13. Johnston A (2013) Equivalence and interchangeability of narrow therapeutic index drugs in organ transplantation. Eur J Hosp Pharm Sci Pract 20: 302-307.


14. Kovarik JM, Mueller EA, van Bree JB, Flückiger SS, Lange H, et al. (1994) Cyclosporine pharmacokinetics and variability from a microemulsion formulation- a multicenter investigation in kidney transplant patients. Transplantation 58: 658-663.


15. Cooney GF, Jeevanandam V, Choudhury S, Feutren G, Mueller EA, et al. (1998) Comparative bioavailability of Neoral and Sandimmune in cardiac transplant recipients over 1 year. TransplantProc 30:1892-1894.


16. Freeman D, Grant D, Levy G, Rochon J, Wong PY, et al. (1995) Pharmacokinetics of a new oral formulation of cyclosporine in liver transplant recipients. Ther Drug Monit 17: 213-216.


17. Cooney GF, Habucky K, Hoppu K (1997) Cyclosporin pharmacokinetics in paediatric transplant recipients. Clin Pharmacokinet 32: 481-495.


18. Dunn S, Cooney G, Sommerauer J, Lindsay C, McDiarmid S, et al. (1997) Pharmacokinetics of an oral solution of the microemulsion formulation of cyclosporine in maintenance pediatric liver transplant recipients. Transplantation 63: 1762-1767.


19. Schroeder TJ, Hariharan S, First MR (1995) Variations in bioavailability of cyclosporine and relationship to clinical outcome in renal transplant subpopulations. Transplant Proc 27: 837-839.


20. Lindholm A, Welsh M, Rutzky L, Kahan BD (1993) The adverse impact of high cyclosporine. Clearance rates on the incidences of acute rejection and graft loss. Transplantation 55: 985-993.


21. Cooney GF, Fiel SB, Shaw LM, Cavarocchi NC (1990) Cyclosporine bioavailability in heart-lung transplant candidates with cystic fibrosis. Transplantation 49: 821-823.


22. Tan KK, Hue KL, Strickland SE, Trull AK, Smyth RL, et al. (1990) Altered pharmacokinetics of cyclosporin in heart-lung transplant recipients with cystic fibrosis. Ther Drug Monit 12: 520-524.


23. Wu KH, Cui YM, Guo JF, Zhou Y, Zhai SD, et al. (2005) Population pharmacokinetics of cyclosporine in clinical renal transplant patients. Drug Metab Dispos 33: 1268-1275.


24. Schädeli F, Marti HP, Frey FJ, Uehlinger DE (2002) Population pharmacokinetic model to predict steady-state exposure to once-daily cyclosporin microemulsion in renal transplant recipients. Clin Pharmacokinet 4: 59-69.


25. Fanta S, Jönsson S, Backman JT, Karlsson MO, Hoppu K (2007) Developmental pharmacokinetics of ciclosporin- a population pharmacokinetic study in paediatric renal transplant candidates. Br J Clin Pharmacol 64: 772-784.


26. Irtan S, Saint-Marcoux F, Rousseau A, Zhang D, Leroy V (2007) Population pharmacokinetics and bayesian estimator of cyclosporine in pediatric renal transplant patients. Ther Drug Monit 29: 96-102.


27. Hesselink DA, van Gelder T, van Schaik RH, Balk AH, van der Heiden IP, et al. (2004) Population pharmacokinetics of cyclosporine in kidney and heart transplant recipients and the influence of ethnicity and genetic polymorphisms in the MDR-1, CYP3A4, and CYP3A5 genes. Clin Pharmacol Ther 76: 545-556.


28. Bourgoin H, Paintaud G, Buchler M, Lebranchu Y, Autret-Leca E, et al. (2005) Bayesian estimation of cyclosporin exposure for routine therapeutic drug monitoring in kidney transplant patients. Br J Clin Pharmacol 59: 18-27.