A model-based algorithm for mycophenolate dose adjustment in paediatric transplantation patients
Alessandro Di Deo (1), Bianca Maria Goffredo (2), Sara Cairoli (2), Raffaele Simeoli (2), Oscar Della Pasqua (1)
(1) Clinical Pharmacology & Therapeutics, University College London, United Kingdom, (2) Bambino Gesù Children's Hospital IRCCS, Italy
Introduction: The immunosuppressant effect of mycophenolate sodium is determined by mycophenolic acid (MPA), which is the active component released in plasma. Even though MPA is recognised to be well tolerated, there is large inter- and intraindividual variability in the dose-exposure relationship [1,2]. In fact, the incidence rate of acute rejection in paediatric renal transplant patients is still relatively high (15 -35%), despite evidence that clinical response is associated with well-defined target exposure ranges [AUC (30-60 mg·h/L) or Cmin (1-3.5 mg/L)]. Attainment of target exposure or requirement for dose adjustment is assessed by therapeutic drug monitoring (TDM) using sparse sampling [3,4]. However, dose adjustment recommendations based on TDM do not consider the effect of other potential covariate factors on drug disposition. Bayesian forecasting is crucial for model-based dose optimisation based on TDM.
Objectives: The present study aimed to assess the influence of covariate factors on the pharmacokinetic disposition of MPA and to develop an algorithm for optimisation of the dosing regimen in paediatric patients receiving solid organ transplantation.
Methods: This was a retrospective, single centre, observational study in paediatric (N=90) and adult (N=7) solid organ transplant patients. Using sparse therapeutic monitoring (TDM) data (N= 539 samples), a nonlinear mixed effects modelling approach was implemented using prior parameter distributions from a population pharmacokinetic model previously developed for MPA [5]. For each patient, we estimated individual PK parameters based on one to up to 10 consecutive segments of data to predict the concentration data of the succeeding segment. Hence, a segment of data was regarded as either historical data if used for MAP estimation, or as prospective data if used to validate the model predictions. We compared different estimation methods: standard MAP, adaptive MAP and weighted MAP [6]. Following the assessment of the predictive performance of the final model, simulations were performed to identify an algorithm for optimised dose adjustment. In addition, alternative dosing regimens (mg/kg vs. mg/m2) have been tested in conjunction with the evaluation of a limited sampling strategy. The proportion of subjects reaching the target therapeutic exposure range as well as the deviation between observed and predicted exposure were used as metrics of model performance.
Results: Systemic exposure to mycophenolate/MPA can be described by a two-compartment model. Inclusion of body weight and organ function were found to be significant covariate factors on clearance, despite considerable inter-occasion variability in absorption. Median AUC0-12 (90%-prediction intervals) in liver and heart transplanted patients was 33.4 (16.8-62.5) and 21.0 (8.8-73.4) mg/L·h, respectively. The proportion of patients within the target range was around 50%, whereas the remainder was below the target. Dosing adjustment using longitudinal data and baseline covariates yields an increased proportion of patients (15%-20%) reaching the desired target therapeutic range (Cmin between 1-3.5 mg/L). Finally, evaluation of different sampling strategies showed a stabilisation of the model predictive performance after the inclusion of 3 samples per patient.
Conclusion: In contrast to empirical dose adjustment, the use of a model-based dosing algorithm allowed the integration of baseline covariate effects into estimation of individual post-hoc estimates of the parameter(s) of interest, ensuring personalised interventions. Moreover, the use of longitudinal data from previous monitoring events can be used within a Bayesian framework to enhance model performance, disentangling the contribution of intraindividual variability in clearance from other sources of variation in drug disposition during chronic therapy.
References:
- Sumethkul V, Na-Bangchang K, Kantachuvesiri S, Jirasiritham S. Standard dose enteric-coated mycophenolate sodium (myfortic) delivers rapid therapeutic mycophenolic acid exposure in kidney transplant recipients. Transplant Proc. 2005; 37(2): 861-863.
- Zeng L, Blair EY, Nath CE, Shaw PJ, Earl JW, Stephen K, et al. Population pharmacokinetics of mycophenolic acid in children and young people undergoing blood or marrow and solid organ transplantation. Br J Clin Pharmacol. 2010; 70(4): 567-579.
- Mourad M, Malaise J, Chaib Eddour D, De Meyer M, Konig J, Schepers R, et al. Correlation of mycophenolic acid pharmacokinetic parameters with side effects in kidney transplant patients treated with mycophenolate mofetil. Clin Chem. 2001; 47(1): 88-94.
- Staatz CE, Tett SE. Clinical pharmacokinetics and pharmacodynamics of mycophenolate in solid organ transplant recipients. Clin Pharmacokinet. 2007; 46(1): 13-58.
- Shum B, Duffull SB, Taylor PJ, Tett SE. Population pharmacokinetic analysis of mycophenolic acid in renal transplant recipients following oral administration of mycophenolate mofetil. Br J Clin Pharmacol. 2003; 56(2): 188-197.
- Guo T, van Hest RM, Zwep LB, Roggeveen LF, Fleuren LM, Bosman R J, et al. Optimizing predictive performance of bayesian forecasting for vancomycin concentration in intensive care patients”. Pharm Res. 2020; 37, 1-9.