055: Population Pharmacokinetic Modeling to Assess the Impacts of Clinical Variables on Mycophenolate Total Body Clearance in Pediatric Kidney Transplant Recipients

Statement of Purpose: Mycophenolic acid (MPA) is an immunosuppressant commonly prescribed in pediatric kidney transplant patients. The plasma exposure of MPA is highly variable in this population and is associated with adverse outcomes. Unfortunately, with only 4 MPA population pharmacokinetic (PK) models available in the literature, the causes and correlates of the PK variability remain unknown. The objective of the current study was to construct a novel non-linear mixed-effects (NLME) population PK model with the goal to elucidate potential significant covariates affecting the PK of MPA in this population.

Description of Methods & Materials: This was a PK modeling study using retrospectively collected data from pediatric kidney transplant patients (≤18 years when MPA data were initially collected) on oral mycophenolate mofetil from Alberta Children’s Hospital (Calgary). NLME modeling was conducted using stochastic approximation expectation-maximization algorithm in Monolix 2021R2 (Lixoft SAS, France) to determine the population PK estimates, inter-individual variabilities, and inter-occasional variabilities of MPA. Covariate modeling was performed using the Model Proposal function in Monolix in conjunction with an already published stepwise inclusion/elimination protocol from our laboratory (e.g., Rong Y, Mayo P, Ensom MHH, Kiang TKL; Clin Pharmacokinet.2019;58(11):1483-1495). A systematic approach was utilized for selecting the optimal model, including assessments of the objective function values, relative standard errors, goodness-of-fit plots, prediction-corrected visual predictive checks, and numerical predictive checks. Confidence intervals of each population PK parameters were determined by bootstrapping analyses (n=500 runs). Estimated glomerular filtration rate (eGFR) was calculated by Schwartz formula (eGFR=0.413×(height[cm]/serum creatinine[mg/dL])).

Data & Results: Our sample population included 50 pediatric kidney transplant patients (female:25) with a total of 219 MPA concentration-time profiles (pre-dose, 1-, 2-, and 4- hours post-dose; with some subjects reporting data from more than one occasion). Mean (±standard deviation) age was 12.8±4.8 years at the time of the first collection. The range of post-transplant times for the sample population (n=50) and all occasions (n=219) were 37-3854 and 37-5056 days, respectively. The majority (~85%) of patients were co-prescribed tacrolimus, with the remaining on cyclosporine. The best structural model (screening 216 variations) was a two-compartment, first-order absorption with lag time, and linear elimination model with lognormal distributed proportional residual errors. The estimated absorption rate constant (2.52h-1), lag time (0.166h), volumes of distributions of the central (22.8L) and peripheral (216L) compartments, and inter-compartment clearance (17.6L/h) were comparable to historical data. However, the total MPA clearance (0.72L/h) in our model was relatively reduced, which may be attributed to concurrent tacrolimus (i.e., lack of cyclosporine interaction) and longer post-transplant time (i.e., stabilized graft function). Of the 16 clinical variables assessed, only eGFR was a significant covariate which positively, exponentially correlated with total MPA clearance.

Interpretation, Conclusion or Significance: We have successfully established and validated a novel population MPA PK model in pediatric kidney transplant patients. A positive, non-linear relationship between eGFR and total MPA clearance has been identified for the first time using this rigorous approach. Our data suggest renal function should be considered when dosing MPA, but further investigations are required to characterize the underlying mechanisms.

Disclosures: The authors declare no conflicts of interest.