Population pharmacokinetic modeling of tacrolimus using real-world data for precision dosing in pediatric liver transplant recipients younger than 36 months during the immediate postoperative period.

Background: Tacrolimus dosing during the immediate postoperative period after pediatric liver transplantation is challenging because clearance can change rapidly as graft function recovers, while conventional weight-based dosing provides limited guidance for incorporating measured concentrations into subsequent doses. We developed a population pharmacokinetic model of tacrolimus in recipients younger than 36 months and evaluated its application to model-informed precision dosing (MIPD). Methods: Routine therapeutic drug monitoring data collected through postoperative day 11 were analyzed using nonlinear mixed-effects modeling. Sampling occasions were stratified into six predefined postoperative intervals, and inter-occasion variability in apparent clearance was estimated separately for each occasion. Model performance was evaluated using goodness-of-fit diagnostics, an occasion-stratified prediction-corrected visual predictive check, and Monte Carlo simulation of individual concentration-time profiles. MIPD was implemented using two complementary approaches: a covariate-based equation for initial dosing and maximum a posteriori Bayesian forecasting for subsequent dose refinement. Forward predictive performance was also evaluated by forecasting the trough concentration at the next occasion using covariates alone or in combination with one or two preceding trough measurements. Results: A total of 731 tacrolimus concentrations from 92 patients were analyzed. A one-compartment model with first-order absorption adequately described the data. Clearance decreased with increasing total bilirubin and aspartate aminotransferase, whereas volume of distribution increased with age. Inter-occasion variability in clearance varied across the postoperative course, from a coefficient of variation of 123.8% on postoperative day (POD) 2%-10.7% on POD 5-6, before increasing to 38.8% on POD 9-11. Monte Carlo simulation placed 89.2% of observed concentrations within the 90% prediction interval. Moreover, incorporating one preceding trough reduced the mean absolute error of next-occasion predictions by approximately one-fifth, while incorporating two preceding troughs reduced it by approximately one-third, according to the performance validation. Conclusion: Tacrolimus clearance in pediatric liver transplant recipients younger than 36 months is characterized by substantial occasion-specific variability during the immediate postoperative period, consistent with dynamic changes in graft function. Dose individualization should therefore incorporate occasion-specific therapeutic drug monitoring data. The developed model provides a quantitative framework for MIPD, implemented here through a publicly accessible web application (https://smjlab.shinyapps.io/fk506_frontiers/).