A Population Pharmacokinetics Model of Busulfan in Pediatric Patients with Thalas-semia Major

Purpose: This research aimed to establish a population pharmacokinetic (PPK) model for busulfan (Bu) in Chinese pediatric patients with thalassemia major. We analyzed pharmacokinetic (PK) parameter variability and explored potential covariates affecting Bu disposition using patient data. These findings are intended to support the optimization and personalization of Bu dosage regimens for children with thalassemia major. Methods: Concentration-time samples were collected retrospectively from 62 pediatric patients with tha-lassemia major. These patients had previously received intravenous Bu as a preparatory regimen for al-logeneic hematopoietic stem cell transplantation (allo-HSCT). A PPK model of Bu was developed through nonlinear mixed-effects modeling. This modeling process, conducted using NONMEM software, concur-rently involved data analysis and examination of the effect of covariates on Bu pharmacokinetics. For validation purposes, the resulting model was evaluated against an external dataset consisting of 20 indi-viduals. Results: The pharmacokinetic results were optimally analyzed using a model that incorporated a one-compartment model with first-order elimination. Body surface area (BSA) was subsequently identified as the most significant factor influencing both Bu clearance (CL) and volume of distribution (V). Diagnostic evaluations, encompassing goodness-of-fit plots, normalized prediction distribution errors, and visual pre-dictive checks, confirmed the satisfactory fit and predictability of the final PPK model. Moreover, predic-tion-based diagnostic indices (MDPE%, 15.75; MAPE%, 22.26; F20%, 45.71; and F30%, 58.57) from external validation showed that no significant bias was detected when comparing the model's predicted concentrations against the observed data. Conclusion: The present study developed the first PPK model characterizing the pharmacokinetics of Bu specifically in children with thalassemia major. This study's final PPK model demonstrated that BSA was the key predictive covariate for CL and V.