A Nonparametric Population Pharmacokinetic Model of Selumetinib in Pediatric Patients Diagnosed With Neurofibromatosis‐I or Plexiform Neurofibromas

A twice‐daily administration of oral selumetinib (SLT) in the fasted state is the only approved pharmaceutical option for treating inoperable neurofibromatosis type I (NF‐1) and plexiform neurofibromas (PN). In children, exposure to SLT is highly variable, and fasting presents a substantial burden. Therapeutic drug monitoring and pharmacokinetic modeling can support individualized therapy accompanied by a more rational alimentary routine. Twenty‐eight children diagnosed with inoperable NF‐1 or PN were recruited at a major pediatric oncological center. Twenty‐two patients donated 156 blood samples in steady state for nonparametric population pharmacokinetic modeling. An equation was developed experimentally for estimating model error. Eleven three‐compartment models were compared in terms of statistical performance. Monte Carlo simulations were performed to validate a limited external model using six additional patients and to compare the trough‐to‐peak SLT concentration ratios simulated for various dosing regimens to develop better control over exposure. A pharmacokinetic model that included total body weight as a covariate provided the best fit between predicted and observed concentrations ( r = 0.994) and the best performance statistics. In the first Monte Carlo simulation, measured concentrations fell within the 0.01%–95% (median: 19.7%) quantiles of the simulated ranges. The second simulation revealed that 6‐h (q6h), 8‐h (q8h), and 12‐h (q12h) dosing intervals would yield comparable trough‐to‐peak concentration ratios, with medians of 0.126 (range: 0.001–0.335), 0.104 (0.000–0.306), and 0.065 (0.000–0.279), respectively. The nonparametric population model provides efficient priors for making individual predictions of SLT concentrations. The simulation did not reveal any disadvantages of q6h or q8h dosing.