An Exploratory Model-Informed Precision Dosing Strategy for Atomoxetine Treatment in Chinese Children with ADHD: Integrating Body Weight and CYP2D6 Pharmacogenetics Across Potential Real-World Clinical Scenarios

Objective: We aimed to develop and evaluate population pharmacokinetic (PopPK) models of atomoxetine incorporating clinically available patient information and to explore model-informed dose selection in Chinese children and adolescents with attention-deficit/hyperactivity disorder (ADHD). Methods: This retrospective single-center study included a model-building dataset of 102 patients (247 concentrations) and a patient-level holdout dataset of 65 patients (65 concentrations collected approximately 2 h post-dose). A one-compartment model with first-order absorption and elimination was developed. Body weight (BW) and alternative representations of CYP2D6 genetic information—rs1065852 genotype, activity score (AS), and predicted phenotype—were evaluated. Monte Carlo simulations evaluated steady-state 2 h post-dose concentrations (C2h) across doses of 0.5–1.4 mg/kg/day; the model-informed dose was the lowest evaluated dose, achieving ≥75% probability of reaching the exploratory C2h threshold of 268 ng/mL. Results: BW affected apparent clearance (CL/F) and volume of distribution, whereas rs1065852 genotype and AS additionally explained CL/F variability. In the holdout dataset, population predictions had limited bias (MdPE, 1.41–3.09%) but modest precision (MdAPE, 33.75–37.83%); CYP2D6-informed models improved selected performance measures. Simulation-derived doses ranged from 1.0 to 1.4 mg/kg/day across subgroups. No evaluated dose achieved the probability of target attainment (PTA) criterion for AS = 2 at any BW, or for the CC+CT and AS = 1.5 subgroups weighing 17–<30 kg. Alternative Ka assumptions changed median C2h by at most 3.99% and PTA by 3.58 percentage points. Conclusions: BW and CYP2D6 genetic information explained part of atomoxetine pharmacokinetic variability and influenced simulated concentration attainment. These model-informed findings are exploratory and require prospective, multicenter validation against efficacy, tolerability, and safety outcomes before clinical implementation.