Model‐Informed Precision Dosing of Infliximab in Korean Inflammatory Bowel Disease Patients: External Validation of Population Pharmacokinetic Models

• Published in: CPT: pharmacometrics and systems pharmacology
• Main Authors: Kim, Yoonjin, Baek, Seung Hwan, Jang, In‐Jin, Chung, Jae‐Yong
• Format: Journal Article
• Language: English
• Published: United States 04.08.2025
• Subjects: model evaluation; individualization; therapeutic drug monitoring; population pharmacokinetics
• ISSN: 2163-8306; 2163-8306
• DOI: 10.1002/psp4.70089

Underexposure to infliximab often leads to loss of response in patients with inflammatory bowel disease (IBD). Model‐informed precision dosing (MIPD) offers a superior approach to maintaining target infliximab concentrations compared to empirical dosage adjustment. This study aims to externally validate the population pharmacokinetic (PK) models implemented in TDMx, an online MIPD dashboard system, for adult and pediatric Korean IBD patients before clinical use. This retrospective study included 199 IBD patients (142 adults, 57 children) treated with intravenous infliximab at Seoul National University Hospital (Seoul, Republic of Korea) from 2019 to 2023. Three adult and seven pediatric models were evaluated based on accuracy, precision, goodness of fit plots, prediction‐corrected visual predictive checks, and normalized prediction distribution errors. For adults, the Passot model showed the best fit (mean percentage error (MPE) 26.4%, mean absolute error (MAE) 1.1 mg/L, relative root‐mean square error (rRMSE) 159.0%), whereas all pediatric models were unsuitable for clinical use (MPE 30.4%–143.4%, MAE 1.4–2.6 mg/L, rRMSE 96.3%–564.0%). Predictive performance was compared between datasets with or without accurate information on antibodies‐toward‐infliximab (ATI), as well as with and without previous concentrations. Assuming all patients were ATI positive improved predictive performance, likely due to the inherent positive bias of the population PK models. Incorporating previous concentrations improved predictions for adult models, achieving acceptable accuracy and precision (Passot model: MPE 17.5%, MAE 1.8 mg/L, rRMSE 80.3% with one concentration). However, pediatric models remained clinically unacceptable, highlighting the need to develop models specifically tailored for this population.