Background and objectives: The pharmacokinetics of infliximab are highly variable and influence clinical response in chronic inflammatory diseases. The goal of this study was to build a Bayesian model allowing predictions of upcoming infliximab concentrations and dosing regimen adjustment, using only one concentration measurement and information regarding the last infliximab infusion.
Methods: This retrospective study was based on data from 218 patients treated with infliximab in Tours University Hospital who were randomly assigned to learning (two-thirds) or validation (one-third) data subsets. One-compartment pharmacokinetic and time since last dose (TLD) models were built and compared using learning and validation subsets. From these models, Bayesian pharmacokinetic and TLD models using one concentration measurement (1C-PK and 1C-TLD) were designed. The predictive performances of the 1C-TLD model were tested on two external validation cohorts.
Results: Pharmacokinetic and TLD models described the data satisfactorily and provided accurate parameter estimations. Comparable predictions of infliximab concentrations were obtained from pharmacokinetic versus TLD models, as well as from Bayesian 1C-PK versus 1C-TLD models. The 1C-TLD model showed satisfactory prediction of future infliximab concentrations and provided satisfactory predictions of infliximab steady-state concentration for up to three upcoming visits after a blood sample.
Conclusions: Accurate individual concentration predictions can be obtained using a single infliximab concentration measurement and information regarding only the last infusion. The 1C-TLD model may help to optimize the dosing regimen of infliximab in routine therapeutic drug monitoring.