Evaluation of the insightRX software package for the pharmacokinetic analysis of infliximab data in patients with inflammatory bowel disease

Abstract

Abstract Background Inflammatory bowel disease (IBD) is identified as chronic intestinal inflammation, mostly ulcerative colitis (UC) and Crohn's disease (CD). The treatment of IBD is mainly known by the use of infliximab (IFX), a chimeric immunoglobulin G monoclonal antibody against tumour necrosis factor alpha. Infliximab trough levels vary between IBD patients; its effectiveness during treatment is associated with maintaining a trough concentration between 3 and 7 µg/mL.9 Therefore, adjustment of dosage regimens to achieve this target range is required in clinical practice. InsightRX® is a maximum; posteriori probability Bayesian software package that can be used to help in designing new dosage regimens.11,12 Objective This study aims to evaluate the performance of previously developed population pharmacokinetic (PK) model in IBD patients for dose individualization for CD and UC patients in a clinical setting. Method Bayesian analysis was conducted among a cohort of 10 adult patients with IBD who had IFX trough concentrations measured using a published population PK model based on data from 112 children in the REACH trial and 450 adults in the ACCENT trial (2011). Infliximab PK parameters and concentrations were then predicted based on each patient’s dosing history and compared with population estimates and actual measured concentrations, respectively, by calculating mean prediction error (MPE) for bias and root square prediction error (RSME) for imprecision. In addition, recommended dosing regimens and model fit for each patient were examined to assess the efficient use of software in a clinical setting. Key findings The predictive performance of the model varied between induction and maintenance phases. Individual clearance prediction was found to be accurate and precise at the induction phase; MPE was 74.0 mL/day (–68.4 to 216.4 mL/day) and RSME was 83.3 mL/day. Concentration prediction was accurate in the maintenance phase, with no bias at 0.6 µg/mL (–0.8 to 2.0 mL/day) and 0.6 µg/mL of precision for first concentration; for the second concentration, this was –1.2 µg/mL (–2.8 to 0.3 mL/day) of MPE and 0.3 of RSME. Conclusion The study highlights the importance of sufficient data information for obtaining good predictions. The software proved its efficient use within clinical practice in terms of designing multiple dosage regimens and displayed the kinetics of the drug, thus enhancing a better understanding of patient outcomes.