Crohn’s & Colitis Congress™

13 - REFINED POPULATION PHARMACOKINETIC MODEL FOR INFLIXIMAB PRECISION DOSING IN PEDIATRIC INFLAMMATORY BOWEL DISEASE (Room Pinyon 4/5)

Background: Standard weight-based infliximab (IFX) dosing for inflammatory bowel disease (IBD) results in long-term clinical remission in only 40-60% of patients. Patient and disease factors account for variability in IFX exposure. Achieving optimal IFX exposure as assessed by pre-dose trough levels is associated with improved outcomes. Clinical tools are needed to personalize IFX dosing in a dynamic fashion. Aims: The aim of this study was to refine the existing models for IFX pharmacokinetics in pediatric patients with IBD on maintenance therapy using real-practice data. Methods: The electronic medical record was queried for patients receiving IFX infusions at a pediatric medical center with diagnoses of IBD, Crohn’s disease (CD), and ulcerative colitis between January 2011 and March 2017. PK analysis with Nonlinear Mixed Effect Modeling using maintenance IFX trough levels assessed disease activity indices, laboratory values, and demographic data as covariates. The predictive performance of the final model was tested by ROC curve analysis. Results: The 147 patients included: 55 (37%) female, 111 (76%) CD, mean 14.2 years old. Mean IFX clearance was in good agreement with previously reported values. Stepwise covariate modeling with backward elimination resulted in a final model including body weight, albumin, ESR, and ordinal antibodies (ATI) to IFX (<22, 22-200, 200-1000, >1000 ng/mL). Lower albumin and higher body weight, ESR, and ATI were associated with increased clearance. These covariates reduced inter-individual variability of clearance by 47% (Fig. 1). The model predicted an IFX level >5 ug/mL with a sensitivity of 83.3% and a specificity of 77.4%, with AUC of 0.907 (95% CI: 0.886-0.937; Fig. 2). Conclusion: A population PK model including weight, albumin, ESR, and ATI categories accurately predicts IFX trough levels during maintenance therapy in pediatric IBD. The model will be prospectively evaluated as part of a Bayesian adaptive control strategy.