B-304 Performance of Alternative Proficiency Testing Program for Bayesian Forecasting Tool of Infliximab Exposure in Clinical Pharmacokinetic Laboratory

Abstract

Abstract Background We have developed a Bayesian (PK) forecasting and dosing tool for Infliximab (IFX), a monoclonal antibody targeting Tumor necrosis factor α. This novel dosing tool employs machine learning and proficiency of the algorithm must be established. Our objective was to implement an alternative proficiency testing program (APT) in the routine clinical (PK) laboratory. Methods The test (PredictrPK® IFX) is validated and measures serum IFX and antibodies to IFX (ATI) concentrations using a homogenous mobility shift assay by size exclusion HPLC and albumin (ALB) by nephelometry. Determinations are inputted with weight, dose, and interdose interval provided on test requisition in the Bayesian forecasting PK tool which uses non-linear mixed effect to calculate the individual parameter estimates, Clearance and predicted trough concentrations. APT is conducted using five previously tested patient samples, blinded to testing personnel and retested in the same manner as the routine method of analysis every six months. APT consists of two components, a component evaluating standalone IFX, ATI, and ALB analytes, and a second component evaluating the Bayesian forecasting PK tool in estimating Clearance and trough concentrations. Criteria for acceptable performance is set at target value ± 20% for standalone analytes and forecasted PK estimates. Acceptable ATP score corresponds to 4/5 samples with passing grade. Results The results from two consecutive APT results with six months intervals is presented in Fig. 1. Comparing the results to the original assayed data at the time of collection, the APT results showed IFX concentration, ATI and ALB within 20% of target value. Clearance and estimated trough concentrations calculated using the Bayesian forecasting tool also met acceptability criteria (100% passing grade). Conclusion We have implemented an APT for Bayesian forecasting tool in the clinical PK laboratory.