Handling interoccasion variability in model‐based dose individualization using therapeutic drug monitoring data

• DOI: 10.1111/bcp.13901
• Publication Date: 2019-02-15T04:49:51Z
• Authors: João A. Abrantes; Siv Jönsson; Mats O. Karlsson; Elisabet I. Nielsen
• Keywords: Adult; Male; Hemostasis; Factor VIII; Adolescent; Coagulants; Middle Aged; Hemophilia A; Models, Biological; Young Adult; 03 medical and health sciences; Treatment Outcome; 0302 clinical medicine; Humans; Computer Simulation; Drug Dosage Calculations; Drug Monitoring; Child; Aged

Aims This study aims to assess approaches handle interoccasion variability (IOV) in a model‐based therapeutic drug monitoring (TDM) context, using population pharmacokinetic model of coagulation factor VIII as example. Methods We assessed 5 TDM approaches: empirical Bayes estimates (EBEs) from including IOV, with individualized doses calculated based on individual parameters either (i) or (ii) excluding related IOV; and EBEs IOV by (iii) setting zero, (iv) summing variances interindividual (IIV) into single IIV term, (v) re‐estimating the without IOV. The impact varying magnitudes (0–50%) number occasions/observations was explored. were compared conventional weight‐based dosing. Predictive performance prediction error percentiles. Results When lower than IIV, accuracy good for all (50 th percentile [P50] <7.4%), but precision varied substantially between (P97.5 61–528%). Approach most precise forecasting method across wide range scenarios, particularly case sparse sampling high Weight‐based dosing led less predictions scenarios. Conclusions Based studied scenarios theoretical expectations, best approach dose individualization is include generation exclude portion unexplained used calculate future dose.