Tamara van Donge

Chapter 6 102 Exploration of feasible TDM strategies The AUC 0-72 of total (R/S)-ibuprofen was simulated for each individual patient using Monte Carlo simulations (n=1000) based on the original study population and dosing design with inclusion of inter-individual variability. We used linear regression to assess the strength of the relationship between AUC 0-72h and ibuprofen concentrations on various sampling time points during the first 24 hours of treatment. For each simulated individual, we assessed whether the target total ibuprofen AUC 0-72h of ≥900 mg·h/L was achieved. 11 The predictive performance of ibuprofen concentrations for reaching this target was quantified for each time point using AUC-ROC curves. Ultimately, we selected the earliest sampling time point where an AUC-ROC >0.8 was reached, which can be considered an adequate predictive performance. Finally, using the logistic regression models, threshold ibuprofen concentrations that are expected to result in attainment of the target exposure (total (R/S)-ibuprofen AUC 0-72h ≥900 mg·h/L) were calculated with different levels of probability for each sampling time point. Results Study population and samples Twenty-three preterm neonates were included in our study of whom 155 PK samples were collected (Figure 1). The median (range) birth weight and gestational age of our study population was 780 (540 – 1150) g and 25.9 (23.4 – 27.4) weeks, respectively, with a PNA of 3.1 (1.2 – 15.6) days at the start of ibuprofen treatment. Suspected adverse events were reported in eight patients, with one developing necrotizing enterocolitis and oliguria that resulted in early discontinuation of ibuprofen therapy after the first dose (Table 1). Population pharmacokinetics of R- and S-ibuprofen A one-compartment model with first-order absorption was identified for both R- and S- ibuprofen, with first-order elimination and a combined error model estimated separately for both enantiomers. For both enantiomers, prior information was only necessary for volume of distribution since we were able to estimate clearance with high precision without priors. After a thorough review of studies on enantiomer specific PK of ibuprofen in neonates, we extracted informative priors from the PK model from Engbers et al. as the patient characteristics were most similar in terms of GA, PNA and birth weight (Supporting Information 2 - Table S1). 17 The absorption rates of R- and S-ibuprofen were estimated at 0.095 h -1 and 0.162 h -1 , respectively (Supporting Information 2 - Table S2). We found a significant impact of PNA and GA on S-ibuprofen clearance. We could not identify the effect of small for gestational