Danica Michaličková (1), Pavla Pokorná (1,2,3), Dick Tibboel (3), Ondřej Slanař (1), Catherijne A.J. Knibbe (4,5), Elke H.J Krekels (4)
(1) Institute of Pharmacology, First Faculty of Medicine & General University Hospital, Charles University, Prague, Czech Republic; (2) Department of Pediatrics, First Faculty of Medicine & General University Hospital, Charles University, Prague, Czech Republic; (3) Intensive Care and Department of Pediatric Surgery, Erasmus MC-Sophia Children’s Hospital, Rotterdam, The Netherlands; (4) Division of Pharmacology, Leiden Academic Center for Drug Research, Leiden, The Netherlands; (5) Department of Clinical Pharmacy, St. Antonius Hospital, Nieuwegein, The Netherlands
Objectives: Phenobarbital is one of most frequently used anticonvulsive drugs in pediatric patients undergoing extracorporeal membrane oxygenation (ECMO). The use of ECMO is associated with significant changes in drug pharmacokinetics (PK) [1]. The aim of this study was to characterize the PK of phenobarbital in neonates on ECMO.
Methods: Data from therapeutic drug monitoring (TDM) were available from 13 (6 female, 7 male) neonates (median (IQR), body weight (BW): 3.62 (2.65-3.80) kg; postnatal age (PNA): 13 (5-21) days; gestational age: 38 (38-41) weeks) receiving veno-venous (VV) or veno-arterial (VA) ECMO, yielding 5 phenobarbital concentrations before ECMO, 31 during ECMO, and 19 concentrations after ECMO. The median loading dose of phenobarbital was 7.5 (8.5-16) mg/kg, while the median maintenance dose was 6.9 (4.5-8.5) mg/kg/day. Phenobarbital levels ranged between 2.8 and 56.4 mg/L.
Population PK analysis was performed using NONMEM 7.3.0 [2]. For the structural model, one and two compartment models were tested. Log-normally distributed inter-individual variability terms were tested on each PK parameter. Proportional, additive and combination error models were tested for the residual error model. To disentangle the impact of maturation from other disease related and treatment related covariates, maturation functions for clearance (CL) and volume of distribution (Vd) were obtained from a previously published model in patients with an overlapping age-range [3].
The following covariates were evaluated: laboratory values, concomitant therapy, and ECMO therapy (on/off, duration, modality (VV, VA), speed, flow, change of circuit, time after start and stop of ECMO). Covariates were included based on a forward inclusion and backward deletion (p<0.05 and p<0.01, respectively). Additional criteria for covariate selection were relative standard error (RSE) of the estimates, physiological plausibility, and absence of bias in goodness-of-fit (GOF) plots. The final model was validated using normalized prediction distribution errors (NPDE) [4].
Results: In a one-compartment model, CL and Vd for a typical neonate of median birth BW (3.21 kg) at median PNA (13 days) off ECMO were 0.0096 L/h (RSE = 11%)) and 2.72 L (16%), respectively. The coefficients of variation for inter-individual variability (IIV) for CL and Vd were 29.4 % (26%) and 45.3 % (17%), respectively. A proportional error with a coefficient of variation of 4.41% (32%) provided the best description of the residual variability.
The maturation function could accurately describe the observed concentrations before ECMO start, as indicated by the lack of bias in GOF plots. During ECMO, CL was found to be increased and this increase was time-dependent. Over the 12-day period observed in this study, the relationship between CL and time since the start of ECMO was best described by a linear function. Furthermore, phenobarbital CL reduced after decannulation compared to CL during ECMO, with an initial decrease, followed by an increase according to the maturation fucntion. Changes in Vd during ECMO could not be identified, possibly due to sparse data collection shortly after the start of ECMO, that would prevent the estimation of changes in this timeframe. The predictive properties of other tested variables were not statistically significant.
The distribution of the NPDEs in this dataset had a mean of -0.0799 and variance = 1.034. Neither of these values were significantly different from the expected values of 0 and 1, respectively. No bias could be observed in NPDE over time and versus the predicted concentration in plots stratified for before, during and after ECMO.
Conclusions: Continuously decreasing phenobarbital exposure in patients during ECMO treatment, resulting from the time-dependent increases in phenobarbital CL, may increase the risk of therapeutic failure over time. Hence, these results strongly indicate a need for regular and repeated TDM measurements for phenobarbital in neonates on ECMO.
References:
[1] Wildschut DE et al. Current Drug Metabolism 2012; 13(6):767-77.
[2] Beal SL et al. 1989-2011. NONMEM Users Guides. Icon Development Solutions, Ellicott City, Maryland, USA.
[3] Völler S et al. European Journal of Pharmaceutical Sciences 2017; 109S:S90-S97.
[4] Comets E et al. Computer Methods and Programs in Biomedicine 2008; 90(2):154-66.
Reference: PAGE 28 (2019) Abstr 9045 [www.page-meeting.org/?abstract=9045]
Poster: Drug/Disease Modelling - Paediatrics