A Population Pharmacokinetic Model of Phenobarbitone in Neonates to Determine Oral Bioavailability and Facilitate Individualised Dosing

Posted on by Angela Williams

Background: Phenobarbitone is the most commonly used first-line drug for the treatment of neonatal seizures. A number of previous studies, with small subject numbers, have identified covariates that may influence the pharmacokinetics of phenobarbitone but results have been inconsistent. In particular, oral bioavailability is relatively poorly described and doses are commonly reported as being the same for both intravenous and oral administration. However 2 recent studies have reported oral bioavailability as 49% and 59% respectively [1, 2]. A model based on a larger data set may assist in identifying covariates that may impact dosing in these patients.

Methods: A population pharmacokinetic model was built based on routine therapeutic drug monitoring data from 112 infants treated at the Royal Brisbane and Women’s Hospital Neonatal Intensive Care Unit. Population modelling was performed using NONMEM 7.3 and PsN 4.7. R studio and the R packages Xpose and VPC were used for data exploration and visualisation. One and two compartment models were tested. Body weight with allometric scaling on Clearance (CL) and Volume of Distribution (V) were included a priori in the structural model. Covariates tested included age (post-menstrual, gestational and post-natal), Apgar scores, concomitant treatment with phenytoin, presence of infection and method of nutrition.

Results: A one-compartment model provided an adequate fit to the data. Typical clearance increased with patient post-natal age (PNA) and was best modelled using the equation CL = 5.1 *WT0.75* (PNA/6.25)0.43(mL/h) were weight is in kg, PNA in days and 6.25 is the median post-natal age. Volume of distribution (V) was best modelled using the equation V = 797 * WT1.0(mL). Oral bioavailability (F) was 85%.  Between-subject variability was 25%, 30% and 21% respectively for CL, V and F. Internal model validation was performed by generating visual predictive check (VPC) and normalised prediction distribution error (npde) plots, and running a non-parametric bootstrap.

Conclusion: This study describes the largest population pharmacokinetic model of phenobarbitone developed to date with estimates of CL and V in line with previously published models. However, the estimate of F is somewhat higher than previously reported but still lower than the assumed F of 100% implied in most recommended dosing regimens. Once externally validated it is intended that this model form the base for the analysis of a larger data set including the covariates of Hypoxic-Ischaemic Encephalopathy and Therapeutic Hypothermia.

References:

  1. Marsot, A., et al., Pharmacokinetics and absolute bioavailability of phenobarbital in neonates and young infants, a population pharmacokinetic modellingapproach.Fundam Clin Pharmacol, 2014. 28(4): p. 465-71.
  2. Voller, S., et al., Model-based clinical dose optimization for phenobarbital in neonates: An illustration of the importance of data sharing and external validation.Eur J Pharm Sci, 2017. 109s: p. S90-s97.

 

2019