Pieter Colin (1,2), Douglas J. Eleveld (1), Johannes P. van den Berg (1), Hugo E.M. Vereecke (1), Michel M.R.F. Struys (1,3), Gustav Schelling (5), Christian C. Apfel (4), Cyrill Hornuss (5)
(1) Department of Anesthesiology, University Medical Center Groningen, University of Groningen, The Netherlands; (2)Laboratory of Medical Biochemistry and Clinical Analysis, Faculty of Pharmaceutical Sciences, Ghent University, Belgium; (3) Department of Anesthesia, Ghent University, Gent, Belgium; (4) Department of Epidemiology and Biostatistics, University of California, San Francisco; (5) Department of Anaesthesiology, Klinikum der Universität München, Munich, Germany
Objectives: To develop an extension to the current state-of-the-art PK model for propofol [1] which allows the simultaneous description of propofol plasma and breath concentrations. To apply this model in a Bayesian forecasting setting to investigate whether breath monitoring of propofol could improve the predictive performance of the current PK model for intraoperative propofol plasma concentrations.
Methods: Propofol breath and plasma concentration measurements were obtained from twenty healthy volunteers who were dosed using a target-controlled infusion system (TCI) based on Schnider et al.[2]. The FOCE algorithm with interaction as implemented in NONMEM® (version 7.3; Icon Development Solutions, Hanover, MD, USA) was used to fit different breath models to our dataset. Afterwards, a simulation study was conducted in which cross-validation was used to compare the predictive performance of the final model when different amounts of breath data were used.
Results: The final model consisted of (i) an effect-site compartment to accommodate the delay between plasma and breath measurements, (ii) a scale parameter (K) to accommodate the unit conversion from µg/mL for the plasma measurements to parts per billion (ppb) for the measured breath concentrations and (iii) a linear time-dependent change in K over time. Goodness-of-fit plots and numerical performance metrics (median prediction error (MdPE) and root mean squared error) demonstrate that the developed model adequately describes exhaled propofol concentrations. When this model is used to predict propofol plasma concentrations from measured breath concentrations, the predictive performance is markedly better than the current state-of-the-art PK model. We found that the MdPE decreased from 42.8% when no breath concentrations were used to -1.05% when more than 35 minutes of breath concentrations were used.
Conclusions: We show that the current state-of-the-art pharmacokinetic model is easily extended to reliably describe propofol kinetics in exhaled breath. Furthermore, we show that the predictive performance of the a-priori model is improved by Bayesian adaptation based on the measured breath concentrations thereby allowing further treatment individualization and a more stringent control on the targeted plasma concentrations during general anesthesia.
References:
[1] Eleveld, D.J., Proost, J.H., Cortinez, L.I., Absalom, A.R. & Struys, M.M. A general purpose pharmacokinetic model for propofol. Anesthesia and analgesia 118, 1221-37 (2014)
[2] Schnider, T.W. et al. The influence of method of administration and covariates on the pharmacokinetics of propofol in adult volunteers. Anesthesiology 88, 1170-82 (1998)
Reference: PAGE 25 (2016) Abstr 5746 [www.page-meeting.org/?abstract=5746]
Poster: Drug/Disease modeling - Other topics