2005 Annual Scientific Meeting - CLICK FOR PROGRAMME

Ketamine Pharmacokinetics in Healthy Volunteers: A Pooled Population Analysis

AE Rigby-Jones¹, JR Sneyd¹, AR Absalom²

¹Anaesthesia Research Group, Peninsula Medical School, Plymouth.

²University Dept. of Anaesthesia, Addenbrookes Hospital, Cambridge.

Introduction

Previous work has shown that the Domino pharmacokinetic model¹ has poor predictive performance when used for target controlled infusion (TCI) of ketamine in healthy volunteers².  Whilst the overall performance of the TCI system was satisfactory, there was marked deviation from the predicted venous concentration at the onset and offset of infusion.  The development of a new model may enhance predictive accuracy for future studies.

Methods

We combined data from 3 studies of neuroimaging and behavioural testing in adult volunteers for which sub-anaesthetic TCI ketamine was administered using the Domino¹ model.  The pooled data set comprised 178 observations from 36 adults (20 males, 16 females).  Age ranged from 18 to 64yrs (median 28yrs) and body weight from 55 to 102kg (median 73kg).  Population pharmacokinetic models were constructed using NONMEM software.  The performance of the final pharmacokinetic model was assessed by calculation of the bias, precision, wobble and divergence, according to Varvel³.

Results

The final model was 2-compartmental and weight-proportional.  Typical population parameter values were CL: 19mL/kg/min, Q2: 121mL/kg/min, V1: 0.79L/kg and V2: 3.3L/kg.  Age and gender were not supported as covariates.  The median performance error was -4.4% and the median absolute performance error was 20.7%.  The median wobble was 11.0% and the median divergence was -0.0327%/min. 

Discussion

To explore our findings, simulations were performed using this model and those developed by Domino and by Clements (Figure 1). The bolus simulation predicts far higher plasma concentrations for the Domino model because of its greatly reduced central compartment volume, relative to the other models.  Our model and the Clements model produce very similar predicted profiles following infusion, as a result of their comparable values for clearance and volume of distribution.  The Domino model describes a significantly faster rate of metabolic clearance thus explaining the observed under-prediction of ketamine concentrations by the Domino TCI system following infusion offset in healthy volunteers. 

Figure 1 Bolus (0.25mg/kg over 10s (left)) and infusion (8hrs, 2mg/kg/min (right)) simulations using 3 pharmacokinetic models.

Acknowledgements

The authors wish to thank collaborators from the Cambridge University Department of Psychiatry.

References

1. Domino EF et al. Clin Pharmacol Ther. 1984; 36:645-53.

2. Lee M, Absalom A. 2004 http://www.sivauk.org/Birmingham/Lee.htm

3. Varvel et al.  J Pharmacokinetic Biopharm. 1992; 20: 63-94.