A Comparison of Four Pharmacokinetic/ Pharmacodynamic Models of Propofol TCI in an Older Population

Jörg Prinzlin¹, Alison Campbell² and Nick Sutcliffe¹

1. Department of Anaesthesia, Golden Jubilee National Hospital, Clydebank G81 4HX, Scotland

2. University Department of Anaesthesia, Glasgow Royal Infirmary

 

 

 

Background and Goal of Study:

Numerous pharmakokinetic/pharmakodynamic (PK/PD) models for propofol, suitable for driving target controlled infusion (TCI) devices, are now available. These models will predict differing blood and effect site concentrations for a given infusion regimen. Propofol effect is more accurately reflected by effect-site concentration than plasma concentration. Postulating that an ideal model would predict the same effect-site concentration at loss, and regaining, of consciousness, we compared the predictive accuracy of 4 different PK/PD models for TCI propofol. We correlated the predicted effect site concentration at loss and return of clinical endpoints in elderly patients.

 

Materials and Methods: 30 ASA I-II patients over the age of 65 scheduled for elective surgery were recruited. A TCI pump (Fresenius) programmed with the Marsh PK/PD model¹ set to a plasma target concentration (Cpt) of 4mcg/ml was used for induction of anaesthesia. Time to loss of eyelash reflex (LOER) and loss of verbal response (LOVR) was recorded respectively. A target of zero mcg/ml was then set and the patients allowed to lighten. Times were recorded at return of eyelash reflex (ROER) and verbal response (ROVR). The patients were then deepened for a final time and the time at loss of responses recorded again (LOER2 and LOVR2). A PK/PD simulation programme (TIVA trainer) was used to reproduce the infusion regimen and predicted effect site concentrations for these clinical endpoints using the Marsh¹, Schnider², Schüttler³ and White-Kenny (WK)4 PK/PD models.  The predicted effect-site concentration at loss of, and return of, each clinical end point for each model were then compared with linear regression to calculate coefficients of determination (CoD).  We also compared the first and second loss of verbal response and eyelash reflex using the same analysis.

 

Results and Discussion: The CoDs are shown in figure 1.

 

 

Marsh

Schuttler

Schnider

White-Kenny

LOER1/ROER

0.29

0.27

0.13

0.46

LOER2/ROER

0.78

0.53

0.36

0.71

LOVR1/ROVR

0.05

0.02

0.02

0.1

LOVR2/ROVR

0.90

0.76

0.47

0.84

 

 

 

 

 

LOER1/LOER2

0.33

0.29

0.1

0.36

LOVR1/LOVR2

0.13

0.09

0.28

0.2

 

The White-Kenny and Marsh models had the best correlation between similar clinical endpoints except for LOVR1/LOVR2. The Schnider model had the lowest correlation in all comparisons except when comparing LOVR1/LOVR2.

 

Conclusion: In this age group, the White-Kenny and Marsh models had the best correlation between predicted effect-site concentration of propofol at loss of and return of consciousness. None of the models showed good correlation for the predicted effect-site concentration when comparing the first with the second occurrence of anaesthesia.

 

References

1) Marsh B. BJA 1991; 67:41-8

2) Schnider T.  Anesthesiology 1999; 90: 1502-16

3) Schüttler J. Anesthesiology 2000; 92:727-73

4) Kenny GN. Personal communication