Use of target controlled infusion to derive age and gender covariates for propofol clearance

Abstract

Background and objective: Attempts to describe the variability of propofol pharmacokinetics in adults and to derive population covariates have been sparse and limited mainly to experiments based on bolus doses or infusions in healthy volunteers. This study aimed to identify age and gender covariates for propofol when given as an infusion in anaesthetized patients.

Study design and setting: One hundred and thirteen patients (American Society of Anesthesiologists class I or II and aged 14-92 years) were anaesthetized for elective surgical procedures with propofol using a target controlled infusion (TCI) system and with alfentanil as a baseline analgesic infusion. Frequent venous blood samples were obtained for measurement of propofol plasma concentrations. PHARMACOKINETIC AND STATISTICAL ANALYSIS: Pharmacokinetic accuracy was determined by the percentage prediction error, bias and precision, as were wobble and divergence. The clearance of propofol from the central compartment was determined for each patient using the computerized record of the infusion profile delivered to each patient, together with relevant blood propofol concentration estimations. For each patient, the nonlinear mixed-effects modelling (NONMEM) objective function was employed to determine the goodness of fit.

Results: The population distribution of propofol clearance was subsequently found to have a Gaussian distribution only in the log domain (mean value equivalent to 26.1 mL/kg/min). The distribution in the normal domain was consequently asymmetric, with a slight predominance of patients with high values of clearance (5% and 95% confidence limits 17.7 and 42.1 mL/kg/min, respectively). Using regression analysis, gender and age covariates were derived that optimized the performance of the target controlled infusion system. The clearance (CL) of propofol in male patients changed little with age (CL [mL/kg/min]=26.88-0.029xAge; r2=0.006) whereas that in female patients had a higher initial value but decreased progressively with age (CL [mL/kg/min]=37.87-0.198xAge; r2=0.246).

Conclusion: We achieved a relatively simple and practical covariate model in which the variability of pharmacokinetics within the study population could be ascribed principally to variability in clearance from the central compartment. Pharmacokinetic simulation predicted an improved performance of the TCI system when employing the derived covariates model, especially in elderly female patients.