Population pharmacokinetic and pharmacodynamic modeling of propofol for long-term sedation in critically ill patients: a comparison between propofol 6% and propofol 1%.

Abstract

A population pharmacokinetic and pharmacodynamic model of propofol for long-term sedation in critically ill patients is described, because limited information is available in these patients. In the models the influence of time-independent covariates, in particular, the propofol formulation (propofol 6% versus propofol 1%), and of time-dependent covariates was investigated. Twenty critically ill, mechanically ventilated patients received propofol formulated as propofol 6% (n = 10) or propofol 1% (n = 10) during a 2- to 5-day period. The level of sedation was assessed with the Ramsay 6-point scale. The data from a short-term sedation study in 24 patients after cardiac surgery were included. Population pharmacokinetic and pharmacodynamic modeling was performed with NONMEM. The pharmacokinetics was adequately described by a 2-compartment model. The propofol formulation was not a significant covariate for the pharmacokinetics, whereas serum triglyceride concentration (TG) and relative body temperature (T(c)) were significant covariates for elimination clearance (CL). The population pharmacokinetic parameters were as follows: CL = 2.2 + 0.27 x T(c) - 0.049 x TG (mean, 2.1 L/min); volume of central compartment, 22.2 L; CL (distribution), 1.5 L/min; and volume of peripheral compartment, 168 L. The addition of other time-independent covariates (long-term versus short-term sedation study, as well as physiologic characteristics) or time-dependent covariates (duration of propofol infusion, additional midazolam rates, and hemodynamic parameters) to the model did not improve the quality of fit. For the pharmacodynamics, the probability that the sedation level was equal to, or more than, a specific score was described with the use of a sigmoid inverse logit of the maximal achievable probability model. The values for the inverse logit of the concentration causing half of the maximal effect for Ramsay sedation scores of 2 through 6 were 0.13 +/- 0.09, 0.31 +/- 0.17, 0.56 +/- 0.24, 0.79 +/- 0.31, and 1.78 +/- 0.65 mg/L, respectively (population mean +/- SE). Interindividual variability was high, with a coefficient of variation of 119% in the 50% effective concentration values. No covariates were identified. The population models in critically ill patients showed no differences in pharmacokinetics or pharmacodynamics between propofol 6% and propofol 1%. TG and T(c) appeared to be significant covariates for elimination clearance. For the pharmacodynamics, when propofol concentrations were between 0.75 and 1.5 mg/L, Ramsay sedation score 6 was most probable (40%-75%) and the probability for Ramsay sedation score 5 was 20% to 40%. Large pharmacodynamic variabilities were observed.