Predictability of processed electroencephalography effects on the basis of pharmacokinetic-pharmacodynamic modeling during repeated propofol infusions in patients with extradural analgesia

Background: Pharmacokinetic-pharmacodynamic (PKPD) modeling can be used to characterize the concentration-effect relation of drags. If the concentration-effect relation of a hypnotic drug is stable over time, an effect parameter derived from the processed electroencephalographic signal may be used to control the Infusion for hypnosis. Therefore, the stability of the propofol concentration-electroencephalographic effect relation over time was investigated under non-steady state conditions. Methods: Three propofol infusions (25 mg (.) kg(-1) (.) h(-1) for 10 min, 22 mg (.) kg(-1) (.) h(-1) for 10 min, and 12.5 mg (.) kg(-1) (.) h(-1) for 20 min) were administered to 10 patients during extradural analgesia. Each successive infusion was started immediately after the patient had regained responsiveness after termination of the preceding infusion. Electroencephalography was recorded from bilateral prefrontal to mastoid leads. Electroencephalographic amplitude in the 11- to 15-Hz band and the Bispectral Index were used as electroencephalographic effect variables. PKPD parameters were calculated with use of parametric and nonparametric models based on electroencephalographic data and arterial propofol concentrations derived during the initial infusion, and these were used to predict electroencephalographic effect during the subsequent infusions. The predictability of the electroencephalographic effects was determined by the coefficient of determination (R-2) and of the - 2 log likelihood of the sequential infusions. Results. The direction of electroencephalographic changes in response to the infusions was reproducible. Although PKPD parameters could be estimated well during the initial infusion (median [range] parametric R-2 = 0.74 [0.56-0.95] for electroencephalographic amplitude and 0.90 [0.27-0-99] for Bispectral index), none of the modeling techniques could predict accurately the electroencephalographic effect during subsequent infusions (R-2 = 0.00 [-0.31-0.46] for electroencephalographic amplitude and 0.15 [-0.46-0.57] for Bispectral Index; P < 0.01). Conclusions. The relation between blood propofol concentrations and the electroencephalographic effect under non-steady state conditions is not stable over time and is too complex to be modeled by any of the applied PKPD models.