Influence of formulation on propofol pharmacokinetics and pharmacodynamics in anesthetized patients

In anesthesia with propofol, variability persists besides sophisticated effect targeting. Drug formulation may be another factor. We have analyzed, retrospectively, the pharmacokinetics (PK) and pharmacodynamics (PD) in monitored surgery patients anesthetized with one each of five formulations of propofol.

Propofol 1% ('form' 1: Diprivan(Zeneca Limited, Macclesfield, UK), 2: Recofol(Schering Espana, Madrid, Spain), 3: Ivofol(Juste, Madrid, Spain), 4: Propofol Abbott (Abbott Laboratories, Madrid, Spain), 5: Fresenius (Fresenius Kabi Espana, Barcelona, Spain)) was administered to 77 ASA I-II patients of age [mean (range) 44 (18-65) years]. Induction of anesthesia was with varying propofol doses up to endpoints of either 60 on the Bispectral Index system (BIS) in group I (n = 48, model development) or standard clinical signs in group II (n = 29, validation). Maintenance was with three 10-min infusions of 10, 8 and 6 mg kg(-1) h(-1). Three blood samples were obtained from each subject, immediately after induction, and at 15 and 30 min on maintenance, with BIS and hemodynamic variables recorded at these times also. Total and free blood concentrations (Cb) of propofol were determined with HPLC. Pharmacokinetic and PD models with link equilibration rate ke0, were studied with a mixed-effects procedure (NONMEM).

The induction dose (group I) showed large interindividual variability [mean (range) 163 (90-290 mg)] that correlated significantly with age, basal systolic blood pressure and formulation. The PK of propofol (basic model) was described by a one-compartment model with (typical value [interindividual coefficient of variation percent (CV%)]) CL=2.30 l min(-1) (27%) and V=8.40 l (80%). Weight (WT) and formulation, within NONMEM, were found to be significant covariates for CL and V, reducing their CV% to 25% and 74%, respectively. The final PK/PD model, which includes formulation, showed a 50% reduction in the CV% for both the ke0 and the residual error. This PK/PD model was validated in group II with 33% precision and no bias.

The PK and PD are not equal for all formulations, which contributes to an increase in variability of the observed effect.