BACKGROUND: This investigation extended the pharmacokinetic analysis of our previous study, of intravenous dexmedetomidine in 10 healthy male volunteers, and prospectively tested the resulting compartmental pharmacokinetics in an additional six subjects using a computer-controlled infusion pump (CCIP) to target four different plasma concentrations of dexmedetomidine for 30 min at each concentration. METHODS: A three-compartment mamillary pharmacokinetic model best described the intravenous dexmedetomidine concentration versus time profile following the 5 min intravenous infusion of 2 micrograms/kg in our previous study. Nonlinear regression was performed using both two-stage and pooled data techniques to determine the population pharmacokinetics. The pooled technique allowed covariates, such as weight, age, and height of the subjects, to be incorporated into the nonlinear regression to test the hypothesis that these additional covariates would reduce the residual error between the measured concentrations and the predicted values. RESULTS: The addition of age, weight, lean body mass, and body surface area as covariates of the pharmacokinetic parameters did not improve the predictive value of the model. However, the model was improved when subject height was a covariate of the volume in the central compartment. The residual error in the pharmacokinetic model was markedly lower with the pooled versus the two-stage approach. The following pharmacokinetic values were obtained from the pooled analysis of the zero-order dexmedetomidine infusion: V1 = 8.05, V2 = 12.4, V3 = 175 (L), Cl1 = (0.0101*height [cm]) -1.33, Cl2 = 2.05, and Cl3 = 2.0 (L/min). Prospective evaluation of the pooled pharmacokinetic parameters using a computer-controlled infusion in six healthy volunteers showed the precision (average [(absolute error)/measured concentration]) of the CCIP to be 31.5% and the bias (average [error/measured concentration]) to be -22.4%. A pooled regression of the combined CCIP and zero-order data confirmed that the covariate, height (cm), was related in linear fashion to Cl1. A striking nonlinearity of dexmedetomidine pharmacokinetics related to concentration was observed during the CCIP infusion. The final pharmacokinetic values for the entire data set were: V1 = 7.99, V2 = 13.8, V3 = 187 (L), Cl1 = (0.00791*height [cm]) -0.927, Cl2 = 2.26, and Cl3 = 1.99 (L/min). CONCLUSIONS: Pharmacokinetics of dexmedetomidine are best described by a three-compartment model. Addition of age, weight, lean body mass, and body surface area do not improve the predictive value of the model. Additional improvement in CCIP accuracy for dexmedetomidine infusions would require magnification modification of the model based on the targeted concentration.
BACKGROUND: This investigation extended the pharmacokinetic analysis of our previous study, of intravenous dexmedetomidine in 10 healthy male volunteers, and prospectively tested the resulting compartmental pharmacokinetics in an additional six subjects using a computer-controlled infusion pump (CCIP) to target four different plasma concentrations of dexmedetomidine for 30 min at each concentration. METHODS: A three-compartment mamillary pharmacokinetic model best described the intravenous dexmedetomidine concentration versus time profile following the 5 min intravenous infusion of 2 micrograms/kg in our previous study. Nonlinear regression was performed using both two-stage and pooled data techniques to determine the population pharmacokinetics. The pooled technique allowed covariates, such as weight, age, and height of the subjects, to be incorporated into the nonlinear regression to test the hypothesis that these additional covariates would reduce the residual error between the measured concentrations and the predicted values. RESULTS: The addition of age, weight, lean body mass, and body surface area as covariates of the pharmacokinetic parameters did not improve the predictive value of the model. However, the model was improved when subject height was a covariate of the volume in the central compartment. The residual error in the pharmacokinetic model was markedly lower with the pooled versus the two-stage approach. The following pharmacokinetic values were obtained from the pooled analysis of the zero-order dexmedetomidine infusion: V1 = 8.05, V2 = 12.4, V3 = 175 (L), Cl1 = (0.0101*height [cm]) -1.33, Cl2 = 2.05, and Cl3 = 2.0 (L/min). Prospective evaluation of the pooled pharmacokinetic parameters using a computer-controlled infusion in six healthy volunteers showed the precision (average [(absolute error)/measured concentration]) of the CCIP to be 31.5% and the bias (average [error/measured concentration]) to be -22.4%. A pooled regression of the combined CCIP and zero-order data confirmed that the covariate, height (cm), was related in linear fashion to Cl1. A striking nonlinearity of dexmedetomidine pharmacokinetics related to concentration was observed during the CCIP infusion. The final pharmacokinetic values for the entire data set were: V1 = 7.99, V2 = 13.8, V3 = 187 (L), Cl1 = (0.00791*height [cm]) -0.927, Cl2 = 2.26, and Cl3 = 1.99 (L/min). CONCLUSIONS: Pharmacokinetics of dexmedetomidine are best described by a three-compartment model. Addition of age, weight, lean body mass, and body surface area do not improve the predictive value of the model. Additional improvement in CCIP accuracy for dexmedetomidine infusions would require magnification modification of the model based on the targeted concentration.