Arterial and venous pharmacokinetics of morphine-6-glucuronide and impact of sample site on pharmacodynamic parameter estimates

Abstract

Background: In pharmacokinetic-pharmacodynamic modeling studies, venous plasma samples are sometimes used to derive pharmacodynamic model parameters. In the current study the extent of arteriovenous concentration differences of morphine-6-glucuronide (M6G) was quantified. We used simulation studies to estimate possible biases in pharmacodynamic model parameters when linking venous versus arterial concentrations to effect.

Methods: Seventeen healthy volunteers received an IV 90-second infusion of 0.3 mg/kg morphine-6-glucuronide (M6G). Arterial and venous blood samples, from the radial artery and cubital vein, respectively, were obtained. An extended pharmacokinetic model was constructed linking arterial and venous compartments. The extent of bias in pharmacodynamic model parameter estimates was explored in simulation studies with NONMEM, simulating M6G effect using first-order effect-compartment-inhibitory sigmoid E(MAX) models. M6G effect was simulated at various values for the arterial blood-effect-site equilibration half-lifes (t(1/2)k(E0)), ranging from 5 to 240 minutes.

Results: Arteriovenous concentration differences were apparent, with higher arterial plasma concentrations just after infusion, whereas at later times (>60 minutes) venous M6G concentrations exceeded arterial concentrations. The extended pharmacokinetic model adequately described the data and consisted of 3 arterial compartments, 1 central venous compartment, and 1 peripheral venous compartment. The simulation studies revealed large biases in model parameters derived from venous concentration data. The biases were dependent on the value of t(1/2)k(E0). Assuming that the true values of M6G t(1/2)k(E0) range from 120 to 240 minutes (depending on the end point measured), we would have underestimated t(1/2)k(E0) by 30%, whereas the potency parameter would have been overestimated by about 40%, when using venous plasma samples.

Conclusions: Because of large arteriovenous differences in M6G plasma, concentration biases in pharmacodynamic model parameters will occur when linking venous concentration to effect, using a traditional effect-compartment model.