Within- and between-subject variations in pharmacokinetic parameters of ethanol by analysis of breath, venous blood and urine

Abstract

Aims: To evaluate the prerequisites for using ethanol dilution to estimate total body water, we studied the within- and between-subject variation in the parameter estimates of a two-compartment model for ethanol pharmacokinetics with parallel Michaelis-Menten and first-order renal elimination. Because sampling of breath might be preferable in some clinical situations the parameter estimates derived from breath and venous blood were compared.

Methods: On two occasions, ethanol 0.4 g kg-1 was given by intravenous infusion to 16 volunteers after they had fasted overnight. The proposed model was fitted by means of nonlinear regression to concentration-time data measured in the breath, venous blood and urine during 360 min. The model contained six parameters: Vmax and Km (Michaelis-Menten elimination constants), CLd (intercompartmental distribution parameter), VC and VT (volumes of the central and tissue compartment, respectively) and CLR (renal clearance). The volume of distribution, Vss, was calculated as the sum of VC and VT.

Results: The mean +/- total s.d. of the parameter estimates derived from blood data were Vmax 95 +/- 25 mg min-1, Km 27 +/- 19 mg l-1, CLd 809 +/- 232 ml min-1, VC 14.5 +/- 4.3 l, VT 21. 2 +/- 4.4 l, CLR 3.6 +/- 2.0 ml min-1 and Vss 35.8 +/- 4.3 l. The variation within subjects amounted to 3%, 9%, 21%, 21%, 17%, 26% and 2%, respectively, of the total variation. Breath samples were associated with a similar or lower variation than blood, both within and between subjects. About 1.5% of the infused ethanol was recovered in the urine.

Conclusions: The low within-subject variation of the key parameter Vss (only 2%) suggests that ethanol dilution analysed by the pharmacokinetic model applied here may be used as an index of the total body water. Breath samples yielded at least as good reproducibility in the model parameters as venous blood.

Figure 1

Two-compartment model for ethanol disposition with parallel Michaelis-Menten and first order renal elimination. C is the concentration in the central compartment as measured in the breath or venous blood and C_T is the concentration in the tissue compartment. A_e is the measured cumulative amount of unchanged ethanol excreted to the urine. V_max is the maximum metabolic rate, K_m the concentration that gives 50% of that rate, CL_d the intercompartmental clearance, V_C and V_T the volumes of the two compartments and CL_R the renal clearance.

Figure 3

Individual concentration-time profiles of ethanol in expired breath (BrAC) and venous blood (VBAC) after a 30 min intravenous infusion of 0.4 g ethanol kg⁻¹ body weight, n = 32.

Figure 2

Measured alcohol concentration in expired air (○), venous blood (•) and cumulative amount excreted by the urine (▪) and the predicted concentrations in the central compartment (C) and peripheral compartment (C_T) and the predicted cumulative renal excretion (A_e). A two-compartment model with parallel Michaelis-Menten and first order renal elimination was fitted to data from one representative experiment.

Figure 4

Partial derivatives, indicating the relative contribution of various time segments to the parameter estimates, based on a simulation using the mean parameter estimates from BrAC data n = 32. The six model parameters are V_max (□), K_m (○), CL_d (▾), V_C (▪), V_T (▵) and CL_R (•).

Figure 5

Within-subject variability for three of the model parameters based on analysis of BrAC (○) and VBAC (•). The x-axis is the mean value for day 1 and day 3 and the y-axis is the difference between them. The mean difference and limits of agreement (± 2.16 s.d., n = 14) are shown as solid lines, and line of identity as a broken line.

Figure 6

Concentration-time profiles of ethanol for two subjects after 0.4 g ethanol kg⁻¹ b.w. was given intravenously over 30 min. In both subjects the modelling failed when VBAC data were used (•) while the profile was normal when BrAC data were studied (○).