What Determines Anesthetic Induction Dose? It’s the Front-End Kinetics, Doctor!

Since IV anesthetics were first used to induce general anesthesia, choosing the appropriate dose has been a combination of the art and science of the specialty. The dramatic evidence of the potential consequences of this dosing challenge is perhaps best illustrated by the tragic outcomes of the administration of a “standard dose” of thiopental to the hypovolemic victims of the Japanese attack on Pearl Harbor (¹). Until recently, little has been accomplished to improve the scientific basis of dosage selection.

Fisher (²), in his editorial, “(Almost) everything you learned about pharmacokinetics was (somewhat) wrong!” presented some of the problems associated with the application of conventional pharmacokinetic compartmental modeling to situations in which the plasma concentrations of the drug are changing rapidly. In this editorial, we shall examine various approaches to modeling early drug distribution (i.e., front-end kinetics), which determines the plasma drug concentration versus time relationship seen by the sites of action of a rapidly administered IV anesthetic. The area under the drug concentration versus time curve (AUC) is an index of the exposure of the sites of drug effect for a time interval during which plasma concentrations are above the minimally effective threshold for a particular drug.

Studies of the distribution and elimination of IV administered drugs used in the practice of anesthesia have mainly used multicompartmental pharmacokinetic models, which are mathematical constructs based on the characterization of the multiexponential drug concentration versus time relationships (Fig. 1) (³). For example, in a three-compartment pharmacokinetic model, the central or initial volume of distribution (V_C) is that volume in which a drug appears to mix instantaneously after administration but before it distributes throughout the remaining volume. From V_C, drug is distributed to the rapidly (fast) and slowly equilibrating volumes of distribution (V_F and V_S, respectively) by a process called intercompartmental clearance. These clearances are volume-independent estimates of drug transfer that are determined by blood flow and transcapillary permeability. The volume of distribution at steady state is the total volume of distribution and, as such, is the sum of V_C, V_F, and V_S. Elimination clearance quantifies the irreversible removal of drug from the body or its metabolism to another form. These parameters have provided the basis for drug dosing in patients, particularly when drugs are administered as multiple doses or by continuous infusion (⁴). They have also led to the development of such useful concepts as context-sensitive half-time, a function of the back-end kinetics (⁵). However, they have been of little use in understanding the pharmacokinetic basis of interindividual variability in response to induction doses or rapidly administered loading doses of these drugs. Variability in response to IV anesthetics is not caused by differences in total volume of distribution, elimination clearance, or elimination half-life, because these minimally affect the plasma drug concentration versus time relationship during the time these drugs exert their maximal effect and while their effect is being terminated. Differences in dose requirements must be caused by altered pharmacodynamics or changes in early drug distribution, because these drugs have rapid onsets of effect.

Figure 1: Three-compartment mammillary pharmacokinetic model characterizing the multiexponential plasma drug concentration versus time relationship: C_p.(t) = Pe^–πt + Ae^–αt + Be^–βt. V_C = central volume of distribution, V_F = fast equilibrating volume of distribution, and Cl_F = fast intercompartmental clearance, V_S = slow equilibrating volumes of distribution, Cl_S = slow intercompartmental clearance, Cl_E = elimination clearance.

The volume estimate of V_C is related to the time of the first blood sample collection after drug administration because the drug is distributed more extensively with the passage of time (⁶). Similarly, the sampling site (e.g., arterial versus venous) and the physiologic state of the subject (e.g., variations in cardiac output) will affect conventional estimates of V_C. A multicompartmental pharmacokinetic model often fails to provide a useful estimate of the V_C of IV anesthetics, because it is based on infrequently collected blood samples beginning after the peak plasma concentrations of the drug have begun to wane. In addition, the traditional model fails to account for the processes responsible for drug distribution and variability in the dose-response relationship: mixing within the vascular volume, flow (both cardiac output and its peripheral distribution), and the diffusion of the drug into both active and indifferent tissues (⁷).

Physiologically based pharmacokinetic models describe measured blood and tissue drug concentration histories by apportioning cardiac output, hence, drug distribution, among tissues or tissue groups with similar perfusion and drug solubility characteristics (³). These models either assume that all tissues are in equilibrium with the blood leaving it (⁸) or invoke the concept of diffusion barriers to flow-limited drug uptake (⁹). Physiologic models have been used to simulate the effect of changes in cardiac output on drug disposition by adjusting regional blood flows in direct proportion to changes in cardiac output (¹⁰) or by setting them arbitrarily (¹¹).

Physiologic models have provided valuable insights into drug disposition. One of the earliest of these models was developed by Price et al. (^8,11) and was used to demonstrate the importance of thiopental redistribution to the termination of its effect. Price (¹¹) also used his model to simulate the fraction of injected dose in the brain of patients in whom the cardiac output and the blood flow to indifferent tissues, such as muscle and portal tissues, are either decreased or increased. These simulations explained the reduced dose requirements of patients in hemorrhagic shock, because, in this condition, the fraction of the dose received by the brain is very large, and its rate of removal is very slow because of decreased blood flow to other tissues. Price (¹¹) also predicted that patients with increased blood flow to indifferent tissues (e.g., apprehensive patients) would require larger doses of thiopental because a smaller fraction of the IV administered drug would appear in the brain.

Wada et al. (¹²) developed a physiologic pharmacokinetic model of thiopental disposition in humans on the basis of a scale up of a model developed in rats. They used their model to simulate arterial plasma thiopental concentrations during and after a simulated infusion of 250 mg of the drug over 1 min to “patients” of different age, gender, and body habitus and to “patients” who differed only in their cardiac output. They predicted slightly higher peak concentrations in women and the elderly and a nearly 50% higher peak concentration in patients with a 50% decrease in cardiac output. Patients who are 50% or 100% overweight or have a 50% increase in cardiac output were predicted to have about 25% lower peak concentrations. This simulation was consistent with the observation that 90% of patient variability in thiopental dose requirements needed to reach electroencephalographic burst suppression could be explained by a multiple linear regression model that included age, lean body mass (or body weight), and cardiac output (¹³).

Upton et al. (¹⁴) looked closely at the relationship between cardiac output and early plasma propofol concentrations in the this issue of the Journal. In an earlier study, Upton and Ludbrook (¹⁵) developed a physiologic model of propofol disposition in sheep. A subsequent analysis of the structure and sensitivity of their model revealed that cardiac output was one of the most important variables determining early propofol concentration histories (¹⁶), which led to the present study in which this hypothesis was tested in several paradigms of altered cardiac output.

In their article, Upton et al. (¹⁴) report the results obtained from two separate groups of studies. In the first group, six sheep were anesthetized with 2%–3% halothane and their mechanical ventilation was controlled to produce different levels of end-tidal CO₂ (nominally 25, 40, and more than 70 mm Hg) resulting in different levels of cardiac output (six-animal means of 3.5, 5.2, and 6.4 L/min, respectively). A second group of five awake sheep were treated as controls or with the sympathomimetic vasoconstrictor, metaraminol (group mean cardiac outputs of 7.3 and 5.1 L/min, respectively). Propofol (100 mg) was infused IV for 2 min, in each animal, under each condition, and the resultant concentrations were determined in frequently obtained pulmonary and carotid arterial blood samples. These values were used to determine the peak concentration and AUC at the various cardiac outputs. As illustrated in the physiologic pharmacokinetic simulation of their experiment, peak concentrations and the AUC are determined by both first-pass and recirculating concentrations (Upton et al., Figure 1) (¹⁴).

The results of their investigation support the hypothesis that cardiac output is a determinant of the initial concentrations of propofol during and shortly after a 2-min infusion. Because positive correlations are more appealing visually, the authors have elected to represent their data as the inverse of the peak concentration or the inverse of the AUC as determined for the pulmonary or arterial concentrations. Over nearly a 4-fold range of cardiac outputs, cardiac output accounted for nearly two thirds of the variability in peak concentration or AUC in the halothane-CO₂ studies. Likewise, the 2.5-fold variability in cardiac output accounted for about half of the arterial concentration and AUC variability in the awake studies, with or without metaraminol. The obvious question relating to these results and their predictive utility is: what accounts for the remainder of the variability in peak concentration or AUC?

One of the most interesting observations of the study of Upton et al. (¹⁴) is the difference in AUC_{0–4 min} between awake and anesthetized animals. Awake animals had significantly lower AUC_{0–4 min} (i.e., the higher 1/AUC_{0–4 min} in their Figure 3, upper right panel, closed circles) than animals anesthetized with 2%–3% halothane (i.e., the lower 1/AUC_{0–4 min} in their Figure 3, upper right panel, open circles). How can this be explained? The most obvious difference between the two groups is the use of halothane anesthesia in the high AUC_{0–4 min} (i.e., the lower 1/AUC_{0–4 min}) group. The authors imply that this difference is attributable to inhibition of pulmonary propofol metabolism by halothane. However, given the apparent lack of difference in peak concentrations between the two groups (their Figure 3, lower right panel) an alternate explanation can be proffered. Studies of regional blood flow during halothane anesthesia demonstrated a very large increase in the shunting of 15-μm radiolabeled microspheres compared with that observed in awake control animals (¹⁷). Preportal and muscle blood flow decreased in these studies, whereas renal blood flow was preserved and cerebral blood flow nearly doubled. Thus, halothane anesthesia not only decreases cardiac output but also affects its peripheral distribution. The differences in propofol AUC_{0–4 min} in awake and anesthetized animals may therefore be caused by halothane-induced alterations of both cardiac output and its peripheral distribution, as suggested by recirculatory multicompartmental analysis of drug disposition in awake and halothane-anesthetized animals.

A recirculatory multicompartmental model of the disposition of physiologic markers based on frequent early arterial blood sampling (Fig. 2) is able to describe drug disposition from the moment of injection (¹⁸). The recirculatory model retains the best aspects of the traditional multicompartmental model and the physiologically based model besides offering significant advantages over both. Both the traditional multicompartmental model and the recirculatory model describe data from individuals (including humans) collected under a variety of conditions, whereas the physiologic model requires tissue drug concentrations and direct measurements of organ blood flow, which are often unavailable in humans. Both the physiologic model and the recirculatory model incorporate physiologic factors, such as cardiac output and its distribution, in a description of drug disposition. The recirculatory multicompartmental model estimates blood flow to tissue compartments based on the calculated intercompartmental clearance of a flow-limited tissue distribution marker (i.e., intercompartmental clearance equals tissue blood flow in the absence of diffusion barriers). However, because of arteriovenous anastomoses or significant diffusion barriers, a fraction of cardiac output returns blood to the central circulation after minimal drug loss caused by tissue distribution. The recirculatory model uses the blood drug concentrations of the recirculation peak to describe this nondistributive blood flow, or clearance, which can be thought of as a pharmacokinetic shunt. This model has been used to demonstrate that halothane-induced changes in cardiac output and its distribution alters the balance of distributive and nondistributive blood flows to various tissues that are not proportional to changes in cardiac output (¹⁹). Because nondistributive blood flow quickly returns the lipophilic marker to the central circulation, the increase in nondistributive blood flow by halothane increases the arterial blood AUC during the early minutes after drug administration. This could explain the difference in the AUC_{0–4 min} for a given cardiac output between awake and anesthetized animals in the report of Upton et al. (¹⁴). Alterations in nondistributive blood flow within the awake or halothane-anesthetized groups caused by the vasoactive effects of metaraminol or of differing arterial CO₂ concentrations, respectively, may also account for the variability in propofol peak concentrations and AUC not accounted for by cardiac output alone.

Figure 2: Multicompartment recirculatory pharmacokinetic model. V_T-F. and V_T-S. are the fast and slow compartments of a three-compartment model (Fig. 1) whereas the tanks-in-series compartments (delay elements) included within the dotted circle are the expanded components of its V_C (¹⁸).

The front-end kinetics of IV anesthesia induction agents, the importance of which has only recently been appreciated, have implications for the practicing anesthesiologist. Early drug distribution kinetics determine the rate and extent of both drug distribution to the brain and its dilution by distribution to indifferent tissues. The report of Upton et al. (¹⁴) in the present issue of this Journal provides further evidence of the importance of cardiac output in determining the early concentration history of drugs after rapid IV administration, whether described as peak concentrations or initial AUCs. Their results may also confirm that not only cardiac output, but also its peripheral distribution, are important determinants of the early drug concentration versus time relationship of IV administered drugs. The relative impact of each may vary, depending on the characteristics of the drug being studied and the physiologic circumstances of the subject.