Anesthetic drug interactions exist when one anesthetic influences the behavior of another anesthetic. The combined effects can be enhanced or diminished, or a new effect may occur. New effects can be therapeutic, or they can be unwanted side effects. During anesthesia, clinicians make therapeutic decisions routinely involving multiple drugs and as such, anesthesiologists are confronted with anesthetic drug–drug interactions on a daily basis.

Anesthetic effects are achieved using either a single drug or combinations of drugs. Drugs in combination all contribute to the overall effect; most combinations decrease the dose of each individual anesthetic when compared to doses of individual drugs required to achieve an equivalent effect. Taking advantage of interaction may decrease the severity or incidence of adverse effects without hampering desired effects.

Despite years of training, clinicians may not appreciate the magnitude of drug interactions, and they may administer excessive doses of anesthetics, hamper recovery, or increase the risk of adverse effects. Recent work has described several anesthetic interactions, and new display technology has been developed to visualize these interactions at the bedside of the patient. These recent innovations may provide a more evidence-based approach to anesthetic drug administration.

This chapter consists of 3 sections and an appendix. The first section will review basic principles of drug interactions. The second section will review the available models for vapor–opioid, hypnotic–opioid, and hypnotic–hypnotic interactions. The third section will discuss methods for applying interaction models in clinical practice. The appendix will present the methodology used to develop interaction models for interested readers.

Two topics concerning interaction will not be covered in this chapter. First, anesthetic drugs are known to influence the pharmacokinetic profile of one another by altering distribution volumes or clearance. Although many studies have demonstrated the importance of such interactions, the ultimate goal in anesthesia practice is to control effect rather than concentration. Pharmacokinetic interactions eventually result in pharmacodynamic changes. Hence, this chapter will focus on the clinical observable expression of interactions rather than on the (more obscure) changes in plasma concentration. Second, the interaction between paralytic and anesthetic agents is not well defined. This is primarily because the interactions have not been well modeled.

Additivity, Synergism, and Infra-additivity

One of the main questions concerning interactions between drugs is the nature of the interaction. Three concepts of drug interaction can be distinguished: (1) additivity, (2) supra-additivity (or synergism), and (3) infra-additivity (or antagonism). For additive drugs with equal potency, the sum of the effects evoked by respectively doses (a) and (b) for drug A and drug B is equal to the effect obtained with a solitary administration of either drug A or B given in a dose (a) + (b). For synergistic interactions, the combination of drugs A and B will result in a more pronounced effect compared to additivity. For infra-additive interactions, the combination of drugs will result in a less pronounced effect compared to additivity conditions (Figure 3–1).

In general, the mechanism behind additive interactions is thought to be activation of effect through a single pathway (ie, an identical receptor). In contrast, the mechanism behind synergistic interactions is thought to be activation of effect through multiple pathways (eg, through simultaneous N-methyl-d-aspartic acid [NMDA] and type A γ-aminobutyric acid [GABA_A] receptor inhibition). The possible advantage of synergism is that a desired effect can be targeted using lower doses of drugs with lower incidence or severity of side effects.

An important descriptor of drug interactions is the term isobole. Isobole refers to an isoeffect line on plots that compare effects between 2 drugs (as presented in Figure 3–1). Similar to the way a single drug pharmacodynamic model characterizes the concentration–effect relations, isoboles characterize concentration pairs with an equivalent effect over a range of drug concentrations for drug A and B. For a single drug, a common point of interest is the concentration associated with 50% of the maximal effect (C₅₀). For multiple interacting drugs, the 50% isobole is commonly used as the reference for potency of the drug combination. It represents the concentration pairs where there is a 50% probability of a given maximum effect. Studying one isobole of the interaction spectrum indirectly provides insight in the underlying mechanism of interactions.

For clinical practice, a wider range of probabilities of response should be available to improve dose-finding during ongoing surgery. For example, other isoboles of interest are the 5% and 95% isoboles that represent the concentration pairs where there is likely to be no effect (ie, < 5%) and there is likely to be a high probability of effect (> 95%), respectively. When dosing an anesthetic, it makes sense to administer anesthetics such that pairs of effect-site concentrations are just above, but not way beyond, the 95% isobole. Administering anesthetics that go far above the 95% isobole do not yield substantially more effect but may unnecessarily prolong the duration of effect and increase the risk of adverse side effects. Therefore, readily available advisory information—at the bedside of our patient—on the spectrum of interactions between major anesthetic drugs, may assist in optimizing the titration of multiple drugs throughout an anesthesia case. It would allow to target drug concentrations in a “just enough to do the job” approach.

Researchers have measured various anesthetic effects over a large range of anesthetic drug combinations. They used this information to build complex mathematical models of drug interactions. These models include a term that describes drug interactions (ie, extent of synergism) and a term that characterizes the transition from no effect to maximal effect. To render these complex model predictions in a form that may be used in clinical practice, researchers have developed a 3-dimensional representation of model predictions called response surfaces. Response surfaces describe the relation between combined drug concentrations (typically on the horizontal x and y axis of a graph) and the clinical effect (plotted in the vertical z axis). As an example, Figure 3–2 presents the propofol–remifentanil interaction surface for tolerance to laryngoscopy as published by Bouillon et al.¹ The response surface describes the complete range of isoboles between 0% and 100% of probability of tolerance to a stimulus with respective drug concentrations as input. As illustrated in this figure, 2 common presentations of response surfaces are used: the 3-dimensional plot described above and a simplified 2-dimensional topographical version of this plot. The topographical version presents drug concentrations on the vertical and horizontal axes with the probability of drug effect as selected isoboles (ie, 5%, 50%, and 95%).

Multiple mathematical approaches have been used to construct response surfaces for combinations of anesthetic drugs, each with specific advantages and disadvantages.² More details on the model building technology can be found in the Appendix.

When using response surfaces to optimize an anesthetic dose, it becomes apparent that it is best to titrate to target effect-site concentrations (ie, concentration pairs that provide a 95% probability of effect) rather than simply starting an infusion or turning on a vaporizer. An easy way to titrate individual effect-site concentration is to use target-controlled infusion (TCI) pumps, programmed with the pharmacokinetic models for each drug. Using TCI and response surfaces provides a means of fine-tuning anesthetic delivery.

Consider the simulations presented in Figure 3–3. It presents 2 response surfaces (in the topographical view): the probability of unresponsiveness and the probability of tolerance to laryngoscopy for combinations of remifentanil and sevoflurane.^3,4 Sevoflurane dosing requirements necessary to achieve a 95% probability of effect for 2 different remifentanil TCIs (2 and 4 ng/mL) are presented on each response surface.

A key point of this simulation is that the interaction models for tolerance to laryngoscopy and loss of responsiveness are not identical. The model of tolerance to laryngoscopy has a more pronounced synergistic interaction, as visualized by a larger bow toward the origin of the plot, and the model of loss of responsiveness has synergistic interactions but to a lesser extent. An example of this phenomenon is the difference in sevoflurane effect-site requirements for each target remifentanil infusion (4 versus 2 ng/mL). Vol% sevoflurane of 1.1 and 1.9 are required to achieve a 95% probability of tolerance to laryngoscopy for the remifentanil TCI set to 4 and 2 ng/mL, respectively (a difference of 0.8%). By contrast, vol% sevoflurane of 0.7 and 0.9 are required to achieve a 95% probability of loss of responsiveness for each remifentanil TCI (a difference of 0.2%).

A second key point is that the probability of effect for each drug alone is ineffective. For example, assuming steady-state conditions, the probability of tolerance to laryngoscopy for the sevoflurane (with no remifentanil) at 1.9 and 1.2 vol% is less than 5% and less than 1%, respectively. Similarly, the probability of loss of responsiveness for sevoflurane at 0.9 and 0.7 vol% is 77% and 48%, respectively. For remifentanil (with no sevoflurane) targeted to 2 and 4 ng/mL, the probability of either effect is zero.

Interaction Models in Anesthesia

For several intravenous anesthetic drugs (ie, propofol, remifentanil, sufentanil, alfentanil), pharmacokinetic models have been developed and used in clinical practice to titrate anesthetic delivery (eg, through TCI pumps) in a more individualized and reproducible way compared to classical dosing schemes (eg, weight-based boluses or continuous infusions). However, when combining anesthetic drugs, clinical responses may be substantially altered. As such, models developed for a single drug may have a large prediction error when used in combination with other anesthetics.

Since anesthesiologists rarely use just one drug, clinical pharmacologists have sought to characterize various responses in the presence of hypnotic–opioid and hypnotic–hypnotic drug combinations. Some of these responses include verbal, tactile stimuli, or painful stimuli; hemodynamic or respiratory effects; and changes in cerebral electrical activity during anesthesia. This line of research has led to the development of numerous interaction models for a variety of anesthetic effects. A summary of selected published interaction models of interest to anesthesiologists for various drug combinations is presented in Table 3–1.

Opioid–Inhaled Anesthetic Interactions

The interactions between potent inhaled agents and opioids have been well characterized. For analgesic effects, there is a pronounced synergistic interaction, for sedation and unresponsiveness, the interaction is synergistic, but to a lesser extent.