A Concept for Assessing Interactions of General Anesthetics

According to classic theories of anesthesia based on unitary nonspecific mechanisms of anesthetic actions, one anesthetic may be replaced freely by another, and, in the case of anesthetic combinations, the anesthetic effect of mixtures is expected to be additive ^[1]. However, even if the state of general anesthesia is narrowed to only one of its components and the type of interacting drugs is restricted to inhaled anesthetics, some deviations from additivity can be found ^[2,3]. A rapid increase in the use of intravenous (IV) drugs for general anesthesia, especially those acting via specific receptors, is the most important factor requiring revision of the above-indicated conceptual approach to anesthetic interactions.

The development of a concept to assess anesthetic interactions is inevitably centered around two major points: 1) identification of the goals of general anesthesia, along with the spectrum of pharmacologic actions through which to achieve them, and 2) determination of the mechanisms of these actions. Although these two aspects of general anesthesia are closely related, they are often discussed separately. The aim of the present article was to combine these aspects to better understand the general principles of anesthetic interactions. To achieve this goal, the article evaluates the growing consensus on the concept, which began its development half a century ago ^[4].

State of General Anesthesia and Its Components

A month after the demonstration of surgical anesthesia by William Morton in 1846, Oliver Holmes suggested the name for it: "The state should, I think, be called anaesthesia" ^[5]. The definition of anesthesia was essentially based on one of the effects of diethyl ether-unconsciousness that provides insensibility in general, and to pain in particular. It was soon noted that in addition to preventing pain, diethyl ether had several other effects that might be useful. Little by little, the other useful effects of diethyl ether were also included in the notion of general anesthesia. The effects listed in Table 1 (left side) are commonly provided in anesthesia, all of which are typical of diethyl ether. As a result, general anesthesia was initially accepted as a condition caused by diethyl ether.

Table 1: State of General Anesthesia: Spectrum of Effects

According to the classic concept of general anesthesia, the state of general anesthesia is a multifeatured phenomenon with one underlying mechanism, and, to be common for all general anesthetics with diverse chemical structures, this mechanism is supposed to be nonspecific. The problem with the classic concept of the state of anesthesia became obvious when neuromuscular blocking agents, opioids, and barbiturates began to be widely used in combination with inhaled anesthetics. One of the first anesthesiologists to identify the problem was Woodbridge ^[4]. In 1957, he indicated that with the use of drugs "having more limited or specific action" for anesthesia, the anesthesiologist must "analyze anew what we have been calling anesthesia." He divided anesthesia into four components: sensory blockade; motor blockade; blockade of undesirable reflexes of the respiratory, cardiovascular, or gastrointestinal systems not usually involving the operative field directly; and mental blockade, sleep, or unconsciousness. Different drugs could be used to achieve each component. Woodbridge believed that the initial definition of anesthesia "properly refers to sensory block only;" therefore, he suggested another term ("nothria") to define the state that reflects all four components of anesthesia.

(Table 2) compares different views of the state of general anesthesia ^[4,6-9]. The various authors listed either one, two, or more effects in the state of anesthesia. Prys-Roberts ^[6] indicated that the state of general anesthesia can be defined as "drug-induced unconsciousness, the patient neither perceives nor recalls noxious stimulation." He wrote, "All other attributes of drugs which produce the state of anesthesia can be classed as alternative pharmacological properties of the drugs and not as components of the state of anesthesia." According the extreme view, all separate effects used to protect the patient from the trauma of surgery and induced by one or several drugs should be termed components of anesthesia ^[9]. The authors "do not see any need to reconcile the contradiction between the initial definition of anesthesia with its new meaning by restricting this meaning: all components of anesthesia should be included in the definition of the state of anesthesia. The requirements for different surgeries or patients can be different, dictating appropriate combinations of components of anesthesia." Thus, anesthetic drugs produce effects that according to the various points of view, can be classified as components of anesthesia or just desirable supplements to anesthesia (Table 1).

Table 2: Definitions of State of Anesthesia

(Figure 1) compares the hypnotic effect (loss of the righting reflex), blockade of somatic motor responses, and suppression of autonomic responses to noxious stimulation (which might be regarded as basic goals or components of anesthesia) produced in rats with the use of drugs representing different classes of drugs. The comparison shows that each of the representative drugs has its own "fingerprint" of the relationships among the end points of anesthesia. The depth of anesthesia can be treated as a passage through ordered stages or planes-a kind of ordered dose-response curve. With such a curve in mind, the comparison of isoflurane with other drugs presented in Figure 1 shows that IV anesthetics are quite different from isoflurane and from one another and that the passage through different end points of anesthesia is far from being similar. For example, thiopental can block the purposeful movement response to noxious stimulation in doses that are far different from those that block the cardiac acceleration response. At the same time, these two end points can be reached with diazepam at dose levels similar to each other and close to lethal doses. Morphine has the "reversed" ranked order of effects, with loss of the righting reflex at larger doses than blockade of the responses to noxious stimulation. This relationship between components of morphine anesthesia obtained in rat experiments confirms the statement by Hug ^[10] that some patients with opioid anesthesia may be completely awake and aware of intraoperative events when there is absolutely no change in hemodynamics or any manifestation of increased sympathetic activity.

Figure 1: Median effective doses of thiopental, diazepam, isoflurane, and morphine for different endpoints of anesthesia in rats. HE = ED₅₀ for hypnotic effect (loss of the righting reflex), PM = ED₅₀ for blockade of purposeful movement response to noxious stimulation, CA = ED₅₀ for suppression of cardiac acceleration response to noxious stimulation. LE = ED₅₀ for lethal effect (LD₅₀). Horizontal lines = 95% confidence limits. From Kissin and Gelman ^[9].

Usually, the term general anesthetic identifies a drug that can fully provide all components of anesthesia when used alone. With the concept of components of anesthesia, the requirements to classify a drug as an anesthetic become less restrictive. For example, midazolam, which cannot block somatic and cardiovascular responses to noxious stimulation, is regarded as an anesthetic drug.

Mechanism of General Anesthesia

Different Drugs Act Via Different Mechanisms

It is possible to provide general anesthesia with a combination of drugs acting via specific receptors. For example, complete anesthesia can result from the combined use of an opioid, a benzodiazepine, and a neuromuscular blocking drug. All of these drugs act on specific receptors, and their anesthetic effects can be reversed by the administration of specific antagonists. This fact alone makes the unitary hypothesis of anesthetic action unconvincing. Several attempts have been made to defend the unitary hypothesis by introducing some modifications and by limiting its scope to inhaled anesthetics. For example, the anesthetics acting via the same mechanism can affect more than one molecular site, each of which may have different physical properties-the multisite hypothesis proposed by Halsey et al. ^[11]. The multisite hypothesis, however, cannot explain much of the evidence against the unitary hypothesis, and Richards ^[12] postulated that different classes of drugs have different mechanisms of anesthetic action.

The fact that general anesthesia is provided by many structurally diverse molecules was usually interpreted as proof that all general anesthetics act via a common nonspecific mechanism. However, the extreme diversity of the cellular and physiologic mechanisms involved in the anesthetic effect suggest another explanation for this fact: multiple potential neuronal mechanisms that could lead to a common final outcome are likely targets for drugs of dissimilar structure. According to the suggestion by Wall ^[13], "the apparent similarity of the effect of various anesthetics may not be a reflection of identical modes of action of all anesthetics, but rather a reflection of the varying stability of different synapses."

Several possibilities for the molecular mechanism of action of general anesthetics are summarized in Table 3. As shown in Table 3, at the molecular level, each anesthetic can act via one of the following mechanisms: one specific mechanism, a combination of different specific mechanisms, or one nonspecific mechanism. Combinations of some of these mechanisms are also possible (e.g., a combination of specific and non-specific mechanisms). For a review of molecular mechanisms of anesthesia, see ^[14-16].

Table 3: Possible Molecular Mechanisms Providing Multifaceted Feature of Anesthetic Action

As far as the cellular basis of anesthesia is concerned, multiple possible mechanisms have been suggested; these were divided into two major categories: mechanisms involving actions on synaptic transmission and on postsynaptic excitability (for review, see ^[17-19]). Although the central nervous system synapses have been regarded as the most probable cellular site of anesthetic action ^[17], other sites and mechanisms are also involved. The anesthetic state may be the result of interactions of multiple different mechanisms that, in combination, provide characteristic behavioral effects.

Different Mechanisms for Different Components of Anesthesia

It was postulated that if different components of anesthesia have the same underlying mechanism of action, general anesthetics should achieve these components in the same proportion. For example, concentrations of different inhaled drugs that are equivalent in terms of preventing movement in response to surgical incision are equipotent in achieving unconsciousness ^[17]. Experiments on animals could not confirm this assumption. Using various inhaled anesthetics in toads, mice, and rats, at least three groups of authors reported that ratios of the 50% effective dose (ED (₅₀)) values for blockade of noxious stimulation-induced movement to the ED₅₀ values for loss of the righting reflex were different with different anesthetics ^[20-22]. The initial determination of similar ratios in humans (the ratio of a concentration of an anesthetic that blocks opening of the eyes on command in 50% of patients [MAC-awake], to a concentration that blocks movement to surgical incision [MAC]) could not demonstrate a statistically significant difference among them; however, a tendency for a difference between the halothane and the diethyl ether ratios has been observed ^[23]. Recent clinical studies have reported a difference in MAC-awake/MAC ratios for several anesthetics. In one study ^[24], the MAC-awake/MAC ratio for isoflurane was significantly lower than that for nitrous oxide. The results of another study ^[25] demonstrated that the MAC-awake/MAC ratio for isoflurane (and enflurane) is significantly less than that for halothane. The initial problem of the discrepancy between results obtained in patients and in animal experiments was probably caused because small differences among various components of anesthesia induced by volatile anesthetics are much easier to detect under laboratory conditions, which permit a greater degree of precision.

It is commonly believed that parallel dose-effect curves are indicative of the identity of the mechanism of action. However, the difference in the slopes of dose-effect curves is a less sensitive index for testing the identity of the mechanism of action than agent-to-agent variability in the potency ratios. With inhaled anesthetics, the difference between slopes of the dose-effect curves for different components of anesthesia was reported only for diethyl ether. This difference was determined in rats between the slopes of the dose-effect curves for blockade of the righting reflex and for blockade of movement to noxious stimulation ^[22]. With IV anesthetics, differences between the slopes of the dose-effect curves of an anesthetic for different components of anesthesia are much more pronounced ^[26].

In 1974, Halsey ^[17] summarized data indicating the importance of the action of general anesthetics not only on the brain but also on the spinal cord. He wrote that in animals with a transected cervical spinal cord, the depressant effects of halothane and nitrous oxide on dorsal horn cells equal or exceed the depressant effect found in animals having intact midbrain and hindbrain cord connections. Recent studies with isoflurane suggest that this drug provides one of the basic components of anesthesia, blockade of movement response to noxious stimulation, by acting primarily on the spinal cord ^[27,28]; this occurs at a time when unconsciousness is achieved by the action of isoflurane on the brain. The difference in the anatomic substrate suggests that different physiologic mechanisms are involved in these two actions of isoflurane.

(Table 4) reflects changing concepts of the state of general anesthesia, from the initial view representing one effect of diethyl ether to the concept of components of anesthesia resulting from separate pharmacologic actions, even if the anesthesia is produced by one drug ^[29].

Table 4: State of General Anesthesia: Changing Concepts

From Mechanism of General Anesthesia to Anesthetic Interactions

Differences in Mechanisms of Actions of General Anesthetics Provide Basis for Supra-And Infraadditive Interactions

Although the principle of additivity in general anesthetic interactions was confirmed in many studies with inhaled anesthetics, it was constantly challenged. When interactions of inhaled anesthetics regarding motor response to noxious stimuli were studied in experimental animals, some deviations from additivity were found, primarily when nitrous oxide was involved ^[2,3]. Small deviations from additivity were also found with the hypnotic component of anesthesia in mice: sulfur hexafluoride-nitrous oxide, argonnitrous oxide combinations (additive), and sulfur hexafluoride-argon combination (infraadditive) ^[30]. It is interesting that the deviations from additivity with combinations of inhaled anesthetics were always toward some degree of antagonism. Findings of the antagonistic interactions among IV anesthetics are usually related to the antinociceptive components of anesthesia. For example, an infraadditive antinociceptive interaction between midazolam and fentanyl was observed in enflurane-anesthetized dogs when the interaction was determined in terms of enflurane MAC reduction ^[31].

Supraadditive interactions were reported only with combinations of IV anesthetics. Synergism was reported in studies on rats (with loss of the righting reflex as an index of hypnosis) for the following combinations: alphaxalone-etomidate ^[32], alphaxalone-methohexital ^[32], and midazolam-thiopental or pentobarbital ^[33]. Synergism was also found in studies of patients with loss of response to voice command as an index of unconsciousness: midazolam-thiopental, midazolam-methohexital, alfentanil-midazolam, fentanyl-midazolam, and propofol-midazolam ^[34]. The study of a propofol-midazolam-alfentanil combination in patients did not support the hypothesis that the triple combination might be even more synergistic than the binary combinations (midazolam-alfentanil, propofol-midazolam, or alfentanil-propofol) due to summation of the binary synergisms; the triple combination produced a profound hypnotic synergism that is not significantly different from that of the binary midazolam-alfentanil combination ^[34]. Hypnotic interactions of IV drugs with different mechanisms of actions should not necessarily be nonadditive; the best examples are ketamine-thiopental ^[35] and ketamine-midazolam ^[36] interactions, which were found to be additive.

The role of the physiologic mechanisms of actions in anesthetic interactions is especially evident in combinations in which an opioid is present. Opioids act as antinociceptive agents, in part by activating the descending inhibitory control systems within the central nervous system ^[37]. By inhibiting the descending inhibitory control system, anesthetics may (in principle) eliminate the supraspinal component of the antinociceptive effect of an opioid and, in addition, the synergism associated with it. For example, a profound morphine-halothane antagonism in relation to suppression of the heart rate increase to noxious stimulation found in rat experiments ^[38] may be the consequence of such interactions. In these experiments, isobolographic analysis used to characterize the interaction indicated that halothane antagonized morphine to an incomparably greater extent than morphine antagonized halothane.

Thus, there are many examples of nonadditive interactions among anesthetics that stem from differences in the nature of their anesthetic actions.

Components of Anesthesia as Grounds for Differences in Anesthetic Interactions

If components of anesthesia represent separate actions with different underlying mechanisms (even if the anesthesia is produced by one drug), a drug combination may result in different interaction outcomes for different components of anesthesia ^[29]. The most striking illustration for such a possibility is that reserpine antagonizes the inhibitory effect of fentanyl on the purposeful movement response to noxious stimulation and, at the same time, strengthens its hypnotic effect (loss of the righting reflex) ^[34]. Another example, an isobolographic analysis of the morphine-diazepam interactions for the hypnotic effect in rats demonstrated a profound synergism; at the same time, the morphine-diazepam interaction characterized by abolition of purposeful movement to noxious stimulation resulted in an antagonism ^[34].

As far as an antinociceptive effect is concerned, results similar to those reported in rats were not observed in clinical studies. Although barbiturates in subanesthetic doses are known to produce an antianalgesic effect in humans, this effect is restricted to small doses of barbiturates. When the effects of alfentanil on the hypnotic (verbal command) and antinociceptive (trapezius muscle squeeze) components of thiopental anesthesia were compared in humans, enhancement of the antinociceptive effect of thiopental by alfentanil was even greater than that of the hypnotic effect ^[34]. These results suggest that antagonistic opioid-barbiturate (or opioid-benzodiazepine) antinociceptive interactions might be typical only for rats; they should be regarded only as an indication of the possibility for opposite interactions (synergism versus antagonism) for different components of anesthesia in principle.

Even such close components of anesthesia as inhibition of somatic motor responses and hemodynamic responses to noxious stimulation may not be affected by anesthetic combinations proportionally. For example, in some of the patients anesthetized with a lorazepam-alfentanil combination, somatic responses to noxious stimulation were present at a time when no hemodynamic responses were observed ^[39].

Different mechanisms of action of an anesthetic for various components of anesthesia can find their expression in the different degrees of efficacy, which would influence the outcome of anesthetic interactions for these components. For example, low opioid efficacy for blockade of responses to phasic noxious stimuli can determine opioid interactions with other drugs when intensity of noxious stimulation is variable. Thus, the outcome of the morphine-pentobarbital interaction for suppression of behavioral awakening (recovery of the righting reflex in rats) caused by noxious stimulation of various strength was reported to be supraadditive or infraadditive depending on the strength of the stimulation ^[34].

Another illustration of the role of low antinociceptive efficacy of opioids in anesthetic interactions is the so-called ceiling effect. In a dog model in which an opioid was tested as a substitute for an inhaled anesthetic by determination of the MAC value (minimum alveolar concentration of an inhaled anesthetic required to prevent 50% of subjects from responding to strong noxious stimulation with gross purposeful movement), it was found that neither morphine, fentanyl, nor sufentanil can produce a reduction in MAC of enflurane or isoflurane by more than approximately two-thirds ^[40]. Similar results were reported with the fentanyl-desflurane ^[41] and sufentanil-isoflurane ^[42] interactions in humans. A ceiling effect with a fentanyl-propofol or alfentanil-propofol interaction was also described ^[43,44]. The presence of the ceiling effect indicates that the extent of an anesthetic interaction with regard to one of the components of anesthesia is limited, and if the dose of one of the interacting drugs exceeds a certain level, the desired effect will not increase; however, the other effects, which are often detrimental, may continue to increase without such limits.

Thus, a combination of anesthetics can interact differently regarding different components of anesthesia; therefore, some of the combinations that seem advantageous according to the outcome for one component of anesthesia may not be beneficial for another component.

Conclusion

When different components of anesthesia are balanced by the combined use of drugs with different mechanisms of action, there is a ground for the wide variability in the anesthetic interaction outcomes (supraadditive, additive, infraadditive) in general. In addition, the same combination of anesthetic drugs could produce different interaction outcomes for different components of anesthesia.

Understanding general anesthesia as a combination of different components achieved through the combined use of drugs with different underlying mechanisms of action has important clinical implications for measuring the depth of anesthesia: different components of anesthesia should be determined specifically for various anesthetic combinations ^[45]. With the use of anesthetic combinations, the search for a reliable index of anesthetic depth is transformed into a search for separate indices of different components of anesthesia.