Temperature Dependence of the Potency of Volatile General Anesthetics: Implications for In Vitro Experiments

THERE is an important practical dilemma facing those who perform in vitro laboratory investigations into the actions of volatile general anesthetics on mammalian preparations. This is the choice of the appropriate general anesthetic concentrations to use when the experimental preparation is at lower than normal body temperature. Different ways of interpreting anesthetic potencies in hypothermic animals can lead to very different ideas as to what constitutes "clinically relevant" concentrations at anything other than normothermia. In this article, we will discuss these different interpretations and offer some practical guidelines we hope will prove useful.

Volatile anesthetics are drugs that are (almost uniquely) administered clinically in the gas phase. Whereas these agents are, of course, applied to mammals at their normal body temperatures (usually about 37 degrees C), for various technical reasons, experiments on simple molecular systems such as ion channels are often carried out at room temperature (20–25 degrees C). Minimum alveolar concentration (MAC) invariably is determined using gas-phase concentrations (usually expressed as vol%=% atm). For example, mammalian MAC values for halothane are, on average, [1]about 0.9% atm. At first sight, it might seem reasonable to apply this same gas-phase concentration to an in vitro preparation at room temperature and assume that the effects observed are equivalent to those in the animal exposed to the same gaseous concentration. Is this procedure reasonable, or should some other method be used?

To answer such questions, it is necessary to consider quantitative animal data on how anesthetic requirements change with temperature. It is well known that, for a variety of volatile anesthetics, MAC (when expressed in % atm) can decrease markedly with decreasing temperature, [2–8]although the size of the effect varies from agent to agent. Because by far the most complete set of data is available for halothane acting on dogs, [2,3,5]we will use these results to illustrate some general principles that are characteristic of most volatile agents. These data (Figure 1, upper line) show that there is a roughly linear decrease in the halothane requirement as the temperature is decreased.

However, it could be argued that MAC values at anything other than normal body temperature cannot be interpreted easily because of the confounding variable of the physiologic effects of temperature. Obviously, mammals have complex mechanisms for maintaining normal body temperatures, and it is not unreasonable to suppose that a change in body temperature per se might influence anesthetic requirement. What is more, the decrease in the anesthetic requirement when linearly extrapolated predicts that one should require no anesthetic at all at a low enough temperature. [2]Remarkably, this does indeed seem to be the case. For example, thoracotomy on dogs has been found [9]to require no anesthetic at body temperatures between 15 degrees C and 22 degrees C, whereas in goats, [8]the MAC for isoflurane reduces to zero at 20 degrees Celsius. That a linear extrapolation of anesthetic concentration predicts this experimental finding seems to be, at first sight, a persuasive reason for believing that the effect of temperature per se is a progressive effect that is altering the anesthetic requirement at all temperatures other than the normal body temperature. We believe this logic is incorrect, for the following two reasons.

First, it has to be appreciated that, while the determination of MAC as a partial pressure is an excellent practical measure of anesthetic potency, choosing to express anesthetic concentrations in the gas phase is arbitrary. Provided the anesthetic is close to equilibrium, [10]the MAC concentration can equally validly be expressed as a concentration in gas or free aqueous solution. For example, in Figure 1(lower line) we have also plotted the halothane MAC data [2,3,5]as concentrations (mM) in water. When the data are looked at in this way, the anesthetic requirement changes very little with temperature (in the range 28–43 degrees C), with the aqueous concentration varying only between 0.20–0.25 mM. (Using available potency [3,6]and partitioning [3,11]data, comparable changes, approximately 30% or less over a 15 degrees C range, can be calculated for the volatile anesthetics isoflurane, methoxyflurane, ether, and fluroxene.) Not only is there relatively little change in anesthetic requirement, but the extrapolation predicts something quite different: a zero anesthetic requirement will not be reached until -40 degrees C. Thus, the agreement between the gas-phase extrapolation and the point at which MAC drops to zero is completely fortuitous. (This also follows from the finding that gas-phase extrapolated temperatures can vary greatly from one anesthetic to another. [2]) Of course, at extreme temperatures (e.g., 20 degrees C in the goat [8]), there clearly are major physiologic changes that result in a vanishing anesthetic requirement. (Similarly, anesthetic EC⁵⁰values in goldfish tend to zero below about 2 degrees C. [12]) For example, [13]canine cerebral metabolism decreases steeply below 27 degrees C (the Q¹⁰for the rate of oxygen consumption between 14–27 degrees C is twice that for the range 27–37 degrees C) and could be the cause of the steep decrease in anesthetic requirement at these extreme temperatures. However, although extreme temperatures may cause such gross physiologic changes in complex organisms (probably involving numerous molecular processes), it seems unlikely to us that such precipitous effects will be seen on, or are relevant to, the effects of general anesthetics on an individual molecule such as an ion channel studied at room temperature.

Second, one must consider that, to produce general anesthesia, volatile anesthetics must bind to, or dissolve in, their targets in the central nervous system, and this process itself will depend on temperature. This raises the question of the extent to which the temperature dependence of volatile anesthetic potency can be accounted for by simple solubility considerations. [2,8,14,15]For example, there is a decrease in solubility with increasing temperature when halothane is transferred from the gas phase to almost any plausible target (lipid, protein, or water). Indeed, because halothane makes essentially no interactions in the gas phase but numerous interactions in any condensed phase, a large temperature dependence in gas-phase potency might be expected from first principles--as the temperature is increased, the favorable interactions in the condensed phase are weakened, and the anesthetic is "driven" into the gas phase. Such interactions show up as a release of heat (that is, as a negative enthalpy change Delta H) on binding/dissolution, and Delta H can be calculated from the slope of a van't Hoff plot of log(solubility) versus reciprocal absolute temperature. For halothane [11,16]dissolving from the gas phase into lipid bilayers or water, or binding to a protein (luciferase), Delta H is between about -35 to -60 kJ mol sup -1 (corresponding to a 2.0- to 3.3-fold increase in solubility/binding when the temperature is decreased from 37 to 22 degrees Celsius). This range, which presumably reflects the differing strengths of the interactions involved, encompasses the apparent Delta H =-52 kJ mol sup -1 that has been calculated [14]from the observed [2,3,5]temperature dependence of halothane MAC for dogs. (The same value of Delta H also has been calculated for halothane anesthesia in goldfish [12]). To exert its effects, an anesthetic must bind to its primary targets, so there is, perhaps surprisingly, little if any temperature dependence left to be accounted for in terms of an additional effect of temperature per se. In other words, it appears that, provided the deviations from normal body temperature are not too great, the effects of temperature per se can, to a first approximation, be ignored.

This conclusion is consistent with the finding that the temperature dependence of anesthetic potency varies considerably among different volatile agents. [2–6,8]Further support for this view comes from the observation [15]that the rat gas-phase MAC for the simple gas nitrous oxide (which might be expected to interact relatively weakly with its targets [17]) decreases only minimally with decreasing temperature. However, the idea that the temperature dependence of animal potency can be accounted for largely by changes in solubility or binding needs to be tested experimentally by studying the effects of temperature on the interactions between volatile anesthetics and their most likely molecular targets, ligand-gated ion channels in nerve membranes. [18].

If gas-phase MAC values (Figure 1) change simply because of changes in solubility, then log(MAC) might be expected from elementary thermodynamics to change linearly with reciprocal absolute temperature. In fact, when the halothane data in Figure 1are replotted in this way (see van't Hoff plots in Figure 2), we see that the fit is equally good. Indeed, fits of the data from published [2–7]mammalian animal temperature studies to linear and van't Hoff straight-line plots cannot be distinguished (F test, P > 0.5; calculations not shown).

To avoid the complications and uncertainties caused by the inevitably large temperature dependences of gas-phase potencies, it is much simpler to consider a concentration in some condensed phase; water is as good as any other, but has the great practical advantage that anesthetics are often applied as aqueous solutions in in vitro experiments. (In experiments where the anesthetic can partition between biologic phases, it is, of course, the free aqueous concentration at equilibrium that is relevant.) When water is chosen, we see that aqueous halothane concentrations in the narrow range 0.18–0.23 mM should be used for canine preparations over the broad temperature range 20–37 degrees C (see Figure 1). (The equivalent gas-phase concentrations can, of course, be used, but the large changes in gas-phase potency with temperature must then be taken into account.) Conversely, we see no justification for applying gas-phase concentrations measured at 37 degrees C to experimental preparations at room temperature. In our view, this can amount to overdosing the preparation severalfold (see also [19]). Many in vitro experiments at "clinically relevant" concentrations actually are using free aqueous concentrations that would be close to lethal in animals at their normal body temperatures (the therapeutic indexes for volatile agents can be as low as 2–4).

It must be stressed, however, that all in vitro experiments on mammalian systems should, ideally, be carried out at normal body temperature. This will not always be either appropriate or convenient, however, and experiments will inevitably be carried out at other temperatures. In such cases, where reliable animal data at different temperatures exist, we would recommend that appropriate experimental anesthetic concentrations be obtained by extrapolating these data logarithmically (Figure 2) to the temperature of the in vitro experiment. However, if reliable animal data are not available, the best procedure would be to regard the free aqueous concentration achieved at 1 MAC in the animal at normal body temperature to be the appropriate benchmark. [1,18]For convenience, these values are given in Table 1for the volatile anesthetics most commonly used in current clinical practice.

The authors thank Edmond Eger and Donald Koblin for their helpful comments on the manuscript, Carl Lynch and Joseph Pancrazio for valuable correspondence, and Robert Dickinson for stimulating discussions.