To the Editor:—

The recent special articles by Eger et al.  1and Eckenhoff and Johansson, 2although different in many specifics, are collectively important for their recognition that, ultimately, elucidating the mechanism of general anesthetic action requires understanding well beyond the binding of these drugs to one or more receptors and the subsequent modulation of that receptor’s activity. Each group of authors has offered simple models of what might occur when multiple receptors are bound by molecules with anesthetic properties. As each has amply demonstrated, such models are easy to critique but difficult to supplant. How do we begin the task of building more realistic models, and what might they reveal of general anesthetic mechanisms?

It is clear that general anesthetics bind and modulate the activity of multiple receptors and receptor classes, in which binding to the voltage-gated receptors primarily affects the membrane properties of individual neurons, and binding to the ligand-gated receptors primarily affects the interaction of these neurons. Therefore, a plausible model of anesthetic action must show how these effects at one or more receptors could map to the general anesthetic state. Unfortunately, at the current time, this is not a well-specified endpoint. Certainly, the induction of neural quiescence would lead to an anesthetic state, but this represents an anesthetic state far deeper than the established activity of the brain and spinal cord during general anesthesia would suggest. Electroencephalographic studies during administration of the volatile anesthetics or ethanol 3consumption exhibit an increasing level of organization, implying a greater degree of synchronous activity. Although this concept of anesthetic action seems to implicate the ligand-gated channels responsible for neural interaction, the situation is more complex. Theorists recognize at least two generic patterns of neural spike activity with distinct topologic signatures, one of which is more likely to lead to synchronous activity when incorporated into a network. 4Importantly, as demonstrated in simple examples of biophysical neuron models, one type of behavior can often be readily transformed into the other with simple parameter changes as could conceivably occur as a consequence of anesthetic interaction with the voltage-gated ion channels. 4Another membrane factor that could conceivably affect cell-to-cell coupling is the pattern of spike activity, in which suppression of bursting behavior, even if the average number of spikes remains relatively constant, could prevent depolarization of the postsynaptic neuron.

One persistent argument in the anesthetic mechanisms literature 5is that the receptors responsible for precipitating the anesthetic state when appropriately occupied should have a concentration–effect relation that parallels that of the in vivo  system. Attention is generally paid to the slope and midpoint of such curves. Experience with theoretical and computational neural models indicates that there is no reason to expect that these slopes should be at all the same. For example, at the single neuron level, consider the Morris-Lecar barnacle muscle fiber model. 6Persistent spike activity can be terminated in an abrupt fashion as the conductance of the Ca++channel is continuously decreased (A. G., unpublished computer simulation of Morris-Lecar model, 2000). At the other extreme are models involving multiple neurons, such as the theoretical solution 7of a now classic problem from the circadian rhythm literature, which demonstrates that there is a specific threshold for neural coupling, below which synchronous activity cannot occur. This could occur either through modulation of synaptic interactions (ligand-gated receptors) or through small decrements in the activity or pattern of activity of the individual neurons (voltage-gated receptors), either of which would lead to decreased coupling with members of the network. Each of these examples suggests more of a threshold type (bifurcation or phase transition) of in vivo  behavior as anesthetic modulation of receptor activity takes place, and that less-than-perfect threshold behavior is the consequence of biologic variability in the population. Already, this concept of a phase transition has been the basis for one large-scale model of general anesthetic action. 8 

Regarding the midpoint of the concentration–effect relation of in vitro  preparations, it has been argued that clinically relevant concentrations should lie close to this midpoint, and consequently, anesthetic modulation of the voltage-gated channels may be not be responsible for the anesthetic state. 5By recognizing the possibility of threshold behavior and linearizing about the midpoint of a hypothetical concentration–effect curve, Eger et al.  1hypothesize that clinically relevant concentrations should lie within a factor of 3 of the midpoint to reach thresholds in the range of 0.1–0.9. For the example of the Morris-Lecar model presented here, persistent neural spike activity ceased with only an 11% decrease in Ca++conductance. In a modification of the Morris-Lecar model to introduce bursting, 4burst duration decreases in a graded fashion as Ca++conductance is decreased so that the burst duration is more than halved when Ca++is reduced by 20% (A. G., unpublished computer simulation of modified Morris-Lecar model, 2000). Similar but more complex behavior is seen for more elaborate single cell models, such as that of the hippocampal CA-3 neuron 9(A. G., unpublished computer simulation of Pinsky-Rinzel CA-3 neuron model, 2000), although conclusions based on these more elaborate models will depend on the ability to introduce anesthetic modulation of all the ion channels accurately.

Thus far, the exchange between Eger et al.  1and Eckenhoff and Johansson 2has highlighted only the acute aspects of the induction of general anesthesia. However, it is now recognized that synaptic reorganization is ongoing in the central nervous system on a continuous basis, shaped by the prevailing pattern of presynaptic and postsynaptic activity. Moreover, there is a growing appreciation, at least in simple systems, that conductances of voltage-gated receptors can be differentially regulated to preserve a given activity pattern. 10Thus, induction of general anesthesia for a period of time could lead to fundamental alterations in central nervous system function. Receptors whose activity is modulated by anesthetics may or may not be essential for induction of the general anesthetic state but could play a role in how the nervous system responds to this state. Whether interactions like these could be the basis for some of the longer-term effects of general anesthesia 11remains speculative.

In summary, although we have come a long way, a full appreciation for the mechanism of general anesthetic action and its consequences will in all probability require a systems level approach emphasizing the collective interactions of multiple neurons in which the activity of one or more receptors has been modulated by an anesthetic and perhaps other drugs that are known to contribute to the anesthetic state. In addition to helping to solve the puzzle of anesthetic action and pave the way to more rational drug design and use, such approaches, through the need to address broad integrative aspects of central nervous system function, could have implications well beyond our specialty. The only thing that may be safe to say at this point is that this is an exciting area in which we could experience a few surprises.

Eger El II, Fisher DM, Dilger JP, Sonner JM, Evers A, Franks NP, Harris RA, Kendig JJ, Lieb WR, Yamakura T: Relevant concentrations of inhaled anesthetics for in vitro  studies of anesthetic mechanisms. A nesthesiology 2001; 94: 915–21
Eckenhoff RG, Johansson JS: On the relevance of “clinically relevant concentrations” of inhaled anesthetics in in vitro  experiments. A nesthesiology 1999; 91: 856–60
Ehlers CL, Havstad J, Prichard D, Theiler J: Low doses of ethanol reduce evidence for nonlinear structure in brain activity. J Neurosci 1998; 18: 7474–85
Rinzel J, Ermentrout B: Analysis of neural excitability and oscillations, Methods in Neuronal Modeling From Ions to Networks. Edited by Koch C, Segev I. Cambridge, Massachusetts, MIT Press, 1998, pp 251–91
Franks NP, Lieb WR: Molecular and cellular mechanisms of general anaesthesia. Nature 1994; 367: 607–14
Morris C, Lecar H: Voltage oscillations in the barnacle giant muscle fiber. Biophys J 1981; 35: 193–213
Mirollo RE, Strogatz SH: Synchornization of pulse-coupled biological oscillators. SIAM J Appl Math 1990; 50: 1645–52
Steyn-Ross ML, Steyn-Ross, DA, Sleigh JW, Jiley DTJ: Theoretical electroencephalogram stationary spectrum for a white-noise-driven cortex: Evidence for a general anesthetic-induced phase transition. Physical Rev E 1999; 60: 7299–311
Pinsky PF, Rinzel J: Intrinsic and network rhythmogenesis in a reduced Traub model for CA3 neurons. J Comput Neurosci 1994; 1: 39–60
Marder E, Abbott LF, Turrigiano GG, Liu Z, Golowasch J: Memory from the dynamics of intrinsic membrane conductances. Proc Natl Acad Sci U S A 1996; 93: 13481–6
Moller JT, Cluitmans P, Rasmussen LS, Houx P, Rasmussen H, Canet J, Rabbitt P, Jolles J, Larsen K, Hanning CD, Langeron O, Johnson T, Lauven PM, Kristensen PA, Biedler A, van Beem H, Fraidakis O, Silverstein JH, Beneken JEW, Gravenstein JS, for the ISPOCD Investigators: Long-term postoperative cognitive dysfunction in the elderly: ISPOCD1 study. Lancet 1998; 351: 857–61
for the ISPOCD Investigators: