To the Editor:—
We are indeed delighted that we struck a nerve (and not an inhibitory one!) in such a distinguished group of investigators. We are also encouraged that there is so much agreement between our special articles. 1,2For example, Eger et al. 2conclude, as did we, that binding isotherms for a molecular target may be significantly shifted from the response isotherms for the same target, whether in vitro or in vivo . Thus, dissociation constants for an anesthetic–target interaction may be very different than EC50values for the function of that target. This should not be contentious because it is already well-documented. 3,4
We also agree that when multiple targets contribute to an effect, the EC50of this concentration–effect relation is shifted to the left of that of any of the individual components (their fig. 3, our fig. 1), even with the overly simplistic integrative paradigm of additivity. We both conclude that the steep slopes of concentration–effect curves represent low population variability, although we differ in the interpretation of this observation. It is possible that some anesthetic targets are so important to organism survival that their conservation is ensured, and in turn, their sites and responses to anesthetics are likewise well-preserved. However, we believe that this remarkably conserved response (across the whole animal kingdom) to inhaled anesthetics arises from multiple interacting targets—the normal heterogeneity in any one of which will only have a small effect on the final integrated behavioral response. This interpretation is consistent with knockout and inhibitor experiments to date: the effects on inhaled anesthetic potency is consistently small and incomplete. However, because our current understanding of the integrated behavior of complex neuronal circuits remains in its infancy, critical testing of these ideas will have to await the appropriate studies and modeling methodology.
Therefore, despite agreement, or at least no argument, with many points in our article that render sensitivity a criterion of questionable validity, Eger et al. 2chose to focus on whether the slope of the minimum alveolar concentration (MAC) response yields insight into the number of anesthetic targets and their sensitivity. This is where we disagree. As noted, the mechanism of low variability is not yet clear, so is it safe to draw conclusions from a simple, untested mathematical model that attempts to link the MAC slope quantitatively to underlying targets? Eger et al. 2do acknowledge that “Nonlinearities, including thresholds, amplification, and feedback, exist in biologic systems and may obscure the true association between sensitivity of the receptor target and sensitivity of the organism to anesthetics.” However, they then dismiss this “important caveat” by assuming “. . . a linear relation between anesthetic interaction with the receptor target and the anesthetic response of the whole organism.” Therefore, their conclusions should be viewed with considerable skepticism because of the admitted invalidity of a major underlying assumption. Even if one accepts this assumption, there are further assumptions that are arbitrary. For example, the conclusion that receptor EC50values “. . . are not expected to differ from clinically relevant concentrations by more than a factor of 3” is based largely on an assumption supported only by intuition:“We suggest that T [threshold] lies within 0.1–0.9.” If their intuition is off by only a little (T is, for example, 0.05–0.95) or if the Hill number is closer to 1 (there is little evidence of cooperative anesthetic binding to any target), the separation of receptor EC50from population EC50exceeds 10-fold, in accordance with our article. Therefore, it is essential for the reader to recall that a model rests on its underlying assumptions, and if they are faulty, any subsequent “analysis” is simply a mathematical exercise. Here, we are presented with a model containing an admittedly invalid assumption, and another that is arbitrary. Therefore, it would seem scientifically prudent to reclassify their conclusions as hopeful speculations.
Eger et al. 2will counter that we, too, used an invalid assumption (equal, additive effects of multiple targets) in our Special Article, 1but we used it as a lower limit of complexity to show that shifts in EC50can occur even with the simplest model of integration. We would welcome a credible argument that more complex, nonlinear models of integration reduce the likelihood and magnitude of shift in EC50.
Some have interpreted our initial Special Article to promote the use of any anesthetic concentration in in vitro research. This is not the case. We, too, believe that the likelihood of a target being an important contributor diminishes with increasing EC50. However, we believe it is premature and scientifically naïve to exclude the possibility of a target contributing to anesthesia based on this criterion alone. Both articles clearly show that once one accepts that more than one target contributes to anesthetic action, the relation between organism and molecular EC50values will diverge (as noted in our article, divergence is common, even with a single target). The magnitude and direction of divergence is not yet predictable. Models are important tools but must be tested by experiment and not by intuition.
What concentrations should be used in in vitro studies? Again, we advocate the construction of complete concentration–effect relations to reliably determine the sensitivity of a system. This occasionally requires very high concentrations to find the maximum response of a system. Confining ourselves to examining only “clinical concentrations” in in vitro studies limits the information derived, may be misleading, and cannot be regarded as rigorous science. We concede that some consistently measurable response should be evident at clinical concentrations for an in vitro system to be judged relevant. However, it is well-known that in vitro conditions can alter and, in some cases, eliminate or reverse responses to drugs, so even this concession must be viewed cautiously.
Eger et al. 2did not take the bait from our article and discuss the relevance of the MAC response to anesthesia and to in vitro targets. As we stated, anesthesia (as defined by the MAC response) is a quantal event and not a continuous, saturable response. Is it valid to quantitatively relate a continuous in vitro variable to this quantal, yes-or-no behavioral response? Perhaps, but only if the underlying mechanism of such a “threshold” is understood. That it is far from understood is exemplified by quotes from a recent article 5in which two of the recent articles’ authors are coauthors:“These results illustrate the difficulties in attributing behavioral responses to drug–receptor interactions . . . ,” and “. . . immobilization and hypnosis produced by volatile anesthetics are complex phenomenon mediated by multiple receptor populations.” This is exactly the argument underlying our original article. 1
We remain convinced that sensitivity of in vitro preparations cannot be rigorously used as a yardstick for relevance to in vivo effects. When constructing their Special Article, Professor Eger informed us that it was an attempt at consensus, normally a desirable corporate goal. However, consensus is a poor strategy for generating knowledge because it inhibits the creative approaches and ideas that move us forward. In the words of Walter Lippman, “Where all men think alike, no one thinks very much.” However, we do need to agree that continued debate and thinking in this area are necessary to foster new questions, further research, and ultimately greater insight into the mystery of how general anesthetics exert their clinical effects.