For decades, it has been known that emergence from anesthesia occurs when the residual anesthetic drug levels are lower than those required for anesthetic induction.1 Hence, hysteresis between entry into, and exit from, the anesthetic state is observed. Hysteresis has traditionally been attributed to pharmacokinetic processes—the lag between plasma and effect-site equilibration. Indeed, pharmacokinetic-pharmacodynamic (PK-PD) models can be constructed such that the hysteresis loop is collapsed. However, recent theoretical and experimental findings postulate that intriguing neurobiology may be hidden by the PK-PD approach.2–4
A system exhibiting hysteresis, or path-dependence, is one that has memory of its previous state. For example, in the hysteresis zone in Figure 1, an individual could either be awake or anesthetized. Along any vertical line through the shaded area in Figure 1B, the effect-site drug concentration is identical. Thus, whether one is awake or anesthetized is not solely determined by drug concentration. Rather, the fate is determined by the previous state of arousal (if formerly anesthetized, then currently anesthetized, and vice versa). A related property of systems with hysteresis is that they exhibit a resistance to state changes. Thus, memory in hysteretic systems is expressed as resistance to state changes. PK-PD models attribute this resistance to state changes to the delay in the effect-site equilibration. While effect-site equilibration undoubtedly contributes to hysteresis, PK-PD models are not able to explain two phenomena that are observed by anesthesiologists every day. First, a patient who remains anesthetized at a low concentration of an anesthetic at the end of a surgical procedure can be immediately roused and show abrupt electroencephalogram (EEG) changes following a verbal stimulus or a light tap. Yet, the anesthetic concentration could not have changed in any appreciable way during the few milliseconds it takes to elicit arousal. Thus, at least in the low concentration range, the patient can be both awake and anesthetized at the same anesthetic concentration!5 This co-existence of the awake and the anesthetized states at the same drug concentration cannot be explained in terms of PK-PD formalism. A second, related observation is that the EEG fluctuates between lighter and deeper activity patterns at a fixed anesthetic concentration even in the absence of surgical stimulation.6,7 These brain activity fluctuations violate a fundamental tenet of PK-PD models—namely that, for a given effect-site concentration, there is only a single response.
What do these EEG fluctuations and abrupt changes in the state of arousal have to do with anesthetic hysteresis? Mathematical modeling suggests that there is a profound connection.8 The model relaxes the central assumption of PK-PD models, which states that, for each effect-site drug concentration, there is a single state of the brain, and asks, what would happen in a system that had more than one state at a given drug concentration. Surprisingly, the result is that hysteresis arises in such systems automatically. The key prediction of this hysteretic model is that, at a fixed anesthetic concentration, the behavioral responses of an individual will fluctuate spontaneously between being responsive and unresponsive (Fig. 1).
Recent experimental work validated key predictions of this mathematical model.8,9 The experimental paradigm is rather simple. Wild-type mice were exposed to an ED50 dose for hypnosis using isoflurane. The state of hypnosis in each mouse was repeatedly tested using a well-validated measure—the righting reflex. One might imagine multiple ways in which a population could average to its ED50 for righting reflex—the rodent surrogate for loss of consciousness. It is possible that, at the population ED50, the “sensitive” half of the animals have lost their righting reflex, while the “resistant” animals remain intact. This oft-assumed possibility was definitively ruled out by experimental findings. As predicted by the mathematical model, behavioral analysis demonstrated that all mice held at their population’s ED50 exhibited transitions into and out of hypnosis. Remarkably, these fluctuations were not entirely random—mice that were unresponsive on the previous trial were far more likely to remain unresponsive on the subsequent anesthetic trial.9 Hence, repeated behavioral testing of rodents under unchanging anesthetic conditions, and in the absence of any additional stimuli, unquestionably demonstrated evidence for a state memory expressed as resistance to state change.9 Stated otherwise, both the awake brain and the unresponsive brain showed a “stickiness”—a tendency to remain in its previous state of arousal. To generalize these observations, parallel experiments were performed on larval zebrafish exposed to propofol while repeatedly assessing their startle reflex. Results were identical to those observed in mice exposed to isoflurane. Each individual fish fluctuated between the responsive and the unresponsive state. Moreover, these fluctuations also exhibited resistance to state transitions.9
Previous animal studies using species ranging from fruit flies to rodents have unambiguously demonstrated experimental evidence for anesthetic hysteresis.2,10 Every attempt was made to control for PK-PD factors, including direct measurement of anesthetic concentration in the brain. However, the most definitive evidence for hysteresis came from genetic manipulations. Modification of single genes that do not affect anesthetic uptake, distribution, metabolism, or elimination has been shown to markedly affect hysteresis.10 These recent experiments definitively rule out the possibility that anesthetic hysteresis is solely due to effect-site equilibration.
The mechanisms that give rise to anesthetic hysteresis remained unclear. Recent experiments showing fluctuations in the state of arousal at a fixed drug concentration offer the first glimpse of these mechanisms. In mathematical models, the key determinant of hysteresis is the degree of resistance to state fluctuations observed at a fixed drug concentration. Remarkably, McKinstry-Wu et al9 find that the degree to which an animal resists state transitions is remarkably conserved amongst individuals. In contrast, individual responsiveness to anesthetics (personalized effective concentration) was markedly variable, even in a genetically homogenous population. Equipped with this new experimental paradigm, future work will be directed at identifying specific neuronal populations that contribute to this resistance to state changes.
Does neural inertia exist in humans? To date, clinical studies have found mixed support.5,11–14 The primary difficulty in providing definitive evidence for or against hysteresis in humans is that effect-site drug concentrations cannot be measured. Thus, simply by adjusting PK-PD parameters, one can either arrive at the conclusion that hysteresis does, or does not, exist based upon identical behavioral observations. However, with phylogenetic conservation of neural inertia found in species ranging from invertebrates to vertebrates, including mammals, the preponderance of evidence would suggest that it should. Identification of neuronal populations responsible for mediating resistance to state transitions could ultimately lead to novel therapies—one aimed at selectively stabilizing the anesthetized state and avoiding the potentially dangerous anesthetic overdose, and a distinct agent to selectively destabilize the state of anesthesia at the end of surgery. Such state (de)stabilizers would add an entirely novel class of therapeutics to the anesthesiologist’s armamentarium.
Andrew R. McKinstry-Wu, MD
Alex Proekt, MD, PhD
Max B. Kelz, MD, PhD
Department of Anesthesiology & Critical Care, Center for the Neuroscience of Unconsciousness and Reanimation Research Alliance, University of Pennsylvania Perelman School of Medicine Philadelphia, PA
The authors are grateful to the National Institute of Health for the following funding: R01 GM088156 (to MBK), R01 GM124023 (to AP), K08 GM123317 (to ARM-W) and to the Department of Anesthesiology & Critical Care at the University of Pennsylvania, Perelman School of Medicine.
1. Koopmans R, Dingemanse J, Danhof M, et al. Pharmacokinetic-pharmacodynamic modeling of midazolam effects on the human central nervous system. Clin Pharmacol Ther. 1988;44:14–22.
2. Friedman EB, Sun Y, Moore JT, et al. A conserved behavioral state barrier impedes transitions between anesthetic-induced unconsciousness and wakefulness: evidence for neural inertia. PLoS One. 2010;5:e11903.
3. Steyn-Ross ML, Steyn-Ross DA, Sleigh JW, et al. Toward a theory of the general-anesthetic-induced phase transition of the cerebral cortex. I. A thermodynamics analogy. Phys Rev E Stat Nonlin Soft Matter Phys. 2001;64:011917.
4. Kim H, Moon J-Y, Mashour GA, et al. Mechanisms of hysteresis in human brain networks during transitions of consciousness and unconsciousness: theoretical principles and empirical evidence. PLoS Comput Biol. 2018;14:e1006424.
5. Proekt A, Kelz M. Schrodinger’s cat: anaesthetised and not!. Br J Anaesth. 2018;120:424–428.
6. Shortal BP, Hickman LB, Mak-McCully RA, et al. Duration of EEG suppression does not predict recovery time or degree of cognitive impairment after general anaesthesia in human volunteers. Br J Anaesth. 2019;123:206–218.
7. Hudson AE, Calderon DP, Pfaff DW, et al. Recovery of consciousness is mediated by a network of discrete metastable activity states. Proc Natl Acad Sci U S A. 2014;111:9283–9288.
8. Proekt A, Hudson AE. A stochastic basis for neural inertia in emergence from general anaesthesia. Br J Anaesth. 2018;121:86–94.
9. McKinstry-Wu AR, Wasilczuk AZ, Harrison BA, et al. Analysis of stochastic fluctuations in responsiveness is a critical step toward personalized anesthesia. Elife. 2019;8:1–25.
10. Joiner WJ, Friedman EB, Hung HT, et al. Genetic and anatomical basis of the barrier separating wakefulness and anesthetic-induced unresponsiveness. PLoS Genet. 2013;9:e1003605.
11. Warnaby CE, Sleigh JW, Hight D, et al. Investigation of slow-wave activity saturation during surgical anesthesia reveals a signature of neural inertia in humans. Anesthesiology. 2017;127:645–657.
12. Kuizenga MH, Colin PJ, Reyntjens K, et al. Test of neural inertia in humans during general anaesthesia. Br J Anaesth. 2018;120:525–536.
13. Sepulveda PO, Tapia LF, Monsalves S. Neural inertia and differences between loss of and recovery from consciousness during total intravenous anaesthesia: a narrative review. Anaesthesia. 2019;74:801–809.
14. Ferreira AL, Correia R, Vide S, et al. Patterns of hysteresis between induction and emergence of neuroanesthesia are present in spinal and intracranial surgeries. J Neurosurg Anesthesiol. 2020;32:82–89.