“…diverse anesthetic agents reversibly perturb functional brain networks in a way that cripples the information transfer on which normal consciousness seems to depend.”
UNDERSTANDING the mechanism of anesthetic-induced unconsciousness remains important for anesthesiology. A precise answer to the longstanding question of how these drugs work would fortify the scientific underpinnings of the field and create new opportunities for improved clinical care through novel anesthetic designs or improved brain monitoring. On the one hand, there is something common to all of the drugs in our armamentarium—they render our patients unconscious or, at least, oblivious to interventional insults. On the other hand, these drugs are structurally, pharmacologically, and neurobiologically diverse. There does not appear to be a trivial explanation at the level of molecular or even neural targets, but could there be some mechanistic Rosetta stone that translates the variety of anesthetic actions to the common language of unconsciousness? The study of dexmedetomidine by Hashmi et al.1 provides further evidence that impaired information transfer in inefficient brain networks might be of central importance.
The investigators reanalyzed functional magnetic resonance imaging data from a previous study of dexmedetomidine-induced unconsciousness2 in which 15 healthy volunteers received a 1-μg/kg loading dose of the drug followed by an infusion at 0.7 μg · kg-1 · h-1. During the experimental protocol, resting-state neuroimaging data were acquired; the designation of resting state denotes that the volunteer was in a relaxed, eyes-closed condition during which the brain was not actively engaged in a cognitive task. After data were acquired during various levels of arousal, they were computationally “sliced and diced” into 131 brain regions, a process known as parcellation. Using these parcels as nodes, a network was reconstructed and analyzed for differences between the states of consciousness (baseline, recovery) and dexmedetomidine-induced unconsciousness.
Modern network science rests on the mathematical technique of graph theory, which extracts common features of distinct networks (e.g., internet, brain, and airport system) in a way that enables standard analysis. Graphs are made up of nodes (i.e., the points on the graph) and links (i.e., the connections between the points). Once defined, the various properties of the nodes and links (table 1) can be mathematically analyzed and visually represented. Although this analysis is clearly technical, we can nonetheless appreciate the principles of network science from experiences in our daily lives. One of the most intuitive examples of a complex network is that of an airport system.
The cities on the map in figure 1 represent the nodes of an airline network, and the lines of travel between those cities represent the links. It is immediately clear that not all nodes are alike: there are sparsely connected airports that are regional nodes and densely connected airports that are hub nodes. Hubs are critically important for networks and tend to be larger nodes that are highly connected and that facilitate efficient pathways. Most of us have had the experience of realizing the importance of network hubs when a major airport is disabled due to a snowstorm. Although a snowstorm in a regional airport might affect local travelers, inclement weather in a hub airport can result in widespread and crippling inefficiency.
Hashmi et al.1 found that dexmedetomidine-induced unconsciousness was characterized by various neural snow-storms in brain networks, with a preferential effect on highly connected hub nodes. The disruption of hubs was accompanied by reduced network efficiency, which likely leaves bits of information stranded in different brain regions in the same way that travelers are stranded in various cities when bad weather disables a major airport. Disruption of hub organization and other elements of functional network architecture—with negative consequences for network efficiency—has been shown to occur during propofol- and isoflurane-induced unconsciousness.3–8 Furthermore, applying computational “lesions” to hub nodes in human brain network models9 recapitulates the changes in information-theoretic measures empirically observed during propofol-, sevoflurane-, and ketamine-induced unconsciousness in humans.10,11 This suggests the possibility that, despite molecular and neurobiologic differences, diverse anesthetic agents reversibly perturb functional brain networks in a way that cripples the information transfer on which normal consciousness seems to depend.
There are two important points to note. First, the preferential target of hub structures in the brain is intriguing given the various lines of evidence suggesting that the primary sensory cortex and sensory thalamocortical networks appear relatively preserved despite anesthetic-induced unconsciousness.12 Of course, turning up the dial will ultimately suppress all brain functions but, at doses consistent with unresponsiveness, information transfer to the primary sensory cortex seems to be resilient to anesthetic exposure.13 This might beg the question, if information is being transferred to sensory cortex, how can a patient be unconscious? The current study, and others like it, provides an answer: the higher-order hubs that facilitate network efficiency and information transfer are disabled. Returning to our airline analogy, if I wanted to fly from rural Michigan to Paris, a snowstorm in the Detroit hub airport would likely prevent the trip, even if my regional airport was minimally affected (although local efficiency was also found to be impaired in this study). Second, although it is tempting to interpret the study of Hashmi et al.1 in the context of the Integrated Information Theory,14 we must exercise restraint for several reasons: (1) we do not need to invoke Integrated Information Theory to assert the importance of synthesizing neural information for the generation of consciousness—this can be argued based on conventional neuroscientific principles alone; (2) showing impaired integration during general anesthesia supports some claims of Integrated Information Theory of consciousness, but leaves other and less obvious predictions (e.g., consciousness has an identity relationship with integrated information; consciousness is a closed system; nonbiologic systems can have some degree of consciousness) unaddressed; and (3) demonstrating anesthetic disruption of functional hub architecture, information transfer, and efficiency is also consistent with a variety of other theories of consciousness, including (for example) global neuronal workspace, higher-order representationalism, and recurrent processing.15 To distinguish among these options, new paradigms and protocols will need to be designed. Furthermore, it will be critical to move beyond mere surrogates of information processing or transfer—which are known to have assumptions and be vulnerable to inaccuracy16 —to the measure of cognitive information itself.
In summary, the study by Hashmi et al.1 demonstrates that a sedative-hypnotic drug with a distinct molecular target and distinct neurobiology can—like propofol and halogenated ethers—disrupt functional hub structure and impair efficiency at multiple scales in large-scale brain networks. This work supports the prediction of graph-theoretical analysis as an approach that “promises to provide a framework for mechanistic studies”17 of anesthesia and, more importantly, sheds further light on a Rosetta stone that reveals how the different molecular languages of diverse anesthetics can all be translated to the common expression of oblivion.
The author is not supported by, nor maintains any financial interest in, any commercial activity that may be associated with the topic of this article.