Correlational studies of unconsciousness under anaesthesia: how far can preclinical studies take us?

Determining the electrophysiological correlates of loss of consciousness (LOC) and recovery of consciousness (ROC) under anaesthesia is a holy grail of biomedical science.¹ From a purely practical perspective, a non-invasive metric that tracks sensory awareness and is robust to the combination of anaesthetic drugs and patient factors would be a welcome addition to the anaesthetist’s perioperative tool kit.² From a theoretical perspective, such a metric would prove invaluable for endeavours to understand the neural basis of consciousness, and in particular, sensory awareness.² The contribution by Plourde and Arseneau³ in this month’s British Journal of Anaesthesia provides evidence towards that end. The authors report that the α₂-agonist dexmedetomidine attenuates high-frequency oscillations in the cortex and thalamus at doses that induce loss of the righting reflex. By comparing these results with their previous work on the thalamocortical effects of isoflurane and propofol,^4,5 the authors conclude that changes in thalamic γ-band signals may be electrophysiological correlates of LOC and ROC. It is interesting to note that some of the results presented here differ from previous studies, in which low γ (∼40 Hz) band power increased under anesthesia.⁶ In contrast, the effects on high γ (>80 Hz) signals observed by Plourde and Arseneau³ are likely to represent the suppression of spontaneous spiking activity under anaesthesia, which is well supported in the literature.⁷

Plourde and Arseneau³ observed a sigmoidal dose-dependent reduction in local field potential (LFP) power in primary somatosensory thalamus and cortex across the γ band (30–200 Hz) for dexmedetomidine, propofol, and isoflurane, but observed interesting differences between agents. Sensitivity of γ-band signals to dexmedetomidine was far lower compared with isoflurane and propofol. In addition, both isoflurane and propofol attenuated γ power in the thalamus more than in cortex; this was especially pronounced for propofol. In contrast, the sensitivity of the cortex and thalamus to dexmedetomidine was similar. The authors conclude that the similar electrophysiological effects of propofol and isoflurane are consistent with these agents readily producing LOC, whereas the more modest electrophysiological effects of dexmedetomidine fit with its lower effectiveness in producing LOC. Indeed, data from volunteers suggest that dexmedetomidine sedation in humans might not be an unconscious state; dreams, a form of conscious experience with disconnection from the environment,² are common under dexmedetomidine.⁸ It may be that changes in γ power are less pronounced with dexmedetomidine as it allows for a maintained conscious state. Furthermore, even under high doses of dexmedetomidine, patients are still rousable, which may reflect a less profound effect on neural dynamics and connectivity. Of course, Plourde and Arseneau³ are limited by the use of behaviour as a metric for consciousness; all the animals lost the righting reflex, and there is no evidence for agent-specific effects on consciousness.

Studying consciousness and its neural correlates is a tricky business. First, studies of anaesthetic-induced loss of consciousness rely heavily on behavioural reports, which may be impaired in the clinical environment by neuromuscular blockade. In addition, researchers have yet to identify specific loci for consciousness (although see Siclari and colleagues⁹ and Koch and colleagues¹⁰), or even for contributory processes, such as arousal and attention. As such, research into the anaesthetic mechanisms of LOC and ROC has been exclusively correlative in nature. Although these studies vary in their specifics, they share the following approach: assay unresponsiveness as a correlate of LOC/ROC (e.g. absence of response to command in human subjects or loss of righting reflex in rodents) and simultaneously measure neuronal activity. By relating these physiological and behavioural metrics as precisely as possible, candidate signatures of LOC/ROC can be identified. However, although it is useful to link in time neural activity and drug dose to LOC/ROC transitions as precisely as possible, this can be difficult to achieve in practice. Large amounts of data are required, and analyses, such as the spectral analysis presented here, require assumptions of stationarity in brain signals, drug doses, and behavioural states.

An additional challenge of studying the mechanisms of LOC/ROC under anaesthesia is choosing where to place the electrodes. For example, depth-of-anaesthesia monitors based on recordings from the frontal cortex fail to track the conscious state under clinical anaesthesia.^11,12 Plourde and Arseneau³ focused their electrophysiological recordings in the cortex and thalamus, reasoning that these structures play a crucial role in the neural basis of consciousness.¹⁰ The focus on the thalamus is also in line with one of the classic models of LOC/ROC under anaesthesia, the thalamic switch hypothesis, supported by imaging, electrophysiological, and microinjection studies.¹³ More recent models, in contrast, emphasize the importance of actions on cortical networks and cortico-cortical connectivity.^2,13,14 However, although there is a strong argument to be made that effects on at least some aspects of consciousness are attributable to direct actions on the thalamus and cortex, substantial evidence suggests that anaesthetics can act in the midbrain and brainstem sleep and arousal centres to produce hypnosis.^15,16 The effects of dexmedetomidine are similar to those observed under non-rapid eye movement sleep.^6,16 Conversely, although isoflurane and propofol are well known to have direct actions in the cortex and thalamus, there is far less evidence along these lines for dexmedetomidine. This supports both the findings of Plourde and Arseneau³ and our recent finding that corticothalamic connectivity was less affected by dexmedetomidine than by propofol.¹⁷

Another common motif in studies of anaesthesia and consciousness, used here by Plourde and Arseneau,³ is to identify common neural effects with LOC/ROC across agents with diverse molecular targets. This approach is invaluable for distinguishing the manifold effects of anaesthetic agents that are extraneous to consciousness from those that are crucial.^15–17 Recent studies have used this approach to identify specific changes in cortico-cortical connectivity and specific thalamic loci as neural correlates of LOC/ROC. We have also undertaken similar analyses for dexmedetomidine, propofol, and stage 3 non-rapid eye movement sleep for functional magnetic resonance imaging connectivity, finding important commonality in thalamic connectivity across states.¹⁷ Although this approach can help to pinpoint changes in neural activity associated with unresponsiveness, it still makes the critical assumption that the behavioural states of unresponsiveness for each agent correspond to equivalent states of unconsciousness.

Loss of the righting reflex as a correlate of LOC in rodents is based on correlation between loss of the righting reflex and unresponsiveness in humans.¹⁸ However, the difficulties in interpreting the findings of Plourde and Arseneau³ show that this is too limited a construct to understand the anaesthetic state.¹⁹ Furthermore, connected consciousness may also occur clinically in the absence of spontaneous movement; for example, patients can exhibit sensory awareness after intubation despite ‘adequate anaesthesia’ with volatile anaesthetics or propofol.^12,20 This issue is exacerbated in experimental animals, which cannot be interrogated in real time or upon awakening as to the content of their experience under anaesthesia.

Like many careful studies, the report by Plourde and Arseneau³ raises many interesting questions to be addressed in future experiments. Establishing a causal link between γ-band activity in the thalamus and consciousness is one crucial outstanding issue. With expanding understanding of thalamic and cortical generators of γ oscillations, we will be able to test hypotheses of causality using pharmacogenetic and optogenetic approaches to manipulate these signals in animal models. In addition, to establish more rigorous assays of consciousness vs connectedness, we require studies in patients and human volunteers. Especially promising are studies involving increasingly precise electrophysiological measurements of neural activity in vivo. For example, studies involving electrocorticogram recordings from epilepsy patients will allow for unprecedented electrophysiological access to brain structures.²¹ Finally, the importance of studying patients in the operating room must not be forgotten; anaesthesia is a humbling profession, and we may yet have much to learn from a few individuals.

Authors’ contributions

M.I.B. wrote the editorial with input from C.M. and R.D.S.

Declaration of interest

R.D.S. is a member of the editorial board of the British Journal of Anaesthesia. MIB and C.M. have no conflicts to declare.

Funding

MIB is supported by R01GM109086. Other support came from departmental resources.

References