ONE of the main effects of general anesthesia is the suppression of consciousness, i.e. , prevention of awareness. This must not be confused with absence of recall. A patient who does not have recall for events during general anesthesia may still have been conscious during the procedure. Recently, a case report in Anesthesiology described a patient who had awareness without recall during electroconvulsive therapy.1In several studies, Veselis et al. 2–4identified amnestic effects at subanesthetic concentrations of anesthetics and the according anatomical structures. In the current issue of Anesthesiology, two research groups try to identify the effect sites of anesthesia-induced loss of consciousness. At first glance, results of the two studies seem to be in conflict with one another.
Velly et al. 5studied patients undergoing intracerebral electrode placement for deep brain stimulation. Scalp electrodes and a deep brain electrode were used for recording of electrical activity of cortical and subcortical structures. During general anesthesia, the first effect was slowing of cortical electroencephalogram with prominent amplitudes in the δ range, whereas the signal obtained from the deep brain electrode showed faster spindle activity with less prominent slowing of the power spectrum. As cortical slowing of electroencephalographic activity was more prominent and occurred earlier, the authors concluded that the main effect site of anesthesia-induced unconsciousness is the cerebral cortex.
In contrast, Alkire et al. 6identified subcortical structures as one of the main effect sites of anesthesia-induced unconsciousness in animals. In their study, microinjection of nicotine into the central medial thalamus, a subcortical structure, reversed loss of righting reflex induced by sevoflurane. Given the specificity of both drug effect and localization, the authors conclude that general anesthesia blocks endogenous cholinergic arousal mechanisms in the central medial nucleus of the thalamus, i.e. , the main targets of general anesthesia are thalamic structures.
Therefore, these reports provide evidence in support of both cortical as well as for subcortical structures as target sites of general anesthesia (fig. 1).
Fig. 1. Diagram of interactions among brain regions during sensory input, modified from John and Prichep. 11,The scheme illustrates the pathway of sensory inputs through the brainstem reticular formation (ascending reticular activating system) and thalamus to the cortex. It is suggested that corticocortical and corticothalamocortical loops (CT–TC loops) are essential for perceptive processes of the brain, providing the basis of “consciousness.” Intralaminar (unspecific) and ventrobasal (specific) nuclei serve as gates to cortical layers I and V. (Cortical) pyramidal neurons ( top ) compare (unspecific) input into layer I and (specific) input into layer V. With concurrent stimulation, corticothalamic discharges are markedly enhanced, and γ frequency is back-propagated to cortical regions where the coincidence had occurred. As suggested by Velly et al. ,5,direct suppression of cortical cells by anesthetics ( red ) induces unconsciousness. This is supported by the coincidence of loss of consciousness with changes at the cortical level. As Alkire et al. 6 show, this mechanism can be reversed by nicotinic stimulation of the ascending reticular activating system ( green ), which reopens thalamic gates, allowing throughput of stimuli to the cortex (arousal).
Fig. 1. Diagram of interactions among brain regions during sensory input, modified from John and Prichep. 11,The scheme illustrates the pathway of sensory inputs through the brainstem reticular formation (ascending reticular activating system) and thalamus to the cortex. It is suggested that corticocortical and corticothalamocortical loops (CT–TC loops) are essential for perceptive processes of the brain, providing the basis of “consciousness.” Intralaminar (unspecific) and ventrobasal (specific) nuclei serve as gates to cortical layers I and V. (Cortical) pyramidal neurons ( top ) compare (unspecific) input into layer I and (specific) input into layer V. With concurrent stimulation, corticothalamic discharges are markedly enhanced, and γ frequency is back-propagated to cortical regions where the coincidence had occurred. As suggested by Velly et al. ,5,direct suppression of cortical cells by anesthetics ( red ) induces unconsciousness. This is supported by the coincidence of loss of consciousness with changes at the cortical level. As Alkire et al. 6 show, this mechanism can be reversed by nicotinic stimulation of the ascending reticular activating system ( green ), which reopens thalamic gates, allowing throughput of stimuli to the cortex (arousal).
Cerebral Cortex as Effect Site for Anesthesia-induced Unconsciousness?
Velly et al. 5found that at loss of consciousness after sevoflurane or propofol administration, cortical δ activity increased and (faster) β activity decreased. As a consequence, spectral edge frequency (SEF90), median frequency, and the nonlinear parameter dimensional activation decreased. Similar changes occurred at the subcortical level, but these changes occurred later and more gradually. The coincidence of loss of consciousness with changes at the cortical level suggests that anesthesia-induced unconsciousness is induced by changes at the cortical level. Even if these aspects seem convincing, there may be several limitations in the current approach. For cortical electroencephalographic analysis, only a frontal channel (F3–C3) was used, which may only in part reflect spatiotemporal distribution of cortical electroencephalographic activity. In addition, differences in the power spectrum may—at least in part—be due to higher frequency spindle activity in deep brain signals. Cortical spindle activity may not primarily appear in frontal leads but may be picked up by central or occipital leads. It remains to be determined whether analysis of a multichannel electroencephalogram would have led to identical differences in the power spectrum.
Another concern relates to the type of signal analysis which may give erroneous results during spindle activity. As used in the current approach, frequency analysis may result in increases in both SEF90and median frequency during slow δ activity. This seems to be crucial because γ-aminobutyric acid type A (GABAA) agonists induce spindle activity (14 Hz) reflecting inhibitory effects rather than less inhibition as suggested by analysis of SEF or median frequency. It must also be remembered that anesthetics induce slowing of cortical activity independent from subcortical structures.7
Timing of signal changes is the strongest point supporting the primary role of the cortex.5Still, it remains unclear whether anesthesia-induced changes at the subcortical level are related to the signal changes seen at the cortical level, i.e. , slowing of electrical activity may not be an uniquely valid representation of anesthetic effect, and Velly et al. 5do not exclude the role of the thalamus but suggest that the “main site” of action is the cerebral cortex.
Thalamus as Effect Site for Anesthesia-induced Unconsciousness?
The thalamus serves as a primary relay station for incoming sensory information and motor output from the brain. Previously, the existence of a thalamic consciousness switch has been suggested.8Alkire et al. 6used an animal model to examine the influence of the cholinergic system at the thalamic level. They found that cholinergic stimulation of the central medial nucleus of the thalamus reversed sevoflurane-induced loss of righting reflex in both receptor- and site-specific manners (fig. 1, green). Cholinergic blockade of this area, however, did not reduce sevoflurane requirements. They concluded that thalamic acetylcholine receptors have a role in regulating the “on” switch, i.e. , arousal, and sevoflurane blocks endogenous cholinergic arousal mechanisms.
As discussed by Alkire et al. ,6the role of cholinergic blockade is mainly supported by changes in the electroencephalogram and suppression of spontaneous motor movement, whereas knockout mice without β2nicotine receptor show no alterations of anesthetic requirements, and cholinergic antagonists per se do not produce unconsciousness.
Despite the meticulous localization experiments performed in this study, the central medial nucleus of the thalamus may not be the most crucial site of action with respect to the cholinergic theory of general anesthesia. Anesthesia may block arousal by inhibiting cells of the central medial nucleus of the thalamus, but this approach neglects the brainstem and the septal–hippocampal system, which in this context may also play an important role.
On the other hand, the cholinergic system may not be the exclusive effect site of general anesthesia. It has been suggested that volatile anesthetics act on GABAA, serotonin, acetylcholine, and possibly glutamate receptors.
Microinjection of the GABAAantagonist gabazine into the tuberomammillary nucleus has been shown to attenuate the sedative response to propofol and pentobarbital, whereas microinjection of the GABAAagonist muscimol led to a dose-dependent sedation.9The role of GABAAis supported by known interactions between GABAAin the central medial nucleus of the thalamus and nicotinic mechanisms (as identified in the current study of Alkire et al. 6).
Integrating cortical and subcortical effects on different receptors, the following mechanism of arousal is consistent with the results of Alkire et al. : If the ascending reticular activation system receives input, the brainstem reticular formation inhibits the nucleus reticularis, which itself opposes GABAergic inhibition by acetylcholine. Thalamic oscillators (in the frequency range of 8–12 Hz) are activated, thalamic gates are opened, and inputs from the exogenous system are transmitted via projection pathways to axosomatic synapses of pyramidal neurons in lower layers of cortex.
Thalamocortical Interactions and Anesthesia-induced Unconsciousness
In contrast to a prominent or exclusive role of either cortex or thalamus, the importance of intact thalamocortical–corticothalamic loops has been highlighted.8,10In the current study, Alkire et al. 6suggest that their results support the role of a corticothalamic reentrant mechanism of neural activity.
In brain metabolism studies using positron emission tomography scans, a general depression from volatile anesthetics was found with pronounced effects on cuneate nucleus, thalamus, midbrain reticular formation, dorsolateral prefrontal cortex, medial frontal gyrus, inferior temporal gyrus, cerebellum, and occipital cortex.8This can be induced by a hyperpolarization block of thalamocortical relay nuclei in thalamic networks. Therefore, general anesthesia induces a change from thalamic throughput to closed thalamocortical gates.
In a recent review, John and Prichep11highlight the importance of synchronized activity of cortical and subcortical structures: Pacemaker neurons in the thalamic region that oscillate in the α frequency range (8–12 Hz) regulate and synchronize the excitability of thalamocortical pathways.
In thalamic nuclei, three main types of neurons interact: thalamocortical relay nuclei, whose axons project to the cortex; reticular nucleus neurons, which interact synaptically with thalamocortical relay cells (GABAergic inhibitory feedback control); and local intrinsic neurons. Thalamocortical relay nuclei consist of relay cells (producing spikes in response to input) and oscillatory cells (producing rhythmic bursts of high frequency spikes, repeated in rhythmic oscillatory pattern). The brainstem reticular formation receives collateral input from all afferent sensory pathways. The importance of this structure is supported by the following experiments: Bilateral transsection of this area induces long-lasting coma, whereas electrical stimulation leads to activation and desynchronization of the electroencephalogram. This system is denoted as the ascending reticular activation system. Cholinergic activation diminishes the influence of GABAergic reticular nucleus neurons, which removes hyperpolarization and thus facilitates throughput to the cortex. This results in cortical arousal, reflected by desynchronization of α waves. As a consequence of event-related desynchronization, corticocortical interactions generate β rhythms (12–25 Hz). With concurrent stimulation (reflecting coincident exogenous and endogenous input), corticothalamic discharges are markedly enhanced, and γ frequency is back-propagated to cortical regions where the coincidence had occurred. This feedback has been suggested to bind distributed fragments, and coherent corticothalamocortical loops reverberate at γ frequencies (25–50 Hz).
As a consequence of these findings, the coherence of β and γ activity seems crucial for consciousness. As suggested by John and Prichep,11fronto-occipital activity is functionally uncoupled with loss of consciousness. General anesthesia interrupts and consciousness restores corticothalamocortical reverberation resulting from detection by the pyramidal neurons of coincidence between the exogenous readout of episodic memories, endowing sensations with meaning.
Inhibition of either cortex or non–sensory-specific diffuse thalamic projection nuclei block corticothalamocortical reverberations hypothesized to be critical for awareness. Closing of the gates of the thalamic diffuse projection system will therefore induce loss of consciousness.
On the background of these mechanisms, John and Prichep11suggest the following “anesthetic cascade”:
Direct hyperpolarization of thalamic and cortical cell membranes12
Suppression of midbrain/pontine areas involved with regulation of arousal, removing excitatory input to thalamocortical loops (inhibiting glutamatergic and cholinergic neurotransmission)13
Enhancement of GABAAsynaptic neurotransmission (inhibitory circuitry within thalamocortical loops) 14
Different anesthetics may only use one of the mechanisms, some agents may use various combinations. This supports Alkire’s view that any push toward hyperpolarization of thalamocortical loops will cause loss of consciousness. Therefore, the state of general anesthesia is produced by enhanced inhibition in a widespread manner, where the thalamus has a “filter” function, blocking the throughput of peripheral stimuli to the cortex.
No single effect on an isolated brain structure can possibly explain how anesthetics produce unconsciousness. Effects of general anesthesia involve a variety of neurotransmission processes, and the effects are widespread throughout the brain. Parts of this complex system have been elucidated by the work of Alkire et al. 6This hypothesis may also be supported by the results of Velly et al. ,5who suggest a primarily cortical effect. Their hypothesis is strongly based on activation of electroencephalographic δ activity, which may, however, not only reflect direct influence of anesthetics on cortical cells. It may also be induced by diminished activation of the cortex by the ascending reticular activation system and extreme depression of thalamic gates.
It seems unlikely that anesthetic effect on a unique neuroanatomical structure is both a necessary and a sufficient condition to produce loss of consciousness. Combinations of positron emission tomography and functional magnetic resonance imaging studies (providing spatial information) on the one hand and electroencephalographic and evoked potential studies (providing temporal information) on the other hand may finally identify brain targets of general anesthesia and their connectivity or disruption during anesthesia-induced loss of consciousness.
* Associate Professor, † Professor, Director, and Chair, Department of Anesthesiology, Klinikum rechts der Isar, Technische Universität München, Munich, Germany. e.f.kochs@lrz.tum.de
