If the in vivo effects of anaesthesia are mediated through a specific receptor system, then a relationship could exist between the regional changes in brain metabolism caused by a particular agent and the underlying regional distribution of the specific receptors affected by that agent. Positron emission tomography data from volunteers studied while unconscious during propofol (n=8) or isoflurane (n=5) anaesthesia were used retrospectively to explore for evidence of relationships between regional anaesthetic effects on brain glucose metabolism and known (ex vivo) regional distribution patterns of human receptor binding sites. The regional metabolic reductions caused by propofol differed significantly from those of isoflurane. Propofol’s reductions negatively correlated most significantly with the regional distribution of [3H]diazepam and [3H]flunitrazepam (benzodiazepine) binding site densities (r=–0.86, P<0.0005; r=–0.79, P<0.005, respectively) and less strongly with [3H]naloxone (opioid) binding density (r=–0.69, P<0.05). Isoflurane’s reductions positively correlated only with muscarinic (acetylcholine) binding density (r=0.85, P<0.05). These findings are consistent with the hypothesis that some of propofol’s in vivo anaesthetic effects may be mediated through a GABAergic mechanism and suggest some of isoflurane’s in vivo effects might involve antagonism of central acetylcholine functioning.
Br J Anaesth 2001; 86: 618–26
Accepted for publication: November 29, 2000
Gamma‐aminobutyric acid (GABA) is the primary inhibitory neurotransmitter in the mammalian brain. GABA and GABA agonists increase neuronal inhibition by interacting with the post‐synaptic GABAA complex receptor and activating an inhibitory chloride conductance.1 Increases in the GABAergic transmission results in the inhibition of post‐synaptic neuronal functioning. The GABAA complex receptor is a large post‐synaptic transmembrane protein with numerous effector sites.2 The primary effector site for benzodiazepines is thought to be on the alpha subunit of this protein.1 It has been suggested that the mechanism of anaesthesia could involve the GABAA complex receptor.3–5 In support of the GABA hypothesis is the fact that both i.v. and inhalation anaesthetic agents are known to have in vitro effects that modify the functioning of the GABAA complex.5–12 In spite of the large literature investigating the in vitro actions of anaesthetics on GABAA complex functioning, the importance of each particular agent’s in vitro GABA agonist activity for ultimately producing the anaesthetic state associated with each particular agent remains to be fully established in vivo during anaesthesia.
We postulated that if an anaesthetic agent’s mechanism of action involves its GABAergic effects, then a relationship might be demonstrable between the regional cerebral metabolic effects of such an anaesthetic and the known regional distribution of GABAA/benzodiazepine binding sites. Such a relationship, coupled with the known inhibitory nature of the GABA system, would seem to predict that regional brain metabolism should decrease more during anaesthesia in those brain areas which have more receptors.
There are, however, a number of caveats to exploring this idea. First, there is no guarantee that a local decrease in brain metabolism would be secondary to anaesthetic actions within the local area itself. It could be that anaesthetic actions on neurons far removed from the area of interest are the ones actually responsible for the observed local effect. Nonetheless, as GABAergic neurons are generally inhibitory interneurons located within various brain structures, it seems reasonable to expect that a substantial portion of an anaesthetic’s effects in any particular structure could be primarily a result of local factors. Second, and conversely, the density of receptors in an area may not determine the local response in that area. It might rather determine a response in some distant area, and the effects in such distant areas could be much larger or smaller in size than the original area of interest. Third, the relationship between regional metabolic reduction during anaesthesia and the various receptor systems involved in producing such effects is likely to be very complex and not well modelled by any attempts at trying to relate metabolic reduction with any single receptor system in isolation. Finally, it remains conceivable that the underlying assumption, regarding regional cerebral metabolic reduction effects during anaesthesia being somehow related to an anaesthetic’s effects on a regionally distributed receptor system, might be in error. Perhaps a lipid, or gap junction‐based model of anaesthesia could better explain the regional cerebral metabolic depressions seen during anaesthesia.
In any event, before undertaking the formal investigation of this receptor‐metabolic reduction hypothesis with current brain imaging technology, it would seem prudent to have at least some suggestion of which receptor systems and anaesthetic agents should be examined in order to yield a positive result. Thus, as a first step towards developing the above hypothesis, we retrospectively examined the regional cerebral metabolic depression effects of propofol and isoflurane measured in vivo with positron emission tomography (PET) during general anaesthesia and correlated these data with the reported ex vivo regional distribution of human benzodiazepine binding sites.1314 In addition, to explore the possibility that other cerebral receptor systems may contribute to the regional cerebral metabolic effects of either propofol or isoflurane, we also correlated a number of those other receptor systems with the regional cerebral metabolic effects of both propofol and isoflurane.15 We chose to use ex vivo data for these initial analyses instead of in vivo PET data because the ex vivo data offers better regional quantification for a number of these ligands of interest.
The analyses presented in this manuscript are primarily based on cerebral metabolic data previously collected and reported for both propofol and isoflurane anaesthesia.1617 However, these data were re‐analysed and updated for this report. This report includes regional metabolic data from three subjects not previously reported in the propofol group. Also, this report includes additional regional metabolic data from brain areas not previously reported, such as the caudate and putamen. For both studies Institutional Review Board Committee approval was obtained and all volunteers subsequently gave their written informed consent.
For the propofol study, eight healthy right‐handed males each underwent two PET scans; one during a mask general anaesthetic with a propofol infusion, and one while awake. The details of the anaesthetic induction technique and the PET scanning procedures used in this study have been reported previously.16 Briefly, a propofol infusion technique was used to anaesthetize spontaneously breathing subjects. The infusion was adjusted to the point where each subject just became unresponsive to verbal or tactile stimulation (mean rate=8.8 mg kg–1 h–1; mean resultant blood level=3.4 µg ml–1 plasma). Regional cerebral glucose metabolism was assessed using the 18fluorodeoxyglucose (FDG) technique. PET scans were done after each FDG uptake period using a NeuroEcat scanner. The PET scanner has a single ring with shadow shields and septa to achieve 7.6‐mm resolution (full‐width half‐maximum) in plane and 9.9 mm resolution in the Z‐dimension. For each PET scan session 13 image slices were obtained parallel to the canthomeatal line. Subjects were positioned using laser guidance and scans started at the level of 85% of head height (vertex to canthomeatal line, usually 12–14 cm) and stepped downward in steps of 10 mm.
For the isoflurane study, five healthy right‐handed males each underwent two PET scans; one during a mask general anaesthetic with isoflurane and one while awake. The details of the anaesthetic induction technique and the PET scanning procedures used in this study have been reported previously.17 Briefly, isoflurane was used to anaesthetize spontaneously breathing subjects. The anaesthetic was adjusted to the point where each subject just became unresponsive to verbal or tactile stimulation (mean expired end‐tidal concentration=0.5±0.2%, range=0.4–0.9%). Regional cerebral glucose metabolism was again assessed using the FDG technique. The PET scans were done after each FDG uptake period using a newer GE 2048 head‐dedicated scanner. Two sets of 15 image planes, resulting in 30 PET images across the whole brain, were obtained per subject. The GE PET scanner has a resolution of 4.5 mm (full‐width half‐maximum) in plane and 6.0 mm axially. Scans were obtained relative to the canthomeatal line. Subjects were positioned using laser guidance and a thermosetting plastic facemask was used to hold each subject’s head stationary during the period of image acquisition. In vivo attenuation correction was obtained by prior transmission scanning using a (68Ge/68Ga) rod source.
Glucose metabolic rate (GMR) values (mg 100 g–1 min–1) were calculated from scan data using the well‐established deoxyglucose kinetic models for autoradiography in animals and modified for humans.18 Relative GMR (rGMR, defined as GMR within a region of interest divided by whole brain GMR) was used to investigate if the pattern of metabolic changes differed between the two anaesthetic agents studied. rGMR normalizes the global effects of each anaesthetic agent on GMR and allows for direct regional comparisons between agents. An unmatched two‐tailed t‐test was used to evaluate the effects of propofol versus isoflurane on the various regions‐of‐interest examined; a P value <0.05 was considered significant.
Absolute per cent changes in glucose metabolism for the selected brain regions were calculated by comparing the percentage change from the baseline condition to the anaesthetized condition for each region examined. This calculation of absolute metabolic percentage change quantifies the magnitude of the metabolic reduction occurring in each particular brain region‐of‐interest during anaesthesia and allows for the most direct comparison of the regional percentage change data between the anaesthetics, which were obtained on different PET cameras. For both data sets the scans were transformed into regional glucose metabolic rate values as previously described.19
Regions‐of‐interest, appropriate for each analysis, were selected with a stereotactic method (see Haier and colleagues who shows the region‐of‐interest templates).20 Briefly, an automated edge‐finding algorithm was used to overlap a template of stereotactically derived region‐of‐interest boxes (25×25 pi) onto each subject’s 2D‐PET data slices. Metabolic rates were taken as the average pixel values within the boxes. The slices and various templates were matched in the Z‐dimension. Regions that spanned more than one PET slice thickness, such as the thalamus or frontal lobes, were averaged together to obtain a single value for a single subject. The percentage change data for each region were determined for each subject. These values were then averaged together across the subjects to determine the mean regional percentage decrease in absolute glucose metabolism for each particular region with each anaesthetic agent.
The regions selected for the correlation analyses corresponded to the regions‐of‐interest examined and reported in the literature for the different receptor systems studied. For determination of human benzodiazepine receptor density ([3H]diazepam), the data of Braestrup and colleagues (1977) were used,13 similar to the analysis performed by Buchsbaum and colleagues.21 In addition, the more recent benzodiazepine receptor data of Zezula and colleagues (1988) were also examined.14 These data examined [3H]flunitrazepam binding in the human brain. To obtain a single number for each region of interest within the data of Zezula and colleagues, the values reported for each named area were simply averaged. Thus, for example, even though the hippocampus lists eight separate areas where [3H]flunitrazepam binding was determined, we took the value of the ‘hippocampus’ to be the average of these eight values. This averaging procedure probably results in a reasonable estimation for the binding sites in each region as it essentially degrades the higher resolution autoradiography sampling to one, which approximates the lower sampling resolution of the PET technique. For determination of GABA ([3H]GABA), acetylcholine (muscarinic antagonist, [3H]quinuclidinyl benzilate), serotonin ([3H]5‐hydroxytryptamine), opiate ([3H]naloxone), and beta‐adrenergic ([3H]dihydroaloprenolol) receptor densities, the data of Enna and colleagues (1977) were used.15 The percentage of absolute metabolic reduction (from baseline awake to anaesthetized) occurring in each brain region was correlated with the different reported receptor densities using Pearson’s r coefficient and Fisher’s r to z transform. Percentage change data are plotted as mean (sem). Regions‐of‐interest that were too small for the spatial resolution of PET (such as the dentate nucleus or hypothalamus) were not selected from each data set.
The pattern of relative regional cerebral metabolism evident during the anaesthetic state differed significantly between the two agents in a number of different brain regions (Fig. 1). Relative glucose metabolism was significantly higher during isoflurane anaesthesia than during propofol anaesthesia in the temporal (P<0.001), parietal (P<0.001), and occipital (P<0.05) cortical areas. Relative glucose metabolism was significantly higher during propofol anaesthesia than during isoflurane anaesthesia in the basal ganglia (P<0.0001), thalamus (P<0.05), and midbrain (P<0.01). This analysis demonstrates that propofol and isoflurane have a differential effect on regional cerebral metabolism in six of the 10 regions examined. In essence, propofol appeared to be associated with lower relative cortical metabolic rates compared with isoflurane and higher relative subcortical metabolic rates, especially in the basal ganglia.
The two benzodiazepine receptor binding data sets revealed a similar rank ordering of regions between the two studies. In general, the cortex had more receptors than the hippocampus and cerebellum, which had more receptors than the basal ganglia and thalamus, which had more receptors than the pons and white matter. However, the rank ordering between the studies was not identical. Nonetheless, the quantified data from each study correlated highly with the other across the 11 regions examined (r=0.934, P<0.0001; Table 1).
The regional metabolic reductions evident during propofol anaesthesia correlated most significantly with [3H]diazepam receptor density (r=–0.86, P<0.0005; Fig. 2 and Table 2). Propofol’s in vivo anaesthetic effects also correlated significantly with [3H]flunitrazepam binding sites (r=–0.79, P<0.005; Table 2). The fact that propofol’s effects correlated with both sets of benzodiazepine receptor density data is not too surprising given the strong degree of correlation between them.
In contrast, the regional metabolic reductions evident during isoflurane anaesthesia did not correlate significantly with the known distribution of human benzodiazepine receptor densities for either diazepam (Fig. 3) or flunitrazepam (Table 2). Isoflurane’s effects did, however, show a positive correlation with muscarinic binding sites (Fig. 4 and Table 2). Because the data are plotted as percentage decreases, it is important to explicitly state that a positive correlation means that brain metabolism during anaesthesia decreases less in those brain regions with more receptors. Conversely, a negative correlation means that brain metabolism during anaesthesia decreases more in those brain regions with more receptors.
Another difference between propofol and isoflurane emerged in the correlation with [3H]naloxone (opioid) binding sites. Propofol’s regional effects showed a weak significant negative correlation with naloxone receptors, whereas there was a positive trend to the relationship between those receptors and the effects of isoflurane (Figs 5 and 6).
These data show that a relationship exists between the regional cerebral metabolic reduction effects of propofol anaesthesia in humans and the underlying regional distribution of human benzodiazepine ([3H]diazepam and [3H]flunitrazepam) receptor densities. The presumed GABA agonist propofol is thought to produce the anaesthetic state by increasing an inhibitory chloride conductance, which is mediated through the GABAA complex,10 in a subunit dependent manner.22 In our data, propofol was found to be inhibitory (i.e. depressed regional cerebral metabolism) in all brain regions examined.16 As the GABAA complex is thought to be the effector site for benzodiazepine binding, our findings offer a link between the functional metabolic (i.e. anaesthetic) effects of propofol in the human brain with its presumed cellular (i.e. GABAA complex mediated) mechanism of action. Thus, those brain areas with more benzodiazepine receptors show a greater decrease in regional cerebral metabolism during propofol anaesthesia than do those brain areas with fewer receptors. The most logical interpretation for this effect is to suggest that the benzodiazepine binding site on the GABAA complex is likely to be strongly regionally co‐localized with the GABAergic site that mediates propofol’s in vivo effects on brain metabolism. Furthermore, as the GABA receptor is only potentiated by propofol if a gamma subunit is present, it would be fair to say that the correlation found between propofol and benzodiazepine receptor densities is with a subset of GABA receptors (N. P. Franks, personal communication). This subset distinction explains why it is possible to have found a correlation with the diazepam and flunitrazepam binding sites without also having found a correlation between propofol’s regional effects and GABA receptors, per se, as assessed with [3H]GABA binding.
In contrast with the findings with propofol, the pattern of regional cerebral metabolic reductions produced by isoflurane was not at all consistent with a GABAergic mediated mechanism for the production of the anaesthetic state associated with isoflurane. The most obvious interpretation of this finding might be to simply conclude that the in vivo regional cerebral metabolic effects of isoflurane are not mediated through a GABAergic mechanism. This would imply that the limited in vitro effects isoflurane does have on GABAergic transmission probably do not play a significant role in producing the anaesthetic state associated with isoflurane. However, for a number of reasons (discussed below), including the low statistical power of n=5, and the fact that a correlation reveals nothing about causation, the present findings can not rule out the possibility that a GABAergic mediated mechanism still does underlie the anaesthetic state produced by inhalation anaesthetics (including isoflurane).
Isoflurane may have failed to show a relationship between benzodiazepine ([3H]diazepam or [3H]flunitrazepam) receptor density and regional cerebral metabolic changes, because isoflurane’s effects on regional metabolism may simply occur at a molecular site on the benzodiazepine receptor, which is functionally different from that of propofol’s. Early evidence for this being a possible explanation of our correlational results appears to be emerging from in vitro work using site‐directed mutagenesis of human GABAA receptors. Krasowski and colleagues report that specific mutations can abolish regulation of the GABAA receptor by isoflurane without affecting propofol mediated receptor interactions.23 Alternatively, it is possible that isoflurane could act on the same molecular site as does propofol, but isoflurane could inhibit the site in vivo in a functionally different manner than does propofol. This ‘functionally different manner’ might not lead to the same degree of regional metabolic suppression for each region as that seen with propofol. Furthermore, it is possible that isoflurane might act in vivo at the same site as does propofol, and in a similar manner to that of propofol, but its effects on regional metabolism might be masked or complicated by its actions within other receptor systems. In other words, the overall effects of isoflurane on regional cerebral metabolism might simply represent a less ‘pure’ picture of regional metabolic effects/receptor interactions than what was found with propofol, as a prototypical GABA agonist. Therefore, simply because this analysis did not reveal a relationship between the regional metabolic changes produced by isoflurane and regional benzodiazepine binding, one cannot rule out an important role for the GABAA receptor in the mechanism of isoflurane anaesthesia (i.e. ‘absence of evidence is not evidence of absence’). In fact, other functional brain imaging evidence is emerging which suggests that the GABAA receptor is modulated, at least to some extent, by isoflurane in vivo.24
The correlation found between isoflurane’s effects on regional cerebral metabolism, and the regional distribution of muscarinic acetylcholine receptors was somewhat unexpected, but nonetheless, appears reasonable. The inverse nature of the correlation between muscarinic receptors and the regional cerebral metabolic changes caused by isoflurane anaesthesia suggests that regional cerebral metabolism decreases less in those brain areas during isoflurane anaesthesia, which have more muscarinic receptors. Muscarinic signalling in the central nervous system is known to be intimately involved with regulating level of arousal and consciousness, and cholinergic signalling systems tend to enhance wakefulness.25 Thus, our findings with isoflurane would seem to imply that brain areas with higher levels of muscarinic receptors are more resistant to the metabolic depressant effects of isoflurane inhalation anaesthesia and may be, in essence, harder to ‘anaesthetize’ with isoflurane.
Unfortunately, because a unified picture of nicotinic receptor binding data remains lacking in humans, the regional cerebral metabolic changes seen with isoflurane and the regional distribution of nicotinic acetylcholine receptors were not directly compared in the present study. Such a comparison might offer important insights into the mechanism of anaesthesia as the nicotinic acetylcholine receptor has been shown to have binding site properties consistent with a potential target site for general anaesthetic action,26 and is known to be very sensitive to clinically relevant doses of inhalation agents.27
The correlation of propofol with naloxone receptors is interesting and might seem to imply that propofol ought to have some opioid‐like activity. However, a more likely explanation of this correlation is to suggest that some degree of regional co‐localization can exist between the various receptors examined. For example, the propofol cerebral metabolism changes, also, weakly trended the serotonin receptor sites. This relationship was not completely unexpected as the naloxone and serotonin binding densities were noted to have a strong correlation with each other across the nine comparable regions examined (r=0.76, P<0.005). Additionally, for the propofol naloxone correlation it should be noted that a large part of the correlation is driven by the one value associated with the frontal lobe. Eliminating this one frontal lobe value as an outlier, or examining this relationship with a non‐parametric (Kruskal–Wallis) statistical approach reduces the correlation value to a non‐significant level (P<0.14). In contrast, using the same non‐parametric statistical approach with the benzodiazepine ([3H]diazepam) receptor density data, the correlation with propofol’s regional cerebral metabolic reduction effect still remains significant (P=0.038).
Whereas these data do support the importance of propofol’s GABAergic effects in producing the anaesthetic state in humans and suggest an involvement of muscarinic signalling with isoflurane anaesthesia, some limitations with the present methodology used in these exploratory analyses need to be considered. Most problematical is the region‐of‐interest selection. We attempted to match regions as reported historically; however, there is no standard definition of a particular region‐of‐interest. For example, ‘caudate’ may mean the whole caudate (including the tail), or just the head of the caudate. We took ‘caudate’ to mean the head of the caudate as this is a large enough brain structure to obtain a reliable PET signal from. With the requirement for similar assumptions to be made before analysis in each region‐of‐interest examined, some inaccuracies are likely in the present comparisons.
Additionally, the historical data may contain some element of sampling error. For example, in some cases, with the Braestrup data the values for some regions‐of‐interest, such as the corpus callosum, represent the measurement of only two subjects.13 However, this is much less likely a problem for the data of Zezula and colleagues. Their findings were obtained in twenty‐one post‐mortem brains. They reported only limited variability from case to case; ‘This variation was, however, limited and in most areas the standard error of mean (sem) did not exceed the 10% of the mean densities’.14 Given the strong correlation between the benzodiazepine binding site density studies, with an r=0.93, it seems unlikely that either investigator was that far off the mark with their ex vivo quantification attempts.
Another limitation is that the number of comparable regions for some of the receptor systems, like the muscarinic system, is small. At best, any conclusions regarding a single agent’s correlation with a single receptor system must be tempered in caution. Nonetheless, the power of the present exploration does not come from any single correlation rather it comes from the fact that the two different agents were compared using the same, albeit potentially flawed, set of rulers. Thus, the most important findings would seem to be related to those comparisons that show one effect with one agent and a different effect with the other agent. In this sense, the relationship between benzodiazepine binding sites and the regional cerebral metabolic effects of propofol is made more interesting because this type of relationship did not occur with isoflurane. Conversely, the relationship between isoflurane and muscarinic receptors is interesting because it did not occur with the propofol. Furthermore, the relationship of both agents with the opioid receptors becomes interesting because it differs between the two agents. It is positive for propofol and negative for isoflurane. Any ultimate explanation for the mechanism of anaesthesia will have to be able to account for these agent‐specific differences in regional cerebral metabolic effects.
Another potential concern for interpreting these data is that we used historical ex vivo human data instead of in vivo data. Ex vivo data may not compare favourably with the real in vivo situation for some of the receptors examined. This is less of a concern for the benzodiazepine receptor; however, as the benzodiazepine receptor appears to be quite stable post‐mortem28 and is unaffected by age, gender, post‐mortem delay or pre‐mortem drug treatment.14 We avoided the in vivo PET data on benzodiazepine binding studies because the semi‐quantitative nature of the in vivo studies performed to date has produced inconsistent results with only limited areas presented for examination. Nonetheless, many of the receptor systems examined in this study could be examined with PET, in vivo. Most importantly, measurement of regional benzodiazepine binding with flumazenil is now quite common. The next logical step in elucidating the relationships suggested in the present study would be to compare benzodiazepine binding sites with propofol’s regional cerebral metabolic effects in vivo using PET in the same set of volunteers. This would allow for more direct comparisons involving the relationship between cerebral metabolic effects and regional receptor systems without confounds associated with region‐of‐interest selections.
In summary, given the assumptions required to make the present comparisons, it is notable that a simple linear relationship between benzodiazepine ([3H]diazepam and [3H]flunitrazepam) binding sites and propofol’s effects on regional cerebral metabolism would be found. It has been proposed that changes in chloride based neuronal inhibition (mediated through the GABAA receptor complex) could be the underlying mechanism of general anaesthesia.45 Our findings with propofol are consistent with that hypothesis; however, our findings with isoflurane inhalation anaesthesia suggest that the anaesthetic state caused by isoflurane may involve a more complex mechanism.
The authors acknowledge N. P. Franks, PhD and W. R. Lieb, PhD for helpful review of the manuscript.
|Brain||[3H]diazepam13 region (pmol mg protein–1)||[3H]flunitrazepam14 (fmol mg protein–1)|
|Brain||[3H]diazepam13 region (pmol mg protein–1)||[3H]flunitrazepam14 (fmol mg protein–1)|
|Receptor system||(Number of comparable regions)||Propofol (r‐value) (n=8)||Isoflurane (r‐value) (n=5)|
|Receptor system||(Number of comparable regions)||Propofol (r‐value) (n=8)||Isoflurane (r‐value) (n=5)|