Functional Brain Imaging during Anesthesia in Humans : Effects of Halothane on Global and Regional Cerebral Glucose Metabolism

The methods of quantitative PET imaging and our standard imaging procedures have been described before. [2,8]Briefly, the study of cerebral glucose metabolism with PET uses the same principles as the deoxyglucose autoradiography technique developed by Sokoloff et al. [9]To study regional cerebral glucose utilization in humans, a positron-labeled deoxyglucose tracer is used, such as FDG. [10]This tracer is taken up by active brain neurons as if it were glucose. However, because of its structure, it cannot be metabolized fully and, therefore, becomes metabolically trapped as FDG-6-phosphate within the intracellular compartment. The amount of radioactive label that eventually remains in each discrete region of the brain is related to the glucose uptake and metabolism of that discrete region. Uptake of FDG and metabolic trapping of FDG in the brain as FDG-6-phosphate is 80% to 90% complete at 32 min. [11]The subsequent PET scan images obtained represent the accumulated regional FDG uptake that occurred during the corresponding radiotracer uptake period.

After we obtained full institutional review committee approval and five adult right-handed male volunteers gave informed consent, each of them underwent two separate PET scan procedures at least 1 week apart. One scan assessed baseline cerebral metabolism associated with the awake state, and the other scan assessed cerebral metabolism associated with unconsciousness induced by halothane anesthesia. The volunteers were healthy nonsmokers, with a mean age of 22 +/− 3 yr. Each was classified as American Society of Anesthesiologists physical status 1 and none had evidence of previous psychiatric problems. The volunteers avoided caffeine and other medications for at least 48 h before each scan, they fasted at least 8 h before each scanning session, and they received an oral antacid (30 ml sodium citrate by mouth) before scans involving anesthesia. Each volunteer had two intravenous catheters inserted, one to administer the FDG-PET tracer and one to sample blood. Blood samples were taken to quantify the uptake of FDG (see Phelps et al. [11]). Monitoring equipment used included a three-lead electrocardiograph, an automated noninvasive blood pressure monitor, a pulse oximeter, end-tidal carbon dioxide monitor, a temperature monitor, and a precordial stethoscope.

Halothane was administered via a tight-fitting face mask using a calibrated vaporizer through a semiclosed non-rebreathing circuit. The end-tidal halothane concentration was monitored using a Poet II agent analyzer (Criticare Systems, Milwaukee, WI). Halothane was administered incrementally in 0.1% expired steps adjusted upward every 5 to 10 min to achieve loss of consciousness with no response to mild prodding. Once loss of consciousness was achieved, the end-tidal halothane concentration was fixed for the rest of the experiment. Airway instrumentation was not used, and the volunteers maintained spontaneous ventilation during anesthesia, although on rare occasion simple assistance (head tilt and jaw thrust) was needed to maintain a patent airway. As the volunteers became nearly unresponsive, the eyelash reflex was tested every 3 min, and they were asked to open their eyes until they no longer followed commands. When the volunteers no longer responded to verbal commands, they were stimulated further by mild prodding and shaking. Loss of consciousness was defined as unresponsiveness to both verbal and tactile stimuli. After the experiment, once halothane was discontinued, the volunteers opened their eyes and were responsive at 6 +/− 2 min (mean +/− SD), and they moved to the scanner with assistance at 14 +/− 3 min.

For the awake control scans, the volunteers lay quietly on a gurney with their eyes closed. Responsiveness to verbal stimulation was assessed occasionally while cerebral metabolism was measured, depending on clinical signs (as many as five times in one volunteer). All the volunteers remained responsive to verbal stimulation during the 32-min uptake period. After the experiment, no volunteer reported any subjective episodes of spontaneous sleep during baseline metabolism assessment.

The volunteers were given anesthesia, as above, while they were in a small darkened, sound-shielded room. Cerebral metabolism was measured during each anesthetic once the volunteers became unresponsive and clinically approached a steady state level of anesthesia (i.e., no changes in heart rate, breathing pattern, or blood pressure occurred for at least 12 min before isotope injection). Once they were stable under the desired conditions, 5 mCi FDG was injected intravenously. The volunteers remained at the targeted level of anesthesia for the next 32 min. After the brain was labeled with the positron-emitting tracer, the anesthetic was discontinued. The volunteers were allowed to emerge from the anesthetic and regain awareness before being taken to the PET scanner. They completed their recovery from the anesthetic while they were in the PET scanner.

For the baseline condition, a similar labeling and scanning sequence, as outlined above, was followed. Scanning of all subjects began within 20 min of the end of each uptake period for each condition. The time between injection of the FDG and the start of each scan itself was standardized across conditions to ensure that it was similar for all volunteers. Those in the awake and anesthetized conditions passively listened through headphones to a prerecorded audiotape of repeated words. The ability of the volunteers to recall the words heard during anesthesia and how such recall correlates with cerebral metabolism will be reported elsewhere.

Because halothane anesthesia is expected to increase global cerebral blood flow, and some degree of hypoventilation might occur in these spontaneously breathing volunteers, which might also increase the global cerebral blood flow, it is conceivable that the resultant increases in global cerebral blood flow could be great enough to confound interpretation of the absolute glucose metabolic rate values eventually obtained. Thus, to quantify the magnitude of this potential confounding factor, middle cerebral artery blood flow velocity (V^MCA) was assessed noninvasively using a MedaSonics transcranial Doppler system (Medasonics, Newark, CA) at baseline and immediately before injection of the FDG radioisotope. Changes in V^MCAare thought to represent reasonable estimates of the changes in global cerebral blood flow caused by anesthesia. [12]The V^MCAwas measured by insonating the left middle cerebral artery through the temporal window, using a 2-MHz pulsed transcranial Doppler probe. Doppler signals from the left MCA were identified and measured at a depth of 45 to 55 mm. The maximum shift in frequency spectra (spectral outline) of the Doppler signals was converted to mean V^MCAby built-in computer software and averaged over four or five cardiac cycles.

The PET scans were done after each uptake period using a GE 2048 head-dedicated scanner (GE/Scanditronix, Stockholm, Sweden). Two sets of 15 image planes, resulting in 30 PET images across the whole brain, were obtained for each volunteer. The PET scanner has a resolution of 4.5 mm (full-width half-maximum) in plane and 6 mm axially. Scans were obtained in relation to the canthomeatal line. The volunteers were positioned using laser guidance, and a thermosetting plastic face mask was used to hold each volunteer's head stationary during image acquisition. In vivo attenuation correction was obtained by previous transmission scanning using a (⁶⁸) Ge/(⁶⁸) Ga) rod source.

To determine absolute whole-brain and regional glucose metabolism, the scans were transformed into glucose metabolic rates (GMR) and relative GMR (rGMR), as previously described. [13]Glucose metabolic rate values (mg [middle dot] 100 g^-1[middle dot] min^-1) were calculated using the deoxyglucose kinetic models of Sokoloff developed for autoradiography in animals and modified for humans. [9-11]Regions of interest were located using stereotaxic coordinates derived from a standard neuroanatomic atlas. [14]Regions selected and subsequent analyses are comparable to our earlier propofol and isoflurane studies. [2,3]However, to allow for direct comparison of relative metabolic differences between the propofol data and the inhalational data, before the region-of-interest analysis for these comparisons, the effective resolution of the inhalation data was degraded using a Gaussian kernel that approximated the lower resolution of the older scanner on which the propofol data were obtained. Differences in the means of whole-brain glucose metabolic rates, various regions of interest, and physiologic variables were compared using paired (two-tailed) t tests. Absolute glucose metabolic data from each region of interest were treated as independent variables. To avoid type 1 errors, we considered these absolute regional metabolic effects significant only for P <or= to 0.01. Region-of-interest comparisons across studies were assessed using one-way analysis of variance, with Bonferroni-Dunn correction for multiple comparisons, with P < 0.05 considered significant.

To determine if the pattern of relative metabolic activity changed significantly within the brain during halothane anesthesia, the data were analyzed using statistical parametric mapping (SPM96) software from the Wellcome Department of Cognitive Neurology, London, United Kingdom, implemented in Matlab (Mathworks, Sherborn MA). [15-17]This process determined regionally significant condition effects for every pixel in a standardized space. This process involved several steps. (1) The data were reconstructed in three-dimensional space. (2) The intercommissural (anterior commissure-posterior commissure) line was identified by an automated routine, and the three-dimensional images were rotated on axis to fit a reference template. A least-squares approach was used to estimate the six parameters of this rigid body transform. [18](3) After realignment, all images were transformed into a standardized space (according to the atlas of Talairach and Tournoux [19]) using nonlinear transformations by matching each scan to the reference template image. [20](4) To increase the signal-to-noise ratio, and to reduce the effect of variable functional anatomy, the images were smoothed using an isotropic (10-mm full-width at half-maximum) Gaussian kernel. (5) Finally, global differences in glucose metabolic rates were normalized across conditions and volunteers using proportional scaling. This correction ensures that variations in activity caused by differences in global metabolic rates among the volunteers and between the conditions did not obscure the relative regional changes caused by the anesthetic itself.

Comparisons of regional relative glucose metabolism were performed between conditions on a pixel-by-pixel basis using t statistics. Regions where pixels reached probability values less than 0.01 were considered significant. To reach significance and to provide protection against type 1 errors, pixels also had to be in a contiguous group extending over more than one transverse plane. [21]The resulting set of pixel values constituted a statistical parametric map of the t statistic SPM{t}. The SPM{t} was transformed to the unit normal distribution (SPM{Z}) and thresholded. The results are displayed as a three-dimensional volume of pixels in coronal, transverse, and sagittal views of the brain.

The expired end-tidal halothane concentrations that induced unresponsiveness in the volunteers ranged from 0.5% to 1% and averaged (+/−SD) 0.7 +/− 0.2%. Table 1shows the physiologic changes produced by this level of halothane. A significant increase in the respiratory rate occurred during halothane anesthesia (Table 1), as did a decrease in mean blood pressure. For each volunteer, the V^MCAtended to increase during halothane anesthesia. However, this effect was not significant on group analysis.

Awake whole-brain GMR averaged 6.3 +/− 1.2 mg [middle dot] 100 g (^-1)[middle dot] min^-1(mean +/− SD). Halothane anesthesia significantly reduced whole-brain GMR to an average of 3.7 +/− 0.6 mg [middle dot] 100 g^-1[middle dot] min^-1(by the two-tailed paired t test, P <or= to 0.005). Halothane significantly decreased absolute regional glucose metabolism in all brain areas studied (Figure 1). No evidence of any regional metabolic increases were found in any brain regions, for any volunteer, during halothane anesthesia. The magnitude and variability of the mean 40 +/− 9% whole-brain metabolic reduction caused by halothane anesthesia can be appreciated in Figure 2. The individual variation in the percentage decrease of whole-brain metabolism caused by halothane anesthesia ranged from 28% to 53% for the five volunteers. Figure 1and Figure 2show that each volunteer had a generalized global cerebral metabolic reduction during anesthesia.

The SPM Decreases (i.e., Areas Most Suppressed by Halothane). Analysis of the regional rGMR effects of halothane showed significant shifts in the pattern of rGMR evident during halothane anesthesia (Figure 3). The Figure showsthe extent of the area of pixels that were significantly decreased in rGMR during halothane anesthesia compared with the awake baseline condition (P < 0.01). Table 2shows the specific anatomic locations and Talairach coordinates for these effects. The magnitude of the regional shifts in rGMR that occurred during halothane anesthesia represent changes of approximately 2% to 4% from baseline.

A trend toward decreased metabolism was noted in several other neuroanatomically interesting brain regions not listed in Table 2. In the medulla, the region around the periaqueductal gray and adjacent areas, such as the dorsal motor nucleus and the solitary nucleus, appeared suppressed. In the pons, the dorsal- and dorsolateral pontine region that encompasses the locus coeruleus, peribrachial nucleus, periaqueductal gray, and caudal dorsal raphe seemed most affected. For the diencephalon, suppression also appeared to include regions of the hypothalamus. In the telencephalon, a continuous band of medial-rostral-dorsal activity through the basal forebrain appeared to show lower rGMR activity. This particular cluster included the substantia innominata, ventral striatum, ventral pallidum, and extended amygdala.

The SPM Increases (i.e., Areas Least Suppressed by Halothane). Regions least depressed by halothane anesthesia were dorsal neocortical areas and included the frontal association cortex involving Broca's area 8, dorsal somatosensory cortex, and superior parietal cortex involving Broca's area 7 (data not shown).

(Figure 4) compares rGMR in regions of interest for halothane, propofol, and isoflurane.

Our results show that halothane anesthesia titrated to a point just beyond the loss of responsiveness in humans is associated with a fairly global 40 +/− 9% reduction of whole-brain glucose metabolism and is not associated with any regional metabolic increases in any brain regions of any volunteer. Nonetheless, the whole-brain metabolic reduction produced by halothane anesthesia is associated with significant shifts in the pattern of regional brain glucose utilization with specific metabolic suppression in brain regions of the thalamus, basal forebrain, cerebellum, occiput, and limbic system.

Shapiro et al. [5]found that halothane anesthesia caused significant decreases in local cerebral glucose metabolism in the baboon. Those regions that were significantly reduced included the occipital cortex, periaqueductal gray matter, reticular formation, inferior colliculus, anterior commissure, cerebellar cortex, and cerebellar white matter. Shapiro et al.'s regional results correspond closely with our current findings. However, in contrast with our results, Shapiro et al. also noted increased absolute glucose metabolism in several brain structures during halothane anesthesia.

Somewhat unexpectedly, therefore, the hope of finding a uniquely interesting “metabolic footprint” pattern of activity with halothane did not materialize. In fact, comparison of the current PET data with the PET data from our previous methodologically similar PET studies of propofol and isoflurane anesthesia reveals that the relative regional changes produced by halothane anesthesia were nearly identical to those found before with isoflurane (Figure 4). [2,3]Indeed, the only regional difference that approached significance between the two inhalational agents was found in the cerebellum, where halothane tended to suppress metabolism more than isoflurane (P < 0.05, uncorrected). In contrast, propofol tended to suppress relative metabolism in the cerebral cortex, especially in the temporal and occipital regions, more than the inhalational agents (Figure 4). In addition, compared with halothane, propofol caused significantly less metabolic suppression in the basal ganglia (caudate + putamen) and midbrain regions.

We hypothesize that these minor variations in regional neuronal effects may be related to the underlying cellular mechanisms through which the different agents work. Propofol has a putative GABAergic mechanism of action, [22]whereas the inhalational agents affect not only GABAergic transmission but also many other receptor systems. [1]Therefore, the effects of propofol on regional cerebral metabolic activity might be explained by simply noting how closely its regional cerebral metabolic effects map to its underlying putative GABAergic receptor sites. [23] 

In addition, clues to the cellular mechanism(s) through which the inhalational agents produce their effects on regional brain metabolism might be found with similar analyses. [24]The large decrease in relative midbrain metabolism evident during inhalational anesthesia is intriguing and may represent the metabolic component of the fact that inhalational agents significantly reduce midbrain levels of acetylcholine. [25]The important role of acetylcholine in excitatory neurotransmission, coupled with the fact that central nicotinic receptors appear sensitive to inhalational anesthesia, [26]suggests that suppression of acetylcholine neurotransmission may play a role in mediating the effects of inhalational agents on consciousness.

Halothane reduced whole-brain metabolism 40 +/− 9%. Isoflurane and propofol were shown before to reduce whole-brain metabolism 46 +/− 11% and 55 +/− 13%, respectively, at a similar clinical end point. [2,3]The magnitude of the whole-brain metabolic reduction seen during these three dissimilar anesthetics, coupled with the recent demonstration that the individual magnitude of the cerebral metabolic reduction caused by propofol or isoflurane anesthesia is correlated with the magnitude of the changes seen in concurrently measured values of various electroencephalographic descriptors, [27]suggests that the percentage of absolute cerebral metabolic reduction of whole-brain glucose utilization caused by anesthesia may be an important component of anesthetic depth related to loss of consciousness. If this is true, the relation previously identified between the EEG and the percentage of absolute cerebral metabolic reduction should continue to hold true when updated with the addition of the current halothane scans. Indeed, the reported correlation between the bispectral index and the percentage of absolute cerebral metabolic reduction remains significant with inclusion of the current halothane data (r =-0.75, P < 0.001, n = 18 scans). [27] 

The individual PET scan data are shown in Figure 2to emphasize several points. (1) It is not known why, but each person starts with a different baseline cerebral metabolic rate. This difference can be as much as two, or threefold among individuals. (2) Although we might hypothesize that a more “metabolically active” brain would be harder to anesthetize than a less metabolically active one (or vice versa), no evidence for this type of relation has emerged. (3) Furthermore, no obvious relation has emerged between a person's anesthetic dose at loss of consciousness and the eventual magnitude of the cerebral metabolic reduction caused by the anesthetic for that person. (4) The only relation that has emerged between cerebral metabolic variables and “anesthetic depth” variables is the relation noted before between electroencephalographic changes and the percentage of absolute cerebral metabolic reduction. [27] 

In this study, a 1 minimum alveolar concentration dose of halothane was necessary to produce unresponsiveness. This compares with a 0.5 minimum alveolar concentration dose previously found for isoflurane. [3]This finding suggests that the anesthetic effects on brain metabolism may not necessarily be equivalent if based solely on minimum alveolar concentration dosing. Indeed, this apparent relation between the minimum alveolar concentration dose and the magnitude of cerebral metabolic reduction produced by halothane versus isoflurane was observed before. [6] 

Examination of the transcranial Doppler data shows that, on average, a 20% increase in blood flow velocity occurred for each volunteer during halothane anesthesia. This level of blood flow velocity change caused by halothane was not unexpected. [6]It is doubtful that a change of this magnitude altered the kinetics of glucose metabolism enough to influence drastically the metabolic rate values reported here. However, further study, such as measuring regional cerebral blood flow with the O-15 PET technique before metabolism assessment, is needed to clarify this issue fully.

In conclusion, the anesthetic state evident during halothane anesthesia titrated to the point of unresponsiveness is associated with a global reduction in brain glucose metabolism and causes significant regional shifts in the pattern of relative metabolism. Comparison with previous results shows that halothane and isoflurane have similar effects on absolute and relative cerebral glucose metabolism, whereas the effects of propofol differ somewhat from those of the inhalational agents.

The authors thank Lisa Smalling for assistance with transcranial Doppler measurements.