Anesthetics may affect the regional cerebral blood flow (rCBF) response associated with increased brain activity in humans. rCBF was measured as auditory stimulus rate was increased during propofol and thiopental administration.
After informed consent, 10 right-handed male volunteer participants (aged 33.5 +/- 10.4 yr, weighing 74.5 +/- 8.4 kg) received thiopental (n = 4) or propofol (n = 6) intravenously at stepwise target concentrations of propofol 1.2 and 2.5-3, or thiopental 4 and 7-9 microg/ml, representing sedative and hypnotic drug concentrations. The latter made volunteers unresponsive to voice or mild stimulation. Quantitative positron emission tomographic brain images were obtained at 0, 20, and 40 auditory words per minute at each drug concentration. Using SPM99 analysis, 10-mm spherical regions of interest were identified by peak covariation of word rate with rCBF across all conditions and drug concentrations. Individual mean rCBF responses in these and primary auditory cortex (Heschl's gyri) were obtained.
Significant increases in rCBF with auditory word rate occurred in temporal lobes bilaterally at baseline (significance, T = 4.95). There was no change in this response during sedation (T = 5.60). During unresponsiveness seven of 10 participants had a diminished response in the left temporal lobe (T = 3.18). Global CBF, corrected for changes in PCO2 (3% .mmHg PCO2), was reduced 15% by sedation and 27% during unresponsiveness.
The presence of propofol or thiopental does not affect the rCBF response to increasing stimulus rate during consciousness. Thus, changes in rCBF activation patterns with sedative concentrations of these drugs represent effects on brain activity itself. The neuroanatomical targets of drug effect on memory and attention may be revealed by changes in rCBF patterns associated with these cognitive activities.
RECENT imaging studies using both functional magnetic resonance imaging and positron emission tomography (PET) techniques demonstrate regional effects of many anesthetic agents on the brain. For example, Heinke et al. 1found that isoflurane affected the brain in very specific locations during a visual stimulus paradigm. Similarly, we have shown that propofol decreases regional cerebral blood flow (rCBF) in regions that demonstrate rCBF increases during a memory task.2These neuroimaging findings support behavioral evidence that the effects of anesthetics on the brain are specific rather than global in nature. The effects of propofol or midazolam on episodic memory are best examined against a drug that produces memory effects on the basis of sedation, such as thiopental. This is because propofol and midazolam will also produce sedation. This comparative approach allows the nonsedative effects of propofol and midazolam on episodic memory to be differentiated from their sedative effects.3The fact that this is possible indicates that these drugs have specific actions on the brain that produce their memory effects.4Thus, possible mechanisms of drug action on episodic memory and attention processes may be revealed by the changes in rCBF patterns associated with these functions during drug effect.
Recently Mintzer et al. 5have demonstrated such a change in rCBF patterns during a memory task in a group of nine volunteer participants receiving triazolam. The rCBF changes that occur with the memory task used in their study are well described and consist of activations on the left side of the brain.6,7Changes in these patterns with triazolam revealed specific locations of drug effect. Whether these represent the action of triazolam on memory processes themselves or some other effect on brain physiology is unclear. The purpose of this study is to help resolve such questions by examining the effect of propofol and thiopental on rCBF responses to increasing rates of auditory stimuli.8The basic physiology of the brain’s response to increased stimulus frequency has been well described and consists of an increase in neuronal activity, metabolism, and ultimately rCBF.9Studies in humans have demonstrated such a relation between auditory word stimulus rate and rCBF or blood oxygen concentration dependent (BOLD) response in the temporal lobes.8,10Interestingly, the BOLD signal demonstrates a more complex nonlinear relationship with stimulus rate,10which may indicate the effect of unknown modifying influences when functional magnetic resonance imaging is used.
As the drug effects on episodic memory occur at sedative doses, we examined the effect of these doses of propofol and thiopental on rCBF response. In addition, there is ample evidence in the animal literature that neuronal activity related to stimulation is still present during even deep degrees of anesthesia.11In humans, a number of studies demonstrate some learning of auditory material presented during clinically adequate degrees of anesthesia.12–14Thus, we also chose to investigate the relation of rCBF with stimulus rate in unresponsive participants receiving hypnotic doses of the study drugs. As significant changes in rCBF associated with cognitive paradigms are measurable by PET imaging in groups of ∼10 participants, this was the sample size studied in the current investigation.15,16Three conditions were tested: no auditory stimulation and words presented at a rate of 20 and 40 words per minute. The rCBF response during these conditions was measured at three degrees of drug effect: baseline, sedative, and hypnotic concentrations. By necessity drug concentrations were administered in this order to avoid carryover effects of a previous concentration of drug.
Materials and Methods
This investigation was approved by the Hospital Institutional Review Board and Radiation Safety Committee of Memorial Sloan Kettering Cancer Center. Informed consented was obtained in writing before accrual of subjects took place.
Ten healthy normal male volunteers were recruited through newspaper advertisements and paid for their participation. Subjects were screened with telephone and subsequent in-person interviews. Exclusion criteria included use of psychoactive medication, history of recreational drug abuse, head trauma resulting in loss of consciousness, neurologic, cardiovascular, or respiratory disease, claustrophobia, hypertension, peripheral vascular disease, hearing deficit, carpal tunnel syndrome, allergy to eggs, or family history of schizophrenia or acute intermittent porphyria.
Four PET scans were obtained at each drug concentration for a total of 12 scans. Every 12 min, 10 mCi of H215O was delivered intravenously at a constant rate over 20 s via an infusion pump. Scans were obtained on a GE Advance scanner (GE Medical Systems, Waukesha, WI) in the three-dimensional “septa out” mode. The resolution of the PET camera in this mode is approximately 5.2 mm in all dimensions. Three 30-s frames were obtained during each scan, corresponding to the highest rate of uptake of tracer into the brain. A single 10 min transmission scan using a rotating rod of 68Ge/68Ga was performed before scanning commenced to correct for attenuation of signal in its passage through bone and cerebral tissue. The images were reconstructed using filtered back projection and standard clinical protocols and stored as “counts” images (counts of coincidence events expressed as nCi/ml). To construct an approximation of the arterial input function to the brain, arterial blood was sampled from the left radial artery through a quantitative radioactivity counter (GE FRQ, GE Medical Systems) starting 20 s before H215O administration. The raw counts data from the PET image were transformed to quantitative CBF data using an autoradiographic method after correction for difference in delay and dispersion of the measured blood-time activity curve, as implemented in the GE Advance Software (GE Medical Systems).
Magnetic Resonance Image Scanning
T1-weighted structural magnetic resonance images were obtained to coregister with individual subject PET data using a 1.5 T GE Horizon scanner (GE, Milwaukee, WI).
Target concentrations were chosen based on previous experience.2,3,17–19Propofol or thiopental was given by intravenous infusion using STANPUMP software**controlling a Harvard22 infusion pump. Propofol was infused using the Schnider kinetic data set. After approximately 10 min, after predicted pseudoequilibration between serum and effect-site concentrations, PET scanning was resumed.
Arterial blood samples were obtained immediately after every scan for blood gas and drug assay. Blood gases were determined in a standard clinical laboratory. Propofol and thiopental concentrations were determined by high performance liquid chromatography with fluorescence detection as previously described.19
Subjects were monitored with electrocardiogram and pulse oximeter. Bispectral Index was measured using a standard clinical montage (Zip-Prep; Aspect Medical Systems, Newton, MA).
Participants were randomized to receive either propofol (n = 6) or thiopental (n = 4). Four PET scans were obtained during baseline drug concentration (no drug). Scanning was performed during 0, 20, and 40 words per minute conditions, and also during a tactile stimulation condition, but data related to tactile stimulation are not reported here.
After baseline imaging, STANPUMP was used to target sedative drug concentrations (thiopental 4 or propofol 1.2 μg/ml), and scanning was repeated. Hypnotic drug concentrations were then targeted (thiopental 7 or propofol 2.5 μg/ml). The sedative concentration was chosen to provide maximal sedative and memory effects but was at a concentration that would allow volunteers to remain responsive or arousable. The ability to achieve this state was variable, as subjects easily fell asleep. The hypnotic concentration was more easily targeted, as a simple criterion of unresponsiveness to touch was tested. If subjects responded, the target concentration was increased by 20% increments. All conditions were counterbalanced between subjects, but the order of these conditions for a given subject was the same at baseline and during drug conditions.
Detailed information was given on study procedures including radiation dose. Tests of handedness (Edinburgh Handedness Inventory)20and vocabulary (vocabulary subtest of Wechsler Adult Intelligence Scale—Revised) were administered, followed by a brief physical examination.
On the study day subjects arrived about 8 am nil per os after midnight. Venous and radial arterial catheters were inserted before transfer to the PET suite. Dextrose 5% half-normal saline at approximately 100 ml/h was administered intravenously. Twelve PET scans were obtained in total. The participant’s total time in the PET scanner was approximately 3–4 h. After the completion of PET scanning, the arterial catheter was removed and the volunteer was returned to the Neuroanesthesia Laboratory, where the intravenous catheter was discontinued after the volunteer was given a light lunch. Participants were discharged home after meeting standard criteria for discharge for ambulatory surgery.
Word stimuli were started approximately 20 s before scanning commenced, and continued until scanning was finished. Subjects wore foam earphones inserted into the auditory canal (EarLink Auditory Systems, Indianapolis, IN). Subjects were screened for normal hearing using the Coren-Hakstian Hearing Screening Inventory before recruitment into the study.21Separate word lists were prepared for baseline and drug concentrations. Two-syllable words from the Toronto Word Pool22(word frequency <100; mean duration 766 ms) were digitized for computer presentation and were presented at 20 words per minute (every 3 s) or 40 words per minute (every 1.5 s). Words were delivered at 80 dB SPL, and subjects were tested for clear and bilaterally equal perception of words before the start of the study. The subjects were instructed to passively listen to the words, and to not perform any task with the words (“let the words float by”). All scans were obtained with eyes closed.
Behavioral and Demographic Variables
Participant related information is presented throughout as median and range (minimum to maximum) because of the small numbers involved. For analyses of these variables, nonparametric tests were employed when indicated.
Statistical Analysis of PET Images: Covariation of rCBF with Word Rate
Quantitative blood flow images were derived off-line from the measured arterial input function and “counts” images using an autoradiographic method, as implemented in GE PET image analysis software (GE Medical Systems). These images were transformed to ANALYZE format for input into standard image analysis software. Statistical analysis was performed using SPM99††implemented in Pro MatLab v. 6.5, release 13 (Mathworks, New York, NY). PET images were coregistered to individual structural T1-weighted magnetic resonance image scans for this analysis.
Images were realigned to the last baseline scan and normalized into Montreal Neurologic Institute brain image space. A 16-mm Gaussian smoothing kernel was used to accommodate interpersonal variations in gyral anatomy and facilitate intersubject averaging with a resultant smoothness of 21.6 × 24.8 × 22.3 (x,y,z) mm. Mean global CBF was normalized to 50 ml·100−1g brain tissue ·min−1. Statistical analysis employed a proportional scaling model, which allows a differing relationship between regional CBF effect depending on global CBF. Word rate and arterial Pco2values obtained at the end of each scan were included in the SPM statistical design matrix as explicit covariates. SPM contrasts were constructed to identify regions of the brain demonstrating covariation of rCBF with the main effect of auditory stimulus rate (fig. 1).
Neuroanatomical labeling of relevant regions was performed using automated anatomical labeling software which interfaces with SPM99.23A region of interest (ROI) analysis was performed using a sphere of 10 mm radius around brain regions demonstrating peak covariation of rCBF with word rate using all scans and a threshold of T = 4.57 (P < 0.05, SPMcorrected for multiple comparisons in brain space). Two ROIs were identified in right and left temporal lobes, located at Montreal Neurologic Institute coordinates −58,−22,2 and 66,−40,14 (515 voxels in each ROI). Each of these spherical regions of interest were modified so that no portion would be outside the Montreal Neurologic Institute brain space. As well, ROIs were defined for primary auditory cortex (Heschl’s gyrus) using the Automated Anatomic Labeling atlas. The mean voxel CBF for each condition and drug concentration over the ROIs in each subject were obtained using MarsBar‡‡and plotted (fig. 2).
Differences in rCBF response with word rate between baseline and sedative drug concentrations were sought by comparing the slopes of the covariation response using a more lenient statistical threshold of voxel-level (P < 0.001, SPMuncorrected for multiple comparisons in brain image space, T = 3.21).
Covariation of rCBF with Pco2was tested at a statistical threshold of voxel-level P < 0.05, SPMcorrected, T = 4.57.
Quantitative Cerebral Blood Flow Analysis
All PET image data were transformed into quantitative images, as described in the section “PET Scanning,” by averaging PET data of the three 30-s frames acquired for each scan. Values at each voxel are the measured CBF in units of ml·100−1g brain tissue·min−1. As we were interested in auditory activation in auditory cortex, mean voxel CBF was measured using MarsBar in gray matter using individual masks created from the T1 structural magnetic resonance image for each volunteer participant using the segmentation algorithm as implemented in SPM99. These values were plotted against the measured Pco2for each scan, and a regression coefficient was determined. All CBF values were corrected to values that would be present at a Pco2of 40. The effect of drug dose (baseline, sedative, hypnotic) and auditory stimulation rate (0, 20 and 40 words per minute) on corrected CBF was tested in a three-way analysis of variance with comparisons of interest performed post hoc using Student t tests with Bonferroni correction.
Ten male subjects participated in this study and had the following characteristics (median, 95% confidence interval): age, 31 (26–41) yr, weight, 73.9 (68.5–80.5) kg, and body mass index, 24.9 (22.7–26.2).
One participant receiving propofol and one receiving thiopental were dozing off during word presentation at the sedation drug concentration. They were responsive to voice command immediately after the end of the scan. Two participants receiving propofol were unresponsive at the sedation drug concentration. Nevertheless, all these participants were included in the sedation drug concentration for analysis. By design, all subjects were unresponsive at the hypnotic dose. Transient myoclonus was noted in one participant receiving propofol in the sedation phase and in two other participants (one propofol, one thiopental) during the hypnotic stage. The degree of myoclonus was not of sufficient extent to require unusual amounts of displacement during realignment of scans, and no activations in motor regions were noted. All participants responded to verbal stimulation within a few minutes after discontinuation of the infusion.
Two regions of brain demonstrated significant covariation with auditory stimulus rate across all drug concentrations (P < 0.05 corrected for multiple comparisons over brain image space, SPMcorrected, T = 4.57). These regions were located in the superior and middle temporal lobes and Heschl’s gyri bilaterally (table 1and figs. 1 and 2). No other regions of brain demonstrated this relation to word rate.
There were qualitative differences between baseline and sedation drug concentrations (fig. 1). In the baseline, covariation occurred principally in the left side of the brain. During sedation, this shifted to the right temporal lobe, at a location more posterior and superior than the left sided activation at baseline (table 2). In both conditions, the peak of activation was posterior to primary auditory cortex (Heschl’s gyrus).
No differences in the slope of covariation of rCBF with word rate were present between the baseline and sedation drug concentrations despite an overall decrease in rCBF in the ROIs. This relationship was true even when tested at a very sensitive and liberal statistical threshold level of voxel-level P < 0.1 uncorrected for multiple comparisons over brain image space (SPMuncorrected, T = 1.29).
There were no regions of brain that covaried with Pco2concentration (increase in rCBF with increase in Pco2) at a statistical threshold level of P < 0.05, SPMcorrected (T = 4.57). It should be noted that this tests for changes in rCBF above and beyond any changes in global CBF (i.e. , regional effects). There is a strong correlation between global CBF and Pco2. Some regions of brain did demonstrate a regional Pco2effect at a statistical threshold level of voxel-level P < 0.001, SPMuncorrected (T = 3.21) (table 3). These regions were primarily in the anterior temporal lobe, distant from the ROIs.
Quantitative Cerebral Blood Flow Analysis
A strong relationship between Pco2and average gray matter CBF was present (fig. 3). The slope of the linear regression for this relationship was 1.15 ml·100−1g brain tissue· min−1·mmHg Pco2−1. This is approximately a 3% change in CBF per mmHg change in Pco2and agrees well with our previous study and the literature.24,25A similar relationship exists between white matter CBF and Pco2with a regression slope of 0.62 ml · 100−1g brain tissue ·−1· mmHg Pco2−1. Corrected gray matter CBF decreased significantly with drug dose (F(2,89)= 10.3, P < 0.001 by analysis of variance), from 44.3 ± 10.5 (SD) ml · 100−1g brain tissue · min−1at baseline to 38.6 ± 6.9 (SD) ml · 100−1g brain tissue · min−1during sedation (P < 0.05 in comparison with baseline, post hoc Student t test) to 33.8 ± 8.4 (SD) ml · 100−1g brain tissue · min−1during unresponsiveness (P < 0.001 in comparison with baseline, not significant compared with sedation by post hoc Student t test). There was no relationship between auditory stimulation rate and quantitative CBF (fig. 4).
Drug Concentrations and Other Parameters
A summary of serum concentration, Bispectral Index, and Pco2values are given in table 4. Sedative drug concentrations (mean and 95% confidence interval) were 1.2 (1.0–1.4) and 4.8 (3.3–6.1) μg/ml for propofol and thiopental respectively. Similarly, hypnotic drug concentrations were 2.5 (2.2–2.8) and 10.6 (7.2–14.6) μg/ml. Pco2values were unchanged from baseline to sedation but increased significantly during the hypnotic drug concentration (P = 0.002 by nonparametric Friedman analysis of variance). Bispectral Index decreased significantly during both sedative and hypnotic drug concentrations (P < 0.05). There were no differences between the two drugs at any drug concentration for these parameters.
The principle finding of this study is that the rCBF response to increasing auditory stimulus rate is unchanged at sedative concentrations of drug, which have been shown by us to have significant memory effects.2,3,17,26A number of studies have demonstrated the preservation of coupling between metabolism and blood flow on a regional basis for various anesthetic agents that either decrease or increase global cerebral blood flow.27–29More relevantly, the relationship between cognitive brain activity and image signal seems to be preserved during sedation with lorazepam or scopolamine at a dose that results in impairment of episodic memory. Sperling et al. 30found no change in the BOLD signal from visual cortex when drug was given, whereas the BOLD signal from hippocampus was inhibited. These findings indicate an action of these drugs on a region of the brain known to be involved with the observed memory effect.
The current study examines more directly the effect of drug on the relation between brain activity and rCBF response. Though neuronal activity was not directly measured, the relation between stimulus frequency and neuronal response has been well characterized.9Information and its transmission in the brain are represented by neuronal spike activity, which is only indirectly measured by rCBF responses. These likely measure changes in synaptic activity associated with underlying brain activity.31rCBF increases are likely to be associated with excitatory brain activity. In this situation spike activity and synaptic activity are closely related.32,33The rCBF response measured in this study represents a summation of transient responses summed over the time of acquisition during imaging.34Thus, with increasing stimulus rate, rCBF will increase linearly. Even when tested by very lenient statistical criteria, no difference in the relation of rCBF to stimulus frequency at sedative concentrations of drug was found. This was true despite a 15% reduction in global cerebral blood flow and probable changes in attention with sedation, which have been shown to modulate this relationship.35There is little likelihood that any differences would be found in larger numbers of participants. Even if present, these would almost certainly be much smaller than the magnitude of rCBF changes induced by cognitive processes, evident in PET studies with approximately 10 participants in each group.5
As may be expected with word stimuli, the region of brain demonstrating peak covariation with word rate was somewhat different from primary auditory cortex. The fact that some automatic language processing occurs with these stimuli is supported by greater covariation on the left side of the brain. Covariation of rCBF with word rate was similarly present in Heschl’s gyri bilaterally. Interestingly, during sedation, the region of peak covariation seems to shift to the right side and becomes larger than at baseline. The number of participants in this study does not allow a test of whether these effects are actually different. If present, some degree of disinhibition on neuronal responses during sedation may account for these observations. We have previously seen this effect both with rCBF and with BOLD measures.
During unresponsiveness, global CBF decreased by 27%, which agrees well with Fiset and colleagues who also measured CBF using O-15 PET imaging during propofol administration.36At hypnotic concentrations in the current study covariation of rCBF with auditory stimulus rate was absent. The paradigm used in the current study is unlikely to critically assess rCBF response to stimulus frequency at degrees of anesthesia past the loss of consciousness, as transmission of sensory stimulation to the cortex is severely diminished or absent as this degree of drug effect.37Supporting this observation is a recent study conducted by Heinke et al. ,38which investigated the BOLD response to auditory sentences at increasing propofol concentrations. BOLD response was absent at concentrations associated with unresponsiveness, though some response was present immediately after the loss of consciousness. As the experimental paradigm used by Heinke et al. resulted in changing concentrations of propofol through the imaging period, it is unknown if this auditory brain response would still be present during unresponsiveness at a stable propofol effect. During natural sleep cortical auditory response is still present, and it may be that with sensitive imaging techniques an auditory response would be discernible with drug-induced unresponsiveness.39,40Interestingly, most participants in the current study did demonstrate a monotonically increasing rCBF response in the left sided region of interest during unresponsiveness in the current study. If the three participants who did not are excluded from analysis, a small area of covarying activation is seen (fig. 1). The rCBF response in these participants is substantially diminished from baseline, as shown in figure 2.
No placebo group was used in this study, as the primary purpose of a placebo group would be to control for the change in rCBF response to stimulation over time. There is some evidence that habituation or diminution of the rCBF response to stimulation occurs with repeated stimulation or over time during a constant stimulus.41The magnitude of this effect is much larger for cognitive activations than with primary sensory stimulation.42In the current study auditory stimulation was present in two of four conditions at each drug concentration. Thus stimulation was repeated at infrequent intervals, with at least 10 min between stimulation conditions. The likelihood that rCBF response would habituate between sedation and hypnotic drug concentrations to the extent that ablation of the response would occur is extremely unlikely. However, if habituation is present, it would tend to increase any differences between baseline and sedation rCBF responses, which were absent in the current study.
Changes in Pco2will alter global CBF dramatically, approximately 3% for every mmHg in current study. This degree of change is very similar that seen in our previous study where midazolam was given.24Although this concentration of drug affected global CBF substantially, no relation between global CBF and stimulus rate was observed. Distinct from the effect on global CBF are regional changes related to Pco2, which can be sought by SPM analysis that normalizes global blood flow changes. In a study by Ito et al. 43where Pco2was independently varied, increased rCBF in the putamen, thalamus and cerebellum, and decreased rCBF in the temporal and occipital cortices were found. In the current study, no statistically significant change in rCBF with Pco2was found, although there may have been a trend for this to occur in the temporal pole and inferior-orbital frontal regions. These potential regional effects of Pco2are distant from those demonstrating covariation of rCBF with word rate in the current study. Thus, we could find no influence of Pco2on the rCBF response to stimulus frequency.
In conclusion, propofol and thiopental in sedative concentrations do not affect the relationship between auditory stimulus rate and the associated rCBF response. Thus, sedative concentrations of drug do not appear to affect the physiology associated with the rCBF response to brain activity. Inasmuch as these processes are the same for cognitive functions of the brain, any changes in rCBF patterns in the presence of propofol or thiopental are indicative of the direct effects of these drugs on the cognitive activity of the brain. Thus, neuroanatomical localization of drug actions on cognitive processes is possible by measuring the interaction of drug effect with rCBF patterns associated with cognitive tasks. These types of studies can begin to elucidate the mechanisms of drug action on these cognitive functions at a basic, neuroanatomical level.
We thank our former research assistants in the Department of Anesthesiology: Anika McPhee, M.A., Avin Lalmansingh, B.A., and Jennifer Parsons, B.A., Research Assistants, Department of Anesthesiology and Critical Care Medicine, Memorial Sloan Kettering Cancer Center, New York, New York, for help in recruitment and data collection. We especially thank our volunteer research participants for their patience and cooperation.