The precise neural mechanisms of propofol anesthesia in humans are still unknown. The authors examined the acute effects of propofol on regional cerebral blood flow (rCBF) using positron emission tomography in patients with severe depression.
In six severely depressed patients (mean age, 55.0 yr) scheduled for electroconvulsive therapy, anesthetic levels were monitored by electroencephalography, and rCBF was serially quantified in the awake, sedated, and anesthetized states. The authors used high-resolution positron emission tomography with 15O-labeled water and statistical parametric mapping 99 for imaging and analysis of the data.
Global cerebral blood flow showed sharp decreases from the awake level during the administration of propofol, decreasing 26.8% in the sedated state and 54.4% in the anesthetized state. Moreover, a dose effect was seen in both parietal cortices and the left lateral prefrontal region with larger regions of relative decrease in rCBF at higher propofol doses. At the higher dose, the values of rCBF in the pulvinar nucleus of the thalamus, the pontine tegmentum, and the cerebellar cortex were also affected. Meanwhile, there were few changes of relative rCBF in the basal frontal lobes during both sedated and anesthetized states.
As in earlier studies using normal subjects, pronounced suppression in rCBF in the brain stem reticular formation, the thalamus, and the parietal association cortex occurred even in severely depressed patients. However, previously reported decreases in rCBF in the basal frontal lobe were absent in depressed patients.
THE metabolic effects of propofol on the cerebral vasculature are reported to resemble those of the barbiturates. 1However, the precise neural mechanisms of propofol anesthesia in humans are still unknown. Positron emission tomography (PET) has been widely used for the determination of regional cerebral blood flow (rCBF) and regional cerebral metabolic rate of glucose (rCMRGlu). Initial studies showed that propofol significantly (by some 40–50%) decreases cerebral metabolic rate of glucose (CMRGlu) and cerebral blood flow (CBF) from their levels in the awake state. 2,3A few PET studies reporting the effects of propofol on CBF and CMRGlu suggested that the patterns of reduction of rCMRGlu and rCBF are not uniform for the whole brain. 4–6It has also been suggested that rCMRGlu is more depressed in the cortical region than in subcortical structures. 4Furthermore, propofol is reported to preferentially decrease rCBF in the brain regions related to the regulation of arousal, memory, and cognition. 4,7However, it is unknown whether propofol affects the same regions even in patients with central nervous system disorders, since most earlier studies were performed in healthy young volunteers. We therefore studied the changes in rCBF with propofol in severely depressed patients who needed electroconvulsive therapy (ECT). We used high-resolution PET in the three-dimensional mode using 15O-labeled water and statistical parametric mapping (SPM-99) 8to clarify which regions of the brain were activated or depressed after propofol administration. It is known that changes in rCBF parallel those in metabolic demand of the corresponding region. 9,10Therefore, we can estimate changes in the activity of specific regions by analysis of rCBF. It is thought that propofol probably produces its effects by depressing certain regions of the brain, and that these regions can be identified by the relative decrease in rCBF beyond changes in global CBF (gCBF). On the other hand, regions showing relative increases in rCBF, i.e. , regions relatively unaffected by propofol, are thought to be identified by fewer decreases or by the absence of change in the rCBF in the face of large decreases in gCBF. This is a neuroimaging study of rCBF changes associated with propofol sedation/anesthesia, and the data regarding the influences of ECT on rCBF will be reported separately. This PET study was performed in the first session of ECT for severe depression.
Materials and Methods
Six inpatients (four men and two women; aged 35–71 yr, mean age = 55.0 yr, SD = 16.1) who fulfilled the Diagnostic and Statistic Manual of Mental Disorders , 4th Edition criteria for a major depressive disorder participated in this study. The mean scores on the Hamilton Rating Scale for Depression were 28.4 ± 10.6 (range, 20–46; range of 7 or less indicates absence of depression). The patients were all right-handed and had no history of serious medical illness, psychotic disorder (except for the current illness), cognitive disorder, psychoactive substance abuse or dependence, or ECT within the past 6 months. Their medication regimens were decreased to the minimum required, namely, lorazepam (1–3 mg), flunitrazepam (2–4 mg), and trazodone (50–150 mg) for anxiety and sleep disturbances. All the patients gave written informed consent before participating in the study, which was approved by the Intramural Research Board of the National Center of Neurology and Psychiatry (Kodaira, Tokyo, Japan).
The PET study was performed at the first session of each patient's ECT course. On the day of the experiment, patients fasted and were free from any medication. The experiment began at 14:00 to decrease the effects of any habitually used medicine, although the drug-free period was not long enough to eliminate all residual effects. We monitored electroencephalograms to estimate the levels of consciousness and depth of anesthesia and to confirm whether convulsions were induced by ECT. Electroencephalograms were recorded from disk electrodes placed at F3, F4, P3, P4, Fz, Cz, and Pz using A1 + A2 for reference. Each patient lay on a scanner couch, with the head stabilized by an individually molded thermoplastic face mask. A venous catheter was inserted into the right median antebrachial vein for transfusion of physiologic saline, administration of propofol, and injection of tracer, and an arterial catheter was inserted into the left radial artery for blood collection to measure radioactivity and blood gas and plasma propofol concentration throughout the scanning. The plasma concentration of propofol in the arterial blood sample of each PET imaging was assessed using high-performance liquid chromatography. Blood pressure, heart rate, and arterial oxygen saturation were monitored throughout the experiment. After the PET transmission scan, acquisition scanning in the awake state with the eyes closed and earplugs in place was performed three times (PET imaging Nos. 1–3;fig. 1). Propofol infusion was started at a rate of 10 mg · kg−1· h−1using an infusion pump until the patients no longer responded promptly to verbal commands in a normal tone. The sedated state was confirmed using a sedation score, which is the responsiveness component of the Observer Assessment of Alertness and Sedation rating scale (OAA/S). 11,12Patients who were drowsy, could still respond to some verbal commands, and could react to mild prodding or shaking were considered to be in a sedated state (OAA/S score of 2 or 3). These were the conditions in which scanning in the sedated state was performed (PET imaging No. 4;fig. 1). As this study included scans associated with ECT treatment (reported separately), only one scan was obtained in the sedated state to minimize total radiation dose and examination time. After the scan, a bolus load of 2 mg/kg propofol was given, and a laryngeal mask was inserted in each patient. The patients were under controlled ventilation with a respirator (tidal volume 10 ml/kg, 8 breaths/min). Next, propofol was infused again at 10 mg · kg−1· h−1for 15 min and then reduced to 5 mg · kg−1· h−1to maintain anesthesia. After that, we confirmed that the patients had been in an anesthetized state by observing their electroencephalograms, and scanning was performed three times (PET imaging Nos. 5–7;fig. 1). A dose of 0.15 mg/kg vecuronium bromide was first administered to the patients, and then additional doses of 0.04 mg/kg were given when necessary.
After this, bilateral ECT was performed using Thymatron DGx with Thymapad disposable electrodes (Somatics Inc., Lake Bluff, IL). A maximum of 12 intravenous injections of H215O for PET images were administered in the relaxed awake (3 injections), sedated (1 injection), and anesthetized (just before ECT; 3 injections) states and also during ECT (1 or 2 injections) and after ECT (3 injections). The intervals between scans were approximately 11 min. To assess the plasma concentration of propofol, arterial blood samples were taken just after each scan was performed. This article concerns data collected from the scans in the awake, sedated, and anesthetized states.
Positron Emission Tomography Procedure
Brain PET images were obtained using a PET camera (ECAT EXACT HR 961 scanner; Siemens, Erlangen, Germany) in the three-dimensional mode in the manner previously reported. 13In brief, the camera, which has an axial field of view of 150 mm, was used to acquire data simultaneously from 47 consecutive axial planes. An image resolution of 3.8 × 3.8 × 4.7 mm was obtained after back-projection and filtering (Hanning filter: cutoff frequency, 0.5 cycles per pixel). The reconstructed image was displayed in a matrix of 128 × 128 × 47 voxel format (voxel size: 1.7 × 1.7 × 3.1 mm). A 10-min transmission scan using a retractable rotating 68 Ga/68 Ge source with three rods was performed to correct the tissue attenuation and background activity prior to acquisition of the emission data. For each imaging, a bolus dose of 259 MBq (7 mCi) H215O was automatically injected intravenously over 15 s. Total radiation dose exposure per subject was less than 1 mSv to the whole body. Scanning was commenced manually 1 s after the initial rise of the head radioactive counts and was continued for 90 s. Arterial blood was sampled automatically throughout the scanning period using a flow-through radioactivity monitor (PICO Count, Bioscan Inc., Washington DC). Functional images of absolute rCBF were produced by utilizing arterial time activity data using autoradiography. 14
Positron emission tomography images were analyzed using SPM-99 software (Wellcome Department of Cognitive Neurology, London, United Kingdom ##) 8implemented in MATLAB version 5.3 (The MathWorks, Inc., Sherborn, MA) for Windows 98SE (Microsoft Co., Redmond, WA) on a personal computer. Spatial normalization was employed to fit each individual brain to a standard template brain in three-dimensional space, so as to correct for differences in brain size and shape and to facilitate intersubject averaging. The stereotactically normalized scans contained 68 planes (voxel size, 2 × 2 × 2 mm), and smoothing with a 10-mm Gaussian kernel produced a resolution of 17 × 17 × 20 mm for the final image. SPM uses a standard brain from the Montreal Neurologic Institute (Montreal, PQ, Canada); therefore, the precise anatomical localizations of significant changes were indicated in accordance with the atlas of Talairach and Tournoux 15by using a numerical transformation formula. ***In this study, the absolute rates of gCBF in the sedated and anesthetized states after induction of propofol anesthesia were analyzed and compared with those in the awake state, and then relative changes of rCBF were investigated in detail using global normalization with proportional scaling. 16
After the appropriate design matrix was specified, the condition of each voxel in each patient was assessed according to the theory of Gaussian fields. The exact level of significance of difference between conditions was characterized by peak amplitude. We focused on a cluster level to detect significantly different regions in the current study because our sample size was too small for random field theory and so would lead to type II errors (pseudonegative). Fortunately, we had a certain amount of evidence for the neural mechanism of propofol anesthesia. 4–7Therefore, we performed a priori studies and, in general, applied a corrected P value of 0.05 as an extent threshold, manipulating peak T values from 2.79 (height threshold: uncorrected P = 0.01) to 3.45 (uncorrected P = 0.001). Finally, a significant level was employed at a height threshold of P = 0.001 (T = 3.45) by reference to the unit distribution and at an extent threshold of corrected P = 0.05.
We compared physiologic variables and gCBF other than rCBF between the three states using repeated-measures analysis of variance followed by Bonferroni t test. Differences were considered significant when P was less than 0.05.
Physiologic Variables and Plasma Propofol Concentration
There were no marked changes in heart rate or in systolic, diastolic, or mean blood pressure in the sedated or anesthetized states, compared with those in the awake state (table 1). The measured concentrations of propofol in the sedated and anesthetized states were 0.851 ± 0.398 and 1.827 ± 0.334 μg/ml (mean ± SD;fig. 2A). Arterial carbon dioxide tension (Paco2) values in the awake and sedated states averaged 44.3 and 44.6 mmHg, respectively. In contrast, Paco2averaged 38.9 mmHg in the anesthetized state because of mechanical ventilation with a tidal volume of 10 ml/kg and a respiratory rate of 8 times/min (fig. 2B).
Absolute Rates of Cerebral Blood Flow
Imaging in the awake state was performed three times for each patient. However, tachypnea developed in a 70-yr-old male patient, probably because of anxiety about the investigation, and his Paco2dropped very sharply, reaching less than 30 mmHg in two of three scans in the awake state. Therefore, we excluded these two images from the analysis. The images of this patient in the sedated and anesthetized states were used for analysis because his respiratory condition returned to normal after the administration of propofol. The imaging in the anesthetized state was again performed three times for each patient. However, since the plasma propofol concentration became more stable, we used the last two images for data analysis. We therefore analyzed totals of 16, 6, and 12 images in the awake (3 images for five patients each and a single scan for one patient with hyperventilation), sedated (1 image for each of six patients), and anesthetized (2 images for each of six patients) states, respectively. The absolute values of gCBF in the awake, sedated, and anesthetized states were 28.1 ± 2.8, 20.5 ± 5.3, and 13.1 ± 2.8 ml · 100 g−1· min−1, respectively (mean ± SD;fig. 2C). The gCBF decreased by 26.8% from the awake state during the sedated state and by 54.4% during the anesthetized state. The absolute rCBF significantly decreased in almost all areas of the brain in both the sedated and anesthetized states, compared with that in the awake state (fig. 3).
Relative (Normalized) Rates of Cerebral Blood Flow
To specify the neural structures related to propofol anesthesia, the distribution patterns of CBF in the sedated and anesthetized states were investigated as follows. The rCBF data in the sedated and anesthetized states were compared with those in the awake state using global normalization with the scaling method. Moreover, we compared the relative rCBF values in the sedated and in the anesthetized states in the same way. Some distinct differences were seen between the relative and absolute rCBF changes.
In the sedated state, in which appropriate reactions to external stimuli, such as mild prodding or shaking, still remained to some extent (OAA/S score of 2 or 3), a significant decrease of relative rCBF occurred in a large area of the cerebral cortex. This occurred especially in the precuneus (Brodmann's area [BA] 39, 40), the supramarginal gyrus (BA 40), the angular gyrus (BA 39), the inferior parietal lobule (BA 39), the posterior parts of both inferiomiddle temporal regions (BA 39, 21), and the left lateral prefrontal region (BA 6, 9). In contrast, the relative rCBF in the sedated state tended to increase bilaterally in the putaminal regions but did not reach the significance level except for a part of the right putamen (table 2and fig. 4A).
In each patient, the anesthetized state was confirmed on the basis of a protocol for propofol administration and characteristic electroencephalogram waves 17,18(high-voltage, slow δ waves superimposed with fast waves [12–18 Hz]). In this anesthetized state, many of the relative rCBF decreases were observed in the same locations as in the sedated state, accompanied by a decrease in the left dorsolateral prefrontal cortex (BA 46, 6, 8, 9, 10) and both posterior cingulates (BA 23, 30). However, the extent of these effects was much larger. In addition to these, significant decreases in rCBF occurred even in regions other than the cerebral cortex, namely, both pulvinar nuclei of the thalamus, the cerebellar cortices, and the pontine tegmentum. No area showed significant increases in relative rCBF in the anesthetized state except for some intracranial blood vessels and some parts of the white matter (table 2and fig. 4B).
Comparison between the anesthetized and sedated states demonstrated that the relative rCBF decreased significantly in the anesthetized state in several discrete regions, namely, the pons, the medial dorsal nucleus and the pulvinar nucleus of the thalamus, the cerebellar cortex, and some parietal regions, in comparison with the sedated state. The only region of relative increase was the internal carotid artery on both sides (table 2and fig. 4C).
Effects of Electroconvulsive Therapy on Regional Cerebral Blood Flow
The absolute rates of gCBF during ECT and after ECT (when propofol was infused at a rate of 5 mg −1· kg · h−1) were 30.9 ± 6.2 and 17.7 ± 2.8 ml · 100 g−1· min−1, respectively. Relative changes of rCBF showed significant increases in the thalamus, basal ganglia, medial temporal regions, mid brain, and the pons bilaterally, compared with pre-ECT conditions, which correspond with the anesthetized state in this research. More details of rCBF during and after ECT will soon be reported separately.
Propofol and Global Cerebral Blood Flow
The gCBF in the awake state was 28.1 ± 2.3 ml · 100 g−1· min−1in the current study. In our previous study, 13the gCBF of 18 healthy subjects (average age, 21.9 yr) obtained using the same equipment was 35.2 ± 7.2 ml · 100 g−1· min−1. This absolute value of gCBF is somewhat low for healthy young volunteers, but it is almost completely in agreement with the other report on gCBF in the awake state using PET. 19It has been reported that depression and aging decrease gCBF. 20,21Since our subjects were fairly old (average age, 55.0 yr) and had severe depression, the decrease in gCBF in the awake state is likely to be associated with these factors.
Earlier reports, 22,23as well as studies using PET in animals and humans, 4,6,24showed that propofol produced a decrease in gCBF. In the current study, the gCBF decreased by 26.8% (20.5 ± 5.3 ml · 100 g−1· min−1) in the sedated state and 54.4% (13.1 ± 2.8 ml · 100 g−1· min−1) in the anesthetized state. The decrease in gCBF in the anesthetized state was somewhat larger than that in the previous reports. 2,6One possible reason was that our subjects had depression and were taking benzodiazepines and antidepressants. Propofol has been considered to promote hypnotic effects through augmentation of γ-aminobutyric acid type A receptor function. 25,26Therefore, the effect of propofol may be enhanced by the drug interactions between propofol and benzodiazepines. The second reason was the decrease in Paco2caused by the controlled ventilation. In a recent PET study, the vascular response to the change in Paco2in the cerebral cortex was reported to be −3.2%/mmHg for hypocapnia. 27In our study, the mean Paco2was 5.4 mmHg lower in the anesthetized state than in the awake state. Therefore, the decrease of Paco2should decrease gCBF by 17.3% in the anesthetized state.
Then, does the decrease of Paco2effect the changes in relative rCBF? It has recently been reported that the responses of rCBF to Paco2changes are not uniform among brain areas, and hypocapnia results in greater reductions of relative rCBF in the pons, putamen, and temporooccipital cortex. 27Therefore, we examined the relation between the values of Paco2and relative rCBF using the multisubject and covariate model in SPM-99. There were no significant differences between these two items. Thus, so far as the current study is concerned, the effects of hypocapnia on relative rCBF appear to be minor in comparison with those of propofol anesthesia.
Depression and Propofol Anesthesia
The subjects of this research were patients with major depression, so there is a possibility that the changes in rCBF caused by propofol differ from those of normal subjects. The reduction of neuronal activity in the frontal cortex is considered to be a characteristic of depression. 21,28Therefore, we compared the rCBF of our subjects in the awake state with that of 11 normal subjects, volunteers with a mean age of 44.4 yr. These normal data had been collected beforehand as the database of this scanner. The rCBF of the inferior frontal gyrus, which is a part of the frontal association cortex, was significantly lower in depressed patients than in normal volunteers. Fiset et al. 6reported that propofol significantly decreased rCBF in the medial thalamus, cuneus, precuneus, and posterior cingulate and right angular and orbitofrontal gyri. However, in our research, the rCBF of the orbitofrontal gyrus did not change significantly. Therefore, the change of rCBF in this region appeared to be lower than that in normal subjects. As far as the other regions are concerned, our results are almost entirely consistent with those reported by Fiset et al. These results suggest that the neural structures that are targeted by propofol are basically the same in severely depressed and normal subjects, with the exception of the basal frontal lobe.
Propofol and Regional Cerebral Blood Flow Analyzed by Statistical Parametric Mapping
Changes in Relative Regional Cerebral Blood Flow in the Sedated State.
When rCBF in the sedated state was compared with that in the awake state, the decrease in relative rCBF was especially great in the supramarginal gyrus, the angular gyrus, and the inferior parietal lobule regions, which are well known as the parietal association cortex (PAC;fig. 5A). The principal function of the PAC is the overall integration of the various somatic sensations and sensory information, including visual, somatosensory, and auditory input. 29,30In the frontal lobe, the decrease of relative rCBF was marked in the left middle frontal gyrus, which is a part of the prefrontal cortex. The prefrontal cortex is considered to be responsible for some of the highest functions of the central nervous system and is associated with higher mental activities and motor control. 29,31On the other hand, the decrease in rCBF was relatively small in the subcortical regions, including both putaminal regions. These findings imply that the activities of the subcortical regions, such as the basal ganglia and the brain stem, are maintained. In other words, the regulation of the autonomic nervous system and arousal reactions to external stimulation are still functional, while the higher functions, such as sensory processing, were strongly suppressed in the sedated state.
Changes in Relative Regional Cerebral Blood Flow in the Anesthetized State.
In the PAC, relative rCBF in the anesthetized state decreased to a larger extent than in the sedated state (fig. 5B). In the frontal lobe, including the prefrontal cortex of the left brain, Broca's area, and a part of the premotor cortex, a significant decrease in relative rCBF was observed. We also observed a significant decrease in the visual association cortex of the occipital lobe, which plays an important role in the identification of visual stimuli. 32However, there was no significant decrease in the primary visual cortex and the primary sensory and motor area. These results suggest that propofol may suppress the higher cognitive functions rather than directly suppressing the recognition of sensory stimuli and the primary motor function.
We also observed significant decreases in relative rCBF in the bilateral posterior cingulate gyrus, the cerebellar cortex, the pontine tegmentum, and the pulvinar nucleus of the thalamus. The brain stem reticular formation is thought to be the regulatory center for respiration and circulation. 33The pontine tegmentum is a part of the brain stem reticular formation, and decreased rCBF in this region may be related to central respiratory and circulatory depression caused by propofol. In addition, the brain stem reticular formation communicates with the cerebral cortex through the thalamus and forms the ascending activating system. 34The pulvinar nucleus has been considered to communicate with the posterior part of the parietal lobes and occipital lobes. 35,36Therefore, propofol appeared to exert its anesthetic effect by acting on the neural network that comprises these areas. A decrease in relative rCBF in the cerebellar cortex was also confirmed in the current study. The cerebellum is in close contact with the brain stem and mainly controls motor function. 37The uneven suppression of the areas influencing motor function may be associated with the spontaneous movements and convulsions occasionally reported during propofol anesthesia. 38,39In the anesthetized state, we used a neuromuscular blockade, and in this way, the study differs from previous PET studies. It was recently reported that muscle relaxation does not alter the hypnotic level during propofol anesthesia. 40Therefore, the influence of a neuromuscular blockade was thought to be minor in comparison with the change of rCBF induced by propofol.
Direct comparison between the anesthetized and sedated states revealed large decreases in the relative rCBF of the thalamus, especially in the medial dorsal nucleus and the pulvinar nucleus (fig. 5C). Veselis et al. 19reported that midazolam significantly decreased the rCBF of the medial dorsal nucleus, which is closely related to the frontal association cortex, and this pathway is reported to maintain vigilance and working memory. 41The relative rCBF of the cerebellar cortex, pontine tegmentum, and PAC also decreased in this combination.
In conclusion, we utilized high-resolution PET and SPM-99 to assess changes caused in rCBF by propofol in severely depressed patients. We found that propofol preferentially decreased rCBF in the bilateral PAC and the left lateral prefrontal region at lower plasma concentrations. At higher plasma concentrations, the extent of these effects was much larger, and substantial reductions in rCBF in the pontine tegmentum, the pulvinar nucleus, and the cerebellar cortex were observed. The brain regions preferentially affected by propofol appeared to be basically the same in severely depressed and normal subject populations, with the exception of the basal frontal lobe.
The authors thank Masato Kobayashi, B.A., and Rei Kurisu, M.Sc. (Technical Assistants, National Center Hospital for Mental, Nervous, and Muscular Disorders, National Center of Neurology and Psychiatry, Kodaira, Tokyo, Japan), for their excellent technical assistance. They also thank Christopher Walton Playford Reynolds (Lecturer in Medical English and Etymology, Yokohama City University School of Medicine, Yokohama, Kanagawa, Japan) for linguistic assistance with this article.