Effects of Xenon Anesthesia on Cerebral Blood Flow in Humans: A Positron Emission Tomography Study

The study protocol was approved by the Ethical Committee of the Hospital District of Southwest Finland (Turku, Finland). After giving written informed consent, nine healthy, right-handed, nonsmoking volunteers aged 20–27 yr with body mass index of 24.8 ± 2.6 kg/m²(mean ± SD, hereinafter presented similarly) were recruited in this open, nonrandomized study. The subjects underwent a thorough physical examination and laboratory testing, including 12-lead electrocardiography. All subjects confirmed having no history of drug allergies or drug abuse, and none had ongoing medications. Magnetic resonance images (1.5-T scanner, Philips Intera system; Philips Medical Systems, Best, The Netherlands) were obtained from each subject in a separate session to exclude structural abnormalities of the brain. Subjects fasted overnight and restrained from using alcohol or any medication for 48 h before anesthesia.

¹⁵O-labeled water was used as a PET tracer to determine rCBF. PET assessment was first performed with the subjects awake to determine baseline rCBF while subjects were breathing room air. The baseline values for vital parameters were determined as mean values during the awake PET assessment. After 1 h of denitrogenation with 100% oxygen, subjects were anesthetized with xenon (Xenon Pro Anesthesia; Air Liquide Deutschland GmbH, Krefeld, Germany). The second rCBF measurement was assessed during approximately 1 MAC anesthesia. After the scan, xenon was discontinued and the subjects were allowed to recover.

No premedication was given. A peripheral vein was cannulated for the administration of ¹⁵O-labeled water and 0.9% saline infusion (100 ml/h). A radial artery cannula was placed during local anesthesia for blood sampling and invasive blood pressure monitoring. Ventilation parameters, breathing gases (oxygen, carbon dioxide, xenon), and airway pressures were monitored throughout the anesthesia. Vital parameters and depth of hypnosis were monitored using a GE Datex-Ohmeda S/5 anesthesia monitor (GE Healthcare, Helsinki, Finland) with plug-in modules recording continuously noninvasive and invasive blood pressures, five-lead electrocardiography, pulse oximetry (E-PRESTIN Module; GE Healthcare), Bispectral Index (E-BIS Module; GE Healthcare; algorithm version 4.0, XP-level), and nasopharyngeal temperature. A portable computer running S/5 Collect software (GE Healthcare) was used for data recording. Arterial blood samples were obtained during scanning, and hematocrit and partial pressures of oxygen and carbon dioxide were repeatedly analyzed with portable equipment (i -STAT; Abbott Laboratories, Birmingham, United Kingdom).

To enable 63% xenon (the estimated MAC value for xenon)12in the closed-system ventilation, the partial pressure of nitrogen solved in the tissues must be considerably decreased. Denitrogenation was performed with subjects breathing spontaneously 100% oxygen through 5 cm H²O continuous positive airway pressure mask for 1 h before the induction. End-tidal oxygen concentration was allowed to reach 92% before xenon administration. Thereafter, the induction was performed by changing the inhaled oxygen concentration from 100% to 21%, thus letting the xenon concentration in the gas mixture to increase. Induction was facilitated by several flushes to gain the target concentration of 63% xenon. During the induction, the subjects breathed xenon spontaneously through a tightly fitting facemask, and they were repeatedly requested to squeeze the investigator’s hand twice. Failure to comply with the request was interpreted as loss of consciousness. A PhysioFlex closed-system ventilator (Dräger, Lübeck, Germany) was used for mask induction and for mechanical ventilation during the maintenance of anesthesia.13After the loss of consciousness, a 0.8-mg/kg bolus of rocuronium (Esmeron, 10 mg/ml; Organon, Helsinki, Finland) was administered for muscle relaxation, and the subjects were endotracheally intubated. Anesthesia was maintained with xenon as a single anesthetic, and additional doses of rocuronium were given to maintain relaxation at one twitch of train-of-four. In case of definite pain reaction due to intubation (rapid increase in blood pressure and heart rate), subjects were given a single 25-μg intravenous dose of remifentanil (Ultiva; GlaxoSmithKline, Espoo, Finland) immediately after the induction. Ventilation was adjusted to maintain the arterial blood partial pressure of carbon dioxide at baseline level, and the BairHugger warming system (Arizant Healthcare Inc., Eden Prairie, MN) was used to stabilize body temperature. After the induction, a 30-min minimum stabilization period was allowed to pass before the second PET scan. After completing the scan, xenon was discontinued and muscle relaxation was reversed with a neostigmine–glycopyrrolate combination (Robinul Neostigmin; Wyeth Lederle, Vantaa, Finland). Subjects were extubated as they recovered spontaneous breathing and regained consciousness. They were monitored for stable vital signs for a minimum of 1 h. The subjects were discharged according to our hospital’s standard criteria for ambulatory surgery patients.

The PET scans were performed with a GE Advance PET Scanner (GE Medical System, Milwaukee, WI), and ¹⁵O-labeled water was used as a tracer, as described in detail previously.14rCBF was assessed at baseline (awake) and during 1 MAC xenon anesthesia. Scans were performed in a dim room, and sudden loud noises were avoided. Positioning of the head was done using anatomical landmarks and laser alignment lights. Automated infusion equipment was used to administer a 300-MBq intravenous bolus of H²¹⁵O within 15 s followed by 90-s static three-dimensional tissue activity image acquisition. A roller pump and two-channel detector device were used for detection of arterial blood activity as previously described.14 

The preprocessing of imaging data were performed using the Statistical Parametric Mapping15software version 99 (SPM99) and Matlab 6.5 for Windows (Math Works, Natick, MA) with a previously described procedure.14Briefly, parametric images of rCBF were calculated from the tracer activity images and plasma activity curves using tracer kinetic modeling as described in our previous article.14Each subject’s images were realigned (motion correction). The normalization parameters for each subject were estimated from the mean image using a default algorithm and a ¹⁵O-water PET template delivered with SPM99.

To quantify rCBF, an automated region-of-interest (ROI) analysis16,17was performed using standardized ROIs defined on magnetic resonance template image representing brain anatomy in accordance with the Montreal Neurologic Institute (MNI) space database. Because this method is based on common stereotactic space, i.e ., spatial normalization of images, the operator-induced error in defining ROIs individually for each subject can be avoided. The ROIs were defined using Imadeus (version 1.20; Forima Inc., Turku, Finland) bilaterally on the frontal (19 planes), parietal (9 planes), medial temporal (9 planes), lateral temporal (10 planes), and occipital (7 planes) cortices; the anterior (10 planes) and posterior (6 planes) cingulate cortices; the insular cortex (8 planes); the thalamus (6 planes); the caudate (8 planes); the putamen (7 planes); the cerebellum (6 planes); and the white matter (5 planes). The average rCBF values were calculated from spatially normalized rCBF images. The average gray matter value was calculated using individual gray matter ROIs.

Voxel-based statistical image analysis was performed with SPM99 as described previously.14Within-subject subtraction analysis with T contrasts was used to test xenon-induced changes in rCBF between the conditions. Because rCBF values of parametric images are quantitative, SPM analyses were performed without global normalization, i.e ., using absolute rCBF values. The analysis was performed as an exploratory analysis covering the whole brain, i.e ., without any a priori  hypothesis or spatial constrictions regarding the location of possible differences. A P  value less than 0.05 (corrected for multiple comparisons) was considered statistically significant. The visualizations were performed with a height threshold T value of at least 3.0. The nonsignificant clusters were discarded from the visualizations by adjusting the minimum cluster size (extend threshold, k). The localization of the results of the SPM analysis was made using the MNI Space utility,##which first converts the MNI coordinates given by SPM to Talairach coordinates using nonlinear transformation18***and then identifies each voxel by the anatomical labels presented in the Talairach Daemon database.19 

All enrolled subjects were included in the statistical analyses. Quantitative rCBF and vital monitoring variables were analyzed with repeated-measures analysis of variance model having the condition (baseline and during anesthesia) as a within factor. The robustness of the statistical analysis was checked by replicating the analysis with nonparametric methods, i.e ., with Wilcoxon signed rank test. To study the effect of remifentanil, another repeated-measures analysis of variance model was fitted, including the condition as a within factor and the use of remifentanil (no/yes) as a between factor. The pairwise comparisons were performed using the linear contrasts of the same model. A two-sided P  value of less than 0.05 was considered statistically significant. Statistical analyses were conducted with SAS (version 8.2; SAS Institute Inc., Cary, NC) by 4Pharma Ltd. (Turku, Finland).

A summary of vital parameters is presented in table 1. The mean xenon concentration (± SD) during anesthesia was 65.2 ± 2.3%. The arterial partial pressure for carbon dioxide and the mean blood pressure and hematocrit remained unchanged between the scans. Peripheral oxygen saturation decreased by 2.0% (P = 0.005) and body temperature decreased by 0.9% (P = 0.02) during xenon anesthesia. The mean partial pressure for arterial oxygen was 92.1 ± 7.7 at baseline and 87.2 ± 11.0 mmHg during anesthesia (not significant). The mean value for the Bispectral Index was 94.9 ± 2.2 with the subjects awake and 24.8 ± 5.0 during the second PET scan (P < 0.001). Five subjects (55%) received a single bolus dose of 25 μg remifentanil (mean dose, 0.30 μg/kg) during intubation because of a clinically evident pain reaction. The mean time interval between opioid bolus and the second PET scan was 34 min (range, 30–39 min).

The baseline values for rCBF in the ROIs were 15.9–59.4 ml · 100 g⁻¹· min⁻¹. The absolute CBF values for each individual region are presented in table 2. rCBF changes individually in the thalamus, the cortical areas, and the insula are presented in figure 1. rCBF decreased by 34.7 ± 9.8% in the cerebellum (P < 0.001), by 22.8 ± 10.4% in the thalamus (P = 0.001), by 16.2 ± 6.2% in the parietal cortex (P < 0.001), by 15.1 ± 7.3% in the frontal cortex (P < 0.001), and by 12.5 ± 8.9% in the posterior cingulate (P = 0.008). rCBF increased in the white matter by 22.1 ± 13.6% (P = 0.001). On average, in the cortical regions the mean rCBF decreased by 8.7 ± 8.3% (P = 0.020). In all gray matter regions (cortical regions and the deep nuclei), the mean rCBF decreased by 11.2 ± 8.6% (P = 0.008). Remifentanil did not affect rCBF changes in any of the brain regions studied.

Xenon anesthesia caused statistically significant reduction in the absolute CBF in the frontal, temporal, parietal, and occipital lobes. The clusters representing the CBF decrease extended also to the limbic lobe, the cerebellum, the thalamus, the brain stem, and the lentiform nucleus. Absolute CBF increases were detected bilaterally in the precentral and postcentral gyri. The increases and decreases are shown in figures 2A and B. Additional information regarding this is available on the Anesthesiology Web site at www.anesthesiology.org.

One minimum alveolar concentration xenon anesthesia decreased rCBF from awake baseline in several brain areas. Greatest reductions were observed in the cerebellum, the thalamus, and the cortical areas. rCBF increased in the white matter. Furthermore, SPM analysis revealed increases in the precentral and postcentral gyri.

Previous animal studies on the CBF effects of xenon have been controversial. Autoradiography study with rats and sagittal sinus outflow study with pigs showed no change in rCBF during steady state inhalation of xenon.11,20However, one study using radiolabeled microspheres in pigs demonstrated a 38% increase in cortical CBF during xenon anesthesia.21In monkeys, subanesthetic xenon caused a global reduction in CBF,10whereas anesthetic concentration (80%) increased absolute CBF by 53%.22Although both of these studies with primates used the sagittal sinus microelectrodes method, the latter was compromised with significant variation in the partial pressure for carbon dioxide (35–46 mmHg). Interestingly, an increase in rCBF has been observed during first few minutes of xenon inhalation in rats but not during steady state anesthesia.20 

To our knowledge, there are no previous PET studies on the rCBF effects of xenon anesthesia in humans. However, few studies have been performed with other methodology. Using xenon-133 clearance, subanesthetic (35%) xenon was shown to increase human rCBF by approximately 12%.23Furthermore, significantly increased middle cerebral artery flow velocities were observed in humans breathing anesthetic (65%) xenon.24The inconsistency of animal and human results may partly be explained by adjuvant anesthetic agents and by nonfixed carbon dioxide levels. It should also be pointed out that interspecies differences in the MAC value for xenon should be taken into consideration when comparing animal and human data.8,12,25 

In the current study, the physiologic variables known to affect cerebral perfusion were carefully standardized. There were no significant changes detected in the partial pressure for arterial carbon dioxide or hematocrit. Also, the mean arterial pressure maintained unchanged and well within the range of intact cerebral autoregulation during the study. Previously, it has been demonstrated with pigs that cerebral autoregulation remains intact during xenon anesthesia.11It has been shown that cerebral metabolic rate of oxygen decreases 5–6% per 1°C.26Because the mean decrease in absolute body temperature was only 0.34°C in the current study, the concomitant changes in CBF can be considered minor. There was a minor decrease in peripheral oxygen saturation as well as in partial pressure for arterial oxygen during xenon anesthesia. The measured values for arterial oxygen were, however, notably higher than values known to increase cerebral blood flow.27 

In the current study, the target level for xenon was 1 MAC (63% xenon),12instead of aiming at a particular depth of hypnosis. Anesthetic depth monitors, e.g ., Bispectral Index, have not been validated for xenon anesthesia, and their usefulness has even been questioned during the emergence from xenon anesthesia.28The actual mean xenon concentration was 65.2%, which was quite close to the target. During steady state anesthesia, the Bispectral Index was remarkably low in all subjects, suggesting adequate depth of anesthesia. In the questionnaire, none of the subjects reported awareness during anesthesia.

To attenuate the pain reaction induced by intubation, five subjects received a bolus of short-acting opioid, remifentanil. Remifentanil has been shown to induce changes in rCBF per se .29On the other hand, major pain responses could cause even more considerable changes in rCBF. Considering the mean time interval between the opioid bolus and the second rCBF assessments and the extremely short half-life of remifentanil (approximately 3 min in healthy young men30), it is unlikely that it would have affected the results. This possibility was taken into consideration in the statistical analysis and, importantly, remifentanil did not affect the CBF results.

In physiologic circumstances, changes in rCBF parallel changes in regional cerebral glucose metabolism, which should, ultimately, reflect alterations in neuronal activity. Assuming that coupling is preserved during xenon anesthesia, the observed reduction in rCBF could represent decreased metabolism. In a recent study, it was shown that surgical concentrations of xenon induce a global reduction in regional cerebral glucose metabolism in humans.31Interestingly, another N -methyl-d-aspartate antagonist, S -ketamine, was previously shown to increase rCBF, but in excess of metabolic needs in anesthetic concentrations.7It is known that general anesthetics can induce vasodilatation in the cerebral vasculature, potentially leading to an imbalance in the relation between CBF and metabolism.7,14,32–35A more fundamental evaluation of the flow–metabolism relation during xenon anesthesia would require concomitant assessment of rCBF and brain metabolism. Concomitant measurement of oxygen consumption using the gaseous PET tracer ¹⁵O-labeled oxygen was not technically possible in the current study because of the applied closed-system ventilation.

SPM analysis revealed bilateral rCBF increases in the precentral and postcentral gyri. The implication of this deviant finding can only be speculated. Assuming intact flow–metabolism coupling, this could reflect neuronal activation in these areas receiving somatosensory information. Xenon monoanesthesia may not be sufficient to suppress all responses to the noxious stimulus in the current study, i.e ., the intubation tube. Concomitant assessment of glucose consumption using ¹⁸F-labeled fluorodeoxyglucose would be needed to resolve whether neuronal activation is involved.

There are also other limitations related to measuring only rCBF. Small changes in vessel diameter can cause considerable changes in CBF, and changes in CBF and cerebral blood volume do not always parallel each other, especially in the injured brain.36Assessment of cerebral blood volume using ¹⁵O-labeled carbon monoxide would have been warranted but was, for technical reasons already explained (gaseous PET tracer), not possible in the current study. Furthermore, the impact and relevance of CBF and cerebral blood volume changes on intracranial pressure can only be assessed in patients with decreased intracranial compliance.

The relevance of the increase in the white matter also remains speculative. It is unlikely that metabolic needs in the white matter would be increased. Our results suggest, however, that xenon anesthesia induces changes in blood flow distribution in the brain, i.e ., a decrease in the gray matter and an increase in the white matter. This could be due to direct vasodilatory effects of the drug.37The vasodilatory effect in the gray matter could be overridden by reduced CBF coupled to decreased metabolism as a consequence of xenon’s anesthetic effects. Nevertheless, the effects of anesthetics on blood flow and metabolism in the gray matter are far better known than in the white matter.

In conclusion, xenon anesthesia decreased rCBF especially in the cerebellum, the thalamus, and cortical areas. rCBF increased in the white matter and in parts of the precentral and postcentral gyri. These results are in clear contradiction with ketamine, another N -methyl-d-aspartate antagonist and neuroprotectant, which induces a general increase in CBF at anesthetic concentrations.

The authors thank the personnel of the Turku PET Centre (Turku, Finland) for excellent technical assistance; Mika Särkelä, M.Sc., and Jyrki Ruotsalainen, M.Sc. (GE Healthcare, Helsinki, Finland), for providing monitoring equipment and for technical advice; Mika Leinonen, M.Sc., and Tanja Huovinen, M.Sc. (4Pharma, Turku, Finland), for performing the statistical analyses; and Carsten Pilger, M.D., Peter Neu, M.D., and Paul Metten, M.B.A. (Air Liquide Deutschland GmbH, Krefeld, Germany), for helping with the regulatory obstacles. Xenon gas was purchased from Air Liquide.