British Journal of Anaesthesia 106 (4): 548–57 (2011)

Advance Access publication 31 January 2011 . doi:10.1093/bja/aeq415

Critical involvement of the thalamus and precuneus during

restoration of consciousness with physostigmine in humans
during propofol anaesthesia: a positron emission tomography
study
G. Xie 1,2,3, A. Deschamps 1, S. B. Backman 1, P. Fiset 1, D. Chartrand 1, A. Dagher 2 and G. Plourde 1*
1
Department of Anesthesia and 2 Department of Neurology/Neurosurgery, McGill University, Montreal, QC, Canada H3A 1A2
3
Present address: Department of Anesthesiology, Perioperative Medicine and Pain Management, Royal University Hospital G525,
103 Hospital Drive, Saskatoon, SK, Canada S7N 0W8
* Corresponding author: Department of Anesthesia, Montreal Neurological Hospital, 3801 University St., Montreal, QC, Canada H3A 2B4.
E-mail: gilles.plourde@mcgill.ca

Background. Functional brain imaging offers a way to investigate how general anaesthetics
Editor’s key points impair consciousness. However, functional imaging changes may result from drug effects
† Functional brain imaging unrelated to hypnosis. Establishing a causal link with loss of consciousness is thus difficult.
shows that anaesthetic- Methods. To identify changes of neuronal activity functionally linked to the level of
induced unconsciousness consciousness, physostigmine was used to restore consciousness without changing the
is associated with anaesthetic concentration in 11 subjects anaesthetized with propofol. Eight subjects
decreased thalamic and (responders) regained consciousness after physostigmine and three did not (non-
cortical activity. responders). Positron emission tomography was used to measure regional cerebral blood
† Unconsciousness and flow (rCBF); during baseline (awake), after anaesthesia-induced loss of consciousness,
functional brain imaging after physostigmine administration, and recovery. In addition to subtraction analyses, we
changes may, however, used conjunction analysis in the responders to identify changes common to the
be unrelated effects of baseline –anaesthesia and physostigmine–anaesthesia contrasts.
general anaesthetics. Results. Complete data were available for seven subjects (four responders and three non-
† Physostigmine can responders). The analyses revealed that unconsciousness was associated with rCBF
restore consciousness decreases in the thalamus and precuneus. Restoration of consciousness by physostigmine
during propofol was associated with rCBF increases in these same structures, with the strongest effect in
anaesthesia even if the the thalamus.
concentration of propofol Conclusions. The results provide strong evidence that reductions in rCBF in the thalamus and
is unchanged. precuneus are functionally related to propofol-induced unconsciousness independently
† This strategy of any non-specific effects of propofol. These observations confirm that the thalamus
demonstrated a and precuneus are key elements to understand how general anaesthetics cause
functional link between unconsciousness and how patients wake up from anaesthesia. Furthermore, they are
unconsciousness and consistent with the notion that anaesthetic-induced unconsciousness is associated with
decreased activity in reduced cholinergic activation.
thalamus and precuneus.
Keywords: brain imaging; cerebral blood flow; cerebral cortex; consciousness; general
anaesthetics; thalamus
Accepted for publication: 29 December 2010

Functional brain imaging studies have been carried out to thalamus and precuneus/posterior cingulate are keys to
identify where neuronal activity is altered during general how general anaesthetics cause unconsciousness.5 – 9
anaesthesia and to understand how general anaesthetic Decreases in blood flow in the frontal and lateral parietal
drugs impair consciousness. An important contribution of cortices have also been observed during anaesthesia,
the early studies was to reveal that anaesthetic-induced but less consistently than in the thalamus and precu-
unconsciousness is consistently associated with a reduction neus.1 – 4 6 – 8
in metabolism or blood flow in the thalamus and in the Functional brain studies carry limitations that are often
midline posterior cortical areas (precuneus or posterior overlooked. They usually rely on a subtraction analysis
cingulate cortex).1 – 4 These observations suggest that the to reveal the difference in brain activity between two

& The Author [2011]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
Functional imaging of anaesthetic action BJA
conditions: normal conscious (awake) baseline and assessed with the responsiveness component of the Obser-
anaesthetic-induced unconsciousness. The difficulty is to ver’s Assessment of Alertness/Sedation (OAA/S) scale.12 Sub-
establish a causal link between the observed differences jects were considered unconscious if they failed to respond to
and the changes in the level of consciousness. Although their name and to follow simple verbal commands (‘open
functional imaging changes may reflect neural events your eyes’). If the subject failed to respond, the requests
that contribute to the loss of consciousness, there may were repeated three times in a progressively louder voice.
also be changes that have nothing to do with conscious- Eight subjects regained consciousness after physostig-
ness but merely reveal non-hypnotic effects of the anaes- mine. The other three subjects showed no behavioural
thetic drug on the brain, such as those involving the changes (Fig. 1B). Of eight subjects who regained conscious-
cerebellum.4 Distinguishing the changes that are function- ness, only four were able to remain immobile during data
ally related to the loss of consciousness from those that acquisition. The other four subjects who regained conscious-
are unrelated to unconsciousness is difficult. This limitation ness after physostigmine did not complete the study because
should not be ignored since there are still unresolved issues excessive head movements during physostigmine-induced
about the role of the thalamus and medial posterior cortices consciousness prevented data acquisition. The present
in mediating anaesthetic-induced unconsciousness.5 9 Simi- results are thus for the seven subjects (six men; 20–35 yr)
larly, this problem also applies to functional brain who completed the study: four ‘responders’, who regained
imaging studies that attempt to identify the affected consciousness 3–7 min after the start of physostigmine
brain regions of patients who are comatose or in a vegeta- administration, and three ‘non-responders’, who showed no
tive state.10 behavioural change after physostigmine. It has not been
How can we demonstrate that the changes observed possible to recruit more subjects because clinical-grade phy-
with functional brain imaging during anaesthesia have any- sostigmine is no longer distributed in Canada.
thing to do with the loss of consciousness? This challenge We initially did not count on non-responders in the study
may be addressed by using physostigmine (a centrally design. Although we are aware that statistical power for
acting anticholinesterase) to restore consciousness during group comparisons is limited by the small size of the
anaesthesia with propofol, thereby changing the level of samples, we felt that it would be preferable to include the
consciousness while the anaesthetic concentration results from the non-responders since they provide an
remains the same.11 Thus, changes common to compari- additional opportunity to assess the role of the observed
sons of anaesthesia vs normal consciousness and anaesthe- changes of rCBF in regards to the level of consciousness.
sia vs physostigmine-induced consciousness likely reflect No major difference between responders and non-
neural events that are functionally related to changes in responders is expected for the BASE –ANES contrast, which
the level of consciousness since the main common includes a difference in the level of consciousness for both
feature of these two comparisons is a difference in the groups. We expect, however, that the differences between
level of consciousness. This approach should allow us to responders and non-responders in the PHYSO –ANES contrast
identify the changes that are functionally linked to the will reveal the critical regions implicated in the change in the
difference in the level of consciousness and, importantly, level of consciousness since the contrast involves a change in
control for any non-specific effects of propofol. The findings the level of consciousness only in the responders.
could thus further define the alterations of thalamocortical We attempted to study the effect of physostigmine on
activity that contribute to (or result from) the hypnotic rCBF when administered to an awake, non-medicated
effect of propofol. subject. Two subjects were tested. Despite the
co-administration of glycopyrrolate to prevent the peripheral
Methods muscarinic side-effects of physostigmine, both subjects
experienced intense nausea and retching that made scan-
Subjects, design, and response to physostigmine ning impossible. Attempts with other subjects were not
The study was approved by the Research Ethics Committee of undertaken given the intensity of the emetic symptoms.
the Montreal Neurological Institute. Eleven healthy right-
handed subjects (20– 36 yr) gave written informed consent
and were tested after a comprehensive medical evaluation. Anaesthesia and drug administration
Serial measures of regional cerebral blood flow (rCBF) were Subjects were under the care of two anaesthesiologists and
obtained with positron emission tomography (PET) during a one nurse. Vascular catheters were inserted under local
single session (starting around 12:00 h and lasting about anaesthesia in a vein of the right forearm for drug adminis-
4 h) comprising four successive periods: awake baseline tration and in the left radial artery for blood sampling. Moni-
(BASE), anaesthesia (subjects unconscious) (ANES), physo- toring included ECG, pulse oximetry, intra-arterial pressure,
stigmine administration (PHYSO), and recovery (20 min and online concentration of oxygen and carbon dioxide in
after end of propofol infusion) (RECO). Two scans were inspired and expired gas. Subjects breathed spontaneously
acquired during each period (Fig. 1A). There was an interval and received supplemental oxygen via a facemask. Gentle
of at least 15 min between successive scans to allow decay chin lift was applied during anaesthesia to ensure patency
of the radioactive tracer. The level of consciousness was of the airway.

549
BJA Xie et al.

A B1 B2 A1 A2 P1 P2 R1 R2

PET scans

Propofol

Physostigmine
Unconscious
Conscious

B
11 subjects anaesthetized
with propofol

3 showed no response to 8 regained consciousness

physostigmine after physostigmine

All 3 yielded 4 yielded 4 yielded

imaging data imaging data no imaging data
(excessive movements)

Final sample of 7 subjects

3 non-responders
4 responders

Fig 1 Experimental plan for drug administration and data acquisition. (A) Two scans (15 min apart) were obtained for each period: awake base-
line (B), anaesthesia (A), physostigmine (P), and recovery (R). After baseline recordings, the infusion of propofol was started and the target
concentration was increased in small steps until unconsciousness occurred. Physostigmine was started after scan A2 as an initial bolus
over 2 min followed by a continuous infusion. Scan P1 was started 2 min after the return of consciousness (or 10 min after the start of phy-
sostigmine if the subject failed to regain consciousness). Scan R1 was acquired about 20 min after the end of the propofol and physostigmine
infusions. Subjects were expected to be conscious during all periods except anaesthesia. Infusion rates not drawn to scale. (B) Response to
physostigmine and data acquisition.

Propofol was infused with a Harvard Apparatus 22 pump subjects would remain awake for at least 20 min. To
controlled by a laptop computer running the Stanpump soft- prevent the peripheral muscarinic side-effects of physostig-
ware (developed by S.L. Shafer, Stanford University, CA, USA) mine, the muscarinic antagonist glycopyrrolate was given
using the data set of Tackley and colleagues,13 which had (4.2 mg kg21 in 2 min followed by 2.1 mg kg21 over 30
been used for our previous study.11 The software combines min). Glycopyrrolate does not cross the blood– brain
boluses and an infusion with an exponentially declining barrier.14 Imaging data were obtained 2 min after the
rate to achieve the desired effect-site drug concentration. return of consciousness or 10 min after the start of the phy-
After acquisition of the baseline data, the propofol infusion sostigmine infusion if the subject failed to regain conscious-
was started, aiming for an effect-site concentration of 2.0 ness. After acquisition of the physostigmine data, the
mg ml21. The concentration was increased by 0.5 mg ml21 propofol and physostigmine infusions were stopped.
steps at 5 min intervals until unconsciousness occurred. Arterial blood samples were obtained at the middle of each
After a delay of 5 min, the first scan during anaesthesia period for blood gas analysis and for subsequent determi-
was acquired. This delay was used for tracer preparation nation of the concentration of propofol by high-performance
and for confirmation that the subject was indeed uncon- liquid chromatography. The assay was conducted by F. Varin,
scious. After acquisition of the anaesthesia data, physostig- PhD (Faculté de Pharmacie, Université de Montréal).
mine was administered (28 mg kg21 in 2 min followed by
14 mg kg21 over 30 min). This dose is higher than in our pre- Effects of physostigmine in conscious subjects
vious study11 and is based on a pilot study (n¼3) which Because intense nausea and retching made it impossible to
showed that an infusion was required to ensure that the assess the effects of physostigmine on CBF in otherwise non-

550
Functional imaging of anaesthetic action BJA
medicated subjects, the effects of physostigmine in con-
scious subjects were estimated by comparing recovery with Responders
baseline. During recovery, the concentration of propofol Non-responders
6
was in the low sedative range (Fig. 2) and the pharmacoki- 5

(µg ml–1)
Propofol
netics of physostigmine suggest that its residual effect 4
3
during recovery was near 80% of peak effect.15 2
1
0
Cerebral blood flow
140 a b c
PET scans were obtained using a CTI/Siemens HR+, 32-rings,

(beats min–1)
120
63-slices tomograph with the H2 15O bolus technique. Counts
100

HR
were measured during a 3 min scan after a 10 mCi H2 15O
80
bolus injection into a vein of the right ventral forearm. To 60
allow for calculation of the absolute CBF, arterial blood 40
samples were acquired throughout the scanning period d e f
100
using the catheter placed into the left radial artery. Arterial
90

(mm Hg)
blood radioactivity was automatically sampled, corrected for

MAP
80
delay and dispersion,16 and calibrated with respect to the 70
tomograph. To facilitate the localization of the regional 60
changes in CBF, each subject underwent a magnetic reson- 50
ance imaging (MRI) scan on a separate occasion (high- 9
resolution, T1-weighted, 1.5 T, 160 contiguous sagittal slices, 8

2
(kPa)
PaCO
1 mm thick). Data analysis was performed with the Brain 7
6
Imaging Software Toolbox of the McConnell Brain Imaging
5
Center (http://www.bic.mni.mcgill.ca/ServicesSoftware/HomePage).
4
60
K1 maps (ml kg min–1)
55
50
gCBF

The absolute CBF was calculated for the whole brain for each
45
scan of each subject. We used a two-compartment, time- 40
weighted integration method.17 Cerebral perfusion maps 35
30
(K1 maps) were generated for each 3 min scan using the
BASE ANES PHYSO RECO
sum of the native PET images across all frames. Mean whole-
brain CBF values were then obtained by averaging the
Fig 2 Propofol concentration in plasma and physiological
K1 maps.
measures (heart rate, mean arterial pressure, arterial partial
pressure of CO2, and global CBF). Each line represents one
Normalized rCBF subject. For simplicity, we only report significant differences
To identify the brain regions where propofol and physostig- between adjacent periods. The only statistical comparisons
mine induced changes in blood flow beyond those observed done for propofol were between anaesthesia and physostigmine
for responders and non-responders (NS in both cases). a, P,0.05
globally, we analysed the effects of propofol on normalized
for non-responders; b, P,0.01 for both groups; c, P,0.005 for
rCBF. The K1 images were normalized for differences in both groups; d, P,0.005 for responders and P,0.05 for non-
global CBF by means of ratio normalization; that is, the responders; e, P,0.01 for responders; f, P,0.01 for responders
count at each voxel (three-dimensional image element) and P,0.05 for non-responders.
was divided by the mean count calculated across all brain
voxels. The normalized K1 images were co-registered with
individual MRIs and transformed into a standardized stereo-
taxic space18 (MNI 305 template) by means of an automated To assess the differences in rCBF distribution between two
feature-matching algorithm.19 Statistical analysis of rCBF periods (i.e. BASE –ANES), we calculated subtraction t-statistic
was performed using normalized rCBF. Unless indicated maps between the two periods using the two scans from each
otherwise, the expression rCBF will denote normalized rCBF. period. Data were smoothed with an isotropic spatial Gaussian
filter (14 mm FWHM). A t-value was calculated for each voxel
Statistical t-maps of rCBF by dividing the mean CBF difference by its standard deviation
All the calculations were carried out for each of the pooled across all brain voxels.20
three-dimensional volume elements (voxels) constituting a To determine whether the differences in the partial
volume. The size of a voxel was 1.34×1.72×1.5 mm in x, y, z pressure of arterial CO2 (PaCO2 ) between periods contributed
dimensions, respectively. Analyses were conducted separately to the differences in rCBF between periods (e.g. BASE –ANES
for responders and non-responders, unless indicated contrast), we obtained regression t-maps of the contrasts
otherwise. between periods as a function of the difference in PaCO2 .

551
BJA Xie et al.

We adjusted the significance criterion for the multiple anaesthesia. Both groups showed a significant decrease in
comparisons involved in searching across a volume with a mean arterial pressure during anaesthesia compared with
method based on three-dimensional Gaussian random field baseline. The responders showed a significant increase in
theory.20 We report all activations significant at P≤0.05 mean arterial pressure after physostigmine. Both groups
based on the combination of peak t-value and cluster showed an increase in mean arterial pressure during recov-
spatial extent.21 The criterion for significance (P,0.05) of ery compared with physostigmine (Fig. 2).
peak t-value independently of cluster size was |t|.4.5.

Conjunction analysis Brain imaging

To identify significant peak activations common to the BASE – In the responders, the BASE – ANES contrast revealed
ANES and PHYSO– ANES contrasts in the responders, we per- decreased rCBF during anaesthesia in the right thalamus
formed a conjunction analysis of the t-maps for voxels with and bilaterally in the precuneus and lateral parietal cortex
t≥4.5. This analysis reveals the voxels where the absolute (Table 1 and Fig. 3A). In comparison with anaesthesia-induced
t-value equals or exceeds 4.5 in both contrasts. These loss of consciousness, there was, during physostigmine-
voxels reveal significant changes common to comparisons induced awakening, an increase in rCBF in the thalamus and
of anaesthesia vs normal consciousness and anaesthesia vs cuneus/precuneus bilaterally and also in the right lateral par-
physostigmine-induced consciousness. These changes likely ietal cortex (PHYSO –ANES contrast; Table 2 and Fig. 3A).
reflect neural events that are functionally related to Conjunction analysis of the above data was used to identify
change in the level of consciousness since the common significant rCBF changes common to both contrasts (BASE –
feature of these two comparisons is a difference in the ANES and PHYSO –ANES). These changes are likely to be func-
level of consciousness. To evaluate the magnitude of the tionally linked to the change in the level of consciousness
changes, we extracted rCBF values using a spherical since the main common feature of these two contrasts is a
region-of-interest (4 mm radius) centred in the middle of difference in the level of consciousness. These changes
the conjunction volume. occurred in the precuneus bilaterally and in the right thalamus
(Fig. 3B), where rCBF during anaesthesia was decreased in
Event regression analysis comparison with baseline or physostigmine-induced awaken-
ing. Increases in rCBF also occurred during anaesthesia in
To compare responders with non-responders in regards to comparison with baseline or physostigmine-induced awaken-
the differences between periods (e.g. BASE –ANES and ing (minima in Tables 1 and 2), but the conjunction analysis
PHYSO– ANES contrasts), the relationship between rCBF and
revealed no significant increase common to the BASES –
the event of regaining consciousness during physostigmine ANES and PHYSO–ANES contrasts.
infusion was evaluated by means of an analysis of covari- The rCBF from 4 mm spheres in the centre of thalamus
ance (ANCONA) with the covariate set to 1 for the responders
and precuneus conjunction areas was measured in each
and 0 for three non-responders. subject to evaluate rCBF across time (Fig. 3B). The rCBF in
the centre of the thalamic conjunction area during
Miscellaneous statistical tests anaesthesia-induced loss of consciousness was reduced by
Paired t-tests or repeated-measures ANOVAs (with the Geisser – 20–25% in comparison with the other three periods, when
Greenhouse adjustment) were used for the data in Figure 2. consciousness was present. For the conjunction area in the
Significant ANOVAs were followed by Tukey’s Honestly Signifi- precuneus, rCBF during anaesthesia-induced unconscious-
cant Difference test. The software used for these tests was ness was reduced by 16 –22% compared with the other
Statistica (version 4.1 for the Macintosh; Statsoft, Tulsa, OK, three periods.
USA). The criterion for statistical significance was P,0.05. In the non-responders, BASE – ANES contrast revealed
decreased rCBF during anaesthesia-induced unconscious-
Results ness in the right thalamus, right lateral parietal cortex, and
in the precuneus bilaterally (Fig. 3C; Supplementary Table
Plasma propofol concentrations and physiological S1). The PHYSO–ANES contrast was however clearly different
parameters from that seen in the responders. It revealed a strong
For the responders as a group, the measured propofol concen- increase in rCBF after physostigmine in the medial occipital
tration was slightly lower during the physostigmine period area bilaterally but no changes in the precuneus and thala-
than during anaesthesia (mean: 2.9 vs 3.2 mg kg21) but the mus (Fig. 3C; Supplementary Table S1). Regression t-maps
difference was not significant (P¼0.094). One non-responder of the subtractions BASE – ANES and PHYSO–ANES as a func-
required an unusually large concentration (.5 mg ml21) of tion of PaCO2 difference between these periods yielded no sig-
propofol for unconsciousness (Fig. 2). nificant activations in either responders or non-responders
Non-responders showed a significant increase in heart (data not shown).
rate during anaesthesia compared with baseline. Both Comparison of responders vs non-responders in regards to
responders and non-responders showed a significant the BASE –ANES contrast revealed no significant activations
increase in heart rate after physostigmine compared with (data not shown). Comparison of responders vs non-

552
Functional imaging of anaesthetic action BJA
because of the possibility that the action of physostigmine
Table 1 BASE – ANES contrast for responders. Coordinates in may not have been the same in both groups.
standard stereotaxic space (mm), approximate anatomical
location, and peak t-value

Anatomical location Stereotaxic t

coordinates Discussion
x y z This is the first human imaging study to provide an evaluation
Maxima of brain activity associated with both anaesthetic-induced
Right precuneus 15 261 48 7.2 unconsciousness and the restoration of consciousness,
Left precuneus 21 250 30 7.0 despite the continued administration of the anaesthetic.
Right precuneus 1 268 38 6.8 This approach allowed us to identify cerebral structures that
Right thalamus (medial dorsal n.) 5 216 11 6.2 are most likely functionally linked to the changes in con-
Right precuneus 9 278 57 5.8 sciousness induced by a general anaesthetic drug. The data
Left inferior parietal lobule 248 250 54 5.4 clearly showed that the difference between consciousness
Right superior parietal lobule 29 259 63 4.6 and unconsciousness involved site-specific changes of rCBF
Right fusiform gyrus 20 288 215 4.6 in the thalamus and precuneus. There was also good evidence
Left orbital gyrus 212 34 223 4.6 for involvement of the lateral parietal cortex. In the respon-
Left cerebellum 221 280 224 4.5 ders, conjunction analysis of the subtractions between base-
Right inferior temporal gyrus 52 249 29 4.4 line and anaesthesia (BASE–ANES contrast) (Fig. 3B and
Right lingual gyrus 1 288 2 4.3 Table 1) and between physostigmine-induced consciousness
Right cuneus 8 288 23 4.1 and anaesthesia (PHYSO–ANES contrast) (Fig. 3B and
Right inferior parietal lobule 46 244 59 4.0 Table 2) demonstrated that unconsciousness is specifically
Right supramarginal gyrus 54 250 32 4.0 associated with reduced rCBF in the thalamus and precuneus.
Right superior parietal lobule 39 268 59 4.0 In addition, outside the conjunction area, a decrease in rCBF in
Right inferior parietal lobule 47 245 57 4.0 the lateral parietal cortex during unconsciousness was
Left cerebellum 225 268 223 4.0 observed with these two contrasts. In the non-responders,
Right superior occipital gyrus 27 285 30 3.9 the BASE –ANES contrast (Fig. 3C; Supplementary Table S1)
Left superior parietal lobule 224 281 51 3.9 also revealed decreased rCBF in the thalamus, precuneus,
Left lingual gyrus 25 283 26 3.9 and lateral parietal cortex during anaesthesia. Moreover, the
Minima comparison of responders vs non-responders in regards to
Left insula 242 27 9 25.6 the PHYSO–ANES contrast (Supplementary Fig. S1 and Table
Right anterior cingulate gyrus 1 32 21 25.0 S2), which involved a change in the level of consciousness
Left superior medial frontal gyrus 23 214 48 24.7 only in the responders, provided additional support, indicating
that the activation of the thalamus, precuneus, and lateral
parietal cortex is specifically associated with the restoration
of consciousness. Finally, regression t-maps as a function of
responders in regards to the PHYSO –ANES contrast revealed the differences in PaCO2 between periods revealed no signifi-
bilateral areas where the difference in rCBF between the two cant activations, indicating that the differences in PaCO2 do
periods was significantly larger in responders than in non- not account for the observed changes in rCBF.
responders. These areas were the thalamus, the putamen, Most previous studies investigating the effects of propofol
the cuneus/precuneus, and the right lateral parietal cortex on relative rCBF had identified the thalamus as a key target.4
22 – 24
(Supplementary Fig. S1 and Table S2). This comparison con- The reduction in rCBF in the thalamus during
firmed that there was a difference between responders and anaesthetic-induced unconsciousness is consistent with the
non-responders in regards to the effect of physostigmine view that the disruption of thalamic function by anaesthetic
on rCBF. Thus, responders and non-responders showed drugs leads to unconsciousness.5 9 First, the thalamus plays a
similar rCBF changes in response to propofol but different critical role in the maintenance of consciousness.25 Secondly,
rCBF changes after physostigmine. the literature on the effects of general anaesthetics on
The effects of physostigmine on rCBF in conscious subjects somatosensory-evoked potentials in humans supports the
were estimated by comparing recovery with baseline (RECO- notion that information transfer through the thalamus is dis-
BASE contrast) in the responders. This contrast showed that rupted, with the ‘non-specific’ nuclei being most affected.26
rCBF during recovery was significantly increased compared Furthermore, there is evidence from animal studies that the
with baseline mainly in the left thalamus and bilaterally in first site of action in the somatosensory system where anaes-
the precentral gyrus, anterior cingulate, and medial superior thetics exert a profound inhibitory effect is the monosynaptic
frontal gyrus (Supplementary Fig. S2 and Table S3). The response of ventrobasal thalamic neurones to cuneothalamic
RECO-BASE contrast in the non-responders also showed an input.27 Thirdly, interventions that increase thalamic activity
increase in rCBF in the left thalamus during recovery (data can restore behavioural responsiveness in anaesthetized
not shown). We did not pool the data from both groups animals. Microinjection of nicotine in the central medial

553
BJA Xie et al.

A Responders
BASE-ANES PHYSO-ANES

7.2 11.5

3.5 3.5

B Conjunction

200
Normalized rCBF (%)

200
150
150
100
100
50
50
0
0
SE ES O O
SE ES SO O
BA AN YS EC
BA AN Y EC PH R
PH R

C Non-Responders
BASE-ANES PHYSO-ANES
7.5 7.5

3.5 3.5

Fig 3 Imaging data for the BASE – ANES and PHYSO– ANES contrasts in responders and non-responders. (A) Activation t-maps overlaid over
average anatomical MRI in standard stereotaxic space. Colour scale shows the t-value. Stereotaxic coordinate of the section plane in
yellow. The right side of the images corresponds to the right side of the subjects in this and all other figures. This contrast reveals the
areas where rCBF was greater during baseline or physostigmine than during anaesthesia. For the BASE –ANES contrast, the areas are the pre-
cuneus and right thalamus. For the PHYSO– ANES contrast, the areas are the cuneus, precuneus, and thalamus. (B) The conjunction between
the two activation t-maps shown in (A) for t.4.5 is shown in pink overlaid over average anatomical MRI. This analysis reveals voxels where the
decrease in rCBF reached significance on the basis of t-values in both contrasts. The bar graphs (precuneus on the left; thalamus on the right)
show rCBF [mean (SEM)] for a spherical 4 mm region-of-interest centred at the middle of the conjunction areas [stereotaxic coordinates (x,y,z):
5, 217, 8 for the thalamus; 1, 267, 32 for the precuneus]. (C) Same as (A) for non-responders. The BASE –ANES contrast shows decreased rCBF
in the precuneus and right thalamus during anaesthesia, as for responders. The PHYSO– ANES contrast show increased rCBF in the cuneus after
physostigmine but no changes in the precuneus and in the thalamus.

554
Functional imaging of anaesthetic action BJA
attenuate thalamic activity by decreasing the influence of
Table 2 PHYSO– ANES contrast for responders. Coordinates in the massive corticothalamic projection system.30 However,
standard stereotaxic space (mm), approximate anatomical
we consider this explanation less likely because the
location, and peak t-value
reductions in rCBF in the cortex do not appear sufficiently
Anatomical location Stereotaxic t extensive to support this possibility.
coordinates The present results confirm the implication of the precu-
x y z neus in anaesthetic-induced unconsciousness. The precu-
Maxima neus is a region where relative rCBF is consistently reduced
Left thalamus (medial dorsal n.) 23 216 8 11.5 during general anaesthesia with propofol4 7 22 – 24 or other
Right thalamus (medial dorsal n.) 8 218 9 11.2 agents6 and during other unconscious states (slow-wave
Right cuneus 5 274 18 9.2 sleep, neurovegetative state).10 31
Right cuneus 3 287 9 8.3 One outstanding issue is that although some rCBF
Right cuneus 24 287 26 6.6 decreases in the precuneus may reveal neural events mediat-
Left cuneus 212 283 35 6.6 ing unconsciousness, others may reflect neural events that
Left cuneus 221 288 23 6.3 result from loss of consciousness. It has been proposed that
Left putamen 224 21 5 6.0 the precuneus and interconnected posterior cingulate are
Left precentral gyrus 240 213 44 5.5 engaged in continuous information gathering and represen-
Right precentral gyrus 56 27 35 5.4 tation of the self and external world.32 33 As such, one would
Left superior medial frontal gyrus 21 226 63 4.9 predict that the impairment of the precuneus could contribute
Right fusiform gyrus 20 283 220 4.6 to unconsciousness. However, there is ample evidence that
Right cuneus 215 271 11 4.5 the precuneus subserves self-centred mental imagery and
Right superior parietal lobule 28 275 48 4.4 episodic memory retrieval.32 These mental activities are
Right putamen 28 5 5 4.2 known to occur spontaneously during wakefulness, particu-
Minima larly when one is waiting passively (during functional brain
Left parahippocampal gyrus 231 219 221 25.0 imaging sessions, for example).33 Thus, loss of consciousness
Right insula 35 1 215 24.9 would interrupt these ongoing mental processes and abolish
Right rectus gyrus 9 20 218 24.9 their neural correlates. Thus, it is possible that the decreases
Right parahippocampal gyrus 35 216 221 24.8 in rCBF in the thalamus during propofol anaesthesia reveal
Left superior medial frontal gyrus 212 30 42 24.7 two types of events: some causing unconsciousness and
Right superior medial frontal 28 29 36 24.7 some resulting from unconsciousness.
gyrus The posterior lateral parietal cortex is a region where rela-
tive rCBF is typically reduced during general anaesthesia with
propofol4 7 22 – 24 or other agents6 and during other uncon-
thalamus of rats restored the righting reflex during inhaled scious states (slow-wave sleep, neurovegetative state).10 31
anaesthesia with volatile agents.28 Although the right posterior lateral parietal lobe did not
The most likely site of peak thalamic activation in our study come out in the conjunction analysis, this structure revealed
is the medial dorsal nucleus, as in our previous reports.4 22 a significant reduction in rCBF for all contrasts involving a
Interestingly, the medial part of this nucleus receives dense change in the level of consciousness (Tables 1 and 2; Sup-
cholinergic projections from both the brainstem and the plementary Table S1). Thus, the present data suggest that
basal forebrain and is close to the site where electrical stimu- propofol-induced unconsciousness involves impaired func-
lation of the thalamic intralaminar nuclei improved behaviour- tion of the posterior lateral parietal lobe, but the evidence
al responsiveness in a patient in the minimally conscious is less robust than for the precuneus and thalamus.
state.29 The medial dorsal thalamus is immediately adjacent How does physostigmine restore consciousness during
to the central medial thalamus where the microinjections of anaesthesia with propofol? We think that the thalamus is a
nicotine restored the righting reflex during inhaled anaesthe- critical site of action for physostigmine based on: (i) the
sia with volatile agents in rats.28 Given the diffusion character- very strong, bilateral thalamic activation seen in the respon-
istics of the microinfusion technique used, the actual effect ders after physostigmine; (ii) the complete absence of thal-
site may be still in the dorsal medial thalamus. amic activation in the non-responders after physostigmine;
The decrease in thalamic rCBF in the BASE –ANES contrast (iii) the increase in rCBF in the left thalamus revealed on
predominantly involved the right side, as was the case for our the RECO– BASE contrast (Supplementary Fig. S2 and Table
previous studies with propofol.4 22 The possibility of a prefer- S3), used to assess the effect of physostigmine alone; (iv)
ential sensitivity of the right thalamus should be further the presence of cholinergic receptors in the thalamus;34
investigated. and (v) animal evidence, indicating that cholinergic acti-
We cannot exclude the possibility that the reductions in vation of the thalamus can restore consciousness during
rCBF in the thalamus observed during anaesthesia result general anaesthesia.28
from decreased corticothalamic input. Direct attenuation of Antagonism of propofol anaesthesia by physostigmine is
cortical activity by an anaesthetic would be expected to consistent with the view that propofol anaesthesia is

555
BJA Xie et al.

associated with reduced cerebral cholinergic tone. This Acknowledgements

reduction of cholinergic tone might be related to a
We thank C. Porlier and L. Ullyatt for their help with clinical
propofol-induced reduction in muscarinic receptor binding
care of the subjects; the staff of Brain Imaging Center for
in thalamocortical systems35 and to reduced levels of
their cooperation; Sylvain Milot and Marc Bouffard for help
endogenous cholinergic drive.36 There is evidence that pro-
with data analysis; Sonia Charbonneau, Patricia Smith Church-
pofol inhibits M1 receptor-mediated signal transduction by
land, Nicholas D. Schiff, Peter Shizgal, and Amir Shmuel for
selectively disrupting interaction between the receptor and
comments on an earlier version of the manuscript.
associated G protein.37 Moreover, our previous study
showed that the restoration of consciousness by physostig-
mine was blocked by the muscarinic antagonist scopola- Conflict of interest
mine.11 Thus, it is plausible that physostigmine increased
None declared.
synaptic levels of acetylcholine and that the resulting
increase in muscarinic M1 receptor binding compensated
for any propofol-induced reduction of muscarinic signalling.
Funding
The following limitations of the present study should be
noted. First, the effective sample size is small. It has not This work was supported by a grant from the Canadian Insti-
been possible to recruit more subjects because clinical-grade tute of Health Research (MOP-49583).
physostigmine is no longer available for research purpose in
Canada. Despite the small sample size, the available data
yielded unequivocal findings with high statistical significance
References
that confirm and expand previous observations. Nevertheless, 1 Alkire MT, Haier RJ, Barker SJ, Shah NK, Wu JC, Kao YJ. Cerebral
metabolism during propofol anesthesia in humans studied with
there is a risk that some relevant changes have been missed
positron emission tomography. Anesthesiology 1995; 82:
because of the small sample size. Secondly, the administration 393– 403
of physostigmine was not blinded and there was no control 2 Alkire MT, Haier RJ, Shah NK, Anderson CT. Positron emission tom-
group. We did not include a control group because we ography study of regional cerebral metabolism in humans during
wanted to maximize the number of scans after physostigmine. isoflurane anesthesia. Anesthesiology 1997; 86: 549–57
Furthermore, we felt that blinding would be difficult to achieve 3 Veselis RA, Reinsel RA, Beattie BJ, et al. Midazolam changes cer-
because we knew from past experience11 that physostigmine ebral blood flow in discrete brain regions. An H152 O positron emis-

increases the rate and depth of breathing in a readily notice- sion tomography study. Anesthesiology 1997; 87: 1106–17
able manner. Thirdly, the responders and non-responders dif- 4 Fiset P, Paus T, Daloze T, et al. Brain mechanisms of
fered in the mean concentration of propofol and in their propofol-induced loss of consciousness in humans: a positron
emission tomographic study. J Neurosci 1999; 19: 5506–13
cardiovascular and respiratory parameters. These differences
5 Alkire MT, Hudetz AG, Tononi G. Consciousness and anesthesia.
constitute a potential, unavoidable confounding variable in
Science 2008; 322: 876–80
the interpretation of differences in brain activity associated
6 Heinke W, Schwarzbauer C. In vivo imaging of anaesthetic action
with physostigmine-induced reversal of unconsciousness. in humans: approaches with positron emission tomography (PET)
Fourthly, although the approach revealed stronger functional and functional magnetic resonance imaging (fMRI). Br J Anaesth
links between rCBF and loss of consciousness, determining 2002; 89: 112– 22
whether the alterations of rCBF reflect events that are the 7 Kaisti KK, Metsahonkala L, Teras M, et al. Effects of surgical levels
cause or consequence of unconsciousness is difficult. of propofol and sevoflurane anesthesia on cerebral blood flow in
On the basis of the differences in rCBF observed among the healthy subjects studied with positron emission tomography.
Anesthesiology 2002; 96: 1358– 70
normal conscious baseline, propofol-induced unconscious-
8 Schlunzen L, Vafaee MS, Cold GE, Rasmussen M, Nielsen JF,
ness, and physostigmine-induced consciousness, we con-
Gjedde A. Effects of subanaesthetic and anaesthetic doses of
clude that unconsciousness caused by propofol is
sevoflurane on regional cerebral blood flow in healthy volunteers.
specifically associated with reductions in rCBF in the thala- A positron emission tomographic study. Acta Anaesthesiol Scand
mus, precuneus, and likely in the lateral parietal cortex as 2004; 48: 1268– 76
well. These observations suggest that these structures are 9 Franks NP. General anaesthesia: from molecular targets to neur-
key elements of thalamocortical–corticothalamic networks onal pathways of sleep and arousal. Nat Rev Neurosci 2008; 9:
that sustain consciousness and that are critically affected 370– 86
during anaesthetic-induced unconsciousness. Furthermore, 10 Laureys S, Owen AM, Schiff ND. Brain function in coma, vegetative
the findings are consistent with the notion that state, and related disorders. Lancet Neurol 2004; 3: 537–46
anaesthetic-induced unconsciousness is associated with 11 Meuret P, Backman SB, Bonhomme V, Plourde G, Fiset P. Physo-
stigmine reverses propofol-induced unconsciousness and
reduced cholinergic activation.
attenuation of the auditory steady state response and bispectral
index in human volunteers. Anesthesiology 2000; 93: 708–17
Supplementary material 12 Chernik DA, Gillings D, Laine H, et al. Validity and reliability of the
observer’s assessment of alertness/sedation scale: study with
Supplementary material is available at British Journal of intravenous midazolam. J Clin Psychopharmacol 1990; 10:
Anaesthesia online. 244– 51

556
Functional imaging of anaesthetic action BJA
13 Tackley RM, Lewis GTR, Prys-Roberts C, Boaden RW, Dixon J, 25 Schiff ND. The neurology of impaired consciousness: challenges
Harvey JT. Computer controlled infusion of propofol. Br J for cognitive neuroscience. In: Gazzaniga MS, ed. The Cognitive
Anaesth 1989; 62: 46– 53 Neurosciences. Cambridge: MIT Press, 2004; 1121– 32
14 Brown JH, Taylor P. Muscarinic receptor agonists and antagonists. 26 Clark DL, Hosick EC, Rosner BS. Neurophysiological effects of
In: Hardman JG, Limbird LE, Molinoff PB, Ruddon RW, Gilman AG, different anesthetics in unconscious man. J Appl Physiol 1971;
eds. Goodman & Gilman’s The Pharmacological Basis of Thera- 31: 884–91
peutics. New York: McGraw-Hill, 1996; 141–60 27 Angel A. The G.L. Brown lecture: adventures in anaesthesia. Exp
15 Asthana S, Greig NH, Hegedus L, et al. Clinical pharmacokinetics Physiol 1991; 76: 1– 38
of physostigmine in patients with Alzheimer’s disease. Clin Phar- 28 Alkire MT, McReynolds JR, Hahn EL, Trivedi AN. Thalamic microin-
macol Ther 1995; 58: 299– 309 jection of nicotine reverses sevoflurane-induced loss of righting
16 Vafaee M, Murase K, Gjedde A, Meyer E. Dispersion correction reflex in the rat. Anesthesiology 2007; 107: 264–72
for automatic sampling of O-15 labeled H2O and red blood 29 Schiff ND, Giacino JT, Kalmar K, et al. Behavioural improvements
cells. In: Myers R, Cunningham V, Bailey D, Jones T, eds. with thalamic stimulation after severe traumatic brain injury.
Quantification of Brain Function using PET. San Diego: Academic Nature 2007; 448: 600– 3
Press, 1996; 72– 5 30 Destexhe A. Modelling corticothalamic feedback and the gating
17 Ohta S, Meyer E, Fujita H, Reutens DC, Evans A, Gjedde A. Cerebral of the thalamus by the cerebral cortex. J Physiol Paris 2000; 94:
[15O]water clearance in humans determined by PET: I. Theory and 391– 410
normal values. J Cereb Blood Flow Metab 1996; 16: 765–80 31 Maquet P. Functional neuroimaging of normal human
18 Talairach J, Tournoux P. Co-planar Stereotactic Atlas of the Human sleep by positron emission tomography. J Sleep Res 2000; 9:
Brain. New York: Thieme, 1988 207– 31
19 Collins DL, Neelin P, Peters TM, Evans AC. Automatic 3D intersub- 32 Cavanna AE, Trimble MR. The precuneus: a review of its
ject registration of MR volumetric data in standardized Talairach functional anatomy and behavioural correlates. Brain 2006;
space. J Comput Assist Tomogr 1994; 18: 192 –205 129: 564–83
20 Worsley KJ, Evans AC, Marrett S, Neelin P. A three-dimensional 33 Gusnard DA, Raichle ME. Searching for a baseline: functional
statistical analysis for CBF activation studies in human brain. imaging and the resting human brain. Nat Rev Neurosci 2001;
J Cereb Blood Flow Metab 1992; 12: 900 –18 2: 685–94
21 Poline JB, Worsley KJ, Evans AC, Friston KJ. Combining spatial 34 Everitt BJ, Robbins TW. Central cholinergic systems and cognition.
extent and peak intensity to test for activations in functional Annu Rev Psychol 1997; 48: 649–84
imaging. Neuroimage 1997; 5: 83 –96 35 Xie G, Gunn RN, Dagher A, et al. PET quantification of muscarinic
22 Bonhomme V, Fiset P, Meuret P, et al. Propofol anesthesia and cholinergic receptors with [N-11C-methyl]-benztropine and appli-
cerebral blood flow changes elicited by vibrotactile stimulation: cation to studies of propofol-induced unconsciousness in healthy
a positron emission tomography study. J Neurophysiol 2001; 85: human volunteers. Synapse 2004; 51: 91–101
1299– 308 36 Kikuchi T, Wang Y, Sato K, Okumura F. In vivo effects of propofol
23 Ogawa K, Uema T, Motohashi N, et al. Neural mechanism of pro- on acetylcholine release from the frontal cortex, hippocampus
pofol anesthesia in severe depression: a positron emission tomo- and striatum studied by intracerebral microdialysis in freely
graphic study. Anesthesiology 2003; 98: 1101–11 moving rats. Br J Anaesth 1998; 80: 644–8
24 Veselis RA, Feshchenko VA, Reinsel RA, Dnistrian AM, Beattie B, 37 Murasaki O, Kaibara M, Nagase Y, et al. Site of action of the
Akhurst TJ. Thiopental and propofol affect different regions of general anesthetic propofol in muscarinic M1 receptor-
the brain at similar pharmacologic effects. Anesth Analg 2004; mediated signal transduction. J Pharmacol Exp Ther 2003; 307:
99: 399–408 995– 1000

557