- Journal List
- Br J Anaesth
- PMC11140841
As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice
Br J Anaesth. 2023 May; 130(5): 557–566.
Published online 2023 Mar 24. doi:10.1016/j.bja.2023.01.021
PMCID: PMC11140841
PMID: 36967282
Christian S. Guay,1,2,∗ Darren Hight,3 Gaurang Gupta,4 MohammadMehdi Kafashan,4 Anhthi H. Luong,5 Michael S. Avidan,4,6 Emery N. Brown,1,2 and Ben Julian A. Palanca4,6,7,8,9
Author information Article notes Copyright and License information PMC Disclaimer
Associated Data
- Supplementary Materials
Abstract
Background
Conscious states are typically inferred through responses to auditory tasks and noxious stimulation. We report the use of a stimulus-free behavioural paradigm to track state transitions in responsiveness during dexmedetomidine sedation. We hypothesised that estimated dexmedetomidine effect-site (Ce) concentrations would be higher at loss of responsiveness (LOR) compared with return of responsiveness (ROR), and both would be lower than comparable studies that used stimulus-based assessments.
Methods
Closed-Loop Acoustic Stimulation during Sedation with Dexmedetomidine data were analysed for secondary analysis. Fourteen healthy volunteers were asked to perform the breathe–squeeze task of gripping a dynamometer when inspiring and releasing it when expiring. LOR was defined as five inspirations without accompanied squeezes; ROR was defined as the return of five inspirations accompanied by squeezes. Brain states were monitored using 64-channel EEG. Dexmedetomidine was administered as a target-controlled infusion, with Ce estimated from a pharmaco*kinetic model.
Results
Counter to our hypothesis, mean estimated dexmedetomidine Ce was lower at LOR (0.92 ng ml−1; 95% confidence interval: 0.69–1.15) than at ROR (1.43 ng ml−1; 95% confidence interval: 1.27–1.58) (paired t-test; P=0.002). LOR was characterised by progressively increasing fronto-occipital EEG power in the 0.5–8 Hz band and loss of occipital alpha (8–12 Hz) and global beta (16–30 Hz) power. These EEG changes reverted at ROR.
Conclusions
The breathe–squeeze task can effectively track changes in responsiveness during sedation without external stimuli and might be more sensitive to state changes than stimulus-based tasks. It should be considered when perturbation of brain states is undesirable.
Clinical trial registration
Keywords: consciousness, dexmedetomidine, EEG, monitor, responsiveness, sedation, sensory disconnection, sleep
Editor's key points
•
The breathe-squeeze task facilitates tracking of spontaneous transitions in responsiveness during sedation.
•
According to the theory of neural inertia, the authors hypothesized that volunteers would regain responsiveness at lower concentrations of dexmedetomidine than those required to induce loss of responsiveness.
•
See AlsoN2 Sleep: K Complexes and Sleep Spindles | SleepopolisSleep Spindle - Knowledge and References | Taylor & FrancisDementia/delerium/sleep disorders/health anxiety disorders Flashcards by Leslie ManciniCounter to the hypothesis, volunteers regained responsiveness at higher concentrations of dexmedetomidine than required to induce hypnosis.
•
Change in responsiveness coincided with broad band changes in electroencephalogram power.
Anaesthetics serve as powerful tools in the clinical arena and the scientific study of consciousness. Our understanding of the neural correlates of conscious state transitions in humans has largely been advanced by sedation studies of healthy volunteers.1 Differentiating drug- and state-related effects of anaesthetics depends on the behavioural paradigm used to infer a participant's state of consciousness.2 Participant performance of behavioural tasks is integral for identification of conscious state transitions and temporal correlation to neurophysiological signals. The behavioural paradigm typically involves presentation of an external stimulus (auditory or noxious somatosensory) with classification of the evoked behavioural response as correct, incorrect, or absent.3 After a predefined number of incorrect or absent responses, the point of loss of responsiveness (LOR) is noted. Conversely, return of responsiveness (ROR) is designated after a series of correct responses. Although this paradigm closely mirrors clinical practice, it suffers from a fundamental limitation: external stimuli used to assess responsiveness disrupt the very system that is being studied. Even when stimuli do not arouse participants, auditory stimuli presented during sedation and anaesthesia are processed in the cortex and drive changes in the EEG.4,5
To overcome this challenge, a task without external intervention was developed to allow tracking of conscious state transitions without risking the disruption of the sleep-onset process: the breathe–squeeze task (BST).6 The task required participants to squeeze a ball embedded with force sensors during inhalation and to release the ball on exhalation. In this way, the group-level EEG changes could effectively be related to the sleep-onset process without iatrogenic arousals.
The EEG correlates during the onset and emergence from the sedative anaesthetic dexmedetomidine resemble those of natural sleep. Nonetheless, previous studies describing the pharmacodynamics and EEG correlates of these processes relied on stimulus-based behavioural paradigms during drug titration.2,7, 8, 9, 10, 11, 12, 13 Thus, questions remain on the impact of an external stimulus-based task on the state transitions induced by dexmedetomidine (i.e. sedation onset and offset). We hypothesised that estimated dexmedetomidine concentrations at LOR and ROR would be lower than those reported previously using stimulus-based assessments. We also hypothesised that estimated dexmedetomidine concentrations would be higher at LOR than at ROR, consistent with the theory of neural inertia.14,15
Here, we address the presented hypotheses and the following aims: (i) assess the feasibility of implementing BST during dexmedetomidine sedation, (ii) characterise dexmedetomidine pharmacodynamics during BST, and (iii) describe the group-level sedation onset and offset processes by time locking EEG data to BST performance.
Methods
Study design and participants
Closed-Loop Acoustic Stimulation during Sedation with Dexmedetomidine (CLASS-D) is a prospective, within-subject, crossover, controlled, interventional trial primarily designed to assess the effects of closed-loop acoustic stimulation (CLAS) during sedation.16 The study was approved by the Human Research Protection Office (HRPO) at Washington University inSt Louis (Institutional Review Board: 2019007086) and registered on ClinicalTrials.gov (NCT04206059) before enrolment. The present paper focuses on the pharmacodynamics and EEG correlates of BST during sedation onset and offset when no acoustic stimulation was provided. The parent trial outcomes, including effects of CLAS during sedation and effects of dexmedetomidine on subsequent sleep, will be reported separately.
Healthy volunteers were recruited using posted fliers, internet advertisem*nts, and university volunteer databases. Participants provided written informed consent before any study procedures. Volunteers were eligible if they were able to provide informed consent, were between the ages of 18 and 40 yr, and were classified as ASA physical status 1–2 by a board-certified anaesthesiologist. Exclusion criteria included diagnosed sleep disorders, psychiatric disorders, and hearing disorders; pregnant or nursing female; habitually short sleeper (<6 h/night); use of psychoactive medication, recreational drugs, or nicotine; BMI >30; neck circumference >40 cm, average heart rate less than 40 beats min−1 during non-rapid eye movement sleep Stage 3 (N3) sleep; and a heat pain tolerance threshold greater than 50°C. Once eligibility was confirmed, each participant was scheduled for a sedation session on a separate day and instructed to adhere to standard ASA fasting guidelines.
Breathe–squeeze task
Breathe–squeeze task was conducted using handgrip dynamometry instead of a ball embedded with force sensors.6 The hand dynamometer (Fig 1a) registered handgrip force and duration (Vernier, Beaverton, OR, USA), as in a previous trial that tracked volitional control during induction of general anaesthesia with propofol.17 During BST, participants were tasked to squeeze the dynamometer briefly during inhalation and release grip during exhalation. This averted handgrip fatigue accompanied by prolonged gripping. Participants were also instructed to keep their eyes closed during the entire experiment and resume the task whenever aware of their breathing or environment. Once the dexmedetomidine infusion was started, participants were no longer instructed on the behavioural task. A research team member monitored the dynamometric time series (Fig 1b) and registered successful or failed BST trials in the EEG record. Loss of responsiveness was defined as five inspirations without accompanied squeezes; return of responsiveness after cessation of dexmedetomidine was defined as five inspirations accompanied by squeezes. These respective surrogates of loss and return of consciousness were time-aligned with EEG data.
Fig 1
Handgrip dynamometer for detecting breath–squeeze task performance. (a) Handgrip dynamometer, adapted from Guay and Plourde.17 (b) Representative participant performing the breathe–squeeze task using a handgrip dynamometer. Participants are asked to briefly squeeze the dynamometer when they breathe in. Force (N) tracings are tracked over time using a graphical user interface, shown on the computer screen.
Target-controlled dexmedetomidine sedation
All study procedures were completed in a clinical environment equipped with standard monitoring equipment, suction, airway equipment, and a dedicated board-certified anaesthesiologist, per ASA standard of practice guidelines. On the experiment day, eligibility and fasting status was confirmed. Each participant had two i.v. catheters placed. Participants were also instrumented for EEG monitoring.
A target-controlled infusion (TCI) protocol was used to initiate and maintain sedation. The TCI pump (Harvard Apparatus 22; Harvard Apparatus, South Natick, MA, USA) was controlled by RugloopII software (Demed, Temse, Belgium) to infuse dexmedetomidine (16 μg ml−1) according to the 2015 Hannivoort pharmaco*kinetic model that accounts for subjects' age, sex, height, and weight.18 To minimise haemodynamic effects, RugloopII limited the maximal dexmedetomidine infusion rate to 6 mcg kg−1 h−1. The TCI protocol began with a target estimated effect-site concentration (Ce) of 1.5 ng ml−1 for all participants. Once this target was reached, the concentration was increased in increments of 0.5 ng ml−1 until two endpoints were met: LOR and induction of large amplitude slow waves of >40 μV amplitude and 0.5–4 Hz frequency. After achieving the desired brain state, the CLAS protocol began and lasted a mean (standard deviation [sd]) of 132 (30) min. The CLAS protocol has been previously described in detail.16 Briefly, acoustic stimuli (60 dB pink noise) were delivered in-phase and anti-phase to EEG slow waves in a balanced, within-subject, crossover design. Participants were aroused three times during the CLAS protocol using a linear thermal ramp: once during in-phase stimulation, once during anti-phase stimulation, and once during sham stimulation. After completion of the CLAS protocol, the TCI was set to 0 ng ml−1, and participants were monitored without intervention until ROR. A modified Brice questionnaire was immediately administered after ROR.19 Participants were discharged after meeting standard institutional post-anaesthesia discharge criteria.
High-density EEG acquisition, preprocessing, and measures
Before sedation, participants were fitted with a 65-electrode EEG geodesic scalp net (Electrical Geodesics, Inc., Eugene, OR, USA). Conductive gel (Nihon Kohden, Tokyo, Japan) was used to maintain impedances ≤50 kΩ.
EEG data were recorded using Net Amps 400 amplifiers and Net Station 5.4 software (Electrical Geodesics, Inc., Eugene, OR, USA) with a sampling rate of 512 Hz. Data were online referenced to Cz and filtered using a 60 Hz notch filter, a 120 Hz low-pass filter, and a 0.1 Hz high-pass filter. Tags were directly entered into the NetStation software during the experiment to note the timing of study events, such as starting and stopping the dexmedetomidine TCI and performance on the behavioural task.
The EEG data were imported into MATLAB (MathWorks, Natick, MA, USA) via EEGLAB.22 EEG preprocessing included a 0.1 high-pass filter, a 30 Hz low-pass filter, and downsampling to 250 Hz. Data points with an amplitude greater than 250 μV were considered as artifact and rejected. Signals were re-referenced to bipolar montages for group-level spectra and spectrograms, and to the average for voltage topograms.
Multi-taper spectral analyses were performed with window lengths of T=4 s with 3 s overlap, time-bandwidth product TW=2, number of tapers K=3, and spectral resolution of 2 Hz in the Chronux toolbox.23 Group-level spectrograms were generated by taking the median across subjects, and group-averaged spectra were computed by taking the median of group-level spectrograms across an entire epoch of interest.12 Scalp power distributions computed from group-averaged spectra were generated using interpolation of the electrode montage with the topoplot function in EEGLAB. Group-level topograms were generated using the median spectrum for six 1 min time windows: LOR–5 (5 min pre-LOR to 4 min pre-LOR), LOR–1 (1 min pre-LOR to LOR), LOR+1 (LOR to 1 min post-LOR), LOR+5 (4 min post-LOR to 5 min post-LOR), ROR–5 (5 min pre-ROR to 4 min pre-ROR), and ROR–1 (1 min pre-ROR to ROR).
Statistical analysis
The pharmacodynamics and EEG correlates of BST performance during dexmedetomidine sedation are exploratory outcomes of the CLASS-D trial. Therefore, no sample size calculation was performed for these outcome measures.
The Kolmogorov–Smirnov test was used to confirm that the group distributions of Ce at LOR and ROR were not significantly different from a normal distribution. Subsequently, paired t-tests were used to compare Ce at LOR with Ce at ROR. The cumulative distribution functions for LOR and ROR were computed using the MATLAB ecdf function.
Group-level spectra and 95% confidence intervals (CIs) were generated according to the bootstrapping approach described by Akeju and colleagues.7 Briefly, we randomly selected spectra with replacement from within spectrograms over the time periods of interest. We then took the median spectra over time from the new spectrograms for each subject, and we calculated differences between the two median spectra (either ‘LOR–5’ minus ‘LOR+5’, or ‘ROR–5’ minus ‘ROR–1’) for each subject. This allowed calculation of a median difference across subjects. This procedure was repeated 2000 times, and we then calculated the 95% CI of the median difference at each frequency.
Results
Between January 23, 2021 and May 22, 2022, 16 participants were recruited, and 14 participants completed all study procedures. Two participants did not contribute data to the analysis, as sessions were terminated early: one because of a strong urge to urinate and the other because of paradoxical agitation.
Estimated dexmedetomidine Ce and Cp at loss and return of responsiveness
Time-aligned data of LOR and ROR for one example participant are shown in Fig 2. The mean (sd) time from the start of the dexmedetomidine TCI to LOR was 12.6 (5.7) min, and from the end of the dexmedetomidine TCI to ROR was 40.5 (14.8) min.
Fig 2
Representative time-locked data for LOR and ROR in an exemplar participant. The top panel shows a power spectrogram of the bipolar channel O1–O2. The bottom panel is an overlay of handgrip dynamometry (yellow tracing; left y-axis) along with estimated plasma (green) and effect-site (blue) concentrations (right y-axis). LOR, defined as five successive breaths without concurrent dynamometer squeezes, and ROR, defined as five successive breaths with concurrent dynamometer squeezes, are indicated in the figure by red lines. In this example, the TCI of dexmedetomidine was set to an effect-site target of 2 ng ml−1. During the sedation-onset process, the plasma concentration quickly increases beyond this target as the effect-site compartment equilibrates with plasma. Once the target is achieved, the infusion rate automatically adjusts to maintain a steady-state equilibrium between plasma and effect-site compartments. During the sedation-offset process, the dexmedetomidine infusion is turned off and the plasma concentration quickly decreases. The decline in effect-site concentration lags behind the plasma concentration as it equilibrates with this compartment. Dex, dexmedetomidine; LOR, loss of responsiveness; ROR, return of responsiveness; TCI, target-controlled infusion.
We hypothesised that estimated Ce would be higher at LOR than at ROR, given the theory of neural inertia.14,15 Contrary with this hypothesis, the mean (95% CI) estimated Ce was 0.92 (0.69–1.15) ng ml−1 at LOR and 1.43 (1.27–1.58) at ROR (P=0.002; paired t-test; Fig 3). These findings are further demonstrated in the cumulative distribution functions of Ce at LOR and ROR, where a clear right shift to higher concentrations is seen for ROR compared with LOR (Fig 3).
Fig 3
Cumulative distribution function for LOR vs ROR according to the estimated effect-site concentration (Ce). The y-axis represents the proportion of participants performing the breathe–squeeze task at the corresponding Ce on the x-axis. The inset box shows that the mean (95% confidence interval) Ce at LOR was lower than at ROR. LOR, loss of responsiveness; ROR, return of responsiveness.
With regard to plasma concentrations (Cp), the mean (95% CI) Cp was 2.50 (2.16–2.85) ng ml−1 at LOR and 1.13 (1.00–1.25) ng ml−1 at ROR. We did not perform statistical comparisons for Cp because of the concern for equilibration delay between plasma and effect-site compartments. Specifically, because LOR and ROR did not occur with dexmedetomidine concentrations at steady states, we expect Cp to overestimate Ce at LOR and to underestimate Ce at ROR (Fig 2).
Time-aligning EEG to loss and return of responsiveness demonstrates spectral changes at the group level
We next leveraged high-density EEG to evaluate topographic and spectral changes at LOR and ROR. Power spectrograms for 5 min periods flanking LOR are shown in Fig 4 with group-level scalp topograms of conventional frequency band-limited power shown for 1 min epochs of EEG. The sedation-onset process from LOR–5 to LOR+5 exhibits gradually increasing slow wave activity (SWA) (0.5–4 Hz) and theta (4–8 Hz) power in a fronto–central–occipital pattern and a global decrease in beta (16–30 Hz) power. The alpha (8–12 Hz) and sigma (12–16 Hz) bands both decrease, leading up to LOR and then increase after LOR, reflecting the appearance of dexmedetomidine spindles. The posterior dominant rhythm (PDR)20 is visible as occipital alpha power and disappears before LOR.
Fig 4
Group-level median occipital spectrogram and topograms during the sedation-onset process. The spectrogram begins 5 min before LOR (denoted by a central black line) and ends 5 min following LOR. Topograms for the traditional EEG frequency bands were generated for 1 min intervals, denoted by dashed lines: LOR–5 (5–4 min before LOR), LOR–1 (the final minute before LOR), LOR+1 (the first minute after LOR), and LOR+5 (4–5 min after LOR). The sedation-onset process is characterised by progressively increasing slow wave activity (0.5–4 Hz) and theta (4–8 Hz) power in the frontal, central, and occipital regions. The posterior dominant rhythm, most prominent in the alpha (8–12 Hz) band, disappears before LOR. Global sigma (12–16 Hz) and beta (16–30 Hz) power also decrease before LOR. The return of sigma power at LOR+5 represents the onset of dexmedetomidine spindles. LOR, loss of responsiveness.
For ROR, power spectrograms and group-level scalp topograms are shown in Fig.5. The sedation-offset process is marked by a reduction in SWA and theta power. Sigma power also declines, consistent with the disappearance of spindles. A reappearance of the PDR as occipital alpha power and return of global beta power are also observed after ROR. Whereas the group-level PDR appears to progressively wane during the LOR process, it reappears abruptly approximately 40 s before ROR, likely heralding the onset of successful breathe–squeeze trials before the fifth squeeze required to fulfill the endpoint of ROR.
Fig 5
Group-level median occipital spectrogram and topograms during the sedation-offset process. The spectrogram begins 5 min before ROR (denoted by a central black line) and at ROR. Topograms for the traditional EEG frequency bands were generated for 1 min intervals, denoted by dashed lines: ROR–5 (5–4 min before ROR) and ROR–1 (the final minute before ROR, defined as five dynamometer squeezes synchronised with breathing). The sedation-offset process is characterised by a decrease in slow wave activity (0.5–4 Hz) and theta (4–8 Hz) power in the frontal, central, and occipital regions; return of the posterior dominant rhythm in the alpha (8–12 Hz) band; disappearance of spindles in the sigma (12–16 Hz) band; and increase in global beta (16–30 Hz) power.
Considering that occipital channels demonstrated the most marked changes during sedation onset and offset, group-level spectra (Supplementary Fig 1) were generated for an occipital bipolar channel (O1–O2). Fig.6 exhibits the bootstrapped 95% confidence bounds for the within-subject difference between LOR–5 and LOR+5 (Fig.6a) and between ROR–5 and ROR–1 (Fig.6b). Statistically significant differences are marked by an asterisk, recapitulating the changes observed in group-level spectrograms and topograms: a reduction in occipital alpha and beta power is accompanied by increasing delta and theta power during sedation onset, whereas an increase in occipital alpha and beta power is accompanied by decreasing delta and theta power during sedation offset.
Fig 6
Spectral transitions for LOR and ROR. (a) Bootstrapped 95% confidence interval bounds for the difference between median occipital (O1–O2) spectra of ‘LOR–5’ and ‘LOR+5’. Asterisks mark statistically significant differences. The sedation-onset process is marked by increasing power in the 0.5–8 Hz range and decreasing power in the 9–14 and 16–30 Hz ranges. (b) Bootstrapped 95% confidence interval bounds for the difference between median occipital (O1–O2) spectra of ‘ROR–5’ and ‘ROR–1’. Asterisks mark statistically significant differences. The sedation-offset process is marked by decreasing power in the 0.5–8 Hz range and scattered increases in frequencies above 9 Hz. LOR, loss of responsiveness; ROR, return of responsiveness.
Subjective report of task performance
Using the Brice questionnaire, we sought qualitative subjective recall of performance of the BST. Responses to the Brice questionnaire were consistent across participants. When asked, ‘What was the last thing you remember before anaesthesia?‘, every participant reported memories related to focusing on completing the BST. Likewise, when asked, ‘What is the first thing you remember after waking up?‘, every participant reported remembering that they should be performing the BST. Remarkably, all participants remembered to squeeze the dynamometer unprompted and did so whilst keeping their eyes closed, consistent with the reappearance of the PDR in occipital EEG channels (Fig.5). Five participants reported dreaming during the study procedures.
Discussion
Human studies probing anaesthetic mechanisms and pharmacologic manipulation almost exclusively use stimulus-based assessments of consciousness. We demonstrated the feasibility of implementing a stimulus-free behavioural paradigm during dexmedetomidine sedation.
This represents a substantial improvement on previous paradigms for two reasons. First, BST detects distinct and earlier changes in brain states than stimulus-based methods. Although states of clinical anaesthesia are designed to prevent responses to noxious environmental stimuli, anaesthetics are also used to study fundamental mechanisms of consciousness and self-initiated behaviours. The BST will be particularly useful in the latter, allowing investigators to probe state changes that precede unresponsiveness to external stimuli. BST could also be combined with stimulus-based tasks to assess sequential state changes during induction of anaesthesia (i.e. transitioning to a stimulus-based task after participants stop performing BST). Second, it allows investigators to study the sedation-onset and sedation-offset processes without the confounding effects of external stimuli. Possible disruption of the sedation-onset process by the stimulus-based task could be inferred by fluctuations of the PDR and reappearance after auditory cues in previous studies.2 Such fluctuations in spectral power may account for the lack of consistent EEG changes during state transitions.7 In the current study, we were able to generate group-level spectra, spectrograms, and topograms demonstrating undisturbed sedation-onset and sedation-offset processes. Spectral and spatial analyses recapitulated and extended previous findings, demonstrating that time-aligning EEG data to performance on BST can enable accurate group-level analyses.
We expected that Ce would be greater at LOR than at ROR because of the theory of neural inertia, which states that ‘transitions into and out of the anaesthetized state exhibit resistance to states transitions’.15 We had also expected that participants would not remember to perform the BST after regaining consciousness, resulting in an even lower Ce at ROR. Contrary to our hypothesis, dexmedetomidine Ce were lower at LOR compared with ROR. One possible explanation is that participants' ability to perform self-initiated behaviours leverages different neural mechanisms than those subserving responses to external stimuli. Another possibility is that the CLAS protocol may have altered neural dynamics in such a way that participants were primed to resume spontaneous behaviour more quickly than if they had not received CLAS. The mean (sd) time from the termination of CLAS and dexmedetomidine infusion to ROR was 40.5 (14.8) min. Therefore, CLAS would have had to induce a lasting impact on neural processes, which seems unlikely based on the sleep literature.21 Nonetheless, we cannot rule out this possibility. Finally, it is worth noting that LOR and ROR did not occur at pharmaco*kinetic steady states, and therefore, our Ce measurements rely heavily on equilibration parameters in the Hannivoort model.18 Considering these limitations and the fact that our experimental protocol was not specifically designed to test neural inertia, our results should be interpreted with caution and sought to be replicated to properly challenge the theory of neural inertia.
Estimated dexmedetomidine concentrations are lower at LOR and ROR for the BST compared with an auditory response task. We expect performance on the BST to be more sensitive to sedatives than any stimulus-based task because the participant is allowed to transition into a state of unresponsiveness without disruption by the external environment. To compare Ce between breathe–squeeze and auditory response tasks, we compared our data with those reported by Scheinin and colleagues2 using unpaired t-tests. In their study, they used ‘target controlled infusions, aiming at stepwise escalating pseudo steady-state plasma concentrations at 7-min intervals until loss of responsiveness’.2 Therefore, Ce and Cp were assumed to be at equilibrium for their predicted Cp values at LOR (i.e. Ce ∼ Cp). In our study, the reported Cp values at LOR were predictably higher than Ce because they had not yet reached equilibrium, as shown for the example subject in Fig.2. Because Scheinin and colleagues2 assumed that Cp ∼ Ce at LOR, we compared their Ce values with our Ce values at LOR. In accordance with our hypothesis, Ce values at LOR determined through the BST (mean 0.92 ng ml−1; sd 0.44) were lower than values at LOR assayed by the auditory response task (mean 1.67 ng ml−1; sd 0.54; Supplementary Fig 2; unpaired t-test; P<0.001).
Conversely, for ROR, Scheinin and colleagues2 stopped the TCI and let dexmedetomidine concentrations decrease passively, as was done in our study. Scheinin and colleagues did not report the predicted Ce at ROR, precluding a comparison of these values with our study. However, because both studies shared identical TCI methods for the sedation-offset portion of the experiments (i.e. turning off the dexmedetomidine infusion and letting Cp decrease without any stepwise pseudo-steady states), the Cp values at ROR could be directly compared between studies. As expected, the predicted Cp at ROR was higher when detected by the stimulus-based task (mean 1.92 ng ml−1; sd 0.85) of Scheinin and colleagues compared with the BST (mean 1.13 ng ml−1; sd 0.23; Supplementary Fig 2; unpaired t-test; P=0.002). In summary, performance of the BST is more sensitive to the sedative effects of dexmedetomidine than performance of an auditory-based task. Whether this finding generalises to other stimulus-based tasks should be investigated in future studies.
Limitations
There are limitations to consider when interpreting our findings. Whilst we estimated Ce and Cp with pharmaco*kinetic modelling, our protocol did not include blood sampling at LOR and ROR. Nonetheless, our results can help guide future studies using dexmedetomidine TCI.
As with most human sedation volunteer studies, our sample size of 14 limits statistical power and generalisability. However, this approach allowed us to study the pharmacodynamics of BST in a ‘clean’ sample of participants without the confounders of comorbidities and polypharmacy inherent in clinical studies. Our group-level spectrograms and topograms also offer insight into the spectral and spatial processes underlying LOR and ROR as measured with BST.
Conclusion
In summary, our investigation supports and extends previous studies of dexmedetomidine pharmacodynamics with a more sensitive, stimulus-free behavioural paradigm. The breathe-squeeze test should be considered when external perturbation of brain states is not feasible or desired. Future studies might also incorporate breathe-squeeze test into behavioural paradigms that probe multiple state transitions using stimulus-free and stimulus-based assessments.
Authors’ contributions
Design/planning: CSG, MSA, ENB, BJAP.
Data acquisition: CSG, AHL, MMK, BJAP.
Data analysis: CSG, DH, GG, MMK.
Data interpretation: CSG, DH, BJAP.
Writing of paper: all authors.
Critically revising of paper: all authors.
Acknowledgements
The authors appreciate the contributions of Mike Prerau, Eric Landsness, and Brendan Lucey in experimental design. Efforts of Alyssa K. Labonte and Thomas Nguyen towards data collection and handling are also acknowledged. The authors also express their gratitude to Harry Scheinin and Max Kelz for providing guidance in interpretation of results and manuscript revisions.
Notes
Handling editor: Robert Sanders
Footnotes
☆A preliminary account of this work was submitted as an abstract and presented at the 11th International Symposium on Memory and Awareness in Anesthesia (Prato, Italy) on June 6, 2022.
Appendix ASupplementary data to this article can be found online at https://doi.org/10.1016/j.bja.2023.01.021.
Declarations of interest
ENB was involved in the start-up company NEURADIA that is investigating drugs to induce wake up from anaesthesia and a start-up company that is developing systems for control of physiological states during anaesthesia. ENB has pending patents US20200187853A1 and US20190374158A1, and he receives royalties from Masimo for pending patent US16373498 and issued patent US10299720. CSG, ENB, and BJAP report pending patent US 17/128,845. BJAP reports a research trial study agreement between Washington University and Elemind Technologies, Inc., specifying Elemind's time-limited exclusive option on intellectual property developed during Closed-Loop Acoustic Stimulation during Sedation with Dexmedetomidine. Elemind did not provide funding for the study. No intellectual property reported by the investigators is directly related to the breathe–squeeze task. The authors report no other conflicts of interest in this work.
Funding
McDonnell Center for Systems Neuroscience at Washington University in St Louis to BJAP.
Appendix A. Supplementary data
The following is the Supplementary data to this article.
Multimedia component 1:
Click here to view.(5.9M, docx)Multimedia component 1
References
1. Casey C.P., Tanabe S., Farahbakhsh Z., et al. Distinct EEG signatures differentiate unconsciousness and disconnection during anaesthesia and sleep. Br J Anaesth. 2022;128:1006–1018. [PMC free article] [PubMed] [Google Scholar]
2. Scheinin A., Kallionpää R.E., Li D., et al. Differentiating drug-related and state-related effects of dexmedetomidine and propofol on the electroencephalogram. Anesthesiology. 2018;129:22–36. [PubMed] [Google Scholar]
3. Mashour G.A., Palanca B.J.A., Basner M., et al. Recovery of consciousness and cognition after general anesthesia in humans. Elife. 2021;10 [PMC free article] [PubMed] [Google Scholar]
4. Plourde G., Belin P., Chartrand D., et al. Cortical processing of complex auditory stimuli during alterations of consciousness with the general anesthetic propofol. Anesthesiology. 2006;104:448–457. [PubMed] [Google Scholar]
5. Yppärilä H., Karhu J., Westerén-Punnonen S., Musialowicz T., Partanen J. Evidence of auditory processing during postoperative propofol sedation. Clin Neurophysiol. 2002;113:1357–1364. [PubMed] [Google Scholar]
6. Prerau M.J., Hartnack K.E., Obregon-Henao G., et al. Tracking the sleep onset process: an empirical model of behavioral and physiological dynamics. PLoS Comput Biol. 2014;10 [PMC free article] [PubMed] [Google Scholar]
7. Akeju O., Kim S.-E., Vazquez R., et al. Spatiotemporal dynamics of dexmedetomidine-induced electroencephalogram oscillations. PLoS One. 2016;11 [PMC free article] [PubMed] [Google Scholar]
8. Sleigh J.W., Vacas S., Flexman A.M., Talke P.O. Electroencephalographic arousal patterns under dexmedetomidine sedation. Anesth Analg. 2018;127:951–959. [PubMed] [Google Scholar]
9. Xi C., Sun S., Pan C., Ji F., Cui X., Li T. Different effects of propofol and dexmedetomidine sedation on electroencephalogram patterns: wakefulness, moderate sedation, deep sedation and recovery. PLoS One. 2018;13 [PMC free article] [PubMed] [Google Scholar]
10. Maksimow A., Snapir A., Särkelä M., et al. Assessing the depth of dexmedetomidine-induced sedation with electroencephalogram (EEG)-based spectral entropy. Acta Anaesthesiol Scand. 2007;51:22–30. [PubMed] [Google Scholar]
11. Kallionpää R.E., Valli K., Scheinin A., et al. Alpha band frontal connectivity is a state-specific electroencephalographic correlate of unresponsiveness during exposure to dexmedetomidine and propofol. Br J Anaesth. 2020;125:518–528. [PubMed] [Google Scholar]
12. Kim W.-H., Cho D., Lee B., Song J.-J., Shin T.J. Changes in brain activation during sedation induced by dexmedetomidine. JInt Med Res. 2017;45:1158–1167. [PMC free article] [PubMed] [Google Scholar]
13. Colin P.J., Hannivoort L.N., Eleveld D.J., et al. Dexmedetomidine pharmaco*kinetic–pharmacodynamic modelling in healthy volunteers: 1. influence of arousal on bispectral index and sedation. Br J Anaesth. 2017;119:200–210. [PubMed] [Google Scholar]
14. Kelz M.B., Sun Y., Chen J., et al. An essential role for orexins in emergence from general anesthesia. Proc Natl Acad Sci U S A. 2008;105:1309–1314. [PMC free article] [PubMed] [Google Scholar]
15. Proekt A., Hudson A.E. Astochastic basis for neural inertia in emergence from general anaesthesia. Br J Anaesth. 2018;121:86–94. [PMC free article] [PubMed] [Google Scholar]
16. Guay C.S., Labonte A.K., Montana M.C., et al. Closed-Loop Acoustic Stimulation during Sedation with Dexmedetomidine (CLASS-D): protocol for a within-subject, crossover, controlled, interventional trial with healthy volunteers. Nat Sci Sleep. 2021;13:303–313. [PMC free article] [PubMed] [Google Scholar]
17. Guay C.S., Plourde G. Handgrip dynamometry for continuous assessment of volitional control during induction of anesthesia: a prospective observational study. Can J Anaesth. 2019;66:48–56. [PubMed] [Google Scholar]
18. Hannivoort L.N., Eleveld D.J., Proost J.H., et al. Development of an optimized pharmaco*kinetic model of dexmedetomidine using target-controlled infusion in healthy volunteers. Anesthesiology. 2015;123:357–367. [PubMed] [Google Scholar]
19. Brice D.D., Hetherington R.R., Utting J.E. Asimple study of awareness and dreaming during anaesthesia. Br J Anaesth. 1970;42:535–542. [PubMed] [Google Scholar]
20. Labonte A.K., Kafashan M., Huels E.R., et al. The posterior dominant rhythm: an electroencephalographic biomarker for cognitive recovery after general anaesthesia. Br J Anaesth. 2023;130:e233–e242. [PMC free article] [PubMed] [Google Scholar]
21. Ngo H.-V.V., Miedema A., Faude I., Martinetz T., Mölle M., Born J. Driving sleep slow oscillations by auditory closed-loop stimulation—a self-limiting process. JNeurosci. 2015;35:6630–6638. [PMC free article] [PubMed] [Google Scholar]
22. Delorme A., Makeig S. EEGLAB: an open source toolbox for analysis of single-trialEEG dynamics including independent component analysis. J Neurosci Methods. 2004;134:9–21. [PubMed] [Google Scholar]
23. Mitra P.P., Bokil H. 1st ed. Oxford University Press; New York: 2008. Observed Brain Dynamics. [Google Scholar]
Articles from BJA: British Journal of Anaesthesia are provided here courtesy of Elsevier
