| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Department of Intensive Care Medicine, University Hospital of Bern, Bern, Switzerland; and VTT Information Technology, Tampere, Finland.
Address correspondence and reprint requests to Stephan Jakob, Department of Intensive Care Medicine, Inselspital, Freiburgstrasse, CH-3010 Bern, Switzerland. Address e-mail to Stephan.Jakob{at}insel.ch.
 |
Abstract |
|---|
 Top Abstract IntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The 
-2 agonist dexmedetomidine (dex) has been approved for short-term sedation in surgical intensive care unit (ICU) patients. Preliminary data suggest that the risk of delirium may be substantially reduced when dexmedetomidine is used to produce sedation (12). Since dexmedetomidine acts via different receptors and brain areas than do benzodiazepines and propofol, its impact on the brain electrophysiology may also be different. The assessment of dexmedetomidine’s effects on the EEG and ERPs at various sedation levels has been limited in humans.
We hypothesized that dexmedetomidine and midazolam (mida), each given in combination with remifentanil (remi), would induce the same changes in EEG and long-latency ERPs during light-to-moderate levels of sedation in healthy subjects, despite the different quality of sedation that they provide. The opioid remifentanil was added because virtually all patients in the ICU have some level of pain and receive an opioid analgesic in combination with a sedative.
|
METHODS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
-glutamyltransferase, serum creatinine) were performed before inclusion in the study to exclude abnormal organ function. A drug screen was performed before each session. The study took place in the ICU of the University Hospital, with a dedicated intensive care medicine specialist responsible exclusively for the volunteers’ safety. The EEG signal was recorded using Ag/AgCl electrodes placed on the scalp according to the international 10–20 system. Two electrode locations—frontal (Fz) and central (Cz)—were used. Both electrodes were referred to the right mastoid, and the electrode-skin impedances were kept below 5 kohm. Automatic impedance checks were performed every 15 min. The EEG signal was amplified and digitalized continuously at 200 Hz with an EEG module attached to an S/5 monitor (Datex-Ohmeda, GE Healthcare, Helsinki, Finland). This EEG module can measure maximal four channels, has a range of ±400 mV, a frequency range of 0.5–40 Hz, a resolution of 60 mV, an input impedance of >10 MW, a noise level of <0.5 mV rms from 0.5 to 40 Hz and a CMRR (common mode rejection ratio) of >100 dB. Bispectral index (BIS) electrodes (Aspect Medical, S/5 BIS Module [XP-level]) were placed on the volunteer’s forehead. Arterial and venous catheters were inserted. Electrocardiogram, invasive arterial blood pressure, oxygen saturation, end-tidal carbon dioxide via nasal probe, and BIS were monitored with a standard patient monitor (S/5) and recorded with data recording software (Wincollect; Datex-Ohmeda, GE Healthcare). Supplemental oxygen was given to all volunteers via a nasal probe.
Four sets of measurements were taken from each volunteer: the first set was taken at baseline, followed by measurements at Ramsay scores (RS) 2, 3, and 4, in that order. Baseline measurements were recorded before any drugs were given; the volunteer remained supine with his eyes closed. Background EEG was recorded for 5 min. Afterwards, auditory stimulation was started and ERPs were recorded. The stimulation was applied according to both a habituation and a single-tone paradigm. In the habituation paradigm, four equal auditory stimuli (800 Hz) were applied through earphones at intervals of 1 s, followed by a pause, and this set of stimuli was repeated after 12 s. Altogether, 40 sets of stimuli were delivered at each measurement, corresponding to a recording time of about 10 min. In the single-tone paradigm, the same standard tone as described above was delivered 600 times with an interstimulus interval of 1 s, which also corresponds to a recording time of 10 min.
The background EEG was first band pass filtered with a finite impulse response type filter using cutoff frequencies of 0.5 and 47 Hz (Matlab, version 6.12, The Mathworks, Natick, MA). Five minutes of the filtered EEG signal was cut into 5-s epochs with 50% overlap. Obvious artifacts were excluded on the basis of the maximum amplitude for each epoch. Epochs with amplitudes more than the predefined limits (100–300 µV) were excluded. The appropriateness of the artifact rejection was manually confirmed.
For each EEG epoch, the root mean squared total power was calculated. The power spectral density was estimated for each epoch using Welsh’s averaged periodogram method, and spectral edge frequency 95% (SEF 95%) and mean power frequency were computed from the power spectral density using a frequency range from 0.5 to 47 Hz. In addition, the following spectral powers were computed: delta power (0.5–4 Hz), theta power (4–8 Hz), alpha power (8–13 Hz), beta1 power (13–20 Hz), Beta2 power (20–47 Hz), and the slow–fast ratio ([Delta + Theta]/[Alpha + Beta1 + Beta2]). The mean parameter values of the accepted epochs were individually computed, and used for further analysis. The EEG signal recorded during the auditory stimulation was first filtered using cutoff frequencies of 1 and 20 Hz, and then transformed to epochs lasting from –100 to 900 ms relative to each stimulus onset. After removal of epochs with artifacts (rejection level ±100 µV), the individual responses to the stimuli of the single-tone paradigm and the individual responses to the first and second stimuli of the habituation paradigm were averaged. The N100 component was defined as a maximum negative deflection appearing 80–150 ms after the stimulus onset. The amplitude and the latency of the prominent N100 components with respect to the prestimulus baseline (50 ms to onset of the tone) were manually recorded. In addition, the first-to-second-tone ratio of the habituation paradigm was computed by dividing the N100 amplitude of the first stimulus by the N100 amplitude of the second stimulus.
After the baseline recordings without sedation, infusion of either midazolam or dexmedetomidine in combination with remifentanil was started. Midazolam (Dormicum®, Roche, Basel, Switzerland), dexmedetomidine (Precedex®, Abbott Laboratories, Abbott Park, IL) and remifentanil (Ultiva®, Glaxo-SmithKline, Schönbühl, Switzerland) were administered by a computer-controlled pump (Alaris Asena®, ALARIS Medical Systems, Carbamed, Bern, Switzerland) driven by the Rugloop II TCI program (BVBA Demed, Temse, Belgium), with the pharmacokinetic parameter sets of Greenblatt et al. for midazolam (13), of Dyck et al. for dexmedetomidine (14), and of Minto et al. for remifentanil (15). In the first step, midazolam or dexmedetomidine was titrated individually to reach the sedation level RS 2 (modified Ramsay Sedation Scale; Table 1). Remifentanil infusion was targeted to reach fixed plasma concentration levels of 2 ng/mL throughout the whole session. We chose this plasma concentration on the basis of our experience from previous trials showing that with this dose hypoventilation does not occur, but an effective analgesia is present. A stabilization period of 15 min was maintained after each change in the doses of midazolam and dexmedetomidine, respectively. After this period, ERPs were measured. The corresponding expected drug concentrations were calculated on the basis of the respective pharmacokinetic models. Then the sedative drug was increased to reach the deeper sedation levels, RS 3 and RS 4. The remifentanil infusion was maintained at the targeted plasma concentration of 2 ng/mL. Given the relatively long half-lives of both sedative drugs, we did not randomize the order of the sedation level, but only increased the sedation level stepwise.
|
The targeted sedation level was reevaluated by two examiners before and after the ERP measurements by assessing the RS, with a glabellar tap and a pinch in the trapezius muscle to discriminate between RS 4 and 5. When discrepancies between the initial and the final determination at each level of sedation appeared, the RS before ERP measurements was used. Indications for discontinuation of drug infusion included too deep sedation (RS 5 or 6), inadequate oxygen saturation (Spo2 <92%), and airway obstruction.
Results are expressed as mean ± sd. Variables were tested for differences in 1) the drug regimes and 2) evolution with increasing sedation by using a two-factorial repeated measures ANOVA. The factors were defined as the following: 1) two drug regimes (dexmedetomidine/remifentanil and midazolam/remifentanil) and 2) four sedation levels (baseline, and RS 2, 3, and 4). In the case of significant differences between the drug regime groups, the evolution of the variable within a drug regime group was tested by a one-way repeated measures ANOVA. The statistics were computed with SPSS 12.0.1 for Windows (SPSS, Chicago, IL). A P-value of <0.05 was considered to be significant.
|
RESULTS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
At baseline, none of the electrophysiologic variables differed between dex/remi and mida/remi. During sedation, most of the EEG-derived parameters and N100 potentials changed similarly at frontal and central electrode locations. Except for the observed differences between the two locations, only the results for the frontal electrode are given for clarity.
During dex/remi sedation, the resting EEG signals intermittently demonstrated a characteristic high-power low-frequency pattern, which became more pronounced as the sedation level deepened (Fig. 1). This resulted in a visible peak at 3 Hz in the power spectra (Fig. 2). This peak was not large enough to cause a significant difference in the relative power in the 
-band (1–3 Hz; P = 0.18), in the SEF 95% (P = 0.55), or in the median frequency power (P = 0.89) between the drug combinations (Table 2).
|
|
|
BIS decreased to lower levels in the dex/remi group (from 94 ± 3 at baseline to 58 ± 14 at RS 4) than in the mida/remi group (from 94 ± 2 to 76 ± 10); main drug effect: P = 0.029, main sedation effect P = 0.001; drug*sedation interaction P = 0.004. Although the evolution of the BIS values demonstrated a significant reduction in BIS for both drug combinations (P = 0.002 for dex/remi and P = 0.004 for mida/remi), only in the dex/remi group did the BIS decrease uniformly, whereas in the mida/remi group some volunteers had an increasing BIS after an initial decrease, and the actual BIS values overlapped widely among the different levels of sedation (Fig. 3).
|
The N100 response of the first tone in the habituation paradigm remained highly variable, with only a trend towards lower values in the dex/remi group (from –5.9 ± 1.3 at baseline to –3.5 ± 2.7 at RS4; n.s.), while it decreased nearly uniformly in the mida/remi group, from –5.3 ± 1.3 at baseline to –0.4 ± 1.1 at RS 4 (P = 0.003 for main drug effect, P = 0.003 for main sedation effect, drug*sedation interaction P = 0.038) (Table 2 and Fig. 4). ANOVA of within-drug-group evolution revealed a change in the N100 amplitudes only in the mida/remi group (P = 0.001). In the dex/remi group, the change was not significant (P = 0.145). A nearly identical picture is seen in the N100 response of the 1-s paradigm (Table 2).
|
In the mida/remi group, five of nine N100 responses derived from the location Fz were identifiable at RS 4, whereas from the location Cz, only two N100 responses could be identified at RS 4 (Fig. 4). Results of the 1-s paradigm were similar; however, N100 amplitudes already became unidentifiable at lower sedation scores (three volunteers at RS 3 and four volunteers at RS 4).
Filtering with cutoff frequencies 1 and 8 Hz did not change the results. Neither the ERP latencies nor the first-to-second-tone ratio of the habituation paradigm changed at different RS (Table 2). As expected, the root mean squared power, SEF 95% and mean power frequency did not change with different drug regimes, although a significant main sedation effect could be observed (Table 2).
The predicted mean plasma concentrations based on the actual infusion rates needed to achieve the target sedation levels for dexmedetomidine were 194 ± 17 pg/mL at RS 2, 544 ± 174 pg/mL at RS 3, and 1033 ± 235 pg/mL at RS 4. Those for midazolam were 16 ± 3.7 ng/mL at RS 2, 31 ± 9.6 ng/mL at RS 3, and 56 ± 11.7 ng/mL at RS 4.
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
How can these striking differences be explained? First of all, the two drugs react with different receptors at different locations in the brain. The sedative effect of dexmedetomidine is mediated via stimulation of the -2 adrenoceptors in the locus coeruleus, whereas midazolam (and propofol) acts through the inhibitory neurotransmitter -aminobutyric acid. There is some evidence that sedation induced with dexmedetomidine resembles normal sleep (16,17); in rats, the pattern of c-Fos expression (a marker of activation of neurons) is qualitatively similar to that seen during normal non-rapid-eye-movement sleep, suggesting that endogenous sleep pathways are causally involved in dexmedetomidine-induced sedation (16). Furthermore, in volunteers breathing a 5% CO2 mixture, hypercapnic arousal phenomena occurred which are similar to those occurring during natural sleep (17). Also, in patients sedated with dexmedetomidine, retention of material by the long-term memory is significantly better than in patients sedated with propofol (18).
During sleep, the brain is able to automatically detect stimulus occurrence and trigger an orienting response towards that stimulus if its degree of novelty is large (19). Although the brain state has been shown to exert a clear effect on external information processing, indicated by a progressive decrease in the amplitude of middle-latency ERPs from wakefulness to Stage 4 of slow wave sleep (20), N100 amplitudes increase and can surpass waking values during rapid-eye-movement sleep (21,22). In the present study, during sedation with dex/remi, the study subjects were fully alert once awoken, even during the highest rates of drug infusion. This phenomenon has also been described in patients: even when on a ventilator, these patients could be easily aroused and could cooperate with procedures such as physiotherapy and radiology without showing irritation (23).
We (11) have demonstrated that the disappearance of N100 can reliably reveal the transition from light/moderate to deep sedation during sedation with propofol alone and in combination with remifentanil. This relationship was uniform among individuals and independent of whether propofol was used alone or in combination with the opioid remifentanil. The concept of changes in N100 amplitude reflecting primarily the level of sedation was supported by the lack of changes during infusion of remifentanil alone (11).
ERPs might be considered ideal for the assessment of sedation, since they represent the cortical response of the brain to a specific stimulus: exogenous components appear after detection of a sensory stimulus, and endogenous components are related to cognitive processes, such as attention and memory. In the present study, we found that in terms of reflecting the level of sedation, ERPs are also susceptible to drug-specific effects. In this respect, ERPs share the weakness of any EEG-based monitoring of sedation: the impact on electrophysiologic variables can be modified both by the specific drug and the level of consciousness. On the other hand, active neuromonitoring with ERPs may reveal important differences between sedation regimens. The preservation of N100 even at deeper sedation levels is consistent with the major difference in the clinical manifestations of the quality of the sedation: during dexmedetomidine/remifentanil, once awoken, the subjects were fully alert, while during mida/remi this was not the case. This supports the concept that the mechanisms of sedation induced by the two drug regimens are fundamentally different.
In the present study, our previous findings that the disappearance of N100 reflected excessively deep sedation with propofol and propofol/remifentanil were confirmed for mida/remi sedation: the N100 amplitudes either disappeared or approached zero at RS 4. Since RS 3–4 reflects the desirable target level of sedation in most clinical scenarios, we believe that ERPs are useful for objective assessment of sedation with these drugs in the ICU. However, this is not true for dexmedetomidine.
The findings of this study, together with known aspects of dexmedetomidine-induced sedation, have consequences for both clinical sedation and electrophysiologic monitoring. First, the observation of full alertness and cooperation of the volunteers in response to stimulation during sedation supports the concept that dexmedetomidine can have major advantages in clinical sedation. Indeed, dexmedetomidine has been used safely during and after surgery (24,25), after trauma (26), and in medical patients (23), also when infused for longer than 24 h (27). Preliminary data also suggest that brain dysfunction, including delirium, is much less common after dexmedetomidine sedation as compared to sedation with lorazepam and propofol (12). It may be argued that amnesia, a key feature of benzodiazepine sedation, is a desired effect in ICU patients. On the other hand, it is conceivable that the preserved cortical responsiveness to external stimuli during dex/remi-induced sedation may be one reason why the risk of brain dysfunction varies with different sedation regimens.
While we demonstrate in this study that ERPs are not useful for dexmedetomidine-induced sedation, the opposite is true for BIS. We can speculate that the consistent increase in EEG power in the low-frequency range, especially the peak at 3 Hz, together with the decrease in the high-frequency range, can explain, to some extent, the consistently decreasing BIS values with increasing sedation in all volunteers. Whether this characteristic explains the relatively low BIS values for a given degree of sedation remains speculative. The wide and overlapping BIS values among various sedation levels observed during midazolam sedation are consistent with other reports (28). A similar pattern has been described for propofol (11), so that even the increase of the BIS in four volunteers in the transition from RS 3 to RS 4 during mida/dex sedation can be explained by variations in these rather low levels of sedation.
The limitations of this study are the small number of volunteers and a wide range of individual N100 amplitudes, which reflect physiologic variability. We tested the hypothesis in young, healthy volunteers who do not have impaired auditory capacity. We do not know how the method performs in patients with organ failure, especially with encephalopathy or traumatic brain injury. Since it is conceivable that a loud and busy ICU environment can have a major impact on the results, our study was performed in a regular ICU. However, the findings with midazolam were similar to those obtained with propofol outside the ICU (11).
In conclusion, we found major differences in resting and processed EEG and ERP characteristics between volunteers sedated with dex/remi and mida/remi. Our findings suggest that ERPs are useful for the objective assessment of sedation with mida/remi, but not with dexmedetomidine/remifentanil. In contrast, dex/remi-titrated clinical sedation states are reflected by concomitant BIS values.
|
ACKNOWLEDGMENTS |
|---|
|
Footnotes |
|---|
Supported by Datex-Ohmeda (now GE Healthcare), Helsinki, Finland.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
2-adrenoceptor agonist dexmedetomidine converges on an endogenous sleep-promoting pathway to exert its sedative effects. Anesthesiology 2003;98:428–36.[ISI][Medline]This article has been cited by other articles:
|
|
 |
|
  V. Bonhomme, P. Maquet, C. Phillips, A. Plenevaux, P. Hans, A. Luxen, M. Lamy, and S. Laureys The Effect of Clonidine Infusion on Distribution of Regional Cerebral Blood Flow in Volunteers Anesth. Analg., March 1, 2008; 106(3): 899 - 909. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Oda, S. Toriyama, K. Tanaka, T. Matsuura, N. Hamaoka, M. Morino, and A. Asada The Effect of Dexmedetomidine on Electrocorticography in Patients with Temporal Lobe Epilepsy Under Sevoflurane Anesthesia Anesth. Analg., November 1, 2007; 105(5): 1272 - 1277. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. M. Yuen, M. G. Irwin, T. W. Hui, M. K. Yuen, and L. H. Y. Lee A Double-Blind, Crossover Assessment of the Sedative and Analgesic Effects of Intranasal Dexmedetomidine Anesth. Analg., August 1, 2007; 105(2): 374 - 380. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
