The authors studied the effects of ketamine and rocuronium on the Bispectral Index, A-Line auditory evoked potential index, state entropy, and response entropy during a calculated steady state anesthesia with propofol and remifentanil.
After ethics committee approval, 42 patients were allocated to four groups. Baseline measurements were performed after implementing a calculated steady state anesthesia with propofol and remifentanil. The control group received no additional medication. The ketamine group received a bolus and continuous infusion of ketamine. The rocuronium group received a bolus of rocuronium. The rocuronium-ketamine group received both. All data were stored during 15 min after baseline. After inspection of the raw data, the authors conducted an explorative statistical analysis.
No significant changes were found in the control group for any of the monitors. Mean values decreased in the rocuronium group for the A-Line auditory evoked potential index, Bispectral Index, and response entropy, but not for state entropy. In the ketamine group, the A-Line auditory evoked potential index and Bispectral Index did not change significantly, but state and response entropy increased. In the rocuronium-ketamine group, the A-Line auditory evoked potential index and Bispectral Index did not decrease as found in the rocuronium group. Response and state entropy increased significantly.
The response of all monitors after ketamine administration is not affected by simultaneous administration of rocuronium. Interpretation of all studied indices must be done cautiously while taking into account the clinical setting during measurement.
THE Bispectral Index (BIS), response entropy (RE), state entropy (SE), and a recent version of the A-Line® auditory evoked potential index (AAI1.6) have all been investigated as surrogate endpoints of hypnotic drug effect. Previously, their behavior during propofol and remifentanil infusion was studied.1–4However, little literature is available on the behavior of the new AAI1.6 in combined anesthesia conditions.
Several reports have been published on the behavior of indices derived from electroencephalography and midlatency auditory evoked potentials (MLAEPs) during clinical conditions characterized by an increased high-frequency electrical activity in the power spectrum of the electroencephalogram.5–11These conditions can be both drug induced (e.g. , by ketamine) or caused by electrophysiologic interference (e.g. , by electromyography).5–12
Ketamine is a dissociative anesthetic drug, inhibiting the N -methyl-d-aspartate receptor. On the raw electroencephalogram, ketamine evokes an epileptiform activation with high-frequency features.13This distortion interferes with the fast extracting index calculations as demonstrated by an increase in mean BIS, RE, and SE.14The mean values of the former version of the A-Line® auditory evoked potential index, which is based solely on MLAEP, is not affected by ketamine administration.5The effects of ketamine have not yet been studied for the new version of the A-Line® monitor (Danmeter A/S, Odense, Denmark), which calculates a composite index (AAI1.6) extracted from the electroencephalogram, MLAEP, and burst suppression.
Because the N -methyl-d-aspartate receptor modulates the basic locomotor rhythmicity in cats, one could hypothesize that the change in the electroencephalographic spectrum evoked by ketamine might be partially derived from altered muscular activity.15Few data are available in current literature on the spectral characteristics of the frontal and retroauricular electromyographic activity in both the awake and anesthetized conditions. However, because other facial muscle groups are able to evoke an electrical activity with a power spectrum ranging from 0.4 to 512 Hz, we can expect a potential overlap of electromyographic activity with the frequency domains used for any of the investigated index calculations.16Because the detection of the electromyogram is not standardized between monitors, electromyographic activity might be present while going unnoticed by any of the monitors under investigation. Only the administration of a neuromuscular blocking agent (NMBA) is able to exclude all electromyographic interferences, in both the high- and low-frequency bands of the power spectrum analysis.
The aim of this study was to investigate whether the effects of ketamine on BIS, AAI1.6, SE, and RE are influenced by the simultaneous administration of rocuronium during propofol and remifentanil pseudo–steady state anesthesia. To answer this question, we performed a two-step analysis on one control group and three study groups receiving rocuronium, ketamine, or both. For each group, we compared the changes of the studied indices over time versus their respectively baseline values. Second, we performed a time-synchronized analysis between groups to find significant differences between the control group and the respective study groups.
Materials and Methods
This study was evaluated and approved by the institutional ethics committee (University Hospital, Ghent, Belgium). After obtaining written informed consent, 42 patients, all with American Society of Anesthesiologists physical status I or II, aged 18–65 yr, and scheduled to undergo gynecologic, urologic, or plastic surgery, were included. Premedication with benzodiazepines or anxiolytic drugs was not allowed. Exclusion criteria were anxiety necessitating benzodiazepines, the use of psychoactive drugs, a history of hearing disorders, or a neuromuscular disease.
A silent operation room was obtained for all patients. Electroencephalographic and MLAEP-derived measures, hemodynamic and ventilation parameters, and accelerometry were monitored as described in a later section. Induction was performed through an 18-gauge intravenous line, placed in a forearm vein. We took care not to infuse more than 200 ml of crystalloids before and during the study period to avoid excessive hemodynamic interference.
Anesthetic Drug Administration
Propofol and remifentanil administration was performed by a Fresenius Base A modular infusion pump (Fresenius Vial Infusion Systems, Brézins, France) connected via an RS-232 interface to a computer running RUGLOOPII.∥This computer-assisted continuous infusion device captured all monitored data while driving the pumps and calculating all pharmacokinetic–pharmacodynamic parameters in a time-synchronized way. Averaging of the data was performed using a 10-s interval.
After appropriate preoxygenation, remifentanil was targeted to a 2-ng/ml effect site concentration using the pharmacokinetic–pharmacodynamic model published by Minto et al. 17,18After 2 min of remifentanil infusion, 1% propofol was started at a constant speed of 300 ml/h until loss of consciousness was detected, evaluated by the clinical signs of “loss of eye lash reflex” and “loss of response to name calling.” At that point, the calculated effect site concentration of propofol, using the pharmacokinetic–pharmacodynamic model published by Schnider et al. ,19,20was locked and maintained in RUGLOOPIIas published previously.5
After insertion of a laryngeal mask and implementation of volume-controlled ventilation with the Datex ADU ventilator (Datex, Helsinki, Finland), ventilation parameters were adjusted to target an end-tidal carbon dioxide tension between 35 and 38 mmHg. After controlled ventilation was commenced, a 10-min equilibration period was maintained (more than three times the time to peak effect for propofol and remifentanil) to reach and maintain a pharmacokinetic–pharmacodynamic pseudo–steady state for both propofol and remifentanil calculated effect site concentrations. We carefully selected this level of anesthetic drug effect because it is a relevant situation for clinical anesthetic practice compatible with a surgical level of anesthesia.
Baseline measurements were registered during 1 min after the equilibration period, and randomization was performed. The patients were allocated to one of four groups. The first group was the control group (CONTROL), meaning that no additional medication was administered during the study period. The second group (KET) received a bolus of ketamine (0.4 mg/kg) followed by a continuous infusion of ketamine (1 mg · kg−1· h−1). The third group (ROC) received a single bolus of rocuronium (0.9 mg/kg), and the fourth group (ROC + KET) received both rocuronium and ketamine simultaneously in the same dose as mentioned above. After administration of the study drug according to randomization, all measurements were logged into the computer for a 15-min study period. After this study period, the laryngeal mask was replaced by an endotracheal tube if necessary, and surgery was commenced.
The BIS XP® (Aspect Medical Systems, Newton, MA), spectral entropy (Datex-Ohmeda, Helsinki, Finland), and A-Line® monitors (Danmeter A/S) were attached to the patients. For BIS, a BIS® sensor (Aspect Medical Systems) was attached to the right side of the patient’s forehead. For A-Line®, three electrodes (A-Line® auditory evoked potential electrodes; Danmeter A/S) were positioned at the mid-forehead (+), left forehead (reference), and left mastoid (−). For RE and SE, we used an Entropy Sensor® (Datex-Ohmeda) attached to the left side of the head. The Entropy Sensor® was always attached closest to the eyebrow, the BIS® electrode was attached higher on the forehead, and the A-Line® auditory evoked potential electrodes were attached in between both.
The BIS was derived from the frontal electroencephalogram (At-Fpzt) and calculated by the BIS XP® monitor (Aspect Medical Systems). The smoothing time of the BIS® monitor was set at 15 s.
The AAI1.6 was calculated using version 1.6 of the A-Line® auditory evoked potential index monitor (Danmeter). The algorithm extracts information from the electroencephalogram, burst suppression, and MLAEP as described elsewhere.1MLAEPs were elicited with headphones producing a bilateral click stimulus of variable click intensity and 2-ms duration. The click intensity is automatically adjusted according to the measured signal-to-noise ratio to avoid interfering startle responses. By using an autoregression method with exogenous input, the AAI1.6 can be calculated within a short delay time of approximately 6 s.
Response entropy and SE are calculated using the Datex M-Entropy® module (Datex). The algorithm has been published elsewhere.21
Hemodynamic and NMBA Monitoring
Electrocardiogram, capnography, oxygen saturation, and noninvasive blood pressure measurements were monitored by the Datex S5® monitor (Datex) and automatically logged into RUGLOOPII. The hemodynamic data were averaged every minute.
An accelerometry monitor (S5® module; Datex) was positioned at the nervus ulnaris on the contralateral side of the intravenous line to detect the train-of-four percentage on the musculus adductor pollicis. After the patient had lost consciousness, a train-of-four percentage calculation was performed every 10 s to evaluate the curarization level during the study period. The accelerometry was calibrated in all patients according to the manufacturer’s guidelines before any NMBA was administered.
The significance level was set at 5% unless otherwise reported. Significant difference in demographic characteristics was tested using an analysis of variance with a Dunnett multiple comparisons test if appropriate. Statistical analysis was performed in an exploratory setting. This means we decided which statistical test to use after having inspected the raw data first.
There is no statistical accepted standard for analysis of trend data. All regression models become very complex, without any assumption about the development of data over time. In our case, such a model would need 16 parameters (16 time points), 3 parameters per group (with one reference group), and 15 times 3 parameters for all possible interactions between time and groups. Moreover, one could argue that the correlation between values within each patient would also have to be considered.
For these reasons, we chose to analyze the data in two steps. In the first step, we analyzed differences within each group (or, more precisely, if there were differences with respect to baseline). In a second analysis, we investigated whether there were differences between groups, without taking into account the time points.
For every individual patient, the raw data obtained from the studied monitors were averaged on a minute-by-minute basis starting with the baseline measurements resulting in one mean BIS, AAI1.6, RE, and SE per minute. By using an individual mean per minute, the natural fluctuation of the raw data obtained from all tested monitors is included in the analysis. Moreover, a change in response of a studied monitor must be consistent during the main part of every minute before it will trigger the statistical test to indicate significant difference. These individual means were averaged within every study group (CONTROL, ROC, KET, and ROC + KET), resulting in a group mean of means and SD for every studied minute. The statistical tests for the within- and between-group analyses were chosen after inspection of the raw data and after testing of normality using the Kolmogorov-Smirnov test.
For the within-group analysis, we performed a repeated-measures analysis of variance with Dunnett post hoc test for multiple comparisons to be able to detect significant differences between the baseline means of AAI1.6, BIS, SE, and RE and the consecutive minute-by-minute means within the CONTROL, ROC, KET, and ROC + KET groups (P < 0.05).
In a second analysis (the between-group analysis), we compared the results of the control group with the respective study groups in a time-synchronized way. We compared all study groups versus CONTROL, ROC versus ROC + KET, and KET versus ROC + KET. All of the baseline values were subtracted from the respective mean value of every consecutive minute to better reflect the amount of change versus baseline and to avoid the drawback of large baseline variability in the data set. For every comparison, the absolute difference of the means with the corresponding 95% confidence interval, based on the t distribution, was calculated. When the confidence interval of the difference of the means includes zero, the measured difference of means between the studied groups is based on coincidence. When zero is not included in the interval, the measured difference between means is statistical significant (P < 0.05).
The detection of electromyographic activity is not comparable between devices because a different definition is used in every monitor. Moreover, in some patients, conflicting results in detection of electromyographic activity were seen between monitors. Therefore, we only present electromyographic measurements results in a descriptive manner.
Forty-two patients were included in this study. One patient was excluded post hoc because of a failure in the train-of-four measurement. Demographic data are shown in table 1. No significant differences were found between groups considering age, weight, height, lean body mass, body surface area, ideal body weight, and the proportion of males versus females.
Hemodynamic parameters did not change in a clinically significant way for any group. The train-of-four percent was 0 for all patients receiving rocuronium from minute 3 until the end of the study period, indicating a clinical comparable drug effect of rocuronium during measurements.
The results for the within-group analysis are depicted in figures 1–4for AAI1.6, BIS, RE, and SE, respectively. In CONTROL, no significant differences were found for all monitors when comparing baseline values with the consecutive minute means. In ROC, a significant decrease was seen for AAI1.6, BIS, and RE but not for SE. In KET, a significant increase was seen for RE and SE but not for AAI1.6 and BIS. Although the mean BIS showed an increasing trend after administration of ketamine, this trend never reached significance because of a wide range of BIS responses within our population. In ROC + KET, the mean RE and SE increased significantly. In ROC + KET, AAI1.6 decreased, but the decreasing response of RE and BIS seen after monoadministration of rocuronium disappeared after ketamine was associated. BIS did not increase significantly in the ROC + KET group. Again, this was probably caused by the large SDs reflecting major differences in response between patients in this study group.
Figures 5–8show the results for the comparison of the absolute change from baseline between all study groups for AAI1.6, BIS, RE, and SE, respectively. Five comparisons were made: CONTROL versus KET, CONTROL versus ROC, CONTROL versus ROC + KET, ROC versus ROC + KET, and KET versus ROC + KET.
When comparing CONTROL versus KET, the absolute change from baseline of the AAI1.6 was comparable at all times. For BIS, RE, and SE, the absolute change from baseline was significantly larger for KET compared with CONTROL, during a considerable duration of time.
When comparing CONTROL versus ROC, AAI1.6, RE, and SE did not change significantly from baseline in ROC compared with CONTROL. For BIS, only on minute 6 was the absolute change from baseline significantly larger in ROC compared with CONTROL.
The comparison between CONTROL versus ROC + KET showed no difference in changes from baseline for AAI1.6 and BIS. RE and SE had a significantly larger change from baseline in ROC + KET compared with CONTROL during several minutes.
When comparing ROC versus ROC + KET, AAI1.6 and BIS did not show a significant change from baseline. RE and SE had a significantly larger change from baseline in ROC + KET compared with ROC during several minutes.
When comparing KET versus ROC + KET, no differences were found in the absolute changes from baseline for AAI1.6, RE, and SE between groups. For BIS, the changes from baseline were significantly larger in KET compared with ROC + KET from minute 3 to minute 8.
Persistent electromyographic activity was detected throughout the study period in two patients of the CONTROL group. One case was detected by the A-Line®, and one was detected by the entropy monitor. For the KET group, no major electromyographic activity was detected by any monitor; however, the difference between RE and SE gradually increased during the study period from a mean difference of 1.3 to 2.2. For the ROC + KET group, no electromyogram was registered by the A-Line® or BIS®, but on the entropy monitor, a comparable gradual increase in mean difference between RE and SE was seen (from 1.13 at baseline to 2.26 at minute 15). For the ROC group, both the A-Line® and the entropy monitor detected electromyography in the same patient from baseline to minute 2 of the study period. The BIS® monitor did not detect any electromyographic activity in any of the studied patients.
One patient of the ROC group kept showing electromyographic activity throughout the study period on the A-Line® monitor, even after the administration of rocuronium. This electromyographic activity was not detected by BIS® or entropy. Unfortunately, a failure in the train-of-four monitor made it impossible to evaluate the intensity of the NMBA effect in this particular patient. Therefore, we excluded this patient from the analysis.
In this study, the response of four electroencephalographic and MLAEP-derived depth of anesthesia monitors have been investigated in clinical anesthetic conditions, during which ketamine and rocuronium were administered in a solitary or combined way. The response of all monitors was compared with baseline values (within-group analysis) and with a control group, receiving a comparable calculated steady state anesthesia as given in all study groups (between-group analysis).
For the CONTROL group, all monitors remained unaltered after baseline measurements, indicating a comparable clinical and pharmacologic calculated steady state anesthesia between patients during the study period.
In the KET group, the within-group analysis suggested that AAI1.6 and BIS were not affected significantly by ketamine. In contrast, RE and SE showed an increasing response after solitary ketamine administration.
One could speculate that the addition of electroencephalogram-extracted information to the MLAEP-derived information, as done for AAI1.6, does not alter the index behavior compared with the former version of the A-Line® monitor, which was based on MLAEP alone.5However, it is inherent in the new algorithm that the electroencephalogram-derived information only becomes an important covariate for index calculations in conditions with low signal-to-noise ratios for MLAEP detection.1In our setting, all environmental interference was avoided, causing a high signal-to-noise ratio, which might have been sufficient to avoid the use of electroencephalogram-derived information. As such, we can not exclude the occurrence of a more pronounced effect of ketamine during more challenging conditions for MLAEP extraction (e.g. , during surgery).
For BIS, the result of the within-group analysis of KET is in contrast with older publications indicating an increase in BIS after ketamine administration.5,14Although figure 2shows a clear increasing trend for BIS in KET, this trend is not significant because of the large SDs, which are a reflection of the large variability in BIS responses within our population. In contrast, the between-group analysis has more power to detect significant differences compared with the within-group analysis, due to the elimination of the baseline variability in the data set. This was done by subtracting the baseline mean from the mean of every studied minute. As such, this new data set reflects the absolute change from baseline. In figure 6, this resulted in a significantly larger change from baseline for BIS in KET compared with CONTROL during several minutes. The less pronounced effects of ketamine on BIS in our study compared with the known literature are probably related to the deeper level of anesthesia chosen in our study. Because no surgical stimulation was present during measurement, the combination of remifentanil and propofol evoked a moderate level of burst suppression patterns in some patients during the study period. The number of patients with burst suppression was comparable between groups, being 3, 4, 4, and 3 patients for CONTROL, ROC, KET, and ROC + KET, respectively. It has been shown before that BIS loses monotonicity when a suppression rate is detected between 0 and 40%.1At that time, an increased interindividual variability in BIS calculations is present compared with lighter levels of anesthesia.1This increased variability, which does not occur for the other studied monitors, might explain the limited response of mean BIS after ketamine administration in our study population.
In the ROC group, rocuronium decreased the AAI1.6 and the RE compared with baseline in the within-group analysis, but this effect could not be confirmed by the between-group analysis. When comparing CONTROL versus ROC, AAI1.6, RE, and SE did not change significantly from baseline in ROC compared with CONTROL. This was in contrast to our findings in the within-group analysis for AAI1.6, BIS, and RE. Apparently, the electromyogram is a main portion of the baseline interindividual variability of AAI1.6, BIS, and RE. By eliminating the baseline variability in our between-group analysis, the test does not detect any difference anymore between CONTROL and ROC. These findings suggest that the effect of rocuronium on the mean values of AAI1.6, RE, and BIS is mainly caused by a drastic reduction in interindividual variability. For BIS, the decreasing effect in ROC is more pronounced compared with the other monitors. Therefore, it is reflected in both the within- and between-group analyses. These results are in agreement with the literature.7,9,12
In the ROC + KET group, the AAI1.6 remained sensitive for the decreasing effects of rocuronium in the within-group analysis. For RE and SE, rocuronium was not able to abolish or dampen the effects of ketamine. These results suggest that the distortions in the RE and SE calculations evoked by ketamine are independent of the electromyographic activity.
For BIS, the within-group analysis was not able to detect any change in ROC + KET. Again, this is probably due to the wide variety in BIS responses in our population. However, the comparison between KET and ROC + KET indicated a larger change from baseline in KET compared with ROC + KET at the beginning of the study period. From minute 9 through minute 15, no differences were found between KET and ROC + KET. Although not significantly different from baseline, a progressive higher trend was seen near the end of the study period for ROC + KET compared with CONTROL. By combining these findings, we conclude that ketamine still has an increasing effect on BIS calculations; however, this effect is much smaller at deep levels of anesthesia compared with the known effects from the literature.5
All electroencephalographic and MLAEP-derived depth of anesthesia monitors are characterized by a large interindividual and intraindividual variability. A major reason for this variability is the occurrence of electromyographic activity. In our study, electromyographic activity was detected only by the A-Line® and entropy monitor in a limited number of patients. This was probably caused by a lower electromyographic activity at deep levels of anesthesia compared with other studies performed at lighter levels of anesthesia, during an ongoing surgical procedure or in the awake patient.8However, anesthesiologists should also be aware of the fact that all studied monitors define the electromyogram in a different frequency band of the power spectrum, ranging from 70 to 110 Hz, 65 to 85 Hz, and 32 to 47 Hz for BIS, AAI1.6, and the entropy monitor, respectively. Although a limited number of data are available in the current literature, one can speculate that electromyographic activity occurs in a much wider range and still can distort index calculations at times where no electromyogram detection is present.
The entropy monitor defines the electromyogram between 32 and 47 Hz in the power spectrum. This frequency domain is included in the algorithm for the calculation of RE, which is propagated by the company commercializing the monitor, as a fast-responding depth of hypnosis index or even as an indicator of arousal. By including a part of the electromyogram in the algorithm, the decreasing RE after solitary rocuronium administration is to be expected. However, if a depth of hypnosis monitor detects the administration of a nonhypnotic NMBA, this drawback might cause a need for redefining cutoff values of the depth of hypnosis indices, whether a NMBA is in use or not. At this time, no commercially available electroencephalogram or MLAEP-derived monitor is able to detect or filter all of the interfering electromyographic activity in a sufficiently accurate way. As such, it is our opinion that the electromyogram should be considered as noise in the data until more evidence is available to support the use of the electromyogram as an indicator of arousal or the hypnotic component of anesthesia. At this time, no evidence exists to support the 32- to 47-Hz frequency band as superior to the other frequency bands, as an indicator of arousal.
Because the effects of ketamine were larger on RE compared with SE, a progressive increase in the difference between RE and SE was seen over time in both KET and ROC + KET. This gives the impression of an increased electromyographic activity. Because rocuronium is not able to reduce the increasing RE in ROC + KET, we consider this progressive increasing difference between RE and SE to be a result of the electroencephalographic alterations evoked by ketamine, not by electromyographic activation.
As a general remark on the results of this study, we must consider the fact that the inconsistent significant levels over time, which are seen in several comparisons, can be related to the limitations in group size combined with the large biologic variability of the studied indices. The fact that the difference in a comparison is not considered to be significant does not always mean that there is no difference. We might have simply missed it.
Statistical analysis for this study was performed in an exploratory, rather than a confirmatory, setting. This means that we decided which statistical test to use after inspecting the raw data first. Although this methodology might be prone to bias, we preferred to do so because we could not be sure how the model would behave in advance. Moreover, because the biologic variability of electroencephalographic and MLAEP-derived indices is large, we preferred to proceed with statistical analysis only after we found a meaningful trend in the raw data. The interpretation of our results should be considered with this drawback in mind.
In conclusion, we investigated the effects of ketamine and rocuronium on four electroencephalographic and MLAEP-derived parameters. Ketamine has no effect on AAI1.6 calculations during anesthetic conditions with a high signal-to-noise ratio for MLAEP extraction. AAI1.6, BIS, and RE decrease after rocuronium administration. RE and SE increase after ketamine administration, independent of the effects of rocuronium. In our population, BIS has a less pronounced response compared with the known literature, after the administration of ketamine. We hypothesize that this is caused by the higher variability of BIS calculations at deep levels of anesthesia compatible with burst suppression patterns.