Changes in the Auditory Evoked Potentials and the Bispectral Index following Propofol or Propofol and Alfentanil

This was a prospective study conducted in 20 paid healthy male or female volunteers. Institutional review board approval was obtained, and all volunteers gave written informed consent. Those with known neurologic disorders, including current use of anticonvulsant or other psychoactive medications, long-term drug or alcohol use, hypertension, or other serious medical conditions that would interfere with response analysis were excluded. The volunteers were divided in two groups of 10 subjects according to which drug they received: Propofol alone or propofol and alfentanil. These volunteers were also part of two other studies, the results of which have already been published. 17,18 

All volunteers underwent a history and physical examination before being enrolled in the study. All were instructed to take nothing by mouth from midnight the night before the study. On the morning of the study, volunteers had an intravenous catheter inserted for fluid and drug administration. A radial arterial catheter was also inserted to monitor blood pressure and to obtain blood samples for subsequent measurement of drug concentration. The volunteers were also monitored using a standard three-lead electrocardiograph and a peripheral pulse oximeter for SpO². They had nasal prongs to receive oxygen and to monitor end-tidal CO²and respiratory rate. An observer monitored respiratory and cardiovascular function and determined the need for interventions such as jaw support to maintain an adequate airway.

EEG signals were acquired using gold cup electrodes applied to the scalp with collodion or cream. Skin impedance was maintained at less than 5 KΩ. The following leads were recorded: Fp1 and Fp2 using Cz as the reference (channels 1 and 2), two additional frontal locations between the preauricular point and the lateral canthus using “Fpz” as the reference (channels 3 and 4), and a ground electrode. The EEG was recorded continuously using an Aspect A-1000 EEG monitor (Aspect Medical Systems, Natick, MA), which also computed the BIS in real time. This instrument has an amplifier band pass of 0.16–80 Hz, a sample rate of 16,384 samples/s, and multiple artifact rejection criteria that are integrated into the BIS algorithm. 19Data averaged from the combined bifrontal leads (channels 1 and 2) are presented here, although similar results were obtained from all channels.

The digitized raw EEG and the computed BIS values were recorded using a personal computer and stored in disk files, along with time-synchronized markers describing all clinical assessments and events. Drug infusion data were also recorded in a similar manner. This allowed a full description of each volunteer’s drug delivery history, EEG, and clinical response profile on a minute-by-minute basis.

The recorded EEG was used to compute the most recent version (revision 3.0) off-line. The BIS 3.0 values corresponding to that time immediately (10–30 s) before the start of each clinical assessment were used for the subsequent analysis.

Auditory evoked potentials (AEP) were recorded using the hardware and software package EP Averager Program version 1.3 (Anesthetic Research Unit, Northwick Park Hospital, UK) on an IBM PC clone. Binaural clicks of 0.4 ms duration at a rate of 6.33 Hz and 75 dB above normal hearing threshold were presented through earphones bilaterally.

Gold cup electrodes were affixed to the scalp at Fpz and bilaterally at the mastoid (linked reference) to record AEP. Impedance of all electrodes was maintained at less than 5 kΩ. The EEG signals were analog filtered with a bandwidth of 0.5–400 Hz, sampled at 1 kHz, and digitally filtered with a high-pass filter of 25 Hz. Recording began simultaneously with stimulation and continued for 125 ms. Responses to 512 stimuli were averaged together to form each evoked potential tracing. Before measuring the waveforms, additional filtering was applied. A high-pass filter of 25 Hz and three-point smoothing were applied to reduce variability due to noise. Four evoked potentials were collected at baseline (before the administration of any drug) and at each drug infusion level. The amplitudes (μV) and latencies (ms) of the waves V, Pa, and Nb were measured manually using a ruler on printed tracings of AEP by an individual (I.C.) who was blinded to which group the volunteer had been randomized. We measured the amplitude as the height of the peak of the wave from the prestimulus baseline (time zero) of the x -axis and the latencies as its distance from the stimulus artifact. Peak V belongs to the brain stem auditory evoked potentials (BAEP), which demonstrates that auditory stimuli are correctly transduced. 20If peak V could not clearly be identified, the evoked potential tracing was rejected.

After all monitoring had been instituted, the volunteers were given a 15-min resting period. Thereafter, baseline readings were obtained. (These readings were also performed following drug administration at each target steady state drug level.) These consisted of a BIS score, AEP, measure of sedation, and two arterial blood samples for drug concentration before and after each assessment.

Sedation was measured using the responsiveness component of the Observer Assessment of Alertness and Sedation rating scale (table 1). This assessment procedure involves presentation of progressively more intense stimulation, ranging from a moderate speaking voice to physical shaking or moderate noxious stimuli (trapezius squeeze) until a response is observed.

The effect of propofol and propofol plus alfentanil on the BIS score, AEP, and level of sedation were evaluated. The intravenous drugs were administered via  a computer-assisted continuous infusion device to a target effect site concentration. 21The pharmacokinetic parameters of Gepts et al.  22were used for propofol administration and those of Scott and Stanski 23were used for alfentanil. The accuracy of these pharmacokinetic parameters has previously been validated. 16,24,25In every subject, propofol was administered in increasing steps to a target effect site concentration of 1, 2, 4, and 6 μg/ml until volunteers lost consciousness as defined by loss of response to a verbal command. The propofol concentration was decreased by the same steps until consciousness occurred and then increased and decreased so that at least two crossovers of consciousness–unconsciousness occurred. Alfentanil or placebo (saline) was administered in a double blind fashion to a preselected effect site concentration of 100 ng/ml. At each step, the target concentration was maintained for a minimum of 10 min to permit equilibration with the effect site, as modeled by the propofol and alfentanil ke0 23,26(fig. 1).

Plasma drug concentrations were determined from the arterial blood samples collected at the beginning and end of each steady state assessment period. The measured concentration at the beginning of each assessment period was used in correlating concentration to MLAEP or BIS.

Propofol determinations were made using high-pressure liquid chromatography analysis 27performed at the Anesthesia Research Laboratory at Duke University Medical Center (Durham, North Carolina). The separation and quantification procedures were conducted with a C-18, 15-cm × 4.6-mm column (Supelcosil LC-18; Supelco, Bellefonte, PA), and detection was fluorometric. Spiked standards (0.5 and 5 μg/ml) were required to be within 20% of the true value with a coefficient of variability of 12%. The duplicate assays of the participant samples were required to be within 20% of each other.

Plasma samples were measured for alfentanil in duplicate by radioimmunoassay using kits from Janssen Pharmaceutical (Olen, Belgium). The coefficient of variability averaged 11.2% over the range of concentrations encountered in this study. 28 

The relation among the different evoked potential variables or BIS, measured drug concentration, and sedation level was analyzed using generalized estimating equations. 29Generalized estimating equations were introduced by Liang and Zeger in 1986 as an extension of generalized linear models to accommodate correlated data using the same link function and linear predictor setup. Generalized estimating equations allow in addition the ability to model the covariance structure of the response measures. It models the first two moments without specifying the complete distribution of the responses. This statistical approach was chosen because it accounts for the within-subject correlation in the face of repeated measurements in the same subject. The effect of the presence of alfentanil was determined by including an indicator (group) in each of the generalized estimating equation models.

Pearson correlations of BIS and AEP variables with sedation and propofol concentration were calculated. Tests for the equality of these correlations for each AEP variable and BIS between the two treatment groups were performed using the Fisher Z transformation. Tests for the equality of these correlations within each group was also done using the single sample test of Olkin. 30Bonferroni correction was done to account for the multiple testing.

All volunteers who responded to any verbal command (sedation scores 3, 4, and 5) were classified as conscious. Those who did not respond (sedation scores of 0, 1, and 2) were considered to be unresponsive and unconscious. Logistic regression techniques were used to analyze these relations for the quantal endpoints of loss of consciousness. We determined for the plasma concentration of propofol and for the MLAEP variables the values at which 50% and 95% of volunteers were unconscious (Cp⁵⁰and Cp⁹⁵; Pa and Nb amplitudes⁵⁰and ⁹⁵and Pa and Nb latencies⁵⁰and ⁹⁵, respectively) for each group. The limits corresponding to the 95% confidence intervals surrounding the 50% and 95% probabilities of unconsciousness were calculated.

The prediction probability (P^k) as described by Smith et al. , 31was calculated for BIS, MLAEP parameters (Pa and Nb latencies and amplitudes), propofol measured, and target concentration and compared within each group using the usual Gaussian test statistic. To enable comparison of P^kvalues, we used 1 − P^kwhen the P^kvalue was less than 0.5. Bonferroni corrections were done to account for multiple tests.

Descriptive statistics were used to compare demographic variables between the study groups. Probability values less than 0.05 were considered significant. Statistical analyses were done using the SAS software of SAS Institute (Cary, NC).

There were no statistically significant differences between the two groups with regard to age, weight, height, and gender (table 2). Typical changes in MLAEP with increasing propofol concentration are illustrated in figure 2. Peak V amplitude of the BAEP did not change significantly either with the increase of the plasma propofol concentration (P = 0.6) or with increasing level of sedation (P = 0.7). Peak V latency was slightly prolonged by both conditions (P = 0.03 with propofol concentration;P = 0.015 with sedation level). All other parameters showed highly significant changes (decrease of BIS index and Pa and Nb amplitudes and increase of Pa and Nb latencies) with increasing plasma propofol concentrations (P < 0.0001) and increasing sedation levels (P < 0.0001) with and without alfentanil (figs. 3 and 4). These changes were comparable between groups for all AEP parameters analyzed except for Pa and Nb amplitudes, which were more pronounced in the group receiving propofol alone.

The BIS index showed a better correlation with propofol concentration (P < 0.00005) and sedation level (P < 0.0000005) than did the MLAEP. This is illustrated in figures 3 and 4, where the relation between change in sedation score and the BIS or MLAEP parameter values are plotted. Among all waveform parameters measured, Pa and Nb latency showed the best correlation with propofol concentration and sedation level (P < 0.005). MLAEP correlations were comparable between the propofol and propofol–alfentanil groups (P > 0.7;tables 3 and 4).

The probability of consciousness versus  BIS, Pa or Nb latency, and PA or Nb amplitude is plotted in figure 5. BIS and latency provided better predictors of consciousness, as reflected by the slope of the probability curves. The addition of alfentanil did not significantly affect the values of Pa and Nb amplitudes and latencies associated with a 50% probability of loss of consciousness (table 5;fig. 5;P > 0.15). Alfentanil did, however, lower the propofol concentration required for loss of consciousness in 50% of subjects (table 5;fig. 5;P = 0.04).

The P^kfor BIS, MLAEP parameters, propofol target, and measured concentration are listed in table 6. The BIS was a better predictor of loss of consciousness than was Pa and Nb amplitude (P < 0.002). The MLAEP parameters were comparable as predictors of loss of consciousness. The P^kvalues calculated for the MLAEP parameters were not affected by the presence of alfentanil (P > 0.1).

In this volunteer trial, increasing doses of propofol resulted in graded changes on the MLAEP which correlated well with the level of sedation. The addition of alfentanil decreased the propofol Cp⁵⁰for loss of consciousness but did not significantly affect the MLAEP wave amplitudes and latencies⁵⁰for the same endpoint. The BIS correlated better with the level of sedation than any of the MLAEP parameters. Among all the MLAEP parameters, Pa and Nb latencies showed the best correlation with the sedation level and were comparable to the BIS as a predictor of loss of consciousness.

Increasing doses of propofol associated with increasing sedation scores affected wave V of BAEP only slightly (no change in peak V amplitude and only a slight increase in peak V latency). Conversely, propofol caused major changes in MLAEP (significant decrease in Pa and Nb amplitudes and increase in Pa and Nb latencies). The MLAEP has been shown to be suppressed by many hypnotic drugs in a dose-dependent manner. 2–10In these studies, the MLAEP changes were analyzed in relation to variable doses or concentrations of hypnotic. General anesthetics alter the MLAEP and, by inference, the processing of auditory stimuli. The effects of general anesthetics can in part be reversed by surgical stimulation, in regard to amplitude more than latency. In patients given a light and stable anesthesia, Thornton et al.  demonstrated an increase in the amplitudes of the MLAEP following surgical stimulation without reversing their latencies. 11In patients administered continuous epidural analgesia with general anesthesia, Schwender et al.  showed a decrease in latencies and an increase in amplitudes of the MLAEP before and during spontaneous movements observed intraoperatively or during emergence of anesthesia. 12It is thus important to establish, in the absence of any stimulus, the relation between changing levels of sedation and changes in MLAEP. In our study, we demonstrated not only a good correlation between the MLAEP changes and propofol concentration but also a good correlation between MLAEP parameters (especially latency) and sedation score.

Our study showed that the MLAEP parameters analyzed were comparable as predictors of loss of consciousness, but Pa and Nb latency had the best correlation with propofol concentration and sedation level. Previous studies have tried to extract from the AEP an index able to predict intraoperative wakefulness, awareness, and recall. 32–39Thornton et al.  analyzed a characteristic MLAEP pattern reflecting the change between wakefulness (response to command detected by isolated arm technique) and light anesthesia (no response) and identified a threshold Nb latency of 44.5 ms as the best feature distinguishing these two states. 32Similar to our study, Newton et al.  demonstrated that the largest changes accompanying the transition from partial to no response to verbal command (consciousness to unconsciousness) were in Pa and Nb latencies. 33Schwender et al.  showed that a threshold of Pa latency increase of greater than or less than 12 ms provided a sensitivity of 100% and a specificity of 77% in distinguishing patients with implicit memory from patient without implicit memory postoperatively. 34In contrast, Heneghan et al.  6compared the slopes of the changes of the MLAEP parameters against minimum alveolar concentration multiples for isoflurane, enflurane, and halothane. They demonstrated that Pa and Nb amplitude were more promising indices of the depth of anesthesia than were Pa and Nb latencies because they were not quantitatively differently affected by the three volatile agents and were qualitatively similarly affected by etomidate and althesin. Mantzaridis and Kenny derived mathematically from the AEP a single numerical parameter reflecting both amplitudes and frequencies known as the auditory evoked potential index (AEPidx). This index proved to be reliable and comparable to Nb latencies in distinguishing consciousness from unconsciousness. 37 

The BIS correlated better than the MLAEP with propofol concentration and sedation level. Similar to our results, Doi et al. 38found that the BIS correlated better with calculated blood propofol concentration during emergence from anesthesia than did the AEPidx mentioned previously. This would indicate that the MLAEP is a good indicator of the transition between consciousness and unconsciousness but it is less reliable in tracking these states. The BIS is derived from several parameters of the EEG and has been optimized to correlate with increasing sedation. It is therefore not surprising that individual parameters of the MLAEP are not as good in correlating with increasing levels of sedation as the BIS. Additionally, our study revealed that the BIS was a better predictor of loss of consciousness than Pa and Nb amplitude but was comparable to the prediction properties of Pa and Nb latencies. Two studies comparing the AEPidx and the BIS as detectors of awareness showed that the AEPidx was the best at distinguishing the transition from unconsciousness to consciousness. 38,39The differences between our study and those previously reported are multiple and include the use of an index versus  the actual MLAEP parameters, the use of different clinical endpoints to define unconsciousness (loss of response to speaking voice vs.  absence of eye opening to command 38and loss of response to a verbal command to squeeze the investigator’s hand 39), and the use of different statistical approaches.

The MLAEP changes with the administration of propofol were not affected by the addition of alfentanil. Our results agree with other studies 13,14showing a preservation of the BAEP and slight changes in the early cortical waves of the MLAEP during administration of high doses of opioids. The MLAEP are the AEP components occurring 10–100 ms after stimulus presentation and originate (mainly Pa) from the primary auditory cortex of the temporal lobe. 40,41In the presence of high doses of opioids, the auditory stimuli can be transduced and conducted in the brain stem and may be processed in the primary auditory cortex. The auditory input is often described as the last sensory modality to be abolished and the first to return at light level of anesthesia. 42As there is a parallel between perception of auditory stimuli and consciousness, the absence of effect of opioids on MLAEP can be related to reports of higher incidence of intraoperative awareness when opioids are the primary anesthetic. 43,44 

At present, MLAEP have four inconveniences which may detract from their use in clinical practice. First, they need considerable time to produce a response (0.5–5 min to build up an average for the AEP compared with continuous display of an index for the BIS). A moving time averaging technique described by Davies et al.  45has overcome this problem, allowing a faster response time (every 3 s). Second, difficulties exist in interpreting the waveforms and on-line measurement of Pa and Nb latencies. The AEPidx mentioned previously can now be derived on-line. Third, technical difficulties exist in obtaining and recording AEP (set-up time > 5 min). Finally, AEP requires auditory stimuli and therefore is applicable only to patients with preserved hearing. However, based on the data from this study, MLAEP are promising as monitors of increasing sedation and loss of consciousness.

MLAEP changes correlated well with increasing sedation produced by propofol and were not affected by the presence of alfentanil. The BIS correlated better with the level of sedation than any of the MLAEP parameters. Among all the MLAEP, Pa and Nb latencies showed the best correlation with propofol concentration and sedation level and were comparable to the BIS as predictors of loss of consciousness.

The authors thank Kevin Weatherwax, Department of Anesthesiology, Duke University Medical Center, for his help with the data and graphs.